Translational relevance of androgen receptor immunohistochemistry scoring systems for data harmonization in triple negative breast cancer (TNBC).

Abstract

1078 Background: Androgen receptor (AR) expressing triple negative breast cancer (TNBC) is a sub-set of TNBC with an evolving prognostic and predictive behaviour. AR immunohistochemical threshold for positivity has not been standardized and a wide range of cut-offs have been used across studies ( > 0% to 75%). In this study we explored AR immunohistochemistry thresholds in relation to disease free survival (DFS) and clinical outcomes in non-metastatic TNBC using the Allred and H-Score systems. Increasing interest in AR as a therapeutic target for TNBC and the use of digital tissue image analysis makes it important to standardize AR immunohistochemistry reporting. Methods: 100 FFPE (formalin-fixed paraffin-embedded) tumour blocks were retrieved for non-metastatic TNBCs diagnosed between January 2015 and May 2017 and immunostained using AR441 (IgG1) mouse monoclonal antibody. Clinical follow-up ranged from 59 to 31 months and DFS was calculated. Cut-off scores were explored using Evaluate Cutpoints ( R maxstat package) and X-tile software. The score with maximum split in DFS (based on log-rank statistics and lowest p-value) was chosen as the cut-off. Descriptive and survival statistics was performed. Results: The median age was 51 (SD 11.262; range 28 to 82) years. Using Evaluate Cutpoints ≥3 was found as the threshold for AR by Allred Score. 36% cases were AR positive using Allred score (HR 0.508; CI 0.234 - 1.11; p-value 0.08). Using Evaluate Cutpoints ≥30 was found as the threshold for AR by H-Score (HR 0.624, CI 0.306 - 1.27; p-value 0.19). 35% cases were AR positive using H-Score. X-tile analysis also found the cut-offs as ≥3 and ≥30 for Allred and H-Score respectively (p < 0.05). A significant correlation was seen between the two scoring systems (Pearson Correlation 0.935; p < 0.01). A significantly higher number of grade III TNBCs were AR negative (n = 55/76) compared to grade II (n = 9/24) (p = 0.002). Cut-off for Ki67 was 75 (HR 1.61, CI 0.85-3.04, p-value 0.141) with a significantly higher number of AR negatives in the Ki67≥75 group (21/26; p < 0.05). The overall median DFS was 51.9 months. There was no significant difference in DFS for the AR negative (median: 47.4 months; mean: 39.39 months) and AR positive (Median survival not reached; mean: 41.3 months) groups(p = 0.23). Conclusions: AR immunohistochemistry cut-offs using the Allred (≥3) and H-Score (≥30) are close to the ones used for ER/PR immunohistochemistry as per ASCO/CAP guidelines, making a strong case for universal application of these systems for harmonization of AR data.