Objective: The purpose of this study was to assess whether use of the PACS cross-referencing tool in 3D MRI improves tumor staging of prostate cancer when pathologic findings are used as the reference standard.
Materials and methods: The institutional review board granted a waiver of informed consent for the study. Endorectal MRI at 1.5 T was performed before radical prostatectomy in 255 consecutive patients. Two radiologists unaware of the clinical data retrospectively and independently interpreted MR images without and with cross-referencing to predict the presence of extracapsular extension (ECE) and seminal vesicle invasion (SVI). Histopathologic findings were used as the reference standard. Area under the receiver operating characteristics curve (AUC), sensitivity and specificity, and weighted kappa statistics were calculated.
Results: At histologic examination, 68 (27%) of the patients were found to have ECE and 13 (5%) of the patients to have SVI; the latter all had ECE. In detecting ECE, both reviewers had a higher AUC using cross-referencing (p < 0.001 for both). The weighted kappa value was 0.56 for MRI alone and 0.76 for MRI with cross-referencing, indicating fair to good interobserver agreement. Sensitivity and specificity for ECE with MRI alone and with cross-referencing were 43% and 94% and 57% and 100% for reviewer 1 and 40% and 93% and 59% and 98% for reviewer 2, respectively. In detecting SVI, both reviewers had a higher AUC with cross-referencing (p = 0.007 and p = 0.056 for reviewers 1 and 2, respectively). Reviewer 1 benefited much more from cross-referencing than did reviewer 2. The weighted kappa statistic was 0.69 for MRI alone and the same with cross-referencing, indicating good interobserver agreement. Sensitivity and specificity for SVI with MRI alone and with cross-referencing, respectively, were 23% and 83% and 46% and 93% for reviewer 1 and 31% and 91% and 54% and 95% for reviewer 2.
Conclusion: PACS cross-referencing significantly improves tumor staging of prostate cancer with 3D MRI. Some reviewers benefit more than others from use of this tool.