Postoperative radiotherapy for limited-stage high-grade neuroendocrine cervical carcinoma: a retrospective multi-center study from two Chinese institutions

Current treatment guidelines rely on limited retrospective data due to the rarity of high-grade neuroendocrine cervical carcinoma, which lacks consensus on the efficacy of postoperative radiotherapy. To facilitate decision-making support for individual patients, a recursive partitioning analysis model was established to reveal prognostic factors and further explore the role of postoperative radiotherapy for patients with neuroendocrine cervical carcinoma. This was a retrospective multi-centre cohort study conducted across 2 institutions (2005–2022). Variables were weighted by HR from univariate and multivariable Cox regression. Recursive partitioning analysis was applied to 199 patients with preselected clinicopathological variables (age, FIGO stage, tumor size, lympho-vascular invasion, deep stromal invasion, and perineural invasion). Postoperative radiotherapy efficacy was evaluated via stratified Kaplan‒Meier analysis and Cox proportional hazards models. Postoperative radiotherapy did not significantly improve survival in the overall cohort (all P > 0.05). Three distinct risk groups emerged after recursive partitioning analysis: Low-risk (n = 96): no lympho-vascular invasion or deep stromal invasion; intermediate-risk (n = 70): lympho-vascular invasion, deep stromal invasion without perineural invasion; High-risk (n = 33): lympho-vascular invasion, deep stromal invasion, perineural invasion (low-risk group: HR = 1.00; intermediate-risk group: HR, 3.60; 95% CI, 1.95–6.67; p-value < 0.001; high-risk group: HR = 6.22; 95% CI, 3.23–11.95; p-value < 0.001). Postoperative radiotherapy significantly improved survival in high-risk patients (HR = 0.22, 95% CI 0.08–0.58; p-value < 0.001), with 5-year overall survival improving from 0% (non- radiotherapy) to 41.9% (Radiotherapy). No mortality reduction was observed in the intermediate- (p-value = 0.890) or low-risk groups (p-value = 0.980). Our recursive partitioning analysis-driven risk stratification model provides a visually interpretable framework for neuroendocrine cervical carcinoma management. Three risk factors (lympho-vascular invasion, deep stromal invasion, and perineural invasion) identify a high-risk subgroup where postoperative radiotherapy confers substantial survival gains. These findings advocate for risk-adapted postoperative radiotherapy protocols in neuroendocrine cervical carcinoma, pending prospective validation. What is already known on this topic: While postoperative pathological features are established indications for radiotherapy in cervical cancer, the indications for postoperative radiotherapy in high-grade neuroendocrine cervical carcinoma remain poorly defined. What this study adds: Lympho-vascular invasion, deep stromal invasion, perineural invasion were the independent risk factors​ for risk stratification. Patients (high-risk subgroup) showed a 41.9% 5-year overall survival with radiotherapy vs. 0% without​ (HR = 0.22), while intermediate-/low-risk groups derived no benefit. How this study might affect research, practice or policy: The recursive partitioning study offered personalized radiotherapy decisions by sparing lower-risk patients overtreatment while prioritizing adjuvant radiotherapy in high-risk neuroendocrine cervical carcinoma with all three factors.