Longitudinal circulating tumor DNA monitoring in predicting response to short-course neoadjuvant radiotherapy in locally advanced rectal cancer: Data from a phase III clinical trial (UNION).

Abstract

LBA3606 Background: The data presented here is from a multicenter, randomized, open-label, controlled Phase III clinical study evaluating the feasibility of short-course radiotherapy (shortRT) sequentially combined with camrelizumab and chemotherapy as neoadjuvant therapy (NAT) for locally advanced rectal cancer (LARC)(UNION). Our aim is to explore the value of circulating tumor DNA (ctDNA)-based minimal residual disease (MRD) in assessing the comparative efficacy of short-course and long-course chemoradiotherapy (CRT). Methods: A total of 244 plasma samples from 79 LARC patients, who underwent NAT prior to curative surgery, were collected at baseline (C1), on-NAT (C2), post-NAT (C3), and post-surgery (C4). Deep targeted panel sequencing of 556 cancer-related genes was performed. The changes in genomic features and ctDNA-MRD status during treatment were monitored, and the relationship between these changes and treatment response were explored. Results: During NAT, the ctDNA-MRD positivity rate showed significant declining trends. Patients with high baseline TMB tend to show a significant inclination towards major pathological response and tumor regression grade 0/1 after NAT, while there is no significant correlation observed between baseline ctDNA-MRD status and treatment response. Interestingly, compared to long-course radiotherapy, microsatellite instability is more pronounced after shortRT ( P=0.042), and ctDNA negativity is significantly associated with pathological complete response (pCR) ( P=0.022). Furthermore, both ctDNA clearance ( P=0.049) and MRD clearance ( P=0.015) after shortRT are significantly correlated with pCR. A risk scoring predictive model based on ctDNA-MRD was established, with achieving the highest C-index at the C2 time point. This model, combining MRD clearance and CEA, outperforms models using only MRD clearance (AUC=0.917, 95% CI=0.753 to 1.000) or only CEA (AUC=0.733, 95% CI=0.449 to 1.000), demonstrating superior performance in predicting pCR/non-pCR (AUC=0.983, 95% CI=0.937 to 1.000). Conclusions: These findings offers valuable insights into the dynamic landscape of NAT for LARC management and emphasize the potential of ctDNA-based MRD assessment as a valuable tool for tailoring treatment strategies. The differences observed between shortCRT and longCRT regimens underscore the need for personalized treatment approaches. Overall, our study contributes valuable insights into optimizing treatment decision-making and predicting treatment response in LARC patients, ultimately advancing the field of rectal cancer management.