康复的能量来源:预测HPV阴性可手术头颈鳞癌患者的胃造瘘管依赖
本研究分析265例HPV阴性头颈鳞癌患者,采用CT测量的颈部椎旁骨骼肌质量(C3)评估肌少症,识别胃造瘘管支持超过28天的患者并建立风险模型,与PEG评分比较。逻辑回归模型AUC最高(0.78),递归分割分析(RPA)模型为0.76,均优于PEG评分(0.67);RPA模型在50例验证队列中保持较高准确度(AUC=0.79)。结论认为RPA模型优于PEG评分,为术前规划和早期干预提供更精确的风险工具。
要点
- 男性、肌少症、既往放疗和辅助治疗是胃造瘘管依赖的关键预测因素。
- 逻辑回归模型AUC为0.78,RPA模型为0.76,均优于PEG评分(0.67)。
- RPA模型在50例验证队列中AUC为0.79,保持较高准确度。
核心结论
- RPA模型优于PEG评分,为术前规划和早期干预提供更精确的风险工具。
研究设计与样本
- 本研究分析265例HPV阴性头颈鳞癌患者,通过CT扫描颈部椎旁骨骼肌质量(C3)评估肌少症,识别需要胃造瘘管支持超过28天的患者,建立风险模型并与PEG评分比较,另用50例验证队列评估模型性能。
主要发现
- 关键预测因素包括男性、肌少症、既往放疗和辅助治疗。
- 逻辑回归模型AUC最高(0.78),其次为递归分割分析(RPA)模型(0.76),两者均优于PEG评分(0.67)。
- RPA模型在验证队列中保持较高准确度(AUC=0.79)。
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: HPV-negative head and neck squamous cell carcinoma (HNSCC) patients can face prolonged gastrostomy tube (g-tube) dependence, affecting quality of life and delaying treatment. Existing models lack key health metrics like sarcopenia; this study develops a risk model to enhance preoperative assessment. METHODS: This study analyzed 265 HNSCC patients. Sarcopenia was assessed via cervical paraspinal skeletal muscle mass (C3) on CT scans. Patients requiring g-tube support > 28 days were identified. Risk models were developed and compared to the PEG Score. A 50-patient validation cohort assessed model performance. RESULTS: Key predictors included male sex, sarcopenia, prior radiotherapy, and adjuvant therapy. Logistic regression (LR) model had the highest AUC (0.78); followed by Recursive Partitioning analysis (RPA) model (0.76), both outperforming the PEG Score (0.67). RPA model maintained superior accuracy in validation (AUC = 0.79). CONCLUSION: RPA model outperformed PEG Score, offering a more precise risk tool for preoperative planning and early intervention.
引用
- McNicoll MM, Mlynarek AM, Richardson K, et al. Fuel for Recovery: Predicting G-Tube Dependence in HPV-Negative Operable HNSCC. Head Neck. 2025. PMID: 40799173. doi:10.1002/hed.70015
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。