P04.6. pCR and ctDNA in SCIENCE: A Phase III Trial of Neoadjuvant Immunotherapy in Resectable LA-ESCC

Abstract

Esophageal Cancer: Adjuvant and Neo-Adjuvant Therapies The optimal neoadjuvant strategy for resectable locally advanced esophageal squamous cell carcinoma (LA-ESCC) remains controversial. Although chemotherapy combined with PD-1 blockade has demonstrated promising activity, it is unclear whether adding radiotherapy further enhances pathological response and long-term outcomes without compromising surgical feasibility. The phase III SCIENCE trial was designed to compare neoadjuvant chemotherapy plus sintilimab, chemoradiotherapy plus sintilimab, and chemoradiotherapy alone in resectable LA-ESCC. The study evaluates pathological complete response and event-free survival as dual primary endpoints and explores circulating tumor DNA (ctDNA) as an early biomarker of treatment response. SCIENCE is a multicentre, randomized, phase III trial conducted at seven centers in China. Treatment-naive patients with histologically confirmed thoracic ESCC (cT1N2–3M0 or cT2–4aN0–3M0; AJCC/UICC 8th edition) were randomized (1:1:1) to neoadjuvant chemotherapy plus sintilimab (A), chemoradiotherapy plus sintilimab (B), or chemoradiotherapy alone (C). Surgery was scheduled 6–8 weeks after neoadjuvant therapy. Dual primary endpoints were pathological complete response (pCR) and event-free survival (EFS). Longitudinal ctDNA was assessed using a personalized tumor-informed assay at baseline, during therapy, and preoperatively. pCR will be compared using the Chi-square test and logistic regression, and EFS using log-rank and Cox models. The overall two-sided type I error is controlled at 0.05 (0.02 for pCR; 0.03 for EFS). Data cutoff is January 19, 2026. Between November 2022 and August 2025, 307 patients were randomized to Group A (n=100), Group B (n=103), or Group C (n=104). Most were male (88.3%) with stage III disease (81.4%); all achieved R0 resection. pCR rates were 18.0% (A), 57.3% (B), and 49.0% (C). Compared with A, pCR was significantly higher in B (OR 6.1, 95% CI 3.3–11.9; p<0.0001) and C (OR 4.4, 95% CI 2.3–8.5; p<0.0001). Among 92 evaluable patients for ctDNA analysis, detectability declined from 96.7% at baseline to 48.9% preoperatively. Preoperative ctDNA positivity was higher in A (78.6%) versus B (40.0%) and C (32.4%) (p<0.001). ctDNA clearance was strongly associated with pCR (62.2% vs 9.1%, p<0.001) and increased stepwise across response categories. In resectable locally advanced ESCC, the addition of immunotherapy to chemoradiotherapy significantly improved pathological complete response compared with chemotherapy plus immunotherapy, without compromising surgical feasibility. Chemoradiotherapy-based regimens achieved superior pathological responses and deeper ctDNA clearance, supporting intensified multimodality approaches. Preoperative ctDNA dynamics strongly correlated with pathological response, highlighting its potential as an early biomarker of treatment efficacy. These findings provide prospective phase III evidence to inform optimal neoadjuvant strategies and support ctDNA-guided response assessment in LA-ESCC.