Bladder MRI quality score was linked to better VI-RADS accuracy in small tumours

Design

Retrospective study with three independent readers

Population

587 patients who had multiparametric bladder MRI before surgery

Primary outcome

Inter-reader agreement of VI-RADS quality score and diagnostic performance for muscle invasion

Effect

Tumours under 3 cm: PPV 51.0% vs 22.7% (p = 0.004); accuracy 81.6% vs 71.7% (p = 0.023); kappa 0.628 to 0.702

This retrospective study included 587 patients who had multiparametric bladder MRI before surgery. Three radiologists, two genitourinary specialists and one non-specialist, independently assigned VI-RADS categories and quality scores. About 1.2% of scans scored 0, 48.4% scored 1 and 50.4% scored 2 on quality. Agreement on the final quality score was substantial (weighted kappa 0.628 to 0.702).

Higher quality scores were associated with higher T2-weighted VI-RADS scores (OR 2.95, 95% CI 1.17 to 7.43). In tumours smaller than 3 cm, particularly for the less experienced reader, higher quality was linked to higher positive predictive value (51.0% vs 22.7%, p = 0.004) and accuracy (81.6% vs 71.7%, p = 0.023) at a VI-RADS threshold of 3 or more. In tumours of 3 cm or more there was no significant difference.

Only one less experienced reader took part, so the authors say the finding needs external validation. It supports recording the quality score with every VI-RADS report and considering a repeat scan when quality is poor and the tumour is small.

• Record the VI-RADS quality score with each report.
• Quality mattered most for tumours under 3 cm and for the less experienced reader.
• In small tumours at VI-RADS 3 or more, positive predictive value was 51.0% with higher quality vs 22.7%.
• Consider a repeat or second read when quality is poor and the tumour is small.
• Agreement on the quality score was substantial among readers.

Why it matters

It turns image quality from a technical footnote into something that changes how far a VI-RADS call can be trusted.

Don't overread it

This is a retrospective, single-reader-comparison study that needs external validation.

The statistics, in plain English

Positive predictive value is the proportion of positive calls that were truly muscle-invasive. A rise from 22.7% to 51.0% means far fewer false alarms on good scans, but the comparison comes from a small subgroup and one less experienced reader, so the exact figures are uncertain.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.