Changes in ctDNA levels as an early indicator of outcomes in advanced NSCLC treated with TKI: Initial findings from a retrospective aggregate analysis of 8 clinical trials

3030 Background: To determine whether changes in circulating tumor DNA (ctDNA) levels reflect treatment outcome, Friends of Cancer Research created the ctDNA to Monitor Treatment Response (ctMoniTR) Project with collaborators from industry, government, academia, and advocacy. A prior ctMoniTR effort analyzing 5 clinical trials (CT) showed an association between decreases in ctDNA levels and improved outcomes in patients with advanced non-small cell lung cancer (aNSCLC) treated with an anti-PD-(L)1. The current study expands that work and focuses on CT investigating tyrosine kinase inhibitors (TKI) treatment in oncogene-driven aNSCLC. Methods: We performed a retrospective analysis of patient-level clinical and ctDNA data across 8 CT representing 1,015 patients with aNSCLC treated with TKI (i.e., anti-EGFR, ALK, RET, or MET). Patients included in the analysis had a baseline ctDNA measurement (T0), an on treatment ctDNA measurement within 10 weeks of treatment initiation (for those with multiple ctDNA measurements within 10 weeks, we used the lowest measurement within 10 weeks) (T1), and overall survival (OS) data (n=749). CT used different ctDNA collection timepoints and assays. We randomly divided the dataset into training (2/3 of the data) and validation (1/3 of the data) datasets stratified by CT cohort (i.e., arm), age, tumor stage, and prior lines of therapy, then ran initial analyses on the training dataset (n=501; reported herein). ctDNA change was calculated as the percent change in maximum variant allele frequency (VAF) between T0 and T1 using tumor-derived variants provided by sponsors for each unique patient sample. CT used either ddPCR or an NGS assay. ctDNA limits of detection were assay specific and varied across CT. Multivariate analyses are ongoing and validation dataset analyses will be conducted. Results: At T0, 141 patients had non-detected (ND) ctDNA and 360 patients had detected (D) ctDNA. Of these, 27% (n=136) had ND ctDNA at both T0 and T1 (“ND/ND”), 52% (n=260) had changes from D at T0 to ND at T1 (“D/ND”), 12% (n=60) had at least a 50% decrease from T0 to T1 (“decrease”) and 9% (n=45) had an increase or a less pronounced decrease in ctDNA. In a univariate analysis, patients with ND/ND and D/ND were associated with improved OS compared to the decrease group. In addition to other characteristics, patients with max VAF ≤0.5% or ND at T0 (n=214, 43%) had improved OS (HR=0.44, P<0.001) compared to those with max VAF >0.5% at T0 (n=287, 57%). Conclusions: In a retrospective aggregate analysis of 8 CT, ND ctDNA at T1 was associated with improved OS in patients with aNSCLC treated with TKI. Changes in ctDNA levels, particularly from D to ND, may provide an early indication of treatment benefit and predict long-term outcomes in this population. Additional ctMoniTR analyses are ongoing to validate the potential use of ctDNA as an early endpoint.