Investigators retrospectively reviewed 519 consecutive robot-assisted radical cystectomy (RARC) cases for bladder cancer performed across four Japanese institutions between October 2014 and March 2026. The team focused on perioperative outcomes, major complications defined as Clavien-Dindo grade ≥3, and all-cause deaths within 90 days of surgery. They observed 64 major complications within 90 days, representing 12.3% of the cohort. Short, stark finding: 13 patients (2.5%) died within 90 days. The authors classified six of those deaths (1.2% overall) as complication-related. These are the primary safety signals reported in this real-world, multicenter implementation series.
The study divided each institution’s series into two chronological halves—early and late—to explore how practice evolved over time. Pelvic lymph node dissection (PLND) was performed substantially less often in the late half: the frequency fell from 95.4% in the early half to 83.3% in the late half. Correspondingly, the proportion of pNx cases rose from 4.6% to 16.7%, and the authors attribute much of this shift to changes at a single institution. Tumor burden at surgery also changed: the proportion of patients with pT3-4 disease decreased from 35.8% in the early phase to 23.8% in the late phase. These temporal shifts in both surgical practice and pathologic stage complicate interpretation of outcome trends.
The study team performed logistic regression that included institution to identify factors associated with major complications. The late phase showed a point estimate consistent with fewer major complications but did not reach statistical significance after adjustment: adjusted odds ratio 0.67 (95% CI 0.39–1.14). In other words, the late period trended toward lower odds of major complications but confidence limits crossed unity. The report also noted that patients who died within 90 days were older and had worse performance status compared with survivors. The abstract did not provide additional multivariable coefficients, nor did it report specific comorbidity indices or perioperative details beyond what is summarized here; the source did not report those additional variables in the abstract.
To avoid immortal-time bias and to focus on longer-term outcomes after the immediate perioperative window, investigators performed a landmark analysis among 90-day survivors with Cox models stratified by institution. The late phase was not significantly associated with improved overall survival in that analysis: hazard ratio 0.70 (95% CI 0.45–1.08). The authors reported similar, non-significant point estimates for cancer-specific survival (hazard ratio 0.65 reported in the abstract) and recurrence-free survival (hazard ratio 0.74 reported). Major complications within 90 days were not associated with subsequent survival in the landmark models. Those are the survival signals available in the abstract; the report framed these analyses as exploratory.
This was a retrospective, multicenter series covering a long time span with variation in institutional practice. The authors explicitly warned that the observed differences between early and late halves in tumor characteristics, rates of pelvic lymph node dissection, and follow-up prevent attributing any change in outcomes solely to increased surgical experience. The abstract did not provide granular operative details, standardized measures of institutional volume, or complete multivariable model tables; those items were not reported in the source abstract. Key unanswered clinical questions remain: whether standardized PLND protocols affect morbidity and oncologic outcomes in RARC across institutions, how case selection changed over time, and which perioperative practices—if any—contribute to the observed trends in complications and mortality.
The dataset documents that roughly one in eight patients experienced a Clavien-Dindo grade ≥3 event and that 2.5% of patients died within 90 days of RARC in this real-world, multicenter Japanese cohort. Differences in baseline tumor stage and in the use of PLND across centers mean that these outcome rates should be interpreted as institutional experience snapshots rather than definitive measures of a learning curve effect. The study team labeled several analyses exploratory and highlighted the constraints inherent to retrospective, multicenter comparisons; those limitations are central to interpreting the findings and guide the specific next questions that require prospective or standardized investigation.