A nomogram for predicting the risk of chemoradiotherapy-associated thrombocytopenia in patients with esophageal cancer: a real-world cohort study

Background: Thrombocytopenia is a common hematological toxicity in esophageal cancer (EC) patients receiving radiation therapy. Objectives: The purpose of this study was to construct and validate a nomogram for predicting the incidence of thrombocytopenia in EC patients receiving radiotherapy. Design: A retrospective study. Methods: All data were collected from 435 EC patients who received radiation therapy from a real-world cohort study from 2016 to 2021. Thrombocytopenia was assessed according to the toxicity criteria of the Radiation Therapy Oncology Group (RTOG). Univariate and multivariate logistic regression analyses were used to identify significant risk factors for thrombocytopenia. The result was presented in the form of a nomogram. The performance of the model was evaluated by the receiver operating characteristic (ROC) curves and calibration curves. External validation was conducted using an independent cohort of 149 EC patients from another cancer center. Propensity score matching was utilized to balance the covariates between the 1:1 matched two groups. Survival analysis was conducted using the Kaplan–Meier method and compared with the log-rank test. Results: A total of 104(23.91%) patients developed thrombocytopenia. Univariate and multivariate logistic regression analysis showed that age, body mass index, planning target volume, tumor location, tumor-node-metastasis stage, platelet count before radiotherapy, history of liver cirrhosis, cycles of induction chemotherapy, and concurrent chemotherapy regimen were the independent predictors of chemoradiotherapy-associated thrombocytopenia and were finally incorporated into our nomogram. The results of the ROC curve showed that the nomogram had high prediction accuracy. The sensitivity of the external validation set was 0.852 (95% CI: 0.663–0.958), and the specificity was 0.820 (95% CI: 0.740–0.883). The calibration curve presented good concordance (Hosmer-Lemeshow test, p = 0.825). An online nomogram-based prediction tool was developed to facilitate clinical implementation. No significant difference was observed in overall survival between patients who experienced severe thrombocytopenia and not (log-rank test, p = 0.390). Conclusion: This nomogram provides a prediction tool of chemoradiotherapy-associated thrombocytopenia in EC patients receiving radiotherapy, which may facilitate the potential candidate’s selection of high-risk EC patients for thrombocytopenia.