Reanalysis of Urothelial Cancer Chemoimmunotherapy Trials With Differential Censoring
Bibliographic record
Abstract
Importance: Three similar phase 3 randomized clinical trials have investigated PD-1/PD-L1 (programmed cell death 1 protein/programmed cell death 1 ligand 1) inhibitors in combination with platinum-based chemotherapy vs chemotherapy alone as first-line treatment for advanced urothelial carcinoma (IMvigor130, atezolizumab; KEYNOTE-361, pembrolizumab; and CheckMate901, nivolumab). Only CheckMate901 reported overall survival (OS) benefit for the combination. The reason for these inconsistent results is unclear. Objective: To explore whether differential censoring-that is, censoring imbalance between the study groups-is a possible explanation for these inconsistent findings. Design, Setting, and Participants: This comparative effectiveness study involved a censoring analysis of data from IMvigor130, KEYNOTE-361, and CheckMate901, which enrolled patients between 2016 and 2022. Participants included patients in these 3 trials. Exposure: Participation in 1 of the 3 trials. Main Outcomes and Measures: The primary outcomes were censoring rates adjusted for treatment effects. Censoring rates were calculated from the Kaplan-Meier (KM) curves. When excess censoring in the control group of open-label trials was found, the hypothesis was that better-performing patients might be dropping out to seek alternative treatments; a sensitivity analysis was conducted in which their survival was assumed to be similar to that of the longest surviving patients in the control group. Treatment effects of the censoring-adjusted KM curves were calculated using the 2-sided log-rank test. Results: The 3 trials involved a total of 2162 patients (1640 male [76%]; age range, 65-69 years) Analysis of progression-free survival (PFS) curves demonstrated no differential censoring in IMvigor130, but there was more than 30% excess censoring in the chemotherapy-only groups in KEYNOTE-361 and CheckMate901 trials. After sensitivity analysis, the PFS benefit was no longer significant in either study (KEYNOTE-361, adjusted hazard ratio [HR], 1.13 [95% CI, 0.95-1.35]; CheckMate901, adjusted HR, 1.17 [0.96-1.44]). Analysis of OS curves demonstrated no differential censoring in IMvigor130 or KEYNOTE-361, but there was more censoring in the chemotherapy-only group in CheckMate901. After sensitivity analysis, the OS benefit of adding nivolumab to chemotherapy was lost (before adjustment, HR, 0.77 [95% CI, 0.63-0.95]; P = .01; adjusted HR, 0.95 [95% CI, 0.77-1.17]; P = .64). Conclusions and Relevance: In this comparative effectiveness study, differential censoring explained the inconsistent results reported in the evaluated trials. The term perceived-inferiority censoring is suggested to describe a phenomenon wherein better-performing patients are aware of their treatment and drop out to pursue alternative therapeutic options; it is possible that this occurred in the open-label KEYNOTE-361 and CheckMate901 trials. Such censoring confounds randomization and interpretation of clinical trials, since a larger experimental group is compared with a selected group of controls with poorer prognosis.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.176 | 0.347 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.004 | 0.017 |
| Bibliometrics | 0.006 | 0.005 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.012 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".