南方医科大学学报 ›› 2023, Vol. 43 ›› Issue (5): 800-806.doi: 10.12122/j.issn.1673-4254.2023.05.16

• • 上一篇    下一篇

透明细胞癌可能性评分v1.0和v2.0的应用价值对比

郝雨薇,高 升,张潇月,崔梦秋,丁效蕙,王 鹤,杨大为,叶慧义,王海屹   

  1. 解放军医学院,北京 100853;中国人民解放军总医院第一医学中心放射诊断科,病理科,北京 100853;临沂市中心医院放射科,山东 临沂 276400;山西医科大学第一医院影像科,山西 太原 030012;北京大学第一医院医学影像科,北京 100035;北京友谊医院放射科,北京 100050
  • 出版日期:2023-05-20 发布日期:2023-06-12

Comparison of diagnostic performance of Clear Cell Likelihood Score v1.0 and v2.0 for clear renal cell carcinoma

HAO Yuwei, GAO Sheng, ZHANG Xiaoyue, CUI Mengqiu, DING Xiaohui, WANG He, YANG Dawei, YE Huiyi, WANG Haiyi   

  1. Medical School of Chinese PLA, Beijing 100853, China; Department of Radiology, Department of Pathology, First Medical Center of Chinese PLA General Hospital, Beijing 100853, China; Department of Radiology, Linyi Central Hospital, Linyi 276400, China; Department of Radiology, First Hospital of Shanxi Medical University, Taiyuan 030012, China; Department of Radiology, Peking University First Hospital, Beijing 100035, China; Department of Radiology, Beijing Friendship Hospital, Beijing 100050, China
  • Online:2023-05-20 Published:2023-06-12

摘要: 目的 探讨透明细胞癌可能性评分(ccLS)v1.0和v2.0对肾脏实性小肿瘤(SRM)中肾透明细胞癌(ccRCC)的应用价值。方法 回顾性收集解放军总医院第一医学中心2018年1月1日~2021年12月31日经病理证实SRM患者的临床资料和MR检查资料,并收集首都医科大学附属友谊医院、北京大学第一医院于2019年1月1日~2021年5月17日行肾脏MR检查的SRM病例。6名放射科医师经ccLS培训后分别运用v1.0和v2.0版本进行独立评分。运用随机效应Logistic回归模型绘制合并受试者工作特征曲线以评估ccLS v1.0和ccLS v2.0诊断ccRCC效能,运用DeLong检验比较两者曲线下面积(AUC)差异。运用加权Kappa检验评价评分结果一致性,并采用Gwet一致性系数对加权Kappa系数的差异进行比较。结果 研究共纳入691例患者(700个肿瘤),其中,男491例(71.1%),女200例(28.9%),年龄54±12岁。对SRM的评分结果显示:6名医师运用ccLS v1.0和v2.0诊断ccRCC的合并准确度、灵敏度、特异度、阳性预测值和阴性预测值分别为77.1%、76.8%、77.7%、90.2%、55.7%和80.9%、79.3%、85.1%、93.4%、60.6%。ccLS v1.0和ccLS v2.0诊断ccRCC的合并AUC分别为0.859(95% CI:0.149,0.793)和0.897(95% CI:0.223,0.768),差异有统计学意义(P<0.01)。ccLS v2.0阅片者间一致性稍高于ccLS v1.0(平均κ=0.60和κ=0.56),差异无统计学意义(P>0.05)。结论 ccLS v2.0诊断ccRCC应用价值优于ccLS v1.0,可以考虑用于辅助影像医师日常诊断工作。

关键词: 多参数磁共振成像, 肾脏小肿瘤, 透明细胞癌可能性评分, 肾透明细胞癌

Abstract: Objective To compare the performance of Clear Cell Likelihood Score (ccLS) v1.0 and v2.0 in diagnosing clear cell renal cell carcinoma (ccRCC) from small renal masses (SRM). Methods We retrospectively analyzed the clinical data and MR images of patients with pathologically confirmed solid SRM from the First Medical Center of the Chinese PLA General Hospital between January 1, 2018, and December 31, 2021, and from Beijing Friendship Hospital of Capital Medical University and Peking University First Hospital between January 1, 2019 and May 17, 2021. Six abdominal radiologists were trained for use of the ccLS algorithm and scored independently using ccLS v1.0 and ccLS v2.0. Random- effects logistic regression modeling was used to generate plot receiver operating characteristic curves (ROC) to evaluate the diagnostic performance of ccLS v1.0 and ccLS v2.0 for ccRCC, and the area under curve (AUC) of these two scoring systems were compared using the DeLong's test. Weighted Kappa test was used to evaluate the interobserver agreement of the ccLS score, and differences in the weighted Kappa coefficients was compared using the Gwet consistency coefficient. Results In total, 691 patients (491 males, 200 females; mean age, 54 ± 12 years) with 700 renal masses were included in this study. The pooled accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of ccLS v1.0 for diagnosing ccRCC were 77.1%, 76.8%, 77.7%, 90.2%, and 55.7%, as compared with 80.9%, 79.3%, 85.1%, 93.4%, 60.6% with ccLS v2.0, respectively. The AUC of ccLS v2.0 was significantly higher than that of ccLS v1.0 for diagnosis of ccRCC (0.897 vs 0.859; P<0.01). The interobserver agreement did not differ significantly between ccLS v1.0 and ccLS v2.0 (0.56 vs 0.60; P>0.05). Conclusion ccLS v2.0 has better performance for diagnosing ccRCC than ccLS v1.0 and can be considered for use to assist radiologists with their routine diagnostic tasks.

Key words: multi-parameter magnetic resonance imaging, small renal masses, Clear Cell Likelihood Score, clear cell renal cell carcinoma