Genomic alterations and dynamic molecular residual disease monitoring predict pathological response to neoadjuvant chemoimmunotherapy in esophageal squamous cell carcinoma

Background Approximately 40-60% of patients with locally advanced esophageal squamous cell carcinoma (ESCC) exhibit suboptimal responses to neoadjuvant chemoimmunotherapy (NCIT), highlighting the need for predictive biomarkers of pathological response. Methods We prospectively enrolled 29 stage II-III ESCC patients receiving NCIT (albumin-paclitaxel/carboplatin + anti-PD-1). Baseline tumor tissues were firstly analyzed via 437-gene targeted sequencing. Serial preoperative plasma samples collected before NCIT, during NCIT and post-NCIT/pre-surgery, along with baseline tissue, were profiled using a 2365-gene panel for tumor-informed molecular residual disease (MRD) monitoring. Associations between clinicopathological features, genomic alterations, MRD status, and major pathological response (MPR) were evaluated. Results No clinicopathological feature significantly correlated with MPR. MPR was significantly associated with baseline NOTCH1 mutations ( p = 0.014) and higher chromosomal instability score (CIS, p = 0.032). Absence of MRD and lower ctDNA levels after NCIT completion, but not before and during, strongly correlated with MPR. Distinct change patterns of ctDNA level from on- to post-NCIT were observed in MPR versus non-MPR patients. A model integrating baseline NOTCH1 mutation and post-NCIT MRD status achieved superior MPR prediction (AUC = 0.954), outperforming either factor alone ( NOTCH1 , AUC = 0.769; post-NCIT MRD, AUC = 0.882). Conclusions Dynamic MRD monitoring, particularly post-treatment, provides strong predictive value for pathological response to NCIT in locally advanced ESCC. Integrating baseline NOTCH1 mutation status with post-NCIT MRD assessment significantly improves MPR prediction, indicating the potential to inform clinical decision-making for these patients.