Diagnostic accuracy of magnetic resonance imaging for tumour staging of bladder cancer: systematic review and meta-analysis.

ABSTRACT

The purpose of this study is to evaluate accuracy of magnetic resonance imaging (MRI) for local staging of bladder cancer for four clinical scenarios (T-stage thresholds) considered against current standards for clinical staging and secondarily to identify sources for variability in accuracy. Systematic review of patients with bladder cancer undergoing T-staging MRI to evaluate the diagnostic accuracy using bivariate random-effects meta-analysis. Sub-group analysis was done to explore variability; risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2 tool. The search identified 30 studies (5156 patients). Pooled accuracy at multiple T-stage thresholds ≤T1 vs ≥T2 = sensitivity 87% (95% confidence interval [CI] 82-91), specificity 79% (95% CI 72-85); T-any vs T0 = sensitivity 65% (95% CI 23-92), specificity 90% (95% CI 83-94); ≤T2 vs ≥T3 = sensitivity 83% (95% CI 75-88), specificity 87% (95% CI 78-93); and sensitivity 85% (95% CI 63-95), specificity 98% (95% CI 95-99). For ≤T1 vs ≥T2, accuracy was higher in studies at low risk of bias. No variability in accuracy was identified for field strength, transurethral resection of bladder tumour status, publication date, index test parameters. For ≤T1 vs ≥T2, accuracy was higher than reported for clinical staging. For T-any vs T0 accuracy was lower than clinical staging. For ≤T2 vs ≥T3, sensitivity was slightly lower than clinical staging but specificity was considerably higher. For sensitivity exceeded the estimated accuracy for clinical staging.

Limitations:

two scenarios had few studies (T-any vs T0; risk of bias. MRI staging for ≤T1 vs ≥T2, ≤T2 vs ≥T3, and risk of bias and sample size; validation in trials comparing clinical staging strategies vs MRI is warranted.