Interobserver Agreement Among Pathologists for Semiquantitative Hormone Receptor Scoring in Breast Carcinoma

Cohen, David A.
Dabbs, David J. MD
Cooper, Kristine L. MS
Amin, Milon MD
Jones, Terrell E.
Jones, Mirka W. MD
Chivukula, Mamatha MD
Trucco, Giuliana A. MD
Bhargava, Rohit MD

¹Department of Pathology, Magee-Womens Hospital of UPMC, Pittsburgh, PA
²University of Pittsburgh Cancer Institute Biostatistics Facility, Pittsburgh, PA

Address correspondence to Dr Bhargava: Magee-Womens Hospital of UPMC, 300 Halket St, Pittsburgh, PA; [email protected].

American Journal of Clinical Pathology 138(6):p 796-802, December 01, 2012. | DOI: 10.1309/AJCP6DKRND5CKVDD

The American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) guidelines recommend reporting of hormone receptor test results in a semiquantitative manner. This study used 74 resected estrogen receptor (ER)–positive invasive breast cancers to determine reproducibility of semiquantitative scoring of hormone receptors using the H-score method. Four pathologists independently scored each slide. Agreement among observers was analyzed via Fleiss κ statistics on ER and progesterone receptor (PR) categorical scores. Intraclass correlation coefficient (ICC) was used to estimate the interobserver agreement for ER and PR H-scores on a continuous scale (0–300). There was 100% agreement for categorical ER results (κ = 1) and 97% agreement (κ = 0.823, P < .001) for categorical PR results. For quantitative H-scores, ICC agreement was 0.85 (95% confidence interval [CI] = 0.79–0.90) for ER and 0.87 (95% CI = 0.82–0.92) for PR. Because the H-score provides a continuous measure of tumor hormone receptor content, we suggest universal adoption of this method.