Prognostic Biomarkers in Ductal Carcinoma in Situ: An Interobserver Agreement Study
Rachel L. Stewart, Daniel S. Hippe, Rebeca Alvarez, Sara H. Javid, Janice N. Kim, Mary Lynn Bryant, Isabella Li, Debosmita Biswas, Anum S. Kazerouni, Savannah C. Partridge, Habib Rahbar, Mark R. KilgoreDuctal carcinoma in situ (DCIS) is a noninvasive form of breast cancer that accounts for 15% to 25% of all new breast cancer diagnoses. Clinical management approaches for DCIS include surgical excision, radiotherapy, and endocrine therapy (for patients with estrogen receptor (ER)-positive disease). Active surveillance is also being investigated for patients with low-risk disease. Pathologist assessment of nuclear grade as well as ER expression by immunohistochemistry is currently used to inform clinical decision-making. Exploratory protein biomarkers in DCIS include HER2, p16, p53, and Ki-67. Our study evaluated pathologist interreader agreement for the following immunohistochemical stains applied across 40 cases of DCIS: ER, PR, HER2, p16, Ki-67, and p53. We found that pathologist agreement was good to excellent for assessment of ER and PR (progesterone receptor), both with regard to the percent of cells staining as well as staining intensity (ICC/kappa: 0.76-0.96 for ER, 0.69-0.98 for PR). Interreader agreement was also good to excellent for HER2 interpretation (ICC/kappa: 0.78-0.97). For additional exploratory biomarkers, including p16, p53, and Ki-67, interreader agreement ranged from fair to good, likely reflecting a lack of pathologist training and less familiarity with these biomarkers in the context of DCIS. Our findings suggest that investigators should take potential interreader variability into consideration when designing clinical trials and exploratory biomarker studies.