Abstract P5-02-10: Factors influencing agreement of Ki67 labeling index between core needle biopsy and surgical resection specimens

Abstract

Abstract Reliable determination of Ki67 labeling index (Ki67 LI) on core needle biopsy specimens (CNB) is essential for determining breast cancer intrinsic subtype (IST), preoperative treatment decisions and short-term treatment response during neoadjuvant therapy. However, analyses investigating robustness of Ki67 LI assessment upon repeated tumor biopsies are scarce and results of studies investigating agreement of KI67 LI between CNB and surgical resection (SR) specimens are conflicting. In the present study, we analysed the role of clinical and pathological factors in influencing concordance of Ki67 LI between CNB and SR specimens. 502 matched pairs of CNB and SR specimens of patients with invasive ER-positive, HER2-negative breast cancer were included in our study. Ki67 LI was determined according to recent recommendations. Luminal B disease was defined by Ki67 LI > 20% according to SR. Ki67 LI values were considered concordant by a divergence of 10% points. Agreement was calculated by Cohen's kappa. Associations with clinicopathological factors were analyzed by Chi square test and logistic regression. Factors investigated included age, menopausal status, CNB method, BIRADS assessment category of imaging abnormality, time between CNB and SR, extent of surgery, histological grade (including individual grading score components according to Elston and Ellis), tumor size, lymph node stage, estrogen- (ER), progesterone- (PR) and HER2-receptor status. A cutoff value of 20% for KI67 LI in SR demonstrated a sensitivity of 90% and a specificity of 60% for identifying luminal B breast cancer in CNB. Upon investigating measurement agreement, we found substantial agreement of Ki67 LI between CNB and SR specimens with a weighted kappa value of 0.837. IST assesment in CNB and SR showed only moderate concordance with a kappa value of 0.587. Concordant diagnosis of IST in CNB and SR was significantly associated with PR expression and histological grade (p>0.05). Agreement of Ki67 LI was higher in tumors expressing low and high levels of PR compared to tumors with intermediate PR score. 27% and 22% of low and intermediate grade breast cancers showed discordant IST in CNB and SR, respectively. In contrast, only 3% of high grade breast cancers differed regarding IST upon repeated measurements. Interestingly, concordance of IST was significantly associated with all separate grading score components in CNB samples such as extent of glandular differentiation (p=0.015), nuclear polymorphism (p=0.005) and number of mitotic figures (p>0.001). We conclude that agreement of breast cancer IST according to Ki67 LI between CNB and SR specimens is significantly influenced by histological tumor grade and PR status. These factors, including all three grading score components are likely to mirror tumor heterogeneity that might compromise obtaining a CNB sample representative of the entire tumor. Our results cast a doubt upon robustness of single CNB-driven measurements of prognostic indicators and outcome predictors in estrogen-receptor positive breast cancer of low or intermediate histological grade. Whether molecular testing of CNB specimens improves classification of luminal breast cancer and helps resolve diagnostic disparities remains to be determined. Citation Format: Zsuzsanna Bago-Horvath, Fabian Roessler, Philipp Wimmer, Martina Mittlboeck, Nicolas Kozakowski, Katja Pinker-Domenig, Rupert Bartsch, Peter Dubsky, Martin Filipits, Margaretha Rudas. Factors influencing agreement of Ki67 labeling index between core needle biopsy and surgical resection specimens [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P5-02-10.