Comparative Study of Ki-67 Labeling Index Quantification by Eye-rolling, Manual Count, and Digital Image Analysis; An Approach with Caution.

Background and objective: An accurate Ki-67 labeling index assessment is critical for managing a few tumors, like breast carcinomas and neuroendocrine tumors. We aimed to determine the degree of agreement between digital image analysis (DIA) and eye-rolling assessment (EE) and DIA and manual count (MC) for Ki-67 LI scoring.

Methods: A total of 120 cases (both tru-cut biopsies and resected specimens) were selected during the study period from the institutional database, wherein the Ki-67 labeling index was performed. The selected cases were divided into two groups, i.e., breast neoplasms and other neoplasms. The correlation between DIA and EE and DIA and MC for Ki-67 LI scoring was calculated in both groups.

Results: A total of 113 cases were analyzed for Ki-67 LI by three different methods (EE, MC, and DIA); 7 cases were rejected due to poor image quality. Ki-67 LI scoring by DIA and EE was highly correlated in both study groups with a Spearman's rank correlation coefficient of 0.809 (P=0.01) and 0.904 (P=0.01), respectively. Correlation between DIA and MC methods was also found to be almost perfect in both study groups with a Spearman's rank correlation coefficient of 0.974 (P=0.01) and 0.955 (P=0.01), respectively.

Conclusion: ImmunoRatio is a free web-based digital image analysis application that can be used for Ki-67 LI assessment with considerable reliability and reproducibility. Yet, it carries a few limitations and demands a careful approach and final confirmation by an expert.