Across 890 Phase III oncology trial records, endpoint selection is dominated by time-to-event efficacy measures. Progression-free survival and related variants form the largest primary endpoint family, while overall survival is used more selectively as a dominant or co-primary endpoint in high-mortality and high-consequence comparative settings.
PFS-family endpoints are the largest dominant primary endpoint family, appearing in 389/890 trials (43.7%). OS is dominant in 176/890 trials (19.8%), while DFS/EFS/iDFS/RFS/MFS-style event endpoints account for 154/890 trials (17.3%).
Phase III oncology endpoint design is primarily built around survival, progression, and recurrence rather than response alone. Response-based primary endpoints remain a minority and appear mainly where response depth, pathological clearance, or residual disease status is clinically central.
PFS remains the most frequent dominant primary endpoint family across all three authorization years: 276/610 trials in 2024 (45.2%), 81/197 trials in 2025 (41.1%), and 32/83 trials in 2026 (38.6%). OS was dominant in 120/610 trials in 2024, 43/197 in 2025, and 13/83 in 2026.
| Year | PFS family | OS | DFS/EFS family | Response / other |
|---|---|---|---|---|
| 2024 | 45.2% | 19.7% | 16.9% | 18.2% |
| 2025 | 41.1% | 21.8% | 19.3% | 17.8% |
| 2026 | 38.6% | 15.7% | 15.7% | 30.1% |
The observed year pattern does not show OS replacing PFS as the default Phase III oncology primary endpoint. Instead, the hierarchy remains stable: PFS for earlier efficacy readout, OS for selected high-stakes settings, and DFS/EFS-family endpoints for curative-intent disease.
Endpoint choice tracks clinical setting. Among advanced or metastatic solid tumor records, 314/548 trials (57.3%) used PFS-family endpoints as the dominant primary endpoint. In adjuvant, neoadjuvant, or curative-intent settings, 111/184 trials (60.3%) used DFS/EFS/iDFS/RFS, pCR, or event-based curative endpoints.
Endpoint selection reflects the clinical question being tested: controlling progression in measurable metastatic disease, preventing recurrence in curative-intent trials, demonstrating survival in near-term mortality settings, and capturing depth of response in hematologic or neoadjuvant disease.
The secondary endpoint layer is broader than the primary endpoint layer. OS appears as secondary evidence in 654/890 trials (73.5%), safety or adverse-event endpoints in 617/890 (69.3%), response/depth-of-response endpoints in 566/890 (63.6%), and HRQoL or PRO endpoints in 321/890 (36.1%).
OS remains embedded in most evidentiary packages even when it is not the primary endpoint. Safety, response, imaging review, HRQoL, biomarkers, and PK/PD endpoints provide the supporting context around the primary efficacy claim.
Most trials remain single-primary, but co-primary designs are structurally important: 686/890 trials (77.1%) listed one primary endpoint, 148/890 trials (16.6%) listed two, and 56/890 trials (6.3%) listed three or more.
Co-primary structures appear where one endpoint does not fully capture the trial question: OS plus PFS in advanced disease, EFS plus pCR in neoadjuvant breast cancer, or response plus PFS/MRD in hematologic malignancies.
Endpoint abbreviations were normalized from trial wording. Definitions below reflect the common measurement patterns observed across the analyzed Phase III oncology records, including RECIST 1.1, Lugano, RANO, IMWG, iwCLL, STEEP 2.0, investigator assessment, blinded independent central review, and CTCAE safety grading.
The same abbreviation can represent different operational definitions by disease. PFS in solid tumors is often RECIST-based, PFS in lymphoma is commonly Lugano-based, PFS in glioma may use RANO, and DFS/EFS endpoints in breast cancer frequently follow STEEP-style recurrence or invasive-event definitions.
Endpoint selection in Phase III oncology is not random variation; it maps to clinical context. PFS is the broad default in measurable advanced disease, OS is selectively used when mortality is decisive and follow-up is feasible, DFS/EFS-family endpoints dominate curative-intent settings, and response/MRD endpoints cluster where tumor clearance or response depth carries major clinical meaning.
The observed endpoint architecture suggests that Phase III oncology programs are increasingly multi-layered: the primary endpoint defines the confirmatory claim, while OS, safety, response, imaging review, HRQoL, biomarker, and PK/PD endpoints build the broader evidentiary package around that claim.