LEUKAEMOGENESIS: QUANTITATIVE ASPECTS AND CO-FACTORS 13 



hundreds of thousands of individuals. What seems to be needed to disprove 

 the necessity of tissue damage is an experiment in which leukaemogenesis 

 (carcinogenesis) was initiated by irradiation of cells in vitro. And the diffi- 

 culties of demonstrating direct cellular action in this way are probably clear 

 to everyone. 



The difficulties may be even greater if the general ideas of Armitage and 

 Doll (1957) and of Fisher (1958) are accepted. In order to account for the 

 change with age in the age-specific mortality rate of leukaemias in man it is 

 postulated (Court Brown and Doll, 1959) that some forms of leukaemia, e.g. 

 chronic myeloid leulcaemia, may be due to two successive cellular events while 

 others, e.g. chronic l}miphoid leukaemia, may be due to three successive 

 cellular events. The events are mathematical abstractions though it is easy 

 to think of them as in some way equivalent to other, more biological formula- 

 tions of successive processes, such as initiation and promotion. If two or, even 

 worse, three successive events or processes are to be made to occur in vitro the 

 experimental problem is clearly even more complex than if only one event is 

 needed. 



Although the theories are hardly more than speculative, they are com- 

 patible with the facts of bone-cancer production by bone-seeking isotopes 

 (Mole, 1962). If they are broadly true, the dose-response curve for radiation- 

 induced leukaemogenesis at low incidence rates, 10~^, and the duration of 

 risk following a single exposure will depend critically on whether the first or 

 second type of event is more easily caused to occur by radiation. If radiation 

 causes the first type of event with a higher probability than the second, then 

 the effect of irradiation w^ill be to "age" the individuals in the irradiated 

 population in the sense that, though not visibly different from an unirradiated 

 population of the same age, the age-specific leukaemia incidence rates wiU be 

 those characteristic of an older population. The increased risk of development 

 of leukaemia would not decrease with time (unless additions are made to the 

 hypothesis, cf. Burch, 1960). Some experimental evidence has been inter- 

 preted m this sense (Upton et at., 1960; Lindop and Rotblat, 1961) but on the 

 w^hole the human data suggest that the risk of development of leukaemia 

 after exposure to radiation reaches a peak a number of years later and then 

 decreases. This is what would be expected if radiation caused the second 

 type of event with a higher probability than the first kind of event, but a 

 second coroUary, that the proportional risk would be the same at all ages and 

 the absolute risk progressively greater the older the individual at the time of 

 exposure, while possibly true of the ankylosing spondylitics, may or may not 

 be true in the Japanese (Doll, 1962). If radiation can cause both kinds of 

 event then the dose response relation ought to be expected to change as the 

 range of radiation-dose changes, a most important practical deduction. 



