252P Impact of informative censoring on the interpretation of progression-free survival (PFS) estimates in phase III randomized trials (RCT) in metastatic breast cancer (MBC)
Bibliographic record
Abstract
PFS is a commonly used primary end point in oncology trials and can be susceptible to potential bias when informative censoring occurs frequently. Here we investigate the effect of informative censoring on the treatment effect in metastatic breast cancer trials. We reviewed phase III RCTs of new therapies in MBC published between the years 2016-2023. We defined informative censoring as patients censored for any reason other than completion of follow (e.g. this included treatment discontinuation for adverse events or withdrawal of consent) Based on the number at risk we estimated hazard ratios (HRs) for time to treatment failure (TTF) where discontinuation of treatment for any reason is considered an event and measured the magnitude of difference between this and the reported HR for PFS. We analyzed 22 phase III randomized controlled trials comprising of 12302 patients. The mean percentage of censored patients was 17.7% (5.5-34) in the control arm and 16.8% (6-29.2) in the experimental arm. The primary reason for censoring was adverse events (AE) with rates of 30% (range: 9.5-64.7) in the control arm, and 40% (range: 14.7-64) in the experimental arm. Among 10 trials exhibiting a greater than 5% difference in censoring between arms; 4 showed more censoring in the control group (mean 12.7%) and 6 showed more censoring in the experimental group (mean 8.1%). The mean difference between the estimated HR for TTF and the reported HR for PFS was 0.1, ranging from (-0.02 to +0.27). These relative differences translated to shorter estimates for TTF compared with PFS with the difference larger in the experimental arm (2.8 months; range: 0.5-6.8) than in the control arm (1.7 months; range 0.2-3.7). Meaningful differences in censoring between arms were observed in almost half of RCTs evaluated. These resulted in, substantial differences between HR for PFS and TTF suggest that treatment effect may be less pronounced than expected. This effect seemed most marked for absolute effects at the median with large differences between the reported median PFS and the estimated median TTF.
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 | 0.544 | 0.764 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.004 | 0.012 |
| Bibliometrics | 0.006 | 0.007 |
| Science and technology studies | 0.001 | 0.005 |
| Scholarly communication | 0.005 | 0.003 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.003 | 0.004 |
| Insufficient payload (model declined to judge) | 0.005 | 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; the direct Gemma label and the distilled Codex classifier agree on what is shown here.
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".