Coelho Fernando Morbeck Almeida, Amaral Lucas Tadashi Wada, Mitsutake Leonardo Kenji Nesi, Mussi Thais Caldara, Baroni Ronaldo Hueb
Department of Radiology, Hospital Israelita Albert Einstein. 627 Albert Einstein Ave. Sao Paulo, SP 05652-900, Brazil.
Eur J Radiol. 2023 Aug;165:110921. doi: 10.1016/j.ejrad.2023.110921. Epub 2023 Jun 8.
To evaluate whether the Prostate Imaging Quality (PI-QUAL) score impacts prostate cancer (PCa) staging on MRI. The secondary goal was to test inter-reader agreement among radiologists experienced in prostate imaging.
A retrospective, single-center study with patients who underwent 3 Tesla prostate MRI scans and were submitted to radical prostatectomy (RP) between January 2018 and November 2021 and were eligible for our study. Extraprostatic extension (EPE) data were collected from original MR reports (EPEm) and pathological reports of RP specimens (EPEp). Three expert prostate radiologists (ESUR/ESUI criteria R1, R2, R3) independently evaluated all MRI exams according to PI-QUAL score for image quality (1 to 5; 1: poor, 5: excellent), blinded to original imaging reports and clinical data. We studied the diagnostic performance of MRI using pooled data from PI-QUAL scores (≤3 vs. ≥4). We also performed univariate and multivariate analyses to assess the PI-QUAL score impact on local PCa staging. Cohen's K and Tau-b Kendall tests were used to assess the inter-reader agreement for PI-QUAL score, T2WI, DWI, and DCE.
Our final cohort included 146 patients, of which 27.4% presented EPE on pathology. We observed no impact of imaging quality on accuracy for EPE prediction: AUC of 0.750 (95% CI 0.26-1) for PI-QUAL ≤ 3 and 0.705 (95% CI 0.618-0.793) for PI-QUAL ≥ 4. The multivariate analysis demonstrated a correlation of EPEm (OR 3.25, p 0.001) and ISUP grade group (OR 1.89, p 0.012) to predict EPEp. The inter-reader agreement was moderate to substantial (0.539 for R1-R2, 0.522 for R2-R3, and 0.694 for R1-R3).
Our clinical impact evaluation showed no direct correlation between MRI quality by PI-QUAL score and accuracy in detecting EPE in patients undergoing RP. Additionally, we had moderate to a substantial inter-reader agreement for the PI-QUAL score.
评估前列腺成像质量（PI-QUAL）评分是否会影响前列腺癌（PCa）在MRI上的分期。次要目标是测试在前列腺成像方面经验丰富的放射科医生之间的阅片者间一致性。
一项回顾性单中心研究，研究对象为2018年1月至2021年11月期间接受3特斯拉前列腺MRI扫描并接受根治性前列腺切除术（RP）且符合本研究条件的患者。从原始MR报告（EPEm）和RP标本的病理报告（EPEp）中收集前列腺外侵犯（EPE）数据。三位前列腺影像专家放射科医生（符合欧洲泌尿生殖放射学会/欧洲泌尿外科学会标准的R1、R2、R3）在不了解原始影像报告和临床数据的情况下，根据PI-QUAL评分对图像质量（1至5分；1分：差，5分：优）独立评估所有MRI检查。我们使用来自PI-QUAL评分的汇总数据（≤3分与≥4分）研究MRI的诊断性能。我们还进行了单因素和多因素分析，以评估PI-QUAL评分对局部PCa分期的影响。使用Cohen's K和Tau-b Kendall检验评估阅片者间在PI-QUAL评分、T2WI、DWI和DCE方面的一致性。
我们的最终队列包括146例患者，其中27.4%在病理检查中出现EPE。我们观察到成像质量对EPE预测准确性没有影响：PI-QUAL≤3分时的AUC为0.750（95%CI 0.26 - 1），PI-QUAL≥4分时的AUC为0.705（95%CI 0.618 - 0.793）。多因素分析表明，EPEm（OR 3.25，p 0.001）和国际泌尿病理学会（ISUP）分级组（OR 1.89，p 0.012）与预测EPEp相关。阅片者间一致性为中等至高度（R1 - R2为0.539，R2 - R3为0.522，R1 - R3为0.694）。
我们的临床影响评估表明，PI-QUAL评分所反映的MRI质量与接受RP患者中检测EPE的准确性之间没有直接相关性。此外，我们在PI-QUAL评分方面有中等至高度的阅片者间一致性。
