Abstract PD06-06: RELATIONSHIP BETWEEN MOLECULAR SUBTYPE AND CHANGE IN KI-67 IN THE PLACEBO ARMS OF WINDOW OF OPPORTUNITY TRIALS

Abstract

Abstract Background: Window-of-opportunity (WOP), presurgical models often use Ki-67 labeling index (LI) as the main surrogate biomarker for screening therapeutic activity of candidate agents and characterize their mechanism of action. Ki-67 LI in the placebo arm has been frequently reported to be higher at surgery than at baseline biopsy in WOP studies, a finding that has been ascribed to the higher proliferative activity in the tumor edge of surgical specimens that cannot be detected in the baseline core biopsy. Since this observation has important implications in sample size calculation and effect interpretation, we investigated which factors affected the changes of Ki-67 LI in a series of patients allocated to the placebo arms of a number of WOP trials carried out at IEO. Methods: Data from 181 patients with pT1-2 invasive breast cancer from the placebo arms of three WOP randomized trials were pooled with those of 98 untreated patients, who had a diagnostic core biopsy preceding surgery but declined participation to the study or were ineligible for histological characteristics. Ki-67 LI was measured by evaluating the prevalence of neoplastic cells showing any definite nuclear immunostaining with the Mib-1 monoclonal antibody, in the whole invasive component of baseline core biopsies and in at least 2000 invasive neoplastic cells of surgical samples, in accordance to the recently licensed international recommendations. Results: We collected data of 273 breast cancer patients with information on the changes in Ki-67 LI: 46 (17%) were Luminal A, 85 (31%) Luminal B/HER2−, 24 (9%) Luminal B/HER2+, 38 (14%) HER2+ and 81 (30%) Triple Negative (TN). Median (IQR) Ki-67LI at baseline in each molecular subtype was 10% (7–11), 20% (16–29), 28% (22–35), 30% (25–45) and 50% (32–75) in the Luminal A, Luminal B/HER2−, Luminal B/HER2+, HER2+ and TN, respectively. Median (IQR) age was 50 years (44–60), median BMI was 24 (22–27), 50% were post-menopausal. The median (range) time elapsed from biopsy to surgery was 41 days (33–48). Overall, the median change in Ki-67 LI between baseline biopsies and surgical samples was 0 (IQR, −2, 5). Results from multivariate analysis showed that none of the factors investigated, including patient and tumor characteristics, time elapsed from biopsy to surgery and circulating biomarkers (IGF-I, SHBG and ultrasensitive CRP) were associated with the Ki-67 LI change except for the immunohistochemically defined molecular subtype, which explained most of the variability of the changes (p = 0.004). As a matter of fact, we observed a 5%, significant increase of Ki-67 LI both in HER2+ and TN tumors, after adjustment for baseline values. Conclusions: We reported a significant increase in Ki-67 LI between baseline biopsy and endpoint surgery in the placebo arms of HER2+ and TN tumors. This association suggests a real biological increase in proliferation rather than an analytical artifact, and should be taken into account in designing future WOP studies. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr PD06-06.