When does a patient-derived tumor model become reliable enough to guide a clinical research decision?
Representative model collections and drug-response assays solve different validation problems; neither model fidelity nor assay agreement alone establishes clinical utility.
Model collections and orthogonal assays are becoming more auditable, while prospective agreement with patient outcomes remains largely unresolved.
- Who encounters this friction
- Precision-oncology researchers, translational pharmacology teams, and trial-design groups.
A pediatric program built 388 patient-derived xenograft models across more than 40 diagnoses, and a separate study compared short-term viability with longer-term imaging across molecularly characterized cancer models.
A model can closely resemble a source tumor yet still produce a result that arrives too late, fails to reproduce, or does not predict what happens in a patient.
Prospective sampling rules, representation of patients who do not generate models, reproducibility across laboratories, and blinded comparison between model predictions and patient outcomes.
What changed
Newest evidence first. Labels describe the bottleneck, not treatment effectiveness.
-
Mixed
Large model collections and complementary drug-response assays improved auditability, but selection bias and prospective clinical utility remained open.
2 linked primary sources
Primary studies and briefings
2 primary sources. Every interpretation remains bounded by the linked evidence.
Discovery
How to read evidence movement
EmergingA newly supported problem worth tracking.
StrengtheningAdditional evidence reinforces the bottleneck framing.
MixedEvidence supports some parts while important conflicts or limits remain.
UnchangedNew evidence does not materially change the open problem.
WeakenedNew evidence reduces confidence that this is the central bottleneck.
These labels track an unresolved research problem. They do not score a treatment or predict benefit for an individual.