Wei Jinyan, Ma Yurong, Liu Jianqiang, Zhao Jianhong, Zhou Junlin
Department of Radiology, Lanzhou University Second Hospital, Lanzhou, China; Key Laboratory of Medical Imaging of Gansu Province, Lanzhou University Second Hospital, Lanzhou, China; Second Clinical School, Lanzhou University, Lanzhou, China; Gansu International Scientific and Technological Cooperation Base of Medical Imaging Artificial Intelligence, Lanzhou, China.
Department of Radiology, Lanzhou University Second Hospital, Lanzhou, China; Key Laboratory of Medical Imaging of Gansu Province, Lanzhou University Second Hospital, Lanzhou, China; Second Clinical School, Lanzhou University, Lanzhou, China; Gansu International Scientific and Technological Cooperation Base of Medical Imaging Artificial Intelligence, Lanzhou, China.
Urol Oncol. 2025 May;43(5):332.e1-332.e10. doi: 10.1016/j.urolonc.2024.11.013. Epub 2024 Dec 8.
To determine the diagnostic value of a comprehensive model based on unenhanced computed tomography (CT) images for distinguishing fat-poor angiomyolipoma (fp-AML) from homogeneous clear cell renal cell carcinoma (hm-ccRCC).
We retrospectively reviewed 27 patients with fp-AML and 63 with hm-ccRCC. Demographic data and conventional CT features of the lesions were recorded (including sex, age, symptoms, lesion location, shape, boundary, unenhanced CT attenuation and so on). Whole tumor regions of interest were drawn on all slices to obtain histogram parameters (including minimum, maximum, mean, percentile, standard deviation, variance, coefficient of variation, skewness, kurtosis, and entropy) by two radiologists. Chi-square test, Mann-Whitney U test, or independent samples t-test were used to compare demographic data, CT features, and histogram parameters. Multivariate logistic regression analyses were used to screen for independent predictors distinguishing fp-AML from hm-ccRCC. Receiver operating characteristic curves were constructed to evaluate the diagnostic performances of the models.
Age, sex, tumor boundary, unenhanced CT attenuation, maximum tumor diameter, and tumor volume significantly differed between patients with fp-AML and those with hm-ccRCC (P < 0.05). The minimum, mean, first percentile (Perc.01), Perc.05, Perc.10, Perc.25, Perc.50, Perc.75, Perc.90, Perc.95, and Perc.99 of the Fp-AML group were higher than those of the hm-ccRCC group (P < 0.05). Coefficient of variance, skewness, and kurtosis were lower than those in the hm-ccRCC group (all P < 0.05). Age, maximum tumor diameter, unenhanced CT attenuation, and Perc.25 were independent predictors for distinguishing fp-AML from hm-ccRCC (all P < 0.05). The comprehensive model, incorporating age, maximum tumor diameter, unenhanced CT attenuation, and Perc.25, showed the best diagnostic performance (AUC = 0.979).
The comprehensive model based on unenhanced CT imaging can accurately distinguish fp-AML from hm-ccRCC and may assist clinicians in tailoring precise therapy, while also helping to improve the diagnosis and management of renal tumors, leading to the selection of effective treatment options.
确定基于平扫计算机断层扫描（CT）图像的综合模型对鉴别乏脂性血管平滑肌脂肪瘤（fp-AML）与均质型透明细胞肾细胞癌（hm-ccRCC）的诊断价值。
我们回顾性分析了27例fp-AML患者和63例hm-ccRCC患者。记录病变的人口统计学数据和常规CT特征（包括性别、年龄、症状、病变位置、形状、边界、平扫CT衰减等）。由两名放射科医生在所有层面上绘制整个肿瘤感兴趣区，以获得直方图参数（包括最小值、最大值、平均值、百分位数、标准差、方差、变异系数、偏度、峰度和熵）。采用卡方检验、曼-惠特尼U检验或独立样本t检验比较人口统计学数据、CT特征和直方图参数。采用多变量逻辑回归分析筛选区分fp-AML与hm-ccRCC的独立预测因素。构建受试者操作特征曲线以评估模型的诊断性能。
fp-AML患者与hm-ccRCC患者在年龄、性别、肿瘤边界、平扫CT衰减、最大肿瘤直径和肿瘤体积方面存在显著差异（P<0.05）。fp-AML组的最小值、平均值、第1百分位数（Perc.01）、Perc.05、Perc.10、Perc.25、Perc.50、Perc.75、Perc.90、Perc.95和Perc.99均高于hm-ccRCC组（P<0.05）。变异系数、偏度和峰度低于hm-ccRCC组（均P<0.05）。年龄、最大肿瘤直径、平扫CT衰减和Perc.25是区分fp-AML与hm-ccRCC的独立预测因素（均P<0.05）。纳入年龄、最大肿瘤直径、平扫CT衰减和Perc.25的综合模型显示出最佳诊断性能（AUC=0.979）。
基于平扫CT成像的综合模型能够准确区分fp-AML与hm-ccRCC，可能有助于临床医生制定精准治疗方案，同时有助于改善肾肿瘤的诊断和管理，并选择有效的治疗方案。
