Gao Jie, Fu Yao, He Kuiqiang, Xu Qinfeng, Wang Feng, Guo Hongqian
Department of Urology, Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School, Institute of Urology Nanjing University, Jiangsu, China.
Department of Pathology, Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School, Jiangsu, China.
Mol Imaging Biol. 2025 Feb;27(1):44-53. doi: 10.1007/s11307-024-01974-2. Epub 2024 Dec 16.
To develop a novel risk model incorporating Ga-PSMA PET/CT parameters for prediction of perineural invasion (PNI) of prostate cancer (PCa).
The study retrospectively enrolled 192 PCa patients with preoperative multiparametric MRI, Ga-PSMA PET/CT and radical specimen. Imaging parameters were derived from both mpMRI and PET/CT images. S100 immunohistochemistry staining was conducted to evaluate PNI of PCa. Significant predictors were derived with univariate and multivariate logistic regression analyses, and the PNI-risk nomogram was constructed with significant predictors. Internal discrimination validation was performed with receiver operating characteristic analysis. Calibration curves were plotted, decision curve and clinical impact curve analysis were performed for clinical benefit exploration.
With the median peritumoral nerve density of 6, patients were stratified as low-PNI group (nerve density < 6, n = 78, 40.6%) and high-PNI group (nerve density ≥ 6, n = 114, 59.4%). Compared with low-PNI PCa, high-PNI PCa harbored significantly larger imaging lesion diameter (P < 0.001), higher PI-RADS score (P = 0.009), higher SUVmax (P < 0.001), larger tumor diameter (P = 0.024) and higher Gleason grade group (P < 0.001). Further, with univariate and multivariate analyses, imaging lesion diameter (OR 2.98, 95% CI 1.73-5.16, P = 0.004) and SUVmax (OR 3.59, 95%CI 2.32-5.55, P < 0.001) and were identified as independent predictors for PNI in PCa, and a PNI-risk nomogram incorporating these two predictors was constructed. The PNI-risk nomogram demonstrated considerable calibration (mean absolute error 0.026) and discrimination (area under the curve = 0.889, sensitivity 73.1%, specificity 97.4%) abilities, harboring net benefits with threshold probabilities range from 0 to 0.80.
Ga-PSMA PET/CT-based model could effectively predict the perineural invasion of PCa. These results may help with the decision-making on active surveillance, focal therapy and surgery approach. Additionally, patients suspicious of high-density PNI PCa should receive more radical treatment than low-PNI PCa.
开发一种结合镓-前列腺特异性膜抗原（Ga-PSMA）正电子发射断层扫描/计算机断层扫描（PET/CT）参数的新型风险模型，用于预测前列腺癌（PCa）的神经周围侵犯（PNI）。
本研究回顾性纳入了192例术前行多参数磁共振成像（mpMRI）、Ga-PSMA PET/CT检查及根治性标本的PCa患者。成像参数来自mpMRI和PET/CT图像。进行S100免疫组织化学染色以评估PCa的PNI。通过单因素和多因素逻辑回归分析得出显著预测因子，并使用显著预测因子构建PNI风险列线图。采用受试者操作特征分析进行内部判别验证。绘制校准曲线，进行决策曲线和临床影响曲线分析以探索临床获益。
肿瘤周围神经密度中位数为6，患者被分为低PNI组（神经密度＜6，n = 78，40.6%）和高PNI组（神经密度≥6，n = 114，59.4%）。与低PNI的PCa相比，高PNI的PCa具有显著更大的成像病变直径（P＜0.001）、更高的前列腺影像报告和数据系统（PI-RADS）评分（P = 0.009）、更高的最大标准摄取值（SUVmax）（P＜0.001）、更大的肿瘤直径（P = 0.024）和更高的Gleason分级组（P＜0.001）。此外，通过单因素和多因素分析，成像病变直径（比值比[OR]2.98，95%置信区间[CI]1.73 - 5.16，P = 0.004）和SUVmax（OR 3.59，95%CI 2.32 - 5.55，P＜0.001）被确定为PCa中PNI的独立预测因子，并构建了包含这两个预测因子的PNI风险列线图。PNI风险列线图显示出良好的校准能力（平均绝对误差0.026）和判别能力（曲线下面积 = 0.889，灵敏度73.1%，特异性97.4%），在阈值概率范围为0至0.80时具有净获益。
基于Ga-PSMA PET/CT的模型可以有效预测PCa的神经周围侵犯。这些结果可能有助于在主动监测、局部治疗和手术方法的决策制定。此外，怀疑为高密度PNI的PCa患者应比低PNI的PCa患者接受更积极的治疗。
