THE LANCET• Vol 353• June 12, 1999 1987 thromboses, five extra endometrial cancers, and no extra deaths. Counterbalancing these side-effects were 29 fewer invasive breast cancers and 17 fewer in-situ carcinomas. For every 19 women treated with tamoxifen, one event was avoided. Over half of all ipsilateral recurrences were treated with mastectomy. By extrapolation from the data presented, 22 mastectomies were prevented by tamoxifen, with one mastectomy avoided for every 40 women treated. From a biological standpoint, ER function and expression are fascinating and of direct relevance to primary prevention as well as to treatment of DCIS. ER expression is not routinely tested for in DCIS, but perhaps it should be now. Data cited by Fisher and colleagues show that ER-positivity rates in DCIS are high, although less so where comedonecrosis is present. 5 Parallels with

NSABP P-1 come to mind, since in that prevention trial, tamoxifen seems to have delayed or prevented ER-positive cancers preferentially. If this finding were to apply in DCIS as well, it would strengthen the notion that ER expression (and subsequent phenotypic behaviour) is an early event in the development of breast cancer. Since not all DCIS is ER positive, this disorder might be a good model for targeting non-ER-mediated molecular pathways in the search for new preventive and therapeutic strategies.