Predicting Severe Radiation Esophagitis in Patients With Locally Advanced Esophageal Squamous Cell Carcinoma Receiving Definitive Chemoradiotherapy: Construction and Validation of a Model Based in the Clinical and Dosimetric Parameters as Well as Inflammatory Indexes

OBJECTIVE: Radiation esophagitis (RE) is common in patients treated with radiotherapy (RT) for locally advanced esophageal squamous cell carcinoma (ESCC). We aim to construct a nomogram predicting the severe RE (grade ≥2) in patients with ESCC receiving definitive chemoradiotherapy (dCRT). MATERIALS AND METHODS: Logistic regression was performed to evaluate the risk factors in predicting RE. Nomogram was built based on the multivariate analysis result. The model was validated using the area under the receiver operating curve (ROC) curve (AUC), calibration curves, and decision curve analyses (DCA). Spearman correlation analysis was used to evaluate the correlation between inflammation indexes. RESULTS: A total of 547 patients with stage II–IVA ESCC treated with dCRT from the retrospective study were included. Two hundred and thirty-two of 547 patients (42.4%) developed grade ≥2 RE. Univariate analysis indicated that gender (p = 0.090), RT dose (p < 0.001), targeted therapy (p = 0.047), tumor thickness (p = 0.013), lymphocyte-monocyte ratio (LMR, p = 0.016), neutrophil-lymphocyte ratio (NLR, p < 0.001), and platelet-lymphocyte ratio (PLR, p < 0.001) were the significant factors for a higher incidence of RE. In multivariate analysis, RT dose [p < 0.001; odds ratio (OR), 4.680; 95% confidence interval (CI), 2.841–6.709], NLR (p < 0.001; OR, 0.384; 95% CI, 0.239–0.619), and PLR (p < 0.001; OR, 3.539; 95% CI: 2.226–5.626) were independently associated grade ≥2 RE and were involved in the nomogram. ROC curves showed the AUC of the nomogram was 0.714 (95% CI, 0.670–0.757), which was greater than each factor alone (RT dose: 0.615; NLR: 0.596; PLR: 0.590). Calibration curves showed good consistency between the actual observation and the predicted RE. DCA showed satisfactory positive net benefits of the nomogram among most threshold probabilities. CONCLUSIONS: The study demonstrated that RT dose, NLR, and PLR were independent risk factors for grade ≥2 RE in patients with locally advanced ESCC receiving dCRT. A predictive model including all these factors was built and performed better than it based on each separately. Further validation in large patient populations is still warranted.