A reanalysis of mutation-carrying breast cancer patients not selected by family history lowered the risk estimates from 71-85 per cent to 45-68 per cent.

Such overestimation may occur because breast cancer risk is not only associated with these mutations, but with other, outside risk factors as well, which may have skewed past estimations.

An accompanying editorial calls for more refined methods of estimating cancer risk. 'Without a healthy respect for the many factors that may influence penetrance [disease incidence],' write the authors, 'we will continue to overestimate the risk conferred by BRCA1 and BRCA2 mutations alone and, thus, miss opportunities to develop truly effective prevention strategies . . . based on a broad understanding of causative factors' (J Natl Cancer Inst, 2002; 94: 1221-6).

* Elsewhere, scientists are questioning estimates of recurring breast cancer in women carrying BRCA1 and BRCA2. One study has suggested that, in women under 42 with the genetic risk, the relapse rate is as high as 49 per cent. But it is not clear whether genetic or other factors are to blame (Lancet, 2002; 359: 1471-7).

Could it be, as suggested by recent correspondence, that aggressive treatment like radiotherapy is the culprit? Was the surgical procedure used to remove the cancer inadequate? Such debates - which are becoming more common - expose how woefully inadequate our understanding of cancer is (Lancet, 2002; 360: 640).