Visual MRI T-category versus VI-RADS evaluation from multiparametric MRI in the detection of muscle-invasion in patients with suspected bladder cancer: single centre registered clinical trial (MIB-trial)

Abstract

Background

Multiparametric Magnetic Resonance Imaging (mpMRI) has been proposed to add value in the diagnostic pathway of bladder cancer (BC). We wanted to evaluate the performance of mpMRI for muscle-invasion detection in BC patients using a subjective MRI visual T-category and the Vesical Imaging-Reporting and Data System (VI-RADS) score.

Methods

This single centre clinical trial included 45 patients with suspected BC (ClinicalTrials.gov Identifier: NCT02662166). All patients had mpMRI prior to transurethral resection of bladder tumour (TUR-BT). The imaging was correlated to histopathological findings. Two individual radiologists evaluated all the mpMRI images. A binary cut-off point for the detection of muscle-invasion in the MRI visual T-category was defined between T1 and T2 and the VI-RADS cut-off score was 3. Cohen’s Kappa values were used to evaluate the agreement between the two radiologists. Sensitivity, Specificity, Area Under Receiver Operator Characteristics Curve (AUC), Positive Predictive Value (PPV) and Negative Predictive Value (NPV) were calculated to evaluate the performance of both radiologists separately.

Results

AUC values for reader A and B using the MRI visual T-category were 0.76 and 0.56, while the corresponding values for VI-RADS were 0.63 and 0.57, respectively. There was no statistically significant difference between the radiologists nor the reporting systems (p > .05) in the detection of muscle-invasion. The inter-reader agreement was substantial (0.61–0.80).

Conclusion

Both the subjective MRI visual T-category and VI-RADS score had only a low to moderate accuracy for the detection of muscle-invasion in BC with no statistically significant difference between the reporting systems.

Keywords:

• Bladder cancer
• multiparametric MRI
• T-stage
• single centre clinical trial
• VI-RADS

Acknowledgements

The corresponding author would like to thank Jarkko Alajääski, PhD. for reviewing statistical analyses used in the current manuscript.

Disclosure statement

The last author, Peter J. Boström, is an associate editor of Scandinavian Journal of Urology.

Additional information

Funding

The article was partly funded by grants from the Finnish Urology Association (Suomen Urologiyhdistys ry) and the Turku University Hospital Research Foundation (Turun yliopistollisen keskussairaalan koulutus- ja tutkimussäätiö).