Mullerad Michael, Hricak Hedvig, Wang Liang, Chen Hui-Ni, Kattan Michael W, Scardino Peter T
Department of Urology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA.
Radiology. 2004 Jul;232(1):140-6. doi: 10.1148/radiol.2321031254. Epub 2004 May 27.
To determine whether predictive value of endorectal magnetic resonance (MR) imaging findings in detection of prostate cancer extracapsular extension (ECE) is significantly affected by the reader's subspecialty experience.
In this cohort study, 344 consecutive patients with biopsy-proved prostate cancer underwent endorectal MR imaging followed by surgery. Likelihood of ECE described in MR imaging reports was compared with clinical predictor variables. ECE was determined from the final pathologic report on specimens resected at surgery. Readers of MR images were classified into genitourinary MR imaging radiologists (n = 4) and general body MR imaging radiologists (n = 6). For data analysis, Wilcoxon rank sum and chi(2) tests, as well as receiver operating characteristic (ROC) curves and univariate and multivariate logistic regression analyses, were used. A difference with P <.05 was considered significant.
Univariate analysis results demonstrated that all predictors except clinical stage were significantly associated with detection of ECE in both groups of readers (P <.05). In the genitourinary MR imaging radiologist group of patients, area under the ROC curve for endorectal MR imaging findings (0.833) was larger than areas under the curves for all other predictors (0.566-0.701). In the general body MR imaging radiologist group of patients, area under the ROC curve for endorectal MR imaging findings (0.646) was not larger than areas under the curves for all other predictors (0.582-0.793). Results of multivariate analysis of two models, one with all predictors and another with all predictors except endorectal MR imaging findings, demonstrated a significant increase in area under the ROC curve with endorectal MR images interpreted by genitourinary MR imaging radiologists (P =.019 and.31, respectively).
Endorectal MR imaging findings are significant predictors for detection of ECE when MR images are interpreted by genitourinary radiologists experienced with MR imaging of the prostate.
确定直肠内磁共振（MR）成像结果在检测前列腺癌包膜外侵犯（ECE）中的预测价值是否会受到阅片者亚专业经验的显著影响。
在这项队列研究中，344例经活检证实为前列腺癌的连续患者接受了直肠内MR成像检查，随后进行了手术。将MR成像报告中描述的ECE可能性与临床预测变量进行比较。ECE由手术切除标本的最终病理报告确定。MR图像的阅片者分为泌尿生殖系统MR成像放射科医生（n = 4）和全身MR成像放射科医生（n = 6）。数据分析采用Wilcoxon秩和检验、卡方检验、受试者操作特征（ROC）曲线以及单因素和多因素逻辑回归分析。P <.05的差异被认为具有统计学意义。
单因素分析结果表明，除临床分期外，所有预测因素在两组阅片者中均与ECE的检测显著相关（P <.05）。在泌尿生殖系统MR成像放射科医生组的患者中，直肠内MR成像结果的ROC曲线下面积（0.833）大于所有其他预测因素的曲线下面积（0.566 - 0.701）。在全身MR成像放射科医生组的患者中，直肠内MR成像结果的ROC曲线下面积（0.646）不大于所有其他预测因素的曲线下面积（0.582 - 0.793）。两个模型的多因素分析结果，一个模型包含所有预测因素，另一个模型包含除直肠内MR成像结果外的所有预测因素，结果显示，由泌尿生殖系统MR成像放射科医生解读直肠内MR图像时，ROC曲线下面积显著增加（分别为P =.019和.31）。
当由对前列腺MR成像有经验的泌尿生殖系统放射科医生解读MR图像时，直肠内MR成像结果是检测ECE的重要预测指标。
