Frequently asked questions on surrogate endpoints in oncology-opportunities, pitfalls, and the way forward
Bibliographic record
Abstract
Patients with cancer expect prolonged life (overall survival, OS) or better life (quality of life, QOL) from cancer treatments. However, majority of new cancer drugs are now being approved not based on improved OS or QOL, but based on surrogate endpoints such as tumor shrinkage or delayed tumor progression. These surrogate endpoints, including their validity as a proxy for overall survival, differ based on disease settings and lines of treatment but in general, most surrogate measures have weak correlation with outcomes that matter to patients. Nevertheless, they are being increasingly used as the basis for regulatory decisions. The current tension in this space is between adoption of surrogate endpoints for early access to cancer drugs versus the need for confirmation that the drugs improve outcomes that matter and not merely improve scan results or surrogate endpoints. In this article, we provide a comprehensive review of surrogate endpoints currently used in oncology trials both in curative and advanced disease settings, including their definition, methodology for validation, existing evidence for their surrogacy, predictive versus prognostic reliability of surrogate endpoints, the promise of surrogate endpoints, their pitfalls, and the way forward. Using a Q&A format, we discuss answers to the most commonly asked questions regarding surrogate endpoints in oncology. Our review answers the following frequently asked questions about surrogate endpoints in oncology: how are surrogate endpoints defined? How are they validated? What is a patient-level versus trial-level surrogate? What are the benefits of using surrogate endpoints? What level of surrogacy is required for regular versus accelerated approval? Are we overusing surrogate endpoints? Should we use surrogate endpoints in adjuvant settings? Can surrogacy be extrapolated from one setting to another? What is the surrogacy of progression-free survival for OS and QOL? Why does PFS not correlated with OS or QOL? Why do regulators rely on surrogate endpoints? We end this article with a proposal on the way forward.
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 | 0.197 | 0.444 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.004 | 0.004 |
| Bibliometrics | 0.005 | 0.007 |
| Science and technology studies | 0.002 | 0.013 |
| Scholarly communication | 0.009 | 0.020 |
| Open science | 0.003 | 0.007 |
| Research integrity | 0.011 | 0.015 |
| Insufficient payload (model declined to judge) | 0.008 | 0.002 |
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".