Assessment of Ki-67 in breast carcinoma: Interobserver variability and comparison between core needle biopsy and resection specimens.

Abstract

Ki-67 is a nuclear protein linked to cellular proliferation and is used as a prognostic and predictive biomarker in breast carcinoma. However, variability in its assessment limits its clinical utility. This study evaluated interobserver reproducibility in Ki-67 scoring and compared proliferative indices between core needle biopsy (CNB) and corresponding resection specimens in invasive breast carcinoma. Sixty-three cases with matched CNB and resection specimens were retrospectively analyzed. Ki-67 immunohistochemistry was independently evaluated by two pathologists using manual image-based counting. Interobserver agreement and CNB-resection concordance were assessed using the intraclass correlation coefficient (ICC), Cohen's kappa, Bland-Altman analysis, and Spearman's correlation. Tumor characteristics were analyzed for their association with variability. Interobserver agreement was excellent, with ICCs of 0.89 for CNB and 0.91 for resection specimens. Cohen's kappa for binary classification (<20 % vs. ≥20 %) showed moderate agreement for CNB (κ = 0.54) and substantial agreement for resections (κ = 0.78). Bland-Altman analysis revealed small but consistent bias, with CNB values slightly higher (+2.89 % and + 2.25 % for raters 1 and 2, respectively). However, discrepancies >10 % were observed in some cases. Tumor characteristics had minimal to no association with scoring variability. Despite excellent statistical agreement, clinically significant variability in Ki-67 scoring may occur. These findings support interpreting Ki-67 as a continuous variable rather than relying on fixed cutoffs. Pathologists should consider specimen type and scoring limitations when reporting Ki-67, and may recommend which value is more reliable or suggest retesting when appropriate.