Findings from a study presented Thursday suggest that the use of biparametric MRI (bpMRI), based on Vesical Imaging-Reporting and Data System (VI-RADS) guidelines, may lead to misdiagnosis of muscle invasion in variant urothelial carcinoma (UC) of the bladder.
Arita
"While bpMRI, which excludes the dynamic contrast-enhanced (DCE) sequence, offers advantages such as reduced scanning times, cost-effectiveness, and the elimination of potential contrast agent-related complications, our study findings suggest that a complete multiparametric MRI (mpMRI), including DCE-MRI, remains the preferred modality for accurately evaluating preoperative muscle invasion in cases of variant UCs," said Yuki Arita, MD, PhD, a research fellow in the Department of Radiology at Memorial Sloan Kettering Cancer Center in New York.
Dr. Arita, who earned an RSNA 2023 Trainee Research Prize for the study, also serves as assistant professor in the Department of Radiology at Keio University School of Medicine in Japan. "An early suspicion of variant UCs should lead to the recommendation of a comprehensive mpMRI assessment, inclusive of DCE-MRI," he said. "This approach is vital to mitigate the risk of underestimating muscle invasion in bladder cancers, especially when considering the possibility of foregoing a second transurethral resection of the bladder."
According to Dr. Arita, since the establishment of VI-RADS, a multitude of studies have validated its high diagnostic accuracy and inter-reader consistency, suggesting its broader future application. "Nonetheless, in my experience as a body MRI fellow, I noted cases where DWI, the primary sequence in VI-RADS, appeared to underrepresent T staging in certain bladder cancer patients," he said.
Dr. Arita and his colleagues aimed to assess and compare the diagnostic accuracy of bpMRI against conventional mpMRI in evaluating muscle invasion in variant UCs. He noted that given the rarity of variant UCs, a multicentric approach to the study was necessary. "Coordinating regular meetings across centers to establish a unified research plan, obtaining IRB approvals, and compiling imaging and clinical data from each institution presented significant challenges," he said. "Additionally, securing accurate pathological assessments of muscle invasion status for each case was critical yet challenging."
Subgroup Analysis Needed to Evaluate Diagnostic Accuracy of DWI in Variant UCs
The researchers performed a retrospective analysis of 118 treatment-naïve patients with pathologically proven variant UCs. The patients underwent bladder mpMRI before transurethral bladder tumor resection between 2010 and 2019. Two sets of images—bpMRI and mpMRI—were independently reviewed by three board-certified radiologists using VI-RADS.
Among the patients, nearly 56% had pathologically confirmed muscle-invasive bladder cancer (MIBC), and just over 44% had non-MIBC.
Using a VI-RADS cut-off score of 4, sensitivity of bpMRI was significantly lower than that of mpMRI for all readers, while no significant differences were observed in specificity. At a VI-RADS cut-off score of 3, no significant differences were found in sensitivity between bpMRI and mpMRI for all readers, but the specificity of bpMRI was significantly lower for one reader.
The diagnostic performance of contrast-free bpMRI, based on VI-RADS, was inferior to that of mpMRI in assessing muscle invasion in patients with variant UCs.
"The diagnostic performance of DWI exhibited notable variability across different subtypes of variant UCs. This variability manifested in the accurate staging of some variant UCs, while others were consistently underestimated," Dr. Arita said. "This observation necessitated a more granular subgroup analysis to evaluate the diagnostic accuracy across diverse UC subtypes. The comprehensive results and insights derived from this analysis are elaborated in our published research."
Access the presentation, "Biparametric v. Multiparametric MRI for Assessing Muscle Invasion in Bladder Urothelial Carcinoma with Variant Histology using the Vesical Imaging-Reporting and Data System: A Multi-institutional Multi-reader Study," (R1-SSGU06-1) on demand at Meeting.RSNA.org.
