657

11

Section of Oncology

Whether or not a precarcinogen can exert a transplacental action depends on the nature and amount of the appropriate activating enzyme that is present in the fetus. An exception to this statement is provided by cycasin (methylazoxymethanol glucoside), which is converted to its proximate carcinogen, methylazoxymethanol, by enzymes in the microbial flora of the gastrointestinal tract (Laqueur & Spatz 1968).

REFERENCES Alexandroy V A (1968) Nature (London) 218, 280-281 Bithel J F, Draper G J & Gorbach P D (1973) British Medical Journal i, 706-708 Curnen M G M, Varma A A 0, Christine B W & Turgeon L R (1974) Journal of the National Cancer Institute 53, 943-947 Druckrey H (1973a) In: Tomatis & Mohr (1973); pp 45-58 (1973b) Xenobiotica 3, 271-303 Fedrick J & Alberman E D (1972) British Medical Journalii, 485-488 Greenwald P, Barlow J J, Nasca P C & Burnett W S (1971) New EnglandJournal ofMedicine 285, 390-392 Herbst A L, Kurman R J, Scully R E & Poskanzer D C (1972) New England Journal of Medicine 287, 1259-1264 Herbst A L, Robboy S J, Scully R E & Poskanzer D C (1974) American Journal of Obstetrics and Gynecology 119, 713-724 Herbst A L & Scully R E (1970) Cancer 25, 745-757 Herbst A L, Ulfelder H & Poskanzer D C (1971) New England Journal ofMedicine 284, 878-881 Ivankovic S & Druckrey H (1968) ZeitschriftffUr Krebsforschung 71, 320-360 Jablon S & Kato H (1970) Lancet i, 1000- 1003 Laqueur G L & Spatz M (1968) Cancer Research 28, 2262-2267 MacMahon B (1962) Journal of the National Cancer Institute 28, 1173-1191 Magee P N (1973) In: Tomatis & Mohr (1973); pp 143-148 (1974) In: Essays in Biochemistry. Vol. 10. Ed. P Campbell & F Dickens. Academic Press, London; pp 105-136 Miller J A (1970) Cancer Research 30, 559-576 Miller R W (1973) In: Tomatis & Mohr (1973); pp 175-180 Robboy S J, Herbst A L & Scully R E (1974) Cancer 34, 606-614 Smith 0 W (1948) American Journal of Obstetrics and Gynecology 56, 821-834 Stewart A & Kneale G W (1968) Lancet i, 104-107 Tanaka T (1973) In: Tomatis & Mohr (1973); pp 100- 111 Tomatis L (1965) Proceedings of the Societyfor Experimental Biology and Medicine 119, 743-747 (1974) Biochemical Society Transactions 2, 703-705 Tomatis L & Goodall C M (1969) International Journal of Cancer 4, 219-225 Tomatis L & Mohr U eds (1973) Transplacental Carcinogenesis. International Agency for Research on Cancer, Lyon

The carcinogenic nitroso compounds include examples of carcinogens, the nitrosamines, that do require metabolic activation, and the nitrosamides, that do not. The nitroso compounds that have been most studied from the standpoint of transplacental carcinogenesis are dimethylnitrosamine and N-methyl- and N-ethylnitrosourea. Dimethylnitrosamine exerts a transplacental carcinogenic action in the rat only when given during the last week of pregnancy (Alexandrov 1968), and evidence for metabolic activation of the compound was found on the 17th day of pregnancy but not on the 15th day (Magee 1973). In contrast, N-ethylnitrosourea administered on the 15th and subsequent days of pregnancy in the rat, in doses well below the lethal level, gave rise to tumours of the central and or peripheral nervous system in all of the progeny.

A most interesting feature of the work by Druckrey and his colleagues was the observation that doses of N-ethylnitrosourea at 1/50 of the LD50 to the pregnant rats still induced some tumours of the nervous system in the offspring. Such doses had no readily detectable toxic effects on the treated mother animals (Druckrey 1973a, b). The transplacental carcinogenic effects of some chemicals may not be confined to the F1 generation. Work by Tanaka (1973) using N-methyl nitrosourethane, and by Tomatis and his colleagues (Tomatis 1965, Tomatis & Goodall 1969) with polycyclic hydrocarbons has shown a higher than expected incidence of tumours in the F2 and in some cases the F3 generations of rats derived from mothers treated during pregnancy with the carcinogens. These very interesting findings have obvious implications and warrant further study.

Conclusion There is no doubt that cancer can be induced transplacentally in human beings and in animals by exposure of pregnant females to some chemical carcinogens. The possible occurrence of such exposure to carcinogens in utero should be considered as an etiological factor in human cancers, particularly those of childhood.

Dr L M Kinnier Wilson (Marie Curie Memorial Foundation, Epidemiology Unit, Research Department, The Chart, Oxted, Surrey, RH8 OTX)

Some Aspects of Epidemiology in Paediatric Oncology The main subject of this paper is the epidemiological evidence on the etiology of childhood oncology. The material used is data accumulated over the years by the Marie Curie/Oxford Survey of Childhood Cancers. This survey, started in Oxford by Dr Alice Stewart in 1956, and now continuing at the Marie Curie Memorial Foundation Research Department and the Childhood Cancer Research Group, is an on-going retrospective survey of all children who have died of neoplastic diseases in England, Scotland and Wales since 1953, and of surviving children since 1962.

658 Proc. roy. Soc. Med. Volume 68 October 1975

12

The relationship between diethylstilbcestrol and Table I vaginal cancer (Herbst et al. 1971, Lanier et al. Numbers of cases and controls irradiated in utero related to the 1973, Herbst et al. 1974) appears to be the only probable number of exposures to X-rays, as recorded by the hospital direct evidence of transplacental chemical carcinoRisk relative to nonirradiated cases genesis in man. There are, however, several reasons for thinking that many, and perhaps most, No. of No. of No. of 95 Y. confidence limits cancers in childhood are initiated before birth. films cases controls Estimate Lower Upper 1 287 239 1.26 1.06 1.50 The epidemiological evidence on the time of 2 199 154 1.35 1.09 1.67 initiation of childhood cancer is based on a 3 96 65 1.54 1.13 2.11 4 59 28 2.18 1.40 3.41 variety of observations. Association between childhood cancer and obstetric radiography: The evidence for an association between the use of abdominal X-rays during pregnancy and subsequent cancer of the child has been presented in a number of papers (Stewart et al. 1958, MacMahon 1962, Stewart & Kneale 1970). The existence and nature of this association have been a matter of controversy for several years. Perhaps the most convincing evidence in favour of a causal relationship is that supplied by recent data from our survey, which shows that there is a marked dose-response relationship (Bithell & Stewart 1975). This is shown in Table 1, from which it can be seen that the relative risk increases with the number of films taken.

Association between childhood cancer and virus infections during pregnancy: Reports of an association between viral infections (mainly influenza) during pregnancy and subsequent leukaemia or cancer in the child have appeared in several papers (Fedrick & Alberman 1972, Leck & Stewart 1972, Bithell et al. 1973). Data from our survey are reproduced in Table 2. It would appear that the proportion of cases associated with virus infections may be quite small. It is, of course, possible that the association is with medication taken for these diseases rather than with the diseases as such.

>5 Unknown

65 475

29 325

2.32 1.53

1.51 1.32

3.58 1.77

All cases

1181

840

1.47

1.34

1.62

Association with poliomyelitis vaccine: A report from the United States (Heinonen et al. 1973) suggests that there may be an association between poliomyelitis vaccination during pregnancy and subsequent tumours of neural origin in the child. This finding will require confirmation from other studies, but if confirmed it provides further evidence of a transplacental factor. At present, data under analysis from our survey tend to support the finding.

Congenital and familial factors: It is well known that some types of malignancy are associated with certain groups of congenital defects. Thus leukaemia has been associated with chromosomal abnormalities; this can be illustrated by the connexion between mongolism, leukemia and trisomy 21, and also by the association between acute myelogenous leuk2emia and chromosomal translocation which has been described by Sakurai et al. (1974). Another example is the association of Wilms's tumour with certain types of growth disturbances, such as hemihypertrophy, maldevelopment of the genitourinary tract and aniridia, though the latter has also recently been associated with partial chromosomal deletion.

Table 2 Numbers of affected children (by tumour) and matched controls where mothers reported virus infections in pregnancyInfection ofmother during pregnancy

Rubella

Chickenpox Influenza

Other virus infections

Leukemia Other lymphatic

9 3

3 0

46 5

10 3

tissue disease Other tumours

5

4

45

13

17 7

7 0

96 64

26 31

All cases Controls Estimated relative risk 95 % confidence limits: lower upper X2 (1 d.f.)

2.43

-

1.52

-

0.96 6.9 3.38

-

1.11 2.14

-

-

6.24

-

* Britain, deaths 1953-67 (as supplied from OPC3)

13

Section of Oncology

Table 3

for girls, in neuroblastoma the girls carried a significantly better survival rate than boys, which was not shown in the other two tumours studied. Preliminary studies on brain tumours suggest that here also girls may have a better prognosis.

Ascertainments of sibling pairs in survey of childhood cancers Total number of ascertaisnents 103 Expected number if there were no familial effect 54.7 Observed and expected numbers according to major diagnostic group Leukwmia Lymphomas Cancers 15 (6.6)@ 10(3.8) Leukemia 25 (21.1) 5 (0.6) 4 (6.0) Lymphomas Cancers 44 (16.6)

659

Suspected therapeutic agents: There is a possible association between childhood cancer and some chronic maternal illnesses, in addition to the viral infections already referred to. If such an * The numbers in brackets are those which would be expected association can be proved, it could either result if there were no familial effect directly from the illness or more probably, it Both these associations are verified by data from could arise from the prolonged usage of certain our survey, and other interesting associations drugs during the course of treatment. Data on do occur. There is also the familial increase in certain possible associations are available from risk. A detailed analysis of such risks (Draper our survey (Bithell & Draper 1975, unpublished), 1974, personal communication) suggests that in but we regard it as premature at this stage to families with one affected child, there is a small produce these data, since only a preliminary risk that a further child will be affected by the same analysis has been carried out. disease or another type of tumour. The results Currently, we are seeking information on the of this analysis are summarized in Table 3, which administration of hormones and ultrasonic tests compares the observed and- expected numbers of and on smoking habits during pregnancy. We are cases in families of children with neoplastic also investigating the occurrence of breastdisease. The figure for the total number of con- feeding and the usage of contraceptive drugs firmed cases in the survey (103) is reached after before conception of the survey child. the children suffering from known familial Finally, it would be a fascinating project to diseases have been eliminated. These figures do apply epidemiological methods to the whole not, therefore, include any sibling pairs with a question of the administration of immunodiagnosis of retinoblastoma, von Reckling- suppressive drugs during pregnancy and immunohausen's disease, or histiocytoses such as Letterer- logical deficiency diseases at birth. The work of Siwe disease. It can be inferred from Table 3 that Cote (1974) suggests that the use of immunothe risk in other sibling-pair families is not quite suppressive drugs may have a transplacental double the risk in the population as a whole. effect, and the subsequent children may have Thus the risk that a child will develop a neoplasm increased susceptibility not only to infections in this country is about one in 600; where there is but also to lymphomas. a previous sibling with a history of such a disease the risk is approximately one in 300, or 0.3 % in REFERENCES Bithell J F, Draper G J & Gorbach P D the population at risk. Data from the USA (Miller (1973) British Medical Journali, 706 J F & Stewart A 1971) show a similar pattern. In such an analysis Bithell (1975) British Journal of Cancer 31, 271 one cannot exclude the possibility that the familial Cote C J (1974) Journal ofPediatrics 85, 324 factor is a consequence of the children sharing a Fedrick J & Alberman E D (1972) British MedicalJournal ii, 485 common environment, which could of course be Heinonen 0 P, Shapiro S, Monson R R, Hartz S C, Rosenberg L & the uterine environment. Alternatively, there may Slone D (1973) International Journal ofEpidemiology 2, 229 A L, Robboy S J, Scully R E & Poskanzer D C be a direct genetic effect, an increased suscepti- Herbst (1974) American Journal of Obstetrics and Gynecology 119, 713 bility or a reduced immunological response in Herbst A L, Ulfelder H & Poskanzer D C (1971) New England Journal ofMedicine 284, 878 certain families. Kinnier Wilson L M & Draper G J Sex and age: It is well known that these factors have an important relationship to the course and prognosis of childhood malignancies. Specific studies of certain tumours have been undertaken in this survey - on Wilm's tumour (Ledlie et al. 1970), neuroblastoma (Kinnier Wilson & Draper 1974) and retinoblastoma (Sanders et al. 1975), in the hope that these may act as base-line descriptions of the natural history of these tumours for possible future trials of new treatment protocols. It appeared that while the morbidity for boys was in all cases greater than

(1974) British MedicalJournal iii, 301 Lanier A P, Noller K L, Decker D G, Elveback L R & Kurland L T i1973) Mayo Clinical Proceedings 48, 793 Leck I & Steward J K (1972) British MedicalJournal iv, 631 Ledlie E M, Mynors L S, Draper G J & Gorbach P D (1970) British MedicalJournal iv, 195 MacMahon B (1962) Journal of the National Cancer Institute 28, 1173

Milier R W (1971) Journal of the National Cancer Institute 46, 203

Sakurai M, Oshimura M, Kakati S & Sandberg A A (1974) Lancet ii, 227 Sanders B, Lennox E M & Draper G J (1975) British MedicalJournal (in press) Stewart A & Kneale G W (1970) Lancet i, 1185 Stewart A, Webb J & Hewitt D (1958) British Medical Journali, 1495