219 Background: In dMMR CRC, predictive biomarkers to guide the use of anti-PD1 single-agent versus anti-PD1 plus anti-CTLA4 combination therapy are a critical unmet need. We investigated the feasibility of cfRNA as a minimally invasive approach to capture tumor- and immune-derived signals associated with benefit from PD1-based therapy in dMMR CRC. Methods: cfRNA was extracted and the average expression profile of purified platelets was proportionally extracted based on the platelet fraction estimated via deconvolution (Zaitsev et al. Cancer Cell 2022). Activity of cell-type specific and process-related signatures, previously tested on independent cfRNA samples, was measured using ssGSEA. Immunotype signatures from peripheral blood (Dyikanov et al. Cancer Cell 2024) were adapted to cfRNA data using linear regression models trained on independent paired blood cell and plasma RNA-seq data. cfRNA donor clusters generated in an independent cohort (N=439) using KMeans clustering on median-scaled scores of 87 immune and cancer signatures were applied to cfRNA data. Association with tumor response and progression-free survival (PFS) were assessed using univariate Cox models. Results: Baseline plasma from 40 dMMR CRC patients (median age 61.4years, 42.5% female, 57.5% metastatic, 30% RAS mutated) treated with PD1-based therapy (82.5% PD1 single-agent, 10% PD1 + CTLA4, and 7.5% PD1 + other) were analyzed. Overall response rate was 77.5% and mean PFS was 40.3 months. To develop a cfRNA-based therapeutic outcome predictor, we evaluated pathway and cell type-related signatures and PFS(Table). Signatures of myeloid suppression, pro-inflammatory cytokines, CD8 T cell (CTL) exhaustion, and checkpoint inhibition were associated with shorter PFS, whereas a B cell signature correlated with longer PFS. The previously identified G2 (primed) immunotype signature was enriched in responders (p-value = 0.06, ROC AUC = 0.7). An epithelial signature correlated with shorter PFS and the presence of metastasis. Finally, a cfRNA cluster enriched in CTL, B cell, and antigen presentation activity with reduced myeloid signal was associated with longer PFS. Conclusions: cfRNA profiling is feasible and captures tumor and immune-derived signals associated with immunotherapy benefit in dMMR CRC. Prospective validation in larger cohorts is warranted. Signature PFS directionality logHR (95% CI) p-value B cell Longer -2.68 (-5.33 to -0.03) 0.047 Epithelium Shorter 2.51 (0.08 to 4.93) 0.043 Stromal suppression Shorter 2.55 (0.09 to 5.02) 0.042 Checkpoint inhibition Shorter 2.80 (0.54 to 5.06) 0.015 Myeloid suppression Shorter 3.13 (0.04 to 6.21) 0.047 CTL Exhaustion signature Shorter 3.20 (0.22 to 6.18) 0.036 TNF response (Acute) Shorter 3.25 (0.03 to 6.19) 0.031 Cytotoxic cell inactivation Shorter 3.85 (0.58 to 7.12) 0.021 Pro-inflammatory cytokines Shorter 8.17 (1.47 to 14.88) 0.017
Overman et al. (Sat,) studied this question.