Objective: To investigate the predictive value of multimode MRI features for nuclear grade of clear cell renal cell carcinoma (ccRCC). Methods: From January 2016 to October 2017, 381 patients (387 tumors) with ccRCC proven by pathology in Chinese PLA General Hospital First Medical Center were enrolled (male 293, female 88, age 24-87 years old). The clinical and imaging data of these patients were retrospectively analyzed, including clinical information (gender, age, BMI, smoke, hypertension) and preoperative renal MRI. Pre-and post-contrast MRI features were subjectively scored. The largest diameter of each lesion was measured. Two-sample t-test,Chi-squared test and binuary Logistic regression analysis were used to evaluate the predictive efficacy of clinical and MRI data. Results: According to WHO/ISUP nuclear grade system,all ccRCCs (n=387) were divided into low grade (n=322) and high grade group (n=65). Between two groups, there were significant differences in age and diameter((54±12) vs (59±10) years old, P=0.001; (4.1±2.2) vs (6.2±3.0) cm, P<0.01). In MRI scores,there were significant differences in scores of pseudocapsule, shape and margin,hemorrhage,enhancement degree,cystic-solid,intratumoral vessel,peritumoral vessel, renal sinus invasion, vein thrombosis, lymphadenopathy, necrosis, perinephric invasion and metastasis, DWI signal intensity between high grade group and low grade group (all P<0.01). Binuary Logistic regression analysis showed that shape and margin, enhancement degree and DWI signal intensity were independent predictors for high grade ccRCC (OR=0.181, 95%CI 0.049-0.666; OR=0.393, 95%CI 0.182-0.846; OR=0.336, 95%CI 0.155-0.728). A nomogram model for predicting the risk of high grade ccRCC was constructed. Conclusions: Multimode MRI features can differentiate low grade and high grade ccRCC. The nomogram developed in this study might aid urologist in the pre-operative prediction of nuclear grade of ccRCC,which might contribute to developing treatment strategy.
目的： 探讨应用多模态磁共振成像（MRI）的图像特征分析肾透明细胞癌核分级的效能。 方法： 回顾性分析2016年1月至2017年10月于解放军总医院第一医学中心经手术病理证实的381例肾透明细胞癌患者（共387个肿瘤）的临床和影像学资料，其中男293例、女88例，年龄24~87岁，包括一般临床资料（性别、年龄、体质指数、有无吸烟史、有无高血压史）、术前肾脏MRI平扫和动态增强检查。对MRI图像特征进行定性评分，测量肿瘤最大径。采用t检验、χ(2)检验及二元Logistic回归分析临床指标及MRI图像特征对肾透明细胞癌核分级的预测效能。 结果： 根据世界卫生组织/国际泌尿病理学会（WHO/ISUP）核分级系统，387个肾透明细胞癌分为低级别组（1级和2级）322个肿瘤和高级别组（3级和4级）65个肿瘤。低级别组和高级别组之间的患者年龄、肿瘤最大径差异有统计学意义[（54±12）比（59±10）岁，P=0.001；（4.1±2.2）比（6.2±3.0）cm，P<0.01]。在MRI图像评分中，两组之间的假包膜、形状与边界、瘤内出血、皮髓质期强化程度、囊实性、瘤内血管、瘤周血管、肾窦受累、静脉瘤栓、淋巴结肿大、瘤内坏死、肾周受累和远处转移、DWI信号强度的评分差异均有统计学意义（均P<0.01）。二元Logistic回归分析显示形状与边界、皮髓质期强化程度及DWI信号强度是预测肾透明细胞癌核分级的独立因素（OR=0.181，95%CI 0.049~0.666；OR=0.393，95%CI 0.182~0.846；OR=0.336，95%CI 0.155~0.728），成功构建用于预测高级别肾透明细胞癌风险的列线图模型。 结论： 多模态MRI图像特征能够鉴别低级别和高级别肾透明细胞癌。本研究建立的列线图模型可以帮助泌尿外科医师术前预测肾透明细胞癌的核分级，有助于制定最佳的治疗方案。.
Keywords: Forecasting; Magnetic resonance imaging; Nomograms; Sarcoma, clear cell.