IN REPLY: We are pleased to see that our recent publication on breast cancer risk in women with atypical hyperplasia has elicited additionalinterestinthissubject.Mascareletalreportalowerabsolute

hyperplasia.Theyreporta5-yearriskof2.8%(95%CI,1.4%to5.5%) and 10-year risk of 5.5% (95% CI, 3.3% to 9.9%) for those found to have epithelial atypia in their group. In the Mayo cohort, we reported a risk of 3.7% (95% CI, 1.6% to 5.8%) at 5 years and 10.9% (95% CI, 7.3% to 14.4%) at 10 years. Mascarel et al suggested two factors that might explain the higher risk reported in our series: (1) inadequate pathologicsamplingofbiopsytissues,suchthatcancerspresentinthe biopsy tissue were not discovered, and (2) inadequate targeting of the highest-risk breast tissues for biopsy because of lack of mammo- graphic screening in early years of the cohort, such that cancers in other areas of the breast were missed/not biopsied. They express their opinion that epithelial atypia is more likely a risk marker of concomitant rather than subsequent cancer. We have concern that what Mascarel et al designate as epithelial atypia may not coincide with atypical hyperplasia as defined for our cohort,inwhichweusedthecriteriaofPageetal. 1 Mascareletalreport finding ductal and/or lobular atypia without concomitant cancer in 670 (24%) of 2,833 of surgical biopsies performed for microcalcifica- tions without a palpable mass. This rate of atypia is at least 10-fold higher than the proportion of patients found to have atypical hyper- plasia in another large series of mammographically detected lesions 2 and suggests that their patients with "epithelial atypia" are different than our patients reported as atypical hyperplasia. Cases that demon- strate only one histologic feature of epithelial atypia—for instance, eithercytologicorarchitecturalatypia—donotconstituteadiagnosis of atypical hyperplasia and thus would not be expected to have as high a risk of subsequent breast cancer. According to their concern regarding inadequate sampling of biopsy tissues, cancers already present might have been missed diag- nostically. If so, then the subsequent breast cancer risk in our cohort should occur primarily in the ipsilateral breast. Although we found slightly more cancers in the ipsilateral breast than the contralateral breastinourpatientswithatypia(relativerisk1.38;95%CI,0.79to 2.21), the difference was not statistically significant. Furthermore, in biomarker investigations in this cohort, we have cut deeper into the blocks for 318 of our 331 women (15 slides of 5-m each). In so doing, only one case was identified as having a small focus of (invasive) cancer that was not seen on the initial diagnostic histology. In terms of the question of mammographically directed biopsy versus not, we found no difference in breast cancer risk for women withatypicalhyperplasiabasedontheindicationforbiopsy(ie,mam- mographic abnormality v palpable concern, P.72 by Cox propor- tional hazards regression). Finally, if Mascarel et al are correct and atypical hyperplasia is really a risk marker of concomitant rather than subsequent breast cancer, then we would expect to see an early increase in risk as occult cancers become clinically apparent, with a plateau effect after several years. In fact we saw just the opposite, with little increase in risk in the first5yearsfollowedbymoredramaticincreasesfromyears5through 15, and no plateau in the increase in breast cancer risk after benign biopsy demonstrating atypical hyperplasia. In summary, our report on atypical hyperplasia is based on the diagnostic criteria of Page et al, 1 rather than isolated features of either cytologic or architectural atypia. This likely accounts for the higher breast cancer risk observed in our cohort compared with that of Mascareletal.Wehaveevidencethatourbiopsytissuesweresampled appropriately and that the breast cancer risk associated with atypical hyperplasia is not related to the indication for biopsy (ie, mammo- graphic v palpable). We continue to maintain the view that atypical hyperplasia is a marker of subsequent breast cancer risk that ap- pliestobothbreasts,andthisriskpersistsforatleast15yearsafterthe diagnostic biopsy.