Document G08xdrEw6ejwrqxvDakBQGz4

EPIDEMIOLOGICAL EVIDENCE Oh' T H E NATURE OF HODGKIN'S DISEASE BRIANMACMAHONM,.D. THE question of whether the disease processes of Hodgkin's disease are fundamentally neoplastic or inflammatory in nature remains unresolved. T h e opinions of pathologists range from the belief that Hodgkin's disease, reticulum-cell sarcoma, and lymphosarcoma are neoplasms originating in a common stem cell and differing only in dominant cell type:! 2%32 to the concept that the tissue changes cannot be differentiated from the granulomatous changes found in tuberculosis, brucellosis, syphilis, and a number of other known infections.21331 Many comprehensive reviews of t h e t o p i c have been assembled.4, 10,11, 28, 31 Recently the interpretation favoring the neoplastic concept of the disease has prevailed. In the Sixth Revision of the International Lists of Diseases and Causes of Death, published in 1948, Hadgkin's disease was included for the first time in the chapter on neoplasms.34 Two facts in p a r t i d a r have favored the development of the neoplastic theory-considerable experimental study has iailed to demonstrate the presence of an infective agent with any degree of regularity, and there has been increasing realization of the frequency with which both Hodgkin's granuloma and the frankly neoplastic Hodgkin's sarcoma may be TABL1E MEAN ANNUAL INCIDENCE OF HODGKIN'S DISEASE I N THE WHITE POPULATION OF RROOCKLYN, 1943 TO 1952, ACCORDING `P3SEX AND AGE O F PATIENTS Age at diag- --Number patients nosis hl F Total Incidence per million M F Total 0-4 5-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 S..S -..S ~ 60-64 65-69 70-74 75 and more ALL AGES 000 2 13 3 58 16 18 34 29 16 45 27 33 60 24 24 48 23 19 42 26 19 45 20 12 32 25 17 42 37 i n 47 .jo i7 47 28 16 44 20 10 30 -11 -8 -19 321 225 546 0.0 2.2 3.5 18.3 29.4 26.1 24.3 22.9 26.8 22.7 30.2 54.1 54.7 69.7 82.3 53.6 25.7 0.0 1.1 6.2 20.0 15.1 29.3 21.9 17 1 18.7 13.4 32is0i ..025 36.8 36.1 30.1 17.4 0.0 1.7 4.8 19.2 22.0 27.8 23.0 19.9 22.7 18 0 25.3 34.8 43.0 52.6 57.7 40.3 21.5 It is the purpose of the present communication to present data on some ieatures of the descriptive epidemiology of Hodgkin's disease and to discuss their relevance to the question of the nature of this disease. found in the same patient at the same or different time~.~,1N*either of these features is conclusive however, and the question cannot be considered settled with any finality. MATERIAL Original data are derived from a study in which an attempt was made to assemble clinical and pathological records of a11 residents of From the Department of Environmental Medicine and Community Health, State University of New York, Downstate Medical Center, Brooklyn, New Pork. Read before the Epidemiology Section of the American Public Health Association, Atlantic City, New Jersey, November 14, 1956. This study was aidrd by a field investigation grant (CS9455) from the National Cancer Institute, of the National Institutes of Health, Public Health Service. The author is indebted to the medical, administrative, and record-room staffs of many hospitals in Brooklyn and other boroughs of New York City. A list of these hospitals was published in a previous communication?` The author is also indebted to the Board of Health of New York City, New York, for permission to review death certificates and to the Bureau of Records and Statistics of the Department of Health, New York City, for providing tabulated information. Received for publication December 26, 1956. Brooklyn, New York, diagnosed as having leukemia or any lymphoma in the period 1943 to 1952. Sources oT data were: (1) a complete survey of the records of thirtyfour Brooklyn hospitals; (2) a selective survey of records of certain hospitals i n other boroughs of New York City; and (3) death certificates of Brooklyn residents dying in any of the five boroughs of New York City. T h e methodology has been more fully described e l s e ~ h e r e . ~ ~ With respect to Hodgkin's disease, records of 573 Brooklyn residents diagnosed for the first time in the period 1943 to 1952 were obtained. When clinical records were found, only pa- 1045 1046 CANCESReptember-October 1957 VOl. 10 60. z i0 - 50. `I a 40. w a w 30;. c d -B R O O K L Y N _ _ - - _U S A z 0 J I 11 w a W v z W Q Vz J 4: 3 z z< w o -o o- m ~o w ~ o w~ o ~m ob m ho y w w m m ~ ~ AGE A T DIAGNOSIS (BROOKLYN) OR D E A T H (UU.S.AA) 2 AGE A T ONSET FIG. 1. Hodgkin's disease: incidence in Brooklyn, 1943 to 1952, and death rate in the United States, 1949 to 1953. EX. 2. Age-specific-iricidence curves for Hodgkin's disease: data from three previous morbidity surveys. Sources of data are Dorn and Cutler (t1,S.A.): Clemmesen, Busk, and Nielaen (DenmaIk)? and Uddstromer ( tierits for whom the diagnosis was supported by biopsy or autopsy were included in the series. A number of patients for whom clinical records were not found, but for whom tlie item "duration of illness" as completed on the death certificate indicated onset within the study period, were included. T h e major criteria .of diagnosis for these patients were as follows: autopsy report, 119; biopsy ieport, 382; and death certificate, 72. hlost 01 the following tabulations are based .on the 546 white patients included in this series. T h e total white population of Rrooklyn averaged 2,540,828. INCIDENCE IN BROOKLYN Mean annual incidence of Hodgkin's disease in the white population of Brooklyn during 1933 to 1952 was 21.5 per million population. Table 1 gives the mean annual-incidence rates according to sex and age a t diagnosis. T h e age trend is illustrated in Fig. 1. T h e striking feature is the bimodality of the curve. There is one peak of incidence in the 25 to 29-year age group, and a second at 70 to 74 years. iZnriual age-specific-incidence rates taken from, or calculated from, data given in three previous reports in which cases from clinical sources can be related to specific populations are illustrated in Fig. 2. T h e studies are those of Uddstromer,30 Clemmesen et al.,3 and Dorn and Cutler,o which presented all ascertained cases 01 Hodgkin's disease in Sweden, 1915 to 1931; Denmark, 1943 to 1947; and ten metropolitan areas of the United States, 1947, respectively. While the individual curves are quite different from each other, each is consistent with the suggestion of one peak of incidence in young adults and a second peak at older ages. i\lthough still present, the bimodal pattern is not so distinct in the material from the state of Washington presented by Thorson and Browm29 T h e number of cases is smaller, however, and the use of ten-year, instead of five-year, age groups tends to obscure the bimodality. DEATHRATEIN THE 1JNITED STATES I n addition to the Brooklyn age curve, mean annual age-specific-death rates from Hodgkin's disease in white persons in the United States are illustrated in Fig. 1. Data are from Vital Statistics of the United States kor 1919 to 1953.23Once again, the curve is definitely bimodal, although the first mode is less distinct than in the curve for the Brooklyn data. T h e same phenomenon, of a sharper peak of incidence in the 20- to %year age group in rates calculated from age at onset or diagnosis, than in rates plotted against age at death, is seen when the Danish data recorded by Clemmesen et al.3 (Fig. 2) are compared to age-specificdeath rates in the same country (Fig. 4). It may be attributable in part to tlie longer sur- Nu. 5 EPIDEMIOLOGICEAVILDENCOEN HODGKIND'SISEASE * kfnrl%fahOTl viva1 of the group ol patients age 0 to 39 than of the group age 40 years or more. Bimodal age-incidence curves are unusual and suggest either that different etiological mechanisms are operating at different periods of the age span, or that different disease entities have been incorporated in the original classification. Furiher analyses were therelore conducted to determine whether the disease occurring in young adults and in older age groups has similar demographic characteristics. 80 70 60 z 40. J 5J 30. z a 1047 SEX INCIDENCE Three hundred twenty-one (58.8 per cent) :I -MALE -__---- FEMALE of the 546 white patients in the Brooklyn ma- terial were males. Age-specific-incidence rates for the sexes separately are given in Table 1 and illustrated in Fig. 3. T h e higher incidcnce 1 T in males is more evident in the later part of the age-incidence curve (40 and more) than in 2?*~- G t UoN l wr r Wo P &O O l l O lo O l o ~ 8 ~ the earlier part (0 to 39). T h e same situation AGE A T DIAGNOSIS is evident in the data of Uddstromer, who rioted that after the fourth decade, the inci- FIG.3. Sex- and age-specific-incidencc rates of Hodgkin's disease in Brooklyn, 1943 to 1952. dence tended to increase in males but not in Shimkin,2* examining United States death females. In Table 2 sex ratios are compared statistics for 1925 to 1950, noted a charac- according to a threefold grouping by age at teristic change in the sex ratio of deaths from onset in the five series in which the data were Hodgkin's disease with age. T h e r e was a available on a community basis. T h e age group marked male predominance in the age group 0 to 9 years is separated because of the observa- 5 to 14 and a lesser peak at 45 to 54. Interpre- tion of Wallhauser,31 confirmed by Smith,27 tation of changes in sex ratio according to age that males predominate to an unusual degree a t death is difficult, since, i n addition to among patients in this age group. This feature changes that are associated with age a t onset is evident from Table 2. I n addition, the sex or diagnosis, the picture will be affected by ratio is lower in the 10 to 39- than in the 40 sex differences in duration of disease aiter diag- and more age group in each series. T h e differ- nosis. There is general agreement that such ence is significant in three of the individual differences exist in Hodgkin's disease in the series and highly so in the combined total for form of a longer expectation of life for females the five series. than for male^.^^^ 25 However, to assess the TABL2E SEX RATIOS, PERCENTAGE MALE, RELATED TO AGE AT ONSET, IN FIVE SERIES OF PATIENTS WITH HODGKIN'S DISEASE .. ____I___^.__ ____ Age groups ~- Series 0-9 Nu. 7; 10-33 (1) No. ~ ~~ /o 40 and more (2) ~ No. 7c. All ages* - No. _____%__ Difference between (1) %6 (2) Uddstroiner30 3 % 91.2 278 53.2 224 6.5.6 536 60.8 Clemmesen et al.3 Dorn and Cutler6 (white patients) 6 . . . 211 54.5 5 . . . 157 48.4 167 58.7 223 62.8 392 5 6 . 6 38.5 56.9 Thorson and B r o w P 5 Present data 3 (white patients) . . . 56 69.6 . . . 237 51.5 72 8 0 . 6 306 64.4 133 75.9 546 58.8 F I V ES E R I E S 53 __8_-1_.9 939 53.2 992 6 4 . 5 1,992 59.7 ... *Figures for "all ages" include 8 patients of unstated age in the series of Clemmesen et d . 3 tDifference exceeds twice its standard error. 12.4 f 4.41 4.2 f 5.1 14.4 f 5.2f 11.0 f 7 . 7 12.9 f 4.3$ 11.3 f 2.07 1048 CANCERSeptem ber-Octobsr 1957 VOl. 10 TABLE3 MEAN ANNUAL DEATH RATE FROM HODGKIN'S DISEASE I N WHITES AND NONWHITES, UNITED STATES. 1949 7'019.53 older age groups acting differentially on the white and nonwhite populations is a ready explanation of this difference. Age 0-39 40 and Death rate per million White Nonwhite* 9.9 7.0 Percent?ge excess In whites 41.4 RELIGION Distribution according to religion, and by implication according to some ethnic tenden- more ALL AGES 32.6 18.0 20.6 11.9 58.3 51.3 *The data for nonwhites are standardized to the age distribution of the white population within each broad age group. cies, is examined in view of the previous demonstration oassociation of leukemia death rate with this variable.16 In Table 4 the distribution by religion of a series consisting of all deaths from Hodgkin's disease in Brooklyn of white Brooklyn residents during 1943 to 1952 effect this will have on sex ratios reIated to is compared with that of a systematic 1-in-200 age at death would require much more de- sample of deaths in Brooklyn of white Brook- tailed information on survival related to sex lyn residents from all causes during the same and age than is presently available. T h e high period. The religious affiliationof the cemetery sex ratio in the 5- to 14-year age group noted of burial as indicated on the death certificate by Shimkin24 is, of course, a reflection of the is used as die index of religion, and has been male predominance in cases with onset in previously shown to be satidactory for this childhood. In the same data, the two age purposc.ls From series of 367 deaths from groups 15 to 24 and 25 to 34 appear to have Hodgkin's disease and 1,119comparison deaths, generally lower sex ratios than any other age patients buried outside the city (six and twenty- groups, except possibly the last two. This three cases), and deaths under 1 year of age again is compatible with the change in sex (zero and fifty cases) have been excluded. The ratio with age at onset noted previously, al- method of examination is more completely de- though, probably for the reasons just outlined, scribed in the study of leukemia referred to the difference between age periods is not so previously. well marked as in the material in which age at Comparing the distributions of the two se- onset was recorded. The reason for the in- ries in Table 4 an excess of deaths of Jew- creasing contribution of females to the total ish patients in the Hodgkin's disease series is death rate seen in the final age groups is not seen. The ratio of the proportions of deaths of clear. It is not evident in age-incidence curves Jewish patients in the Hodgkin's disease and showing age at onset or diagnosis. comparison series suggests that after standard- ization for age differences, Hodgkin's disease COLOR appears as cause of death about one and a half times more frequently among Jewish There were twenty-sevenNegroes among the than among non-Jewish patients. 573 Brooklyn patients. On the basis of the age- specific rates observed in the white population, TABL4E thirty-four Negro patients would have been PERCENTAGE DISTRTRTJTTON. R V expected. The incidence in white persons was apparently higher therefore than in Negroes, RELIGION, OF DEATHS FRoM-HODGKIN'S DISEASE AND A COMPARISON SAMPLE OF ALL DEATHS, BROOKLYN. 1943 TO 19.52 but not so much so as has been evident in previous studies of death statistics.9,24 For ex- Religion, % - ample, in the years 1949 to 1953 age-standard- Age at death* Pt. group No. Cath- Jew- Protespt. olic is11 tant Total ized death rates in white and nonwhite persons Hodgkin's in the United States were 18.0 and 11.9 per 0-39 disease Compari- 115 43.5 35.7 20.8 million per annum respectively-a white ex- 40 and sonsample Hodekin's 68 4 9 . 4 31.9 18.7 cess of 51.3 per cent. more disease Compati- 246 34.2 45.9 19.9 In these same United States statistics, the sonsample 978 42.9 32.2 24.9 ALL Hodgkin's excess of the death rate in whites compared to AGES disease 361 37.1 42.7 20.2 that in nonwhites is somewhat lower in the Comparison sample 1,046 45.0 32.1 22.9 100 100 100 100 100 100 age group 0 to 39 than after that period (Table *In each age group, figures for the comparison sample are stand- 3). However, less complete diagnosis in the ardized t o the age distribution of the Hodgkin's disease patients within the samc group. KO. 5 EPIDEMIOIXIGICALEVIDENCEON HODGKIND'ISSEASE M a c M a h n 1049 TABLE5 countries besides the United States for which DEATH RATES FROM HODGKIN'S DISEASE data are published by the World Health Or- IN SELECTED COUNTRIES, 1950 TO 1952 g a n i ~ a t i o n .W~ ~ith the exception of that for ~~ Death rate per Percentage Japan, all the age trends were found to be - _ _m_illi~on*_ US. rate distinctly bimodal, as noted in the United ________I Age Age Age 40 & All Age 40 & All States data. Death rates standardized to the age distribu- Country 0-39 more ages 0-39 more ages tion of the United States'white population are United States (white) 9.9 3 2 . 6 18.1 100 100 100 Canada 8 . 7 29.1 16.0 88 89 88 Denmark 1 5 . 0 25.7 18.8 152 79 104 Finland 9 . 5 2 1 . 5 13.8 96 66 76 Xorway 11.4 22.8 15.5 115 70 86 Sweden 7 . 9 22.6 13.2 80 69 73 England & \Vales 10.1 25.0 15.4 102 77 85 Ireland 9 . 5 27.8 1 6 . 1 96 85 89 Scotland 10.7 30.2 17.7 108 93 98 France 10.1 20.6 13.9 102 63 77 Germany 9 . 0 21.7 13.6 91 67 75 Italy 10.7 31.1 18.0 108 95 99 Netherlands 17.6 23.7 19.8 178 73 109 Switzerland 12.9 25 7 17.5 130 79 97 Australia 5 . 9 2 1 3 11 5 60 66 64 Japan 1.8 1 3 . 3 5.9 18 41 33 *Within each broad age group all rates have been standardized to the age distribution of the United States white population in five-year groups. shown in Table 5. When the total death rate for all ages is considered, two countries-the Netherlands and Denmark-are seen to have rates slightly higher than that of the United States. T h e total range, however, is small, parLicularly if the data for Japan are excluded. A wider range of variation is evident if rates i n the age groups 0 to 39 and 40 and more are considered separately. There is little correlation between rates for the same countries in the two age groups. For example, the rate for the United States, which is the highest in the 40 and more group, is cxceeded by that of eight of the fifteen other countries in the 0- to 39-year group. T h e Netherlands, with the highest rate in the 0 to 39 period, is ninth in the older age group. I n Table 4 this association is also examined Two interesting patterns are illustrated in separately for the two age groups 0 to 39 and Fig. 4. T h e first, typified by the curves for the 40 and more. T h e excess proportion of deaths Nctherlands and Denmark, and evident also among Jews in the Hodgkin's disease series is seen to be almost entirely restricted to the later age period. T h e difference between the two age group$ in the proportion of deaths in the data for Switzerland, is characterized by high rates in the younger age group and relatively low rates in old age. I n the case of Denmark, this pattern can also be demon- among Jews in the Hodgkin's disease series ap- proaches significance (10.2iz5.6). By contrast, 50 r .,----- the proportion of deaths among Jews in the comparison sample is about the same in the two age periods. `That this is not merely a cir- cumstance of the small number of comparison deaths under 40 years of age is suggested by the data of Deardorff,5 in which only minor changes with age are seen in the proportion of 5 10 Jewish residents of New York City as a whole. a I t is suggested therefore that Hadgkin's disease in patients more than 40 years of age shares with leukemia18 and lymphosarcomalG the feature of more frequent diagnosis among Jewish than among non-Jewish people in this community. T h e disease as affecting the w IQ a5 x4 I- 4wn 3 I I- USA [WHITE) - -- -_-- D E N M A R K NETHERLANDS _ _ _ _J A- P A N younger ages does not share this characteristic, or, if it does, the association is much less marked. 2 I -., --' -*' ' I` ' DEATHRATESIN OTHERCOUNTRIES EGS2g%gikgE$R:2 AGE A T D E A T H Death rates from Hodgkin's disease accord- FIG.4. Age-spccifir4eath rates from Hodgkin's dising to age have been calculated for fifteen ease in four selected counlries, 1950 to 1953. 1050 CANCERSeptember-October 1957 Vol. 10 strated by comparing incidence rates in that young and older adult age groups, it is of in- country (Fig. 2) with the Brooklyn incidence terest to examine for trends in length of sur- data (Fig. 1). vival related to age. Jackson and Parker13 and T h e second inieresting pattern is seen in the Shimkin et al.25 examined this question on se- age trend for Japan. Of course, the complete- ries of 171 and 254 patients respectively. Each ness of the data from this country may be reported n o change in prognosis with age. T h e questioned. Neverthelcss, the disease is diag- latter authors grouped cases in patients less nosed in Japan with some frequency, as evi- and more than 40 years of age without reveal- denced by the fact that the death rate in the ing significant diffcrences in survival. HOW- oIder age groups approximates that of the ever, in the data of Jackson and ParkerI3 a Netherlands and Denmark (Fig. 4) and a num- definitely shorter median survival is seen in the ber of other European countries. T h e virtual oIder age groups, beginning with the group 40 absence of the disease in the age groups less to 49. Some irregularity is evident, but in gen- than 40 is therefore particularly interesting, eral the patients in the age groups more than even when reservations related to diagnostic 40 have median survivals half or less than those difficulties are considered. T h e age trend for in the younger age groups. T h e authors' con- deaths from Hodgkin's disease in Australia clusion that age has no effect on pro<gnosisis shows a pattern somewhat similar to that for unexplained. Uddstromer, examining data on Japan, although the disease is appreciably more 494 patients, reported definitely shorter sur- common in individuals less than 40 years of vival among those in the age groups more than age in Australia than i t is in Japan. 55. Although the trend is most marked at more than age 55, examination of the data reveals SECUI.AR CHANGES IN INCIDENCE definite worsening of prognosis for those as young as 40 to 44. Gall and Mallory8 a190 T h e recorded death rate from Hodgkin's noted much shorter survival for patients who disease has increased progressively in the were more than 50 years of age at onset. United States since 1921.** There was n o sig- In the present series, follow-up to the end nificant increase in Australia between 1908 and of 1952 was available for 77 per cent of the 1950.14 I n the United States death rates from 474 white patients ascertained from hospital Hodgkin's disease in childhood have decreased, records. (Patients ascertained from death cer- in contrast to the regular increases observed in tificates are excluded since no information on all other age groups. However, adult age time of diagnosis was availabIe for this group groups more than and less than 40 years of age and they are of course selected in respect to have been equally affected by the increase.24 survival.) Using the life-table method, and in- T h a t the same generalizations apply to the cluding untraced patients as "at risk" during situation in Sweden is suggested by comparison the total period of observation, the percentage of the incidence data for this country in 1915 of survivors at various periods u p to three to 1931 reported by U d d ~ t r i i m e rw~it~h pres- years after diagnosis are shown in Fig. 5. Sepa- ent-day death rates. Uddstriimer pointed out rate trends are shown for males and females the increased incidence evident between the and for age cgroups less than and more than early and late years of his study. age 40. T h e data indicate a definite difference Irregularities in age curves that are based on in prognosis between the two age groups. Cor- data derived from cross sections of the popula- rection ol the figures for the group 40 and tion may sometimes be explained by the fact more, by eliminating the number of deaths that changes over a period of time have affected that would have been expected in the group particular cohorts of the population.15 HOW- on the basis of current age-specificdeath rates, ever, in the circumstance of a regular increase raises the three-year-survival percentage from in incidence, as noted for Hodgkin's disease, it 15.8 (both sexes) to only 17.6, compared to is not possible to explain the bimodal age 38.1 per cent for patients less than 40. T h e curve in this way. Such an explanation would difference bctween the two age groups ir there- require several changes in the direction of the fore not explained by the higher death rates trend with time. from other causes in the older age groups. T h e reliability of these findings is qualified SURVIVAL by the high percentage of failures in follow-up. However, these failures will affect the absolute In comparing the behavior of this disease in values rather than the comparison of the two No. 5 EPIDEM~OLOGEICVAIDLENCOEN HODGKIND'ISSEASE* MacMahon I051 90; ,\ 80 70 z0 6 0 -> a> 50 3 0) c 40. Z w U (wr 30 a - MALE .-.--F E M A L E 0` 3 6 9 12 18 24 36 MONTHS OF SURVIVAL FIG.5 . PerceriLage survival of 474 white patients after diagnosiq of Hodgkin's disease, related to age and Sex. age groups, unless the probability of follow-up is related to age. The percentage of losses is higher in the less than 40 age group (28 per cent of 218 patients) than in those 40 and more (18 per cent of 256 patients). Threeyear-percentage survivals have therefore been recalculated on two hypotheses: (1) that patients lost to follow-up died at the time of loss: and (2) that such patients survived the wholc three-year period. Under these two hypotheses, three-year survivals for the group 0 to 39 years of age were 24 and 53 per cent respectively; for the group 40 years and more corresponding figures were 10 and 28 per cent. Although the absolute-survival rates vary considerably under these two hypotheses, it is not likely that the difference between the two age groups is related to this circumstance, since, in order to explain the difference in this way, it would be necessary to assume that all of the patients less than 40 years of age who were lost to followup died at the time of loss, but that all of the patients more than 40 who were lost survived the three-year period. It seems more likely that the higher percentage of patients traced in the older age group is related to the poorer survival of this group, since the method of the study was such as to give somewhat greater likelihood of follow-up for dead than for surviving patients. DISCUSSION The term "Hodgkin's disease" is descriptive of certain characteristic though variable pathological and clinical findings. It is without etiological implications. 7 he pathological and clinical features of the condition include aspects suggestiveof both infective and neoplastic forms of disease. I h c problem posed by this situation has been resolved by some writers with the hypothesis that the condition occupies an intermediate position between these two disease groups-either that a neoplastic virus, such as that of the Rous sarcoma, is the causative 19 or that neoplastic changes may be superimposed upon a primarily infective lesion.2.7, 33 The evidence reviewed here suggests an alternative hypothesis-that Hodgkin's disease is a syndrome that includes the common clinical and pathological end results of distinct etiological processes. Specifically, the trend in age incidence suggests the possibility of at least two etiological entities-the first with maximum incidence in the 20- to 54-year age groups and the second showing increasing incidence with age after 35 or 40. T h e epidemiological features of the disease in these two age periods differ in many particulars. In the present study, differences have been noted in sex ratio, geographic incidence, survival, and possibly racial and religious incidence. Only in secular change in incidence has the behavior of the disease in the two age groups been comparable. T o these features of difference may be added the fact that patients in whom sarcomatous changes are found are predominantly in the older age groups. In the present material, specific histological diagnoses were not recorded with sufficient frequency for adequate analysis. However, in the material of Jackson and Parker,13 6 per cent of the 169 patients less than 40 years of age were diagnosed as having the sarcomatous form of the disease, compared to 26 per cent of the 160 patients of 40 or more years of age. It is clear that no one of the above listed features is in itself conclusive evidence of fundamental difference in the nature of the disease at different ages. Each can be interpreted as due to relatively minor factors modifying incidence of the same disease process at different ages. However, the number of the differences between the two age groups and the markedly bimodal nature of the age-incidence curve itself suggest that the difference between 1052 CANCERSeptember-October 1957 Vol. 10 the disease as it occurs in the two age groups may be in such fundamental areas that it would not seem appropriate to consider the condition as a single disease. The question may then be raised-to what extent is the suggested etiological differentiation by age group interchangeable with the common pathological one into sarcoma, granuloma and paragranuloma? In the Brooklyn ma- terial, 44 per cent of the patients were less than 40 years of age. (Of course, sharp delineation between the two groups at age 40 is not in- tended. This age is merely suggested as a point above or below which the majority of the cases of any one type seem likely to fall. Age curves for diffeient countries might suggest points of delineation other than the one selected for use here on the basis of the Brooklyn data.) From this fact alone-that there are approximately equal numbers of patients older and younger than age 40-4 is obvious that the pathological and epidemiological classifications are not interchangeable, since in the pathological classification the granuloma is by far the most common form of the disease. Thus, although the sarcoma occurs almost solely among those in the age groups more than 40, 75 per cent of the patients in this group do not have the sarcoma (according to the data ol Jackson and Parker13). (It should be noted that an accurate estimate of the incidence of sarcomatous changes in Hodgkin's disease in patients more than 40 years of age is not available. The granuloma and the sarcoma may be found in the same patient at different times or at different biopsy sites,4S10 and complete ascertainment would require complete autopsy examination of a series of patients of this age group.) I n the data of Jackson and Parker,I3 paragranuloma occurs relative to the granuloma with almost equal frequency in patients older or younger than age 40 (14 and 15per cent respectively), and there is therefore no reason to suspect that the granuloma-paragranuloma distinction coincides with any part of the epidemiological differentiation. In short, the pathological classification, as presently understood, does not coincide with the epidemiological differentiation by age group, except insofar as the sarcomatous form is restricted to those who are more than 40 years of age. However, we are unaware of any specific search for pathoIogica1 differences between the disease occurring in patients younger and older than 40 years, 60 50 40 Z 30 0 J 2 20. 2 IT Wa 10 ' W Uz w 0 -2 5 4. -1 2zz 3 . Q 2. :_.___, 3` ,---._,' ; ! - HODGKINS DISEASE ----- RETICULUM C E L L SARCOMA 11' -'''' o,g$8ggt$K886; AGE A T DIAGNOSIS FIG.6. Age-specific-incidencerates for IIodgkin'sdisease and reticulum-cellsarcoma, Brooklyn, I943 to 1952. The further question of the neoplastic or infective nature of Hodgkin's disease can then be posed separately or each of the two broad age groups for which separate etiologies are hypothesized. At the outset it should be made clear that the terms "infective" and "neoplastic'' are not used to describe a dichotomy. `The fact that neoplastic conditions may be associated with "infective" agents of a specialized type does not detract from the utility of an attempt to categorize the disease into a broad group of diseases from which we expect particular types of behavior and particular types of response to preventive measures. With respect to those cases falling into the later part of the age-incidence curve, two features are suggestive of neoplasm: (1) the frequency with which changes recognized by pathologists as frankly neoplastic (sarcoma) are found; and (2) the age-incidence curve, which, showing regular increase with age after early middle age, is typical of a neoplasm. Neither of these features is conclusive, but they are each more suggestive of neoplasm than of infection. The similarity26 or even the identity" of the histological pictures of Hodgkin's sarcoma and reticulum-cell sarcoma has been attested to frequently. In Fig. 6 the age-incidence curve for Hodgkin's disease (all forms) in No. 5 EPiDEMIOLOGICAL EVIDENCOEN HODGKINDI'SSEASE MacMalzon 1053 Brooklyn is compared with that for reticulumcell sarcoma in the same population. T h e similarity of the two curves in the second half of life is remarkable, since, even between known neoplastic disorders, differences in shape and slope of the age-incidence curve are usual. The sex ratio of patients with reticulum-cell sarcoma in Brooklyn (193 patients, 60.1 per cent male) is also similar to that of Hodgkin's disease in patients more than 40 years 01 age (306 patients, 64.4 per cent male} (difference 4.32 4.4). As just notcd the frequency with which sarcomatous changes may be found with complete examination in patients with Hodgkin's disease more than 40 years of age is unknown but is not less than 25 per cent. These features suggest that the relationship between Hodgkin's granuloma occurring after 40 years of age, Hodgkin's sarcoma, and reticulum-cell sarcoma is a close one. On the other hand, none of the forementioned features suggestive of neoplastic disease arc found in the group of patients less than 40 years of age among whom the maximum incidence is in the 25 to 29-year age group. The sarcomatous form is unusual, and the age-incidence curve is quite unlike that of a neoplastic disorder. For example, the trend in the first half of liIe stands out strikingly compared to that of reticulum-cell sarcoma (Fig. 5). T h e high rates for the disease in this age group in the three traditional dairy-producing countries (Netherlands, Denmark, and Switzerland) are provocative and warrant further in- vestigation. SUMMARY By survey of hospital records and death certificates, a series of 573 residents of Brooklyn, New York, diagnosed as having Hodgkin's disease during the period 1943 to 1952 was assembled. For all but seventy-two of these, diagnosis was supported by biopsy or autopsy. The age-incidence curve was distinctly bimodal, with one peak in the age group 25- to 29- and a second in the 70- to 74-year group. Similar bimodality is evident in mortality statistics for the United States and in data from previous morbidity surveys. Further examinations were conducted separately for patients more than and less than 40 years of age. The disease as it affects these two groups appears to differ in sex ratio, geographic incidence, survival, frequency of sarcomatous change, and possibly racial and religious incidence. It is suggested that Hodgkin's disease as presently understood may be a syndrome including the common clinical and pathological end results of at least two distinct etiological processes. There is no evidence as to the nature of Hodgkin's disease as it occurs in persons less than 40, and this portion of the age-incidence curve for the disease is quite unlike that of any known neoplastic disorder. On the other hand, thc disease as it affects the older age groups (40 and more) has leatiires characteristic of known neoplasms. It is suggested that there is a close relationship between Hodgkin's granuloma in persons more than 40, Hodgkin's sarcoma, and reticulum-cell sarcoma. REFER.ENCES 1. BERMAN, L.: Malignant lymphomas; their classification and relation to leukemia. Blood 8: 195.210, 1953. 2. CHEVALLIEPR., and BERNARDJ.,: La maladie de Hodgkin; lymphadhie aleucbmique-Iymphosarcomepseudo-Ieudmie de Bonfils-morbus Hodgkini de Wilks-adhie de Trousseau-granulome malin de Benda-lymphogranulome malin-lymphogranulomatose maligne, etc. Paris. Masson et Cie. 1932. 3. CLEMMESEJN.; , BUSKT, ., and NIELSENA,.: Topographical distribution of leukemia and Hodgkin's disease in Denmark, 1942-46.Acta radiol. 37: 223-230, 1952. 4. CUSTER, R. P., and RCRNHARWD., G.: Interrelationship of Hodgkin's disease and other lymphatic tumors. Am. J . Med. Sc. 216: 625-642, 1948. 5. DFARDORFNF., R.: Religio-cultural background of New York City's po ulation. Milbank M e m . Fund Quart. 33: 152-160, 1&5. 6. DORN,H. F., and CUTLORS,. J.: Morbidity from Cancer in the United States; Pt. I, Variation in Incidence by Age, Sex, Race, Marital Status, and Geographic Region. Pub. Health Monograph No. 29. Washington. U. S. Government Printing Office. 1955. 7. EWIXG,J.: Neoplastic Diseases; a Treatise on Tumors, 2d ed. Philadelphia, M`.R. Saunders Co. 1922. 8. GALL, E. A., and MALLORYT, . B.: Malignant lymphoma; clinico-pathologic survey of 618 cases. A m . J, Path. 18: 381-429, 1942. 9. GILLIAMA, . G.: Age, sex, and race selection at death from leukemia and lymphomas. Blood 8: 693-702, 1953. 10. HERBUTP, . A.; MILLERF, . R., and ERF, I,. A.: Relation of Hodgkin's diseasc, lymphosarcoma and reticulum cell sarcoma. A m . J . Path. 21: 233-253, 1945. 11. HOSTFRH, . A., and DRATMANM,. B.; edited by GRAVERL,. F., and ROLNICKH,. A.: Hodgkin's disease 1832-1947. Cancer Research 8: 1-48, 49-78, 1948. 12. HOSTERH, . A., and ZANESR, . P., JR.: Personal communication to HOSTCRH, . A., ET ~ ~ 1 1 13. JACKSONH,., JR., and PARKERF,., JR.: Hodgkin's Disease and Allied Disorders. New York. Oxford University Press. 1947. 14. LANCASTEHR., 0.: Mortality in Australia from leukaemia. Med. J . Aust. 2: 1064-1065, 1955. 15. MACMAHONB,.: Breast cancer at menopausal ages; explanation of observed incidence changes. Can- 1054 CANCESReptember-October 1957 VOl. 10 cer 10: 1037-1011, 1957. 11: 211-224, 19.56. 16. hfACMAHON, U.: Unpublished data. 17. hfACkfAHOK, B., and CLARKD, .: Incidence of common forms of human leukemia. Blood 11: 871-881, 195G. 18. MACMAHONB,., and KOLLERE,. K.: Ethnic differences in incidence of leukemia. Hood 12: 1-10, 1957. 19. hfCJUNKIN, F. A.: Histologic resemblance of Rous chicken sarcoma No, 1 to Hodgkin's granuloma. J . Cancer Research 12: 47-52, 1928. 20. OLIVLR,J.: Relation of Hodgkin's disease to lymphosarcoma and endothelioma. J . M e d . Research 29: 191-207, 1913. 21. PARSON,P. B.; POSTONM, . A., and WISE, B.: Pathology of human brucelliasis. [Abstr.] Am, J. Path. 15: 634-635, 1939. 27. SMITH, C. A.: Hodgkin's disease in childhood; clinical study with rCsumP: of literature to date. J. Pediat. 4: 12-38, 1934. 28. SYMYERS,D.: Lymphoid diseases; Hodgkin's granuloma, giant follicular lymphadenopathy, lymphoid leukemia, lymphosarcoma and gastrointestinal pseudoleukemia. Arch. Path. 45: 73-131, 1948. 29. THORSONT, . A., and BROWN,D. V.: Stud of lymphomas; I, distribution and incidence. A . Arch. Path. 60: 353-358, 1955. A. 30. UDDSTROMEMR.,: O n occurrence of lymphogranu- lomatosis (Sternberg); in Sweden 1915-1931 and some considerations as to its relation to tuberculosis. Acta tuberr. Scandinav. Su-p~pl. 1: 1-225, 1934. 31. WALLHAUSEAR.:, Hoclgkin's disease. Arch. Path. 22. PATERSONR,., and PATERSONE,.: Hodgkin's dis- 16: 522-562; 672-712, 1933. eaw. Brit. M . J . 2: 1315-1318, 1934. 32. WARTIIINA, . S.: Genetic neoplastic relationships 23. PUBLICH E A L I HSERVICEF. FDERALSECURITY of Hodgkin's disease, aleukaemic and leukaemic lym- AGLNCY:Vital Statistics of the United States Washing- phoblastoma, and mycosis fungoides. Ann. Surg. 93: ton. U. S. Government Printing Office. Annuals, 1949- 153-161, 1931, 1933. 33. \VISEMAN, U. K.: Blood pictures in primary dis- 24. SHIMKINM, . B.: Hodgkin's discase; mortality in eases of lymphatic system; their character and signifiUnitcd States, 1921-1951; race, sex and age distribution; cance. J . A. M. A . 107: 2016-2022, 1936. comparison with leukemia. Blood 10: 1214-1227, 395.5. 34. WORLDHEALTHORGANIZATIOMNa:nual of inter25. SHIMKINM, . B.; OPELRMANKN.,C.: BOSTICKW, . L., national statistical classification of diseases, injuries, and and LOW-BEERB, . V. A,: Hodgkin's disease; analysis of causes of death; sixth revision of international lists of frequency, distribution and mortality at University of diseases and cause of death: adopted 1948. Bull. World California Hospital, 1914-1951. Ann. Int. Med. 42: 136- Health Orgun., Suppl. 1, vol. 1, p. 89, 1948. 153, 1955. 33. WORLDHEALTHORGANIZATIOMN:ortality from 26. SMETANAH,. F., and CoHm, R . M.: Mortality in Hodgkin's disease and from leukaemia and aleukaemia. relation to histologic type in Hodgkin's disease. Blood Epideiniol. & Vital Statict. R e p . 8: 81-114, 1953.