Abstract
Xpert Bladder Cancer (BC) is an mRNA-based urine assay for diagnosis and surveillance of urothelial carcinoma. Although previous meta-analyses have examined Xpert, none has evaluated its accuracy in both BC detection and non-muscle-invasive bladder cancer (NMIBC) surveillance, while also exploring its potential role in guiding repeat transurethral resection (re-TURBT) and in the diagnostic evaluation of upper tract urothelial carcinoma (UTUC). We systematically assessed the diagnostic accuracy of Xpert in all these clinical settings, integrating grade-specific analysis and formal evidence-certainty assessment. PubMed/MEDLINE, Embase, Scopus, Web of Science, and the Cochrane Library were searched to January 2026 (PROSPERO: CRD420251184275). Sensitivity and specificity were pooled using bivariate random-effects models; PPV and NPV using univariate models. Grade-specific sensitivity was pooled across 13 studies with tumour-grade-stratified data. Evidence certainty was assessed using GRADE-DTA. Publication bias was evaluated with Deeks' funnel plots. Twenty-six studies (8613 patients) were included. For Detection (k = 8, where k denotes the number of cohorts pooled), pooled sensitivity was 0.83 (95% CI 0.73-0.90) and specificity 0.77 (0.65-0.86); NPV was 0.96. For Monitor (k = 15), sensitivity was 0.71 (0.63-0.77) and specificity 0.78 (0.70-0.85); NPV was 0.91. Grade-specific pooling showed high-grade (HG) sensitivity of 0.89 (0.83-0.93) in Detection and 0.80 (0.71-0.87) in Monitor, versus 0.60 for low-grade (LG) in both settings; heterogeneity was not statistically significant for HG estimates. Re-TURBT (k = 2) and UTUC (k = 2) were exploratory, with encouraging NPV but limited by sparse events. Evidence certainty was low for sensitivity/NPV and very low for specificity/PPV. No publication bias was detected. Xpert demonstrates high sensitivity and NPV, particularly for HG disease, supporting its role as an adjunctive rule-out tool. Grade-adapted interpretation is warranted. Prospective validation with standardised grade-specific reporting is needed.