乳腺癌术后唑来膦酸/地诺单抗相关颌骨坏死风险预测模型的构建与验证OA
Construction and Verification of Risk Prediction Model for Zoledronic Acid/Denosumab-Related Osteonecrosis of the Jaw after Breast Cancer Surgery
目的:筛选乳腺癌术后使用抗骨吸收药物(唑来膦酸/地诺单抗)发生相关性颌骨坏死的高危因素,并尝试构建与验证风险预测模型.方法:回顾性收集 2020 年 6 月至 2025 年 6 月在成都市第六人民医院口腔科、四川大学华西口腔医院就诊,既往行乳腺癌根治术后接受唑来膦酸或地诺单抗进行辅助性抗骨转移治疗的135 例患者的临床资料,7:3 随机分为训练集与验证集,采用 Lasso 回归筛选发生药物相关性颌骨坏死的预测指标,使用一致性指数、AUC、校准曲线、风险得分图及再分组验证评估预测性能,最终绘制列线图.结果:Lasso 回归模型得出口腔侵入性操作、糖皮质激素使用、绝经史、药物使用年限及联合放疗对使用抗骨吸收药物(唑来膦酸/地诺单抗)发生相关性颌骨坏死具有重要的预测价值.在训练集及验证集中构建了 3 年、5 年的时点 ROC 曲线.在训练集中 3 年、5 年的AUC 值分别为 0.81、0.79;在验证集中 3 年、5 年的 AUC 值分别为 0.78、0.79.预测指标在训练集与验证集中预测 3年及 5 年颌骨坏死的发生概率与实际概率均表现出良好的一致性.风险得分图及分组再验证分析得出,5 项预测指标具有良好的预测效能.结论:本研究尝试构建的预测列线图能有效预测乳腺癌术后使用抗骨吸收药物(唑来膦酸/地诺单抗)相关性颌骨坏死的发生风险.
Objective:To identify high-risk factors for medication-related osteonecrosis of the jaw(MRONJ)in postoper-ative breast cancer patients treated with anti-resorptive agents(zoledronic acid/denosumab),and to construct and validate a risk prediction model.Methods:We retrospectively enrolled 135 patients who visited the Department of Stomatology at Chengdu Sixth People's Hospital and West China Hospital of Stomatology,Sichuan University between June 2020 and June 2025.All patients had previously undergone radical mastecto-my for breast cancer and received adjuvant anti-bone metastasis therapy with zoledronic acid or denosumab.They were ran-domly assigned to training and validation sets at a ratio of 7:3.Lasso regression was employed to screen predictors of MRONJ.Predictive performance was assessed using the C-index,area under the curve(AUC),calibration curves,risk score plots,and stratified validation.Finally,a nomogram was constructed.Results:The Lasso regression model identified invasive oral procedures,glucocorticoid use,history of menopause,duration of medication use and combined radiotherapy as significant predictors of MRONJ in patients receiving anti-resorptive agents(zoledronic acid/denosumab).Time-dependent ROC curves at 3 and 5 years were plotted in both the training and validation sets.The AUC values were 0.81(3-year)and 0.79(5-year)in the training set,and 0.78(3-year)and 0.79(5-year)in the validation set.The predicted probabilities of MRONJ at 3 and 5 years showed good agreement with the actual probabilities in both the training and validation sets.Risk score plots and stratified re-validation demonstrated that the five predictors had good predictive performance.Conclusion:The predictive nomogram developed in this study can effectively predict the risk of MRONJ in postoperative breast cancer pa-tients receiving anti-resorptive agents(zoledronic acid/denosumab).
袁蒙姜;吴赵多;吴云龙;廖培希
610061 成都,成都市第六人民医院 口腔科300070 天津,天津医科大学临床医学院 口腔系610041 成都,四川大学华西口腔医院 颌面外科610061 成都,成都市第六人民医院 口腔科
医药卫生
乳腺癌唑来膦酸地诺单抗相关颌骨坏死预测模型
Breast cancerZoledronic acidDenosumabMedication-related osteonecrosis of the jawPrediction model
《肿瘤预防与治疗》 2026 (7)
535-546,12
This study was supported by grants from Chengdu Science and Technology Bureau(No.2024YF0500910SN).成都市科技计划项目(编号:2024YF0500910SN)
评论