Analytical validation of a standardized scoring protocol for Ki67: phase 3 of an international multicenter collaboration

Pathological analysis of the nuclear proliferation biomarker Ki67 has multiple potential roles in breast and other cancers. However, clinical utility of the immunohistochemical (IHC) assay for Ki67 immunohistochemistry has been hampered by unacceptable between-laboratory analytical variability. The International Ki67 Working Group has conducted a series of studies aiming to decrease this variability and improve the evaluation of Ki67. This study tries to assess whether acceptable performance can be achieved on prestained core-cut biopsies using a standardized scoring method. Sections from 30 primary ER+ breast cancer core biopsies were centrally stained for Ki67 and circulated among 22 laboratories in 11 countries. Each laboratory scored Ki67 using three methods: (1) global (4 fields of 100 cells each); (2) weighted global (same as global but weighted by estimated percentages of total area); and (3) hot-spot (single field of 500 cells). The intraclass correlation coefficient (ICC), a measure of interlaboratory agreement, for the unweighted global method (0.87; 95% credible interval (CI): 0.81-0.93) met the prespecified success criterion for scoring reproducibility, whereas that for the weighted global (0.87; 95% CI: 0.7999-0.93) and hot-spot methods (0.84; 95% CI: 0.77-0.92) marginally failed to do so. The unweighted global assessment of Ki67 IHC analysis on core biopsies met the prespecified criterion of success for scoring reproducibility. A few cases still showed large scoring discrepancies. Establishment of external quality assessment schemes is likely to improve the agreement between laboratories further. Additional evaluations are needed to assess staining variability and clinical validity in appropriate cohorts of samples.