LEUKAEMOGENESIS: QUANTITATIVE ASPECTS AND CO-FACTORS 7 



the square of the dose. If we are thinking of the smallest possible effects of 

 radiation, the distinction may be illusory since a small part of any curve is a 

 straight line, but from the point of view of mechanism the distinction is 

 important. It may be worthwhile to recall those aspects of the human observa- 

 tions that do not fit easily with any simple hypothesis. 



The best documented and most detailed study is still that of Court Brown 

 and Doll (1957) on ankylosing spondylitics given part body X-irradiation in 

 fractionated doses spread over several weeks and at dose-rates of '-^lO-SO r/ 

 min. The study is continuing and more data and more accurate deductions 

 should soon be forthcoming. The published data give a linear regression of 

 leukaemia incidence on dose wliich of itself suggests that a dose of 54 r 

 would cause no leukaemia (Court Brown and Doll, 1958) but are of course 

 compatible with the non-threshold linear hypothesis. They appear also just 

 as compatible with a squared-dose hypothesis (Brues, 1959; Burch, 1960). 

 The leukaemogenic effect of a given exposure seemed to be greater the older 

 the individual at the time of irradiation (Doll, 1962). There were unexplained 

 differences between different radiotherapy centres. Cases began to appear 

 when a latent period of 2 years or so had elapsed, but the way in which the 

 risk of leukaemia changes with time thereafter is not yet properly known (cf. 

 Wise, 1961). 



The data from Japan on leukaemia in those exposed to atomic bomb 

 explosions are of first importance because exposure to radiation was not 

 correlated with other factors which might conceivably affect leukaemia 

 incidence or radiosensitivity. When radiation is given to people with ankylos- 

 ing spondylitis, with thymic enlargement, with carcinoma of the cervix, or 

 because they are pregnant and there is a possible obstetric difficulty, this is 

 not necessarily so. 



Generally, experience in Hiroshima and Nagasaki has been similar 

 (Heyssel et at., 1960; Tomonaga, 1962). In each the dose-response relationship 

 seems broadly linear with a similar and considerable uncertainty at doses of 

 less than 100 r. However the two regression lines have very different slopes 

 (Fig. 1). The over-all increase in nine years of observation has exceeded the 

 whole life-time expectancy of leukaemia by three-fold and six-fold respectively 

 so that the radiation exposure has really caused leukaemia, not merely 

 accelerated the appearance of leukaemias which w^ere going to occur anyway. 

 In each city the proportional increase in incidence rate for a given exposure 

 seems greater in younger people than in older, the opposite of what occurred 

 in ankylosing spoldylitis (cf. Doll. 1962). There may be differences between 

 the cities in the rate of change with time since exposure of the risk of 

 contracting leukaemia; there are also quite complex changes with time in the 

 relative incidence of chronic granulocytic leukaemia and the different forms 

 of acute leukaemia. 



