We thank Drs Gallos and Gupta for their interest in our article and for underscoring the need for research on biomarkers that can predict progestin therapy resistance and endometrial hyperplasia progression. We agree that intrauterine progestin is commonly being used in some communities to treat endometrial hyperplasia, but it was uncommon during the time frame of our study, 1985-2005. However, our study contributes relevant information given the continued use of oral progestins in practice.

We also agree that it would be ideal to account for progestin therapy compliance or other qualities of treatment including dose and duration in future studies of biomarkers of progestin therapy resistance. However, we contend that it is important to do so because these qualities may modify the association between baseline biomarker expression and hyperplasia persistence/progression among women treated with progestin, not confound it. In our study, study pathologists evaluated biomarker expression, blinded to outcome status, using samples collected at index biopsy, prior to progestin treatment.

We appreciate Drs Gallos and Gupta bringing to our attention the issue of spectrum bias and their experience of attenuated findings in a large prospective cohort. We would like to note that we reported the odds ratio as a measure of association, not as a diagnostic odds ratio. Regardless, we agree that our results should be confirmed in adequately powered studies that are well designed to minimize bias.