Abstract PD06-02: Standardized Assessment of Ki-67 Using Virtual Slides and Automated Analyzer for Breast Cancer Patients: Comparison of Automated and Central/Local Pathology Assessment

Abstract

Abstract Background: Ki-67 is useful in determining the effect of hormonal therapy and also as a prognostic factor. However, the method for its evaluation has not been standardized, and the results vary depending on the assessment by pathologists. We reported that the discordance rate of Ki-67 in preoperative core needle biopsy and the surgical specimens was higher than that of other biological markers, and this was considered to be due to the discrepancies in the evaluation methods (SABCS 2011). In this report, we reviewed the possibility of introducing an automated analyzer by comparing the diagnosis made by local pathologists and reassessment by central review to standardize the evaluation method for Ki-67. Subjects and Method: Of the 354 cases of primary breast cancer that had their first operation from October 2008 to March 2012, 225 cases that had preoperative core needle biopsy at Tokyo-West Tokushukai Hospital were selected as the subjects of this study. Cases with ductal carcinoma in situ (DCIS) and cases with neoadjuvant chemotherapy were excluded. A cut-off value of 20% was used for Ki-67 positive criteria. The concordance rates of estrogen receptors (ER), progesterone receptor (PgR), human epidermal growth factor receptor 2 (HER2), and Ki-67 by local pathologists were reviewed, and in cases of non-matching Ki-67, tumor diameter (≈ heterogeneity) and operative method (≈ condition of formalin fixation) were studied. Next, non-matching cases (∼ to October 2011) were reassessed both by central review and by an automated analyzer (Ventana Digital Pathology) using virtual slides, and the results were studied and compared with the evaluation by local pathologists. The Wilcoxon-t test was used to review the p value for statistical analysis. Parameters such as tumor diameter, operative method were analyzed by χ2 analysis. Results: The concordance rate of Ki-67 in preoperative core needle biopsy and surgical specimens was 76.9% in 225 cases of local pathology assessment from October 2008 to March 2012. This rate was significantly lower (p < 0.001) than that of ER (95.6%), PgR (88.0%), and HER2 (91.5%). In non-matching cases, no difference was observed in tumor diameter and operative method. Reassessment of the 39 cases of non-matching Ki-67 to October 2011 using central review and an automated analyzer resulted in matching 27 cases (94.8%) and 32 (96.3%) out of 39 cases, respectively. Conclusion: The concordance rate of Ki-67 in preoperative core needle biopsy and surgical specimens was lower compared with other biological markers. However, they were nearly equal with reassessment using central review and an automated analyzer. Clinical application of Ki-67 and its standardization are challenging due to variable counting methods and measurement sites. Conducting central reviews in general clinical facilities is difficult, but introducing an automated analyzer with a Ki-67 labeling index is shown to achieve that goal. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr PD06-02.