Abstract: Background The World Health Organization（WHO）/International Society of Urological Pathology（ISUP） grading standard for clear cell renal cell carcinoma（ccRCC） serves as a critical basis for formulating treatment plans and evaluating patient prognosis. Conventional ultrasound lacks sufficient sensitivity to detect tumor micro-heterogeneity，and traditional needle biopsy carries inherent risks of trauma. Consequently，there is an urgent need to identify a reliable and non-invasive diagnostic method. Objective To construct a nomogram model for preoperative prediction of WHO/ISUP grading of clear cell renal cell carcinoma（ccRCC） by conventional ultrasound combined with radiomics，providing a basis for precise clinical diagnosis and treatment. Methods Clinical data of 296 patients with pathologically confirmed ccRCC were retrospectively collected from the Affiliated Hospital of North China University of Science and Technology and the Second Hospital of Tianjin Medical University between May 2018 and June 2025. Based on the inclusion and exclusion criteria，a total of 258 patients with 258 tumors were ultimately enrolled. Among them，196 pathologically confirmed ccRCC lesions from the Affiliated Hospital of North China University of Science and Technology were divided into a training set（146 lesions） and an internal validation set （50 lesions），while 62 ccRCC lesions from the Second Hospital of Tianjin Medical University served as an external test set. Conventional ultrasound features of renal lesions were recorded，and 3D-Slicer was used to segment tumor boundaries and extract radiomic features. Dimensionality reduction was applied to identify robust features. Binary logistic regression was performed to identify independent factors influencing ccRCC nuclear grading，and nomograms were constructed. Model performance was evaluated using receiver operating characteristic（ROC） curves，calibration curves，and decision curve analysis（DCA）. Results A significant difference in internal echo patterns of tumors was observed between the training and validation sets （P<0.05）. In the training set，statistically significant differences were identified in lesion size and lesion contour regularity between the low-grade and high-grade groups（P<0.05）. Binary logistic regression analysis revealed that lesion size（OR=0.319，95%CI=0.114-0.894，P=0.030），lesion contour regularity（OR=2.905，95%CI=1.243-6.790，P=0.014），Original_ shape_Sphericity（OR=1.927，95%CI=1.121-3.312，P=0.018），and Wavelet_LLL_glcm_InverseVariance（OR=1.824， 95%CI=1.117-2.979，P=0.016） were independent predictors of ccRCC nuclear grading.Based on these findings，three predictive nomogram models were constructed：a conventional ultrasound feature model，a radiomic feature model，and a combined model. ROC curve analysis demonstrated that the areas under the curve（AUC） for predicting WHO/ISUP grading of ccRCC in the training set were 0.732，0.754，and 0.817，respectively. In the validation set，the AUC values were 0.733，0.748，and 0.829，respectively，and in the test set，they were 0.676，0.742，and 0.817，respectively. Calibration curves indicated good agreement between predicted and actual probabilities for all models（all P>0.05）. Decision curve analysis showed that the combined model provided higher net clinical benefit across a threshold probability range of 0% to 60%（training set：0.297；validation set：0.291；test set：0.291）. Conclusion Conventional ultrasound combined with radiomics nomogram models can effectively predict the WHO/ISUP grading of ccRCC，providing a reliable basis for preoperative evaluation.

Key words: Clear cell renal cell carcinoma, WHO/ISUP classification, Conventional ultrasound, Radiomics, nomogram

摘要： 背景 肾透明细胞癌（ccRCC）的WHO/国际泌尿病理学会（ISUP）分级标准是制定治疗方案及评估患者预后的关键依据。常规超声对于肿瘤微观异质性敏感度不足，且传统穿刺活检存在创伤风险，寻找一种可靠、无创的诊断方法迫在眉睫。目的 构建常规超声联合影像组学术前预测ccRCC WHO/ISUP分级的列线图模型，为临床精准诊疗提供依据。方法 回顾性收集2018年5月—2025年6月华北理工大学附属医院和天津医科大学第二医院经手术病理确诊的296例ccRCC患者的临床资料。按照纳入和排除标准，最终纳入ccRCC患者共258例，共258个肿物。其中来自华北理工大学附属医院经病理确诊的196个ccRCC肿物分为训练集（146个）和验证集（50个），来自天津医科大学第二医院的62个ccRCC肿物作为测试集；采集肾肿物常规超声特征，并利用3D-Slicer勾画肿物边界、提取影像组学特征，经降维处理获得稳定的特征。通过二元Logistic回归分析ccRCC核分级的独立影响因素，并绘制列线图，采用ROC曲线、校准曲线及临床决策曲线（DCA）评估模型性能。结果 训练集和验证集之间肿物内部回声比较，差异有统计学意义（P<0.05）；训练集中低级别组与高级别组患者病灶大小、病灶形态是否规整比较，差异有统计学意义（P<0.05）；二元Logisitc回归分析结果显示，病灶大小（OR=0.319，95%CI=0.114~0.894，P=0.030）、病灶形态是 否 规 整（OR=2.905，95%CI=1.243~6.790，P=0.014）、Original_shape_Sphericity（OR=1.927，95%CI=1.121~3.312，P=0.018）、Wavelet_LLL_glcm_InverseVariance（OR=1.824，95%CI=1.117~2.979，P=0.016）为ccRCC核分级的独立影响因素。基于上述结果，建立常规超声特征列线图预测模型、影像组学特征列线图预测模型及联合列线图预测模型，ROC曲线分析结果显示，训练集中三者预测ccRCC WHO/ISUP分级的ROC曲线下面积（AUC）分别为0.732、0.754、0.817；验证集中三者的AUC分别为0.733、0.748、0.829；测试集中三者的AUC分别为0.676、0.742、0.817。校准曲线表明各模型预测概率与实际预测概率一致性良好（P均>0.05）。决策曲线分析结果显示，联合预测模型在0%~60%的临床决策关键区间内能提供更高的临床净收益（训练集：0.297，验证集：0.291，测试集：0.291）。结论 常规超声联合影像组学列线图模型可有效预测ccRCC WHO/ISUP分级，为术前评估提供可视化可靠依据。

关键词: 肾透明细胞癌, WHO/ISUP 分级, 常规超声, 影像组学, 列线图