Structured expert elicitation to inform long-term survival extrapolations using alternative parametric distributions: a case study of CAR T therapy for relapsed/ refractory multiple myeloma
Bibliographic record
Abstract
BACKGROUND: Our aim was to extend traditional parametric models used to extrapolate survival in cost-effectiveness analyses (CEAs) by integrating individual-level patient data (IPD) from a clinical trial with estimates from experts regarding long-term survival. This was illustrated using a case study evaluating survival of patients with triple-class exposed relapsed/refractory multiple myeloma treated with the chimeric antigen receptor (CAR) T cell therapy idecabtagene vicleucel (ide-cel, bb2121) in KarMMa (a phase 2, single-arm trial). METHODS: The distribution of patients expected to be alive at 3, 5, and 10 years given the observed survival from KarMMa (13.3 months of follow-up) was elicited from 6 experts using the SHeffield ELicitation Framework. Quantities of interest were elicited from each expert individually, which informed the consensus elicitation including all experts. Estimates for each time point were assumed to follow a truncated normal distribution. These distributions were incorporated into survival models, which constrained the expected survival based on standard survival distributions informed by IPD from KarMMa. RESULTS: Models for ide-cel that combined KarMMa data with expert opinion were more consistent in terms of survival as well as mean survival at 10 years (survival point estimates under different parametric models were 29-33% at 3 years, 5-17% at 5 years, and 0-6% at 10 years) versus models with KarMMa data alone (11-39% at 3 years, 0-25% at 5 years, and 0-11% at 10 years). CONCLUSION: This case study demonstrates a transparent approach to integrate IPD from trials with expert opinion using traditional parametric distributions to ensure long-term survival extrapolations are clinically plausible.
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 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 | 0.090 | 0.185 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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, unvalidatedMachine predicted; both teacher heads 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".