Automated and Visual Assessment of Ki-67 IHC in Breast Carcinoma: Comparison of 3 Antibody Clones
Eugen C Minca, Sheppard A Elizabeth, Tsuta Koji, Free Heather, Tubbs R Raymond. Cleveland Clinic Foundation, Cleveland, OH; Ventana Medical Systems, Tucson, AZ; National Cancer Center Research Institute and Hospital, Tokyo, Japan
Background: The expression of Ki-67 is widely used as a marker for tumor cell proliferation and is an established prognostic factor in breast cancer. The immunohistochemical (IHC) Ki-67 labeling index (LI) is commonly visually estimated as percentage of tumor cells with positive nuclear staining. Although the mouse MIB-1 is the most referenced antibody clone for Ki-67 IHC, several newer rabbit monoclonal antibodies are commercially available, with little data comparing their performance. Furthermore, an automated platform has been recently FDA-approved for Ki-67 analysis in breast cancer (Ventana). Here we compare 3 antibodies for Ki-67 LI by both visual and automated analysis, on a collection of breast carcinomas.
Design: Three tissue microarrays containing 235 formalin-fixed paraffin-embedded breast cancer samples from 3 institutions (UK, Japan, Canada) were stained for Ki-67 with the SP6 (Neomarkers/LabVision), 30-9 (Ventana), and MIB-1 (Dako) antibody clones using optimized protocols. The LI as percentage cells with positive staining was analyzed for each antibody on each array by using the Virtuoso Software and iScan Coreo Au scanner platform (Ventana) and manually by two independent pathologists. Statistical analysis was performed using Stata 12/SE software (StataCorp).
Results: MIB-1 antibody showed mean Ki-67 LIs of 9% (automated analysis), 7.3% (manual #1) and 22.9% (manual #2). 30-9 and SP6 rabbit antibodies showed similar and significantly higher Ki-67 LIs compared to MIB-1: automated 25% and 27.5%, manual #1 21% and 21.4%, manual #2 30.9% and 39.1%, respectively (p<0.001 in all six instances). Furthermore, the LI read variation was lower for all three antibodies with the automated analysis (SD 12.8% [MIB-1], 22.3% [30-9], 23.2% [SP6]) compared to the manual reads (SD 17.4% and 17.8% [MIB-1], 25.5% and 22.5% [30-9], 28.4% and 25.4% [SP6]).
Conclusions: The 30-9 and SP6 rabbit monoclonal antibodies show similarly higher sensitivities for Ki-67 LI analysis in breast carcinomas compared to MIB-1 antibody. Automated analysis for tumor proliferation index is a more accurate alternative to manual counts. Further correlation with the clinical outcome is required to validate these tools as highly-sensitive and accurate for detecting tumor proliferation index.
Tuesday, March 5, 2013 1:00 PM
Poster Session IV # 56, Tuesday Afternoon