Trimodal treatment vs radical cystectomy for muscle-invasive bladder cancer: The neglected impact of informative censoring. A systematic review and meta-analysis.

INTRODUCTION

Informative censoring (IC) is a statistical bias that occurs when there is an imbalance in the dropout rate among study arms and this imbalance is not random, but can be ascribed to information that is not accounted for. Studies on trimodal therapy (TMT) versus radical cystectomy (RC) yielded conflicting results. We conducted a meta-analysis to assess the potential impact of informative censoring in influencing study outcomes.

MATERIALS AND

METHODS

A systematic literature search identified studies involving patients with cT2-4 any N, M0 muscle-invasive bladder cancer treated with TMT versus RC to assess the risk of informative censoring in assessing oncological measures. We reconstructed published Kaplan-Meier curves to obtain time-to-event data and applied the inverse Kaplan-Meier method alongside the log-rank test to evaluate the presence of informative censoring. Finally, we performed a simulation analysis to assess the minimum proportion of events required among censored patients to compensate for the potential effect of IC.

RESULTS

Overall, 6 studies with 8594 patients were included; 4 used propensity score matching, while 2 did not. An imbalance in censoring was present in 10 of 12 reported outcomes, with bias favoring RC in 5 instances and TMT in 5 instances. Overall, censoring favored RC in overall survival (OS), metastasis-free survival, and disease-free survival (Log-rank: all p < 0.01). These findings were consistent after stratifying for propensity score matching and neoadjuvant chemotherapy. For OS, the minimum proportion of events required compensate for the potential effect of IC ranged from 16 % to 33 %. After adjusting for potential presence of IC, RC was still superior in terms of OS relative to TMT (5-year OS: 42 % vs. 30 %; p < 0.001).

CONCLUSIONS

Most retrospective studies comparing TMT to RC are at high risk of informative censoring. IC tends to favor radical cystectomy. After correcting for IC, the oncological outcomes of patients undergoing RC were still superior to the ones who undergo TMT. Our data underline the complexity of comparing oncological outcomes between TMT and RC, and challenge the acceptance of TMT based on retrospective comparison in spite of randomization.