Predictive model based on folate receptor-positive circulating tumor cells in neoadjuvant immunochemotherapy for esophageal cancer.

Background: Effective biomarkers for predicting outcomes in patients with esophageal squamous cell carcinoma (ESCC) receiving neoadjuvant immunochemotherapy (nICT) remain limited. In this study, we investigated the prognostic value of baseline folate receptor-positive circulating tumor cells (FR-CTCs) in locally advanced ESCC. Methods: This cohort study retrospectively analyzed ESCC patients who underwent nICT between August 2020 and August 2023. Factors associated with disease-free survival (DFS) were evaluated using univariable and multivariable Cox proportional hazards regression. Candidate variables were further screened by univariable Cox and least absolute shrinkage and selection operator regression to construct a predictive model. A nomogram incorporating FR-CTCs levels and key clinical parameters was developed and comprehensively assessed for discrimination, calibration, and clinical utility. Results: Analysis of clinical data from 64 patients identified baseline FR-CTCs level emerged as a strong and independent predictor of postoperative recurrence. FR-CTCs were significantly associated with DFS when analyzed as either a continuous or categorical variable. As a continuous variable, each one-unit increase in FR-CTCs level was associated with an increased risk of recurrence both before [hazard ratio (HR) =1.26, 95% confidence interval (CI): 1.12-1.42; P<0.001] and after multivariable adjustment (HR =1.81, 95% CI: 1.34-2.45; P<0.001). When dichotomized, patients with high FR-CTCs levels had a markedly higher recurrence risk than those with low levels in unadjusted (HR =5.47, 95% CI: 1.69-19.45; P=0.005) and adjusted analyses (HR =14.35, 95% CI: 2.28-90.17; P=0.005). The final nomogram achieved a concordance index of 0.821. Time-dependent receiver operating characteristic analysis demonstrated excellent discrimination at 1-, 2-, and 3-year [areas under the curve (AUC): 0.933, 0.904, and 0.780], with good calibration. At the optimal cut-off value determined by the Youden's index of the model, the sensitivity, specificity, positive predictive value, and negative predictive value were 1.00, 0.85, 0.31, and 1.00 at 1-year, and 0.89, 0.76, 0.70, and 0.92 at 2-year. Brier scores at 1-, 2-, and 3-year were 0.05, 0.14, and 0.17, respectively. Conclusions: In summary, baseline FR-CTCs represent a powerful biomarker for postoperative recurrence risk stratification in nICT-treated ESCC patients. The proposed nomogram exhibits robust predictive accuracy and provides a strong foundation for future prospective validation and potential clinical application.