一种新型列线图预测头颈癌幸存者放射性脑坏死后的总生存
研究基于放射并发症登记研究纳入495例放射性脑坏死(RN)头颈癌幸存者,按7:3随机分为训练队列与内部验证队列,用LASSO回归筛选预测因子并构建多变量Cox列线图,另用88例RN患者外部验证。列线图内部验证C指数0.761(95%CI 0.676至0.847),外部验证C指数0.795(95%CI 0.717至0.874)。
要点
- 训练数据识别出5个RN后生存预测因子:年龄、RN前肿瘤进展、低位颅神经损伤、双侧坏死、卒中史。
- 列线图内部验证C指数0.761(95%CI 0.676至0.847),外部验证C指数0.795(95%CI 0.717至0.874)。
- 决策曲线分析显示RN诊断后1年死亡概率1%至48%、3年3%至50%、5年超过2%时列线图有临床实用性。
核心结论
- 该基于LASSO-Cox模型的列线图研究开发并验证了一种易于应用的模型,用于预测头颈癌幸存者RN诊断后的总生存。
研究设计与样本
- 研究纳入来自放射并发症登记研究的495例RN头颈癌幸存者,按7:3随机分配至训练队列和内部验证队列。
- 训练队列采用LASSO回归筛选RN后生存的显著预测因子,并用多变量Cox模型开发列线图。
- 列线图用内部验证队列评估性能,并用额外88例RN患者进行外部验证。
主要发现
- 训练数据识别出5个RN后生存预测因子:年龄、RN前肿瘤进展、低位颅神经损伤、双侧坏死和卒中史。
- 列线图内部验证队列C指数0.761(95%CI 0.676至0.847),外部验证队列C指数0.795(95%CI 0.717至0.874)。
- 决策曲线分析显示列线图在RN诊断后1年死亡概率1%至48%、3年3%至50%、5年超过2%时具有临床实用性。
关键数值与推荐
- 内部验证队列C指数0.761(95%CI 0.676至0.847)。
- 外部验证队列C指数0.795(95%CI 0.717至0.874)。
- 决策曲线分析临床有用概率范围:RN诊断后1年死亡概率1%至48%,3年3%至50%,5年超过2%。
临床怎么用
- 原文未据此给出推荐。
局限性
- 原文未报告局限性。
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 AND PURPOSE: The study aimed to develop and validate a novel nomogram to predict overall survival in head and neck cancer survivors following the diagnosis of radiation-induced brain necrosis (RN). MATERIALS AND METHODS: We included head and neck cancer survivors with RN from a radiation complications registry study. A total of 495 eligible patients were 7:3 randomly allocated to a training cohort and an internal validation cohort. The Least Absolute Shrinkage and Selection Operator (LASSO) regression was applied to select significant predictors of post-RN survival in the training cohort, and a multivariable Cox model was used to develop the nomogram. The performance of the nomogram was assessed using the internal validation cohort and externally validated using additional 88 RN patients. RESULTS: We identified five predictors of post-RN survival using the training data: age, tumor progression before RN, lower cranial nerves injury, bilateral necrosis, and history of stroke. The nomogram showed favorable performance in the internal validation cohort (C-index 0.761, 95% CI 0.676 to 0.847) and in the external validation cohort (C-index 0.795, 95% CI 0.717 to 0.874). The decision curve analysis indicated that the nomogram was clinically useful when the probabilities of death ranging from 1% to 48% at 1 year, from 3% to 50% at 3 years, and exceeding 2% at 5 years after being diagnosed with RN. CONCLUSION: In this LASSO-Cox model-based nomogram study, we developed and validated an easily applied model to predict overall survival in head and neck cancer survivors following an RN diagnosis.
引用
- Pan D, Rong X, Zhu Y, et al. A novel nomogram to predict overall survival in head and neck cancer survivors with radiation-induced brain necrosis. Radiother Oncol. 2022. PMID: 35121029. doi:10.1016/j.radonc.2022.01.033
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。