Circulating tumor DNA (ctDNA) monitoring to inform maintenance outcomes in patients (pts) with advanced NSCLC treated with induction atezolizumab+carboplatin+nab-paclitaxel (A+CnP).

9075 Background: Chemoimmunotherapy (ChemoIO) is a prevalent first-line treatment for advanced NSCLC without driver mutations, with maintenance therapy recommended after induction. However, real-world data suggests variability in maintenance therapy’s timing, intensity, and duration. Motivated by evidence that ctDNA monitoring can predict outcomes in pts receiving IO for advanced cancers, we hypothesized ctDNA monitoring could inform outcomes in advanced NSCLC prior to the start of maintenance therapy. Methods: This retrospective study included 98 pts from a completed phase III trial of A+CnP in squamous NSCLC (IMPower131; GO29437). Pts were treated with 4 or 6 cycles of induction A+CnP followed by maintenance atezolizumab. ctDNA monitoring utilized FoundationOneTracker, involving 1) comprehensive genomic profiling of pretreatment tumor tissue, 2) variant selection using an algorithm to filter out non-tumor variants, and 3) multiplex PCR of up to 16 variants to detect ctDNA and quantify plasma mean tumor molecules per mL (MTM/mL). Progression free-survival (PFS) and overall survival (OS) were estimated with Kaplan-Meier analysis and hazard ratios (HR) were calculated using multivariate Cox proportional hazard models adjusting for age, smoking history, PD-L1, and performance status. PFS and OS analyses were landmarked from C4D1 unless otherwise stated; pts with progression/death before the landmark date were excluded. Results: TP53 (91%), CDKN2A (41%) and PIK3CA (32%) were frequently altered in baseline tissue; 51% had TMB ≥10 mut/Mb. A median of 10 variants (range 2-16) were tracked per pt. ctDNA was detected (ctDNA+) in 43% of C4D1 samples (median 15.2 MTM/ml, range 0.3-787.3). ctDNA+ at C4D1 was associated with a lower objective response rate (41%) compared to undetectable ctDNA (ctDNA-, 77%, p < 0.001). Pts with ctDNA- at C4D1 (56/96) had better PFS than their ctDNA+ counterparts (median PFS from C4D1 7.7 vs 2.8 months; HR: 3.07 [1.88-5.02]). Median OS from C4D1 was not reached for ctDNA- pts and was 8.3 months for ctDNA+ pts (HR: 4.31 [2.31-8.06]). In a C6D1 landmark analysis of 34 pts who were ctDNA+ at C4D1, prolonged induction (6 cycles vs 4 cycles) was not associated with any improvement in PFS (1.6 vs 2.2 months, HR 1.41 [0.58-3.39]) or OS (4.7 vs 9.8 months, HR 1.84 [0.71-4.76]). Additional analysis of plasma samples at baseline and on-treatment time points is ongoing. Conclusions: On-treatment ctDNA monitoring during induction chemoIO can inform outcome on subsequent maintenance therapy in pts with advanced NSCLC. For pts with ctDNA+, prolonged induction chemoIO was not associated with improved outcomes. ctDNA testing during induction chemoIO may offer an opportunity to identify pts at higher risk for disease progression and inform selection for novel personalized maintenance treatment strategies.