文献速递头颈部肿瘤放疗

基于腮腺干细胞区剂量学参数的鼻咽癌放疗后口干恢复预测模型

A predictive model based on the dosimetric parameters of the parotid stem cell region to assess the recovery of radiation-induced xerostomia in long-term survivors of nasopharyngeal carcinoma after radical radiotherapy.
PubMed 文献 · Radiother Oncol · 2025-09-26

多中心分析648例接受根治性放疗的鼻咽癌患者,利用Cox模型筛选出口干恢复的关键因素包括年龄、诱导化疗、腮腺干细胞区Dmean和V20,并构建列线图预测模型。该模型对患者报告和医生评估的口干恢复均具有良好预测效能,可将患者分为高低风险组,低风险组恢复率显著更高。

要点

  • 年龄、诱导化疗、腮腺干细胞区Dmean和V20是鼻咽癌放疗后口干恢复的关键预测因素。
  • 基于上述因素构建的列线图模型预测口干恢复的3年/5年AUC在0.75~0.81之间。
  • 按模型评分中位数分组,低风险组口干恢复率显著高于高风险组(p<0.001)。

核心结论

  • 结合腮腺干细胞区剂量学参数与临床因素的模型能有效预测鼻咽癌根治性放疗后长期生存者的口干恢复。

研究设计与样本

  • 多中心研究纳入中国四家中心648例接受根治性放疗的鼻咽癌患者,按7:3随机分为训练集和验证集。
  • 采用Cox比例风险模型筛选口干恢复的关键临床与剂量学因素,并构建整合预测模型。
  • 通过时间依赖ROC曲线下面积(AUC)评估预后性能,用累积发生率曲线估计不同风险组的恢复情况。

主要发现

  • 多因素Cox分析确定年龄(p=0.000)、诱导化疗(p=0.003)、腮腺干细胞区Dmean(p=0.047)和V20(p=0.000)为口干恢复的关键预测因素。
  • 患者报告口干恢复的3年/5年AUC在训练集为0.77/0.806,验证集为0.773/0.81。
  • 医生评估口干恢复的3年/5年AUC在训练集为0.756/0.786,验证集为0.748/0.786。
  • 按模型评分中位数分层,低风险组在训练集和验证集中均显示显著更高的恢复率(p<0.001)。

关键数值与推荐

  • 腮腺干细胞区Dmean的p=0.047,V20的p=0.000,均与口干恢复显著相关。
  • 列线图模型预测患者报告口干恢复的3年/5年AUC:训练集0.77/0.806,验证集0.773/0.81。
  • 列线图模型预测医生评估口干恢复的3年/5年AUC:训练集0.756/0.786,验证集0.748/0.786。
  • 基于模型评分中位数将全队列分为高风险组和低风险组,低风险组恢复率显著更高(p<0.001)。

临床怎么用

  • 可利用年龄、诱导化疗及腮腺干细胞区Dmean、V20等参数构建列线图,预测鼻咽癌放疗后口干恢复概率。
  • 根据模型评分将患者分为高低风险组,对高风险患者可考虑优化放疗计划或加强口干管理。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: For nasopharyngeal carcinoma (NPC) patients with bilateral parotids near high - dose areas, more research is needed on how radiation to the parotid stem cell region (SCR) impacts xerostomia. This multicenter study aims to explore SCR radiation dosimetric parameters and clinical factors that influence recovery from radiation-induced xerostomia in long-term survivors. METHODS: We analyzed 648 NPC patients from four Chinese centers who underwent radical radiotherapy. Recovery was assessed via patient-reported (PRO) and physician-assessed (PAO) outcomes. The Cox proportional hazards model identified key clinical and dosimetric factors for recovery, aiming to develop an integrated prediction model. Prognostic performance was evaluated with time-dependent area under the receiver operating characteristic curve (AUC), and Cumulative incidence curve estimated recovery across different risk groups. RESULTS: Patients were randomly allocated to training and validation sets at 7:3. Multivariate COX analysis identified age (p=0.000), induction chemotherapy (p=0.003), SCR Dmean (p=0.047), and SCR V20 (p=0.000) as key predictors for xerostomia recovery. A nomogram model using these factors outperformed individual ones in predicting xerostomia recovery for PRO recovery, 3-year/5-year AUC was 0.77/0.806 (training) and 0.773/0.81 (validation); for PAO recovery, it was 0.756/0.786 (training) and 0.748/0.786 (validation). Based on the median score derived from the model, the entire cohort was stratified into high-risk and low-risk groups. Notably, the low-risk group demonstrated a significantly higher recovery rate in both the training set and validation set (p < 0.001). CONCLUSION: A model combining SCR dosimetry and clinical factors effectively predicts xerostomia recovery.

来源:Radiother Oncol,查看原文。版权归原作者所有。

引用

  1. Zhang M, Gong J, Zhu Y, et al. A predictive model based on the dosimetric parameters of the parotid stem cell region to assess the recovery of radiation-induced xerostomia in long-term survivors of nasopharyngeal carcinoma after radical radiotherapy. Radiother Oncol. 2025. PMID: 41016666. doi:10.1016/j.radonc.2025.111163

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。