Impact of circulating tumor DNA (ctDNA) detection on survival outcomes of patients (pts) treated with immune-checkpoint inhibitors (ICIs) in early clinical trials.

Abstract

2542 Background: Detection of ctDNA is a promising tool for managing pts in oncology. Most methods require whole-genome sequencing of tumor samples followed by the design of personalized panels for tracking purposes. In this work, we evaluated the prognostic and predictive value of total ctDNA quantification, using shallow whole-genome sequencing (shWGS) exclusively from plasma samples, in a prospective cohort of pts treated with ICIs in early clinical trials. Methods: IchorCNA pipeline was used to quantify ctDNA of shWGS from plasma ctDNA samples of pts treated with ICIs in phase 1 trials, collected at baseline and prior to cycle 2 (prec2). We investigated the association and correlation of ctDNA levels with surrogate markers for tumor burden (LDH levels, summatory of target lesions (TL), liver metastasis) using Spearman and Kruskal-Wallis tests. Kaplan-Meier estimates of overall survival (OS) of pts with baseline detectable ctDNA levels versus undetectable ctDNA were calculated. A multivariate Cox proportional hazards model, including continuous classical prognostic factors (LDH, albumin, hemoglobin, derived Neutrophil-to-Lymphocyte ratio (dNLR), platelets, number of metastases sites, ECOG PS) and ctDNA was performed. An estimate of progression free survival (PFS) of pts with ctDNA increase levels in preC2 versus a non-increase group was evaluated. Results: Since January 2018, 113 pts with no standard-treatment options were included. Median (m) follow up was 14.8 months (mo). Baseline ctDNA levels correlated significantly with baseline TL (R = 0.4, p < 0.001) and LDH levels (R = 0.61, p < 0.001). Pts with liver metastasis had higher levels of ctDNA (11,68 ng/ml) versus pts with no liver disease (2,31 ng/ml) (p < 0,001). In the survival analysis pts with detectable baseline ctDNA (74 pts) had significantly shorter OS compared with pts with undetectable ctDNA (39 pts); median 9.6 m (8.4 – 16.4) and NA m (13.6-NA), respectively (HR = 2.25 [1.18-4.29] p < 0.01). In the multivariate analysis, only ctDNA and albumin levels maintained the impact in OS (HR = 1.03, p < 0.05 and HR = 0.22, p < 0.05, respectively). Pts with early increases in ctDNA had a shorter PFS compared with those with a stabilization or decrease, median 1.9 m (1.6 – 4.0) and 3.0 m (2.6 – 3.8), respectively (HR = 2.19 [1.31-3.67], p < 0.01). Conclusions: Quantification of baseline ctDNA using shWGS is a strong independent prognostic factor. Early dynamic changes of ctDNA could be a useful tool to predict PFS outcomes.