超声组学对前列腺癌碘125粒子植入治疗预后的价值

Prognostic Value of Ultrasonography-based Radiomics in the Treatment of Prostate Cancer with Iodine-125 Implantation

目的:探讨超声组学对前列腺癌碘125粒子植入治疗预后的价值。方法:回顾性分析2022年8月至2024年1月283例前列腺癌患者资料。将入组患者按7∶3的比例随机分为训练集(n=198)和验证集(n=85),将总前列腺特异性抗原(tPSA)>10 ng/mL认定为预后不良,tPSA<10 ng/mL认定为预后良好。Logistic回归分析临床协变量;组内相关系数(ICC)和线性相关性检验,建立LASSO回归筛选较优特征,支持向量机(SVM)构建影像组学标签;利用Logistic回归构建出一个可视化的列线图模型,ROC曲线评估区分度,校准曲线评估准确度,决策曲线评估临床收益。结果:tPSA、前列腺特异性抗原密度(PSAD)、影像组学评分、前列腺体积是前列腺癌患者植入碘125粒子后预后良好的独立影响因素(P<0.05),其中tPSA、PSAD、影像组学评分每增加一个单位,患者预后良好的概率分别增加1.114、1.582、1.873倍,95%CI分别为1.109~1.283、1.431~1.624、1.656~5.361;而前列腺体积每增加一个单位,患者预后良好的概率减少3.3%,95%CI为0.943~0.985。最终筛选出与前列腺癌预后相关的影像组学标签5个特征(ICC>0.750)。在训练集中,列线图模型、影像组学模型、临床模型AUC分别为0.846(0.802~0.884)、0.788(0.752~0.823)、0.767(0.685~0.817),其中列线图模型AUC最大,预测效能最好。校准曲线和Hosmer-Lemeshow检验结果均显示列线图模型具有较高的预测性能,检验结果显示预测值与实际值拟合度为(χ^(2)=1.827,P=0.216),差异不具有统计学意义,拟合度较高。模型净收益远高于影像学和临床特征。结论:综合上述,tPSA、PSAD、影像组学评分、前列腺体积是前列腺癌患者植入碘125粒子后预后良好的独立影响因素;超声影像组学联合临床协变量构建的列线图模型在预测前列腺癌患者植入碘125粒子后的效果中具有较高的区分度和准确度,在临床应用中具有较高的临床决策价值。

Objective:To investigate the value of ultrasonography in prognosis of prostate cancer treated with iodine-125 particle implantation. Methods: The data of 283 patients with PCa from August 2022 to January 2024 were retrospectively analyzed. The enrolled patients were randomly divided into the training set ( n =198) and the test set ( n =85) at a ratio of 7∶3. tPSA>10 ng/mL was defined as poor prognosis, and tPSA<10 ng/mL was defined as good prognosis. Logistic regression was used to analyze clinical covariates. Intraclass correlation coefficient (ICC) and linear correlation test were used to establish LASSO regression to select the best features, and support vector machine (SVM) was used to construct the radiomics signature. Logistic regression was used to construct a visual nomogram model. The ROC curve was used to evaluate the discrimination, the calibration curve was used to evaluate the accuracy, and the decision curve was used to evaluate the clinical benefit. Results: The tPSA, PSAD, radiomics score, and prostate volume are independent factors associated with a favorable prognosis after implantation of iodine-125 particles in patients with prostate cancer ( P <0.05). For each unit increase in tPSA, PSAD, and radiomics score, the likelihood of a favorable prognosis for patients increases by 1.114, 1.582, and 1.873 fold, respectively, with 95% confidence intervals of 1.109-1.283, 1.431-1.624 , and 1.656-5.361, respectively. However, for each unit increase in prostate volume, the likelihood of a favorable prognosis for patients decreases by 3.3%, with a 95% confidence interval of 0.943-0.985. Finally, five radiomics features (ICC>0.750) related to the prognosis of prostate cancer were selected. In the training set, the AUC values of the nomogram model, radiomics model, and clinical model were 0.846 (0.802-0.884), 0.788 (0.752-0.823), and 0.767 (0.685-0.817), respectively. Among them, the nomogram model had the highest AUC value and the best predictive performance. Both the calibration curve and the Hosmer-Lemeshow test results showed that the nomogram model had high predictive performance. The test results showed that the goodness-of-fit between the predicted and actual values was ( χ^(2)=1.827, P =0.216), indicating no statistically significant difference and a high degree of fit. The net benefit of the model was much higher than that of imaging and clinical features. Conclusion: The tPSA, PSAD, radiomics score, and prostate volume are independent factors associated with a favorable prognosis after implantation of iodine-125 particles in patients with prostate cancer. The nomogram model constructed by combining ultrasonic radiomics with clinical covariates has high discrimination and accuracy in predicting the effect of iodine-125 particle implantation in patients with prostate cancer, and has high clinical decision-making value in clinical applications.