Early radical cystectomy in high-risk non-muscle-invasive bladder cancer: a systematic review and meta-analysis

High-risk non-muscle-invasive bladder cancer (HR-NMIBC) carries substantial risks of recurrence and progression. The comparative oncological effect of radical cystectomy (RC) performed early in the disease course versus bladder-preserving or delayed surgical strategies remains uncertain because the literature spans different BCG states, surgical decision points, and treatment eras. PubMed, Cochrane Library, Web of Science, and Scopus were searched from inception through May 2025. Eligibility was re-audited against the prespecified PICOS criteria. To restore independence, each study contributed only one estimate per outcome. For fixed-time analyses, 5 years was selected before the revised re-analysis as the single clinically relevant time point used most consistently across studies. Events were defined as all-cause death for OS, cancer-specific death for CSS, and disease progression for PFS; odds ratios (ORs) < 1 therefore favor early RC for all three analyses. Random-effects models were prioritized. Adjusted hazard ratios (HRs) were synthesized separately where sufficiently comparable. Risk of bias was assessed with ROBINS-I or RoB 2, and certainty was evaluated using GRADE. After eligibility re-audit, 18 primary clinical studies comprising 11,763 unique participants were retained in the systematic review, and 12 contributed to at least one quantitative synthesis. At 5 years, OS included 6 independent studies ( n = 9,800) and showed no statistically significant difference under random effects (OR 0.85, 95% CI 0.46–1.58; I²=96.2%). CSS included 11 independent studies ( n = 6,358); the event was cancer-specific death, and the random-effects estimate was also non-significant (OR 1.22, 95% CI 0.72–2.06; I²=85.2%). PFS included 2 studies ( n = 185) and was non-significant (OR 1.59, 95% CI 0.76–3.32; I²≈0%). An adjusted-HR sensitivity analysis for OS (3 studies) was consistent with the primary result (HR 0.82, 95% CI 0.45–1.49). Sensitivity analyses excluding critical-risk studies did not produce statistically significant pooled effects. Certainty of evidence was very low for all outcomes. The rebuilt quantitative synthesis does not establish a consistent OS, CSS, or PFS advantage for early RC. The evidence is very uncertain because of substantial clinical heterogeneity, observational confounding, high risk of bias, and imprecision. Treatment decisions should therefore remain individualized according to tumor risk, BCG exposure and response, surgical fitness, and patient preferences.