Plasma-TMB (pTMB) and circulating tumor fraction (ctF) dynamics as biomarkers of benefit from the addition of atezolizumab to first-line FOLFOXIRI/bevacizumab in metastatic colorectal cancer (mCRC): A translational analysis of the AtezoTRIBE study.

Abstract

2575 Background: TMB-high pMMR mCRC patients (pts) do not seem to derive benefit from immune checkpoint inhibitors (ICIs). In the CO.26 trial, pMMR/MSS mCRC pts with high baseline pTMB had better outcome with ICIs than those with low pTMB. Moreover, in the AtezoTRIBE study, tissue TMB was identified as a promising biomarker able to identify a subgroup with potential benefit from the addition of atezolizumab to first-line FOLFOXIRI/bev. In that trial, the addition of atezolizumab did not improve ORR, and PFS curves separated only after approximately 6 months. Given this delayed treatment effect observed, we hypothesized that dynamic liquid biopsy biomarkers could identify early molecular changes associated with subsequent immune-sensitivity. Here we evaluated the prognostic and predictive role of both pTMB and on-treatment ctF dynamics in the AtezoTRIBE trial. Methods: Pts with untreated mCRC enrolled in the AtezoTRIBE trial (NCT03721653) and with available plasma at baseline (T0) and at the end of induction (T1) were included. Samples were analyzed using the Tempus xF assay, a 105-gene cell-free liquid biopsy panel, to determine ctF by Tempus xFv2 cTF algorithm, and pTMB. pTMB was assessed as a dichotomous variable (high: ≥26 mut/Mb). ctF dynamics between T0 and T1 identified molecular responders (MRs), i.e. pts with the top 30% relative ctF reduction including clearance below quantification/detection limits at T1, and non responders (nMRs). Cox models stratified by liver metastases, and including treatment-by-biomarker interaction when appropriate, were used for PFS analyses (likelihood ratio test, p<0.15). Results: pTMB was evaluable in 123/155 (79%) T0 samples, and 20 (16%) were classified as pTMB-high. Baseline characteristics were balanced between pTMB-high and pTMB-low groups. High pTMB was associated with longer PFS (HR: 0.69 [90%CI 0.43-1.10], p=0.095), while no interaction between pTMB and treatment arm was observed (p int =0.871). Among 89 pts with evaluable ctF at T0 and T1, 63 (71%) were MRs. Baseline clinical and molecular characteristics did not differ in the MR and nMR groups. A significant interaction (p=0.132) between ctF dynamics and MR was reported, with nMRs achieving higher benefit from the addition of atezolizumab to FOLFOXIRI/bev (HR: 0.33 [90%CI: 0.16-0.69]) than MRs (HR: 0.76 [90%CI: 0.03- 3.95). A sensitivity analysis including 6 additional pts with undetectable ctF both at T0 and T1 provided consistent findings (p int =0.108). Conclusions: The addition of ICI to first-line chemotherapy may be more efficacious among mCRC pts with lower molecular response to the upfront induction regimen, as measured by ctF dynamics. If independently validated, ctF molecular response biomarker might be useful for future trials’ design.