晚期葡萄膜黑色素瘤患者在替本他福普治疗中出现影像学进展时与临床获益相关的影像组学标志物
本研究基于2期试验(NCT02570308),将127例二线及以上晚期葡萄膜黑色素瘤患者中34例影像学进展(PD)者按ctDNA是否下降≥0.5 log分组,利用基线和第8周CT/MRI影像组学特征构建机器学习标志物,患者水平ROC AUC为0.71,病灶水平为0.70,最具影响力的预测因子为肿瘤异质性下降。
要点
- 在PD患者中,基于体积的影像组学特征区分ctDNA下降组的患者水平ROC AUC为0.71。
- 病灶水平影像组学特征区分ctDNA下降组的ROC AUC为0.70。
- 肿瘤异质性下降是预测ctDNA下降最具影响力的影像组学预测因子。
核心结论
- 影像组学分析在患者水平和病灶水平均能预测PD患者的ctDNA下降。
- 最具影响力的预测因子是CT/MRI上观察到的肿瘤异质性下降。
研究设计与样本
- 研究为2期试验(NCT02570308)中的影像组学分析,纳入127例二线及以上晚期葡萄膜黑色素瘤患者中的34例影像学进展患者。
- 按ctDNA是否下降≥0.5 log将34例患者分为两组,分析基线和第8周CT/MRI病灶的影像组学特征,得到两个机器学习衍生的患者特征标签(16个特征)。
- 评估患者水平分析(n=32)和病灶水平分析(n=148)的ROC AUC(95%CI)。
主要发现
- 患者水平分析中,体积特征将患者分组的ROC AUC为0.71[0.53-0.90],最佳阈值0.57时特异度63%、敏感度81%。
- 病灶水平分析中,影像组学特征的ROC AUC为0.70[0.58-0.81],最佳阈值0.53时特异度66%、敏感度74%。
- B组基线肿瘤病灶体积更低(ROC AUC=0.65),基线肿瘤异质性不同(ROC AUC=0.66),第8周肿瘤异质性变化在CT/MRI上ROC AUC分别为0.66/0.69。
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
STUDY AIM: Tebentafusp, a bispecific fusion protein consisting of affinity-enhanced T cell receptor fused to anti-CD3 effector, has shown overall survival (OS) benefits across all RECIST response categories, including progressive disease (PD). In a phase 2 trial (NCT02570308) for advanced uveal melanoma (mUM), 35 % of PD patients experienced ≥ 0.5 log ctDNA reduction, resulting in a median overall survival (OS) of ∼17 months, compared to ∼8 months in the non-ctDNA reduction group. METHODS: A total of 34 of 127 2L+ mUM patients with PD were split into two groups based on absence or presence of ctDNA reduction (≥0.5 log reduction). Lesions from CT and MRI scans were analyzed using radiomics features at baseline and week eight, yielding two machine-learning-derived patient signatures (16 features). Performance of per-patient analysis (n = 32) and per-lesion analysis (n = 148) was assessed using ROC AUC (95 % confidence interval [CI]). RESULTS: In the per-patient analysis, a volumetric signature classified patients into groups with ROC AUC of 0.71 [0.53-0.90] with 63 % specificity and 81 % sensitivity at the optimal threshold (0.57). In the per-lesion analysis, a radiomic signature reached an ROC AUC of 0.70 [0.58-0.81] with 66 % specificity and 74 % sensitivity at the optimal threshold (0.53). Group B had lower baseline tumor lesion volume (ROC AUC=0.65), distinct baseline (ROC AUC=0.66), and change by week eight (ROC AUC=0.66/0.69 on CT/MRI) in tumor heterogeneity. CONCLUSION: Radiomic analysis accurately predicted ctDNA reduction in PD patients at both the patient and lesion level. The most influential predictor was decreased tumor heterogeneity observed on CT/MRI.
引用
- Beylergil V, Collins L, Schwartz LH, et al. Radiomic markers associated with clinical benefit in patients with radiographic progression of advanced uveal melanoma on tebentafusp. Eur J Cancer. 2025. PMID: 40174442. doi:10.1016/j.ejca.2025.115386
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。