Background Ki-67 is an important proliferative biomarker in breast carcinoma; however, methodological variability in its assessment limits reproducibility and clinical applicability. The International Ki-67 in Breast Cancer Working Group (IKWG) has proposed a standardized global scoring method with a freely available Android application to address this issue. This study aimed to compare hotspot and weighted-global Ki-67 scoring methods using the IKWG Android application in invasive breast carcinoma, assess their correlation and categorization, and evaluate their diagnostic performance in predicting histological grade, with potential relevance to clinical decision-making and treatment stratification. Methodology A hospital-based, cross-sectional study was conducted on 53 cases of invasive breast carcinoma. Ki-67 immunohistochemical slides were digitally scanned and scored using both hotspot and weighted-global methods via the IKWG Android application. Statistical analysis was performed using the Mann-Whitney U test, Spearman’s correlation, and receiver operating characteristic analysis. Results The hotspot method produced a markedly elevated mean Ki-67 index (57.1 ± 22.6%) in contrast to the weighted-global method (43.1 ± 24.0%; p = 0.004). Using IKWG cutoffs, 44 (83.0%) tumors were classified as high proliferative by hotspot versus 36 (67.9%) by weighted-global scoring. Both methods showed a very strong correlation (Spearman’s r = 0.954, p < 0.001). The weighted-global Ki-67 demonstrated excellent diagnostic performance for predicting high-grade tumors (area under the curve = 0.991), with 100% specificity and 84.9% overall accuracy. Grade 2 tumors predominated (26 (49.1%)), followed by Grade 3 (18 (34.0%)) and Grade 1 (9 (17.0%)). Conclusions Hotspot scoring systematically overestimates tumor proliferation. The weighted-global method, incorporating intratumoral heterogeneity, provides a more balanced and biologically representative Ki-67 estimate. Standardization via the IKWG Android application improves reproducibility and supports reliable clinical decision-making.
Patel et al. (Thu,) studied this question.