P2.01-23 Baseline Plasma Biomarkers Predict Long-Term Responses to ALK-TKIs in ALK+ Advanced Non-Small Cell Lung Cancer (NSCLC)

Abstract

Median progression free survival (PFS) for ALK tyrosine kinase inhibitors (ALK-TKIs) range from 10.9-25.7 months (mos)[1],[2],[3],[4]. While most ALK+ NSCLC patients develop resistance to ALK-TKIs within 1-2 years, a subset of patients experience a response >2 years. Given the lack of proteogenomic markers predicting long-term responses to ALK-TKIs, we conducted whole-exome sequencing (WES) and proteomic profiling to define molecular determinants that could predict a long-term ALK-TKI response, help guide the sequentiality of therapies and optimize treatment strategies. [1]Soria, J.C. et al. Lancet389,917–929 (2017) [2]Peters, S. et al. N. Engl. J. Med.377, 829-838(2017) [3]Kim, D.-W. et al.J. Clin. Oncol.35,2490–2498 (2017) [4]Solomon, B. J. et al.N. Engl. J. Med.371,2167–2177 (2014). Twenty-four patients with advanced ALK+ NSCLC were enrolled in our study. WES was performed on primary and post-treatment metastatic tissue to identify genomic aberrations and ALK fusions. MRM-MS was used to analyze 327 protein candidates in plasma collected from patients at baseline. Patients were categorized into 3 groups based on duration of response: long-term responders [LR; PFS ≥24 mos (n=8)], normal responders [R; 3 < PFS < 24 mos (n=10)] and non-responders [NR;PFS <3 mos (n=6)]. At data cutoff (30 April 2018), median PFS was 1.6 mos for NR, 11.7 mos for R and 37.1 mos for LR. Two LR remain on treatment and have experienced a PFS > 37.4 mos. Despite detecting novel ALK fusion partners, multiple somatic mutations and copy number aberrations by WES, we could not define a genomic signature predictive of a long-term response to ALK-TKIs from our small cohort. However, MRM-MS identified 15 proteins differentially regulated between LR and NR, including SODE, F13A, LYAM1, FCGBP, PGBM and LUM. Differences in protein levels were further pronounced between LR and NR. To determine whether our protein signature can discriminate according to response groups, we performed principal component and hierarchical clustering analyses. Both analyses successfully segregated LR from R and NR. Moreover, we used our set of 15 proteins to generate single-sample Gene Set Enrichment Analysis scores which distinguished LR from R and NR as distinct groups. Targeted proteomic profiling of baseline plasma from ALK+ NSCLC patients identified a protein signature that may predict a long-term response or resistance to ALK-TKIs. A collaboration is in development to confirm the validity of this protein signature in a larger cohort, and with next-generation ALK-TKIs.