Association of circulating tumor DNA (ctDNA) variant allelic frequency (VAF) with outcomes on matched targeted therapies (TT) in advanced non-small cell lung cancer (aNSCLC).

Abstract

8628 Background: There is a critical gap in our understanding of the correlation between ctDNA driver VAF and outcomes in patients (pts) with aNSCLC treated with TT. We explore the landscape of driver alterations (alts) in aNSCLC by ctDNA VAF and assess the association with outcomes in pts treated with matched TT. Methods: We analyzed 3,146 pts with aNSCLC with a driver positive liquid biopsy (LBx) (FoundationOneLiquid CDx) Targetable driver altswere defined as alts listed in NCCN guidelines and were stratified by VAF. Clinical outcomes were assessed for pts included in the nationwide (US-based) de-identified Flatiron Health-Foundation Medicine aNSCLC clinico-genomic database (FH-FMI CGDB) originating from approximately 280 US cancer clinics (~800 sites of care). For 224 pts receiving matched TT within 60 days of LBx collection, multivariate Cox proportional hazard models were employed to evaluate the association of VAF on real-world progression-free (rwPFS) and overall (rwOS) survival adjusting for clinical and demographic factors. Results: Among 3,146 pts with targetable alts detected in LBx, the frequency of drivers with VAF <1% in ctDNA was 36% (1,185/3,262 alts). Distribution of drivers by VAF is shown in the table. For pts in the FH-FMI CGDB treated with matched TT following driver positive LBx, clinical and demographic characteristics were balanced between pts with driver VAF <1% (n = 75) and those with VAF ≥1% (n = 147), except for the presence of liver metastases, which were more common in pts with VAF <1% (12% v 26%; p = 0.0002). There was no significant difference in the median rwPFS for pts with driver VAF <1% vs those with VAF ≥1% (10.8 vs 8.7 months; HR = 1.40 [0.92-2.00]; p=0.12). Similarly, there was no significant difference in median rwOS (32.4 vs 23.2 months; HR 1.20 [0.74-2.00]; p=0.45). To account for potential bias due to varying effectiveness of TT by driver, we limited the analysis to pts treated with EGFR inhibitors: VAF of the EGFR mutation did not affect rwPFS (10.8 vs 9.8 months; HR 1.14 [0.69-1.90]; p=0.61) or rwOS (18.3 vs 23.2 months; HR 1.1 [0.61-2.00]; p=0.74). Conclusions: Outcomes in pts with aNSCLC receiving matched TT after LBx were comparable between pts with driver VAF <1% and those with VAF ≥1%. Our findings highlight that the presence of a detectable targetable driver alt in aNSCLC is actionable, regardless of ctDNA VAF. Alteration VAF <1%(n = 1,185) VAF ≥1%(n = 2,077) Percentage <1% VAF [95% Confidence interval] VAF Range KRAS G12C 288 566 34 [31-37] 0.1% - 78% EGFR 457 1,065 30 [28-32] 0.09% - 98% ALK 129 126 51 [44-57] 0.04% - 49% RET 47 32 59 [48-70] 0.04% - 36% ROS1 41 22 65 [52-76] 0.05% - 47% MET exon 14 75 94 44 [37-52] 0.09% - 80% NTRK 1/2/3 13 7 65 [41-84] 0.11% - 17% BRAF V600E 71 59 55 [46-63] 0.09% - 35% ERBB2 64 106 38 [30-45] 0.09% - 77%