This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
STRIVE was a randomized, triple-masked, parallel-group phase 2 trial comparing enzalutamide with bicalutamide in men with prostate cancer. Its primary endpoint was progression-free survival defined by the earliest of PSA progression, radiographic progression or death.
| Feature | STRIVE |
|---|---|
| Official title | Safety and Efficacy Study of Enzalutamide Versus Bicalutamide in Men With Prostate Cancer |
| Phase | Phase 2 |
| Condition | Prostate cancer |
| Design | Randomized, parallel-group, triple-masked; primary purpose treatment |
| Arms | 2: enzalutamide and bicalutamide |
| Enrollment | 396 |
| Primary endpoint | Progression-free survival (PSA progression, radiographic progression or death) |
| Analysis stratification | Disease stage at study entry |
| Start / primary completion | 2012-08 / 2015-02 |
| Status | Completed; results posted |
| ClinicalTrials.gov | NCT01664923 |
| Lead sponsor | Pfizer (industry) |
2. Clinical Question
Enzalutamide and bicalutamide both block the androgen receptor, but differ in potency and mechanism. STRIVE was a head-to-head comparison: rather than testing a new agent against placebo, it asked whether enzalutamide delayed disease progression more than an established active comparator.
Population
Men with prostate cancer, including participants with and without metastatic disease at study entry; disease stage was used as a stratification factor.
Intervention
Enzalutamide, administered under triple masking.
Comparator
Bicalutamide, an active androgen-receptor inhibitor comparator.
Primary question
Does enzalutamide prolong progression-free survival, defined by PSA progression, radiographic progression or death, relative to bicalutamide?
3. Trial Design
Enzalutamide
- Androgen-receptor inhibitor
- Administered under triple masking
- Median duration of treatment 14.7 months at the data cut-off
Bicalutamide
- Active comparator androgen-receptor inhibitor
- Same masking as the enzalutamide arm
- Median duration of treatment 8.4 months at the data cut-off
4. Randomization, Stratification, and Analysis Populations
The primary analysis was stratified by disease stage at study entry. The stratified Cox model and stratified log-rank test compare treatments within each stage stratum and then combine the within-stratum information, so that differences in prognosis between stages do not blur the treatment comparison.
| Analysis population | Definition | Used for |
|---|---|---|
| Intent-to-treat | All participants randomly assigned to study treatment | PFS, time to PSA progression, FACT-P degradation |
| Evaluable intent-to-treat | Randomized participants with a baseline and at least 1 post-baseline PSA measurement | PSA response ≥50% |
| Metastatic subset | Randomized participants with metastatic disease at study entry | Radiographic PFS |
| Measurable soft-tissue subset | Randomized participants with metastatic and measurable soft tissue disease at screening | Best overall soft tissue response |
Only the primary PFS analysis and the time-based secondary endpoints use the full randomized population. The radiographic PFS and soft-tissue response analyses are restricted to baseline-defined subsets; because those subsets are defined by characteristics measured before randomization, the comparisons remain randomized, but they are smaller and less precise.
5. Endpoints
For every endpoint, the registry time frame is: from randomization until the data cut-off date of 09 February 2015 (median duration of treatment 14.7 months in the enzalutamide arm and 8.4 months in the bicalutamide arm).
| Endpoint | Role | Type |
|---|---|---|
| Progression Free Survival (PFS) | Primary | Time-to-event |
| Time to PSA Progression | Secondary | Time-to-event |
| Percentage of Participants With a PSA Response ≥ 50% | Secondary | Binary |
| Duration of Radiographic PFS | Secondary | Time-to-event (metastatic subset) |
| Quality of Life: Time to Degradation of FACT-P | Secondary | Time-to-event |
| Best Overall Soft Tissue Response | Secondary | Binary (measurable-disease subset) |
Primary endpoint definition
The registry defines PFS as time from randomization to earliest objective evidence of prostate-specific antigen (PSA) progression, radiographic progression, or death on study.
- PSA progression: ≥25% increase in PSA with an absolute increase ≥2 ng/mL above the nadir, confirmed by a second consecutive assessment.
- Radiographic progression in bone: based on Prostate Cancer Clinical Trials Working Group (PCWG2) guidelines, defined as at least 2 new lesions on bone scan.
- Radiographic progression in soft tissue: on CT/MRI, based on RECIST 1.1.
- Reading: CT/MRI and bone scans were read locally by the same radiologist (or nuclear medicine physician for bone scans) whenever possible.
- Censoring: participants not known to have had a PFS event at the analysis data cut-off were censored at the date of last assessment.
This is a composite endpoint. Because PSA typically rises before imaging changes, PSA progression is likely to drive many events, which is why the separate time-to-PSA-progression and radiographic PFS analyses are useful for understanding which component carries the effect.
6. Primary Result: Progression-Free Survival
Hazard ratio for progression or death, enzalutamide vs bicalutamide
95% CI (two-sided): 0.181–0.320 · P < 0.0001 (log-rank)
Cox regression stratified by disease stage at study entry · Intent-to-treat · Superiority
What the estimate means. An HR of 0.240 means that, over the observed follow-up and under the stratified Cox model, the estimated instantaneous rate of PSA progression, radiographic progression or death with enzalutamide was about one quarter of that with bicalutamide, a 76% lower estimated hazard.
What it does not mean. It does not mean 76% of patients avoided progression, nor that median PFS was four times longer. The registry analysis record does not report median PFS or event counts by arm, and the hazard ratio alone cannot be translated into absolute time gained.
Precision. The 95% CI of 0.181 to 0.320 is narrow and lies far below 1. Even the upper limit corresponds to a 68% lower estimated hazard, so the direction and approximate magnitude of the effect are well determined.
The p-value. P < 0.0001 from the log-rank test indicates the survival curves are very unlikely to be identical; it does not quantify how far apart they are. The HR and its interval carry the magnitude.
Cautions. A single HR assumes roughly proportional hazards across follow-up. Local, not central, radiologic reading introduces some risk of assessment variability, though triple masking limits systematic bias. The endpoint is a composite heavily informed by PSA, which is a biomarker rather than a direct measure of clinical benefit. Censoring at last assessment assumes that censored participants have the same future progression risk as those still followed.
7. Secondary Results
Time-to-event secondary endpoints
| Endpoint | Population | HR (95% CI) | P-value | Model |
|---|---|---|---|---|
| Time to PSA progression | ITT | 0.190 (0.137–0.264) | <0.0001 | Cox stratified by disease stage |
| Radiographic PFS | Metastatic at entry | 0.324 (0.211–0.497) | <0.0001 | Unstratified Cox; unstratified log-rank |
| Time to FACT-P degradation | ITT | 0.910 (0.695–1.192) | 0.4945 | Cox and log-rank stratified by disease stage |
The PSA-based endpoint shows the strongest relative effect (HR 0.190), and radiographic PFS in the metastatic subset also favours enzalutamide (HR 0.324), with a wider interval reflecting the smaller population. The consistency of direction between the PSA and imaging components supports the conclusion that the primary composite was not driven solely by a biomarker signal.
Time to FACT-P degradation behaves differently. The HR of 0.910 with a 95% CI of 0.695 to 1.192 includes 1, and P = 0.4945. The trial did not show a difference in time to quality-of-life deterioration. The interval is compatible with a modest benefit or a modest harm, so this is an inconclusive rather than a demonstrated null result.
Response endpoints
| Endpoint | Population | Effect measure | Estimate (95% CI) | P-value | Test |
|---|---|---|---|---|---|
| PSA response ≥50% | Evaluable ITT | Difference in rates (enzalutamide − bicalutamide) | 50.0 (41.4 to 58.5) | <0.0001 | CMH mean score, stratified by disease stage |
| Best overall soft tissue response | Metastatic, measurable soft tissue disease | Difference in objective response rate | 46.05 (26.79 to 65.30) | <0.0001 | Unstratified CMH mean score |
Both response differences are expressed in percentage points. The PSA response difference of 50.0 points has a relatively tight interval; the soft-tissue response difference of 46.05 points has a much wider interval (26.79 to 65.30), as expected from the smaller subset with measurable soft-tissue disease. The registry analysis records do not list the arm-specific response rates alongside these differences.
8. Safety: Serious Adverse Events
| Group | Participants with serious adverse events / at risk |
|---|---|
| Enzalutamide | 76 / 197 |
| DB phase: bicalutamide | 60 / 198 |
| Open-label phase: bicalutamide crossover | 16 / 37 |
Serious adverse events were reported for 76 of 197 enzalutamide-treated participants and 60 of 198 bicalutamide-treated participants in the double-blind phase. Because median treatment duration was 14.7 versus 8.4 months, crude proportions compare groups with different exposure time; exposure-adjusted rates would be needed for a fairer comparison. The open-label crossover group is small and not randomized, so it should not be compared directly with either double-blind arm. The registry does not post a formal statistical comparison of serious adverse events.
9. Statistical Methodology
Kaplan-Meier estimation and censoring
Time-to-event endpoints (PFS, time to PSA progression, radiographic PFS, time to FACT-P degradation) are summarized with Kaplan-Meier methods, which accommodate participants censored at their last assessment before the 09 February 2015 cut-off.
where di is the number of events at time ti and ni is the number at risk just before it.
Stratified log-rank test
The primary p-value comes from a log-rank test. For PFS, time to PSA progression and FACT-P degradation, the comparison is stratified by disease stage at study entry: observed-minus-expected events are computed within each stratum and summed. Radiographic PFS, analyzed only in participants with metastatic disease, used an unstratified test.
Stratified Cox proportional-hazards model
Hazard ratios were estimated from a Cox regression with treatment as the only covariate, stratified by disease stage at study entry. Stratification allows each stage to have its own baseline hazard while assuming a common treatment HR across strata.
where s indexes disease-stage strata, h0s is the stratum-specific baseline hazard and Z = 1 for enzalutamide. An HR below 1 favours enzalutamide.
Cochran-Mantel-Haenszel test for response
Binary response endpoints were compared with the Cochran-Mantel-Haenszel mean score test, stratified by disease stage for PSA response and unstratified for soft-tissue response. The effect is reported as a difference in response rates (enzalutamide minus bicalutamide).
Intention-to-treat
The primary analysis included all randomized participants, analyzed by assigned treatment. This preserves randomization regardless of treatment discontinuation or later switching.
10. Statistical Methods Explained
Under the stratified Cox model, the estimated hazard of progression or death with enzalutamide was 0.240 times that with bicalutamide, a 76% lower estimated hazard, averaged over follow-up. It is a relative rate, not a probability or a ratio of median times.
Participants with and without metastatic disease have different underlying progression risks. Stratification gives each stage its own baseline hazard, avoiding the need to model how stage affects risk while still pooling the treatment effect.
It was restricted to participants with metastatic disease at entry. Within that subset, disease stage no longer varies in the way it does in the full population, so a stratified analysis offers little benefit.
It compares response between treatments within strata and combines the evidence across strata, protecting against confounding by stage. The accompanying effect estimate here is a risk difference: 50.0 percentage points more participants achieved a ≥50% PSA decline with enzalutamide.
They measure different things. PFS captures biomarker and imaging progression; FACT-P captures participant-reported function and well-being. A treatment can delay measured progression without a detectable difference in time to quality-of-life decline, especially over a limited follow-up and when both arms receive active therapy.
Progression was read locally rather than centrally. Masking of multiple parties involved in the trial reduces the risk that knowledge of treatment influences when scans are ordered or how equivocal findings are called, a key protection for an assessor-dependent endpoint.
11. Important Limitations and Interpretation Issues
- Phase 2 design: STRIVE was a phase 2 trial; its results are informative but not designed with the size of a confirmatory phase 3 program.
- Composite, PSA-weighted endpoint: PFS includes PSA progression, a biomarker, alongside radiographic progression and death.
- Local radiologic reading: scans were read locally, not by blinded independent central review.
- Absolute effects: the posted analyses give hazard ratios and response differences but not median PFS or arm-specific rates in the analysis records.
- Proportional hazards: a single HR summarizes the effect across follow-up; if hazards change over time the summary is less descriptive.
- Subset analyses: radiographic PFS and soft-tissue response are estimated in smaller baseline-defined subsets with wider intervals.
- Multiplicity: no multiplicity-control procedure is described for secondary endpoints.
- Unequal exposure: median treatment durations of 14.7 and 8.4 months complicate crude safety comparisons.
- Overall survival: the registry does not report overall survival analyses for this trial.
12. Why This Trial Matters Statistically
STRIVE is a clear example of an active-comparator oncology trial in which stratified survival methods, composite endpoints and response-rate differences each answer a different question.
| Concept | How it appears in STRIVE |
|---|---|
| Time-to-event endpoints | PFS, time to PSA progression, radiographic PFS, FACT-P degradation |
| Kaplan-Meier estimation | Handles censoring at last assessment before the data cut-off |
| Log-rank test | Stratified by disease stage for primary PFS; unstratified for radiographic PFS |
| Hazard ratio | Primary PFS HR 0.240 (95% CI 0.181–0.320) |
| Cochran-Mantel-Haenszel test | PSA and soft-tissue response comparisons |
| Risk difference | PSA response difference 50.0; soft-tissue response difference 46.05 percentage points |
| Intention-to-treat | All randomized participants for the primary analysis |
| Confidence intervals | Contrast between precise PFS effects and the inconclusive FACT-P result |
13. Related Tutorials
Learn more about the methods used in this trial:
14. Related Calculators
15. Sources
- ClinicalTrials.gov: NCT01664923 — Safety and Efficacy Study of Enzalutamide Versus Bicalutamide in Men With Prostate Cancer.
- PubMed: PMID 34621011
- PubMed: PMID 30189902
- PubMed: PMID 26811535
Continue exploring clinical trial statistics
Each trial page connects its endpoints and methods to deeper statistical tutorials and to calculators for the same analyses.
16. Record Summary
STRIVE randomized 396 men with prostate cancer to enzalutamide or bicalutamide under triple masking. In the intent-to-treat population, the stratified Cox HR for progression-free survival was 0.240 (95% CI 0.181–0.320; log-rank P < 0.0001). Time to PSA progression (HR 0.190), radiographic PFS in metastatic participants (HR 0.324), PSA response ≥50% (difference 50.0 percentage points) and soft-tissue response (difference 46.05 percentage points) all favoured enzalutamide, whereas time to FACT-P degradation showed no clear difference (HR 0.910, 95% CI 0.695–1.192). Serious adverse events occurred in 76/197 enzalutamide and 60/198 double-blind bicalutamide participants, against a background of longer median exposure to enzalutamide. Interpretation should weigh the PSA-weighted composite endpoint, local radiologic reading and the phase 2 setting.