I was surprised to read the editorial by T Abe on infant acute leukaemia in Leukemia.1 Abe suggests that genistein, a topoisomerase II inhibitor, which occurs abundantly in soybeans, may be largely responsible for the development of infant acute leukemia. Although he has provided evidence that genistein can induce chromatid-type chromosomal aberrations, the epidemiological evidence is clearly against an association between genistein ingestion and infant leukemia. As stated, the incidence of infant acute leukaemia in Japan is comparable to that in other countries in which large amounts of soybean products are not consumed. To keep his hypothesis viable Abe proposed an unsubstantiatable post-hoc modification, which suggests that Japanese infants are protected from the putative harmful effects of genistein by favourable genetic polymorphisms. Overall I felt that the hypothesis was undeveloped and I believe that Abe and Leukemia were premature in printing his ideas. At the very least the final dogmatic sentence of his abstract should have been modified.
As we all know, however the progress of science relies on the generation of new ideas, and perhaps the publication of such hypotheses may generate some more useful ideas and discussion. The issue however is more complex. Since the biomedical literature is frequently misquoted, I predict that in the future the literature will cite Abe's article as if it was based on some ‘truth’, and entrench the concept that genistein has a role in the onset of infant leukaemia. We all generate numerous hypotheses, mostly poorly defined and perhaps irrational, occasionally useful. But, before we publish any or each of these ideas we must provide logical, convincing and biologically plausible arguments, or even better, concrete experimental or epidemiological evidence, in support of our hypotheses.
This is a preview of subscription content, access via your institution