Preoperative Circulating Tumor DNA Detection and Risk Stratification in Esophageal Squamous Cell Carcinoma.

Importance: Neoadjuvant chemoradiotherapy is recommended for clinical stage II (T2N0) esophageal squamous cell carcinoma (ESCC) with high-risk features such as tumor size 3 cm or greater, lymphovascular invasion (LVI), and poor differentiation. However, these criteria have limited predictive value, and there is an unmet need for more accurate preoperative risk stratification tools. Objective: To evaluate whether preoperative detection of circulating tumor DNA (ctDNA) is associated with nodal upstaging and postoperative recurrence in patients with clinical stage I (T1b) or T2N0 ESCC. Design, Setting, and Participants: This cohort study included 2 independent cohorts in Seoul, Korea: Samsung Medical Center (SMC; n = 50), with data from January 2015 through December 2019, and Yonsei University Severance Hospital (YUSH; n = 24), with data from January 2023 through December 2024. All patients had T1b or T2N0 ESCC, underwent radical esophagectomy and lymph node dissection without neoadjuvant therapy, and provided tumor and preoperative plasma samples for tumor-informed ctDNA sequencing. Exposure: Presence or absence of ctDNA in preoperative plasma samples. Main Outcomes and Measures: Primary outcomes were pathologic nodal upstaging and survival (recurrence-free survival [RFS] and overall survival [OS]). Associations between ctDNA status and outcomes were analyzed using χ2, logistic regression, and Cox proportional hazard models. Results: In the SMC cohort, the median (IQR) age was 68 (60-74) years, and 47 participants (94%) were male; in the YUSH cohort, the median (IQR) age was 67 (61-69) years, and 21 participants (87.5%) were male. Preoperative ctDNA was detected in 36 patients (48.6%) (27 [54.0%] in SMC and 9 [37.5%] in YUSH). Detection was more frequent in T2N0 than T1b: 26 [57.8%] vs 1 [20.0%] in SMC and 9 [42.9%] vs 0 in YUSH. Over a median (IQR) follow-up of 37.7 (24.9-49.6) months, patients with positive ctDNA results had worse RFS (hazard ratio [HR], 4.15; 95% CI, 1.54-11.22; P = .005) and OS (HR, 4.02; 95% CI, 1.50-10.74; P = .006). Among patients with T2N0 disease-a subgroup where neoadjuvant therapy decisions remain uncertain-ctDNA positivity predicted occult nodal metastasis with a positive predictive value of 100% (95% CI, 71.51-100) in the SMC cohort and 88.89% (95% CI, 51.75-99.72) in the YUSH cohort. Preoperative ctDNA remained significantly associated with nodal metastasis (odds ratio [OR], 19.98; 3.90-211.42; P < .001), outperforming guideline-based criteria (LVI, poor differentiation, tumor size ≥3 cm). Adding ctDNA to models with guideline risk factors improved the area under the receiver operating characteristic curve from 0.66 to 0.91 (P = .048) in the SMC cohort, and showed consistent performance in the YUSH cohort (from 0.67 to 0.89; P = .03). Conclusions and Relevance: Preoperative ctDNA detection was significantly associated with occult nodal metastasis and recurrence risk in clinical N0 early-stage ESCC in this study. Among patients with T2N0 ESCC, ctDNA may serve as a complementary biomarker to inform neoadjuvant treatment escalation and support more personalized therapeutic strategies.