Response: Re: Ki67 Index, HER2 Status, and Prognosis of Patients With Luminal B Breast Cancer

Howell et al. are interested in using our dataset to address a different but related hypothesis to the one addressed in our recent article, which concerned the value of Ki67 index as a proliferation marker in the context of the intrinsic subtype approach to breast cancer risk assessment. A response to their hypothesis requires quantitative estrogen receptor (ER) and progesterone receptor data for more than 4000 tissue samples in tissue microarrays. Our study on automated quantitative ER assessment ( 1 ) did not reveal evidence that additional prognostic information could be extracted from quantitative ER beyond that already captured by a binary cut point set at approximately 1%. Currently, our dataset has captured complete immunohistochemical information on progesterone receptor in a semiquantitative fashion around visually assessed cut points rather than as a quantitative continuous variable. An extensive reanalysis of the primary image data for ER (possibly by using an improved image analysis approach) and progesterone receptor in a quantitative fashion would, therefore, be necessary to address their inquiry. Our primary image data have been captured, have been published ( 2 , 3 ), and are publicly accessible at http://www.gpecimage.ubc.ca/tma/web/viewer.php . We have therefore invited Howell et al. to access these published digital images and generate the quantitative ER and progesterone receptor scores by whatever system that they feel is appropriate. We are also willing to consider collaborations with other academic groups who wish to conduct similar exercises. We will then test the hypothesis that quantitative hormone receptor data add useful prognostic information in multivariable models incorporating human epidermal growth factor receptor 2 (HER2) and Ki67 data.