OA06.05 Proteomic Analysis of ERCC1 Predicts Benefit of Platinum Therapy in NSCLC: A Reevaluation of Samples from the TASTE Trial

Abstract

It is hypothesized that low or absent expression of the excision repair cross-complementation group 1 (ERCC1) protein predicts improved survival in NSCLC patients treated with platinum-based therapy. However, the International Adjuvant Lung Cancer Trial Collaborative Group concluded that current ERCC1 assessment methods are inadequate for clinical decision-making. Due to the unreliability of ERCC1 immunohistochemistry (IHC), the IFCT-0801 TASTE (Tailored Postsurgical Therapy in Early-Stage NSCLC) trial of adjuvant therapy for NSCLC was discontinued. We reevaluated a subset of samples from the TASTE trial using mass spectrometry-based proteomics to quantitate ERCC1 protein. We correlated ERCC1 proteomic status with survival after chemotherapy with cisplatin/pemetrexed and compared it to ERCC1 IHC ranking. Formalin-fixed, paraffin-embedded NSCLC tumor tissues were laser microdissected, solubilized, digested, and proteomically analyzed. A multiplexed, selected reaction monitoring mass spectrometric assay was used to quantitate levels of multiple proteins including ERCC1. The Kaplan-Meier method and univariate Cox analysis assessed overall survival (OS) and relapse-free survival (RFS). A chi-squared test compared binary proteomic levels of ERCC1 (detectable vs. undetectable) with the IHC status assessed using an anti-ERCC1 antibody (8F1) during the TASTE trial. Of 146 evaluable patients, 33 (22.6%) had undetectable ERCC1 by quantitative proteomics. Proteomics found no detectable ERCC1 protein in 8/36 (22.2%) IHC-positive patients nor in 8/22 (19.3%) IHC-indeterminate patients. ERCC1 was detected in 71/88 (80.7%) IHC-negative patients (range: 36-137 amol/μg total tumor protein). Undetectable ERCC1 by proteomics was prognostic of OS (hazard ratio [HR]: 5.45; p=0.031). In survival analyses of cisplatin-treated patients (n=122), only one of the 15 deaths occurred among the patients with undetectable ERCC1 protein. These patients had better OS than cisplatin-treated patients with detectable ERCC1, although the difference statistically nonsignificant (HR: 3.98; p=0.102). RFS was similar between patients with and without detectable ERCC1. GARFT protein (predictive of response to pemetrexed) was quantified in 100% of patients (range: 492-4006 amol/μg). The 10 cisplatin/pemetrexed-treated patients with GARFT levels >900 amol/μg had nonsignificantly worse OS than their counterparts with lower GARFT levels (p=0.08). Although underpowered to detect statistically significant survival differences, this study clearly demonstrates that quantitative proteomics can increase accuracy in identifying NSCLC patients who will respond to platinum-based therapy because they do not express ERCC1. Approximately 28% of such patients were misclassified by ERCC1 IHC in the TASTE trial. Clinicians should be aware that multiplexed quantitative proteomics can quantitate ERCC1 simultaneously with multiple clinically relevant proteins in lung tumors and small biopsies.