ML20099J490

Applicant Exhibit A-146,consisting of Undated Annex G, Radiation Carcinogenesis in Man

ML20099J490
Person / Time
Site:	Limerick
Issue date:	05/22/1984
From:	AFFILIATION NOT ASSIGNED
To:
References
OL-A-146, NUDOCS 8411290079
Download: ML20099J490 (64)

Text

{{#Wiki_filter:$., 3 NUC EA9 REGtJLATORY C0?.tMissicN m 1 Docket No $c_ . eg,3,, g, g, g}g Q 7-il ' f q fn the r t;.tte< cf g. 1 l p Staff / [j-A[ f ? N 'O Ap:N; ant - - -_l ' ~ g,.. s \\ _-] ~ ee s_. -s n, ~ Interw-r _ ~~ l'~ if G n- ~.7. . Ntun m o ,-j\\ pf ' C;!*;;;t -.. _ _ p n C*3 Q f, C'hsr__ '>'n w~ .. s.:.c ~ p e Reputes p g9 b Radiationi carcinogenesis isp anan e Ii-br a e CONTENTS

==m,a, INTRODUC770N................... I.2 IV. BREAST CANCER,.............. 151 199 151 164 A. Breast cancer in AM mrvers L GENERAL CONSIDERATIONS 3-62 B. Breast cancer following exposure in 165-174 A. IAngth of survey and latent period for a,,,,,,,a, nadiology 4 12 C Breast enacer following redsothesary to development of mehgnancies 175 188 1318 the breast................. B. Ascertausenent.............. 339.g99 C Information on absorbed does 19-27 p, g,,,,,y D. Suitabihty of does levet and does 200-226 28 36 V. LUNG CANCER .1-distreution E. 37 42 A. IJang cancer in A-bomb servvvors.... 201 207 F. Sustabshty of compennon populations Further factors affecting the preusson B. Long cancer in workers exposed to 43-47 208 222 of risk estimates............. @ redon M..... G. weighting factors for neutrons 4842 223-226 C Semsnary II. LEUKAEMIA 63-96 VI. BONE TUMOURS 227 236 A. IAskaemia in A-bosnb survivors 63 76 VilOTHER CANCERS.............. 237 302 B. leukaemia in populations in the Mar. 253-256 shan hiands irradiesed froen fallout 77 A. Brain.................. 257 263 C leuksenism fot'owmg pelvic irradiation 78-82 B. Sahvary glands C. Mucosa of cranial sinuses 264 265 - D. IAskaems followins treatment of anty. lomas spondyletis by x irradiation 33 34 D. Dissettve organs.*............. 266 276 277 282 E. Fotvic oryne E. leuksemia fotowing other ramologleal 283-285 procedures................. 85 39 F. liver................... 90596 G. "W** lysmphoma and multiple mye-F. Summary 286 292 lonia III. THYROID CANCER.. 97 150 A. Dysoid cancer in A. bomb survivors 100 107 Vill. MALIGNANCIES IN FRE. NATALLY ~ EX. 303-308 B. Thyroid cancer in a - in the POSED CHILDREN Marshall Islands exposed to thysoid 309-332 irradiation from fauout 108 111 IX. CONCLUSIONS C Dyroid cancer in patients therapeuti. A. C--- ^ in efferent orpas.... 309-319 cally irradiated froen external sources 112-138 B. Indications for future work 320 332 l D. Thyroid cancer in patients tressed with "'I 139 141 417 142-150 References................... t' _L Summary t Introduction will occur following a given radiation exposure. Since the radiation may involve the whole body more or less -1. In assessing the harm that might result from any uniformly or individual organs or tissues selectively. it is form of human exposure to ionizing radiation, it is necessary to examine the serisitivity of the different necessary to identify the types of injury that may be tissues as well as of the body as a whole. Moreover, to be s caused and to estimate the frequency with which each of value in estimating the effect of environmental or 361 8411290079 840522 DRADOCK05000g

occupational exposures, the estimates should be A. LENGTH OF SURVEY AND 1.ATENT PERIOD applicable over a rany of doses down to very low ones FOR DEVEl.OPMENT OF MALIGNANCIES delivered either at low dose rates, or in separate repeated fractions delivered at high or low dose rates. 4. It has repeatedly been shown that after radiation exposure, malignant tumours may continue to become Q 2. There is increasing evidence that in human beings detectable in excess in the exposed population for long (as in animals; see Annex I, paragraph 26), the induction Periods, often several decades. Much of the " latent l of mahgnancies represents the most important effect Period" observed in practice between the exposure and l produced at low doses in the exposed individual, and the the detection of a tumour must be due to the time 1 frequency with which such chanps a.re induced in required for sufficient increase in size of the tumour to l different tissues has been exammed by the Committee in make it detectable. The estimated mean latency of a 4 its previous reports. It is at present impoeible to deduce given type of tumour will therefore depend in part upon these frequencies of radiation carcinopnesis in man the methods used to detect it and upon the closeness of l } from experimental work in animals (see AnnexI, surveillance of the exposed population. It will depend paragraph 5). The following review, therefore, is of upon the rate of cell division, cell survival and forms of 5 estimates of the induction of malignancies by radiation I Cal 8Pread or metastasis of the particular type of that can be denved from studies of human populations tumour, it will also depend upon the ease of in which the whole body or individual organs have been examination f the organ in which the tumour arises. irradiated; it is intended to update information detection of a small tumour obviously being much more - l presented i.s the 1972 report of the Committee efYicient in the thyroid than, say, in the h,ver or 2 pancreas. In statistics based on mortality from tumours, { the interval between irradiation and death will vary also 3 with the widely differing speed at which different a clinically diagnosable tumours cause death, and this may I. GENERAL CONSIDERATIONS depend on the forms of treatment available. It is uncertain whether part of the latency is also due to any f rm f induction period before the initially affected 3. The strength of the available information depends heavily upon the consistency of estimates obtained cell or cells start to divide and form a tumour, or before I under different conditions, even though individual the tum ur assuma "rnaugnant, charactuisu,cs of i estimates may be subject to a variety of defects. gr wth and spreading. It is clear from some investiga-j Optimally, a valid estimate of the carcinogenic effect of "' I" 8nimals that at short intugals aftu a,rradiation radiation on a given tissue will depend upon the the thyroid gland is found to contam tumours that are l following factors: nly of ** benign, histological character (22), and that 4 malignant nimours become detectable at later stages. In (a) Study of the irradiated population over a a recent detailed study in man (58) on the other hand. prolonged period of several d: cades after exposure this sequence was not evident, bemgn tumours being during which mahgnancies may become detectable; detected after a longer latency, particularly in males. than malignant tumours. Some biological transition /b/ A cnterion of diagnosis which ensures that period may indeed be involved, as well as the simple all malignancies induced or all fatalities from such growth of an initially malignant tumour to detectable malignancies are recorded; size. It has, however, been shown that lung tissue from (c/ Knowledge of the absorbed dose and dose irradiated rats may show no tumours detectable by serial rate, or any fractionation of the dose and of the varia. sectioning at short intervals of 2-3 months after tion of absorbed doses in different individuals; exposure. Yet if sections of these lungs are transplanted I into mice of a type lacking the immune mechanisms that (d) A sufficiently uniform distribution of dose would prevent the growth of transplanted tissues, 3 { through the body or through the tissue ofinterest; characteristic tumours develop in the recipient mice, l I (e) - A valid control population; showing that abnormal processes are in fact acting in the irradiated lung earlier than it is practicable to detect /// A number of malignancies in the irradiated them, even by detailed microscopic examination (79). population sufficient to give statistical reliability to the excess over that expected on the basis of the control 5. The progressive increase in size which causes Population; human tumours to become detectable is likely to l (g) An adequate basis for comparing the depend, as in animals (see Annex 1, paragraph 74), upon carcinogenicity of radiations of different quality at the rate of tumour cell division and the balance between different dose levels. Processes of cell production and cell destruction. The f estimated ** doubling time" differs widely for tumours of 6fferent types,in man as well as in animals. i 'In this Annes, the term " carcinogenesis" is used to e inchade the indaction not only of carcinoma but also of leukaemis of any other form of mahgnancy (sarcoma, lyn.phoma 6. The mean latency and the distribution of latent { 4 I etc.). The word "malappancy"is used when the reference is to all Periods of any type of tumour may thus depend upon a such mahanant conditions including leukaemia as well as sohd number of factors characteristic both of the tumour and canars, sance "canar"or "mahynant tumour" should refer only of the circumstances of surveillance and ascertainment in to solid or focal mahanancnes. The term ** tumour" is used without quahhcaten when it is either clear from the context, or the expased population (Annex 1, para.105 L Moreover. unimportant, whether a malagnant or a benign tumour is m some instances the latency of tumour development intended. appears to depend upon the age of the person irradiated ,f 3e .. _ _ _ - - _ _ _. - - ~ _ -. _ _. _. -. - _ _ _ _ - _ _ _ _, _ m

i t l (for example as discussed later for thyroid and breast were made had remained constant for long periods (e.g. 2 cancers), and the latency for development cf tumours is 50 years) and if the likelihood of ascertainment and certainly short following irradiation of the foetus in presentation for diagnosis were independent cf time utero (148). It is also likely that latency may vary with since irradiation (apart from the possibility of prior mthe size of the absorbed dose, as appears to be the case death from other causes, for which correction could l ~or example for osteosarcomas following radium-226 ordinarily be made). These conditions failin many cases a Vtcorporation in the skeleton (41) and for thyroid (for example, as discussed later for the development of cancers following external irradiation (58), although thyroid malignancies following thymic irradiation), since t' apparently not significantly for leukaemia resulting from this procedure was practised only for a limited period, A-bomb irradiation at Hiroshima and Nagasaki (para. 64 largely prior to 1960, and the distribution of observed and table 3). A similar variation of latency with latencies is probably still truncated by exclusion of the i' I dose is observed with chemically induced cancers in longer latent periods. Similarly, many " retrospective- ) animals. In all such studies it is of course important to prospective" surveys (for example,of the late effect of a I base conclusions upon the mean latency, or the latency period of radiotherapeutic practice), will underestimate t' l f;r a given percentage of tumours, as studied over latency until they have been continued for several l prolonged periods after irradiation. The " latent period" decades after the period of the practice. that elapses 'ntil the first tumour is detected will, on l purely statistical grounds, be shorter after large doses

10. In many instances, particularly of the latency with than after smaller ones if larger doses produce larger which tumours are detected after irradiation of patients numbers of tumours a.2d if tumours appear after varying of different ages, no correction is in fact made for time intervals. The relationship between latency of diminution of the exposed population due to death from l

tumour detection and dose size in experimental other causes. This results in an underestimation of the carcinogenesis is discussed in detail m Annex I number of long latencies that would otherwise be f (para.109). observed, particularly in studies on those irradiated at i ' " *E'** 7. The distribution of latent periods for different tumours ctr.not therefore be expected to conform to

11. Values quoted as mean latencies for tumour any single ietationship of frequency with time (7).

detection are, however, unlikely to overestimate the true Additional :ases of cancer aoparently induced by value, and certain surveys indicate that this value must radiation may however continue to be detected for at be large, at least for some types of malignancy and least 30 years following exposure. Estimates of the conditions for ascertainment. A number of the published carcinogenic effect of radiation, based on shorter periods estimates of mean latency of radiation-induced malig-of observation, may thus require correction, and no nancies give values exceeding 20 years (table 1). O reliable distribution curve oflatencies may be available on ") which to base an exact correction factor. For leukaemia, (, several surveys are now sufficiently prolonged from the TABLE 1. EXAMPLES OF LONG MEAN LATENCIES RE-time of exposure to demonstrate that further cases are D MM6ERAPEW IRM AN ceasing to be detected.The distnbution oflatent periods under the conditions of exposure and ascertainment of m,,,,3,,,f 3,,, of utan those surveys can thus be defined. In particular, data Reference subrects concer derency (ycers) published by the Japanese Life Span Study indicates that by 1972 the incidence of leukaemia in exposed 30 20 Thyroid 20.3 89 to stadder 20.7 survivors was close to that in the comparison group (see 10 Br st 22.6 paragraph 63). From the data of Moriyama et al. (97), the mean time interval from irradiation to death has head and neck 22.8 been about 14 years (see table 3). 47 37 Pharynx and larynx 23.4 8. For other malignancies, apart from those following V$$ g 77 II3 m irradiation in utero, no prospective survey appears yet to 4 38 Skin 24.5 have been sufficiently prolonged to record allmalignan. 121 10 Pharynx 25.0 167 130 Pharynx and cies attributable to the exposure.The Life Span Study survey shows a small reduction in the rate at which 147 40 Sk excess deaths from malignancies other than leukaemia (basal ceu) 41.5 were occurring during 1970-1972 as compared with earlier periods (in those exposed at over 10 rad T65 in both cities, both sexes and all ages, in excess of the comparison populations (97); see paragraph 239), but

12. As aa approximation in the absence of better this fall is not statistically significant.

information,it seems appropriate therefore to assume a median latency of about 25 years, so that the total of 9. Nor is the necessary information available from cancers diagnosed within this time after radiation, at any retrospective survey (for example, of the dste of least in young subjects with a long life expectancy, may represent abmt half of all cancers likely to have been p] radiation exposure ot' subjects subscquently diagnosed as i developing cancers that are presumed to be radiation-induced. Data on otherwise undiagnosed cancers found operation on the thyroid 24 years after neck 3 b induced). In such cases the distribution oflatency would st f be unbiased only if the frequency of significant irradiation in childhood support this estimate (see irradiation in the population within which the diagnoses paragraph 133). 363

~ - ~ - -,- B. ASCERTAINMENT types. In a fsw forms of cancer however, prolonged

+

survival is common, whether because cf the effectiveness

13. Problems of complete or correct ascertainment of surgical or other treatment, or because of slow arise in a number of surveys referred to below, on which progress of the cancer. In most cases, the accuracy of risk estimates depend. Moriyama and his colleagues (97) present risk estimates of cancer induction is insufficient

.O _j - that, although death certification diapostic to justify the separate estimation of fatal and non. fatal a Q of carcinoma is about 90 per cent accurate, the accuracy cancer induction. The following values were obnerved of specincation of particular sites is much less precise. (124) for the percentap of patients registered as having l Reir analysis has therefore been limited to sites where developed cancer in 1954 in England and Wales and who ' the accuracy of specification is high. De reliability of were ' surviving after 15 years (values quoted are means death certification of primary malignant tumours of for males and females): leukaemia. 2; lung, trachea and bone is also regarded as being poor in many countries, bronchus, 3: myeloma,10; brain and nervous system, owing to death from carcinomatosis with metastases to 13; bone, 20; large intestine, 24; rectum, 24; bladder, bone often being recorded as due to primary bone - 25. I tumours. his difficulty is likely to apply also to a t number of other oigans. in.which metastases from

18. The thyroid average value, 28 per cent, results tumours of other tissues commonly occur, and death essentially from the very short survival of patients with certificates may give quite unreliable information as to anaplastic cancers (50 per. cent survwal after 3 months the primary site of canar development.

(115)). De type of thyroid cancer induced by radiation, 4 however,. an-adeno. carcinoma arising from follicular 2

14. In surveys based on malignant tumour incidence cells, has an unusually slow progress and a high cure rate, rather than on mortality, retrospective analyses can and here the difference between risks of cancer sometimes be reliably based on efficient tumour induction and of fatal cancer induction 4 very large (see

. registries. Modan (92) however, points out that his paragraph 150). For the female breast and for the uterus estimates for the frequency of tumoun that can also, the 15-year survival rates are substantial, and are - commonly be removed by operation, such as those of estimated as 29 and 55 per cent respectively,while that the thyroid and skin, are likely to be underestimates, for the salivary glands is 70 per cent. l since the National Israeli Tumour Registry only came 4 - into operation 11 years after the start of the irradiations he studied and his additional check through death C. INFORMATIONON ABSORBEDDOSE i i l certificates did.not detect the incidence of these j operable tumours. The lack of estimates for skin tumour

19. A number of epidemiological surveys demonstrate induction, in general, is probably due to the ease with the occurrence of radiation carcinogenesis but cannot be which these tumours can be removed surgically as sun used in risk estimation because the size of the original j

as they are diagnosed, as well as to the low incidence of ~ doses is not known. It is important to recogmze, such tumours when only small skin areas are irradsated. however, that these surveys may still have considerable I value if a number of different organs can be assumed to

15. In the thyroid, the ease with which small nodules have received about equal doses. Given an equally can be detected implies that ascertainment is likely to be efficient ascertaintnent of tumours arising m these high in a population under close surveillance because of organs, therefore, they can be graded quantitatively for 4

a known previous radiation exposure. Thus for example, relative sensitivity to tumour induction. If, moreover, ,!~ the Marshall Islands populations exposed by fallout are the carcinogenic risk is known for one, inferences can be i examined annually and operation is suggested if palpable drawn for the others. Thus,it may be possible to infer } thyroid nodules develop. Indeed, if it is the case that in the sensitivity of the salivary glands from the ratio of the thyroid, radiation. induced malignant tumours may salivary cancers to thyroid canars following neck i i develop from nodules which initially are histologically irradiation and from independent data on thyroid benign, this surveillance involves, as intended, a reduced sensitivity, risk of the development of malignant tumours. i

20. It is equally valuable to identify the tissues in

{

16. On the other hand, the frequent lack of symptoms which no tumours develop when tumours do occur in e

i or distinctive signs from small malignant nodules of the equally irradiated organs, and so to establish an upper thyroid may imply that the incidence of thyroid cancers limit to the risk for relatively insensitive tissues. This is also underestimated in populations which are not inference can of course only be made ifit is clear that repeatedly examined medically, since these tumours may the information examined is not selectively reporting i not progress to cause death. This applies particularly to only tumours of significantly increased incidence. the so. called " occult sclerosing papillary" thyroid tumours. For the purposes of risk estimation, however,

21. In some instances, dose estimates are only

- the om smon of tumours which do not even become obtainalele on a very indirect basis. Thus for example, } - clinically detectable during life, is of doubtful the external irradiation estimates of the thyroid glands importance. of the Marshall Island populations (paras.108-1I1) are j derived from measurements which began within a few

17. In the.ase of most types of cancer discussed in days of the start of exposure and a theoretical basis for l

this Annex, an average of less than 25 per cent of extrapolation to deduce earlier doses. The internal doses patients remain alwe 15 years after detection of the from concentration of radiciodine in the gland,however, t cancer-st least as judged by the survival statistics of are largely based on the activity that was being excreted patients with " naturally occurring" cancers of those in the urine subsequent to the ingestion of the l 364 i .-e-,..--m--e,----m,,,-wmm,.n,.,ne--v,w-w-,.,-., ,-n-,,,.n,mm..m.,_ m - ,,,,~.~wm,- u

. ~ = . ~. ,.4*q radionuclides and from data en the likely relationship

26. Problems cf a different type arise in attempting to

- between excretion rates and retained activity at a given derive the risk per unit absorbed dose of tissue exposure - interval after the period of ingestion. to radiations of 1:w hnear energy transfer (LET) from " f f examP e, on the populations exposed l l

22. In some instances of non-uniform internal 8'""""

I "' radiation, there may be. uncertainties as to dose 8* " ** "8 Y " " * " " " '" I8'"""' [' (( distribution or as to the tinues over which absorbed ta at n P" unit aMed dme, as shown in mious doses should be averaged. Rus, it has become evident experimental studies in animals, prevents any simple that osteogenic sarcomas arise mainly from the endosteal in AC8 enc 7 tWn uc n P" und cells of bone and that the dose as averaged over tissues f energy absorbed (but see paragraphs 48 to 61). It has - lying within 10pm of the bone surface is the relevant in n sugguted M that unain typu of ^ [ one, while the mean dose throughout the bone is not. hus, it is necessary to use different models for '"h*may be preferentially induced by radiation j i dosametric calculations for short-lived radionuclides like radium-224, from which most of the energy of

27. Even in the nnalysis of effects of medical exposure disintegration is given up while the isotope is still present to low-LET radiation there may be difficulties in at its uutial site of deposition on bone surfaces, and for estimating the likely absorbed dose, for example in the long. lived nuclides like radium-226, which become lung, resulting from known skin exposure with specified distributed throughout bone during the course of their field size and position. De absorbed dose in bone from disintegration. Materials with long effective half-lives in radiation oflow energy may also be particularly difficult tissue present further problems m determination of the to estimate with confidence.

1 dose relevant to risk estimation, since it cannot be known how much of the dose delivered during the long latent period of any tumours that occur is relevant to D. SUITABILITY OF DOSE LEVEL tumour induction and how much is " wasted radiation" AND DOSE DISTRIBUTION occurring during the subsequent development of an l. established tumour.

28. Valuable information has been derived from eP emiological studies of patients previously treated for id

23. The occurrence of osteosarcomas following incor.

ankylosing spondylitis by local irradiation of affected poration of radium-226 is associated with the develop. Parts of the spine. These studies show an increase in ment of about one third as many carcinomas arising cancer in certain. organs that are likely to have been from the comparatively small mass of mucous membrane which covers bone surfaces in the mastoid and other heavily inadiated as a result of their frequent inclusion in the direct beam. Dese increases, or at least those intracranial cavities (41). It is not possible at present, however, to deduce the sensitivity of such membranes to following single courses of therapy, can potentially be used to evaluate the sensitivity of these organs, once carcinogenesis, since radon is formed from the radium estimates are made of the dose they will have received and retained in these cavities in conantrations that are and if due allowance can be made for the likely normal at present still undetermined, and the relevant dose is incidence of different forms of cancer in this condition. therefore unknown.

24. Rather similar problems arise in connection with

29. Estimates have also been made of the frequency

)

the carcinogenicity of " thorotrast" where the incidence with which leukaemia is induced by this treatment. ij of tumours in the liver (mainly cholangio-These estimates have then been related to the mean endotheliomas) can be related to the level of absorbed dose to the bone marrow by averaging. !I concentration of the thorium dioxide in the liver. ne throughout the, total mass of the marrow, the dose l. sensitivity of the liver to radiation carcinogenesis cannot delivered to the fra'ctions of marrow irradiated in the be established with certainty, however, for several various treatments (see paragrsphs 78-80 and 83-84). reasons. Particles of the thorium dioxide tend to Such estimates of leukaemia induction per unit dose as aggregate and produce necrosis of the immediately averaged over the whole marrow are not unduly l surrounding tissue, so that much of the alpha radiation is discrepant with those denved from conditions of more or absorbed either in dead cells or in the particles less uniform manow irradiation. It must be recognized, themselves. Moreover, it remains possible, although not however, that these conditions involve an entirely likely, that the chemical properties of the thorium non-uniform exposure of the tissue at risk and that, on dioxide, as well as the radiation emitted by the thorium biological grounds, there is ample evidence that the t and its daughter products, contribute to its carcino. protection of a small volume of marrow from irradiation genicity (see paragraph 283)- may allow " recolonization" of inadiated marrow by uninadiated marrow cells, and that this is of importance

25. De problerns in deriving a risk per unit absorbed to continued manow function, at least following high l-dose for lung tinue from statistics of lung cancer local exposures.

development in uranium miners are sirnilar, since the doses to the bronchi ard alveoli cf the lung depend

30. Dese ditriculties may,however,be more apparent critically upon the depositi,m and location of products than realif carcinogenesis is related to events induced by of radon decay. Although a relatioruhip is established rad;ation in single cel!s cr small groups of cells without

) between the indon concentration in air samples in mines significant reaction between affected and unaffected and the mortality amongst mmers who have worked in cells. To the extent that this is true, we are concerned them, any radsation risk estimates denved for lung with the number of cells involved and with the estimated tissues as a whole are necessarily indirect. dose to each. In this way it may be valid to use 365 ---,mw. ---m.,,.-~.m_m,-. -,,m _ m -,, --,,n.--.-

observations in which, for example, part cf the gut or level applies strictly only at that dose level, and that the part of the t:tal skin area has been irradiated to derive likely frequency per ra> at low dose levels of a few rad estimates for the carcinogenic effect of irradiating the or less, which are of

• ost concem in radiation protec-whole gut or the whole skin, tion, cannot be assumed to be equal to the frequency observed per unit absorbed dose at higher levels.

b

31. It may be noted that a gross non-uniformity of the Similarly, the frequency observed in populations irradiation of cells of an organ will always occur with exposed at different doses cannot be taken as that alpha radiation, since even a uniform concentration of applicable at the mean dose, unless the dose range within an alpha emitting nuclide is likely to involve a high the group is small. Dese considerations apply deposition of energy in a few cells and no deposition in particularly to radiation oflow LET.

the many cells through which no alpha track passes, unless the nuclide is present at very high activities.

35. If the carcinogenic effects of radiatbn are in fact i

attributable only to the sum of one-and twoevent i

32. If data are being analysed on the basis of an processes, with a decrease of effectiveness at very high assumed linearity between dose and frequency of effect, doses due to cell death and dependent on similar the arithmetic mean dose throughout the population of contributions, the relationship of frequency of effect E q

cells at risk will in any case form the appropriate index to absorbed dose D has been described by with which the tumour incidence should be compared. If, however, the range of doses to which cells are E = /aD+ bD 1 c-48"8D l 2 exposed exceeds that over which a linearity of the At very low doses, the effect per unit dose E/D tends dose effect relationship can be assumed to apply, then towards a. For moderate doses, this ratio would be given t the mean de e over the tissue as a whole can at best be approximately by l only an approximation to the relevant parameter (see Annex I, chapter 111. section B). E/D = a [I + (b/a) DJ [1 - (c + dDI D] He value of (c + dD)D is likely to be small for all values

33. For many forms of tumour induction in animals, of D which are substantially below that at which the l

there is evidence (Annex I, paras.140154) that, for maximum induction of tumours occurs. On this basis the uniform irradiation of various tissues, the number of yield per unit absorbed dose at low doses is likely to be tumours induced increases to a maximum (often at absorbed doses of the order of a few hundred rad or overestimated by that at rather higher doses by the factor I + (b/a)D). moce) and then decreases, presumably as the cell-killing effect of the radiation predominates over the tumour.

36. Data on the pnetic effects oflow LET radiation i

inducing effect. (The maximum yield per unit absorbed in the mouse and on the induction of chromosome i' dose occurs at a somewhat lower dose.) It follows from aberrations in several mammahan species including man this, firstly, that for highly non-uniform radiation, which have been analysed in this way suggest values of estimates based on mean tissue dose may overestimate b/a in the range 0.01 0.03 (20), and it has been the risk of non-uniform irradiation and, secondly, that suggested' that similar values may apply for carcino-r estimates based on uniform high doses may under-genesis. If this is so, it would indicate that estimates of estimate the risk per rad of uniform low doses. The carcinogenic effects per rad derived at doses of 100 rad paucity or absence of thyroid cancers following the of low.LET radiation could only overestimate the i radioiodine therapy of hyperthyroidism, for example,is frequency of effects per rad at low dose by a factor of thus not necessarily in conflict with the relatively between 2 and 4. frequent induction of thyroid cancers by moderate or low doses. The absence of any consistent increase in I leukaemia following the radium treatment of cancer of E. SUITABILI'IY OF COMPARISON l the cervix uteri has similarly been attributed to the POPULATIONS l effect of cell killing in the parts of the bone marrow which are most heavily irradiated (see paragraph 82).

37. The excess incidence of malignant disease in an irradiated population can be reliably estimated only by

34. It follows also that,at very high doses (e.g.,several comparison with a control population which is similar in thousand rad) the number of tumours induced per unit all respects except that it has not been equally

[ dose may be lower than that at lower doses (e.g. a few irradiated, and this condition is seldom fully achievable. F hundred rad). Moreowr, t' ere is now abundant i evidence in animals, and some evidence in man, that the

38. In the important studies on late effects of

+ dose effect relationship at these lower doses is radiation in patients after radiotherapy the problem is f non-linear, at least for certain tumours (Annex 1, particularly difficult, since it is usually hard to establish [.. paras.30 and i12). Indeed, it has been suggested on and follow up a control group of patients,with the same ( theoretical and microdosimetric grounds, that the disease and of the same severity, but untreated by [ tumour-inducing effect of radiation is likely to be radiotherapy, since the selection of patients for such y represented substantially by the sum of a knear term in treatrrent usually ccrrelates with the stage or severity of e i dose corresponding to the consequences of single events the disease. Yet, it is difficult to be sure that some form j due to ionization tracks passing through sensitive cell of mahynant diseaw may not be slightly increased in e structures, and of a quadratic term in (dose): incidence in patients with the disease being treated.The corresponding to damage due to two events (125.157). radiation exposure itself rnay be unlikely to produce It must be emphasized therefore that the frequency of more than a small increase of this sort, so that spurious ) tumours induced per unit absorbed dose at a given dose results may be caused if comparisons are made with the g + =

Cl p3 9 ' ncidence cf mahanant disease in a sample of the pneral with radioactive iodine. The incidence cfleukaemia was i I~ population, matched only for age and sex. If the excess fcund to be slightly higher in the group cf patients so incidence in the irradiated, compared with the pneral, trented than in members of the general population of the population conelates with the radiation dose received, same age and sex. Fortunately, however, a control series - {' ( and if this dose does not correlate with the severity (or had been studied of patients with the same condition type) of disease treated, the comparison may be valid. It treated by surpry in which a similar slight excess of may also be valid when the disease treated is one such as leukaemia was found. It would how twen hard to . ringworm of the scalp, which appears wry unlikely to be predict, and perhaps even to suppose, that leukaemia associated with increased mahanancy, at least of tissues would be increased in fre<pency' in hyperthyroid . cther than the skin, t>ut here controls by unitradiated patients, and particularly in hyperthyroid patients who e patients with ringworm have in fact been possible. had been cured of their disease by adequate trestrnent, and the excess might have been used as a basis for

39. In the treated ankylosing spondylitics,it has been radiation risk estimation if the appropriate control series pointed out that the excess incidence of cancer of the had not been studied.

colon cannot be regarded as evidence of radiation induction of this form of malignancy,since ankylosing

41. In the very important epidemiological studies of spondylitis is known to be associated with ulcerative the inddence of cancer in A. bomb survivors in l

. colitis and the incidence of cancer of the colon is Hirosiuma and Nagasaki, the establishment of an increased in patients with ulcerative colitis. In addition, app:opriate control group has always presented diffi. little was originally known of any association between culties. In mortality studies, the pneral mortality rates either the disease itself, or the drugs used in control of at appropriate age and sex in tha Japanese population as its symptoms, and the incidence of other malignan'cies, a whole have been used, but are subject to criticism in except that in an initial survey (37) one case of several respects. Firstly, living conditions in the two leukaemia occurred in a thousand patients with cities were very otmously difficult for a long period ankylosing spondylitis who were not treated by after the explosions, and may have involwd factors radiotherapy. Further data have, however, now been which might inGuence mortality rates even for mahanant obtained (120) showing no increased mortality from disease. Secondly, the populations exposed were i l leukaemia, cancer of the col,m or other malignancies as depleted of men of military age, and particularly those l compared with that in the pneral population,in 1021 of this age and in good health;and,while the difference l patients with ankylosing spondylitis followed during a in sex distribution could be adjusted, the adjustment i mean period of 8.5 years, and who had not been given would not necessanly be accurate if based on statistics

x. ray therapy. Similarly, the raised incidence of cancer for the whole of Japan, even if these were confined to l

of the breast in patients irradiated for acute post partum the years at issue, since the distribution may well have mastitis could only be attributed reliably to the been different in urban and rural areas. Thirdly, and j radiation if it was known that there is no increase in particularly in regard to carcinoma of the bronchus,it is sucl. cancers in untreated patients with this form of known that the incidence of this condition differs in

• mastitis.. It is particularly valuable therefore that a urban and rural communities so that comparisons survey has now been made (138) on the breast canar between the survivors in the two cities and those for incidence in such untreated patients (see paragraph 176).

Japan as a whole are not necessarily valid. Moreover,it is The increased incidence of thyroid cancer in children known that the mortality from tumours of the lung in Japan rose considerably between 1967 and 1972 (from following neck irradiation for enlarpment of the 12.910-s to 17.010 per year). The use of the 1967 i thymus was similarly regarded as possibly being due to some effect of thymic malfunction, associated both with values (97) as controls for estimating more recent the enlargement that was being treated, and with a radiation induction of lung tumours in Hiroshima and failure of the immune reactions to a developing cancer, Nagasaki may therefore overestimate this induction. *Ihe so that cancer incidence might be increased in such use of the mortality in those exposed at 0 9 rad may patients. It was only when it was found that the therefore be preferable. In various estimates made in this increased thyroid cancer incidence was associated with report, both bases of comparison are shown, even neck irradiation for a variety of other conditions in though the small numbers of death occurnns in the infancy that it could be asserted that the radiation was 0 9 rad group may considerably reduce the accuracy of hkely to be responsible for the excess.In the same way, estimates based upon comparison with this group (see leukaemia is known to occur in patients with paragraph 45). 38 polycythaemia vera whether treated with P or not.

42. In addition, the immediate mortality at the time Control series of untreated polycythaemic patients sre of the explosions must necessarily have been selective to only valid if the duration of the disease when untreated some extent, involving those in different states of health I

with 38P is as long as when it is so treated. In pneral, to different degrecs, and this in itself may have the 8'P treatment is effective in prolonging life and innuenced the mortality rates for different conditions may, therefore, be apparently associated with a higher amongst the survivors. Moreover, those surviving in the I incidence of leukaemia purely because the patients so high exposure groups are likely not caly to have been treated live long enough to aDow the development of subjected to a more powerful selection of this type at I what may be a natural sequel of prolonyd polycythae. the time of the explosions. but in addition may well (. mg,' have had other injuries, ailments or difficulties in living ,l

40. A further example of the hazards of using the conditions which may have correlated stror. gly with the li pneral population as a comparison group for patients magnitude of the exposure that they received, or with t[

with a particular disease treated by radiation is given by the position in the city at which these exposures were follow-up studies of hyperthyroid patients (128) treated received. L 367 s h

' F. FURTHER FACTORS AFFECTING THE PRECISION OF RISK ESTIMATES remembered that pr:babilities cannot necessarily be derived from values of the standard error on the I

43. Even when populations irradiated at known dose are followed over sufficiently long periods of time,with full ascertainment of resulting malignancies and with an

47. In quoting the amount by which an observed adequate control group for comparison, several other number of cases (e.g. of deaths from a disease) exceeds factors will affect the precision of the resulting estimate.

the expected number, the size of the exass may be 4 The size of the exposed population is of great negative, or one of the confidence limits with which it is importance for detecting an excess of most types of estimated may be so. In such cases the excess is noted tumour following irradiation. Since the acidence of with its negative value, or is quoted simply as being i many types of tumours following radiation appears to be negative (with the abbreviation "neg."). Thus for in the range of 5 20 per million per rad, radiation. example, for an observed number cf 3 that has 90% induced tumours will usually not occur in more than, Poisson confidence limits of 0.8 and 7.8 and an expected say I per cent of those exposed. Such an excess may be number of 2.2, the exass would be expressed as 0.8 readily detectable in an exposed population of a few (-1.4 5.6)or 0.8(neg. 5.6), and rates derived from such hundreds if the type of tumourinduced occurs rarely m an excess are quoted similarly. Since so many estimates the absence of radiation. If, however, a small frequency in this Annex are necessarily based on small excesses of of induced tumours has to be distinguished statistically numbers observed over those expected,it is important to i from a large number of similar tumours occurring indicate in this way the accuracy of the stated e'xcess or naturally, populations of many thousands may need t excess rate on statistical grounds and the confidence be studied. with which it can be asserted even to be positive. i 1

44. The statistical uncertainties involved in estimates based on small numbers of tumours are increased if the i

incidence in the comparison population itself involves G. WEIGHTING FACTORS FOR NEUTRONS i . small numbers of tumours also.

48. For neutrons and for alpha radiation, estimates of

45. For example, it has been emphasized (pars.41) risk per unit absorbed dose cannot be used in any direct that the incidence of malignant disease in the irradiated way estimating n,d pu urut abewbed dose (m n j

survivors at Hiroshima and Nagasaki cannot necessarily radiamns of W W fw two wasons. Fintly, j be estimated accurately by comparison with the general radiations of high ET are known to be more harmful mortality rates for Japan as a whole. The sarne applies than those oflow MT pn unit absabed dose, d h with equal emphasis for morbidity rates, particularly RBE, or ratio of absorbed doses causing a given since the populations of Hiroshima and Nagasaki are frequency of effect, may vary for different types of likely to have been under much closer rnedical effect. Moreover, even for a given type of effect and surveillance for the particular purposes of the health quality of radiation, the RBE is likely to vary with the i 'j study than other populations. For these dose levels at which the companson is made or with the reasons therefore, the incidence of malignant disease m the frequency of effects (76,125). exposed groups has often been compared with the incidence, either in residents of the two cities who were

49. There are considerable difficulties therefore in not in the cities at the time of the explosion, or in expressing the carcinogenic risks observed in A bomb groups exposed only to radiation at doses of 0 9 rad.

survivors in any form which is applicabie to radiation in When such comparisons are made however, the incidence general. In Hiroshima in particular, a substantial of diseases in the control groups may be subject to PmPortion (23 35 per cent in the different dosage substantial statistical uncertainty in view of the small groups) of the estimated kerma (tissue Kerma irt free air, numbers of tumours occurring in these groups. The T65 estimates)is attributable to neutrons. t est? mate of risk based on these comparison groups may, therefore, be more valid than when the comparison is

50. In Nar===ki this percentage is much lower, less 4

with the Japanese health statistics as a whole, but may than 2 per cent in all dose groups. In principle it would be so imprecise that this advantage is lost. be possible to obtain approximate risk estimates for L i

46. Throughout the following sections, attempts have low-LET radiation for various forms of malignancy from the mortality data in Nagasaki. In fact, however, many been made to indicate the statistical reliability of the of the estirnates from this city are of low statistical

[ estimates and rates quoted, usually by stating the 90% reliability, those from Hiroshima being determined with 3 confidence limits involved, as determined by the Poisson considerably greatst accuracy. distributions depending upon the number of cases observed. Thus, whenever an estimate is followed by a

51. For leukaemia, however, relatively accurate risk statement of two altematin values within parentheses, for example "Si (26 88) 10d" it is implied that the estimates can be made for several dose groups in each expectation 5110 has 90% confidence limits of city. In this case, therefore,it is possible to compare the i

2610'* and 8810'*. These values give only a minimum relative effectiveness of neutrons and gamma radiation in i estimate of the confider.ce tones if other sources of these dose groups in inducing ieuksemia,ifit is assumed l O error affect the estimate. Where several such sources of that a given absorbed dose of high LET radiation was i V random error can be evaluated and where standard errors equally effective m each city, and that the same applies cf a mean value are quoted in consequence,it must be to equal absorbed dos-s of low.LET radiation. For this j, comparison two steps are required. Firstly, an estimate 368 I T 1

1 lapping) groups cf d those exposed at ever 100 rad,. must be made cf the mean absorbed' doses cf high, and' ' over 50 rad, or over 10 rad. On this basis, the values are

low.LET radiation in bone marrow resulting from the

~ as follows: ' asutron and gamma radiation in each city and for each group examuwd. For bone marrow, values for the ratio Kerma group (rad) >200 >100 >50 >10

i W mean absorbed dose /mean kerma are under investigation Weighting factor W,

' (V(the ratio for gamma radiation islikely to be abaut 0.55.

' 71). Prelimmary estimates however (70) indicate that by comparison with: 09 rad group 7.6 11.4 14.4 19.3 For neutrons, the corresponding ratio is estimated as SE 14.2 15.2 16.6 19.4 0.26 for the absorbed does of high.LET radi tion (from Japanese National protons induced by neutron capture or by recoil) with Statistics. 7.3 10.6 13.0 15.8 en additional 0.07 by induced gamma radiation.s SE i4.1 24.7 i5.5 16.6

52. As the second step, a biological weighting factor W

56. These values clearly suggest an increase in j

for the effectiveness of high-relative to low LET effectiveness of high. relative to low.LET radiation in j ' radiation in inducing leukaemia can be derived by leukaemia induction as progressively lower dose levels equating the induction rates-in excess - cases per are included (also see Annex I, paragraph 176). As person. year per unit absorbed dose-in the two cities, stated, however, no adequate estimate can be made of giving the component of absorbed dose due to high.LET the value of W at lower doses, owing to the uncertainty l radiation in each city the same weighting factor. in risk estimates in these dose poups that results from the small excess of observed over expected cases. Thus,

53. Thus,if the mean absorbed doses in marrow from for example, the estimate of W for those exposed at garuua radiation and from neutron. induced protons in a 100-199 rad kerma (using comparison with the 0-9 rad given dose (kerma) poup are D(H.7) and D(H.p) for group)is 47,with the standard error about 67.

Hiroshima, and D(N.7) and D(N.p) for Negn=ld, the i value of k' for this dose group may be derived from the

57.. These estimates are based on the mortality from equation leukaemia in the two cities in rnembers of the Life Span Af (H.obst - M iH.exp, Study, for whom the kerma has been estimated.

= PtH) [D(H.yl + WD(H.p)] Substantially larger numbers of deaths from this disease i have been recorded in the leukaemia Registry, but M (N.obst - M (N.exp) values of tlw kerma are not known for many of these l P(N) [D(N 7)+ WD(N p)] individuals. If approximate kerma estimates are made for where M(...) represents the number of deaths from this larger population, however, on the basis of the leukaemia observed -(obs) or expected (exp) in distribution of the whole A. bomb exposed population in. + 1950. the excess mortality from leukaemia in this group O Hiroshima (H) or Nagasaki (N), and P(H) and P(N) are is more nearly proportional to estimated kerms in both the populations exposed in the relevant dose groups in cities (13). On this basis the variation of W with kerma ~ Hiroshima and NagasakL might therefore be considerably reduced.

54. For example, in those exposed at over 200 rad

58. The values quoted in ps.rapaph 55 can however be (kerma) the mean absorbed doses in marrow oflow.and used to obtain tentative risk estimates for the induction llI high-LET radiation can be estimated as 155 and of leukaemia by low.LET radiation over the stated dose 24.4 rad, respectively, in Hiroshima and 181 and 1.4 rad ranges in the two cities.* lhus,in table 4.in addition to in Nagasaki. The number of deaths from leukaemia in estimates of excess mortality rate per rad kerma, rates f

this dose group in Hiroshima was 28 as compared with are quoted also per unit of weighted absorbed dose,in .1.3 expected on the basis of the 09 rad group, the which the high.LET component of the tissue dose is corresponding values being 15 and 1.2 in Nagasaki. The weighted by the value appropriate to the done group . number of persons was 1301 in Hiroshima (the average considered. l number of survivors during the period of survey) and 't 1191 in Nagasaki. For this kerma poup therefore

59. For malignancies other than leukaemia, the risk 28-I3 IS-I2 estimates.in particular for Nagasaki, are too uncertain to ii ll 1301 (155 + 24.4 W) 1191 (181 + 1.4 W) allow any adequate values to be derived for separate il-weighting factors appropriate to each malignancy. In M.

'whence W = 7.6. previous reports, risk rates have been quoted per rad jj kerms without an estimated allewance for the neutron

j

$5. Estimates of W can be derived in the same way for component of the exposure. The reports of the joint l. lower dose groups considered separately, but are very .g,,,m,,h shou'd are be iriade at W 1* imprecise, with the standard error (SE) of sampling weight:d itsorbed done, rather than at equal kerma. when ruch f approaching or exceeding the expected value itself. The compansons are made, estimates of w ue somewhat lower in the values of such a weighting factor can, however, be hash dow rarse with values of 5.4,7.3 and 14.2 as weighted 3' estimated with reasonable confidence for the (over. absorbed doses of 190.130 and 90 tad respectively. The necessity for averagus doses and leukaemia rates over wide dose

• A recent study by Hashizume es at t$)a) est. mates that m:ervals m this analyss aiso introduces errors, to the extent that for bone marrow the ratio of mean absorbed done to mean the low IIT done eflect relationship departs from e atmple hnear
• w kerirs would he about 0 65. For neutrons, the ratios wovM te proportionahty between dose and effect. These errors prove not 0.25 for proton induction. and 0.10 for samma radiation to be large however, when calculated on the basis of hkcly tr. duction by neutrons The use of these ratios would lower the estimates for the factors, discussed in paragraph 35. which may l,

values of the weighting factors given in paragraph 55 by about determine hnear and quadratic components in such relationships 10 per tent. 369

,11. - p' Atomic Bomb Casualty Ccanusson'(ABCC)-Japanese National lastitute cf Health (JNIH) surwys also give - f:r neutrons (n) as inducing proton (p) or gamma (7) ' i [ dose ranges in rad (kIrma) but refer in some instances to ' radiation in tissues. Mean values for vari:us modes of "RBE. dose estimates in rem" (e.g., reference 67, irradiation are as follows: table 5), using the RBE of 5 which 1shimaru et al. (63) o,)g, o,fg, o,fg, note as being applicable for neutrons in leuksemo-f- as 80 0.5 0.M5 genesia.

60. In the absence of other estimates - for the These ratios haw been used in deriving risk rates for the effectiveness of neutrons in inducing malignancies in weighted absorbed doses.

man, and since the estimates ofleukaemia mortality now yield weighting factors for different dose groups as

62. ~ For evaluating the risks of alpha radiation, either shown in paragraph 55, these weighting factors haw in bone sarcoma induction by radium or lung carcinoma been used in later chapters to_ derive risk estimates for by daughter products of radon, no comparable biological other mahanancies,in addition to those quoted per rad basis is available for estimating any RBE, or weighting i

(kerma). The use of this procedure is not intended to factor,that should be applied to the high MT radiation imply that the RBE for neutrons in inducing malignant of tissues in these cases. Risk rates are quoted, therefore, ! change is necessarily the same at a giwn dose for all per rad of alpha radiation and require to be interpreted types of mahanancy. It is cf interest however that the in the light of other evidence as to the likely RBE for induction rates-in the two cities for a rar.ge of alphas at the dose levels concerned. mahanancies for which approximate risk estimates can be made, do appear to be statistically consistent with the intention in using these factors for other malignancies is H. LEUKAEMIA use of the same weighting factors for all types. The thus solely in order to derive from the important observations in Hiroshima, risk estimates that may be more applicable to low ET radiation than those derived A. GUKAEMIA IN A BOMB SURVIVORS directly from kerma, or absorbed dose, having large 63. components of high ET radiation. For this purpose it The seventh report on' mortality data in the appears preferable to use factors which reflect a dose JNIH ABCC Life Span Study (97)includedinformation l ' dependent RBE and which are based on other human on deaths in exposed populations to the end e f 1972, carcinogenic data observed within the same dose groups, and indicated that the annual frequency of deaths from j than to apply a constant value for RBE derived from leukaemia had retumed to about the level observed in quite different sources, the unexposed or only lightly exposed comparison populations (table 2 and fig. III). During the three-year period 19701972, only 3 deaths occurred in the groups 61. In the case of the breast and the thyroid, estimates exposed at have been published (9 53) for the ratios of absorbed known doses greater _ than 10 rad, as done D to kerma K. both for gamma radiation (7) and compared with 2.7 expected by comparison with the 0.9 rad group, and with 2.1 expect. J on the basis of 1 TABI.E 2. EXCESS MORTALITY FROM LEUKAEMIA IN HEROSHIMA AND N 2 Dose poups 10 >200 rad T65 Males and females, all ages Mens red Men I 06 Ex-year Exerar mee Ok En. i md' Prried served preted Excese (10*) (10'*y" nd") Prried arrued preted Exces (10*) (10 y" md *l yeer Esenerute s HIROSHIMA NAGASAKI Compared Mth 09 redgroup Compared Mth 09 redgroup 19541954 16 2.9 13.1 (6.9 21.6) 5.0 2.6 (1.4-4.4) 1950 1954 8 1.5 6.5(2.2 13.2) 3.7 2.0(07-4.0) 1955-1959 22 5.5 16.5 (9.0 26.3) 5.6 3.0(1.64.7) I955 1959 5 1.1 3.9 (0.7-9.6) 3.7 1.1 (0.2-2.6) i 196&l964 12 1.5 10.5(5.3 18.0) 5.2 2.0(1.03.4) 1960 1964 4 0.7 3.3 (0.5-8.7) 3.5 0.9 (0. I-2.5) 1965-1969 11 2.7 8.3 (3.2-15.8) 4.9 I.7(0.73.2) 1965-l % 9 3 2.6 0.4 ines.-6.0) 3.4 0.1 (nes..I.8) 1970 1972_ 3 1.5 1.5 (nes.-6.8) 2.7 0.5 (nas. 2.5) 1970 1972 0 1.2 -1.2 tnes. 3.2) 1.9 -0.6 (nes.-l.7) 19541972 64 14 2 49.B (36.7 65.1) 23.4 2.1(1.72.7) 1950 1972 20 7.1 12.9 (5.4-22.7) 16.2 1.3(0.72.1) [ Compared Mth JapaneseNationeelSnetutks Compared with Japanese NetionalStatistics { 1950 1954 1.1 14.9 (8.9-23.2) 3.0 (1.8-4.7) 1950 1954 0.6 7.4(3.4 13.8s 2.311.04.2) .1955 1959 2.0 20.0412.7-29.2) 36(2.3-5.3) 1955 1959 1960 1964 2.3 9.7(4.6 17.1) 1.9(0.93.3) 19641964 h, 0.7, 4.3 (1.3 9.8) 1.2(0.42.6) 1 % 5 1969 2.6 8.4 (3.6 15.6) 1.7(0.73.2) 1965 1969 1.1 1.9 (nes. 6.7) 0.6 (nes. 2.0) t 1.1 2.9 10.3-8.1) aboes 0.8 (0.4 2.3) l 1970 1972 1.5 1.5 (nes.-6.3) 0.5 (nes.-2.3) 1970 1972 0.6 -0 6 (nes. 2.4) .-0.3 tnes 1.2) 1950 1972 9.6 54.4(41.8-69.2) 2.0 (1.5-2.5) 1950 1972 4.1 15.9(9.1-25.0) 1.0(0.31.9) Source: Reference 97. Note: The 90T, conn 4ence heits are indscated in perentheses. 370 p l e

p! .n

~

Japanese N tional Statistics.While these differences are last 3 ye:rs of thi 22.25 years cf the survey appears to [ not significant, it b n;ted that a small significant excess have been only about 1.5 per cent of the total excess p was still present in the two subgroups that had leukaemia mortality. previously shown the highest frequencies. Thus, in g Hiroshima in the groups exposed at 100200 rad and at 64 As shown in table 2, the annual death rate per man g er 200 rad,2 and I deaths occurred during this latest rad rose sorr.ewhat from the first (19501954) to the y period as compared with 0.2 and 0.1 expected on the second (1955-1959) five. year period of survey,then fell { i basis of the 0.9 rad group or of Japanese National progressively. Ignoring any cases that may have arisen Statistics. With this expectation,3 or more deaths would prior to October 1950, the mean interval from radiation f occur by chance with a prcbability of only 0.0035, exposure until death from leukaemia has been 13.7 years although the testing of these two subgroups selectively (with median 12.5 years), as judged from the annual i because they showed higher incidences must somewhat incidence rates. If this mean interval is determined for I increase the probability of this difference being excess deaths of those exposed in each dose group l fortuitous. While, therefore, some mortality may have (table 3), no relationship between interval to death and l been persisting, the contribution to deaths during the size of dose is evident. TABLE 3. MEAN INTERVAL FROM EXPOSURE TO DEATH FROM LEUKAEMIA 19501972 IN i HIROSHIMA AND NAGASAKI i Males and fererles, an ages (Years) Dose group (T63 kerma) Populerton 09 10 49 30 99 200 199 >200 All Haroshima Excess deaths 12.0 10.8 5.5 ILO 27.2 66.5 Mean interval 10.1 15.8 12.5 15.0 14.1 13.7 SE8 s 0.9 a 1.2 23.6 22.0 t 1.1 0.6 Negasaki Exceu deaths 4.7 0.0 (nes.) 2.3 14.3 20.6 Mean interval 19.1 10.2 13.8 13.8 SEs a 4.5 1.8 21.6

• 2.1 Hiroshima andNagannki Mean mterval

. 12.6 15.8 (12.5) 14.1 14.0 13.7 SE' e l.4 21.2 (s 3.6) 21.8 20.9 20.6 Source: Rererence 97. (,) dSE = standard error. The values of standard error are based on the number of deaths from leukee. (\\j maa in the dirferent groups.

65. Ichimaru et al (61) show the number of cases of

66. As judged by recorded month of onset there acute and chronic leukaemia with onset in each year appear to be no significant differences in latency from 1947 to 1971 including all confirmed cases in the between different types of acute leukaemia, although ABCC leukaemia Registry. For those who had received that for acute moneytic is slightly shorter than the a kerma of 100 rad or over, the mean interval between average for other acute types, as seen from these data exposure and onset (mean latency) of leukaemia was as (years):

l follows (in years): Type of acute Number of Mean leuksenua cases latency SD K l Typr of AB Excess Approximate t leukaemia cases cases

• SDb SE median latency Granulocytic 19 12.4 4.9 1.1 Stem cell 5

13.2 6.1 2.7 l Acute 11.8 11.5 5.9 0.9 11 1.ymphatic 15 10.8 6.9 1.8 i Chronic 11.1 11.0 6.4 1.7 9 Mon cytic 7 9.2 3.8 1,4 All 11.6 11.4 6.0 0.8 10 All 11.5 5.9 0.9 8As compared with onsets in unexposed grovos.

67. Clear evidence is now obtained, however, of a 6SD = standard deviation.

variation in latency with age at exposure, at least for Since these values are based on the onset of individual acute leukaemias and possibly also for the chronic cases, and since the numbers of persons in the fixed (granulocytic) type induced. From the recorded month sample was decreasing slowly with time, the mean and year of onset ofleukaemia between October 1950 intervals based upon incidence rares would be slightly and December 1971 in those who had received a kerma longer. Excess numbers of cases in die groep exposed at of 100 rad or more in either city, the following mean 199 rad are too small for analysis oflatracy. latencies are determined: Acute leuksemnas Chronicleuksenua Agent Number Meen Number Mean exposure of latency of latency (years) utses (years) SD SE cases (years) SD SE / ("/ <l5 11 9.4 33 1.0 5 93 0.8 0.4 15 29 15 11.0 53 1.4 4 11 3 9.1 4.6 30 44 10 14.7 6.4 2.0 4 12.4 9.1 4.6 >45 7 15.2 4.2 1.6 1 12.6 371

~ ~ ~ i I J d O u TABLE 4. EXCESS MORTALI1Y FROM LEUKAEMIA IN HIROSHIMA AND NAGASAKI,19501972 Males and females, all ases - i Excess use Excese rete per unit Excear rest per unit wekhtat Excese rete per unit Doseswup herme absorbat done Dosegroup per unit watchend (red kerme) Obs#rn d Expected Excese (20** red'*) (20'* red'*) (red herme) Oberrwd Expected Excear (20** red'*) (20** red **) herme absorbeddoor HIROSHIMA Compared with 69redgroup NAGASAKI Compared with 69 redgroup. 1449 17 9.2 7.8(1.6-16.8) 37(5-80) 10-49 2 3.5 -1.5 (nes. 3.3) -22 (negA 7) 1 SG99 7 2.2 4.8 (1.I-11.0) 29(747) 5499 0 1.2 -1.2 (nes.-l.9) -15 feeg.-24) 100-199 12 1.5 10.5 (5.4-17.9) SI (26 88) >200 28 1.3 26.7(18.6-37.1) 57 (39-79) 61(41 44) >200 15 1.2 13.8(8.424.4) 35(2452) 61 (35-91) 100 199 3 1.2 1.8 (nes.-6.7) 1I(nes A I) >100 40 2.8 37.2 (27.2-49.5) 55(40 73) 47(3443) >100 18 2.4 15.6 (9.2-23.5) 28 (16-44) 48 (28-76) i > 50 47 5.0 42.0 (31.3-54.9) 50(3745) 38 (28-50) > 50 18 3.6 14.4 (8.0 23.1) 22 (12-36) 38(2041) l > 10 64 14.2 49.8(36.745.1) 47(3541) 31(22-39) > 10 20 7.1 12.9 (6.1-22.0) 18 (6-33) 30 (10 55) Compared with Japenese NaionelStatistics Compared with Japanese NationalStatistics j 1449 17 52 10.8 (4.6-19.3) 52(22 92) 1449 2 2.0 0.0 (neg.-4.3) 0 (nes.-62) 5499 7 1.5 5.5 (I.8-II.7) 33 (11-71) 5499 0 0.7 -0.7 (nes.-2.3) -9 (nes.-29) 104199 12 1.0 11.0(5.9-18.4) 54(29-901 100 199 3 0.7- .2.3 (0.1-7.1) 14(1-43) >200 28 0.9 27.I(19.0-37.5) 57(40-79) 62(43-86) >200 15 0.7 14.3(8.5-22.4) 36 (21-56) 63(37-98) >100 40 1.9 38.1 (28.3-30.2) 56(52-74) 50(3547) >100 18 1.4 16.6 (10.2-25.3) 29 (18-45) 50 (31-78) { > 50 47 3.4 43.6(32.9-56.5) 52(3947) 42(32 54) > 50 18 2.1 15.9 (9.5-24.6) 25 (15-38) 42(2544) > 10 64 9.6 54.4(41.849.2) 52(4046) 38 (29-48) > 10 20 4.1 15.9 (9.1 25.0) 22 (I3-35) 37(22-59) i l Source: Referencs 91. Nose: The 90% confidence limits are indicated in parentheses, i l W $ M T4 9*9 F4 W 'o^# h .. _..,.y i

I i

68. 1hs effect cf age at exposure on the frequency.

However, the numbers involwd are small and the age with which leukaemia has beem induced in Hiroshima structure cf the male and female populations differed and Nagasaki is now clearly shown (14) by a study of appreciably, with a relative deficiency of males aged j the number of excess deaths per onmaa person. year. rad 20 35 and a slight plative preponderance of males at ages f 'in diffemnt age groups. As estimated by the number of over 35. j deaths from leukaemia from October 1950 to September i 1974 in the Adult Health Study Populations exposed at

70. 'the natural annual incideem rate of leukaemia in i

Himahuna and Nagasaki, the induction rate is highest in Japan, as judged by cancer registry data (35) is the groups aged 010 years and over 50 years at somewhat lower in females than in males, with values of t

exposure, with excess mortality rates of 2.7 and 3.710-8 y ' in females for age standardized f

3.2 (2.8 3.6)10-* and 3.4 (2.8 4.1) 10-' y-8 rad-' populations in two Japanese registries, as compared with respectively. 'the rate is lowest in those who were aged 4.4 and 4.310-s y ' in males. it is ratherlower than in 10 to 20, with a value of 1.0 (0.71.2) 10-* y-8 red-8 registries in many other countries (which have mean r 1 and intermediate for age groups 20 35 and 35 507 ears rates for 58 other registnes of 4.4 and 6.810-s y'n at exposure,with values of about 1.9 (1.5 2.2) 10, y ' females and males).1he incidence rates in members of red-' in each group.These differences in mortality rate, the Master Sample exposed at less than 5 rad at as estinisted in the period up to 29 years from exposure, Hiroshima and Nagasaki were 3.1 and 5.710-s y-a, i are not clearly attributable to differences in latency v.th respectively. The incidence rate of chronic lyr..;'istic age, since excess deaths have been occurring in all age leukaemia in this group was only 0.1 ITs y ' G.7 per groups in a comparable way since the beginning of the cent of all cases) as compared with a mean of 1.210-5 period narveyed (i.e., from 5 years from exposure). y ' (22 per cent of all cases) in 22 registries in other t

69. The mortality from leukaemia in Hiroshima and countries reporting leukaemia by type (35).

. Nagasakiis not reported separately for the two sexes in current JNIH.ABCC Life Span Studies but the appendix

71. During the total period 19501972, the number of of the latest review of the ABCC 1.auksemia Registry deaths from leuksemin in the Life Span Study Group in data (61) records the sex of patients developing any Hirosiuma (97), at known kerma of over 10 rad 4

form of this disease. Of 57 cases with onset from 1950 exceeded the number expected on the basis of the 0-9 1 to 1971 in people estimated to have raaived a kerma of rad group by about 50 (with 90% confidence limits 37 100 rad or more, 32 occurred in males and 25 in and 65). The corresponding excess for Nagasaki was 13 females. From Report 7 of the Life Span Study (97) the (6-22). Expressed as rates per nuthon exposed and per number of person years at risk dunng this period appears unit kerma, these excesses correspond to 47 (37 65) to have been about 47 370 in males and 61390 in go-* rad-8 and 18 (6 33) 10-* rad-'; and table 4 gives females (the Registry study uses total person. years rates for the different (kerma) dose groups compared which are about 7 per cent higher but does not quot* with both the 0 9 rad group and the Japanese National ~ i values separately for the two sexes). On this basis, the Statistics. t s leukaemia incidence is somewhat higher in males than in females. It is of some interest, however, that the gg g meidence of acute leukaemias in this does group appears over 10,50,100 and 200 rad. For these 8roups, table 4 i to be equal in the two sexes, while that of chronic gives estimates also of the induction rate per rad (granulocytic) leukaemia is considerably higher in the (weighted) absorbed dose, using the weighting factors ,,g,. Msier Femeier for neutrons derived above in paragraph 55. These estimates assume the ratios of absorbed dose to kerma TW twe I=sewe for bone marrow of 0.55 and 0.26 for gamma and isuAnemie flo-* y-') Caser (10" y-8) Ces,, neutron radiation, respectively, as discussed in paragraph 3 Acute 4.2 (2.8-6.1) 20 3.7 (2.6-5.3) 23

51. The distribution of mean numbers exposed, and l-Chronic 2.5 (1.5-4.1) 12 0.3(0.11.0) 2 neutron component of kerma, for each dose group is l

All 6.8(4.99.1) 32 4.1(2.85.7) 25 given in table 5. l TABl.E S. EST1 MATED KERMA AND NUMBERS EXPOSED IN LIFE SPAN STUDY DOSE GROUFS Males and females, all ages.19501972 inreeMme Neeennkt Mann M eent Meen Moon N eent Meen Dose erveur kernen f>om number of kerns f>om number of a tred) (mn) neutrons persons (oss) neutnms persons l 10 49 21.9 20 9 533 21.0 00 3 302 i { 50 99 70.2 19 2 350 70.5 0.3 1 120 100 199 138.6 22 1473 145.7 1.0 1127 l >200 363.2 26 1301 334.7 1.7 1191 Seewee: Reference 91. I 373 L

' I s 1

73. 13 the eighth report on mortality data in the Life

75. When the groups which were heavily exposed Cc Span Study (14), the excess mortality rate per unit (at or over 100 rad kerma) in the two cities are compared, ris k.erma from leukaemia in the 24. year period 1950 1974 there is some indication of a significant difference in d'

has been relative frequency of types ofleukaemia(with x = 81, di 8 cy 56 (5161)l0.e rad. in Hirosiu.ms,

V 35(29 41)l0* rad in Nagasaki,and 4 d.f., P = 0.09 comparing specified types and x = 8.3, i

8 ar 46(42 50)l0-* rad-m both cities com. 5 d.f., P = 0.15 including unspecified groups), the main

bined, contributica to the difference being again a relative excess of chronic granulocytic leukaemia in Hiroshima these rates being estimated from the slopes of (variance-weighted) regression lines of mortality rate and of acute granulocytic leukaemia in Nagasaki. De I

upon kerma. They correspond closely with the excess difference is much clearer when the incidence in all those exposed at over I rad is compared in the two cities rates for the period 19501972 as based on comparison 2 between the > 200 rad group and rates in the 0 9 rad (x = 17.4, 4 d.f., P < 0.005 and x = 17.1, 5 d.f., 2 group, as would be expected if P=0.005), the main difference again being due to a relative excess in Hiroshima of the chronic, and in (a) no substantial excess mortality has occurred Nagasaki of the acute granulocytic types. 3 In 1972 74; E (b) the regression line is influenced substan.

76. Since, however, this difference is also evident tially by the large numbers in the 0-9 rad group and the between the control series for the two cities, a difference y

large excess in the > 200 rad group. The additional in the types of leukaemia induced by radiation in these a deaths from leukaemia in 1972 74 have in fact been 6 in cities can only be established by estimating the r those exposed at 10 rad or over in both cities, u frequencies of excess cases of d fferent types in each ( compared with about 3.7 expected on the basis of the city,in each city, the leukaemia rate in those who were c 09 rad group (see paragraph 240). The previous excess unexposed was about 12 per cent of that in those who f r 1950 72 was 63 deaths. i were exposed at or over i rad. Part of the i difference between the types of leukaemia observed in

74. The types of leukaemia that had developed in those exposed at Hiroshima and at Nagasaki may members of the ABCC Master Sample exposed at 5 rad therefore merely reflect the similar differences observed cr over at Hiroshima and Nagasaki differed significantly between the control series in the two cities.

from the types occurring in the groups who were relatively unexposed (at less than 5 rad), according to the report of Ishimaru et al (63), analysing the B. LEUKAEMIA IN POPULATIONS IN THE incidence-of leukaemia from October 1950 to MARSHALL ISLANDS 1RRADIATED FROM September 1966. The types of leukaemia occurring FALLOUT tween October 1950 and December 1971 are now ported by Ichimaru et al. (61). Their data show

77. Leukaemia has developed and has proved fatalin differences in relative frequency of the different types in one inhabitant of the Marshall Islands wno was exposed those exposed to radiation in the two cities, but to radiation from fallout in 1954. The leukaemia was of corresponding differences are also seen between the the acute myelogenous (promyelocytic) type (26). Quite

" control" groups for these cities. Thus, no differences clearly the occurrence of one case might be due to are seen, either in Hiroshima or in Nagasaki between the chance rather than to radiation. In a population relative frequencies in those exposed at less than I rad, estimated to have received about 11870 man rad of and in those who were not in the city at the time of the exposure, however, if the expectation ofleukaemia was, bomb. for example,20 cases per million per rad, the chance of A comparison of relative frequencies in < ' rad 0, I and over 1 cases occurring would be 0.82,0.16 and group and the National Institute of Cancer (NIC) group 0.02. No excess incidence of other forms of malignancy, shows the following: apart from that of the thyroid (see paragraphs 108111) (a) For all specified types ofleukaemia: has been observed. 2 Hiroshima, x = 3.4, 5 degrees of freedom (d.f.), not sip =ificant; Nagasaki, x 8.2, 5 d.f., not significant; C. LEUKAEMIA FOLLOWING PELVIC (b) For specified types and groups of acute and IRRAMAN chronic tmspecified types:

78. Following x-radiation of the pelvisin the treatment 2

Hiroshima, x = 3.8, 7 d.f., not significant; of metropathia haemorrhagica, Smith and Doll (142) 2 Napsaki,x = 9.8,7 d.f.,not significant. observed an excess death rate from leukaemia of 16.3 i C:mparing the comtined control series for liiroshima per 10' wom:n years at risk, during the period of 5 or with that for Nagasaki, however, there appears to be a m re years after treatment.Since the mean raanow dose clear diffe:ence in relative frequencies: was 134 rad, this rate corresponds e 1.22 (a) For all specified types, x = 24.8, 5 d.f., (0.25 2.8)l0

• rad 8 y '. Since the mean period of 2

P < 0.001 I Il W-up from the time of treatment was 19.0 years, ~ and h no excess was esened in the Gnt 5 yean, th (b) For specified and unspecified groups' s = 29.5,7 d.f.,P < 0.001. estirrnted leukaemia nsk to 19 years m women becomes 17(3 36)l0 rad-' at this dose level. 4 difference is due mainly to a relative excess of acute mphatic and chronic granulocytic types in Hiroshima

79. The 7 cases of leukaemia (2.33 expected) were and of acute granulocytic and chronic lymphatic types diagnosed at 11.0 (2 2.1 SE) years after exposure and l

in Nagasaki o death occurred at 11.6 (12.0 SE) years after exposure. i l 374

- Correcting for the decressmg number of patient years at

84. In a further survey of the patients who had risk [fws mean intrvals cf 12.4 and 12.9 years to received only one course of adistherapy (36),an excess diagnosis and to death, values which clearly are not of 316.5 = 24.5 (15.9 35.3) deaths from leukaemia were discrepent with those observed in the leukaemia registry observed in 14109 patients followed for a mean period

. /3 and in the mortality data from Hiroshima and Nagasaki. of 9.5 years. This survey cowred about 134 000 V. Bnnkley and Haybittle (18) made a follow.up of person-years at risk,11900 of them being within the 80 period starting 17.5 years after treatment during which 277 women for a mean period of 16.1 years after an no further excess deaths from leukaemia occuned

x. ray induced menopause involving a mean pelvic dm (1 observed, 0.% expected). With an estimated mean cf about 735 rad. No deaths from leukaemia were 4

marrow dose of 321 rad, this excess conosponds to an obsened, but only 0.24 were expected on the basis of annual r sk of 0.5710-* y-8 rad-8 owr the period of normal incidence at the ages concemed, and only a ,,,,y, this rate having been 0.7810-* y-' red-' further 0.83 cases would have been expected if the during the first 6 years and 0.4910-* y-1 rad-8 during induction rate per woman. year and per rad found by the subsequent period, until the time at which excess Smith and DoD had applied. deaths ceased to occur. ne total risk is 5.410-' red-8 l

81. Alderson and Jackson (2) followed up 2049 as estimated from the whole group, which was studied patients of whom 15 per cent were treated with radium for a mean period of 9.5 years. This value, however,is an and 85 per cent with x rays in the therapy of undemstimate of the true expectation because de.

menorrhagia. No deaths from leukaemia were obeved cmasing numbers were at risk during the penods when during the first 5 years but 4 occurred subsequently as leukaernia incidence may have been maximal, since compared with 1.09 expected. Risk estimates are many patients were withdrawn from survey when they difficult to obtain owing to the non. uniform pelvic mceived a second course of treatment. A more valid risk arradiation from radium. Dickson (32) observed no estirmte may be derived, however, by summmg the l significant increase in leukaemia in 4010 women treated excess mortality rates, rather than the number of deaths, with radium for bemen uterine haemorrhage, but less at =e=*ive time intervals after exposure, as if a cohort than 50 per cent of the patients were adequately traced of patients were being followed without withdrawals and this negative finding is therefore of doubtful from survey except as a result of death. When this is significance, done the total risk until 20 years, when exass leukaemia deaths had ceased to occur, is 11.4 (7.5 16.4)l0-*

82. A manber of studies, which are reviewed by Smith rad - 8.* The mean interval from arradiation to death -

and Doll W2) show no significant increase in leukaemia from leukaemia was 6.610.9 (SE) years, or 7.4 years as i following pQic irradiation for carcinoma of the cenix, based on excels leukaemia rate per person. year at the in many cases by radium. It was, however, suggested by different time intervals after exposure. These intervals, Hutchinson (60) that the negative findings in these and those noted in paragraph 83, are thus clearly shorter i studies are due to the much higher -doses given in than those observed in Hiroshima and Nagasaki treatment cf cervical cancer than those used for metropathia, so that considerable cell killing is likely to take place in the areas of pelvic marrow which are irradiated. E. LEUKAEMIA FOLLOWING OTHER RADIOLOGICAL PROCEDURES i D. LEUKAEMIA FOLLOWING TREATMENT OF

85. An increased inciderce of leukaemia has been

] ANKYLOSING SPONDYLITIS BY X.lRRADIATION observed in the various series of patients followed up j" after the use of Thorotrast as a diagnostic contrast

83. In a survey of patients who had received one or more medium (see paragraphs 283 ff.). As shown in table 6,a courses of radiotherapy for ankylosing spondylitis, total excess of 53.7 cases has occurred in series including Court Brown and Doll (29) observed an excess mortalit{

4594 subjects followed for a mean period of 27 years rate from leukaemia and splastic anaemia of 4.210-after receiving a mean dose of 26 ml of Thorotrast. Risk i during an average penod of 9.7 years in patients fully estimation from these data is subject to considerable l followed up, and 4.710-a in a rather longer but less uncertainties in respect to irradiation of already dead complete follow.up for an average of 11.4 years. Since cells, or to any possible chemical carcinogenic action of 10 years appears to be about the median interval thorium oxide, as discussed in paragraph 282. Subject to between irradiation and death from radiation induced these reservations, the induction rate from the absorbed leukaemia, a total induction in the region of 910-3 dose 4: livered to the end of the period of fol!ow-up cases might be expected. It was likely that an average would be 33 (24 39)10-* rad-' of alpha radiation. dose of about 880 rad was delivered to 40 per cent of-Since the thorium remains in the tissues and continues the bone marrow. If this can be regarded as equivalent to to irradiate the bone marrow, however, much of this an average dose of 350 rad, the total induction rete per unit absorbed dose is estimated as 2510-' rad-', the

• An additional 5 pet;ents died with leukaemia recorded number of excess cases observed determining 90%

on the death cer incate, although not as the primary cause of confidence limits of (20 31)10-' red-8 The mean deeth; and for 2 patients, death was recorded as due to spleet6c interval in years from irradiation to death had been as

• • ""* but, on resw. H wu conadened that a mon appropriate diagnosis would have been of leukaemia. If these 7 IOIIO*8 patients are included with those recorded as dying ofleuksemis, Folkwup By deaths By death refer and if 3 further patients are excluded in whom leukaemia is known to have been present at the time of treatment, the total Complete to 1960 6.2 *0.5 (SE) 8.6 excess leukaemia mortality rate per unit absorbed does within 20 Incomplete to 1%3 6.610.5 (SE) 7.9 years from treatment becomes 13.5 (9 20) to-* red-'.

375

TABLE 6. LEtJK AEMIA FOLIJDWING THOROTRASTINJECTION Dost to bone marrow inut) Cases ofleukaemia Followup A t and 10 years from Number and snapiastic pened Dose of end of of /,... ) Reference Country (years) (ml) foRowup followup subjects Obsermt Expected anaemse k j 141 Portugal 30 26 387 25 8 1 231 20 0 I I 42 Denmark 28 23 322 207 756 11 (3.7) 69 IJnited States 10 24 120 0 724 3 0 l 72

Germany, l

Fed. Rep. of 30 30 444 296 1750 25 2.4 96 Japan 30 16 271 189 133 1 0.2 Mean 27 26 353 221 } Total 4 594 60 6.3 radiation is likely to be " wasted" in the r,ense that it is

89. Additional studies have been made of the j

delivered after leukaemia may have been induced. If a mean latency of 10 years is assumed for leukaemia, and exposures likely to have been received by radiologists in the dose estimated for the period of 10 years less than the course of their occupation (49) but no further the mean duration of survey, the rate would be 53 evidence has been obtained from which corresponding risk rates are derived. (41-67) 10-' rad'8 of alpha radiation.

86. A small excess incidence ofleukaemia occurred in a series of 2872 children who had had scalp irradiation for tines capitis, but only a very tentative risk estimate F.

SUMMARY

can be made from these data. De cranial marrow dose 90. was estimated (131) as 385 rad, but the proportion of 12ukaemia is thus induced by radiation with a the active marrow which is in the cranium of a child is mean latency to diagnosis of about 10 years. Estimates uncertain. Atkinson (8a) has estimated that 7 per cent of of the mean or median interval from exposure to death, the active bone marrow of children is in the skull,and on under widely different conditions of irradiation, vary this basis the mean marrow dose may have been in the from about 8 to 13 years. These mean intervals are [mT region of 30 rad. The excess leukaemia incidence wu of probably about half those which apply for other ancers. ( V 4.7 (1.010.9) cases, the confidence limits ignoring They do not appear to differ considerably for different i variability in estimation of the control rites. These types of leukaemia. They do, however, vary with age at values give a risk estimate of 50 (10130) 10 rad-8, exposure, estimates from Hiroshima and Nagasaki but there are obviously considerable uncertainties, apart increasing from 9 to 15 years with increasing age at the from the statisticalimprecision in this estimate. time of exposure; and the latency is short after pre. natal exposure. The mean latency does not appear to vary

87. No additional information has become available with dose in a consistent manner, and the increased since the Committee's 1972 report on the possible acidence after exposure appears to have almost ceased induction ofleukaemia by P as used in the treatment after about 30 years.

cf palycythemia vera. As noted in that report (Annex H, para. 85)it remains difficult to derive valid radiation risk

91. All acute forms, and the chronic granulocytic type estimates from study of the sequels of this treatment.

of leukaemia are induced, the former types predo-The position is further complicated by the evidence that minating. No evidence has yet appeared indicating that the frequency of subsequent acute leukaenua is not chronic lymphatic leukaemia is induced by radiation. directly related to the amount of P administered,and va ies according to the size of the spleen at the time of

92. The frequency ofinduction per unit dose increases treatment (21).

with age from early adult life, but is also relatively high following exposure in childhood, and perhaps also irr

88. No further data have been published on the utem (see chapter VIII). There is no clear evidence of a follow-up of hyperthyroid patients treated with '31 or difference in induction rate in the two sexes, 8

by surgery (1972 report, Annex H, para. 78). Since the initial survey of hyperthyroid patients continued for a

93. Here are no reliable bases for determining directly mean period of only about 10 years, terminating in the frequency with which leukaemia is induced in the mid-1967, a further study might be informative, but the adult at absorbed doses of low LET radirtion of less estimated mean marrow dose, of only about 10 rad, than about 100 rad. The total deaths from !euksenua in makes it doubtful whether useful results would be the Life Span Study 1950-1972 in those estimated to obtained. In the series of patients treated for thyroid have been exposed at 09 rad were recorded (97) as cancer at high dose levels with 31(1972 report, Annex higher than those in people who were not in the cities

, para. 73), no further cases of leukaemia have developed despite a further 9 years of follow-up of this (P= 0.065) and were also higher than expected on the basis of Japanese National Statistics (P= 0.004), but series (118). neither companson allows estimates of any precision. 376

.? i:

94. The mortality rates for leukaemia in Hiroshima in

99. In its previous reports, the Committee has kerma ranges cf 10 to 50 rad and higher allow induction reviewId information indicating the induction of thyroid

},[. rates to be estimated, but here the neutron component cancer by radiation in Japan se A. bomb survivors, in jy of the dose complicates the interpretation of the residents of Marshall Islands exposed to radioactive ,b 8 tes. For Nagasaki, the estimates for several dose fallout. in patients therapeutically irradiated from 9 are too imprecise to be useful. Here however, extemal sources, and in children treated for hyper. 4 es imates derived from the inclusive groups of all over thyroidism with radioiodine. 4 200 rad, over 100 rad etc. are based on adequate' ! L cumbers. These estimates (table 4) suggest clearly that I the frequency of leukaemia induction per unit kerma A. THYROID CANCERIN A-BOMB l[9 decr:ases with decreasing kerma, and hence with SURVIVORS decreasing absorbed dose. As the inclusive dose groups ig 'are widened to include a progressively greater contri-100. From the report of Wood et al. (163) it was h 5; bution from those. exposed at lower dose, the rate per evident that thyroid cancer had been induced by unit kerma falls from 3510-' rad-' at mean kerma radiation in members of the Adult Health Study Sample 330 rad, to 1810-' rad-8 at mean kerma 100 rad, the of 11 roshima and Nagasaki. It was clear that the rate of rate pr unit absorbed dose in this group being 3310-' induction was greater in females than in males, and the rad at mean dose about 55 rad.His decrease of effect sex ratio of affected females to males was 2.2 in the per rad with decreasing dose is perhaps to be expected group exposed at distances to 1400 m. The rates of for the predominantly low.LET radiation and suggests a induction were much higher in those who were exposed 8 value of about 3010-' rad as an upper limit for the proximally than in those distally exposed as examined likely effectiveness of low absorbed doses of low.LET in females, the number of male cases being too small for radiation. statistical tests. For females of all ages combined, the group exposed within 1400 m had a rate of 2.5 times as

95. His inference would be consistent with estimates high as that for those exposed at distances greater than derived above for the rates, per million per rad, of 25 3000 m.

h and cf 11 folicwing x ray treatment of ankylosing spondylitis, and of 17 after pelvic x. irradiation. The 101. -The rates were also expressed in terms of estimates ii tentative estimates of 50 55 rad-' of alpha radiation of kerma at different ranges, but it was difficult to from thorotrast deposits and of 50 (10130) following derive values for the annual incidences per unit kerma treatment of tinea in children are not inconsistent. since available records were only of the total number of f' L! cases diagnosed at examination about 20 years after k-

96. An estimate in the range (15 25) 10- rad-exposure. With assumptions as to the mean dose

/ 7t thus be taken for the induction rate forlow.LET corresponding to the stated dose ranges, however, the h Catson in a population of all ages, probably with a Committee made tentative estimates of induction rates 2 somewhat higher rate in the young and in the elderly, corresponding to between 10 20 cases per million per and lower in early adult life. It must be emphasized, rad for males, and 20 40 for females, for kermas in the j however, that this estimate is derived predominantly range 25 200 rad. The variation of induction rate with be k-from rates observed following absorbed doses of over age at the time of exposure was not clear from the data Ei 100 rad. While the rate per unit dose from doses of a few then available.

l rads is unlikely ta be higher than this value,it might be

!I substantially lower. 102. Since the last report of the Committee, a further detailed study has been published (109) of thyroid ?* [j ~: cancers diagnosed between 13 and 26 years after exposure in members of the Adult llealth Study III. THYROID CANCER population.nis examination clarifies further the greater j. 6 induction rate in females, the histological type of tumour induced, the rate ofinduction per unit absorbed .h

97. De Committee's previous reports have shown the dose and, to some extent,the variation of induction rate 3

thyroid gland to be an crgan of high sensitivity for with age at the time of exposure. Of the 74 thyroid l radiation carcinogenesis. Indeed, considering that the cancers recorded in the population studied only 40 were 4 mass of the Wand is typically about 20 g, much of which diagnosed during life (see paragraph 104). The other 34 yl b cell-free colloid, the sensitivity per cell is likely to be were found first at autopsy and were rvither detected considerably higher than for most other types of tissue. durinF life, nor judged to have been detectable by 1 I normal clinical examination. They were usually small

98. At the same time, the malignant tumoursinduced and of the papillary sclerosing type,although sometimes have conshtently been of a histologically well.differen-showing metastases to !c, cal lymph nodes. The study by

?. tiated type which usually develops only slowly and Sampson et al. (130) of tumours of this type was j which can often he completely removed by operation or discussed in the Committee's 1972 report, where it was j successfully treated with radioiodine even if meta-noted that the frequency of these tumours was increased . g stasized. For this reason the Committee has previously by about 23 per cent in autopsies in females receiving k ted risks in terms of morbidity rather than of kermas of over 50 rad. He high

• prevalence in

!g ality. It will, however, be necessary to assess the unirradiated subjects, however, in about 16 per cent of i (L ely contribution to fatal cancers as well as to cancer autopsies, was regarded as casting doubt on the clinical {$ induction, from thyroid irradiation (paras.148 and importance of these tumours.Indeed, Woolner (164)has reported a study of 140 cases of occult papillary 149). 377

carcinoma cf the thyroid c.bserved during a 30-year period and regards the presence Cf these tumours as 105. The relationship between excess incidence and age 1 having no effect on mortality. at the time of exposure could not be examined in males, g since too few cases were observed. In females, the ratio 103. The autopsy findings of Parker er d (109) rely on of the incidence in those exposed at over 50 rad to that 3 p routine autopsy methods, rather than on serial sections in the group with minimal exposure was noted as being ( jof the thyroid, and the author? consequently report a greater in those younger than 20 at the time of exposure 3 V lower incidence of such " clinically silent" tumours in than in older age groups. The relative risk is difficult to the unitradiated group (receiving kermas of less than interpret, however, in view of the few cases occurring in the unexposed group, in three of the four age groups g I rad). He observed frequencies at autopsy were 1.3 only one case having been observed. The absolute and 1.1 per ten thousand person. years of observation in increase, therefore, appears to be a more reliable index males and females respectively, tumours of this type of tumour induction, and was probably, although not having been detected in about 0.15 per cent of autopsies. Two thirds of these tumours were of the necessarily, higher in those aged less than 10 years at the tiine of exposure than in the other three age groups. papillary sclerosing type, the remainder being either papillary or follicular. As in the study by Sampson etd Rus, the excess numbers of cases per 10 000 person. (130), there was a significant increase in the frequency years in the four age groups, as compared with the of these " clinically silent" tumours in females who had incidences obsesved in the unexposed group, were 14.2 (5.0 32.7), 4.9(1.9 10.3), 3.6 (1.4-6.9), and received 50 rad or more, but no such increase was 5.4(0 16.2) in those aged less than 10 years, I detectible in males. The numbers of tumours detected 1019 years, 20-49 years, and 50 years and over, i only at autopsy in the low. dose (less than I rad and respectively. The observed occurrence of 4 cases in l not.in-city) and high. dose (greater than 50 rad) groups females aged 10 years or less at the time of exposure appear to have been as follows: would only have been equalled with a chance of P = 0.04 l e Lowdoretap High-doscE*P if the rate of inductica at this age were equal to the l average of rates at an other stages. A fall of induction Maie Female Male

remal, rate D, with age A is also of probable significance i

Tumo" (r = -0.85, P = 0.08, the data being fitted by the l equation D = ll.6,0.22A). De suggestion that the Papillary 0 0 1 3 gland in childhood may have several times the sensitivity Follicular 0 4 0 0 Sclerosing of the adult gland thus continues to support the earlier indication from study of the same subjects (130) in papillary 5 3 I 8 which the percentage incidence of thyroid cancers in Person years (102) 38 63 20 23 those exposed at less than 1400m was 1.110.62 in

1. r sclerosing papillary tumours in females.dicating an increase in incidence rate fo those aged less than 10 at the time of exposure, as compared with 0.2210.13 in those aged 20 or over. The excess incidence in those aged between 10 and 20 had an intermediate value of 0.8310.29.

104. D e remair'ing 40 thyroid cancers had been diagnosed cluncally during the course of the Adult 106. Of the 40 tumours diagnosed clinically, histo, Health Study. Of these,9 occurred m persons who had logical examination of 34 tumours in high and low dose received less than 1 rad. Six cases, with 3.9 expected, ranges showed that papillary and follicular cancers both occurred in those exposed to between 1 and 49 rad. A had higher incidences per person-year in the high. dose further 25 cases (4.5 expected) occurred in those group, of 50 rad or over, than in the group who were not exposed to 50 rad or more and the analysis m the report in the cities or who received less than I rad, clearly I b based essentially upon this more heavuy exposed

g,,

g g g group. Comparing this group with those receiving less clinical cancers were as follows: than 1 rad or not in the city at the time of the bomb, 1**d *"#

1. ish'd#' "#

the excess of cases diagnosed clinically corresponds to a Male Female Male Female rate of 2.2 (0.7 5.0) per 10 000 person-years in the male, ,pil7a and 5.0 (2.9 7.9)in the female. The mean dose for the i 6 4 l' exposed group is not stated, but appears, from the Follieu r 0 2 1 7 distribution of person-years at risk in the three exposure Sclerosig groups quoted, to have been about 200 rad kerma. If s ' papillary 0 0 0 1 and since observations covered thirteen years, the total Person years (103) 38 63 20 3*' excess incidence for the period 13 26 years from As in previous series of radiation-induced thyroid exposure corresponds to rates of 14 (5 31) 10-' and cancers, no 33(19 52)10 per rad kenna in males and females, anaplastic or medullary tumours were rtspectively. The rates per rad absorbed dose would be observed (see paragraph 138). The absence of anaplastic cancers is of importance, since these are rapidly growing 20(6 44)10-5 and 47 (27 73) 10, assuming ratios of and commonly inoperable tumours. The absence of absorbed dose to kerma as in paragraph 59 for gamma medullary carcinomas may suggest that the parafollicular radiation and neutrons; or of 11 (3 24) 10-* and cells of the thyroid are less sensitive to tumour induction 24(14 38)l0 per rad of absorbed dose weighted than are follicular cells, but this difference may arise tding to the weighting factors for neutrons given in 63710-* and 1910 per rad weighted absorbed i;raph 55. The rates for both cities combined would merely because fewer parafollicular cells are present in the thyroid and at risk. Only one of the 40 clinically dose. diagnosed cancers had caused death at the time of the report. 378 r

M 107. Tumour registry data for the period 1959 1970 C. THYROID CANCER IN PATIENTS THERAPEU-(14) show incidence rates of thyroid cancer per unit TICALLY IRRADIATED FROM EXTERNAL t kerma of about 10(5-15)10-' rad-' for Hiroshima and SOURCES 5-17) 10-* rad-8 for Nagasaki. These would corre-l d to rates per unit absorbed dose of about 112. Since the Committee's last review of this subject in ,)j j 7 20) l0-' rad-8, and 14 (6 23) l0-' rad-8, with its 1972 report, further information has become equal rates in the two cities. The rates per unit weighted available on the carcinogenic effect of radiation of the b absorbed dose would however differ substantially in thyroid in infancy and in childhood. The earliest reports Hiroshima and Nagasaki, with values of about 4(2 6) had been of radiotherapy given for supposed enlarge. d and 13 (6 20) l0-* rad-8, respectively. ment of the thymus, and the possibility could not be .e excluded that some abnormality of the thymus, for example in relation to immune mechanisms, was in itself '[ B. THYROID CANCER IN POPULATIONS IN THE determir.ing an increased incidence of subsequent d MARSHALL ISLANDS EXPOSED TO THYROID thyroid carcinoma, since control series did not include Bh IRRADIATION FROM FALLOUT children with a corre.ponding thymic enlargement (see Annex 1, paragraph 59). j f. e 108. The Committee's 1972 ' report described the development of 4 thyroid cancers in residents of certain 113. Subsequent evidsnee, extensively reviaed by c f the MarshallIslands who had been exposed in 1954 to Lindsay and Chaikoff (80) however, made it clear that

[

fallout (25), both by extema' radiation and by ingestion thyroid cancers were developing after irradiation for a in wide variety of clinical conditions in infancy or jf cf radioiodine, childhood, so that the development of the cancers could

• Q 109. By September 1976,3 further thyroid cancers had not be attributed simply to pre existing thymic Q:}

developed and the total incidence at that time abnormality. Indeed, it seems likely in retrospect that I corresponded to 7 such tumours in 243 subjects exposed most of the thymus glands irradiated for a supposed t3 both extemal and intemal radiation (27). To within enlargement may in fact have been normal. 22 years from exposure, therefore, the diagnosed 114. The prespective study of Pifer et al (112) was g incidence rate per unit absorbed dose was J 145 (70 270) 10-' rad-' (table 7). described in the Committee's 1964 report, which also noted that the analysis by Toyooks et al. (153) } indicated that 29 of 34 thyroid tumours (malignant and j TABLE 7. INCIDENCE OF THYROID CANCER IN MAR. SHALL ISLANDS POPtJLATIONS benign) had developed in 472 children treated by .d combined anterior and posterior radiation fields, ' (* whereas only 5 tumours developed in 2111 children j ^ cYrs dose humber Rare per unt, e at Number c Qx~posure of to thyroid of absorbed dose treated by anterior fields only. i-tyearsi subjects (man rad) eencers i!0* rad") 115. A further report of this study was published in h 1967 (56). when 2876 children had been followed for the mean period of 16.5 years, during which 19 thyroid d 1 8 25 8 10 18 34 7 980 2 250145 790) >l8 122 14 645 3 205 (55-535) carcinomas had been diagnosed as compared with only a Tttal 0.14 expected. Since the mean absorbed dose to the exposed 243 48 535 7 145(70 270) thyroid was about 160 rad with the field size ordinarily p used, this incidence corresponded to a mean rate of .f (Jnexposed 504 0 about 4010-' rad-8 h3 Sourcer Reference 27. 116. In reviewing these and other series of irradiations 1 ' i^ involving the thyroid in infancy or childhood, Dolphin E l10. These data neither support nor exclude a greater (38)in 1968 suggested that the frequency of tumours 5 sensitivity in childhood, the induction rates per unit observed by various authors after different periods of f absorbed dose in those aged less than 10 years and in follow.up could be corrected to allow for tumours which i 20510,d more than 18 years at exposure being 80 and might develop subsequently. This correction was made 3 those age rad. Both estimates, however, have wide on the basis of information on an observed distribution 4 confidence limits and are consistent with the average of latent periods between irradiation and the diagnosis d[f rate for all ages (table 7). of thyroid cancers. On Jtis basis he inferred that the ultimate incidence would be of about 10010-* rad-' 111. All the cancers have occurred in females (27), following mean doses of a few hundred rads delivered in ]E indicating a significantly higher mduction rate than in infancy. M,.! males: 117. In 1968, Hempelmann (57) reported further on 7 cancers in 130 females exposed, rate 5.4 the series previously described by Pifer etal (112), Q (2.510.2) per cent extending the mean period of follow.up to about 0 cancers in 113 males exposed. rate 0.0(0.02.7) 25 years and reporting the diagnosis of further thyroid r 4o cancers were observed in 5(M subjects who were per cent cancers in the "high. risk subgroup C" involving irradia. p tion by both anterior and posterior fields with an [ unerpsed, and no data are available on the relative estimated mean thyroid dose of 335 rad. In this group [ freque cy of naturally occurring thyroid cancers in the the excess i:icidence rate of thyroid carcers per unit L.- absorbed dose corresponded to about 13010-* rad-' as d c two sexes in this population. y p 379 ,r

~' m diagnosed - within an awrap of 25 years following in subgroup C as compared with other groups is thought irradiation. Sewral surwys, however, indiate much likely to be due to the closer surveillance of children lower incidences. For example, Janower and Miettinen from subgroup C and probably also to the longer mean g (68) found evidence (by postal questionnaims) of only follow-up of this group (31 years) than for. others 2 thyroid cancers, as compared with 0.4 expected on (23.5 years). The dose effect relationship for thyroid ( . the basis of control populations, in 466 subjects . cancers was ensistent with, but not necessarily irradiated in infacy or childhood at a mean thyroid diagnostic of, a linear function (although that for benign exposure of about 400 R, surwyed at 30 years after tumours appeared not to be) For the dose range and exposure. These values correspond to a risk of only follow-up times involved, these data indicate rnean risk 3 ' 10 (0 30) 10** red-'. estimates of about 16010 rad *8 and 5010-' red-8 for Jews and non. Jews in the United States of America 118. A further report by Hempelmann et af. (58) has irradiated in childhood. His difference may, of course, ^ madafied the pneral applicability of the risk estanates be due to cultural or environnental circumstances, for derned from the high-risk subgroup C. It is now example, influencing diet or surveillance, ratlwr than to suspstod that much of the incseased risk in this group, any pnetically determmed racial factor. Hempelmana as compared with the other groups studied,is likely to refers to reports (103,133) indicating a somewhat have been due to a higher induction rate per unit. higher mortality fr9m spontaneous '4yroid cancer in i . absorbed dose in Jewish than in nonJewish children, Jews than in nonJews in New York City. lsraeli cancer and to the fact that a lary proportion of children in Registry data (35) show no increased incidence for Jews subgroup C were Jewah. born in Israel but* moderate increases for Jews born i elsewhere, as compared with registries from other i 119. Table 8 gives separate estimates of these risks.The countries, and particularly if the unusually high values somewhat higher rates obserwd for each racial category from the 5 Hawaiian registries are excluded (table 9). t TABLE 8. THYROID CANCER FOLLOWING X-RAY EXPOSURE IN CHILDHOOD ' Expensre Jteedel Namenaerof Nasse6er of Mean door Aare greaar orkbe anecene anbferse fred) (20** (neene red)**) " Subgroup C Non Jewish 4 136 388 76 (26-174) I Jewish 9 125 410 176 (92-306) - Other groups Non-Jewish 9 2 506 87 41 (22-72) j Jewish 2 105 153 124 (22-392) All groups Non Jewish 13 2 642 102 48 (28-76) i Jewish 1I 230 293 163(92 270) .Soance: Rererence 58. TABLE 9. THYROID CANCER REGISTRATION 120. The frequency with which thyroid cancer was E'"*"' *I '" *8'* diagnosed in irradiated females was 2.3 times that in Idaa"*l8 Per 100 000/ males. This difference was significant and arose mainly ow ng to a 5-fold difference in frequency in subjects apd 15-29 at the time of diagnosis, the frequencies arn t a,e s.,imee,r being equal in the two sexes for those younger or older than these ages. The ratio of 2.3 was considerably lower Aegdwry Medes Femmies Medes Franses than that for " naturally occurring" cancer of the thyroid in the two sexes in the same district. No relationship was AR Jews 1.6 4.1 1.8 4.2 f und between latency and size of dose. i Jews born in: Israel 0.2 1.0 0.3 2.0 l Africa or Asia 1.8 5.2 1.6 4.6 121. A significant excess was obserwd of 4.7 (1.0-10.9) Europe or cases of leukaemia but no estimate of mean marrow dose United States 3.2 6.6 2.0 4.0 is available. With regard to other tissues within the Non Jews - 0.9 1.2 1.2 2.0 radiation field, it may be noted that no malignant J3 ether resurrn,, tumours of the salivary glands are now regarded as i Mean 1.1 3.2 1.2 3.4 having occurred, the four salivary tumours noted [ a3D 1.0 s 3.1 s 1.1 23.2 previously now being regarded as benign mixed tumours "'8' of these glands. Only one cancer of breast, diagnosed { 8ince Comp etion of the main survey,has occurred. l O.8 2.5 0.8 2.6 aSD s 0.5 a 1.3 s 0.6 s 1.4 122. Observations published by Modan et al. (92) in 5 seance: Rererence 3s. 1974 suggest that thyroid cancer may be induced by

• The ss resistr=.>= above sees the s Howeiten resisaries.

irradiation in childhood (at ages less than 15) at very t 380 =,

much lower tbsorbed doses.Dey report a fallow.up for little from the original best estimate cf 6.5 rad. It 12 to 23 years cf 10902 children irradiated for remains apparent therefore, that mahgnant thyroid ringworm of the scalp and subsequently reported as tumours may be induced by quite low doses from showing an incidence of thyroid cancer in excess of that external inadiation in children, and that the rate of

~

observed in control populations. One such populaticn induction per unit absorbed dose may be comparable was of an equal number of controls matched for age, with that following mean doses higher by a factor of 50 sex, muntry of origin and date of immigration into or so. It may be noted however that in a smaller series of y -Israel. The second control group consisted of sibhngs 2215 children irradiated at comparable doses for the who had not been inadiated and who were within same condition (139), six thyroid tumours developed, of 5 years of age of the irradiated child. which all were classed as benign and none mahanant, after a mean follow.up period of 20.5 years. 123. The development of thyroid cancers was ascer-tained in part by reference to a centraltumour registry, 126. The induction of tumours of brain, sahvary gland which was regarded as having 95.per-cent completeness and possibly skin in this series is dealt with in of ascertamment, but which had only been in operation appropriate sections (see paragraphs 254,257 and 294). since 1960. Since the arradiations were carried out 4 between 1949 and 1960, this check was supplemented 127. It should be noted that, despite the evidence for by examination of death certificates, and review of the carcinogenic effect of radiation on the thyroid in hospital records in the event of deaths involving nfancy and in childhood, no excess of thyroid cancers malignancies or suspicions of mahgnancies. In both has been observed by Stewart (152) or by McMahon methods of ascertainment, the investigator was unaware (83) to result from irradiation in utero. Both these of whether the name sought was that of an irradiated or studies, however, were of rnortality,in contrast to those a control subject. By these means, twelve thyroid of Hempelmann (58) and Modan (92) which recorded - cancers were shown to have developed, as against 2.0 morbidity. The slow progress and low mortality of expected on the basis of the rates observed in the two radiation induced thyroid cancers could therefore control series. It was emphasized that this frequency of acmunt for this difference in findmgs. induction represents an underestimate sincs ascertain-ment may not have been complete during the initial 128. Reference has already been made to the long l period of 011 years after irradiation. The minimum latency between irradiation and clinical diagnosis of induction rate is thus 0.09 (0.04 0.16) per cent. many malignant tumours, including those of the thyroid. j 'the occurrence of the small occult sclerosing papdlary 124. Careful estimates were made (159) of the absorbed thyroid cancers, found in high frequency at autopsies on 'g dose expected in the thyroid. A " phantom" was subjects without evidence of thyroid disease duringlife, irradiated in conditions similar to those used originally, in itself shows that tumours with histological charceters D, with ! :calp fields at 350 R per field. The absorbed dose of malignancy and with metastases to lymph nodes may was estimated to vary somewhat in different parts of the not develop to a clinically diagnosable stats duriag the thyroid gland from 6.2 to 7.4 rad with an average value lifetime of the subject. taken as 6.5 rad. It is of very considerable interest that on this basis the incidence rate is in close agreement with 129. Even for clinically significant thyroid cancers, the that observed by Hempelmann following much higher mean latency appears to be long. Prospective determina-mean doses, Modan's (minimum) estimate for a period tions of this mean latency are liable to be truncated by of 13-24 years from irradiation being 140 (70 240) inadequate periods of observation. Retrospective deter-10-' rad, but for an estimated mean dose of a few minations may be affected by the fact that thymus rads only. It is to be noted that this high risk derives (as irradiation in infancy was practised only over a limited l* in paragraph 118) from data on the irradiation of Jewish period of time so that estimates, which have been based largely on resultmg tumours developing during child-ll children. hood, tend to be dependent upon the date of the survey. l' ll 125. During irradiation, the neck, face and parts of the De Groot and Paloyan (30) however, record a mean age head were shielded with lead-rubber sheetir.g. It is clear of 20.3 years in 20 patients presenting with thyroid however, and is emphasized by the authors, that the cancer whose glands had received absorbed doses, estimate of dose depends critically on the assumption probably in the region of 500 900 rad, when these that the children maintained a constant position during patients had a mean age of 7.4 years. irradiation and did not move so that the thyroid came j into the primary field. Since these fields involved an air 130. Moreover, the available evidence suggests that the f exposure (focus / skin distance of 25-30 cm) of latency increases with age at the time of irradiation. l. 350-400 R per field, such errors in alignment would only Raventos and Winship (122) indicate a rise in mean require to have occurred during radiation of, for latency from 10 to about 17 years following irradiation example, 5 per cent of fields (and each child was in infancy or during adolescence respectively. Greene irradiated with 5 fields) for the mean thyroid dose to (48) obtained similar results in 59 subjects developing have been 3 times as high as estimated. Modan has in thyroid carcinoma, with mean latencies of 9.8 t 0.7 fact repeated the thyroid dose estimation (160) using a (3.E.), 12.412.3, and 19.013.2 years followinf, variety of different head positions simulated on the irradiation at ages of less than 1 year,117 years and phantom, and with different filtrations (01 mm of over 17 years respectively. Analysis (118) of a number aluminium). With air doses of 350400 R to the scalp, of separate publications also suggests alatency(LJ rising the maxiinum range of estimates for either thyroid lobe with age at irradiation (A/ until adult life, according to was 417 rad; but the average value (8.8 rad) differed the equation L = 9.4 + 0.27A. 381 _. ~. _ _ _ - _ _ _ _. _ _ _. _ _ _ _ _. _.. _ _. _., _

131. Park:r et st (109) point cut that new cases the tumours being cf the occult papillary type, of subt continue to be detected in the Adult Health Study diameter less than 1.5 cm,in 40 (11 with metastases to irra population and that 25 new cases have been ascertained nodes), luger papillary tumours being present in iI (4 in the 5. year period since June 1966(i.e.at 2126 years with nodal metastases) and follicular cancers in the I3 after exposure) as compared with 49 in the previous remaining 5 (2 with nodal metastases). These obser.

68. year period. It is of some significance however, that vations do not allow risk estimation, since the original 8*

his ascertainment largely involves cases only detected at radiation exposures are not known and since the persons autopsy. For clinically ascertained cases, only 6 have examined may have been subject to some selection.De 17 ) been detected in this 5-year period as compared with 34 detection of such tumours in 4 per cent of irradiated in the previous 8 years. If the ascertainment truly subjects, with 1.1 per cent of other than the sclerosmg reflects the rate of new clinical presentations during the papillary type, at 2 or 3 decades after irradiation, corresponding periods, this may suggest some reduction however, supports both the high incidence and the slow 13 1 in the incidence of new cases with time. progress of such induced tumours. Becker et al (12) mi report similar I'mdings in adults who had been irradiated e> 132. This point is of great importance for estimating in infancy or childhood, and who were operated upon id the total expected development of tumours during the because of scintigraphic abnormalities at a mean time of be whole lifetime of irradiated subjects. A report of 25 years after irradiation. Of 15 operated,8 were found di considerable interest in this connection is that of to have thyroid cancers. (1 Refetoff et al (123). These authors desenbe the findings U on examination and at operation on patients who had 135. Favus et al (45)also examined during 1974,1056 h been irradiated for a variety of reasons (commonly for subjects who had had radiation therapy dunng the 1940s ti tonsillar and thymic enlargement)in childhood and who and 1950s and who were examined in order to detect a had come for examination only because they were aware any resulting abnormality, and not because of any n of the possible consequences of thyroid irradiation,and symptoms or known evidence of thyroid disease. If a not because of any abnormality detected clinically or by palpable or scintigraphic abnormalities were found, 8 themselves. Of 100 patients so examined, operation was operation was suggested and the findings at operation t recommended in 18 patients who had either discrete were known in 182 of the 270 subjects in whom this ( thyroid nodules, diffuse enlargement of the gland or advice was given, a total of 60 thyroid cancers being abnormal consistency in the absence of raised thyroid found. Of these,15 were microscopic in size,6 were less autoantibodies. Operation was performed in 15 of these than 5 mm in diameter and 39 were larger. Capsular patients, a total extracapsular thyroidectomy being invasion was noted in 12, blood vessel invasion in 5, and carried out. In 7 cases thyroid carcinomas were found,5 lymph node involvement in 17. Of the tumours larger being shown to be invasive, while 5 were multifocal in than 5 mm in diameter,10 were papillary and 27 were aracter. Four of the glands removed were multino. of mixed papillary and follicular structure, only 2 being ar, I was diffusely enlarged and I showed a single follicular. odule on palpation, so that none of these tumours appear to have been of the occult sclerosing papillary 136. The thyroid dose could be most reliably esttmated type. All were either papillary, follicular of mixed in in a group of 859 subjects who had had x. irradiation of character, a the tonsil or nasopharynx only. Their mean age at I exposure was 5.4 years, and at exammation 32.8 years. 133. The mean age at irradiation was 4.7 2 5.2 (SD) The mean dose to the thyroid was estimated at 725 rad, years and the mean interval since irradiation was and 57 thyroid cancers were found in this group, 24.415.2 years. The estimated dose to the thyroid correspondmg to a risk of 85 (65105) 10~' rad at i 1 varied with the reasons for irradiation, but had an this dose level, the rate for cancers greater than 5 mm l average throughout the whole group of about 750 rad. It diameter being about two thirds this value and the rate is c f considerable interest that this incidence of thyroid in an unirradiated population being small. Three cancers cancer, undetected clinically at 24 years after irra. occurred in the remaining subjects with an estimated dittion, in 7 per cent of subjects corresponds to a rate of 42 600 man rad to the thyroid giving a similar total risk, 6 95(45 175) 10 rad-', which is close to estimates of of 70 (20185) 10 rad-8 after a slightly shorter the frequency with which such tumours are diagnosed period, of 26 years since irradiation at a mean age of clinically at intervals of up to about 20 years following 6.7 years. Taking account of the fact that histological irradiation in childhood or infancy, as in the series findings were known in only two thirds of those subjects l reported by Pifer Hempelmann and others. If all the for whom surgery was recommended, the authors I tumours detected by Refetoff and his colleagues would estimate an annual risk of 4.510 y" rad" for the have become diagnosable clinically during the lives of group receiving tonsillar and nasopharyngeal irradiation ( i these subjects, these observations would appear to only, and emphasize that this may be an underestimate confirm that a median latency for clinical detection of of the final risk. If no bias was introduced in the actual g thyroid cancer following irradiation in infancy may be selection for surgery of those for whom it was s cf the order of 25 years. recommended on clinical or scintigraphic evidence, the l + risk by 27.4 years would be 125 (100-155) 10-* rad", 134. Similar observations are reported by Amold et al with microscopic tumours forming about 25 per cent, (8) on 1452 persons with a history of x. ray therapy to and tumours larger than 5 mm forming about 65 per neck for benign disease 18 35 years previously. In cent of this total. r cent, clinical or imaging studies indicated the possibihty of abnormality and 193 were explored 137. Paloyan er al (108) report the histological findmgs i ( surgically. Thyroid mahgnancy was found in 56 of these, in 38 thyroid cancers found at operation upon 70 ,j l 382 ,3g l n

V o s subjects who had nodular thyroids and had had radioiodine uptake, has fallen to below about 5 per cent f' irradiation ta the neck in childhood: of n:rmal, it is to be presumed that considerable cell l killing or inhibition must have taken place in these A 13 were papillary (7 m. males, 6 m. females), 6 havm.g patients. nodal metastases but none with remote metastases re follicular (4 in males,4 in females),3 with nodal, 141. In adults no excess of thyroid cancer has been } and 2 with remote metastases observed following the radioiodine treatment of 17 were of mixed papillary and follicular structure (9 in hyperthyroidism in a large series of patients followed for males and 8 in females),11 with nodal and 2 with an average time of about 10 years (33). Isolated -{ instances are reported in which such a cancer has ] remote metastases. followed radioiodine treatment in an adult (87),11 such .g 138. Rese histological findings are characteristic of cases are noted in one review (90). In view of the very 'j most reports of radiation induced thyroid cancer, and no large numbers of patients who have been treated in this )q excess of anaplastic or medullary cancers has been way, however, with a predominance of females in the identified. Only two anaplastic tumours appear to have higher age groups at which thyroid cancer incidence is been reported as following radiation exposure, one being normally greatest, the numbers at present reported do diagnosed 6 years after external irradiation in infancy not suggest any excess above a chance expectation. On 4 (140), and the other following radiciodine treatment the basis of the age and sex distribution of patients being 4 (10). In both cases the previous irradiation may perhaps treated for hyperthyroidism with radioiodine, an J have been fortuitous, but it still cannot be excluded that incidence of 44, and a mortality of 16, cases per million .I the types of tumour that develop are those characteristic patient-years were estimated to be likely to occur by

]t at the age at which they are diagnosed, since the chance among these patients (114) if the normal majority of irradiations studied have been in childhood, incidence of thyroid cancer is the same in hyperthyroid i

and tumours arising in early adult life are rarely patients as in the general population. Since the total anaplastic. One case of a probable fibrosarcoma of the experience from this treatment is 11ely considerably thyroid after irradiation has been noted by Hempelmann to exceed I million patient years, the present evidence e (see paragraph 148). appears to exclude any substantial cancer induction by d irradiation at these high doses in adults also. It is, however, important to note that the period of a ' follow up of many patients is still relatively short,since D. TilYROID CANCER IN PATIENTS these treatments were only started 35 years ago and only j; TREATED WITH l became common within the last 25 years.Moreover,the s activity ordinarily administered has been reduced within d (D. The Committee rr. ported in 1964 (154) that a the last 10-15 years and it will remain important to ] U Jroid carcinoma of low. grade malignancy had confirm that no increased cancer incidence occurs developed in a child treated for hyperthyroidism with following this use of the rather lower absorbed doses, 4 now commonly in the region of about 4000 rad. p ia:I but noted that, apart from this instance, no association had been reported between this treatment ji and the development of thyroid cancer. 3 140. In a recent study, Safa et al. (129) have followed E. SWMARY 888 f 87 patients treated in 1949 with 1 for hyper-thyroidism at ages of between 3 and 18 years. During 142. In summary, therefore, it is evident that thyroid f 1 cancers are induced by radiation at absorbed doses of the period of follow-up, 5-25 years (with mean 12.3 years), no deaths and no cancers were observed in over 100 rad and probably even at less than 10 rad.The 'j these patients. The authors refer to a total of seven induction rate per unit absorbed dose appears to be t reports, including their own,in which such patients have somewhat higher in females than in males, and is been followed clinically for average periods of about prob 61y also rather higher in infants and children than '} 10 years, and in which 2 cases of cancer had developed in adults. The mortality from radiation-induced thyroid among the total of 273 patients reported. They cancess is, however. low, at least during the first 20 years ?! emphasize the need for continued and longer periods of after irradiation, so that although the thyroid shows a follow up and indeed,if the mean latent period for the high rate for radiation induction of malignancies,it hu a d; development of thyroid cancer following irradiation in relatively low rate for induction of fatal malignancies. A childhood is in fact 20-25 years, this estimate must be j.t treated with reserve. It does,however, tend to exclude a 143. The cancers induced are ordinarily of papillary or g high rate of induction following high doses in children. follicular type; anaplastic or medullary thyroid cancers it remains uncertain whether this is attributable to the have not been reported. The frequency of small occult f}; c.ver. active state of the irradiated gland in hyper. sclerosing papillary tumours is increased by radiation, j thyroidism or to the high absorbed doses used in this but these appear to have little or no effect clinically or J treatment, probably of the order of several thousands of rad, with which substantial cell killing must be on mortality. y ciated it is in fact noted that hypothyroidism velop i in 40 per cent of the children treated in this 144. The induction rate per unit absorbed dose in ,j ries, and in,8 per cent in the six other series noted. females has been found to be between 2 and 2.5 times y e Since el"tical hypothyroidism ordmarily develops only that in males in surveys in which this point has been j when the thyroid function. as measured by its exanuned. Thus, a ratio of 2.3 is derived from the data gf, es 383 t'

of Parker et al (para.104) f;r clinically diagnosable less than 10 tt exposure having onl- 0.4 times the tumours, and also from those cf Hempelmann et al induction rate per unit absorbed dose + diose aged ever I (para.120). In the Marshall Island populations, Conard

18. The confidence limits of these estimates are, ct al (para.1I1) observe rates of 5.4 (2.5-10.2) per cent however, too wide to exclude the possibility of some age f; males, and 0.0 (0.0 2.7) per cent in males.

effect in the direction observed in Hiroshima and Nagasaki. In the Marshall Islands populations the thyroid

5. None of the studies of therapeutic irradiation doses tended to be considerably higher in children than irr olving the thyroid give evidence on an effect of age at m adults owing to the greater concentations ofingested e-osure on cancer induction ' rate, since adequate radiciodine in the smaller glands of children. A lower groups of people irradiated at different ages have not ind.iction rate might perhaps therefore be associated been equally followed up. However, the findings of with doses causing substantial cell killing, and cases of Parker et al in Hiroshima and Nagasaki described in impairment of thyroid function were in fact observed in paragraph 103 indicate an apparent dependence upon several of these children.

age at exposure, for tumours diagnosed clinically 13-26 years subsequently. The mean rate of detection per ten thousand person-years of survey of those aged 147. In the populations studied, estimated induction less than 10 at exposure was 4.112.1 times that for rates until about 25 years have mainly been in the range 5015010 radd (table 10). The rates in Hiroshima those aged 20 or over at exposure, considering males and fimales together. The ratio in males could not be ane Nagasaki are, however, conspicuously lower, with determined with adequate accuracy, but that in females values of 3710** rad" absorbed dose on the data of l e: was 3.712.2. The rate for the males and females who Parker et al (109) or 1410-* rad-' on tl'cse of the it were exposed at age 10-20 is perhaps slightly higher Tumor Registry (14). The lower rates in these cities l s( (1.620.8 times) than in those who were aged 20 or nvolve some ascertainments by autopsy but depend also o on repeated clinical examinations in the Adult Health is over. It must be recognized, however, that these inferences as to a variation of induction rate with age at Survey, so should truly reflect morbidity and not only t mortality. It is to be noted that the incidence rates of i i cxposure, depend critically upon the assumption that the latent period between irradiation and detection of a clinically diagnosed thyroid cancer in the reference c tumour does not in itself depend upon age at trradiation. populations (those not in the cities, or exposed at less I t The observations recorded in paragraph 130 suggest that than I rad) appear to be substantially higher, at least in i in fact the mean latency may increase with increasing females, than in other Japanese, or other national, } age at exposure, and this effect might explain the tumour registries (table 11). These relatively high rates for unirradiated populations in Hiroshima and Nagasaki present apparent difference in induction rates at may well reflect the close surveillance in the Adult erent ages. Health Survey; if so, they should apply equally in the exposed and the unexposed groups. Moreover, the . On the other hand, the induction rates observed in estimated induction rates would only be increased by the irradiated Marshallisiands population (para.110) do about 20 per cent even if the control rates were taken as not support this indication of an age effect, those aged zero. There appears, therefore, to be no obvious i 'e TABLE 10. TilYROID CANCER-INCIDENCE RISK ESTIMATES 6 Risk in period cfstudy T:st Period ofstudy Dose trad) (per unst of reference }%pulation (years ssnce Type of eksorbed dose) ipere.) studied irradiation) radiation hn Range (10'* rn ~*) 102-106 Hiroshima amt 13-26 Gamma and neutron about 200 50->200 37(23-54); Nagasaki (kerma) 19(12 28), (38% male) weighted dose 107 Hiroshima 14-25 Gamma and neutron by 14 (7 20) l Nagasaki regression. 14 (6-23) 0 >200 [ 108-111 Marshall !slands To 22 Internal and 200 27-1150 145 (70 270) ? (47% male) external (Iow-LET) 114-120 Infants and To 24 (mean) Radiotherapy 119 63%<100 48(28-76), ) children to neck 29%>200 non-Jewish; s 163 (92-270), 5 Jewish 135 136 Mainly children To 27 (mean) Radiotherapy 725 aI85 85 (65-105), to neck operated cases; v 125 (100 155), inferred, all cases [j 133 Children Operations at 24 Radiotherapy 725 180-1 500 95 (45-175) o [ (mean) to neck i n 225 Children To about i8 (mean) Radiotherapy 6.5 4 17 140 (70-240) { to scalp (estimated) g .e. 384 'k

._..-~ TABI.E 11. THYROID CANCER INCIDENCE hrsons cf all ages - (Annual rete per 100 000) MeJet Fenannes ~ JAroshinerandNessand, IJnirradiated (chnical diagnoses) 2.7 (0.1-12J) 12.6 (6.3-22.8) Uncorrected Corrected

• Uncorrected Corrected
• Jepenese reetstries 0.7 0.8 2.0 2.0 Miyagi prefecture Okayamis prefecture 1.3 1.1 3.8 3.3 Mean s 3D 1.1 s 1.0

-1.221.2 3.0 s 3.I 3.4 s 3.3 ~ J# otAerresistrars 6 Mean

• SD 0.9
• 0.6 0.9 s 0.6 2.4 s 1.3 2.6 s 1.4 JJ other resistrars Seewer: Rererence 3s.

Corrected to populattom or standard eso. 8

• Escludies s Hawanan seelstries (see parayaph a19).

IV. BREAST CANCER difference between explanation - for the apparent induction rates estunated for these cities and from other A. BREAST CANCERIN A. BOMB SURVIVORS s'ources. The tumour registry data, and the rates observed in untrradsated populations in Huosinma and study of Wanebo et al (158) of the 151. The Nagasaki, tend to exclude a low rate of spontaneous occurrena of breast cancer in A. bomb survivors in thyroid cancer as a possible explanation for the low Huoehlms and Nagasaki was described in the Commit. = induction rate (table 11), although Parker et al quote tee's 1972 report. In view of the low mortality of most 1 - one report (132) as indicating a low mortality rate for forms of breast cancer during the early years after ' this malignancy in Japan. diagnosis, Wanebo et al examined the incirlance of this suase in memben of the morbidity sample of the 148. Few radiation-induced thyroid cancers are reported to have caused death. De Groot and Paloyan JNIH-ABCC Adult Health Study, and based their (30) record that one patient, out of 20 with thyroid evidence on findings at the 6-month health examinations this group. This clinical evidena was cancers following childhood irradiation, had died during made on j a (mean) period of about 22 years from irradiation, supplemented by records of local tumour registries, g although 4 had post. operative recurrences of the disease

• surgical pathology diagnoses, autopsy diagnoses and Parker et al (109) note 6 deaths within 26 years of death certificates. As compared with incidence rates exposure at Haroshuna and Nagasaki in their 40 patients observed in groups exposed to 0 9 rad or not in the city whose thyroid cancer was diagnosed clinically, but only at the time of the bomb explosion, a significant excess one of these deaths wu attributable to the cancer.

of cases was found in the exposed groups, the relative Wilson et al (162) record 2 deaths from the cancer risk increasing with the estimated kerma. ~1he total + 11.6 (15 observed compared with 3.4 ' amongst 58 patients with thyroid malignancies following : expected) in females exposed at known doses in excess excess was neck it adiation, occurring within 23 years of expause. J Six of their patients, however, had lung metastases and of 10 rad. If, as suggested by the numbers in each dosage ,[ one had bone metastases. Hempelmann (58) observed no group, the nwan dose was in the region of 130 rad, an ,j deaths during a mean follow.up of 24 years on 24 induction rate of 24 (12 40) 10-' rad-8 can be inferred patients, with apparently radiation. induced thyroid (table 12). j, cancer. One other irradiated patient was said to have 152. While this excess was noted as being statistically died subsequently from a fibrosarcoma of the thyroid, significant, it was emphasized that part of the excess might possibly be due to factors other than the ' but histological confirmation was not obtained. radiation, since parity, duration of lactations and .~ g 149. These four reports, therefore, enumerate 4 deaths frequency of married status were not equal in the (or 5, including the fibrosarcoma) occurring in 142 exposed and comparison groups and these factors are c 7' L subjects with apparently radiation-induced ~ thyroid known to affect breast cancer incidence (85). However,

I cancer, indicating a fatality rate of about 3 per cent Wanebo et al (158) found no differences in marital F

within a meen period of 24 years from irradiation. status, parity or length of lactation' among the 16 women with, and 7819 women without, breast cancer l 150. If, ignoring the lower values from Japan, the risk who answered a questionnaire on these points. The l . of induction is taken as 5015010-* rad-8 within about average period of lactstion in 10 fertile patients with 25 years ofirradiation(table 10),with equal numbers to breast cancer who had lactated was shorter than in be detected subsequently (para.132), the lifetime risk of women without breast cancer (36 months compared a fatal induced thyroid cancer should be of the order of with 50 months), but most of this difference was 51510-* rad-8, assuming a 3-per. cent fatality risk per explainable by the fact that there were f4wer children in 1 25 years,both for the earlier and for the later developing the breast. cancer group. No relationship was found tumours. Such estimates are necessarily tentative but between the frequency of any of these epidemiological i . reflect the apparently high frequency and low mortality factors and radiation dose. j-of radiation. induced thyroid cancer. 385

.[.

y -

• --,-____--r.-.~~..---m-._.--~%._m__,-.

..w., _..m,,, ,,m,.w,,,.--.w.rmy-,,v,-m,-.,.-.,--_#.-y--+m.__#-.

_( .._~. ?p nTABLE 12. INCIDENCE AND MORTALITY FROM BREAdT CANCER IN HIROSHIMA AND NAGASAKI COstIINED g i Females, all ages I Dese -indeenonrare- 'O asma (casse or deer 4s greary 06 ~ Ex - sed Emerar row .(sed) annen r gotteer Escene (20*} (20** red"). 9er thoanaand persons espoend) INCIDENCE 1958-1966 ~ 10 39 - 4 1.14 ' 2.9 (0.2-8.I) 1 0.023 128 (44-358) 2.3 (0.2-6.4 ) 4049. 2 0.77 1.2 (nes. 5.5) 0.051 24 (nes.108) 1.5 (nes.-6.9) 90 199 4 ~ 0.72 3.3(0.68.5) 0.110 '30 (5-77) 4.1 (0.7-10.6) >200 5 0.76 4.2(1.29.7). .0.294 14 (4-78) 4.9(1.4 10.3) - Total 15 3.39 11.6 (5.9 19.7) 0.478 24 (12 41) 3.1 (1.6-5.3) INCIDENCE 19541972 Conapsrud with 0 9 sodsroesp 10 49 - I9 15.2-3.8 (nes.-l3.4) 0.172 ' 22 (nes. 78) 0.5 (nes.-l.8) a . 50 99 -6 4.0 2.0 (nes. 7.9) 0.150 13 (nes. 51) 0.9 (nes. 3.9) + l 100 199 8 2.5 3.5(l.5-11.9) 0.205 27(7-59) 3.8(l.08.2) >200 4 2.3 1.7 (nes.-6.9) 0.480 4 (nes. l4) 1.2 (nes.-4.9) Total 37 24.0 . 13.0 (3.1-25.1) 1.011 13 (3-25) 1.0 t0.21.9) - Conupered with JapaneseNarna nel5 stistks 10 49-12.0 7.0 (0.4-15.9) 41 (2-93) 0.9 (0.1-2.0) 50 99 as 3.2 2.8 (nes.-8.6) as 19 (nes.-58) 1.3 (nes.-4.0) 100 199 above 2.0 6.0 (2.0 I2.4) above 29(1040) 4.1(1.48.5) F '>200 1.8 2.2 (nes.-7.4) 5(nes.15) 1.6 (nes.-5.4) Total 37 19.0 18.0 (8.6-29.6) 1.011 18 (9 30) 1.4(0.72.3) Joanese: Rererences 97, a ss. Neer: The 9e% coendence,imits are indicated la parentheses. } 153. No further data how been published on the status included deaths occurring up to 1972, and so will haw ~ of the exposed populations in respect of these factors. t O Exanaination of the mortality experience in the Life incorporated records of patients alive in 1966 when the

l morbidity study terminated.

d ; Span Study (97), however, now indicates a significant i excess also of deaths from breast cancer occurring in 155. Both in the study of excess morbidity (158) and in 'these survivors. } that of excess mortality (97) from breast cancer (fig. I), . 154. For females of all ages and from both cities, the the excess rate per unit kerma in the lowest dose groups deaths from breast cancer observed from 1950 to 1972 considered (10 39 or 10 49) is higher than the average, in groups at known doses greater than 10 rad totalled whereas that in the highest dose group (> 200 rad) is

37. The number expected, on the basis of the 0 9 rad lower than the awrage rate per unit kerma (table 12). It group, was 24: as compared with Japanese National is difficult to interpret this observation (see Annex 1, Statistics,19..From evidence as to the. mean doses paragraphs 140 151), which is of uncertain significance I

within these grr ops, the excess mortality corresponds to statistically, and the agreement of the two reports on rates per - unit kerms of 19 (6 35) 10-* rad ~' in this point may be due in large part to their study of the l Hiroshima compared with Japanese National Statistics, same population. When the rate of induction in various and 16 (+34) 10-' rad-8 in Nagasaki. Table 13 giws does groups (cases or deaths occurring per 1000 p the mean rates for the different dose groups. The-exposed) is considered in relation to the mean received I in eacL group, there is no clear indication of a kerma corresponding rates per unit absorbed dose (see paragraph 61) for those exposed at 10 rad or owr would - giving a maximum yield of tumours (fig.1), since the fall at kerma abow 200 rad is.tatistically nnt significant. in be 26 (8 48)10-8 rad-8 for Hiroshima and 20 - (5-42) 10-8 red-' for Nagasaki, with close agreement the mortality studies the total yield falls as the mean between the two cities.'the rates would, however, differ kerma rises from 140 to 350 rad in the two highest dose substantially if expressed per unit weighted absorbed groups, but the estimate in the highest group depends does (using weighting factors as in paragra upon only 4 deaths as compared with 2.3 or I.8 f i . values of 6 (211) and 17 (4 37)10 p 55),with expected, and the significance of this fall is very ( rad-' in doubtful. Hiroshima and Nasssaki, espectiwly. Similar rates apply I E 4 for higher dose groups in Hiroshima but somewhat lower 156. The mortality data t,aggest that the frequency with ones in Nagasaki. While mortality rates per unit kerma of. which breast cancer has been induced by radiation at 1020 deaths per milhon per rad are consistent with a Haroshima and Nagasaki depends considerably upon age morbidity rate of 24 (12 41) cases per million per rad, at the time of exposure. No deaths from breast cancer the consistency must be due in part to overlap in the had occurred by the end of 1972 in those who were less 4 v information on which each estimate depends. Thus the than 10 years of age at the time of the bomb,of whom morbidity study made use of data from death 2757 (as at October 1,1950) had been exposed at i certificates and autopsy data, and the mortality study known doses of 10 rad or more. Few deaths in excess of j 386-4 -t ns n ~ - - + m-c a - A,n,. - w.-r -e m.--.-w.ws. ne n-w-~ --+.~,,-ne-n,n..-n.,-,~,.--, .---n-nn.--nm....~-,,nw--n-e A v

u. 9.. 4 TABLE 13. EXCESS MORTALITY FROM BREAST CANCER IN HIROSHIMA AND NAGASAKI, 1950-1972 Females, all ages A (Compared with Japanese National Statistics) ( ) ' Dose group Excese rete Excese rate per unit 7.-. ~} trad Ob-Er per unit kernee absorbed dose L3 kernme) served preted Excese (10** red) (2 0'* red **) ' :p x HIROSHIMA l!. 1049 13 9.5 3.5 (neg.-l0.2) 27 (nes.-78) 100 199 6 1.3 4.7(1.3-10,5) 39(11 87) <~,r 50 99 4 2.4 1.6 (nes.4.8) 15 (nes.-65) >200 3 1.1 1.9 (nes.4.7) 7 (nes.-25) 9 (nes. 34) y >100 9 2.4 6.6 (2.3-13.3) 17 (6-34) 22(845) !r f > 50 13 4.8 8.2 (2.9-15.9) 17(6-32) 23(843) f > 10 26 14.3 11.7 (3.9-21.8) 19 (6-35) 25(847) D e NAGASAKI u 1049 6 2.5 3.5 (0.1-9.3) 86 (2 228) 50 99 2 0.8 1.2 (nes.-5.5) 26 (nes.-121) p. 100-199 2 0.7 1.3 (nes.-5.6) 14 (nes.62) t ii >200 1 0.7 0.3 (nes.-4.0) 1 (neg.-19) I (nes.-23) >100 3 1.4 1.6 (neg.4.4) 5(nes.21) 6 (nes.-26) l I > 50 5 2.2 2.8 (nes.-8.3) 8 (neg.-24) 10 (nes.-30) i > 10 11 4.7 6.3(1.5-13.5) 16 (4 34) 20(5-42) 'g U Source: Reference 91. b Note: The 905 conridence limits are indacated in parentheses. fi 157. In view of the small number of deaths that have ( expectation had occurred in those aged 20 or over, most resulted from breast cancer, it is difficult to make any J f cf the excess occurring in those who were ID 19 years reliable estimate of induction rates, except for age 9@ old at the time of the bomb explosion. The following 1019. Table 14 shows that, for most dose groups, the ata refers to exposures at or greater than 10 rad kerma: estimates are non-significant or negative and, even where ~P i s gnificant, have wide confidence limits. For those who N,',*,hn",r Total xcess were aged 20-34, only 2 out of 7 dose groups indicate .{ f,,,,

• Age group 195M972 served peered (10-* ntd-8, apparently significant induction, at rates likely to lie d<

OA Er.

erm, between I and 4010}' rad *8 for ages at or over 50, the 155 (17-457) 10

( 0 9 years 49 740 0 0 (neg.19) rad,199 rad group gives a va.a of, all others being neg 100-1019 years 64 253 10 0.4 34 (18 59) >20 years 174 319 27 23.7 6 (neg. 23) no group gives a significant result for those aged 35-49. All 288 312 37 24.1 13 (3-34) It can only be stated, therefore, that exposures have i 5 TABLE 14. EXCESS MORTA1.lTY FROM BREAST CANCER IN lilROSHIMA AND N AGASAKI,19N 1972 Females, all ages, dose > 10 rad (Compared with mortality in 0 9 red group) (Deaths per milhon per rad kermaj ',t Torsi cases y Ageet fyrers) 10 19 20 34 33 49 >30 320 >10 Observed Espected Escess exposere Qgar,,,,9 f 10649 141 ($1-305) nes nes 9 (ns) nes 26 (ns) 19 15.4 ns $0 99 94 (24-250) 27 (ns) neg ne ner 16 (ns) 6 4.0 ns i 100 199 nes 26 (ns) 37 (ns) 155 (174 57)48(13-103) 31(846) 8 2.5 l >200 13(240) 10 (ns) neg neg neg 4 (ns) 4 2.2 ns >100 9 (1 28) 15 (1-28) 4 (ns) 47 (ns) I4 (2-33) 12 (4-25) 12 4.7 > $0 19(741) 17(242) nes 18 (ng) 9 (ns) 13 (4-25) 18 8.7 > 10 34 (18-59) 10 (ns) neg 16 (ns) 6 (ns) 15 (4-87) 37 24.1 q,

)

Total cases: Observed 10 8 12 7 27 37 Expected 0.4 5.3 12.6 5.8 23.7 24.1 a!

Excess, ns nes ns ns

.a [ Mean kerms trad) 98 81 72 61 74 81 (! >10 tad groups Source: Reference 97. Noter neg = estimated escesa negat6ve. = encesa not signincant (at P = 0.0s). t ns 387 't .t* b

l ' 480R2DITY ' C13"' red'8 CW* H>- e 120 l}R): ,n r 100 so 8 .r 80 - 8 u 40 - g t, 4-i l j i t 1 ' *d H l N 4h 2-t i d H \\ 0

== 1v 0 AsORTALITY 010* rad-8 D 10~8 t I 40< 4-t ( s l \\ W t 20- ). 2- + d & a i W i j' 0 { I 0-100 200 300 rad 100 200 300 rad Figue L Var 6stion of breast cancer induction rate with T65 karma (Hiroshams and Nagasaki fe l I - Excess rates as compared with 0 9 rad group mains, all aoss, periods as in Table 12) Upper diagrams-morbidity (158), excess cases (C) induced Lower dangrams-mortality (97), excess fatal cases (D) induced Left daagrams-eacess cases or deaths per milhon man rad Right diagrams-excess cases per thousand persons l (Ranges indicate 90% mnfidenes brnits) -}. g l 388 4

.~ --- F ' nduced breast cancer, fatal within 29 years in those who 162. The' variation cf incidence with dose, for all those I i were apd 1019, probably also at a low rate in those apd 10 years or more, was consistent with a linear who were older, but not, or not yet, in those who were relationship. It is of considerable interest that the slope of this agression was closely similar in the two cities, younpr. despite the diffemacs in neutron component of the 158. It has, however, been emphasized (13) that radiation. Thus, the risk estimates per unit kerms for all [- \\ mortahty is far inferior to incidence as an indication of aged 10 or over were 1.810.6 and 2.010.710-* y-8 radiation induction of breast canar and that, whereas red-' in Hernahnma and Nagasaki; or, when the distribution, were 1.9010-* y-' red-' gard to age

-, ' "= were standardized with re F

. the annual

• excess mortality rates in Hiroshima and Nagneski in 19501974 were 0.43 and 0.2610-* y-'

in Hiroshima rad-8, the excess incidence rate during the period and 1.8810-* y-' rad-' in Nagasaki The corre. -8 absorbed dose would be 19501969 was about 1.510-8 y rad-' in both cities. spondig rates per unit The incidence of breast cancer in A. bomb survivors has 2.6 10 y ' red-' in Hiroshima and 2.410-* y-' now been reviewed (88)in considerable depth and detail rad-8 in Nagasaki. i using numerous sources to ascertain the frequency with i which bmast cans has occured in membas of me M 163. The mean latency after which induced cancers Span udy dunns % W were detected did not vary with does,50 per cent of all 159. This report makes a full analysis of tumours diagnosed between 1950 and 1969 being detected by about 18 years from exposure in each age i (s) The frequency with which breast cancer is 5 induced in females of each of five ap groups (09,1019, group (0 to 9,10 to 99 and > 100 rad, for women less &an 30 yean old at &e time of exposum). 20 34,35-49,> 50); (b) the distribution in time since exposure (and t therefore also in ap) with which induced cancers are 164. Wanebo et al. (158) report that the histological i detectable in each age group; types of tumours occurring in subjects who had been (c) the way in which the frequency of induced exposed in Hiroshima and Nagasaki at over 60 rad (T65 breast cancers increases with estimated kerma or kerma estimate) were similar to those occurnng in those who were exposed at lower dose. Of 12 cancers from the absorbed dose in breast tissue. former group,11 were duct carcinomas, one was a - 4 160. In the exposed group as a whole, the breast cancer comodo carcinoma, and all were ofinfiltrating character. incidence started to increase above expectation within la the latter group also, all were infiltrating,8 being i 10, but probably not within 5, years of exposure. In ductal, one a comodo type and one an infiltrating colloid women of different ages at exposure, however, the carcinoma. A rnore recent survey (74) confirms this 4 subsequent time course of the detection of breast indication that types of breast cancers are induced in j cancers differed. In those who had been less than 10 about the proportions in which they occur "sponta. years old, no breast cancers had been detected (by neously" in the same population, the great majority 1969). In those who had been 1019, few were being infiltrating ductal carcinomas. This type repre. ~ diagnosed prior to 1960, but in 19651969 an eceses was .sented 85,85 and 76 per cent of cancers examined in detectable at an annual rate which exceeded that unirradiated penons (at 0 rad, or not in the cities),in observed in those who had been older at the time of the those exposed at 199 rad, and in those exposed at i bombs. These differenas in time course prevent any 100 rad or over respectively. The tumour had remamed simple comparison of the way in which ultimate total introductal, and therefore presumably of good prognosis, i risk may vary with age. For example,it is not yet known in 13,9 and 21 per cent in these groups. Other types of whether those exposed at ap 09 will show a large cancer repmsented only a small proportion of each excess when they reach the age at which hormonal or series. (1he total series was of 225 cancers classified other influences determine a full expression of " latent" histologically, with 118, 74 and 33 from the three cancers, or will show only a minimal excsts because only groups). Yosiuzawa and Kusama (168) have surveyed a 5 _ few breast cells had developed, and were exposed to number of published series of radiation-induced breast radiation, at the time of the bomb explosion. cancers and note that most such cases are reported as of 0 ductalorigin. l 161. For breast cancers detected in the period surveyed, j however, of 5 24 years after exposure, the annual risk 4 per unit kerms of inducing breast cancer varied with age ffY,r, B. BREAST CANCER FOLLOWING EXPOSURE apo re hrsoni years errimste SD IN DIAGNOSTIC RADIOLOGY l ' (years; coers W (W? y mr*) O.9 1 0.198 0 165.1hs occurrence of breast cancer in patients treated 10 19 38 0.250 2.4 1 0.8 g,r-tuberculosis by artificial pneumothorax 20 34 76 0.292 2.1i0.8 ,, reported by Mackenzie in 1%5 (82) to be in excess 35 49 76 0.252 0.6 1 0.8 of expectation, and this excess was regarded as probably attributable to the large number of fluoroscopic 2 .9 5 examinations requimd to control the lung collapse. the incidence rates being determined by estimated slopes Breast irradiation will have been substantial, since of maximum likelihood regression of annual incidena Mackenzie's patients commonly faced towards the x.rsy on kerma, tube rather than towards the screen during the 389

1; fluorescopies. Myrden and Hiltz (99) extended this fluoroscopies during pneumothorax or pneumoperito. surwy, and reported 22 cases cf breast cancer in 300 neuen traitment and who were followed up for a mean petsents tressed with pneumothoras (7.3 per cent) and _ period of 26.8 years. Just owr half of the total person.

four cases in 483 patients not so treated (0.8 per cent).

years at risk were in the period of 15 45 years after 1 The annual incidence in the pneumothorax treated exposure. No excess occurred (15 observed,14.1 (" patients was about 6.5 tismos that in the general female expected) in 717 women from the same sanatoriums i t populatson of Nova Scotia, but that in the tuberculous who had not had such treatment for their tuberculosis. patients untreated by pneumothorax did not differ i t alenificantly from this general rate. Here was commonly an agreement between the side on which the carcinoma 169. The excess breast cancer incidence started to dewloped and that on which the pneumothorax, when appear !$ years after the exposures and was continuing unilateral,was estabbshed, at 40 years from exposure. The risk increased with estimated does in a way consistent with a linear 166. In its 1972 mport the Committee discused dose.effect relationship. For the whole group, the mean estunates of the dose likely to have been dehvered to dose to breast was estimated as 150 rad assuming beast tissue and concluded that it may have been in the a 15-s exposure yielding a 1.5-red dose to breast i ? range 4 20 rad per exanma=Han. On the basis of this per fluoroscopy. His implied a total risk during the 26.8 year-period of follow.up of estimate, and of the number of examinations involwd, which commonly exceeded 100, it was concluded that (4123.3)/(1047x 150) = 113 (50190) 10** rad *'. The authors estimane an annual risk rate of 6.2 the observed excess of 20 cases in 300 treated patients (2.810.7) 10-* y"8 rad-' by comparison with New j. would correspond to a rate of(20120) l0 cases per York State health statistics (or 5.6 relative to the rad as ^af in the 20 years following initial i exposure with 90% confidence limits of(15160)10-* group of patients). A rate of 2.910-* y-' com rad' is obtained by fitting a linear regression to all L rad *8. Myrden and Quinlad (100) state that a further follow-up which now extends for 22 32 years, of 326 data, or 8.2 if this analysis is confined to doses of less I i patients with pneumothorax shows 32 breast cancers to than 300 rad. For doses less than 100 rad an increased i risk was observed but was not statistically significant. have dewloped (as compared with 22 in the group in However, the relationship between excess rate and 1%7). A control group of $35 patients with pulmonary estimated absorbed dose in the breast was consistent tuberculosis but without pneumothorax or multiple fluoroscopies how shown sewn breast cancers compared with linearity our the whole range of doses used, i' although some reduction of risk is suggested by the data with four in 1 % 7. Without correcting for the at high doses (corresponding to over !!O fluoroscopies). patient. year total of surviving patients, the excess cancer i incidence rate appears now to be of (30140)10-* } grad-8 on the dosimetric basis assumed in the last report 170. Most of the observed excess,10.7 f 4.918.8) cases. j 2f the Comrnittee, with confidence limits (20 200) 10-' occurnd in 341 women who were younger than 20 rad *'. His estimate may be low since it is uncorrected (although older than 13) at the time of the exposure. I f:r the death of 104 of the original 326 patients, and the The remainder, 7.0 (-0.8-17.1) cases, was observed in j filure to trace 18 others. Brown (19) has used the 706 women of 20 or older at exposure, who were j frequency with which the tumour and the pneumo-followed up for a mean time (26.9 years) comparable j thorax were on the same or on different sides of the with that for those aged under 20 (26.4 years). The total i body, to estimate the dose that is likely to have been excess rate during these periods of time taking account of [ deliwred to the breast on the side opposite to the the estimated mean breast dose at each age (17a) is pneumothorax. By this method he obtains a factor of 162(74 285)10** rad *' for those who were under 20 1.5 by which he considers that the estimate of risk per and 78(-8189)10 rad for those who were 20 or i unit absorbed dose to the breast (on the side of the over. The wide 90% confidence zones clearly prevent i pneumothorax) should be increased, to determine the any determination of risk rates at higher ages than 20 j risk for squal irradiation of both breasts. (although the excess rate is just significant for ages 20 29, at 108(11297) 10-* rad 8). They do, however, I 167. The additional 10 cancers have developed in appear to indicate that the rates are highest in those i patients who won aged 20 to 39 at fluoroscopy'and exposed in adolescence. For all cases, the mean interval j crude incidences suggest a rather lower induction rate in between exposure and diagnosis of a breast cancer was those over 30 at the time of exposure: 24.4 years. Allowing for the expected incidence at i l different periods, the mean latency for induced cases is l M""' estimated as about 27 years. 4e theitsel hanst becMenace ( l freers/ member / esmeers fW L 0 19 '59 6 10(4 20) 171. The authors emphasize the substantial uncer. { l 20 29 181 23 13(9 18) tainties in estunating retrospectively the doses to the >30 86 3 4(19) breast. From questions to patients and physicians it was ( concluded that the patient faced the tube in about one The cancers how proved fatal in 14 of the 32 fluorce oped patients, and in 3 of the 7 others. quarter of all the examinations. Interviews with physicians led to the acceptance of a mean exposure time of 15 s, and to the conclusion that 69 per cent of ,68. Boice and Monson (17)have reported a significant physicians had worked with the fluoroscope shutters i excess of breast cancer (41 obserwd and 23.3 expected) open so that both breasts were exposed to radiation. I b 1047 women who had had an awrape of 102 81 per cent always also scanned the opposite lung. i 390 r ( l f 1 i - ~ ~

Estimates cf dose, and thus cf risk, are therefore more and the mean exposure cf breast tissues as a whole was nearly applicable per patPnt than per single breast; and estimatId at about 210 R. This incidence, which was it was in fact found that the cancer was on the same sideobserved during follow-up periods of 10-25 years, was as the pneumothorax in 17 cases, and on the opposite clearly in excess of that (5.9 cases) expected in a general female population at the same ages. No data were f side in 12. available however of the incidence of breast cancer ( following post.partum mastitis untreated by radiation. lf 172. As noted in the 1972 report of the Committee radiation were, however, the sole factor responsible,the (Annex H. para.120), the high incidence of breast excers would correspond to an induction rate of cancer in the series reviewed by Myrden and Hiltz (99) 55(15 115) 10-* rad to all breast tissue. may have resulted from the procedure adopted,with the patient facing the x-ray tube so that breast tissue received high doses. Similar frequencies have not been 176. An important further study (138)on this group of observed m several other series, in which the same patients has now been reported. in which it was possible position was not used, or if mortality rather than to examine the incidence of breast cancer in two groups morbidity was studied, as in the report of Kitabatake er g ,ggg g al (78). Here 568 patients, who had received an average had been treated by x-ray therapy in one group but not of 75 pneumothorax refills between 1941 and 1961, in the other. He incidence was examined also in further were reviewed m 1975. No deaths had resulted either control groups consisting of sisters of members of each from breast cancer (0.4 expected) or from lung cancer of these groups. This procedure is of value in controlling. (1.6 expected). not only against possible differences in breast cancer incidence for genetic reasons in different families, but 173. Similarly Delarue et al. (31) found no increased also against any such differences in different localities since the irradiated patients were mainly from one town incidence of breast cancer in patients followed for a and the unitradiated patients from another. All minimum of 20 years, and who had had multiple diagnoses were confirmed histologically. fluoroscopies in the supine position (i.e. with back tube and receiving 17 rad per towards the x ray examination, whereas the prone position would have 177. The dose to the irradiated breast was commonly in involved 308 rad per examination). Six breast cancers a range corresponding to an exposure of 100 400 R in l had developed in 269 fluoroscoped patients as compared air, and the mean absorbed dose was estimated as with 8 m 260 patients who had not been fluoroscoped. 377 rad to this breast. De average dose to both breasts i. l was estimated as 247 rad per treatment. 174, it is of interest that Janower and Miettinen(68) A the occurrence of 3 breast cancers in 466 178. De breast cancer incidence in irradiated women ( ) report individuals who were irradiated for thymic enlargement started to exceed that expected on the basis of the during childhood or infancy. No estimate !s given for the control groups at 1215 years after irradiation. There. dose to breast tissue but that to the thymus was about after, new cases appeared annually in excess of the l 400 R and the antenor radiation field used in 92 per expectation in 0.27 per cent of exposed subjects. There i cent of cases was planned to extend down from the level was some suggestion of a decrease in latency with of the suprastemal notch, with a maximum cone increasing exposure, but this was of uncertain diameter of 12.7 cm. Two breast cancers were recordedsignificance. De time intervals until detection of 25 per i in 2604 umrradiated controls, the period of follow-up cent of excess cases (percentage of all cases diagnosed by (from irradiation to the date of the postal questionnaire) 30 years, determir ed by~ life. table methods), with 80% being 30 years. On the basis of the control senes, the confidence zones were: l probability of 3 or more breast cancers occurringin the Dostpowp kremi to M of cares exposed group would be 0.006-but this estimate is not /R/ /7'8"1 strictly valid s%ce the incidence of breast cancers is tested for statistical significance because it is raised. If 50 199 20.5 (17.7 22.0) breast tissue received the same exposure as the thymus, a 200 400 19.4 (13.0-21.6) risk estimate of (3 0.4)/(466 X 400), or 400.I100 33,9 (g,3 16.7) 14 (2-40) 10~' R, is obtained. De risk per rad cannot be derived with confidence, but is unlikely to be higher 179. The relationship between relative risk and mean by a factor of more than 1.25. exposure to both breasts remained closely consistent with linearity until a mean exposure of 234 R (group I 200 229 R). The relative risk at the highest exposure, at ] 401 R (group 300-550 R), however,was lower than that at 234 R, with a value about 0.57 (0.31 1.03, 80% r C. BREAST CANCER FOLLOWING confidence zone)of that expected if the linear regression RADIOTilERAPY TO TIIE BREAST had continued. This indication c,f a maximum yield of cancer at a mean absorbed dose in the region of 300 rad 175. Mettler et al. (9) l 3rted 13 confirmed cases of is not as clearly established in other human epidemio-lob cal findings (but see figuresi and II) but is >( breast cancer occurnns among 606 patients treated for i acute post.partum mastitis by radiation to the breast commonly demonstrable for cancers in animals (see Annex I). The dose.effect relationship for estimated area. The mean exposure of the irradiated breast was dose to single breasts (rather than mean dose to both 346 R. Only one breast was irradiated in many cases, y e 391

breasts) does not show as clear an initial linearity, but the relatise risk again reaches maximal values of 3.53 (at as the authors note, the effict cf radiation on inflamed and lactating breast might differ quantitatively from that 264 R) and 3.55 (at 452 R) and then falls to 2.29 (at on normal breast tissue. 733 R). [l 180. No excess of cancers was detected within 10 years 185. Another prolonged follow.tp study (11) has been C/ of irradiation, and life-table methods were used to derive made of patients who had had x-ray treatment of breast the su5 sequent annual risk of cancer induction until conditions, and in whom an increased incidence of 34 years from irradiation, the population (571 women) breast cancer was detected subsequently. Radiotherapy having been followed up over a mean period of was given to 1115 women in the period 19271957 et the Radiumhemmet in Stockholm for benign diseases of 25.2 years. De estimated annual rate in the whole the breast, mainly for fibroadenomatosis, but for acute exposed gr during this 24 year period was 8.310-* y,oup8 mastitis in 120 cases and for chronic mastitis in 49. In R where the value for the exposure is taken u the mean value for both breasts. In a separate most instances only one breast was irradiated, the mean analysis, the annual risk from irradiating one breast was dose to this breast being relatively high, about 845 rad. In 1023 of these patients, who were followed up for a estimated to be 5.0'10-* y-8R-8. Over 80 per cent of mean period of 31.5 years,115 cancers occurred in the the total number of cancers induced would be diagnosed only after 20 years following exposure. irradiated breasts, and 20 in unitradiated breasts, at periods of over 5 years after irradiation. From Swedish brent-cancer incidence data (available since 1970), the 181. The risk of cancerinduction was somewhat greater expected number in each breast having regard to the ages in those exposed at an age of over 30 than in those who f patients at risk would have been 19.9; or, since m youn g 14 per cent of patients had irradiation to both breasts, 28.7 per irradiated breast. Ap at expomre Number of induenon rate (years) ab/ects (10-* y-' R-'l 186. The excess in unitradiated breasts is not 15 29 397 7.3 significant. That in irradiated breasts corresponds to a >30 173 12.0 risk rate of (115 28.7)/(lI68 X 845) = 8710 rad-' per breast to 31.5 years or about 17510** rad-' per patient if both breasts had been irradiated. I 182. In the period of survey,7 of the 37 breast cancers l occurring in the exposed group had proved fatal, j as had 8 of 34 cancers in the comparison groups.The 187. Excluding 9 cancers which were detected within 6 time intervals between irradiation and death, and years of treatment, the mean latency from irradiation to i tween detection of the cancer and death, were not diagnosis was 23.6 years. The distribution of latencies j eported, but the figures indicate a mortality risk from (uncorrected for a presumably dimin,ishmg number of breast cancer within a mean penod of 25 years from penon years at risk duringlater periods) was shown by i exposure which is about one fifth the incidence risk the observation that 13 per cent of cancers developed at during the same period. This unusually low fraction of 614 years from irradiation, 38 per cent at 15-23 years, breast cancers proving to be fatal (see paragraph 18)is 34 per cent at 24-32 years and 15 per cent at over 32 likely to be due to the short average period of follow-up yeart The mean latent penod decreased somewLt with i since many of the cancers were only recently diagnosed, increasing dose; and to the fact that most of the women treated were in urency the younger age groups at the time of exposure. Dose group mean : SE trad) (years) 1 499 25.8 i 1.4 183. It is of interest that the risk per unit exposure 500-999 25.5 i 1.2 proved to be about equal following treatments given in 1 1 000 1 499 21.6 i 1.9 or 2 fractions (mean total exposure 143 R, risk 1 500 3 999 19.3 1.4 8.310-* y R*') or in 3 or more fractions (mean total exposure 235 R, risk 9.610 y R-8), suggesting that flowever, the mean admmutered dose increased increasing the number of fractions does not reduce the carcinogenic effect of the total dose given, although here markedly with age at the time of treatment (from the exposure per fraction does not appear to have 420 rad at age 10 29, to 770 rad at age 3049 and differed greatly in the two groups. The risks per unit 960 rad at age 50 or over), and it is uncertain whether dose are unlikely to be higher than those quoted per the mean period of follow.up varied with age at roentgen by factors of more than 1.1. exposure, or whether latency varies with this age perse. Either factor, or the increased death rate of older patients from other causes, might possibly be responsible 184. This study is of particular value in presenting for the apparent relationship between latency and dose. adequate controls for any increased incidence of breast A small apparent reduction in rnean latency with age at cancer following acute post partum mastitis per se, as exposure (from 26 years following irradiation at ages les: well as tendmg to exclude genetic or geographical bias, than 30, to 22 years for greater ages) may well be due to ce the two groups of patients with mastitis and their the confounding of dose size with age. ling co itrols were drawn from different parts of New ork State, it appears to determine the observed 188. The authors show that the numbers of cancers carcinogenic effect as being due to radiation, although, induced per rad and per breast vary markedly with age at 392

h r much cf the estimated breast dose applied bilat: rally,but [ exposure in a manner similar t2 that seen in cther rat:s for uniform inadiation (f both breasts might be cudies, with apparently higher incidence rates following somewhat higher. De dose estimates must hcwever be exposure at ages less than 30: 4e at

• Number Excess rate evidence as to the duration of the fluoroscopies.

f istion of Ob-Ex-per brasst ) breasts served pected - (M* rad'*) 193. In an ther study (para.165), dose estimates are an an ute 6 h n@n d 10 19 28 2 0.25 219 (18-757) 20 29 205 31 3.4 308 (212 429) (1-7) l0-s"y-rad-is suggested (with an integral 8 8 n mer e year Pesd 30 39 333 28 8.3 89 (52 136) 40 49 311 20 8.9 40(16 73) < present follow-up). The rates in this study should >50 119 7 3.1 34 (1 88) probably be increased by about 50 per cent to apply for All 996 88 24.0 90 (69-114) bilateral breast irradiation. Rese values relate to the total excess numbers of 194. Following radiotherapy, a survey (para.176 ff.) cancers detected, and indicate an increased sensitivity of with detailed control studies yielded an estimated annual the breast to cancer induction at younger ages, unless incidence rate of 8.310 y-' R-' for bilateral breast the periods of follow-up varied so markedly with age irradiation (implying a total incidence rate of that the reduction at greater ages merely reflects alower 21010~' R-' in the 25-year mean period of follow.up). ascertainment of induced tumours. This appears most Analysis based on single. breast exposure gave the improbable in view of the magnitude of the reduction in somewhat higher annual rate of 1010 y-8 R~'. The further study following irradiation for various benign rate. breast conditions (para.185) yielded a rate of about D.

SUMMARY

2.810-' y-' rad for single. breast exposure (corres-189. In summary, therefore, it is clear that breast pondmg' to a total risk for bilateral exposure of 17510 rad-' during the mean period of 31.5 years cancers may be induced with relatively high frequency by radiation, particularly in adolescence and early adult follow.up). life. De cancer is of the type arising initially from duct 195. There is thus a substantial discrepancy between cells, but is commonly found to be infiltrating breast the induction rates estimated for the Japanese tissue. Cases start to appear in excess of normal Populations and those for patients exposed to expectation within 10 years of irradiation, and new cases either diagnostic or therapeutic radiological procedures. continue to appear for over 30 years more, the mean Although surveyed over comparable periods of time, latency probably being in the region of 25 yean. 25 30 years after exposure, the Japanese incidence rates /3190. Three bases can be used for estimating the per unit kerma for those older than 10 at exposure, have been 1.910 y'* rad or about 2.510 y-8 rad-8 irradiation in women: (/ carcinogenic risk of breast namely, from the Japanese Ufe Span Study, from (absorb.d dose). The rates estimated following radialo. t 9 i> er re t a a ere' cerv. na retiewt== r a<e-81cai grecedures ran,e frem 2 te 10. with a <>,ure ef 0 therapy to tlie breast. From Hiroshima and Nagasaki (6 8)10 y-' rad being consistent with all of these mortality records, the risk rate per rad kerma for series. 19501972 was about 0.610** per year, or 1310-' 196. Mortality risk rates are estimated at about 0.25 times durmg the whole period by comparison with the 0 9 rad the incidence rates in the Japanese Ufe Span Study. In group, or 1910- as compared with Japanese National gg gg g 3 Statistics. The rates per rad for 19501974 as determined was 0.45 times the corresponding m.cidence rate, by regression analysis over all dose groups were 0.43 and whueas {ollowing one radiotherapy series it was 0.2 0.2610 y~' rad for liiroshima and Nagasaki, times the meidence rate. To a period of 25-30 years after c respectively,or 0.36 for the two cities combined. exposure therefore, approximate values for the total risk rates per unit absorbed dose (bilateral absorbed dose) to f 191. On incidence data, however, the mean annual rate l per unit kerma for 19501969 was I.910 y-' red-' adults would be of an incidence of about 5010 on for those aged 10 or over at exposure, of Japanese data, or 20010 on surveys following radiological procedures. The corre pondmg mortality 1.510-' y-' rad-' for the whole exposed population, risk rates within this period would be about 1010** giving an average incidence rate over the period of about from the Japanese studies,or about 6010~' rad-8 from 3010 rad *' for all ages. This incidence rate, of 3 or 4 radiological surveys. times the mortality rate in the same period, appears I consistent with a rather slow course and high cure rate 197. Several factors may contribute to the discrepancy for breast cancer of the types induced (see paragraph 18 between these sources of risk estimates. Firstly, the and reference 40). naturalincidence of breast cancer in Japan is somewhat icwer than in most other countries, with an average 192. The risk rates inferred from studies following annual incidence rate in females in 2 J:panese cancer l radiological proceduies, however, are considerably registries of 11 and 1410-8 y-8, as compared with a higher. After multiple fluoroscopic examinations, one study (para.168 ff.) gives an annual rate of mean value in 41 registries in other countries of 46 (SD 23) 10-s y '. This difference in incidence is due l p) 6.210** y-' rad for those patients survivmg 10 years after the examination, and about 4 (2 7) 10** y*' mainly to a lower incidence at older ages in Japan than l elsewhere. This is strikingly demonstrated by estimating ( for the whole series (with an incidence rate per the ratio between the meaa incidence rate for age 60 75 L rad'8 unit absorbed dose of about 110 (50190) 10-* rad to that for age 40 55. In the 2 Japanese registries, this durmg the mean follow-up of 27 years). It is likely that 393 8

ratio b less than 1 (0.59 and 0.69, and 0.64 for Japanese V. LUNG CANCER populations in Hawaii). In the 41 cther registries for which the ratio can be derived, the lowest value is 0.97 a and the mean is 1.48 (SD 0.24). It is not obvious, 200. In the 1972 report the Committee discussed however, how this striking difference would affect the estimates for lung cancer incidence in survivors of the 8 Onduction of tumours by radiation or their expression. A bomb explosions at Hiroshima and Nagasaki in uranium miners, in patients irradiated for ankylosing 198. It must also be recognized that the whole basisof spondylitis and in certain groups of tuberculous patients I the dose-and therefore the risk-estimation for the who were likely to have received repeated diagnostic I chest x rays. surveys following fluoroscopy, depends upon the opinion of physicians that the average time that they ? spent on pneumothorax fluoroscopies, which in most A. LUNG CANCER IN A BOMB SURVIVORS cases ceased to be performed many years previously, was 201. Further information has been published on 15 s. It is hard to feel sure that this estimate would be mortality from lung cancer in the JNIH-ABCC Life Span accurate, and easy to imagine that it could be Study (97). During the period 1950-1972,100 deaths systematically biased in one direction or the other. In addition, in the studies following radiotherapy, the occuned from malignant neoplasms of trachea, bronchus abnormality of the breast tissue clearly also represents and lung (referred to in this section as " lung cancer")in l another source of possible bias since, although it has the groups exposed to known doses in excess of 10 rad now been clearly shown that acute post partum mastitis (table 15). The numbers expected on the basis of the i mortality experience in the 0-9 rad group are 78.4, so b not followed by any increased breast cancer incidence, that the excess is 21.6 (6 40). However, it must be the inadiation of an infected and lactating gland could possibly be more carcinogenic than that of the normal noted, firstly, that the induction of any cancers in the gland. group exposed at 0-9 rad will reduce the estimate of risk for the higher dose groups, and secondly, that the small 199. It remains clear, however, that the breast is of number of deaths occurnng in this 0 9 rad group will,in relatively high susceptibility to radiation carcinogenesis, itself, increase the statistical uncertainty of the control i particularly when exposed during adolescence. rate and therefore of the estimate of excess in other groups. /9 ? ) TABLE 15. EXCESS MORTALITY FROM CANCERS OF THE TRACHEA. BRONCHUS AND LUNG IN HIROSHIMA AND NAG ASAKI BY DOSE GROUP,19501972 Males and females, all ages (Compared with o-9 rad group) Dose Excess rate Excess rete I group per umt per thousand kerma tred) Obsermt Expected Excess (20** red") (10 * *] persons HIROSHIMA 1049 49 34.4 14.6 (2.8 21.6) 70(13 137) 1.5 (0.3-2.9) t 50 99 13 9.0 4.0 (nes. l 2.3) 24 (nes. 75) 1.7 (nes. 5.3) 100 199 10 5.8 4.2 (nes. l1.3) 21 (nes. 56) 2.9 (nes.-7.7) >200 8 4.4 3.6 (nes.10.0) 8 (nes -2!) 2.8 (nes. 7.7) >100 18 10.2 7.8 (1.2 16.7) 12(2 26) > $0 31 19.2 11.8 (3.2 22.6) 14 (4 27) > 10 80 53.6 26.4 (10.4-44.6) 25(1042) NAGASAKI 10-49 7 12.4 -5.4 (nes. l.9) -78 (nes. 27) neg.(nes.0.5) 50 99 I 4.2 -3.2 (nes. 0.8) -41 (nes.10) neg.(nes.0.6) 100 199 7 4.0 3.2 (ntg. 9.4) 19 (neg. 57) 2.8 (nes. 8.4) >200 5 4.2 0.8 (nes.-6.5) 2 (nes.17) 0.7 (nes. 5.5) >100 12 8.2 3.8 (nes. II.6) 7 (nes. 21) > $0 13 12.4 0.8 (nes. 8.3) 1(neg.13) I (\\s) > 10 2o 24.8 -4.8 (nes.-6.6) - 7 (neg. 9) s r Source: Referente 97. Note: The 9oT, conndence hmits are indicated in parentheses. 394 r

I e d s 202. In Hiroshima,80 deaths occurred in those exposed TABLE 16. EXCESS MORTALITY FROM LUNG CANCER ~ PER UNIT OF t*!EIGl!TED ABSORBED DOSE,e i I at known doses cf 10 rad or more, as compared with HmOSHIMA. IslM2 4 H 53.6 expected on the basis of the 0-9 rad group, M21n and fernales, an agu indicating an excess of 26.4 (10.4-44.6). This excess "**"# * ""'"'" " "# "N [. rresponds to an induction rate.of ' 25(10 42) j Excess ute per undt weighted g

• • rad-'.(kerma). No estimates have been published absorbed dose (20** rad *')

j r the ratio of absorbed dose in lung to kerma for u,m,.3,o,oed dose /acrme retto o ) gamma and neutron radiation. Table 16, however, shows f,,"' f,",'"#,,",',',,,,, j, g,,,,,,, 3, g, } } that the rates per rad kerma do not differ considerably fr dj (10-*,.c ; depth stomach a a from those per rad weighted absorbed dose, using f weighting factors as in paragraph SS, and absorbed. >200

(nes.21) 7 (nes.-Is) 11 (nes. 28) dose / kerma ratios based either on the Committee's

$I$ Ih[$ Ilh ] assumption in its 1972 report of the lung as being at a > 10 25(1042) 11 (4 19) 24 (10-40) depth of 4 cm, or on the values quoted for a deep-sited organ such as the stomach (53).

• see parasraph 20a.

3 D 10~' red *' D 10'8 N Il V h 150-6- i: My (h 100-4-

2 N

q>--. O b N 50-2- b ~ -4>- I O) l il u ('v' J 100 250 300 red 100 200 300 rad } Figure II. Variation oflung cancer induction rate in throshima with T65 kerma h* Excess cances; as compared with 0 9 rad group Left diagram-excess deaths per milhon man rad Right diagram-excess deaths per thousand persons l IRanges indicate 90% confidence limits) {q

e 203. The number of deaths observed in Nagasaki are excluded, but the average rate over a wide dose range A}

too few for further detailed analysis. For liiroshima, appears to be in the region (10-25) 10-* rad-8 ikerma or however, the difference between the effectiveness per weighted dose). 3 unit kerma at low doses and at high doses is the same as 2) was noted in the 1972 report (table 15). Thus, the 204. Ishimaru et al (64) have examined the possibility mortality rate per rad falls from 70 (13137) l0-' rad that those surviving exposure at high dose might have 9 for those exposed at 1049 rad, to 8 (neg. 21) smoked more heavily than others,and that the apparent p 10-' rad-8 following exposures at over 200 rad (fig. II). association of an increased lung cancer mortality with i Similarly, the excess rate per thousand persons exposed radiation might be due to this. They found no evidence j does not continue to rise with increasing mean dose, to support this possibility,or that environmental factors although the wide confidence limits of all these values other than radiation were responsible for the increase. do not allow the assumption that a maximum mortality The smoking history was available in 204 subjects from 3 ; rate for these cancers has been reached at the mean Iliroshima and Nagasaki in whom a lung cancer was 1 absorbed doses (of about 200 rad) attained in the higher veril.ed at autopsy, 61 having been exposed at low y exposure groups. When the induction rate in the whole radiation dose and 13 at high dose. The relative risksof group of those exposed at over 10 rad is compared with lung cancer (in 1323 autopsies) were 8.6 and 3.0 in that for those exposed only at higher levels (table 16), smokers and non smokers who had been exposed at over a fG rates per rad thus still differ by a factor of 3 when 200 rad (T65 kerma) and 6.2 and 1.0 in smokers and 4; i td on kerma, or by a factor of about 2 when based non smokers exposed at less than I rad. The relative risk

• f

'N weighted absorbed dose, on the assumptions made in (relative to unitradiated nor> smokers and standardized for i paragraph 202. The likelihood of a falhng induction rate sex and age at death)was thus increased by about as much M with increasing mean absorbed dose is thus not by radiations whether the subject was a smoker or not. T 395

205. De histological charact:r c.f the " lung" cancers was cxamined in the same cries (24), which included cancer associated with the high radon concentrations in 1700 autopsies on subjects exposed at I rad or over the atmosphere cf many mines. It was stated, however, (with 127 at over 200 rad) and 1196 with exposures of that the incidence of or mortality from lung cancer in less than I rad. The prevalence rate of lung cancer at groups of miners could only be related to the radon concentrations to which they had been exposed in the ("3 autopsy was significantly increased in those who hsd case of those working in the uranium mines of the /~ ) been exposed at 200 rad or over, with a rate of 10.2 per cent at autopsy as compared with 5.1 per cent in those Colorado Plateau in the United States, where extensive exposed at less than I rad, and 5.1 per cent also in those studies had been made both of the frequency of lung who had not been in the cities at the time of the bomb. cancer and of the associated radon levels. Dree types of lung cancer were increased in this highly exposed group, but the increase was only significant for 209. Since the time of the last report, two further small-cell anaplastic tumours. The excess numbers, populations of miners with increased incidence of lung relative to expectation based on the <l rad group, are cancer have been described, and the increase related to however too small to exclude equal increases in other the corresponding radon levels. Each of the three major tumour types (table 17), particularly the epidermoid and studies shows a dose effect relationship between excess incidence of lung cancer and estimated exposure, the bronchogenic carcinomas, or even a proportionate data being consistent in each case with a linear increase of all types as seen in the control group. relationship between the, excess cancer and the TABLE 17. EXCESS NUMBER OF LUNG CANCERS AT cumulative working level months (WLM)s of exposure toradon. AUTOPSY (compared with expected numbers as judsed by types observed 210. It remains difficult to separate adequately the le the <3 rad group) contributory effects of radon and of smoking in causing

1. """' *"N the cancers, since there are too few non-smoking miners to form a sufficient control group, but Archer et al (5)

>200 1 199 attempt to make a correction for the cigarette-smoking Epidermoid 1.2 (-I.0-6.0) 7.6 (- 0.8 18.3) habits of miners. They find that, among smokers,15.5 splasue 3.7 (0.7-9.2) -1.8 (-7.3-6.1) deaths from respiratory cancer should hase occurred (among white underFround uranjura miners) as com-adeno <aremoma 2.0 (-0.6-7.2) -4.0 (-10.9-5.2) pared with 58 observed. In non smokers, 0.5 deaths Bronctuo alveolar adeno caremoma 0.0 (-0.6 2.4) 3.1 (-1.7-10.3) should have occurred as compared with 2 observed. i r ell caremoma 0.0 (-0.3 2.7) -0.9 (-3.1 3.9) While the proportionate increase in mortality from lung cancers is thus about the same in the two groups, they unclasufied 0.6 (-0.4-4.3) -0.3 (-3.3-5.2) emphasize that the sample size of non-smokers is too (^) Totat 7.5 (2.2-15.2) 3.7 (-10.8-20.4) small to be certain of this peint. Rey point out, however, that in the absence of any interaction between source: sererence 24. d**

• d Ci8 #* '#"8' * "OI #8 Note: The 907,conridence timits are indicated si parentheses accout b no mre dan a Opu-cent inmase a C

lung cancer rate, so that some interaction between the 206. De number of lung cancers induced, or at least effects f radiation and of cigarette seking appears the number causing deaths within 24 years of the pr bable and is bemg further investigated, exposure, appears to depend critically upon the age at exposure (14). For those who were aged 35 or over, the 211. $ eve et al (136) found that cigarette smokers i total number of deaths has been increasing about formed about 70per cent of a random sample of 700 linearly with time since about 10 years after exposure, uranium miners, and that this was equal to the that rate ofincrease being somewhat (about 50 per cent)frequency in the general male population in Czecho-faster in those who were over 50 than in those aged slovakia.Dey noted that there was no reason to suppose 35-49. Here has, however, been no increase in rate in that consumption of cigarettes correlated with radiation those who were aged less than 20, and little,if any,in e.posure otherwise than through age, and that the subjects aged 20 34 at exposure, comparison of lung cancer rates m miners with age standardized general vital statistics therefore permitted 207. The increased mortality in Hiroshima and Nagasaki the exclusion of smoking as a major causal factor in the over 1950-1972 is slightly, but not significantly, greater excess oflung cancers that they observed in miners, i in females than in rnales (mortahty per 1000 persons exposed at 10 rad or over, as compared with 0 9 rad 212. Archer et al (6) have now shown that the excess i gnup): rnortality from lung cancer increases about linearly with cumulative WLM in non smokers, but to an extent which Males 0.85 (neg. 2.27) (43% were aged 35 or over) is only of the order of one eighth that which is observed i Females 1.09(0.37-l.97) (38% were aged 35 or over) in miners smoking 20 cigarettes per day. B. LUNG CANCER IN WORKERS 213. At least in regard to miners who smoke,however, i EXPOSED TO IllGH RADON LEVELS there are now three estimates of the lung-cancer risk rate

68. In its 1972 report, the Comnuttee discussed the

• • Workmg level" as defined as any combmation of II

'h "'h"d d " d'"8h"' P' d"' ' " evidence obtained from studies on uranium and otherresult in the ulumate emisuon of I.'310, Me% of potential s!pha "' h '" ' r that will hard rock miners, of an increased incidence of lung enersy, workms level month (wLM) is the esposure to one e workms level dunns l70 hours. [' 396 b L

} for Cfferent exposure levels from radon and two further associating radon measurements at cert'in positions in estimates are available from cther mining groups, giving mines with alpha radiation dose to cells cf the bronchial a mean risk for a stated averop exposure, epithelium, as well as possible differences in smoking habits, urban or rural residence, typical duration of work l. 214. For uranium mines of the Colorado Plateau, the in mining, age at starting and other circumstances. s dose.effect relationship is consistent with a linear Moreover, it has been shown that differences in risk may wensanon through zero. The total experience amongst apply to those with long or shon total periods of white underground workers in uranium and other hard exposure (135). To the precision required for risk estimation. however, a rate of 510 lung cancers per rock mines, from July 1950 to September 1968, showed an excess of 58.3 deaths from lung cancer in a million per year and per W1.M appears representative of available data. population equivalent to 37 958 person-years. Assuming a mean exposure at the midpoints of the various o 218. A study of lung cancer in Ontario uranium miners exposure groups with stated limits, and a mean exposure L l cf 5000 WLM in the group exposed at over 3720 WLM, in the period 19551974 (50) records a death rate from the mean exposure would be of 740W1.M, implying an this malignancy which rises from about 0.3 per cent in excess mortality of 3810-* per WLM durias the period the unexposed subjects to 3.7 per cent in those with cf observation. Average annual rates are about 180WLM of cumulated exposure. The authors empha. size, however, that these data cannot properly be used 2.910 per WLM year for those who have worked as miners for over 10 years, or 2.1 10 per WLM year for for risk estimation, owing to the limited total periods of wrvey (of within 20 years from the start of exposure) the whole group. and for lack of information on smoking habits. He Wrwy is f particular value, however, for showing 215. Studies from uranium mines in Czechoslovakia c nsistency f the done effect mlati nship at low dose l showed an excess rate corresponding to 17010 deaths w th a knear regression which is taken to exclude any from lung cancer per WLM over an exposure period 'h" 8"** **"

• II 'I'* '"E**'" 'h*

varying from 19 to 23 years (135), indicating an annual although the absolute excess oflung cancers was lower rate of about 810-' per WLM year. Most of these in those under 35 than over 35 at entry to mining the expmuru neeMd pior to @mm d M 8t** CM*5e was e mPanW in the two groups. ventilation of mines in about 1958. A further report on a group of these nuners who started work in the uranium mines between 1948 and 1952 and have now been 219. Several steps are involved in deriving the total risk followed for 2126 years (136) shows an excess rate of of lung cancer from estimates of induction per WLM. 23010 per WLM, the dose.effect relationship again Since lung tumours may develop with a long latency after the relevant exposure, the annual rates of incidence being consistent with a linear regression through zero. must be regarded as continuing over long periods of l The rate varied with age: time, and the mean rates used for calculation of total g,,,,,, mortality must be appropriate for the average of these E h eststser and 93 confidence timits t} ofminuqr (10** WLM*') periods of time. In the case of the least two of the estimates quoted above, this condition appears to be 9"' ' *** 39 1 ut" c n"PW to awnga om pWs M Weanor 3 Over 40 370 (280 460) more. If this is regarded as comparable to the median All 230 (155 305) latency for tumour development and diagnosis. and the 4 total of cancers is doubled to allow for later developing t indicating rates per millior. per WLM year of about 6,10 malignancies, the final figure (of about 45010-' per l and 16 in the three age groups. De authors suggest that WLM) should approximate the total expression of the these rates may be higher than those inferred for the radsation effect. Colorado miners because of a younger mean age of the latter at the start of mining. Dey emphasize that their 220. For the Swedish data also, although mean own estimates of exposure rely on lary numbers (over exposure times are not stated, the " latencies" observed 120 000) of measurements of radon concentrations in from the start of mining to the diagnosis of the tumours mines in the period since 1948, with at least 100 observed, which range up to 40 years and average deternunations per year in each mine, and that the 26 years, show that the rates derive from populations additional hard rock exposure of the miners was very with prolonged exposure. The annual rate might thus be small multiplMd by 40(years) to allow for full expression. I* giving a total in the region of 14010-* per WLM. 216. In non. uranium mines in Sweden the excess mortality shows a linear regression,with slope equivalent 221. He values quoted for miners in the Colorado to 3.4 10-* excess deaths from lung cancer per Plateau represent rates averaged over somewhat shorter WLM year. For fluorspar mines in Newfoundland the exposures since, for example, the data of Archer et al. mean rate is estimated as 2.210-' perWLM year,while (5) refer to 3366 miners and 37 958 person. years of for iron mines in the United Kingdom it is 6.010 per exposure, giving an average exposure of about 10 years. WLM year (144). From information given (81) on the excess deaths and person. years at risk in the various exposure groups, it 217. These estimates may be regarded as reasonably appears that the risk for a given exposure r'ses with the consistent in view of the difficulties and likely mean penod since the start of uranium mining.as shown differences m ascertainment of lung cancer, and of in table 18. While the mean rate for all periods since 397 El ~.

TABLE I8. MORTALITY IN URANIUM MINING Perted ssner start af mining ryears) <3 >3 >10 - >t3 >20 >23 em Excess deaths from ( ) respiratory cancer 0.6 8.6 20.6 17.6 5.3 5.7 (,) Person year WLM product (10') 3.4 7.8 7.7 4.3 2.4 2.6 Excess deaths per 10* person year WLM 0.2 t.1 2.7 4.0 2.2 2.2 90% confidence linuts 0.01.4 0.4-2.0 1.74.0 2.6-6.2 0.8-4.6 0.8-4.6 Source:Rererence 81. start of mining is 58.3/28.3 = 2.06 deaths per 10' other miners. There are, however, a number of factors person year WDi, that for miners who have worked for that might explain this discrepancy: 10 years or rnore is 49.1/17.1 =2.86 per 10' person year Wul. Or, if the rates for the various intervals are (a) The rate per unit kern.2 is not, and that per used to estimate the total of ail deaths, say to 30 unit absorbed dose (weighted according to data on from start of exposure, a value of about 7010 years leukaemia induction) may not be, the relevant basis for per Wul is obtained. It is not clear, however, whether the risk estimation. However, even expressing the rates as apparent fall in rate after 20 years of exposure is per unit absorbed dose (with RBE= 1) would only significant, or what contribution would be made at neresse the Japanese estimates by about 75 per cent; periods longer than 30 years. fb) The Himshima population was of all ages, i and the carcinogenic effect appears to have been 222. For a full, e.g. 40. year, expression of the total expressed predominantly in those aged over 35 at carcinogenic effect on lung tissue of radon and of its exposure. Since 40per cent of these populations (exposed at over 10 rad) were over that age, the observed daughter products. therefore, an incidence of 200 450 10-' per WDi can be regarded as probable. rates should probably be multiplied by a factor of up to Since 1 WI,M has been estimated as delivering about 2.5 to correspond to the rates observed in uranium I miners; i rad of alpha radiation to bronchial epithelium i (Annex B),the risk of exposure to low.LET radiation (c) The Hiroshima rates are based on informa-would on this basis be in the region of 50 (or from 20 t tion in the period from 5 years to 27 or 29 years from / 1150)10

• rad, if a weighting factor of 510is assumed exposure and, since deaths continue to occur in the (jfor alpha radiation at the high dose levels on which the exposed groups, should be increased (perhaps by up to estimates are based. If estimated in terms of mean lung 0per cent) to emespond to de % ear peM dose, with 0.5 rad per WDi. the risk per rad would be c nsidered for uranium miners; twice as high, but this criterion seems less appropriate (d) It has been noted (13) that many deaths since the cancers are likely to arise from bronchial i

epithelium. It must be recognized, however, that these from lung cancer are not identified as such on death i estimates depend on very indirect indications of tissue certificates and that adjustment for such errors would dose, as derived from air concentrations at monitoring raise the rates in Hiroshima by about 50 per cent; Ctes in mines and strictly may apply only to miners who (e) The induction rate estimated for those smoke. If smoking has a co. carcinogenic and not simply exposed at 1049 rad in Hiroshima was an additive caremogeme effect the same increas',.umot 70(13 137)10 per rad kerma (table 15), and would I necessari!y be assumed for non. smokers. Uncertamties as be equivalent to about 120(23-240)10 per rad t) the lung tissue over which dose should be averaged, unweighted absorbed dose. This rate is much greater the relationship between radon concentration and tissue then those estimated at higher dose but has wide dose, and the value of RBE appropriate for lung confidence limits. It is improbable, however, that this carcinogenesis from moderate doses of alpha radiation rate in Piroshima occurs at a level which corresponds to add to the uncertainty of the estimate (ref.51 and those on which the uranium miners mortality are based, para. 215). since the former was associated with a 0.2.per. cent excess incidence of lung cancer (fig. II), while in the studies of seve et al (136), the data are consistent with I the higher incidence rate per WDI up to excess C.

SUMMARY

m rtruties of 10 per cent; (f) It might be questioned whether, for the lung I. Comparison of estimates as for the breast, spontaneous cancer rates in Japan are unusually low, and that a lower induction rate might be 223. There is an apparent discrepancy between the low in some way associated. In females the annual rate of lung cancer registration in two Japanese cancer registries an induction rates in Iliroshima and Nagasaki, was 6.0 and 5.2 cases per 100 000, as compared with a 2510 per rad of kerma or of weighted absorbed mean rate of 7.615.5(SD) in 54 other registries with ose, and the higher values, 40 180 10 per rad, that have been inferred from the mortality in uranium and comparable critena. The annual rate in males was somewhat lower than that in many other countries, with 398 ( g

? L" " age standardized rates in these mjstries cf.15.6 (in 225. In view cf the unusually short and fatal course of 19621964) and 15.3 (in 1966), as compend with lung cancer, particularly of the ansplastsc type L '35.5 t 19.4 per 100 000 at comparable periods in the (pers.205),' the excess incidence of this disease is . other registries,.but Japanese rates have been noted unlikely to differ materially from the mortality. The incidence rates in Hiroshima and Nagasakiin the period 1 (para.41) to be increasing; 1959 1970 corresponded to 1310-* and 1210~' per (g) Islumaru et al (64) record a history of rad kerma mspectively. The corresponding rates per unit . igarette smolung in 52per ant of a series of 1323 absorbed dose of low.LET radiation would be about c people from Hiroshuna and Nagasaki (males and 2010-* rad in Nagasaki, and 10,16 or 2510 rad-' females) from whom a smoking histwy was availaWe, in Hiroshinu, according to whether an RBE of 20,10 or l whoseas invc records 70 per cent of a random sample of 1 is used for the high.LET component of the absorbed miners as being cigarette smokers. The difference, do e. presumably due in part to the proportion of women iri the Japanese populations, might account for some of G" 226. From the rather discordant souras ofinformation l discrepancy in induction rates, but the data of Moris ama available, the risk of lung cancer induction by bw LET .and Kato (97) do not indicate a clear difference in radiation would. appar to be in the agion of . induction rates in the two sexes in Hiroshima; 5010** rad ** for exposure at sys greater than about (h) On the other hand, estimates of the rad 35, and presumably about half this rate as the average risk for exposure of a whole population of all aps. l. equivalent of the WLM at the tissues of interest haw Further guidance as to this estimate should result from l varied rather widely and the estimate of 0.5 rad mean determination of a mean lung does in the irradiated l~ lung dose may be too low, or if the dose to bronchial spondyhtics, since lung cancer incidence is clearly raised i-epithelium (para. 222) were the appropriate value to use, h these patients but not in unirradiated patients with the risk estimates for miners would be halved; tha disease (120). !I-(I) Finally, the values of 5 and 10 taken for l RBE (in paragraph 222) for calculation of mortality per l rad of low-LET radiati6n in miners may be unduly low I at the dose levels involved and an RBE of 20 would i. bring these estimates into the range of (20 50)10-*. VI. BONETUMOURS However, since these dose levels were associated with high incidences of lung cancer, this appears improbable. 227. In its Ib72 report, the Committee referred to the absence of evidence of induction of bone cancer in 2 survivon at Hiroshima and Nagasaki, where there l

2. Conclusions appeared to be no excess of diagnoses to 1%5 (165).

There is still no good basis for estimating the risk of 224. Apart from the findings in Hiroshima and Nagasaki bone cancer induction by external radiation, except that and in miners, no other sources of information are results of scalp irradiation for ringworm suggest that it is available to confirm the higher or lower value of the small in children. Modan et al. (92) record 2 cancers of induction rate. In its 1972 report,the Committee noted bones of the head (with about 0.5 expected)in 10 902 the excess of 41.8 deaths from lung cancer among children. Shore et al. (139) record 4 tumours of scalp and 13 940 patients treated for ankylosing spondylitis-jaw,1 of which was malignant (138),in 2215 irradiated (within a mean period of 13 years). A further study of children with none expected. Assuming that the skull such patients treated with only a single course of forms about 25 per' cent of the totalbone massin children radiotherapy has shown an excess mortality from lung (reference 62, quoting 40 per cent at birth and 15 per cent cancer in 0.26 per cent of 14109 patients followed for a in the adult) and from the dose estimates given for each period of 9.5 years from the time of the survey (52,131,159) risks per unit absorbed dose for mean treatment. The exceu mortality from lung cancer was bone cancer of 3 (neg 10) 10-' rad and 0.46 per cent during the period starting 6 years after the 5(0.2 21)10** rad *' can be derived. In the recent treatment in 6838 patients (143) who were followed for survey of 14109 patients irradiated for ankylosing a further mean period of 11.3 years from this time (36). spondylitis and followed for a mean period of a.5 years l No risk estimates can, however, be derived with (36), 4 deaths occurred from malignancies arising in confidence, since the mean lung dose will depend heavily irradiated bone areas. Since only 1.3 such deaths critically upon the position and size of the radiotherapy were expected, this observation indicates a significant ' fields, and is at present under investigation.- The induction of bone tumours in the adult. No estimate is 3' Committee's 1972 report referred to a tentative estimate yet available for the mean dose to bone from these j of bronchial dose by Dolphin and Marley (39) of 80 red, treatments (as distinct from the mean bone marrow dose '(- derived by inference from the mean spinal marrow dose of 321 rad) so no risk estimate can be derived; and it of 880 rad. This estimate, if applicable to patients who could only be very approximate in view of the small l had a single course only with estimated mean total excess number of cancers. Moreover, the distribution of marrow dose of 321 rad, would imply a ri*.k for the " spontaneous" osteosarcomas in man and probably also mean period 617 years of 0.0046/80 er 6010** rad-8 that of radiation. induced sarcomas, varies in different The mean lung or bronchiaf dose will. however, depend bones, and the long bones of the limbs will have been critically upon the position and size of the fields used in little exposed in the spondylitis treatments. The excess i treatment, and must await a more accurate deter. mortality, of only 0.02 per cent of patients however, mination. sugests a low induction rate at least in the spine. 399 I

228. The Committee also reported on the incidence cf cumulative mean bone doses cf 700 rad or less. It was osteosarcomas following intravenous therapeutic ad-pointed cut that the number of people studied at doses ministrations cf Peteosthor, a preparation containing lower than 700 rad corresponded only to 4.610* man 28 Ra. The risk estimates for bone from this source have been extended in a valuable way by Spiess and rad and that only 1.8 sarcomas would be expected in this group, as judged from the incidence at higher doses, I sys (146) who have identified the difference in risks ifincidence were linearly proportional to dose. r those who were juveniles (younger than 20) or adults 92 the time of treatment. They also show that the risk appeared to be higher from a given total dose if the 232. Meanwhile, however, Rowland et al. (127) have p-administrations were given over a time span of a year or noted that the incidence of osteosarcomas in the groups more than if given during a short period, i.e. of a few studied is better represented by the relationship in which la months only. By June 1970, the average time since incidence is proportional to the square of the dose than patients had received their first injections had been to the dose itself, each relationship involving also an 4 about 22 years for juveniles and 19 for the adults, and exponential term to correspond with the decreased sarcomas had developed in 35 of the 208 juveniles for incidence at higher doses. The former relationship would whom dose and injection span were known, and in 12 of imply an even lower expectation of sarcoma at low 607 adults (who had in general received lower doses). doses. For example, according to the equation given in paragraph 233, the expectation in the group at doses 229. For juveniles, the estimated risk rises from about lower than 700 rad would be only 0.3 sarcorras. 4010** rad-' (mean bone dose) for injections given during a short span to about 22010~' rad ~' forlonger 233. The best fit for incidence of sarcomas (/= frac-time spans. For aduits, the risk rises from 30 to tional incidence) was to the equation 17010-* rad-'. In each case the data are consistent with the risk E(deaths per million per rad) rising from 1 - 3.9 10-

• D c-8 '" '"

2 the lower to the higher value with an interval diminishing with a half period of about 8 months: where D is the total rad dose to bone from exposure to diagnosis of a tumour. The incidence of carcinomas (of paranasal sinuses and mastoid air cells) was adequately - Forjuveniles, E = 40 + 180 (1 - c-0 o'") fitted by For adults, E = 30 + 140 (I - c - 0 O'*) l=3.1 10-S De-a i.24 io, where m is the time span in months during which the ( irgections were given (146). 234. It is of interest that sarcomas have occurred A o

2. Two points need emphasis with regard to these following mean bone doses as low as O rad from 8 8 ' Ra.

kJ data. Firstly, since the dose is of alpha radiation (see but only at doses above ll60 rad from 8 8

• Ra. However,

[ l Annex 1, paragraph 269), the appropriate risk estimates when these hmits are expressed as dose to endosteal cells for bone from radiation oflow LET might be lower by a rather than as rnean bone doses, this discrepancy fact r of 5 20. Secoadly, there is now ample evidence disappears since sarcomas are seen at endosteal dose's that osteosarcomas arise predominantly from endosteal above 810 rad from as*Ra and above 760 rad from I cells and that the relevant dose for sarcoma risk s a + Ra (127). estimation is therefore that to these cells, which lie at a distance of up to 10pm from bone surface, rather than 135. In a review of 261 cases of bone tumour which the mean dose averaged through bone,rs used in the risk were regarded as having been induced by therapeutic estimates quoted. Radium.224 has a short half. life of radiation, Yoshizawa et al. (169) note that 50 per cent 3.64 days and its radiation and that of its daughter

,f the induced malignancies were described as products is largely delivered while these radionuclides osteosarcomas, 25 per cent as fibrosarcomas and 7 per are tillp' resent on bone surface. This contrasts with that as chrondrosarcomas. A further 8.5 per cent cent from 8 Ra, which becomes distributed throughout included spindle cell sarcomas and mixed polymorphic l

bone dunng its r:riod of radioactive decay. The dose to cell sarcomas. cndosteal cells from 88* Ra is about 9 times that as averaged throughout bone, whereas that from 8 8

• Ra is

\\ about two thirds the mean bone value (127). The l appropriate risk estimates for sarcoma induction by

SUMMARY

i irradiation of endosgal cells from 8 8* Ra in juveniles are thus about 25 10 rad (of alpha radiation) t 236. The risk of inducing malignancies of bone by i low.LET radiation thus a endosteal cells for long spans of injection, and region of (2 5)10** rad'ppears to be low, and in the to endosteal cells (table 19) I 0 10 rad

• m adults. An approximate figure for prolonged bone irradiation from low.LET radiation for absorbed doses of a few hundred to 1500 rad in woul locally irradiated bone. Risk estimates for alpha rad,d thus probably be in the region of (150)10*

radiation at doses of a few hundred to many thousand l rads (table 19) are consistent with a risk of this order f p) in its 1972 report, the Committee discussed theassuming an RBE of about 10. Spiess and Mays found l induction rates per unit absorbed dose to be about equal ka available from groups of people with as*Ra m males and females, and about 30per cent higher in } i burdens and the lack of bone sarcomas at estimated Juveniles than in adults. 400 0 0

[ TABLE 19. BONE.TUtsOUR INCIDENCE RISK ESTIMATES M i f h=' dose (sedl A w,f g W**** ngene benese Resist (20'* *'d l 4 e 1 92 GiGdson About 18 about 1008 to1500 3 (nasa 0) low LET t39 Ould8" N about t008 - to 800 $(0.2-2t) low LET About 23

• Ra I

2 juve.nele.s ,M, high LET 146

e. e uits a

die,4$e ^ Admits About 35

• Ra abent 2 m mMM 85 e t 60' high LET a

n'% 2s per cent or bone to be irredissed at a sneen does or 40s red. 6 ased on quadratic formule (pare. 233). i 3 8assed on kneer rormula (pare. 231). ossen eens VH. OTHER CANCERS se - i t 7 237. While some very approximate indication can be 'given for the risk of cancer induction in certam other 8 - organs by radiation, the total carcinogenic risk cannot i . yet be deriwd by summing the individual contributions 9 e j to this risk from every organ. A clearer guide to the size I 1 i of the total risk is obtained by estimating the total l mortality from all forms of mahomant disease followag effectively whole-body radiation at known does lowls, Q lP i for example in the exposed populations of Hiroshima 4,. l -i >-' and Nagasaki, and comparing this with the expected noenber of such cancers. With reservations owing to the 4 l partial-cody nature of exposure, some estimate can also I be derived from patients treated for ankylosing te l spondylitis. In each case, correction has to be applied to allow for tumours that may become diagnosable after $g the s.ailable penod of follow-up, although in some o. studes, for example of the effects of foetalirradiation, g-the full number of tumours appears to how been .-d>d i nached. Wh i 238. In the case of mahanancies resulting from ,ess iss. ,m occupational exposum of @, M W ,,,,,,Ly,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, cannot be adopted because the exposums are not osier wah assie since exposine known. How it is possible, howewr, to moord the total g excess of deaths from malignant disease other than JNIH-ABCC surwy leukaemia, and to compare this total with the excess dus tsuksenile: open circles and hatched colusans to leukaemia. The ratio is of value, firstly because long An other n. += risied circles and dashed columns penods of follow up have occurred since the exposures (Renses ind6cate 90s coendence lunho) that are likely to have been relevant, and secondly, because the radiation induction of leukaemia can be estimated from other sources, so that that for other 240. The further increases in the period (of 21 months) malipant diseases can be dertwd. between ti.e 1950 1972 and the 1950 1974 estimates (see table below) are consistent with those indications, l ll-239. From the JN1H AllCC 1.ife Span Study, the neither the increase in leukaemia nor indeed that in )- mortality rates are shown in figure ill for successive other malignancies being much (or sipificantly) greater j periods, ; both for leukaemia and for all other than that expected on the basis of the 0 9 rad group.(if 9. malignancies. There is a strong suggestion that the excess this expectation is estimated simply in proportion to j mortality from leukaemia is now ceasing, and a possible person years at risk, but without any correction for age !, \\ indication that the excess rate from other malignant and ax distribution, the 1972 1974 increases for disease sney have reached its peak. This would,if so,be leukaemia and other malignancies became 3.7 and 145.3, respect.wly. The corresponding excesses of the >10 rsd consistent with the indications noted above that the group over the 0 9 rad group are then 6 - 3.7 = 2.3 and needian latency for diagnoens of radiation-induced cancer 156 - 145.3 = 10.7). may be about 25 years. i 401 n

---

---..~

(' HiroshimaamtNorsaki Males and females, all ages Dosegroup(red): 49 .{v) 19301972 193SI974 1972197a _ _ >10 193S1974 193&l974 - 1972 1974 Person-years (10 ) 3 1 092 1166 74 476 522 46 Deaths from: Leukaemia 48 54 6 84 90 6 Other malignancies ~ 2 232 2 467 235 1 075 1 075 156 Ratio 241. FigureIV shows the ratio of excess deaths from other malignancies to those from leukaemia. The upper 3 - Ain other malienancies curve, obtained by comparison with Japanese National ^83 8"*****.* Mortality Statistics, may be the more valid,in the sense that some deaths attributable to malignant disease may have occurred even in the group exposed at only 0 9 rad. For example, the mortality ratio for leukaemia was 1.5 even in this group. On the other hand, the 0 9 rad group 2 is much more comparable with those exposed in the Y methods of study and ascertainment, as well as in place ofresidence. [ ( 242. It is probable that this ratio of mortality rates may O differ for those exposed in different dose groups, since ~ the form of the dose.effect relationship is unlikely to be O 4 similu for all forms of malignancy. The precision with / /0 / which the excess number of deaths is determined in ,y lower dosage groups, particularly for solid cancers, v ~ a however,is insufficient for this estimation to be made a 01 O adequately. i i 1945 1955 1965 1975 243. In a further report of the life Span Study Figure IV. Ratio of numbers of deaths from all otaur recordmg mortality data from October 1950 to Septem-motiansacios othee than inheemia to those rrom ievkaemie her 1974 (14), the excess mortality rates are estimated Rataos are of all excess deaths during the period from 1950 until from the slope of the imest regression of deaths per ! t d*'" indic*t'd person year on estimated dose (T65 kerma). On this upper curve (filled circles): escess deaths as cornpared withbasis, the relationship between the excess of solid g,,,,,,, gop,#$,(',",,",$,3***$,**,, compared with cancers to that of leukaemia it similar to that derived 4,, frequencies observed in persons exposed at 0 9 rad (fig.IV and table 20) by comparison with the rates observed in th 0 9 rad group, as follows: Periodserveyed 19301934 1934-1938 1938 1962 I962 1966 1966 1970 197&1974 Excess rate (10-dy-' rad-8): leuksemia 4.06 2.28 1.78 0.88 1.13 0.49 Other malignancies 1.61 - 0.38 2.35 2.99 3.32 4.19 Cumulative total, October 1950 to September of final year of period: leukaemia p) Other malignancies 4.06 6.34 8.12 9.00 10.13 10.62 I.61 1.23 3.58 6.57 9.89 14.08 Ratio of cumulative totals, other malignancies / leukaemia 0.40 0.19 0.44

• 0.73 0.98 1.33 4

402 e

~'~ s. < amen. __ - =, - _. m ( ) / ) \\' %/ j EXCESS MORTALITY FROM LEUK AEMIA AND FROM ALLOTilER CANCERS IN lilROSillMA AND NAGASAKI BY PERIOD TABLE 20. Dme groups 10->200 rad T65 Males and females, all ages ALL OTIIER CANCERS Ratio LEUKAEMIA other cancers} Duration year Ob-Ex-Excess rete Ob-En-Excess rate leukaemia Aten red Prried tyrers) (a 10*] u. J pected Excess (10

• y** red)

served pected Excess (20'* y red) at end of period Gmspared with 69 radgroup 1954 I954 4.25 8.24 24 4.5 19.5 (11.8-29.61 2.37(I 43 360) 146 126.8 19.2 (nes -44.3) 2.3) (nes.-5.36) 0.99 (nes.-2.22) 1955-1959 5 9.26 27 7.0 20.0 (11.6 30.7) 2.16(1.25-3.31) 193 209.0 - 16 (nes.-13.6) - 1.73 (nes.-l.47) 0.08 (nes.-l.00) 1960 1964 5 8.77 16 2.2 13.8 (7.6-22.3) I 57(0.86-2.54 258 223.0 35.0 (4.5-67.7) 3.99(0.51 7.71) 0.72 (nes.-l.61) 1965-1969 5 8.25 4 4.8 9.2(3.3-17.5) 1.1240.40-2.13) 301 268.1 32.9 (neg48.1) 3.99 (nes.-8.26) 1.12(0.14-2.096 1970 1972 3 4.68 3 27 0.3 (~g.-5 6) 0.06 (nes.-l.20) 177 162.0 15.0 (nes.-42.4) 3.2I (nes.-9.06) 1.37(0.30 2.44) Compared wtth Japanese NationalStatistics 1.7 22.3(14.8-32.1) 2.71 (f.80 3.90) 128.4 17.6 (nes.-39.1) 2.14 (nes.-4.75) 0.79 (nes.-l.73) 2.8 24.3 (l6.3-34.4) 2.62 (l.76-3.7 I) 184.0 9.0 (nes.-33.5) 0.97 (nes.-3.62) 0.57 (nes.-l.24) 1950 1954 1960-1964 as above as above as above 3.3 12.7 (6.7-21.0) 1.45(0 76 2.39) as above 215.2 42.8(17.0-70.8) 4.88(l.944.07) 1.17 (0.43-l.90) 1955 1959 3.7 10.3(4.8-18.2) 1.25(0.58-2.21) 238.4 62.6(34.7 92.7) 7.59 (4.21-11.24) 1.90(1.09-2.71) 2.3 0.7 (nes.-5.51 0.15 (nes.-l.ls) 153.0 24.0 (2.7-47.5) 5.13(0.58-10.15) 2.22(1.31-3.13) 1 % 5-1969 1970-1972 Source: Refesence 91. Note: The 90% confidence tiraits are indicated in parentheses. b m -Q E-- ,~.-' -.;, ,_ d,;,.,,,,,,,,,, _ a _ ,. ~..,,,,..

7 244. It has been pointed out (para.63) that the frequency with which leukaemia b induced by radiation current Life Span Study data, therefore, show that the varies with age at the time of exposure, with a moderate ratio of deaths from other malignancies to those from

w%.

,3 ' qp rate of induction at age 0 9 years, a lower rate for ages leukaemia, as occurring in the total period from 5 to I A 1019 and higher rates at greater ages. It is important to 29 years from exposure, differs for different ages at y (c 4 tt ) note that the induction of other malignancies as a whole exposure (see table below). He ratio discussed in 4 ta ( paragraph 242 and refened to in the following 3I.\\ V varies in a similar manner with age, and that the relative paragraphs will therefore depend on the age structure of gg increase at older ages is greater than for leukaemia.De the population exposed. 4 e Hiroshima andNagasaki, 1950 1974 Male and female s Age at exposwe(yeen) l c Mortality per unit kerma 69 W 2A34 3549 > 30 Au (10" rad-8): g Leukaemia 76 23 54 44 82 46 r (66 85) (17 29) (38-53) (34-53) -(66 99). (42 50) Other malignancies 33 20 59 74 153 53 Ratio, other malignancies / leukaemia 0.4 0.9 1.1 1.7 1.9 1.2 1 245. He findings of Doll and Smith (36) on the irradiation of ankylosing spondylitis patients, show a below about 20. Their values for excess risk at older similar relationship between the induction rate of ages, however, show a clear increasing trend with age for cancers, and probably also of leukaemia, and the age at cancers of these heavily inadia'ted sites (x! trend = 15.1) exposure. Heir data refer only to cancers of certain and for leukaemia (x trend = 6 7). He excess rates for s organs, essentially those in the heavily irradiated parts of cancen at >6 years after treatment and for leukaemia at g > 2 years are as follows: the body; relatively few treatments were given at ages 6 I Age er treatment (yean) s t <23 534 3544 4534 >33 AU ( Excess rate (10" y-'): Cancers 14 34 127 182 339 91 i Leukaemia 4 13 19 43 52 90 l Ratio, cancers / leukaemia 4 3 7 4 6 47 6 246. If it is assumed that the median latency for deaths from malignant disease other than leukaemia is 25 years,mean time of 13 yean and to 3.7 at 25 years from the I and that the mortality from leukaemia has ceased by this time of inadiation. He ratio based on numbers of l time, then the final expected ratio of excess deaths from deaths, however, is subject to bias owing to the l cancers to those from leukaemia should finally reach decreasing number of patient. years at risk at the longer l 6 twice the value reached by 25 years, or about 3 on the intervals after exposure, when leukaemia deaths have l t basis of the Japanese data (fig. IV). ceased to occur but when deaths from induced cancer continue. If allowance is made for this (by using the 247. His estimate of the ratio can be compared with nuo of cumulant nto nthu than ohumulant numben i i l the value 4.3 derivable from the mortality studies on f deaths), the ratio has a value of 3.6 at 13 years and American radiologists, among whom an estimated excess at 25 yean. Mowing pehic x inadiadon for of 11.3 deaths from leukaemia occuned during the metr pathis haemorrhagica, Smith and Doll (142) have period 1935 1958, as compared with an excess of f und a ratio of excess deaths from cancers to those 48.2 deaths from all other cancers (134). For ankylosing fr m leukaemia of 3.9 (to a mean of 19 years from spondylitis, it is difficult to estimate the proportion of single courses of irradiation). liere again, however, the organs sensitive to radiation carcinogenesis which lay rati will depend on the proportion of marrow and of within the radiation fields, as compared with the rgans sensitive to radiat on caremogenesis within the proportion of manow in those fields. From the pelvis. In the light of these values, however, and I publication of Court Brown and Doll in 1965 (29), particularly of the ratios derived from effectively l however, the ratios of excess deaths from all other whole. body irradiation, it appears probable that the malignancies to those from leukaemia were mate total number of deaths from all malignant u 3.1 i 1.1 (S.E.) for patients completely followed up at U"" I"" '""'8 *I

• time and 5.511.4 for those with incomplete I

ow.up. In the further report on patients who had ceived only one course of treatment (36) and for 248. His estimate might be fahified if,in populations J e vhom the relevant time of exposure is therefore under study, deaths started to occur from any group of ( tumours for which the latency, from exposure to death, i unequivocal, the mean ratio rises to a value of 3.0 by a was unusually long. As noted already, it is difficult to j i. L 404

L qy-

establish 'mean latencies for any type of solid tumour, 249. In any case. the median latency f:t fatal induc d

~ because of the wry long periods of follow-up required cancers in the populations studied at Hiroshima and _ for full ascertainment. There is some suggestion that Nagasaki, which had an initial mean age of about 30 at } tumours of skin, and perhaps also of pharynx and the time of exposure, could not greatly exceed 25 years, i larynx, may how long latencies, but tinis is uncertam. It ' since the mean expectation oflife would in itself wart to s is clear, however, that some malipant tumours have exclude a much lonpr median latency unless the exceptionally long mean survival times from dispons to distribution of latencies was very skewed positively;and u death, and this applies particularly to cancers of the ' it is of interest that that of leukaemia appears to be thyrood of the type induced by radi=*ian, and probably negativ 8y skewed. also to tumours of breast. The high induction rate for 4 such cancers by radiation might possibly, therefore, 250. ihe distribution of mahpancies fatal' within cause a bimodal distribution with time of deaths from - 27.5 3ars of exposure at Haroahuna and Nagasaki is t solid tumours, although the probable low fatality rate show. in table 21. As compared with Japanese National i from these thyroid tumours with adequate treatment Statistics, the individual fatal malagnancies shown on this makes.this unlikely. For the breast, the data of Wanebo basis to be significantly increased in frequency were et al (158) suggest an induction rate in the Adult Health ~ leukaemia, lung cancer and breast cancer. There was a i Study population exanuned between 1958 and 1966 probable increase in cancer of digestive organs other which is several times the mortahty recorded in the Life than the stomach and clear evidence of increase of Span Study by 1972. This ratio, however, could weg be cancers not separately listed. Of the total increase in consistent with the fact that many breast tumours may des'.hs from solid tumours, lung cancers accounted for be fully removed at operation. There is thus no a priart 28 per cent, those of the breast for 11 per cent and L season to expect any large number of deaths from a those of digestive organs other than the stomach for . separate group of malignant diseases to start to occur, about 18 per cent. Rather over half of all fatal induced ~ although the proportion of deathsthat are due to breast solid tumours therefore probably arose from these cancer and perhaps to thyroid cancer may well rise. tissues. TABLE 21. EXCESS MORTALITY FROM ALL CAUSES IN HIROSHIMA AND NAGASAKI,19501972 5 Matos and females, all ages l (Yeberfor alt =p==d at > 10 red rotes are per unit kerme} l t e I cw e-ed w a o.e mde w c==< wa >=p= ae uenoa.:s== ace l i-I'! 06 Es-Esener rere Es-Eseear mee 4 l,Q Canne ofdeath arrwed pereed Eseear (10** red") preted Eseear fl0** red") laukaemia 84 21.0 63(48 40) 36 (27-46) 13.7 70(5687) 40(32 50) All other mahsnant neoplasms 1 075 987.6 87 (24 153) 50(14-87) 918.8 156(103 212) 89 (59-12D Cancers of: I j, Stomach 417 411.8 5 (-34 47) 3(-20 28) 419.6 -3 (-36-33) -2 (-28-25) Otswr disestive organs 261 255.3 6 (- 26-39) 3 (-13-20) 232.8 28(2 56) 16(1 32) Trachea. bronchus and luns 100 76.2 24 (5 44) 13(3 24) 56.5 44 (28 62) 25 (16-36) 29 26.0 3 (-7-15) 2 (-5-10) 17.2 12(4 22) 7 (2 13) i Other respiratory organs' 37 24.0 13 (2-26) 7 (1 14) 19.0. 18 (9-30) - 10 (5-17) i Breast. 86 77J B (-9-28) 4 (-4-14) 79.1 7 (-8-24) 4(-5 14) Cervix and uterus Other mahonant neoplasms 145 !!7.0 28 (6 53) 16(3 30) 94.6 50(31 72) 29 (18 41) Benien or unspecified neopisems 53 52.6 0.4(-1316) 0.2 (-6 8) 42.9 10 (- 1-24) 6 (-1 14) AE diseases except neoplasms 3 970 3 947.7 22 (-101 148) 13(-6044) 4 592 -622 (-725--517) -354 (-413--294) sowee: Referease o7. Nere: The 90s, conne. ace nomise are omdiented in pareneneses-251. Mortality rates for cancer of the stomach, cancers as discussed above, and with the probably low oesophagus, urinary orgsns, and lymphatic and haemato-rates for other organs. It must be recognised, however, poietic tissues are also sipificantly increased in exposed that some bias towards high estimates for individual members of the Ufe Span Study during the period types of cancer may be introduced, since significant 1950 1974 as discussed below. Table 22 shows the induction might be suspected if random high, but not estimated mortalities per unit kerma in this period, and random low,incidences occurred. Indicates also the total rates for all malipancies. Although individual estimates are imprecise, it appears r ' that the types of malipancies listed may account for 252. For a number of these other organs or tissues, owr 90per cent of all deaths from malignant disease however, radiation has been shown to induce malipant that have occurred, and suggests that fatal malignancies tumours after moderate doses of a few hundred rads. [, of all other organs, for which increased rates are not yet While no convincing estimate can be made of the detectable, may together contribute only a small carcinogenic nsk for most of them,it is often possible to proportion of all fatal radiationinduced rnalignancies. show that the risk is unlikely to be high.The position [ ~ This would be consistent with tho low induction rate of can be reviewed briefly with regard to several such - bone cancers, and the low fatality rate of thyroid organs or tissues. 405 i . -. ~, - - - - - - - -,. -, - -., - .,---.-.-c

3 TABLE 22. EXCESS MORTALITY FROM MALIGNANCIES IN HIROSillMA AND NAGASAXI, 1950-1974 2 Males and females, all ages C (Ents rare per unir kerma.10'* red **) e Mehrnancy a n Hiroaktme Neenanki Both ettire ( I i Lauksemia 56(5141) 35 (29-41) 46(4250) e Other (solid) mahgnancaes: g Lung 13 (5-21) 3 (nes..ll) 8 (3-14) Stomach 19 (3 36) 10 (nege25) 15 (4-27) e Oesophagus 9 (4 14) -2 (neg. 3) 5(I-8) } Large bowel 4 (nes.-8) 0.5 (nes.-4) 2 (nes. 5) g Other digestive organs 9(0 18) 3 (nes. l3) 6 (nes.13) i Bladder and urinary organs 4 (nes.-8) 2 (nes.4) 3(0.56) [ t Lymphatic and haematopoieticorgans 5(f4) 4 (0-9) 4(17) Breast (females only) 10 (2 !8) 6 (nes.-17) 9 (2-15) Cervix and uterus (females only) 10 (nes.-24) -9 (nes.4) 3 (nes 12) Residue (by difference) 5 0 3 Totalsolid malignancies 77 ($1 102) 21 (nes.-47) 53 (35-71) Source: Lire Span Study 89so-1974 (14). Note: The 907, conridenct limits are indicated in perentheses. A. BRAIN irradiation, if, as the Committee noted in its 1972 253. Although tumours of the brain do not metastasize report, the total induction of malignancies in the foetus and may not therefore be classified as " malignant", theycorresponds to a rate of 20010~' rad'8, this would commonly constitute a hazard to life because of their suggest a high risk to the foetal brain. However, since situatien in a way that non-malignant tumoun of other only 19 cases were observed as compared with 14.3 organs ordinarily do not. For this reason, estimates in expected, the estimate, of 50(neg.145) 10-' rad-' has this section (paras. 253-256) are of total brain tumour wide confidence limits and is not inconsistent with that induction, where this information is available. for the brain during childhood. Moreover, it has been noted that many pelvimetries are canied out when the I 254. Modan et al (92) record the development of eight f etus is in a breech presentation, when the absorbed brain tumoun (six definite and two probable) during dose in the foetal brain is likely (106) to tw considerably l their follow-up from 12 to 23 years of 10 902 children Nghu than the nwan foetal dose. diated for tinea capitis. Irradiation was by 5 fields to V he scalp, with an estimated dose to brain of 120-140 rad at the surface, and 95120 at a depth of B. SALIVARY Gl. ANDS 2.5 cm. On the basis of two control studies,1.5 certain 257. Modan et al (92) record 4 malignant tumoun of l or probable tumours would have been expected. His salivary (parotid) glands in their follow up of heidence for a mean dose of, say,120 rad would 10 902 ch9dren inadtated for tinea capitis, with none in g correspond to an induction rate cf 5 (210) l0'* rad *8 either of the control series. The dose to the salivary 255. In its 1972 report, the Committee recorded the glands wn not esumated. The mean thyroid dose of j-t follow up study by Albert et al (1) of 2043 children 6.5 rad, however, is similar to that of 6 rad estimated in i treated for tinea capitis by radiation epiiation of the the New York study of tinea capitis irradiation in scalp and in whom the occunence of malignancies was children, for whom a parotid dose of 39 tad has now c2 pared with that in 1413 children who had had the been estimated by Harley et al (52). If this parotid dose l same infection untreated by radiation. Dese authors is taken as applying also in Modan's series, the induction i (139) now record the development of 6 intracranial rate for salivary cancers would be 9(3 22)l0 rad-8 tumours in 2215 irradiated children with none in 1395 controls, ne cruc'e incidence rate is quoted as 2.810-8 258. Four salivary tumours were observed in the New De estimated brain dose varied from 70 to 175 rad with York series (139) of which one was malignant (138), l 4 a mean of 140 rad, so that this rate per unit absorbed none being seen in the contrul series. De authors quote a crude rate of 1.810-5 for all such tumours, which i dose conesponds to 20(9-39) 10-' rad *' for a mean 4 would imply risk rates for salivary cancers of { follow up of 20.5 yean. A long latent period may 12 (135) l0-' rad-' to 20 yean. however elapse before diagnosis of radiation-induced j intracranial tumours and a mean value of about 27 years 259. Following neck irradiation in infancy or child- .i was noted by Munk et al (98). Indeed, Shore et al estimate from their data by actuarial methods that the hood, Saenger et al (128) reported an excess of 2 8 salivary cancers in association with n thyroid cancers. risk of intracranial tumour development to 29 years flempelmann (58) reports 4 salivary tumours, all now J would be about 2.5 times that already observed in their regarded as benign, in association with 24 thyroid series. cancen. If the salivary glands were as frequently exposed . He sensitivity of the foetal brain may, however, as the thyroid in the radiation fields used for neck high, since MacMahon (83) found tumours of the irradiations in childhood, these observations would t suggest an induction rate for salivary cancers ofless than i rvous system to constitute 24 per cent of "all one tenth that for the thyroid, or in the range y malignanews" apparently induced by intra-uterine (510) 10 rad. 406 L (

f i g h. fewer in numbers than are osteosarecmas in patients 260. Belsky (16) reports a significant excess cf salivary following incorporation of 28'Ra with only about tw2 cancers in A-bomb survivors in Iliroshima and. Nagasaki fifths the frequency of the sarcomas (127), so that the I. exposed at over 300 rad kerma, with two cases observedcontribution to fatal malignancies from this source ' Ses compared with 0.12 expected. Since 1340 persons would, in the case of whole body radiation, be even 16172 person years) were exposed in this group, this smaller than that from bone cancers, assuming equal excess rate is of 1.410-81 or,if the mean k'erma in the mortality from each type of tumour. There is some group was between 400 and 500 rad. about 310-' indication that the number of these carcinomas is now rad over a period of 12 years (19571970). This estimate would have confidence timits(I-8) l0-* rad. increasing more than is that of the sarcomas, corresponding to a longer latency until these tumours No excess was observed (3 observed,3.6 expected) in become detectable. The mucosa is closely applied to those exposed at 1299 rad, but absence of effect in this bone, so is likely to receive a dose at least equal to that group is not inconsistent with the significant excess in delivered to endosteal cells, or larger if accumulated the higher dose group. The rate in the latter would be radon makes any substantial contnbution. about the same ti estimated as per rad weighted absorbed dose (with W= 7 at this dose level), or 265. The relationship between incidence and mean somewhat higner (by about 40per < nt) per rad 28 ' Ra, quoted in paragraph 233, would l bone dose from unweighted absorbed dose. imply a risk rate of (2 5) 10~' rad-' if an RBE of 10 or 261. Takeichi et al. (152a) now report a total of 17 20 was taken for the alpha radiation from radium and if cancers of salivary glands that have been diagnosed in no added dose was derived from radon retention. If hospitals in Hirosluma during a 19-year period in additional irradiation resulted from radon accumulation. A-bomb survivors exposed within 5000 m of the ground the risk rate would of course be lower. While it seems zero. The number expected was 1.7, as judged by the likely, therefore, that these areas of mucosa may have a estimated numbers of survivors and the rate (1.210-' risk comparable with or lower than that from the whole y ')in those who were over 5000 m datant.The excess endosteum, no inference can be drawn as to the relative rate in those exposed at less than 1500 m was sensitivity of two tissues per unit area or per unit mass 21(9-41) 10-* y-' while for those exped at between or ceig3, 1500 and 5000 m,it was 7(414)l0' y-'. Risk rates I cannot be derived, however, since mean doses are not D. DIGESTIVEORGANS known for the total population of survivors. For 266. The mortality statistics of the Life Span Study to ' members of the Life Sps.n Study, the total air dose at gg72 (97) sitowed a possibly significant excess of 1500 m in Hiroshima was about 32 rad (67J and the A mean dose over all shorter distances would be about cancers of digestive organs other than the stomach, but 135 rad. If these dese levels apply also for all survivors, did not indicate any high incidence for the stomach (] it would suggest a low induction rate of a few cases per itself. Thus by the end of 1972,261 deaths had been attributed to cancers of digestive organs other than the nillion per rad in 19 years, but shielding factors become important at short ranges and are not known for stomach, as compared with 232.8 expected on the basis of Japanese National Statistics. This excess of 28.2 has survivors as a whole. 90% confidence limits of 2.2 and 56.4. 262. All these estimates are thus somewh2t lower than the values derived in paragraphs 257-259, perhaps 267. Nakamura has now identified a significant increase because the record is of cases following exposure at all in cancer of the stomach in irradiated populations in ages, and diagnosed only between 13 arid 25 years from Hiroshima (101).1te Life Span Study data (14) for exposure. All estimates have wide confidence limits, both cities (for the 24 years to September 1974) however, and would be consistent with a risk rate m the confirm an increased mortality from cancers of the digestive organs, with significant increases for oeso-terp$o ged phagus as well as stomach in Hiroshima, possible clu d d, with perha twi sy ue follow-up, and with no evidence of any greater rate after incuases f r large mtestine, and an meresse for the exposure in adult life. group of other digestive organs (see table 22). l 263. No evidence is available as to the mortality from 268. Tumor Registry data for the period of 1959-1970 l any radiation-induced salivary tumours. The 15 year suggest a signifier.ntly (P < 0.02) increased incidence in survival of naturally occurring cancers of the salivary H r shima of cancers of the stomach, oesophagus, liver l glands is, however, high (70 per cent, see paragraph 18 and large mtestine, and perhaps in Nagasaki of cancers of and reference 23). 'Ihe average risk rate per unit the pancreas (P = 0.06) and large intestine (P = 0.085). !e' absorbed dose for fatal salivary cancers may thus be j n,rosh,m, y,saraki about 510 rad. (Incidence rate, per unit kerma.10** raf) l Stomach 18(3-34) 10 (-5 26) C. MUCOSA OF CRANIAL SINI>SES Desophagus 6 (1-10) 4(27) 4 (-2 10) Large. testine 6(1 11) 3 (-t 6) A' 264. As already stated, there is clear evidence of the m ( ) carcinogenic effect of radiation on the mucosa of the Pancmas O (-4 4% { mastoid and other air sinuses, but risk estimates cannot it must be emphasized, however, that incidence rates, as be derived since no determinations have been made of based on procedures of tumour registration, are liable to t, radon accumulatien in these sinuses. It seems c: car, vanous possibilities of bias. These are not reflected in fI however, that the risk must be small. Such cancers are 407

T i. ,the confidence limits or probabilities cf significance showed an sxcess cf deaths (occurring at more than 6 _ given here, which are determined only by the random years after treatment) from cancer of several organs of errors involved in estimating these rates (by contingency the digestive tract which were heavily irradiated, table methods (66)). although at unascertained dose. For the groups of patients completely followed up to 1 January 1960, or 269. 'Ihe studies of Court Brown and Doll (29) on incompletely followed up to 1 January 1%3, values . patients treated for ankylosing spondylitis by irradiation were as follows: Compierefonowup incompeere foCowup (141796 personyears) (163 631 personyears) 06 Er Ok Ex. sernd pected Extwss sermi pected Excess Cancers of: Pharynx 4 0.70 33(0.7-8.5) 5 1.05 3.95 (0.9 9.5) Oesoohagus 3 2.25 0.75 (nes. 5.6) 3 337 -0.6(nes 43) Stomach 28 16.0 12.0 (3.9-22.4) 38 23.6 14.4 (4.9 25.1) Pancreas 9 3.8 5.2(0.91I.9) 12 5.7 63(l.213.7) 270. In a follow.up of the patients who had only one either in the total series (of 14109 patients with mean course of radsotherapy for ankylosing spondylitis, Doll follow.up of 9.5 years) or in deaths at 6 or more years .and Smith (36) observed numbers of deaths from after treatment (6838 patients with a further follow.up stomach, pancreas, large intestine and probably oeso. phagus, which were significantly increased as shown to a mean time of 173 years from exposure as follows: AE doeths Deathsafter 6 years Ok Ex. Ok Ex. sernd peered Excess served peered Excess Cancers of: Stomach 45 34.2 10.8 (0.4 24) 36 24.6 11.4 (2.1 23) Oesophagus 10 5.6 4.4 (-0.2 11) 9 4.3 4.7 (- 0.4 11) Pancreas 18 9.5 8.5(1.2 17) 12 7.5 4.5 (-0.6 12) large intestine 28 17.3 10.7(2.6 21) 18 13.1 4.9 (-1.5 13.6) 271. In the earlier report, a significantly increased years after pelvic irradiation for benign disorders. About nurnber of deaths from cancers of the large intestine was half of the large intestine is likely to have been in the also reported (25 observed and 14.8 expected) but the irradiation field and to have received a mean dose in th: authors hesitated to ascribe this excess to radiation in order of 800 R(54). Certainly 3, probably all 4, of the l view of the known increase of cancer of the colon in cancers of the later intestine arose from the area l ulcerative cohtis, and the association of ulcerative colitis included within the field. No cancers were observed as with ankylosing spondylitis. Similarly, it was thorght having occurred in the small intestine, of which about that the increased rate of stomach cancer might possibly half is likely to have been in the field. The observations be attributable to drugs taken in the relief of pain in ankylosing spondylitis. suggest a substantial induction rate in large intestine and rectum, perhaps of the orderof 2510-' rad-' from half the intestine, but these indications are necessarily very 272. Now, however, a series of patients with ankylosing approximate. spondylitis, but not treated by radiotherapy, has been found to have no increased cancer incidence as 274. Smith and Doll (142) observed an increased compared with a normal population (120). During a number of deaths from cancer of the intestine in 2067 l mean observation period of 7.9 years on 1021 patients, patients followed for a mean period of 19 years after l the total number of deaths from mahanant disease was treatment for metropathia haemorrhagica by pelvic 18, compared with 19.1 expected. Of these,3 were from x irradiation. From cancer of sne intestines,24 deaths stomach cancer (2.6 expected) and none from cancer of occurred at 5 years or more after irradiation, with 13.86 large intestine (1.4 expected; one colon cancer occurred during a further 3-year period which added 8 per e n! to expected; and for cancer of the rectum,8, with 5.23 the person years surveyed). It does appear likely expected. It is of interest that of the intestinal cancers,3 therefore that the increases in cancer rate for stomach were of small intestine (143). With 0.4 expected,3 or more would only occur by chance with P= 0.008. For and large intestine, as well as for oesophagus and cancers of the large intestine and rectum, the excess of l I pancreas, may be attributable to radiation, and that risk 29 - 18.69 = 1031 (2.120.8) would only occur by estimates may be obtainable from this source when the chance with P = 0.028. absorbed doses to these organs from the courses of treatment have been derived. j 275. While these data indicate the likelihood of cancer J induction in both small and large intestine, no reliable 273. Bnnkley and Haybittle (18) noted an increase of nsk estimate can be derived because of the small r fatal cancer of colon and rectum (7 observed,1.5 numbers of cases and the uncertainty as to the expected) in 227 patients within an average period of 16 proportion of small or large intestine lying within the 408 i

l5 i pelvis and exposed in tha primary beam. Ifit b asswned, after aoses which presumably were about 700 rad. He however, that only about one third of each intestine estimated excess incidence cf all pelvic cancers wu 6 per (small or large), but all the rectum, was directly exposed cent, whereas that following irradiation for cervical c3 a mean dose of 400 rad, approximate risk estimates cancer in 471 patients, at dose levels higher by factors of I would be: small intestine 10(2 25) 10-' rad-8 ; large 2 or 3, was only 0.35 per cent. He latter estimate is 8 ~ intestine 25 (2-60)l0-* rad-8; rectum 3 (010)l0-* imprecise, as only 13 cancers were found as compared - - rad-8 with 11.4 expected, but does suggest that the induction of these tumours may be lower at high doses. 276. In summary, therefore, it is clear that radiation induces cancers of the gastro-intestinal tract and 279. Palmer et al (107) quote five surveys including a digestive organs, although the total risk rate for all these total of 3968 patients in whom 27 uterine,10 cervical tissues is not high, the estimate for Hiroshima and and 8 ovarian cancers were obseged following pelvic Nagasaki mortality 1950 1974 being about irradiation, but they note thar mean periods of 25(10 45)10-* rad-' kerma. Estimates for individual follow.up were commonly less than 10 years and no data sections of the gut are in the region of 510-' red-' for are given of the local dosimetry or control series. oesophagus, small intestine and perhaps rectum. De rate for large intestine is similar on Japanese data, but is 280. Following pelvic irradiation for metropathia i somewhat higher as inferred from pelvic irradiation haemorrhagica, Smith and Doli (142) observed the (paras. 273 275), but these estimates depend upon following numbers of deaths from pelvic malignancies, assumptions as to the proportion of bowel present in the occurring at 5 or more years after irradiation: primary beam. A value of (510) 10-' rad-' would however be consistent with these data also. For the Site ICD Code Observed Exprered stomach "a rate of(10-20) 10-' red-8 is suggested by the Uterus (171/174) 16 10.34 current Japanese information, but a further estimate for Ovary (175) 8 7.66 low-LET radiation will be availab!: as soon as dosimetric Bladder (181) 3 2.15 estimates are obtained for the stomach irradiation during Other pelvic (176) 0 0.85 the treatment of ankylosing spondylitis.His applies also to the pancreas. for which a significant excess of cancers Rese data suggest (P = 0.08) an excess of fatal uterine la detectable following this therapy. Estimates for the cancer, with an esti nated risk-for death between 5 and liver are discussed below (paras. 283 285). 19 years after 400 rad-of 6.8 (016.9) 10 rad-' if the incidence of uterine cancer is not increased in patients with metropathis haemorrhagica. Data for other pelvic organs indicate that risks higher than 810-' rad-' for E. PELVIC ORGANS ovary and 710-* ra(' for bladder are imprcbable i (P < 0.05). 277. Palmer et al. (107) also noticed an excess of rectal 0 cancers following the irradiation of the pelvis by radium 281. Data for Hiroshima and Nagasaki (14) still show or x rays or both in 651 patients with uterine fibroids no significant increase in mortality from cancer of cervix g g and 80 with other benign pelvic disorders. Seven rectal and uterus, and appear to exclude a risk rate per unit y cancers occurred within a mean follow-up period of 16.i kerma higher than about 1010-* rad-8-the estimates years as compared with 2.1 expected. No risk estimate for excess mortality in both cities to 1974 being can be derived, both because of uncertainty as to dosage 3(-7 12)10-* rad-8 He mortality rate per unit and because ascertainment was by letters sent to kerma for cancers of bladder and urinary tract is, previously irradiated patients to which replies wm however, just significantly in excess of expectation, with received in rather less than 50 per cent of cases. local a value of 3 (0.5 6) 10-' rad-'. Tumor Registry data for doses were quoted as of 2700 R, and 650 700 R, at 2 1959 1970 confirm an increasing incidence with I and 5 cm from the radium source in the cervix, and of increasing exposure for bladder (and urinary organs) { "the equivalent of 1600 to 1800 mg.hr" if x rays alone cancers in both cities, and-for certain dose groups in were used. He cancers observed and expected were: Hiroshima-suggest this also for the ovary (tube and i ligament) and for the cervix. No significant excess is g, g,,,,, Observed Expected (yearsj observed for cancers of the prostate or rectum. stimates rates during this 12. year period have been: Fundus uteri 29 4.9 9.7

1. '## A' 6

10 Rectum 7 2.1 10.7 II"'id'" '*l' per unit kenna.10* rad") Bladder 3 0.8 14.0 Bladder 4 (0-8) 4 (0-8) -i Vagina 2 0.2 12.7 Ovary 7 (2 12) 0(-3 3) i Vulva 1 0.3 14.5 Cervix 13(2 24) -3 (-17 11) Vl ,i 278. He excess number for the rectum is thus about 282. A suggestion of sensitivity of the bladder to cancer equal to those for the cervix and ovary, although the induction by radiation at much higher doses is given by O excess is about one fifth of that doses will not necessarily have been the same. His data of McIntyre and Pointon, who observed 16 for cancer of the epithelial cancers of the bladder, with 9 expected, in fundus uteri, in which malignancies occurred in 4 per 8950 patients treated for cancer of the cervix (89). cent of patients compared with 0.7 per cent expected, Ascertainment was likely to have been better than [ e 409 sih

M, 50per cent, but was certainly not known to be 5 per cent within 20 yIars cf injections involving liver complete. he dose ta the bladder, of 4000 rad or more, absorbed doses of about 500 rad b this time. This is too high to allow any useful risk assessment of the would imply a risk rate of 10010'{ rad-' of alpha carcinogenic effects of moderate doses, radiation if the carcinogenic effect of Thorotrast was [,] due essentially to its radiation, rather than to its chemical, effect. It is suggested that this is so in the light (,/ F. LIVER of observations on animals injected with a " thorotrast" in which the 832 Th content is enriched with asoTh. 283. Only a tentative risk assessment appears to be Rese surveys would suggest that the lifetime carcino-possible for the livet. Cancer of the liver has repeatedly been observed following Thorotrast administration but, genic risk ofliver irradiation oflow LET would be in the l region of (10 20)10~' rad-8 if an RBE of 10 or 20 l as stated earlier, muel. of the energy of the decay of the were applicable to alpha radiation from '82% at these contained thorium is expended in necrotic tissue and in dose levels. It remains uncertain however, how much of the Borotrast deposits themselves. Estimates have however been made (75) of the mean dose toliver from the estimated dose is " wasted" radiation, either because a given injection of Dorotrast, with corrections for of irradiation of already necrotic cells, or because the self-absorption depending upon the mass injected. In the tumours have already been induced while exposure is 8 light of these dosimetric estimates, it appears from a continuing. It also remains uncertain whether any of the l' number of epidemiological surveys (see table 23) that cance s observed are chemically or mechanically induced I by the thorium oxide preparation (see also Annex I, the excess frequency of liver cancer is in the region of paragraph 253). TABl.E 23. OCCURRENCE OF 1.IVER CANCERS FOLLOWING TIIOROTRAST INJECTION Approximate Mean dose Menn solume A ccumulated I meen of of Thororrest dose in folJowp Thororrest infected 2C years s Reference Country (years) (ml) (mt) (red) 141 Portugal 30 26 6.4 530 42 Dentnark 28 23 3.6 470 69 Unned States 10 24 72

Germany, Fed. Rep.of 30 30 7.0 570 96 Japan 33 16 4.5 370 N

V 284. The induced tumours are mainly cholangiocar-cinomas and haemangioendothehomas, with smaller standardize for population distribution and for the numbers of hepatomas arising from liver cells them-skewed distribution of age in the population at the time of the bomb owing to the paucity of civilian males of selves. His distribution may of course reflect the military age. For multiple myeloma, 6 cases were relative sensitivity of cpithelial and parenchymal tissues observed with 2.6 expected on this basis, an excess of to cell killing by radiation rather than to tumour induction alone. 3.4 (0.0 9.2). The numbers in the two cities were: Mahanant lymphoma Multiple myeloma 285. De incidence of liver cancer in Hiroshima and Nagasaki, as observed in Tumor Registry data for Observed Expected observed Expected 1959-1970, increases with increasing estimated kerma, Hiroshima the excess mcidence during tiu,s period bem 1-99 rad 7 11.0 3 1.6 4 (2 7)10 rad in Hiroshima and 4 (neg.10) 10'f >100 8 1.3 1 0.2 rad *' in Nagasaki. Nagasaki 1-99 rad 6 4.7 2 0.7 >100 2 1.0 0 0.15 G. MALIGNANT LYMPHOMA AND Total 23 18.0 6 2.6 MULTIPLE MYELOMA 286. An apparent increase in both these types of 288. He total excess of cases of malignant lymphoma tumour was r'oted in the Committee's last report, with is thus only 5.0 (0-14.6). It is suggestive, however, that the reservation that more data were needed to conclude the greatest excess occurred in the highest exposure that A-bomb survivors were at increased risk of developing these forms of malignancy. group at Hiroshima. In this group of 3138 survivors there were 4 cases oflyrnphosarcoma (0.4 expected on a i 287. Data published by Nishiyama et al. (104) show basis of those exposed at less than I rad at Hiroshima), I that 23 cases of malignant lymphoma were recorded in case of reticulum cell sarcoma (0.5 expected) and three i 1945-1965 among 44 509 survivors in the extended Life cases of Hodgkin's disease (0.2 expected). Since this Sumber can be compared with 18.0 expected on the group corresponds to about 0.7610' man rad, the pan Study sample at Hiroshima and Nagasaki. This excess per unit kerma for all malignant lymphomas basis of findings in 34 675 survivors exposed at less than would, durina the period of observation, represent about k 9 (418) 10" rad-', or 12 (5 23) 10 rad ~' absorbed 1 rad in both cities, adjustments being made to dose, assuming absorbed-dose / kerma ratios as for the I 410 Yr

( 3 with which they are induced. The absence of reports of . d comach (53). If the high.LET component wne skin tumours occurring either in the populations weighted by about 10 (para.55), the rate would be exposed at Hiroshima and Nagasaki, or in the closely ?'} 10(4-20) 10" rad, studied populations on Rongelap Island or following }

89. Court Brown and Doll (29) found an excess of neck radiation in infancy, strongly suggest that such s

7(111) deaths from cancers of lymphatic and tumours must be infrequent at moderate dosage. In g matopoietic tissues to be associated with 30(2141) these populations, the low mortality from skin tumours deaths from leukaemia at periods of 6 years or more would presumably not prevent their occurrence being y l~ after treatments for ankylosing spondylitis.nis ratio of detected on examination, even if they had been excised h rates appears to be broadly consistent with estimates for or cured by radiation therapy. It can be argued that y lymphoma noted in paragraph 288 and for leukaemia in nfrequent reporting of skin tumours following radiation ) chapter 11 of this annex, assuming that the relevant therapy in adults does not constitute evidence against I lymphatic and marrow tissues received about equal doses the sensitivity of the skin as a whole for moderate doses, in view of the small fields, high levels of skin irradiation b these treatments. and often of the short survival of patients after 290. Malignant neoplasms of lymphatic and haemato-treatment. These observations,however,do not apply to poietic tissues (apart from leukaemia) have now been populations exposed to moderate whole body radiation. identified as contributing a small increasing mortality in j the 1.ife Span Study cohort from 1950-1974,the excess h h skin h b M W rates r unit kerma over tius period being: (10 to carcinogenesis by radiation is suggested by the findings of Modan et al. (92) of only one skin cancer (on Hiroshima 5(18) the scalp) in the 10902 children given scalp irradiation Nagasaki 4(09) (with 5 fields at 350 R per field). No skin tumours were Both cities 4 (I 7) seen in the control series. It is likely, in view of the known variation of head circumference and the body 291. In the study of Doll and Smith on patients treated t 4 pu cent oMe total m ace area age, t a with a single course of radiotherapy for ankylosing skin area may have been irradiated, so that a low rate a' spondylitis (36), no dose estimates are yet available for w uld apply even for whole skm arradiation. Modan et I lymphatic tissues (excluding those from Hodgkin's 81-emphasize, however, that skm tumours, being f i Disease which were not increased), but the excess deaths au ciate w aI wm ty, may en mined d from malignant tumours of these tissues can be

• • ""'. during the early years after irradiation

'Y b-compared with those from leuknemia in the s'ame series,

• *" "* "" " ' N* 'Y *** "*
• h suggesting that lymphatic malignancies are less frequent 295. Albert and Omran (I) record the occurrence of and develop later than leukaemia:

two basal cell carcinomas of the scalp within 12 years of r Excers deaths irradiation of 2043 children of average age 7.5 yeart Mean period From l with none in their control group of 1413 children.Since Number during which mahrnancies of deaths occurred From oflymphatic the scalp dose was of 500 800 rad, this implies a low risk level of I or 2 skin cancers per million per rad for parents (years) Jeuk cmaa nssu 14 109 0-9.2 24.5(16 35) 8.2 f 0.3-16) irradiation of this area. Further follow.up to a mean ( 6 838 617.3 10.4(5-18) 7.6(2.5-15) time of 20.5 years (139) now shows 10 skin tumours to u have occurred within the unshielded areas (in contrast to only one in the control group), the majority being basal 292. Matanoski et al. (89a) have recorded an increase in cell carcinomas. The totalincidence of epithelial cancers standardized mortality ratio (SMR) for certain tumours corresponds to an induction rate per unit absorbed dose arising from lymphoid tissues in members of the of about 5(210)l0 rad-' by 20 years for this area 4 j Radiological Societies of North America, as compared with those found during the same periods in members of exposed in childhood, although with no excess mortality from this cause. I pathologists' and oto-rhino-laryngologists' societics. The increases were observed for lymphosarcoma and 296. Rowell (126) notes the occurrence of 5 skin reticulum cell carcoma (ICD 200) for 1920-1929 and cancers (one squamous and four basal cell) in 100 1930-1939, although not for 1940 D49; and for patients treated for benign dermatoses during 19301964 Hodgkin's Disease (ICD 201) for the last of these periods and examined in 1964. Patients were selected for only. For other neoplasms of lymphatic and haemato-examination if the treatment doses had exceeded poietic tissues (ICD 202), increases occurred during the 1500 R to hands or 1600 R to face, nails or feet. Skin last two of these periods, and particularly during cancers occurred in 1 of 59 patients whose hands were 1940-1949 when the SMR was 5.7 for radiologists as treated, and in 4 of 25 patients who had treatment to compared with 1.0 for pathologists and 2.2 for the feet.The mean doses are not stated,but the tumours oto.thino-laryngologists. Risk estimates cannot be developed after exposures of 2000-3200 R.The fields,if derived from these data, however, in the absence of used for treatment of dermatoses, may have myolved the adequate estimates of the doses received at work. whole of hands or face but would still represent only a small percentage of body area, and a S.per. cent tumour incidence following, say,2500 R to a few per cent of the O H. SKIN skin surface would suggest a substantial sensitivity of the skin to carcinogenesis at these dose levels. It is clear, 293. Malignant tumours of skin were the first cancers however, that more and larger series require to be reported to have been induced by radiation (46), but no estimate yet appears to be available of the frequency studied. 411

~5 5.' 297. Delarue et al (31) record the occurrence cf only I at 1 one skin cancer in the area likely t2 have been exposed no excess incidence cf laryngeal cancer was observed in during multiple fluoroscopies done in management of Czechoslovak uranium miners by Placek and Seve in appe 1975(113). mah pneumothorax treatment of 308 patients. Since an in : average of 142 examinations were made, at a mean skin 301. Any estimates of this type, however, are x.ra ose of 612 rad for each, this corresponds to an necessarily approximate and should be treated with thot x.ra incidence rate of about 4(0.2 18)10 rad-' following considerable reserve in view of the long latencies that i diff exposure presumably of about 3-5 per cent of the total have been reported for radiation. induced tumours of the skin area. the skin. Pegum (110) has noted a mean latency of 31 years l (1.1 after treatment of tinea capitis; Andrews (4), a mean l in t 298. In patients who received a single course of value of 27 yean 11 patients fo!!owing radiotherapy; and radiotherapy for ankylosing spondylitis, Doll and Smith Petersen (111), one of 27.5 years in 21 patients; while (36) have recorded no deaths from skin cancer, with 1.6 Sp ttle (para.12, table 1) observed the very long mean expected, in the 6838 patients followed from 6 years latency of 41.5 years for basal cell cancer in a group of D' after the course for a further mean period of I1.3 years. patients after treatment of tinea capitis by radiotherapy. anc A typical treatment is likely to have involved a total skin rati dose in the region of 10001500 rad to 3-5 per cent of 302. It is important, however, to note the low mortality E the total skin area. Since the upper limit of the 90% associated with skin cancer. Andrews (4), for example, I II* confidence zone (with 0 observed and 1.6 expected) has observed a mortality of only 6per cent dunng would correspond to a possibility of 1.4 cases, the 10 years after diagn6 sis. It is likely, therefore, that the l Ib maximum estimated annual rate for whole-skin ex contribution of the sliin to a fatal radiation-induced j I"' at these dose levels would be about 0.510-* y posurerad , cancer rate must be low,

• I i

implying a relatively low mortality even if the rate l th operated over several decades. in i I T8 VIIL MALIGNANCIESIN 299. Sevcova et al (137) observed the incidence during PRE-NATALLY EXPOSED CHILDREN l 3< a 5-year period of skin cancer in uranium miners at a g rate of 18.7 (8.0 38.0) l0-* (95% confidence limits), as 303. In its 1972 report, the Committee described in i compared with an expected rate of 4.110-* for Czech detail the results of studies'of Stewart and Kneale (149, c: males of the same age distribution. De rates in other 150) and of MacMahon and Hutchinson (84), showing rt uranium workers (not miners) were 4.8 (0.617.0) 10-4 an increased risk in the development of malignancies II observed and 7.210 expected. The tumours in the during the first 10 years oflife of children who had been 0 'ners were predominantly basal cell cancers of the irradiated in utero during pelvimetric or other x ray k and forehead. De estimated absorbed dose to examinations involving the pelvis of the mother. It was basal cell layers of the epidermis from the alpha stated that, although children born from mothers 2 radiation from radon daughters was estimated as 6 rad

x. rayed while pregnant seem to have an increased risk of 1

per year, as derived from epidermal thickness data of cancer after birth, a possibility still remamed that the ( Whitton (161) and from measured average skin association, or at least part of it, was caused by factors a contanunations cf 6pCicm-8 under current exposure other than radiation, and that further studies were i conditions. Having regard to average lengths of mining needed to clarify tnis point. i experience prior to skin cancer diagnosis (14.2 years) and the probable occurrence of higher contamination 3% nis uncertainty was partly because of a i t I levels previously, the accumulated dose to basal cells wa discrepancy in the frequency of such malignancies estimated as 100 rad or r. ore. On this basis, the f 11 wing irradiation in utcro in the studies referred to i induction rate for facial skin can be estimated as 310 6 and in those of Jablon and Kato (65)of children whose rad *' y-' or less. Alternatively,since an estimated dose m thers were pregnant at the time of the A-bomb l rate of 610 rad-' y was leading to an incidence of explosions in Hiroshima and Nagasaki. A second possible 29010-' y, the total induction rate could be criticism of the evidence, however, was that a subgroup regarded as at least 30 5010-* rad ~' of alpha radiation, of mothers might have been more frequently x. rayed and this estimate should be increased owing to the than others because of heredituily determined abnor-relatively short rnean times of exposure and survey and malities (for example, of the pelvis) and that this i the known long latency for many skin cancers. subgroup might have a preponderance of children who developed malignant disease as a result of some associated and transmitted congenital abnormalities. 300. It may be questioned however whether the ascertainment of skin cancer in the mining populations, surveyed for the development of this condition, might 305. Mole (93) has now pointed out, by a further have differed in efficiency from that for the general analysis of the data of Stewart (151,152), that pelvic population as derived from nationwide oncological x rays are much more frequently carried out (in 55 per a cent of cases) on mothers with twin pregnancies, than on notifications, but the rates recorded for the non-mining mothers with other, or " singleton", pregnancies (of J ranium workers, from registrations at the same health whom only 10per cent are x-rayed during the "tute, were in fact somewhat lower than those pregnancy). Despite this, the excess frequency of j v2cted on the nationwide basis. All tumours were removed surgically and no signs of recurrence had been malignancies in twins who were irradiated m utero is no observed at the time of reporting. It may be noted that higher than that in singletons so irradiated. The factors v r, which called for a greater frequency of pre. natal x rays, h 412 8

iw at least in this type of pregnancy, do not therefore IX. CONCLUSIONS l' appen to be associated with an increase in childhood malignanci s. Thus, in singletons, leukaemia developed A. CARCINOGENESIS IN DIFFERENT ORGANS 3 i in 3.410 of all live-bom children who had been j x rsyed in utero, but in only 2.310-* of those not 309. The evidence on radiation carcinogenesis in man !.. x.rsyed. In twins, leukaemia developed in 2.610-* of that has been reviewed in this Annex indicates a those x-rayed and in 1.210-* of those not x rayed.The progressive kicrease in information u to the tissues and difference in rate apparently attributable to radiation is thus about equal in singletons and twins, being organs in which malignancies may be induced at (1.18 2 0.15)10-* in the former and (1.4310.46) 10 absorbed doses in the region of a few hundred rads or less. It also allows approximate estimates to be made of in the latter. the risk of inducing a malignancy at these dose levels for an increasing number of such organs if selectively 306. For other malignancies a similar result was tradiated, or for the whole body if more uniformly l observed. In x-rayed singletons the rate was 3.910 exposed. and 2.710 in those not x-rayed in x-rayed twins the rate was'3.110, and 2.010-* in those not x-rayed. 310. The risk of the development of a malignancy he difference in rate attributable to radiation is following irradiation of individual tissues at dose levels l (1.24 i 0.17) 10-* in singletons, and (1.14 i 0.53) 10-4 of the order of 100 rad may vary with the ET of the in twins. Provided the higher still-birth rate in twins radiation, sometimes with the age and sex of the subject (5.7 per cent) than in singletons (2.1 per cent) did not exposed (see Annex 1, paragraph 289), and probably also l introduce bias by including an excess of still-bom twins with the dose rate or "fractionation" with which the who would have developed leukaemia or solid cancers, dose is delivered (see Annex 1, paragraph 179). the evidence clearly supports a likelihood that the I increase in malignancies was attributable to radiation 311. De risk estimates obtainable for individual organs l rather than to selective factors. or tissues cannot be determined with great accuracy, but for several tissues, such as those of the bone marrow and i 307. Newcombe and McGregor (102) and Holford (59) the thyroid gland, approximate estimates are obtainable i have discussed the apparently litear relationship from several different sources and are mutually j l observed by Stewart and Kneale (150) between excess consistent within the accuracy of their determination. lt cancer risk and number of films per examination and is thus becoming possible to classify different body regard the data as showing a close correlation to a simple tissues into groups with different degrees of sensitivity l linear relationship with dose,down to doses in the range to induction of malignancies by radiation. It must be of 0.2 0.25 rad. emphastzed, however, that such classifications, and the values for the induction rates themselves, are derived 308. Mole (93) discusses the apparent discrepancy from the results of exposure at high dose levels, between the excess of childhood cancers following ordinarily in excess of 100 rad (of low.ET radiation). diagnostic x rays in utero, and the apparently lack of As stated above for leukaemia (paragraph 96) the any such significant excess in Japanese A-bomb survivors corresponding rates per unit absorbed dose at doses of a who had been irradiated in utero (G,73). He suggests few rad, while unlikely to be higher, may be that the higher foetal doses received in Hiroshima and substantially lower. Nagasaki may have been associated with cell. killing effects which may have reduced the subsequent 312. The t!'yroid and the female breast appear to have incidence of cancers that would otherwise have been high rates of cancer induction by radiation at levels expected. He points out also that, with only I childhood usually in excess of 100 rad, at least for certain ages at malignancy observed after irradiation in utero in these exposure. Average risk rates for all ages appear to be in cities, as compared with 0.4 expected on the basis of the region of 10010-' induced cancers per rad absorbed Japanese National Statistics, the confidence limits of the dose of low ET radiation. 7he low mortality rate for resulting risk estimate would in any case have been wide radiation-induced thyroid cancers and the moderately and might be consistent with the Committee's estimate, low rate for breast cancers probably bring the in its 1972 report, of the risk of malignancies following corresponding average risk rates fer fatal cancer foetal irradiation. With 34 900 man rad to mothers and induction for these two issues to about 1010-' and with absorbed doses to the foetus as given by Auxier et 5010-* rad-8 respectively, al (9), the foetal risk per unit kerma would be 43 (neg.-316) 10-' rad-8,or 25 (neg.-186) 10-' rad-' if 313. De induction rate for lung cancer also is high for J the high-ET component of absorbed dose were males of owr 35, insofar as a value can be derived from I weighted as in paragraph 55. Such estimates are not experience of uranium mining and p obably also for l necessarily inconsistent therefore with the Committee's females at these ages. Rates at other ages are difficult to risk rate per unit absorbed dose for foetalirradiation, as establish but seem clearly to be lower, and a mean rate given in its 1972 report,of 23010-' rad-' (for doses in for all ages in the region of (25-50)10-' rad-' of the range of 0.2 20 rad). In summary therefore,the risk low.MT radiation seems piobable. per unit absorbed dose of fatal induced malignancies by foetal irradiation may be in the region of 314. The induction of leukaemia, of the acute and 200 250 10-' rad-', half of these malignancies being chronic granulocytic forms, occurs with a risk which attributed to leukaeia and one quarter to tumours of appears to fall, probably from about 5010-5 rad-' at the nervous system (see am Annex 1, paragraph 303). moderately high doses, to' the region of 2010-' rad-' 413 i

as the dose level 1[ reduced. F:r radi: tion-induced much lower dose levels involved in occupational leukaemia, the mean interval between exposure and ben death L1in the region cf 10 years, and is much shorter exposure, and even more so in environmental exposures than the mean interval for other malignancies, which ta radiation, may well be substantially less; and this pati typically appears to be 25 years or more; The total reservation applies also to values quoted for individual dos ,, induced mortality from leukaemia can thus be estimated organs or tissues. Moreover, these values, although often (ps consistent with several different sources of evidence, are rada ( ) with greater confidence in most present surveys than only very approximately determined, and indicate the V that from other malignancies. need for further investigation of many points. They do, 32: 315. He risk rates for stomach, liver and large intestine however, form some basis for assessing the possible bas low significance of occupational exposure and for planning are less reliably determined, but appear to be rad of protection measures. substantially lower, and are probably in the region of ext (1015) 10-' rad-' for fatal cancers. For the brain, 319. It is likely that malignancies may be induced in the im-quantitative information is scanty, but the rate for all foetus by exposure in utero at average absorbed doses in yie tumours, which may prove fatal in this situation, may be the range d 0.2 20 rad from diagnostic x rays bo hig similar. He saliva y glands appear to have an induction rate in about this range also, but the risk rate for fatal (para. 308). The induction rate of fatal malignancies per cancers is likely to be lower. unit absorbed dose is difficult to determine with any ut: thc confidence, but is estimated as being in the region of (200 250) 10 rad ~' po 316. For bone, oesophagus, small intestine, urinary do bladder, pancreas, rectum and the lymphatic tissues, in-risks of cancer induction are identifiable but appear to wi be lower again, with values probably in the region of ce (2 5)10-' rad. He risk for the mucosa of cranial B. INDICATIONS FOR FUTURE WORK ra. sinuses appears to be similar. No good estimate is irt available for skin cancer induction, but the induction of 320. Quite clearly, however, much further information be fital cancers of skin appears also to be low. All estimates a needed before the carcinogenic risks of radiation can di quoted in these paragraphs (312 to 316) are " rounded be estimated with accuracy and confidence, particularly ar at low doses. De Committee wishes to emphasize the values which are broadly consistent with figures ty obtainable from different sources, but reference should importance of further mvestigations, particularly in four areas: ar be made to.the paragraphs cf the annex which deal with w the various tissues for reviews of the different bases for (a) ne continuation of surveys of cancer risk estimation and the reservations or limitations whichinduction following radiation exposure, forlong periods 3 apply to each. of time-ideally.hroughout the lifetimes of those ci nitially exposed; 1

17. Evidence is given in paragraph 247 which suggests (b) Further study of the basis for assessing risk u

t the total of all fatal malignancies which may ultimately result from a given uniform whole body from low 1.ET radiation at various dose levels, from that 3 resulting from high-l.ET radiation; e exposure, may be in the region of 4-6 times that for e leukaemia alone. The average rates quoted in para-(c) Continued examinations of the effects upon t. graphs 312 317 are necessarily tentative, but do not the risk of malignancy, of the unifortnity of radiation appear inconsistent with this possibility. Rus, the total exposure, both within a tissue of which the cells all have 2 rate of all fatal solid cancers for the various tissues or a similar sensitivity to tumour induction, and within s organs for which some estimate has been made, amounts organs in which the malignancy may result only from to about 20010 rad-' for moderately high doses (e.g. irradiation of certain types of cell; 1 t cf a hundred to a few hundred rads of low LET (d) Further investigation of the proper basis for 1 radiation). If these tissues have been identified becauseinferring risk rates at low doses from those observed ( they are the main contributors to the carcinogenic following high doses, and in assessing the influence of effects of radiation, and ifleukaemia is induced with a rate of 5010-* rad at these doselevels, the final ratio Rese fields of investigation are discussed in the cf a'l total malignancies to leukaemias could thus be in following paragraphs. the region of 6. 318. It is to be expected that low LET radiation is 321. The need for prolonged continuation of epidemio-likely to be !:ss carcinogenic per unit absorbed dose at logical studies is obvious, since it cannot otherwise be doses of a few rads than at levels of one or a few known what types or frequencies of cancer may occur at hundred rads. For dose levels at which a leukaemiaseveral decades after exposure, or what cancers may develop when those who were exposed in childhood induction rate of (15-25) 10 rad-8 may apply (see paragraph %), a ratio of 4-6 between the frequency of reach an age at which hormonal or other influences other induced fatal malignancies and that for leukaemia facilitate the expression of cancers induced many years would imply a total for all fatal induced malignancies, previously (see Annex 1, paragraph 68). Rese consider-including leukaemia, of (5-7) times (15 25) 10~' rad ~', ations apply very strongly to the extremely important esting a value of about 10010 rad-' at such dose observations in Hiroshima and Nagasaki, of which the Is. It must be emphasized again, however, that such a prolonged continuation is regarded by the Committee as is derived essentially from mortalities induced at being of the greatest value. The same applies also to the doses in excess of 100 rad. De value appropriate to the studies of uranium miners and to those exposed in childhood or in adult life to local radiotherapy for r 414 I

benign conditions. Further possible studies ofirradiated small increase in a common form of cancer is to be ~ patients are discussed in Annex F with review of the distinguished from the naturalincidence of the cancer in dosimetric information available (para.138) or needed the control population. The raised incidence of thyroid .110) in evaluating the carcinogenic effects of cancer following an estimated dose of 6.5 rad to the gland (para.122) resulted from a study of the records of tion on other organs. about 11000 irradiated children and about 16 000 con-trols. The increase in childhood cancers after irradiation p: i 322. There is an urgent need for establishing a reliable in utero at a probable mean dose of only a few rads z basis for infening the risk from a given absorbed dose of entailed identification of whether x ray examinations ? gew.LET radiation from that resulting from high LET had occurred in utero on 7500 children who died of radiation. Most occupational and enviramental exposure is to radiation oflow LET. However, L.r very malignancies, in Stewart's survey (para. 303). I important sourc:s Of human epidemiological evidence yield risk estimates anly for, or largely in terms of, 326. Extended s:.idies of malignancies in occupa- ?- high-LET radiation, namely, following the irradiation of tionally exposed workers would be informative and bone cells with 88*Ra and 88'Ra, of lung tissues of valuable, if exposures to both external and internal uranium miners with radon daughters. ofliver cells with radiation were adequately known, if an appropriate 1 thorium, and particularly of the whole body in the control population was studied, and if the numbers populations of Hiroshima who were exposed to ah rbed observed-and the time of observation-were sufficient. O#~ doses from neutrons of up to a hundred or more rads. It can be estimated that the number of deaths from Investigations in animals have compared the frequencies " spontaneous" cancer occurring, for example, in 3' with which mutations, chromosomal aberrations and 30 years in a male population of 100000 of age 18-65 cell-killing effects are induced by high-and low-LET (distributed by age as in the general population) would radiations at various dose levels. It appears most be 11200 (as bas d on mortality rates in the United 1 2 important that similar comparisons should be made Kingdom of Greas uitain and Northem Ireland). In a i between the carcinogenic effects of such radiations at group of this size, therefore, and with a sufficient I l different dose levels, and should extend over various control group, cancer induction by radiation would only I animal species (see Annex 1, paragraphs 13 and 339) and be detectable (at the 2 SD levels) if it caused 212 types of tumour, so that the RBE for carcinogenesis, and additional deaths within this period, corresponding to an + any variations of it with dose, should be explored on as induction rate of over 7010-' y~8 from the doses 'J 1 wide a basis as is practicable. received by this population. To evaluate the excess rate with any accuracy would require 1.5-2 times this rate. 323. It would be valuable also if svailable human data unless unusual forms of cancer were detected. Unless the l-uld be examined for any evidence as to the RBE for present estimates of total cancer induction we forms of human carcinogenesis, for example,by the' substantially in error, therefore,it would require studies l, i of epidemiological surveys yielding better values for covering some millions of persen-years of workers sarcoma induction in bone following extemal radio-exposed at annual average whole-body doses of over therapy, or by a more sophisticated intercomparison of I rad, to evaluate directly the risk of such doses. Equal the leukaemia induction rates in Hiroshima and Nagasaki or smaller surveys might, of course, still be valuable in l' o h than has been possible in the present report. excluding higher risks than have been estimated, or = unexpected forms of malignancy due to undetected local ? l 324. The identification of the cell types at risk within tissue irradiation. t j various organs and of the relative importance of homogeneous or non-homogeneous irradiation of a given 327. In surveys of patients who have had diagnostic tissue are essential to an undersunding of the process examinations with radionuclides involving moderate or and of the frequency of tumour induction and to the low tissue absorbed doses, the interpretation of epi- [ estimation of the relevant dose for risk evaluation (see demiological surveys for the induction of malignancies Annex t, paragraph 301). The approximate consistency may involve two problems in particular. In the first of leukaemia risk estimates derived from irradiation of place, it will be difficult to know-without suitable j= the whole marrow or of only a fraction of the marrow, control series-whether the patients' diseases may not be suggests that a given integral dose of moderate size may associated naturally with an excess incidence of certain { be about equally effective whether distributed uniformly types of malignancy. Secondly, it may be difficult to be or not. When smaller fractions of a tissue are exposed to sure that the radionuclide test itself was not occasionally 2 much higher local doses, the effects are likely to depend done to investigate a condition which was in fact due to on the form of the dose-effect relationship at high doses a developing malignance. For thyroid scans, for T and the way in which the yield of tumours falls when example, this may be particularly difficult, since thyroid local doses are very high. The problem has been closely cancers may cause modularity or enlargement of the .=_ "~ examined on a theoretical basis (516) but further gland which, in the absence of operation, may only ? experimental examinations are needed. become diagnosable as malignant many years later. Any l survey of the sequels of diagnostic radionuclide tests 325. The direct determination of the risk to man from should only be undertaken if it is clear that such j posure of the whole body or of particular body tissues difficulties will not necessarily make the results useless low doses presents considerable difficulties, but the when they have finally been obtained. j ommittee wishes to emphasize the importance of any practicable studies that would be likely to yield 328. The potential importance of examining malig- .I statistically reliable information on this question. Large nancy rates in populations livmg in areas of high natural irradiated populations need to be studied, however,if a background radiation has frequently been emphasized, j 415 3

.~

• ~

provided that adequate rnedical records were cbtainable, unlikely that either eituation gives opportunities for and that parallel studies could be made in a comparison useful epidemiological studies. With increased internal ares differing only in being exposed to lower radiation from dietary radioactivity, the known background radiation. However, when the raised background radiction is due to increased extemal populations are small and dosimetric evaluations might (O radiation, populations hitherto identified have been of be difficult to make on a reliable basis, even for the 1 () such a size,in relation to the elevation of background organs that might be selectively irradiated. For l radiation, as to require a very prolonged observation populations living at high altitudes,it would probably be difficult to identify control populations with the same with efficient ascertainment of causes of death to detect cultural, medical and dietary practices. or evaluate an increase in total cancer mortality above that to be expected from natural causes in a comparison Population. 331. Some guidance as to the relative risk of high and l 1' Iow exposures may also be obtainable from any l 329. It has been estimated (117) that an increased practicable analyses of the form of the dose <ffect cancer mortality is likely to be detectable in a relationship for cancer induction. Particular importance considerably shorter time if studied on the basis of attaches, therefore, to any studies of human carcino-deaths of those aged less than a certain optimum age, genesis in which that ts possible, since it will be seen that which is likely to vary with the pcpulation studied, than m much of the work reviewed m this Annex,the risks 2 if based on deaths at all ages, when the rising natural determined at lower doses are evaluated too imprecisely cancer mortality at higher ages is likely to make the f r the likely values at very low dose to be estimated radiation. induced mortality harder to detect. ( A similar with conGdence. 3 argument may apply to surveys of occupationally irradiated groups.) Problems of ascertainment, however, 332. It remains of considerable importance, therefore, continue to be the major difficulty in existing high-background areas with adequate populations, and it that the likely form of the dose <ffect relationships for a may be that studaes on larger populations exposed to number of different types of radiation carcinogenesis d should be closely studied in experimental animals, and only moderate elevations of background radiation rnight preferably in a range of. species (see Annex 1, be more informative if good records of death rates from malignant diseases were already available and had paragraphs 13 and 339), so that any quantitative generalization that might apply to different species previously 1,een maintained in these areas. (For given might be recognized (as appears to hold very rates of natural cancer incidence and of cancer inductionapproximately for the induction of chromosomal per unit dose, the duration of survey required to detect a aberrations (20)). In particular, if the form of such radiation-induced increase in rate should vary inversely dose <ffect relationships could be reliably shown to with the size of exposed population and also inversely depend mainly on terms in dose and (dose)*, with 'th the square of the elevation of background radiation rate as compared with that of the control population.) modification at high doses from cell killmg (as discussed in paragraph 35), and if the coefficient of the linear and 330. Populations are also exposed to raised radiation quadratic terms were determmed for a number of different types of tumour in different species, some levels in certain an. - in which various dietary indication might be obtained of the amount by which constituents are unusually r>Joactive or when living at the risk per unit absorbed dose at moderately high doses altitudes at which increased cosmic radiation causes is likely to exceed that at very low doses for elevation of the extemal radiation background. It seems carcinogenesis in general. f e e t D(v 416 i

I& u l REFERENCES 1. Albert, R. E. and A.R.Omran. Follow.up study

13. Beebe,G.W., H.Kato and C.E. Land. Mortality of patients treated by X. ray epilation for Tinea experience of A-bomb survivors 19501974. 24th Capitis. Arch. Environm. Health 17: 899 918 Annual Scientific Meeting of Radiation Research (1968).

Society, San Francisco, June 1976. i 2. Alderson,M. R. and S.M. Jackson. Long-term

14. Beebe,G.W., H.Kato and C.E. Land. Studies of e

i follow.up of patients with menorrhagia treated by the mortality of A. bomb survivors. 8. Mortality irradiation. Br. J. Radiol. 44: 295 298 (1971). experience of A. bomb survivors, 1950 74. Radia. tion Effects Research Foundation, RERF Techni. 3. Anderson, R.E. and K.Ishida. Malignant lym-cal Report 177 (1977). phoma in si.rvivors of the atomic bomb in Hiroshima. Ann. Int. Med. 61: 853 862 (1964).

15. Belsky,J.L, K.Tachikawa, R.W.Cihak et aL Salivary gland tumours in atomic bomb survivors, 4.

Andrews,P. Rodent ulcers induced by X. rays and Hiroshima-Nagasaki 1957 1970. J. Am. Med. radium. Thesis for M.D., University of London, Assoc. 219: 864-868 (1972). March 1957. i

16. Belsky,J.L, N.Takeichi, T.Yar tamoto et al I

Archer,V.E. and J.K. Wagoner. Lung cancer Salivary gland neoplasms following atomic radia. 5. among uranium miners in the United States. tion. Additional cases and reanalysis of combined Health Phys. 25: 351371 (1973). data in fixed' population. Cancer 35: 555 559 (1975). i 6. Archer,V.E., J. D.Gi!!iam and J.K. Wagoner. I ; Respiratory disease mortality among uranium

17. Boice,J.D.Jr., and R.R. Monson. Breast cancer l

miners. Ann. N.Y. Acad. Sci. 271: 280 293 following repeated fluoroscopic examinations of i the chest. J. Natl. Cancer Inst. (In course of f (1976). publication). o 7. Armenian,11.K. and A.M.Lilienfeld. The distri-bution of incubation periods of neoplastic 17a. Boice, J. D. Personal communication. diseases. Am. J. Epidemiol. 99: 91 100 (1974).

18. Brinkley, D. and J. L Haybittle. The late effects 8.

Arnold, J., S. Pinsky, U. Y. Ryo et al "'"Tc-of artificial menopause by X. radiation. Br.J. pertechnetate thyroid scintigraphy in patients pre. Radiol.42: 519 521 (1969). disposed to thyroid neoplasms by prior radio. therapy to the head and neck. Radiology 115:

19. Brown, J. M. Personal communication (1976).

653-657 (1975).

20. Brown,J.M. Linearity vs non.linearity of dose 8a. Atkinson,H.R. Bone marrow distribution as a n carcinogenesis. Health Phys.

factor in estimating radiation to bloodforming '*'.Po" g 23 45 6~ organs. J. Coll. Radiol. Austr. 6: 149154 (1962).

21. Campbel'. 4... E.W. Emery, J.N.Godlee et al i

9. Auxier, J. A., F. F. Haywood, T. D. Jones et al Diagnosis and treatment of primary polycy. Dosimetry for human exposures, p.79 82 in Oak thaemia. Lancet 1: 1074 1077(1970). Ridge National Laboratory report ORNL 5056 (1975).

22. Casarett,G.W. Radionuclide Car inogenesis. U.S.

10. Baker,H.W. Anaplastic thyroid cancer 12 years Atomic Energy Commission Symposium Series y

J] after radioiodine therapy. Cancer 23: 885 890 No. 29, Washington,1973. (19(9).

23. Chong, G '

O. H. Beahrs, M. L C. Chen et al ] I1. Baral,E., L.E. Larsson and B.Mattsson. Occur. rence of breast carinoma in women trested with Management of parotid gland tumours in infants Ji

x. rays for benign diseases of the breast (in press).

and children. Mayo Clin. Proc. 50: 279 283 j (1975). 1 } - Becker, F. O., S. G. Economou, H. W. Southwick ct al Adult thyroid cancer after head and neck

24. Cihak, R. W., T. lshimaru. A. Steer et al Autopsy 1

irradiation in infancy and childhood. Ann. Intem. findings and relation to radiation. Cancer 33: Med.83: 347 351 (1975). 1580 1588 (1974). j 417 7it. s

25; Conard, R. A., B.M.Dobyns and W.W.Sut:w. nyroid neoplasia as late affect cf cxposure to accidents at nuclear intallations. Unit:d Kingdom radioective iodine in fallout. J. Am. Med. Assoc. Atomic Energy Authority report AHSB (RP)R95, Harwell,1969. 214: 316 324 (1970). m

40. Duncan,W. and G.R.Kerr. De curability of

-[V)

26. Conard,R. A. Acute myelogenous leukaemia fol-lowing fallout radiation exposure. J. Am. Med.

breast cancer. Br. Med. J. 2: 781 783 (1976). Assoc. 232: 13561357 (1975).

41. Evans,R.D. The effect of skeletaHy deposited

27. Conard, R. A. Personal communication (1976).

28. Cook, D. C., O. Dent and D. Hewitt. Breast cancer 41a. Evans, R. D.,' A. T. Keane, R.J.Kclenkow et d foDowing multiple chest Iluoroscopy: the Ontario Radiogenic _ tumours in - the radium and meso-experience. Can. Med. Assoc J. lll: 406 409 and thorium cases studied at M.I.T., p. 157-194 in 412 (1974).

Delayed effects of boneseeking radionuclides (C.W. Mays, W. S. S. Jee, R. D.1.loyd et d, eds.).

29. Court Brown,W.M. and R. Doll. Mortality from University of Utah Press,1969.

cancer and other causes after radiothersp> for ankylosing spondylitis. Br. Med.J. ii: 1327 1332 42. Faber,M. Follow.up of Danish thorotrast cases, (1%5). p.137150 in Pmwdu of the Third Inter. national Meeting on the Toxicity of norotrast.

30. De Groot,Ia J. and E.Paloyan. Thyroid Carcin.

Damsh Atomic Energy Commission report Ris4 oma and Radiation. A Chicago Epidemic. J. Am. 294 (1973). Med. Assoc.225: 487 491 (1973). 4 4

43. Faber,M. De effect of thorotrast enriched with

31. Delarue,N.C., G. Gale and A. Ronald. Multiple 2"Th. p. 294 302 in Proceedings of the Third fluoroscopy of the chest: carcinogenicity for the

. international ofecting on the Toxicity of Thoro. 1 A female breast and implications for breast screening trast, Danish Atomic Energy. Commission report programs. Can. Med. Assoc. J.112: 1405 1413 Riss 294 (1973). (1975). 44.

32. Dickson, R. J. De late results of radium treatment Faber, M. Dose effect relations in hepatic carcino.

for benign uterine haemorrhage. Br.J. Radiol.42: genesis. p.308-315 in Proceedings of the Third 582 594 (1969). International Meeting on the Toxicity of Thoro. trast, Danish Atomic Energy Commission report

33. Dobyns, B. M., G. W. Sheline, J. B. Workman et d Risp 294 (1973).

Malignant and benign neoplasms of the thyroid in 45. patients treated for hyperthyroidism: a report of Favus, M. J.. A. B. Schneider, M. E. Stachura et d the Coop:rative nyrotoxicosis Therapy Follow-Thyroid cancer occurring as a late consequence of up Study. J. Chn. Endocrinol. 38: 976 998 head-and. neck irradiation. New Engl.J. Med. 294: (1974). 1019-1025(1976). i

34. Doll, R. and P. G. Smith. The long. term effects of 45s. Finkel, A.J.,

C. E. Miller and R. J. Hasterlik. x-irradiation in patients treated for metropathis Radium-induced malignant tumors in children and haemorrhagica. Br. J. Radiol. 41: 362 368 (1968). adults. p. 195-225 in Delayed effects of bone. seeking radionuclides (C.W. Mays, W.S.S.Jee,

35. Doll, R., C. Muir and J. Waterhouse (eds.). Canar R. D. Lloyd et d, eds.). University of Utah Press, 1969.

incidence in 5 continents Vol.H. International J U. ~ 'n Against Cancer Springer Verlag Berlin, 1970.

46. heuen, A. Demonstration eines Cancroids des recht;n HandrGekens, das sich nach langdauernder

36. Doll,R. and P.G. Smith. Mortality from cancer Einwirkung von R5ntgenstrahlen entwickelt hat.

and other causes af er radiotherapy for ankylosing Fortschr. Roentgenstr. 6: 106111 (1902). t spondylitis: further observations. (To be pub-

47. Gooiden, A.W.G. Radiation cancer. A review with special reference to radiation tumours in the 37.~

Doll,R. Cancer following therapeutic external ax, and thyroid. Br.J. Radiot. 30: irradiation. Paper presented to the 10th Inter. g national Cancer Congress in Houston (1970).

48. Greene,M. De treatment of thyrotoxicosis with

38. Dolphin,G.W. The risk of thyroid cancers I*E"*I3I and partial thyroidectomy. M.D.

following irradiation. Health Phys. 15: 219 228 Thesis, Un.versity of Sheflield,1964. ( (1968). 49. j

39. Dolphin,G.W. and W.G.Marley. Risk evaluation Gouskova, A. K., V. A. Soldstova, E. A. Denisova et d late effects of occupational irradiation.

t to the protection of the public in the event of Gov. Report 34 40 VII,1975. j 418 m C. -++>yeg

3mog, g,nw,

,,.,---,9-w,yryyg wp,. w w. pg, w e,ws.-,,pr,, cyes.y, gee,-c.,,-.---,s- ,-e,,.a,- .e,-way, ,,w.emenu+i+ce%wme,n.m-te--o=.ep.-w.*'ee=Nerm-To-e m-w we-+g--e-g-*

a. mea--T-m-*e*-e e

~ ll l u

63. Ishimaru, T.,

T. Hoshina, M. Ichimaru et al. Ham,J. M., F. R.Hume, L C. Gray et cl Report $0. J of the Royal Commission on th Health and leukaemia in atomic bomb survivors, Hiroshima Safety of Workers in Mines. Government of and Nagasaki,1 October 1950 30 September r 1966. Radiat. Res 45: 216-233 (1971). 4 Ontario, Canada,1976. Hamrick,P.E. and P.J.Walsh. Environmental

64. Ishimaru T., R.W.Cihak, C. E. Land et al Lung cancer at autopsy in A bomb survivors and Radiation and the Lung. Environ. Health Perspect.

controls, Hiroshima and Nagasaki, 1961 1970. H-9:33 52(1974), Smoking, occupation and A. bomb exposure. i

52. Harley, N. H.,

R. E. Albert, R. E. Shore et al Cancer 36: 17231728 (1975). I l Follow-up study of patients treated by X. ray epilation for tinea capitis. Estimation of the dose

65. Jablon,S. and H.Kato. Childhood cancer in to the thyroid and pituitary glands and other relation to prenatal exposure to Atomic bomb I

structures of the head and neck. Phys. Med. Biol. radiation. Lancet 2: 1000(1970). 21: 631 642 (1976). I

66. Land, C. E.Penonal communication.

53. Hashizume,T. and T.Maruyama. Physical dose estimates for A. bomb survivors-Studies at Chiba,

67. Jablon,S. and H.Kato. Mortality among atomic bomb survivors, 1950 1970 Atomic Bcmb Casu.

Japan. J.Radiat. Res., Vol.16, Supplement 12: alty Commsssion report ABClu R 10 71 (1971). 12 23 (1975). l 53a. Hashizume,T., T.Maruyama, K.Nishizawa et al

68. Janower,M.L and O.S.Miettinen. Neoplasms Mean bone marrow dose of atomic bomb survivors after childhood irradiation of the thymus gland.

in Hiroshima and Nagasaki. J.Radiat. Res.18: J. Am.Med Assoc.215: 753 756(1971). 67 83(1977).

69. Janower,M.L Follow-up of Thorotrast patients 1

54. Hiybittie, J. L Personal communication.

from Bostm, Massachusetts and Ann Arbor,

55. Hazen, R. W., J. W. Pifer, E. T. Toyooka et al Michigan, UaA p. 126 136 in Proceedings of the Third international Meeting on the Toxicity of Neoplasms following irradiation of the head.

Dorotrast. Danish Atomic Energy Commission Cancer Res. 26: 305 311 (1966). report Risd No.294(1973).

56. Hempelmann, L H., J.W. Pifer, G. J. Burke et al

70. Jone T.D. A CHORD Simulation for insult Neoplasms in persons treated with X. rays in assessment to the red bone marrow. Oak Ridge infancy for thymic enlargement. A Report of the National Laboratory report ORNL-519) (1976).

O. nird Follow-up Survey. J.Nati. Cancer Inst. 38: 317 336(1967).

71. Jones,T. D... J. A. Auxier, J. S. Cheka et al in vivo dose estimates for A. bomb survivors shielded

57. Hempelmann,L H. Risk of thyroid neoplasms by typical Japanese houses. Health Phys. 28:

after irradiation in childhood. Science 160: 267 381 (1975). 159 163 (1968).

72. Kaick.G. van and K.E.Scheer. Actual status of j

58. Hempelmann, L H., W.J. Hall, M.Phillips et al the German Thorotrast study, p. 157162 in Neoplasms in persons treated with X-rays in Proceedings of the Third International Meeting on f

infancy: fourth survey m 20 years.J. Natl. Cancer the Toxicity of Thorotrast. Danish Atomic Energy Inst. 55: 519-530 (1975). Commission port Rise No.294 (1973).

59. Holford,R.M. The relation between juvenile

73. Kato,H. Mortality in children exposed to the I

cancer and obstetric radiography. Health Phys. 28: A. bomb while in utero. Am.J. Epidemiol. 93: 153 156 (1975). 435ff(1971). I

60. Hutchison.G. B. leukaemia in patients with Kato,H. Personal communication on work in 74.

cancer of the cervix uteri treated with radiation. A report covering the first five years of an preparation by M. Tokunaga, M. Asano, S.Tokuoka et al. Malignant breast tumours and international study. J. Natl. Cancer Inst. 40: their relation to radiation. Hiroshima and 951 982(1968). Nagasaki,1950 74

61. Ichimaru,M., T.Ishimaru, J. L Belsky et al incidence of leukaemia in A-tomb survivors by

75. Kaul, A. Mean organ dose rates in man following intravascular injection of thorotrast, p.40 51 in dose, years after exposure, age and type of Proceedings of the nird International Meeting on leukaemia 1950 71, Hiroshima and Nagasaki.

the Toxicity of norotrast. Danish Atomic Energy RERF Technical Report 10 76. Commission report Rise No. 294 (1973). O. International Commission on Radiological Prote.e-tion. Report of the Task Group on Reference

76. Kellerer, A.M. and H.H. Rossi. RBE and the Man: ICRP Publication 23. Pergamon Press, primary mechanism of radiation action. Radiat.

Res. 47: 15 34 (1971). Oxford,1975. 419 eh

I

77. Kikuchi, A., C.Watanabe, M. Abe et al Head and '

89a.Matanoski,G.N., R.Seltser, P.E.Sartwell et at neck cancer following therapeutic irradiation and a The current mortality rates of radiologists and brief review on those in Japan. Nippon Acta other physician specialists: specific causes of Radiologica 34: 491501 (1974). death. Am.J.Epidemiol. 101: 199 210(1975). ^

78. Kitabstake, T., S. Kurokawa, M. Yamasaki et al A

90. ludical Joumal of Australia 1: 1171 (1973).

I( f prospective survey on the incidence of chest lodine.131 and caremogenesis (editorial). malignancies after repeated fluoroscopy during artificial pneumothorax therapy for pulmonary

91. Mettier, F. A., L H. Hempelmann, A. M. Dutton tuberculosis. Nippon Acta Radiologica 35:

895 899 (1975). et al Breast neoplasms 'in women treated with 1,

x. rays for acute postpartum mastitis, a pilot study.

79. Kolibe.D., R. Masse, P. Fritsch et al Greffe chez J. Natl. Cancer Inst. 43: 803-811 (1969).

~ la souris athymique de tissu pulmonaire de rat

92. Modan, B., D. Baidatz, H. Mart et al Radia; ion.

1 contamind par l'oxyde de plutonium, p. 95107 in induced head and neck tumours. Lancet 1: t Biological and Environmental Effects of low-277 279 (1974). 1.svel Radiations, Vol.2. IAEA publication STIl i PUB /409. Vienna,1976. t

93. Mole,R.H. Antenatal, irradiation and childhood j

1 cancer: causation or coincidence? Br.J. Cancer

80. Lindsay,S. and L L. Chaikoff. De effects of irradiation of the thyroid gland with particular 30: 199 208 {l974).

i reference to the induction of thyroid neoplasms: a i review. Cancer Res. 24: 10991107 (1964). ?4. Mole, R.H. lonizing radiation as a carcinogen: l practical questions and academic pursuits. j

81. Lundin, F.E.,

J.K. Wagoner and V.E. Archer. Br.J. Radiol. 48: 157169 (1975). 5 Radon daughter exposure and respiratory cancer,

95. Mole, R. H. Personal communication.

I quantitative and temporal aspects. U.S. Depart-ment of Health, Education and Welfare, National l Institute for Occupational Safety, and National

96. Mori,T., Y.Nozue, T.Miyazi et af Thorotrast Institute of Environmental Health Sciences joint injury in Japan, p.175192 in Proceedings of the monograph I (1971).

Third International Meeting on the Toxicity of j 4 Dorotrast. Danish Atomic Energy Commission report Risd No.294 (1973).

82. Mackenzie,1. Breast cancer following multiple fluoroscopies. Brit.J. Cancer 19: 1 8 (1965).

97. Moriyama,I. M. and H. Kato. Mortality experience of A. bomb survivors 1970 1972, 1950-1972;

83. MacMahon. B. Prenatal X. ray exposure niid child-hood cancer. J. Natl. Cancer Inst. 28: 11731191 JNIH.ABCC Life Span Study Report 7, Technical (1%2).

Report 15 73.

84. MacMahon, B. and A. B. Hutchison. Prenatal X. ray

98. Munk, J., E. Peyser and J. Grusskiewicz. Radiation and childhood cancer: a review. Acta Unio. Int.

induced intracranial menmgiomas. Cim. Radiol. 20: 90 94 (1969). i Contra Cancrum 20: 11721174 (1964).

85. MacMahon, B., T. M. Lin, C. R. Iowe et al Lacta-

99. Myrden, J. A. and J. E. Hiltz. Breast cancer follow.

l 4 tion and cancer of the breast. A summary of an ng multiple fluoroscopies durmg artificial pneu-international study. Bulletin of the Wodd Health mothorax treatment of pulmonary tuberculosis. l Can. Med. Assoc. J. 100:1032 1034(1969). Organization 42: 185194 (1970). i

86. Mayneord, W. V. and R. H. Clarke. Carcinogenesis

10. Myrden,J. A. and J.J.Quinlan. Breast carcinoma and radiation risk: a bio. mathematical reconnais-following multiple fluoroscopies with pneumo.

thorax treatment of pulmonary tuberculosis. sance. Br. J. Radiol., Suppl. No. 12,1975. Annals of Royal College of Physicians and i

87. McDougall, L R.,J. S. Kennedy and J. A. Thompson.

Surgeons of Canada 7: 45 (1974). i Thyroid carcinoma following iodine.131 therapy. Report of a case and review ~of the hterature. 101. Nakamura, K. Personal communication. J. Clin. Endocrinol. 33: 287 292 (1971). 102. Newwmbe, H. B. and J. F. McGregor. Childhood

88. McGregor, D. H., C. E. Land, K. Choi et al Breast cancer following obstetric radiography. Lancet 2:

i151 1152 (1971). cancer incidence among A-bomb survivors, Hiroshima and Nagasaki, 1950 69. Atomic Bomb I Casualty Commsesson, Technical Report 32 71. 103. Newill,V. A. Distribution of cancer mortality among ethnic sub groups of the white population of New York City, 1953 1958. J. Natl. Cancer 9. McIntyre.D. and R.C.S. Pointon. Vesical neo. Inst. 26: 405-417 (1961). plasms occurring after radiation treatment for carcinoma of the uterine cervix. J. Royal College 104. Nishiyama, H., R. E. Anderson, T. Ishimaru et al of Surgeons. Edinburgh l 6: 141 146 (1971). The incidence of malignant lymphoma and 420

roultiple myeloma in Hiroshima and Nagasaki 120. Smith,P.C., R. Doll and E.P. Radford. Cancer A. bomb survivors 1945 1965. Cancer 32: mortality among patients with ankylosing spon. 1301 1309(1973). dylitis not given x ray therapy (in press). I e '05. Nitze,H.R. Radiogene Tumoren in H.N.O. Be-121. Raven, R.W. and V. B. levison. Radiatio *: cancer r reich. Archiv fur Klinische und-Experimentelle of the pharynx. Iancet 2: 683 684 (1954). I (O Ohren., Nasen-und Kehlko' fheilkunde 199: p 122. Raventos, A. and T.Winship. The latent interval l 634 638(1,971). for thyroid cancer following irradiation. Radiology 106. Osborn,S. B. Radiation doses in radiographic 83: 501 508 (1964). 1 ,i pelvimetry. Br. J. Radiol. 24: 174 (1951). 123. Refetoff,S., J. Harrison, B.T.Karanfilski et al 107. Palmer,J.P. and D.W.Spratt. Pelvic casemoma Continuing occurrence of thyroid carcinoma after following irradiation for benign gynaecological irradiation to the neck in infancy and childhood. D diseases. Am. J. Obstet. Gynaecol. 72: 497 505 New Engl.J. Med. 292: 171 175 (1975). t (1956). 124. Registrar General's Statistical Review, England and t j i 108. Paloyan,E., A.M. Lawrence, M.H. Brooks and Wales, for 1968 70. Supplement on Cancer. Her J.R.Picieman. Total thyroidectomy and para. Majesty's Stationery Office, lendon,1975. i p thyroid autotransplantation for radiation asso. 125. Rossi, H. H. and A. M. Kellerer. Radiation carcino. ( ciated thyroid cancer. Surgery 80 70 76(1976). genesis at low doses. Science 175: 200 202 (1972). f 'l 109. Parker,l N. J.L Belsky, T.Yamamoto et aL 126. Rowell,N.R. Microanatomy and thickness of the Thyroid carcinoma diagnosed between 13 and skin. Br. J. Radiol. 45: 610 (1972). [ 26 years after exposure to atomic radiation. A study of the ABCC.JNIH Adult Health Study 127. Rowland, R.E., P.M. Failla, A.T.Keane et al p Population Hiroshima and Nagasaki 1958 71. Some dose. response relationships for tumour k Technical Report 5-73. incidence in radium patients, p.117 in Radio- [ I gical Physics Division, Center for Human 110. Pegum,J.S. Radiation induced skin cancer. Br.J. Radiobiology Annual

Report, July 1969-Radiol.45: 613 614 (1972).

June 1970. U.S. Atonn,e Energy Commission a report ANL.7760(Pt.2). ill. Petersen,0. Radiation cancer. Report of 21 cases. p Acta Radiol.42: 221236 (1954). 128. Saenger,E.1-., F. N. Silverman, T. D. Sterling et ( al Neoplasia following therapeutic irradiation for N 112. Pifer, J.W., E. T. Toyooka, R. W. Murray et al benign conditions in childhood. Radiology 74: Neoplasms in children treated with X. rays for 889 904 (1960). thymic enlargement. I. Neoplasms and mortality. J. Natl. Cancer Inst. 31: 13571377 (1963). 129. Safa. A.M., O.P. Schumacher, A. Rodriguez-Ant 6nez. Long-term follow.up results in children and adolescents treated with radioactive iodine 113. Placek,V. and J.5 eve. K otizce rizika rakoviny (i dine.131) for hyperthyroidism. New Engl. J. hrtanu u hornika uranovych doln. Prac. Lik. 27: Med. 292: 167171 (1975). 113-115 (1975). l 130. Sampson, R.J., C. R. Key, C. R. Buncher et al 114. Pochin,E.E. Untoward effects of radiciodine Prevalence of thyroid carcinoma at autopsy, treatment of hyperthyroidism. Minerva Nucleare Hiroshima 1957 1968, Nagasaki 1951 1967. 8: 106 110 (1964). Atomic Bomb Casualty Commission report ABCC.TR 25 58(1968). 115. Pochin E.E. Prospects from the treatment of thyroid carcinoma with radiciodine. Clin. Radiol. 131. Schulz, R.J. and R. E. Albert. III. Dose to organs 18: 113 125 (1967). of the head from the x-ray treatment of tinea capitis. Arch. Environ. Health 17: 935 950(1968). 116. Pochin,E.E. Frequency ofinduction of malignan. l-cies in man by ionizing radiation, p. 341355 in 132. Segi,M., M.Kurihara and T.Matsuyama. Cancer Handbuch der medizinischen Radiologie. Springer mortality for selected sites in 24 countries Verlag, Berlin,1972. 1964-1965, No. 5 Sendai, Japan, Dept. of Public Health. Tohoku Univ. School of Med.,1969. 117. Pochin, E. E. Problems involved in detecting increased mahgnancu. rates in areas of high natural t radiation background. Health Phys. 31: 148 151 133. Seideman, H. Cancer death rate by site and sex for rehgious and socio. economic groups in New York n (1976), City. Environ. Res. 3: 234 250 (1970). '{ (~ lI8. Pochin E.E. Personal communication. 134. Seltser.R. and P.E.Sartwell. The influence of I19. Pointon, R. C. S. Personal communication. occupational exposure to radiation on the P 421 i i i

? I mortality of American radiologists and cther 148. St: wart A.M. Measuring latent periods of juven. I medical specialists. Am. J. Epidemiol. 81: 2 22 (1965). ile cancer. Proc. R. Soc. Med. 62: 684 (1969). 135. Seve,J. and. V.Plaaek. Radiation induced lung 149. Stewart, A.M. and G.W.Kneale. letter to the I editor. Lancet ii: 1190(1970). p cancer; relation between lung cancer and long term g exposure to radon daughters. Proceedings of Sixth 150. Stewart, A.M. and G.W.Kneale. Radiation dose Conference on Radiatien Hygiene, CSSR,1973. effects in relation to obstetric X-rays and s childhood cancers. Lancet li: 1185 1188 (1970). 136. Seve,J., E.Kunz and V.Plaaek. Lung cancer in uranium miners and long-term exposure to radon 151. Stewart, A.M. Cancer as the cause of abortion and daughter products. Health Phys. 30: 433-437 stillbirths; the effects of these early deaths on the (1976). recognition of radiogenetic leukaemias. Br. J. Canar 23: 465 (1973). 137. Sevcovi, M., J. Seve and J. Thomas. Alpha irradia-tion of the skin and the possibility oflate effects 152. Stewart, A.M. Factors controlling the recognition (in press). of leukaemia and childhood cancer in Health Physics and Healing Arts. 7th Media Topical 138. Shore, R. E. Personal communication (1976). Symposium of the Health Physics Society. U.S. Department of Health, Education and Welfare, 138a. Shore, R., L Hempelmann, E.Kowaluk et al 1973. Breast neoplasms in women treated with x-rays for e postpartum mastitis.J. Natl.CancerInst (m 152a.Takeichi, N., F. Hirone and H. Yamamoto a ~ Salivary gland tumours in atomic bomb survivors. Hiroshima, Japan. I. Epidemiological observations. 139. Shore, R. E., R. E. Albert and B.S.Pasternack. Carner 38: 2462 2468 (1976). Follow-up study of patients treated by X-ray epilation for tinea capitis. Arch. Environ. Health I53. Toyooka, E. T., J. W. Pifer, S. L. Crump et al. 31: 21 28 (1976). Neoplasms m, children treated with X rays for thymic enlargement. II. Tumour mcidence as a 140. 3ilink,K., J.Soumar and J.Nemec. Occurrence function of radiation factors. J. Natl. Cancer inst. and pathogenesis of childhood thyroid cancer. 31: 139137W63). ~ Neoplasma 20: 465 469 (1973). 154. United Nations. General Assembly. Report of the 3 United Nations Scientific Committee on the 141. Silva Horta,J.da. Tumours developed in people Effects of Atomic Radiation, 1964. Official injected with shorium dioxide (Thorotrast)(Portu-Records of the General Assembly, Nineteenth guese experience). p. 233 258 in Proceedings of Session. Supplement No.14 (A/5814). the Third lntemational Meeting on the Toxicity of Thorotrast. Danish Atomic Energy Commission 155. United Nations. A report of the United Nations report Rise No. 294 (1973). Scientific Committee on the Effects of Atomic Radiation to the General Assembly,with annexes. 142. Smith, P. G. and R. Doll. I2te effects of X.irradia. Vol.II: Effects (United Nations publication, sales tion in patients treated for metropathia haemcrrh. No. E.723X.18), New York,1972. agica. Br. J. Radiol. 49: 224-232 (1976). ,157. Vanderlaan,M., F.J. Burns and R.E. Albert. A 143. Smith, P. G. Personal communication. m del describing the effects of dose and dose rate l on tumour induction by radiation in rat skin. l

p. 253 263 in Biological and Environmental 144. Snihs,J.O. The approach to radon problems m.

Effects of Low. level Radiation. Vol.1. IAEA e non-uranium mines in Sweden, p.900 911 in Proceedings of Third International Congress of the publication STI/PUBJ409, Vienna,1976. l IRPA. U.S. Atomic Energy Commission report CONF 730907 (1973). 158. Wanebo, C. K., K. G. Johnson, K. Sato et al { Breast cancer after exposure to the atomic 145. Spiess,H. and C.W. Mays. Bone cancers induced bombing of Hiroshima and Nagasaki.New Engl.J. by 224-Ra (7hX) in children and adults. Health Med. 279: 667 671 (1968). Phys.19: 713 729 (1970). 159. Werner, A., B.Modan and D.Davidoff. Doses to 146. Spiess,H. and C.W. Mays. Protraction effect on brain, skull and thyroil following x. ray therapy for tinea capitis. Phys. Med. Biol.13: 247 258 l bone sarcoma induction of 224.Ra in children and (1968). adults, p. 437 450 in Radionuclide Carcinogenesis. l U.S. Atomic Energy Agency Symposium Series 160. Werner, A., B.Modan and E. Ron. 7hyroid dosi-No. 29, Washington,1973. metry re-evaluation after treatment for scalp tmea. (p Communication given at 14th International 147. Spittle, Margaret F. Personal communication Conference on Medical Physics, July 1976, (1976). Ottawa. 422 . ~

pr [. 161. Whitton,J. R. New values for epidermal thickness 166. Yamamoto,T. and T.Wakabayashi. Bone tumoun and their importance. Health Phys. 24: 18 (1973), among atomic bomb survivors of Hiroshima and Nagasaki. Acta Pathol. Jpn. 19: 201 212 (1969). 162. Wilson,S.M., C.Platz and G.M. Block. Thyroid 5 carcinoma after irradiation. Arch. Surg.100: 167. Yoshizawa,Y. and T.Takeuchi. Search for the s \\j) '/ 330 337 (1970). lowest irradiation dose from literatures on 163. Wood,J.W., H.Tanagaki, S.Nerrishi et al Thy-radiation induced cancer in pharynx and larynx. roid carcinoma in atomic bomb survivon. Nippon Acta Radio!ogica 34: 903 909 (1974). l Hiroshima and Nagasaki. Am. J. Epidemiol. 89: 4 14 (1969). 168. Yoshizawa,Y. and T.Kusama. Search for the l i 1 164. Woolner, L. B. et al Occult papalary carcinoma of lowest irradiation dose from literatures on the thyroid; a study of 140 cases observed in a 30 radiation induced breast cancer. Nippon Acta year period. J. Clin. Endocrinol. 20: 89 105 Radiologica 35: 11251130(1975). (1960). 165. Yamamoto,T. and T.Wakabayashi. Bone tumoun 169. Yoshizawa,Y., T. Kusama and K. Morimoto. l among atomic bomb survivors, 1950 1965, Search for the lowest irradiation dose from Hiroshima and Nagasaki. Atomic Bomb Casualty literatures on radiation induced bone tumoun. . Commission report ABCC.TR 26 28 (1968). Nippon Acta Radiologica 37(4)(1977). i s e i f% c 1 1 e\\ 1 ) I. t 'v) \\ 423 _-}}