Dr. P. Muti draws attention to several important features relating to studies of hormones and disease etiology. Although it is true that diurnal and menstrual cycle fluctuations exist in the levels of many hormones, our study was directed at the relationships between the levels of hormones and receptor expression in breast epithelium. Our major concern therefore was that the blood samples for hormone assays be obtained as close as possible to the time of acquisition of the breast tissue samples. All samples were drawn either in the operating room or within 1 h prior to the patient entering the operating room (generally between 9 a.m. and 2 p.m.). We did not attempt to compare hormone levels between groups of patients and therefore do not feel that diurnal fluctuations would be relevant to the major point of analysis in our study (i.e., a comparison of hormones with simultaneously obtained tissue samples).

With regard to menstrual cycle variations in the levels of estradiol and progesterone, there is undoubtedly large variation in the levels of these hormones through the menstrual cycle. In previous studies of breast epithelial hormone receptor expression and proliferation, menstrual cycle phase has been used as a surrogate for hormone levels. However, there is considerable interindividual variation in estradiol and progesterone levels within the same phase of the menstrual cycle. For example, in our population, the mean serum estradiol level was 135 pg/ml in women who were in the second week of a standardized 28-day cycle, but the range was 44–310 pg/ml. Because our study was designed to directly measure hormone levels in each individual rather than use menstrual cycle phase as a surrogate for expected hormone levels within each phase, we felt that further adjustment for menstrual cycle phase was not necessary.

Lastly, Dr. P. Muti raises the very germane issue of variability in hormone and receptor determinations. Hormone assays were performed in batches as they were acquired chronologically. Duplicate serum samples were included in different runs, and the variation between duplicates was less than 10%. Immunohistochemical assays for receptors and proliferation were not repeated, but repeat counts were performed by at least two observers. Although there was some variation between observers (up to 20%), the rank correlation between observers was good.

We would like to thank Dr. P. Muti for her comments and again point out that our main purpose was to explore case-control differences in the relationships between parameters (e.g., hormones and receptors) rather than differences in individual parameters, as has been done in earlier studies. We feel that additional such studies of organ responsiveness will yield further insights into disease etiology.