Real-world (rw) analysis of quantitative MET copy number (CN) as a biomarker in NSCLC.

Abstract

9123 Background: MET amplification (amp) can be a de novo driver or mechanism of resistance to targeted therapy. We explored the genomic landscape of MET amp NSCLC, including co-driver alterations and amplicon size, and its association with outcomes to MET tyrosine kinase inhibitors (TKIs) using rw data. Methods: Comprehensive genomic profiling (CGP) results from 64,521 tissue and 5,177 liquid NSCLC samples were queried for MET amp (focal amplicons ≥4 copies above specimen ploidy). Sub analysis comparing focal (≤20Mbp) and non-focal (> 20Mbp) amp was performed. Using the nationwide (̃280 US cancer clinics) de-identified EHR-derived Flatiron Health-Foundation Medicine aNSCLC clinicogenomic database (CGDB) linked to tissue CGP (8/2014-9/2021), we assessed co-driver presence ( EGFR, ALK, RET, ROS1, NTRK, BRAF, KRAS) prior to therapy exposure and outcomes to MET TKIs. Results: MET amp was detected in 2,102 (3.3%) NSCLC tissue samples (median CN of 11, range 6-168). Evaluating full gene focal amplicons as well as non-focal MET amps (n = 398, median CN 7), smaller amplicon size significantly correlated with higher CN (p < 0.001). In 5,177 liquid samples, MET amp was detected in 40 (0.77%) cases; frequency increased to 3.2% (33/1,033) in those with a tumor fraction above 10%. In the CGDB, 261 (3.2%) pts had NSCLC with MET amp (median CN 11). Among 241 pts with MET amp detected prior to receipt of any targeted therapy, MET amp co-occurred with another driver in 32 (13%) cases (81% KRAS, 16% EGFR, 3.1% BRAF) and with a MET exon 14 skipping alteration ( METex14) in 25 (10%). MET CN negatively correlated with the presence of a co-driver (median 11 with MET amp alone vs 9 with a non- METex14 co-driver, p < 0.001); however, MET CN was similar in MET amp cases with and without METex14 (median 11 for both, p = 0.60). Co-occurring non- METex14 drivers were detected in 26% of MET amp cases with CN < 8 and 19% with CN 8-9, but in only 9.1% with CN 10-20 and 2.6% with CN > 20. Tumor mutational burden was not significantly correlated with MET CN. In 14 cases with an EGFR or ALK co-driver and MET amp detected post-treatment with an EGFR or ALK TKI, median MET CN was 13. Of 241 targeted therapy naïve MET amp pts, 39 received a MET TKI after CGP (30 crizotinib, 9 capmatinib) and 26 had rw response assessment available. Excluding 5 pts with co- METex14, 12/21 (57%) pts had a partial response and median rwPFS was 3.6 months. This rw cohort analysis was underpowered to assess the relationship between CN and outcome. Conclusions: MET amp was detected in 3.3% of NSCLC tissue samples and 3.2% of high tumor fraction blood samples. In TKI-naïve pts, MET CN negatively correlated with the presence of a concurrent NSCLC driver. MET amp was associated with response to MET TKIs. Further studies evaluating MET CN, amplicon size and presence of other potential drivers in both blood and tissue, as predictive biomarkers for MET TKIs and potential combination strategies for targeting MET amp, are warranted.