Alice Stewart

A survey of childhood malignancies.

SECTION I. BACKGROUND TO THE SURVEY The present survey is based on an earlier study of the vital statistics relating to leukaemia (Hewitt, 1955). This had revealed an unusual peak of mortality in the third and fourth years of life which indicated that the subsequent survey should, in the first instance, be restricted to children. The earlier investigation had also led to the suggestion that it might be particularly worth while to study modern innovations, such as radiology.

RADIATION DOSE EFFECTS IN RELATION TO OBSTETRIC X-RAYS AND CHILDHOOD CANCERS

Abstract Epidemiological data from the Oxford Survey of Childhood Cancers has been analysed in respect of in-utero exposure to X-rays during obstetric investigations. The risk of cancer was greatest when exposure was during the first trimester. The excess cancer risk from obstetric X-ray examination was directly related to the fetal dose. It is suggested that this dose-response relationship fits in with a previously published hypothesis that cancers caused in this way are due to the propagation of one cell whose controlling gene had experienced a small but irreversible change at the moment of exposure to X-rays.

Radiation exposures of Hanford workers dying from cancer and other causes.

AbstractData from the Hanford study have shown that sensitivity to the cancer-induction effects of radiation is at a tow ebb between 25 and 45 yr of age. Nevertheless, at younger and older ages there is probably a cancer hazard associated with low level radiation which affects bone marrow cancers mo

Aetiology of Childhood Malignancies

The interpretation of this distribution of Rh groups among the children with gliomas is much more difficult than that of the ABO groups, not only because of the various subdivisions of the Rh factor, but also because their geographical variations have not been determined (A. C. Kopec, 1959, personal communication). Thus the blood-group distributions among children in the Register show the following pattern: (a) There is a very significant excess of group A over group 0 in children with gliomas and a less significant excess of group A over group 0 in children with epithelial tumours. Children with gliomas also show an excess of Rh positive which may be significant. (b) Possibly children with tumours of the sympathetic nervous system and retinoblastomas may have an abnormal 0: A ratio. (c) Children with other tumours seem to have a normal ABO and Rh distribution. Mayr, Diamond, Levine, and Mayr (1956) did not find any abnormal blood-group distributions in brain tumours except in pituitary adenomas. Their series was collected from various neurosurgical centres in Boston and New York and was not broken down according to age. The nature of the hospitals and the types of glioma suggest, however, that the number of children included was so small that any variation from normal would not be apparent. As a result of our finding an abnormal O: A ratio among children with gliomas in the Register, Yates and Pearce (1960) examined the blood-groups in patients of all ages with astrocytomas in Manchester, diagnosed between 1933 and 1958, and found an abnormat 0: A ratio in those under 20 years old who were treated after 1945. However, their cases included those in the Register, so they could not be described as forming a completely independent series.

Prenatal irradiation and childhood cancer

Estimates of the relative risk of childhood cancer, following irradiation during fetal life, are reported. They are based upon extended case-control investigations of childhood cancer deaths in England, Wales and Scotland between 1953 and 1979 comprising 14759 geographically-matched and birth-date-matched case/control pairs. The estimates were calculated using Conditional Logistic Regression (Miettinen-Breslow) techniques. This method of risk-estimation limits the distortions caused by confounding factors or by biased selection of controls. Through analysing a range of reported exposures other than radiation, levels of general reporting and recording biases between cases and controls were also assessed. There was no evidence among cases or controls of any systematic reduction in the frequency of pregnancy x-rays between 1950 and 1979. During this period of time, about 7 per cent of all childhood cancers, and 8 per cent of those with onset between the ages of 4 and 7 years, were caused by X-ray examinations. The dose-response relationship was one death per 990 obstetric X-ray examinations; or 2000 deaths per 104 man-Gy.

Pregnancy x-rays and childhood cancers: effects of exposure age and radiation dose

Using data from the Oxford Survey of Childhood Cancers this paper examines the effect of foetal age and number of films used on the subsequent risk of childhood cancer associated with prenatal x-rays. X-rays early in pregnancy were taken for different reasons and required more films when compared with those taken in late pregnancy. Therefore, Mantel-Haenszel techniques were used to estimate the independent effects of (a) exposure age and (b) number of films. Age at exposure had a clearly significant effect; x-rays taken in the first trimester of pregnancy were 2.69 times as effective as x-rays taken in the third trimester. First trimester exposures were often the result of maternal illnesses, so these maternal illnesses were then included among the controlling factors. When this was done the first trimester x-rays were 2.73 times as effective as later exposures. First trimester x-rays were most strongly associated with the cancers which were diagnosed between 4 and 5 years of age. Although the number of films had no detectable effect upon relative risk calculated over all ages, multiple exposures were demonstrably associated with early age at diagnosis.

CHANGES IN THE CANCER RISK ASSOCIATED WITH OBSTETRIC RADIOGRAPHY

Abstract Summary The prenatal experiences of 12,694 children who were born between 1943 and 1965 were the subject of an inquiry, the purpose of which was to discover how the cancer hazard associated with obstetric radiography varied over the years 1946 to 1962. Half of the children had died from a malignant disease before the age of ten years during the period 1953-65 and the other half were alive and well when the data were collected. The controls were individually matched for sex, date of birth, and region with the cases. The case group contained 985 children who were X-rayed in utero, and in the control group 645, and it is suggested that about 370, or 6%, of the traced cases were caused by these examinations. If it had been possible to include in the survey all the children who were born in England, Scot- land, and Wales between 1946 and 1962 and subsequently died from cancers before the age of ten years, it is estimated that there would have been in the region of 10,500 non-radiogenic and 650 radiogenic cases (i.e., 94.5% and 5.5%, respectively). In this country, juvenile cancer deaths affect about 1 in every 1200 liveborn children or 620 members of each birth cohort. At the beginning of the study period children who were X-rayed in utero were over twice as likely to encounter such a death as their non-X-rayed contemporaries, and at the end of the period 1.4 times as likely; but the risk of being X-rayed was lowest at the beginning of the period and higher in the middle than at the end. Compared with the immediately preceding and following years, the period 1955-57 was a hazardous one in which to be X-rayed, and the period 1960-62 compared unfavourably with the period 1958-59. Temporary interruptions in the trend towards safer obstetric radiography are ascribed partly to changes in X-ray procedures and partly to premature rebuttal of the cancer hazard. The available evidence suggests that small doses of X-rays are sufficient to initiate cancer processes in immature tissues, and that all such tissues are equally radiosensitive. It is not, however, possible to say whether this effect is confined to immature tissues since it is possible that the timing of obstetric X-rays in relation to the children concerned (and the whole body exposures) were only of importance in revealing a wider hazard.

Childhood cancer and parental use of alcohol and tobacco

Reported consumptions of alcohol and tobacco for the parents of 1641 children who died with cancer in England and Wales during the period 1977 to 1981 were compared with similar information for the parents of 1641 control subjects. Consumption of alcohol by fathers was not associated with an increased risk of childhood cancer (relative risk (RR)) = 1.05; 95% confidence interval (CI): 0.86 to 1.28), but for daily consumption of cigarettes was not shown to be associated with an increased risk and consumption of alcohol was associated with a relatively low cancer risk (RR = 0.82; 95% CI: 0.70 to 0.96). Relations between maternal consumption of cigarettes and birth weights suggested that the smoking data were equally reliable for case patients and control subjects.

Adult leukaemias and diagnostic x rays.

A survey was made to detect possible associations between exposure to medical x rays and the subsequent development of malignant diseases in adults. Data were obtained from 963 cases of leukemia, 60 of lymphosarcoma, 951 of other cancers, and 974 control subjects. The leukemias included 512 lymphatic leukemias and 511 other leukemias (referred to as the M series). X-ray exposures which involved the trunk (chest and abdomen) were reported more often by M series leukemias than other subjects. The difference between M series leukemias and other subjects with respect to trunk x-ray exposures was greater for therapeutic than diagnostic exposures. For diagnostic exposure there was a peak incidence for M series leukemias 3 to 5 yr before the onset of symptoms. It was concluded that about 8% of leukemias other than lymphatic leukemias were caused by diagnostic x rays, and a further 3.6% by therapeutic x rays. It was estimated that 1 leukemia death occurred for every 46000 x-ray examinations of the chest or abdomen in Britain in 1951-5.

Background radiation and childhood cancers

Outdoor terrestrial gamma radiation exposure levels (TGR), estimated for each of the 10 km squares of the Great Britain National Grid, were related to local cancer death rates in childhood The examination was based upon the prior hypothesis that an association ought to be detectable This was itself based upon an examination of geographical TGR variations and upon a recently reported recalculation of the dose-response relationship between the risk of childhood cancer and foetal exposure to medical x-rays The analysis was pressed through several stages in which the effects of sociodemographic and medical confounding factors and their temporal changes were identified and separated. TGR was then shown to exert an independent statistically significant effect.