Neoadjuvant Chemoradiotherapy Versus Neoadjuvant Immunochemotherapy for Resectable Esophageal Cancer: A Target Trial Emulation Study

Detailed Description This study is a target trial emulation (TTE) designed to compare the causal effects of neoadjuvant immunochemotherapy (nICT) versus neoadjuvant chemoradiotherapy (nCRT) in patients with resectable locally advanced esophageal squamous cell carcinoma (ESCC). The TTE framework, as proposed by Hernán and Robins, is used to emulate a hypothetical randomized controlled trial using observational data.

Data Source: Electronic health records (EHR) from Henan Cancer Hospital, including medical records, pathology reports, imaging reports, prescription systems, radiotherapy records, and follow-up systems, covering the period from January 2013 to December 2025.

Target Trial Specification:

• Eligibility: Age ≥18 years; histologically confirmed ESCC; clinical stage cT3-4a N0-2 M0 or cT2N+ M0 (AJCC 8th edition); ECOG PS 0-1; no prior antitumor therapy for esophageal cancer; curative surgical intent at Time Zero.
• Treatment Strategies: Strategy A (nCRT): platinum-based chemotherapy with concurrent radiotherapy (41.4-50.4 Gy). Strategy B (nICT): PD-1 inhibitor combined with platinum-based chemotherapy (2-4 cycles).
• Assignment: In the target trial, patients are randomized 1:1. In the emulation, patients are assigned based on actual treatment received, with propensity score overlap weighting to simulate randomization.
• Follow-up: Time Zero is defined as the day before the first neoadjuvant treatment order. Follow-up continues until death, loss to follow-up, or administrative end of study (December 31, 2026). A grace period of up to 180 days is allowed for treatment initiation (i.e., the order date must be within 180 days from Time Zero). Patients who initiate treatment beyond this window are excluded from the primary analysis or evaluated separately in sensitivity analyses.
• Outcomes: Primary: Overall Survival (OS). Secondary: Event-Free Survival (EFS), pathological complete response (pCR) rate, tumor regression grade (TRG), R0 resection rate, and safety (≥grade 3 adverse events per CTCAE v5.0, postoperative complications per Clavien-Dindo classification).

Statistical Analysis:

• Primary Analysis: Overlap weighting using propensity scores to balance baseline covariates; weighted Cox proportional hazards models to estimate hazard ratios (HR) with 95% confidence intervals for OS and EFS.
• Secondary Analysis: pCR, TRG, and R0 rates compared between groups in the surgical subgroup.
• Sensitivity Analyses: Per-protocol analysis, clone-censor-weight method, E-value for unmeasured confounding, different Time Zero definitions, different grace periods (4 weeks, 12 weeks), complete-case analysis vs multiple imputation.
• Sensitivity Analyses: Per-protocol analysis, clone-censor-weight method, E-value for unmeasured confounding, different Time Zero definitions, different grace periods (4 weeks, 12 weeks), complete-case analysis vs multiple imputation.
• Subgroup Analyses: By age, sex, tumor location, clinical stage, ECOG PS, and diagnosis period (2013-2019 vs 2020-2025).

Sample Size: Based on historical data, the target sample size is approximately 150-200 patients per group, with a total enrollment of approximately 400 patients (final sample size determined after data extraction and pilot feasibility assessment).

Ethics: Approved by the Medical Ethics Committee of Henan Cancer Hospital (Approval No. 2026-366-001). A waiver of informed consent was granted for this retrospective study.