1UCLA Brain Tumor Imaging Laboratory (BTIL), Center for Computer Vision and Imaging Biomarkers, University of California, Los Angeles.
Departments of2Radiological Sciences.
J Neurosurg. 2023 Aug 4;140(2):338-349. doi: 10.3171/2023.5.JNS23285. Print 2024 Feb 1.
The objective of this study was to identify baseline clinical and radiological characteristics of brain metastases (BMs) associated with a higher probability of lesion-specific progression-free survival (PFS-L) after laser interstitial thermal therapy (LITT).
A total of 47 lesions in 42 patients with BMs treated with LITT were retrospectively examined, including newly diagnosed BM, suspected recurrent BM, and suspected radiation necrosis. The association of baseline clinical and radiological features with PFS-L was assessed using survival analyses. Radiological features included lesion size measurements, diffusion and perfusion metrics, and sphericity, which is a radiomic feature ranging from 1 (perfect sphere) to 0.
The probability of PFS-L for the entire cohort was 88.0% at 3 months, 70.6% at 6 months, 67.4% at 1 and 2 years, and 62.2% at 3 years. For lesions progressing after LITT (n = 13), the median time to progression was 3.9 months, and most lesions (n = 11) progressed within 6 months after LITT. In lesions showing response to LITT (n = 17), the median time to response was 12.1 months. All 3 newly diagnosed BMs showed a long-term response. The mean (± SD) follow-up duration for all censored lesions (n = 34) was 20.7 ± 19.4 months (range 12 days to 6.1 years). The mean pretreatment enhancing volume was 2.68 cm3 and the mean sphericity was 0.70. Pretreatment small enhancing volume (p = 0.003) and high sphericity (p = 0.024) computed from lesion segmentation predicted a longer PFS-L after LITT. Lesions meeting optimal cutoffs of either enhancing volume < 2.5 cm3 (adjusted p = 0.004) or sphericity ≥ 0.705 (adjusted p = 0.019) had longer PFS-L, and their probability of PFS-L was 86.8% at 3 years. Lesions meeting both cutoffs showed a cumulative benefit (p < 0.0001), with a 100% probability of PFS-L at 3 years, which was unchanged at the end of follow-up (4.1 years). Manually computed estimates of lesion size (maximal axial diameter, p = 0.011) and sphericity (p = 0.043) were also predictors of PFS-L. Optimal cutoffs of diameter < 2 cm (adjusted p = 0.035) or manual sphericity ≥ 0.91 (adjusted p = 0.092) identified lesions with longer PFS-L, and lesions meeting both cutoffs showed a cumulative benefit (p = 0.0023). Baseline diffusion imaging did not predict PFS-L. A subset of lesions (n = 7) with highly perfused hotspots had worse PFS-L (adjusted p = 0.010), but perfusion signal contamination from vessels and cortex and underlying size differences were possible confounders.
Small size and high sphericity are ideal baseline features for lesions considered for LITT treatment, with a cumulative PFS-L benefit when both features are present, that could aid patient selection.
本研究旨在确定与激光间质热疗（LITT）后病变特异性无进展生存期（PFS-L）较高相关的脑转移瘤（BMs）的基线临床和影像学特征。
回顾性分析了 42 例接受 LITT 治疗的 BMs 患者的 47 个病灶，包括新诊断的 BM、疑似复发性 BM 和疑似放射性坏死。使用生存分析评估基线临床和影像学特征与 PFS-L 的相关性。影像学特征包括病变大小测量、弥散和灌注指标以及各向异性，各向异性是一种从 1（完美球体）到 0 的放射组学特征。
整个队列的 PFS-L 概率在 3 个月时为 88.0%，在 6 个月时为 70.6%，在 1 年和 2 年时为 67.4%，在 3 年时为 62.2%。对于 LITT 后进展的病变（n=13），进展中位时间为 3.9 个月，大多数病变（n=11）在 LITT 后 6 个月内进展。在显示对 LITT 有反应的病变（n=17）中，反应的中位时间为 12.1 个月。所有 3 个新诊断的 BM 均表现出长期反应。所有 34 个有记录的病变（n=34）的中位随访时间为 20.7±19.4 个月（范围 12 天至 6.1 年）。预处理强化体积的平均值（±SD）为 2.68cm3，平均各向异性为 0.70。LITT 后 PFS-L 较长的预测因素为预处理小强化体积（p=0.003）和高各向异性（p=0.024），其病变分割计算得出。满足强化体积<2.5cm3（调整后 p=0.004）或各向异性≥0.705（调整后 p=0.019）的最佳截值的病变具有更长的 PFS-L，其 3 年 PFS-L 概率为 86.8%。满足两个截值的病变显示出累积获益（p<0.0001），3 年的 PFS-L 概率为 100%，在随访结束时（4.1 年）保持不变。病变大小（最大轴向直径，p=0.011）和各向异性（p=0.043）的手动计算估计也是 PFS-L 的预测因子。直径<2cm（调整后 p=0.035）或手动各向异性≥0.91（调整后 p=0.092）的最佳截值确定了具有更长 PFS-L 的病变，满足两个截值的病变显示出累积获益（p=0.0023）。基线弥散成像不能预测 PFS-L。具有高灌注热点的病变亚组（n=7）的 PFS-L 更差（调整后 p=0.010），但来自血管和皮质的灌注信号污染以及潜在的大小差异可能是混杂因素。
小尺寸和高各向异性是考虑接受 LITT 治疗的病变的理想基线特征，当同时存在这两个特征时，具有累积的 PFS-L 获益，这可能有助于患者选择。
