开发基于放射组学的模型，使用 [68Ga]Ga-PSMA PET/CT 预测活检格里森分级组 (GGG) 1-2 前列腺癌 (PCa) 患者的术后不良病理 (AP)，协助选择患者主动监测（AS）。共有 75 名活检 GGG 1-2 PCa 并接受根治性前列腺切除术（RP）的男性入组。患者被随机分为训练组（70%）和测试组（30%）。从[68Ga]Ga-PSMA PET扫描中提取整个前列腺的放射组学特征，并使用最小冗余最大相关算法和最小绝对收缩和选择算子回归模型进行选择。进行逻辑回归分析以构建预测模型。分别采用受试者工作特征（ROC）曲线、决策曲线分析（DCA）和校准曲线评价模型的诊断价值、临床实用性和预测准确性。75例患者中，30例经RP确诊为AP。临床模型显示训练集中的曲线下面积 (AUC) 为 0.821 (0.695-0.947)，测试集中的曲线下面积 (AUC) 为 0.795 (0.603-0.987)。放射组学模型在训练集中达到 0.830 (0.720-0.941) 的 AUC 值，在测试集中达到 0.829 (0.624-1.000) 的 AUC 值。该组合模型结合了放射组学评分 (Radscore) 和游离前列腺特异性抗原 (FPSA)/总前列腺特异性抗原 (TPSA)，表现出比临床和放射组学模型更高的诊断功效，AUC 值为 0.875 (0.780 -0.970）在训练集中，0.872（0.678-1.000）在测试集中。 DCA 表明，组合模型和放射组学模型的净效益超过了临床模型。组合模型显示出根据最终病理学中 AP 的存在对活检 GGG 1-2 PCa 的男性进行分层的潜力，并且优于仅基于临床或放射组学特征。它有望帮助泌尿科医生更好地选择适合 AS 的患者。© 2024。作者。

To develop a radiomics-based model using [68Ga]Ga-PSMA PET/CT to predict postoperative adverse pathology (AP) in patients with biopsy Gleason Grade Group (GGG) 1-2 prostate cancer (PCa), assisting in the selection of patients for active surveillance (AS).A total of 75 men with biopsy GGG 1-2 PCa who underwent radical prostatectomy (RP) were enrolled. The patients were randomly divided into a training group (70%) and a testing group (30%). Radiomics features of entire prostate were extracted from the [68Ga]Ga-PSMA PET scans and selected using the minimum redundancy maximum relevance algorithm and the least absolute shrinkage and selection operator regression model. Logistic regression analyses were conducted to construct the prediction models. Receiver operating characteristic (ROC) curve, decision curve analysis (DCA), and calibration curve were employed to evaluate the diagnostic value, clinical utility, and predictive accuracy of the models, respectively.Among the 75 patients, 30 had AP confirmed by RP. The clinical model showed an area under the curve (AUC) of 0.821 (0.695-0.947) in the training set and 0.795 (0.603-0.987) in the testing set. The radiomics model achieved AUC values of 0.830 (0.720-0.941) in the training set and 0.829 (0.624-1.000) in the testing set. The combined model, which incorporated the Radiomics score (Radscore) and free prostate-specific antigen (FPSA)/total prostate-specific antigen (TPSA), demonstrated higher diagnostic efficacy than both the clinical and radiomics models, with AUC values of 0.875 (0.780-0.970) in the training set and 0.872 (0.678-1.000) in the testing set. DCA showed that the net benefits of the combined model and radiomics model exceeded those of the clinical model.The combined model shows potential in stratifying men with biopsy GGG 1-2 PCa based on the presence of AP at final pathology and outperforms models based solely on clinical or radiomics features. It may be expected to aid urologists in better selecting suitable patients for AS.© 2024. The Author(s).