Impact of retrospective data verification to prepare the ICON6 trial for use in a marketing authorization application
Bibliographic record
Abstract
BACKGROUND: The ICON6 trial (ISRCTN68510403) is a phase III academic-led, international, randomized, three-arm, double-blind, placebo-controlled trial of the addition of cediranib to chemotherapy in recurrent ovarian cancer. It investigated the use of placebo during chemotherapy and maintenance (arm A), cediranib alongside chemotherapy followed by placebo maintenance (arm B) and cediranib throughout both periods (arm C). Results of the primary comparison showed a meaningful gain in progression-free survival (time to progression or death from any cause) when comparing arm A (placebo) with arm C (cediranib). As a consequence of the positive results, AstraZeneca was engaged with the Medical Research Council trials unit to discuss regulatory submission using ICON6 as the single pivotal trial. METHODS: A relatively limited level of on-site monitoring, single data entry and investigator's local evaluation of progression were used on trial. In order to submit a license application, it was decided that (a) extensive retrospective source data verification of medical records against case report forms should be performed, (b) further quality control checks for accuracy of data entry should be performed and (c) blinded independent central review of images used to define progression should be undertaken. To assess the value of these extra activities, we summarize the impact on both efficacy and safety outcomes. RESULTS: Data point changes were minimal; those key to the primary results had a 0.47% error rate (36/7686), and supporting data points had a 0.18% error rate (109/59,261). The impact of the source data verification and quality control processes were analyzed jointly. The conclusion drawn for the primary outcome measure of progression-free survival between arm A and arm C was unchanged. The log-rank test p-value changed only at the sixth decimal place, the hazard ratio does not change from 0.57 with the exception of a marginal change in its upper bound (0.74-0.73) and the median progression-free survival benefit from arm C remained at 2.4 months. Separately, the blinded independent central review of progression scans was performed as a sensitivity analysis. Estimates and p values varied slightly but overall demonstrated a difference in arms, which is consistent with the initial result. Some increases in toxicity were observed, though these were generally minor, with the exception of hypertension. However, none of these increases were systematically biased toward one arm. CONCLUSION: The conduct of this pragmatic, academic-sponsored trial was sufficient given the robustness of the results, shown by the results remaining largely unchanged following retrospective verification despite not being designed for use in a marketing authorization. The burden of such comprehensive retrospective effort required to ensure the results of ICON6 were acceptable to regulators is difficult to justify.
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
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | Metaresearch Domain: Reproducibility · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Observational | high |
| gpt | Metaresearch Domain: Reproducibility · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Other design | high |
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.016 | 0.036 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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, unvalidatedLabeled directly by 2 models reading the full record.
The models disagree on parts of this classification; every voice is preserved in the section at the end of the page.
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".