Oncology Endpoint Overview
Definition
FDA recognizes two broad categories of endpoints for cancer drug approval:
Clinical benefit endpoints (support traditional/regular approval):
- Overall Survival (OS) — "the time from randomization until death from any cause and is measured in the intent-to-treat population." (FDA Cancer Endpoints 2018, §III.A)
- Disease-Free Survival (DFS) — "the time from randomization until disease recurrence or death from any cause." Used primarily in adjuvant settings.
- Event-Free Survival (EFS) — "time from randomization to any of the following events: progression of disease that precludes surgery, local or distant recurrence, or death due to any cause." Used in neoadjuvant settings.
- Symptom improvement / palliation — direct measure of clinical benefit; requires symptomatic population at baseline.
Tumor assessment endpoints (may support traditional or accelerated approval):
- Objective Response Rate (ORR) — "the proportion of patients with tumor size reduction of a predefined amount and for a minimum time period… the sum of partial responses plus CRs." (§III.B.2)
- Complete Response (CR) — "no detectable evidence of tumor." (§III.B.3)
- Progression-Free Survival (PFS) — "the time from randomization until objective tumor progression or death, whichever occurs first." (§III.B.4)
- Duration of Response (DOR) — time from first response to documented tumor progression; accompanies ORR.
- Time to Progression (TTP) — time from randomization until objective tumor progression; does not include deaths. Inferior to PFS; rarely used.
- Time to Treatment Failure (TTF) — composite of time to discontinuation for any reason; "generally not recommended as a regulatory endpoint for new molecular entity drug approval." (§III.B.5)
Regulatory Position
| Endpoint | Traditional Approval | Accelerated Approval | Notes |
|---|---|---|---|
| OS | Yes — gold standard | N/A | Direct clinical benefit; preferred when feasible |
| DFS/EFS | Yes — adjuvant/neoadjuvant settings | Yes — as surrogate | Context-dependent; magnitude and risk-benefit weighed |
| PFS | Yes — large effect + acceptable safety | Yes — as surrogate | Most common primary endpoint in Phase 3 (NSCLC, Ovarian, Myeloma) |
| ORR | Yes — if durable and high CR rate | Yes — most common surrogate | Single-arm studies acceptable for accelerated approval |
| CR | Yes — hematologic malignancies | Yes | For AML and lymphoma traditional approval |
| Symptom endpoints | Yes | No | Must be pre-specified; symptomatic population required |
| TTF | Not recommended | Not recommended | — |
Source: FDA Cancer Endpoints 2018 (Final guidance). "Traditional approval" equates to "regular approval" throughout this KB.
The accelerated approval regulations (21 CFR part 314, subpart H) "allow use of additional endpoints for approval of drugs or biological products for serious or life-threatening illnesses... based on an effect on a surrogate endpoint that is reasonably likely to predict clinical benefit." Post-marketing confirmatory trials are required.
2025 Update (Draft guidance): The August 2025 OS draft adds OS as a mandatory safety endpoint in every randomized approval trial regardless of primary endpoint choice — a new layer beyond the 2018 framework. See Overall Survival (OS).
When to Use
OS as primary endpoint:
- Pancreatic cancer (47/129 trials, ~36%), colorectal cancer (54/300, 18%), melanoma (34/121, 28%)
- Any setting with short post-progression survival (< 12 months expected)
- When crossover is absent or minimal
- GI malignancies where PFS-to-OS correlation is strong
PFS as primary endpoint:
- NSCLC (109/300 trials, 36%), ovarian cancer (76/230, 33%), myeloma (99/256, 39%)
- Targeted therapies with large HR reductions (HR < 0.60)
- Maintenance/consolidation settings (PARP inhibitors, antiangiogenics)
- Indications with long post-progression survival making OS impractical
ORR / CR as primary endpoint:
- AML (CR: 46/489 trials, 9%; ORR: 48/489, 10%) — CR is the standard hematologic endpoint for traditional approval
- Lymphoma (ORR: 27/300, 9%; EFS: 42/300, 14%)
- Accelerated approval across all tumor types when unmet need is high and surrogate validity is plausible
- Single-arm Phase 2/3 trials in heavily pretreated patients
DFS as primary endpoint:
- Breast adjuvant (49/300 trials, 16%), colorectal adjuvant (38/300, 13%)
- Melanoma adjuvant (RFS: 6/121, 5%)
- Settings with high cure fractions where OS follow-up would take > 10 years
EFS as primary endpoint:
- Neoadjuvant breast cancer settings
- Lymphoma (42/300, 14%) — particularly EFS-24 or EFS-12 as early landmark
- AML (16/489, 3%)
Design Considerations
Assessment schedules and criteria
- Solid tumors: RECIST 1.1 for PFS/ORR/DFS; every 6–9 weeks standard for 1L metastatic, every 8–12 weeks for maintenance
- Hematologic malignancies: IMWG criteria (myeloma), Lugano 2014 (lymphoma), IWG/ELN criteria (AML/MDS)
- Assessments must be symmetric between arms; unscheduled assessments introduce bias
IRC requirements
- Required when primary endpoint is a tumor assessment endpoint (PFS, ORR, DFS) in unblinded trials
- IRC must be blinded to treatment assignment; blind maintained throughout recruitment
- Blinded IRC may be unnecessary when: (1) trial is double-blind, or (2) effect size is robust in large trial with sensitivity analyses confirming no investigator bias
Censoring rules (PFS)
- No baseline tumor assessment → censor at Day 1
- No post-baseline assessment before new anti-cancer therapy → censor at Day 1
- New anti-cancer therapy before progression → censor at last tumor assessment prior to new therapy
- Progression ≥2 assessment windows after last adequate assessment → censor at last adequate assessment
- Death without prior progression → count as PFS event
Intercurrent event handling
- Subsequent therapy: Treatment policy (OS) or hypothetical (PFS) are most common
- Treatment discontinuation: Treatment policy in most cases; composite if toxicity-driven discontinuation is informative
- See Intercurrent Events in Oncology Trials and Intercurrent Events in Oncology Trials for full IE tables
Sample size drivers
- OS: typically driven by number of deaths; 300–500 events common in metastatic trials
- PFS: 200–350 events typical; faster to accumulate
- ORR (single-arm): often 30–100 patients; hypothesis test against historical control rate
Intercurrent Events
| IE Type | Most Affected Endpoints | Recommended Strategy | Key Issue |
|---|---|---|---|
| Subsequent anticancer therapy | OS | Treatment policy (primary), hypothetical (sensitivity) | Post-progression therapy dilutes OS benefit |
| Treatment discontinuation | PFS, DFS | Treatment policy | Discontinuation not informative for tumor growth |
| Protocol-specified crossover | OS | Hypothetical + statistical adjustment (RPSFT/TSE/IPCW) | Crossover nullifies OS signal |
| Death before progression | PFS | Composite (count as event) | Standard; PFS includes death |
| AML transformation | EFS (MDS) | Composite (count as event) | Endorsed by FDA 2025 MDS draft |
Regulatory Precedent
From ClinicalTrials.gov Phase 3 oncology dataset (2024 snapshot):
| NCT# | Drug | Indication | Primary Endpoint | Approval Pathway |
|---|---|---|---|---|
| NCT04129502 | TAK-788 (mobocertinib) | 1L metastatic NSCLC EGFR exon 20 | PFS by IRC (RECIST 1.1) | Traditional |
| NCT01828112 | Ceritinib | 2L+ ALK+ NSCLC post-crizotinib | PFS by IRC | Traditional |
| NCT02864251 | Nivolumab + ipilimumab | EGFR-mutant NSCLC post-TKI | OS (co-primary) | Traditional |
Note: Full regulatory precedent tables for each endpoint type are in the endpoint-specific articles.
Limitations and Pitfalls
OS:
- Long follow-up required (often 5–10 years in adjuvant)
- Confounded by subsequent therapies, particularly crossover
- 2025 draft now requires OS collection even when not primary endpoint
PFS:
- Surrogate-to-OS correlation varies widely by indication (strong in CRC/ovarian, weaker in NSCLC/breast)
- Subject to assessment frequency bias (more frequent scans in one arm inflate events)
- PFS benefit does not guarantee OS benefit; FDA may not accept PFS alone in some settings
ORR:
- Magnitude and durability matter; a 20% ORR with 2-month median DOR is insufficient
- Stable disease must not be included
- Single-arm ORR requires well-characterized historical control rate
DFS:
- Counting non-cancer deaths as events vs. censoring them changes the estimate — FDA prefers counting all deaths
- Long post-treatment follow-up required; unscheduled assessment bias is a real risk
Endpoint inflation:
- Using tumor assessment endpoints as sole efficacy evidence for traditional approval typically requires confirmatory trial or IRC verification
Backlinks
- Overall Survival (OS)
- Progression-Free Survival (PFS)
- Response-Based Endpoints (ORR, CR, DOR)
- DFS and EFS Endpoints
- FDA Approval Pathways in Oncology
- Intercurrent Events in Oncology Trials
Source: FDA Guidance for Industry — Clinical Trial Endpoints for the Approval of Cancer Drugs and Biologics (December 2018, Final) Status: Final guidance Updated context: FDA OS Draft Guidance (August 2025, Draft) Compiled from retrieved FDA chunks + ClinicalTrials.gov records (2,454 Phase 3/2 oncology studies)