In the present issue of the Journal, 2 papers (Agha et al. and van der Horst et al.) describe trends in incidence of childhood cancer, one from Ontario, Canada, and the other from Denmark. The Ontario investigators also estimated projections to 2015, while the Danish investigators included consideration of cancers up to the age of 34, an age group that often seems to be ignored in consideration of both childhood and adult cancers. Both sets of authors indicate that they have presented their data as they may provide clues as to aetiology as well as progress in other areas. Neither included data on childhood cancer mortality; had they done so, there is little doubt that they would have confirmed the progress made in treatment, currently the mainstay of childhood cancer control. Both groups emphasise the increase in tumours of the central nervous system, the Ontario investigators attributing this, at least in part, to increases in the efficiency in diagnosis. The Danish group emphasise gender differences; the Ontario group did not consider them.

The trends in incidence presented, including the projections in Ontario, suggest that there has been little or no progress in identifying new causes of childhood cancer on which action can be taken, both groups documenting increasing incidence at approximately the same rate per annum even though the 7time periods studied were different. With the exception of leukaemia induced by radiation to the foetus in utero, and possibly exposure to high levels of electro-magnetic fields, we know very little about the environmental causes of childhood cancer, so that prevention is largely ignored in considerations of control of these diseases. Even early detection, or screening, for the one childhood cancer where early studies suggested some benefit, neuroblastoma, on further evaluation was found not to be effective, the early apparent benefit having been confounded by over-diagnosis. So why should there be lack of knowledge on childhood cancer causation? Childhood cancers are rare, so that the most favoured study design used in observational epidemiology, cohort studies, would be difficult if not impossible to perform. There are cohorts of children under observation in some countries, but they are largely designed to evaluate developmental issues, and their relatively limited size is likely to preclude learning much about aetiology of cancer from them. That leaves case-control studies, yet few have been performed for childhood cancer, except for those that have provided the knowledge we have on leukaemia aetiology, so that their success, even if regarded as limited, should encourage other investigators to follow this approach. That they have not done so points to a paucity of testable hypotheses, even from molecular biology. From where will hypotheses come? The Danish study confirms that the types of cancers that are seen in young adults are very different from those that occur in children, so that studies of this age group are unlikely to provide clues to aetiology in younger children. Yet the adult cancer field shows that there are multiple routes from which hypotheses can be derived, including large case-control studies collecting large amounts of data on possible aetiologic factors. There are many groups, including those represented by the authors featured in this issue of the Journal, who could perform such studies. The major hurdle would not be in collecting detailed data from cases, but from comparable controls. Yet it seems unlikely that it would be difficult to obtain hospital-based or friend- or relative-based controls, while the investigators of the studies of electro-magnetic fields in Canada were able to collect data from population controls. Large hypothesis-deriving studies sometimes have difficulty in publishing their findings, especially if they are negative. However, given the lack of information on childhood cancer aetiology, many Journals, including the International Journal of Cancer, would be interested in receiving such papers for consideration for publication.

Turning briefly to the trends of cancer in young adults in Denmark, it is striking how much commoner these cancers are than that in children. Again there seem to be major research agendas that should be pursued, including further consideration of the aetiology of testis cancer and lymphomas; and here, it is likely that knowledge gained would be very relevant to causation of cancer in older adults, though one would have to concentrate on external factors in childhood and early adult life, as well as during gestation, and even parental exposures, the latter two being the potential periods of carcinogenic exposure relevant to childhood cancer.