← Clinical Trials
Urothelial Cancer Phase 3 Randomized NCT04223856

EV-302: Complete Statistical Analysis of Enfortumab Vedotin and Pembrolizumab in Urothelial Cancer

An independent statistical analysis of the randomized phase 3 EV-302 trial comparing enfortumab vedotin plus pembrolizumab with standard of care in untreated locally advanced or metastatic urothelial cancer.

Trial start: March 30, 2020  ·  Primary completion: August 8, 2023  ·  Enrollment: 886
Scope of this record

This page separates reported trial results from statistical interpretation. The numerical efficacy results presented here are limited to the statistical analyses posted for the two registered primary endpoints. ClinicalTrials.gov provides the official trial registry record. View the EV-302 registry record.

1. Trial at a Glance

EV-302 was a randomized, parallel, open-label phase 3 trial evaluating enfortumab vedotin plus pembrolizumab versus standard of care in untreated locally advanced or metastatic urothelial cancer. The registry reports two primary time-to-event endpoints: progression-free survival and overall survival.

886
Enrolled
Randomized trial
2
Arms
Parallel design
0.450
PFS HR
95% CI 0.377–0.538
0.468
OS HR
95% CI 0.376–0.582
FeatureEV-302
Trial nameEV-302
PhasePhase 3
ConditionUrothelial Cancer
PopulationUntreated locally advanced or metastatic urothelial cancer
AllocationRandomized
DesignParallel
MaskingNone
Primary purposeTreatment
Enrollment886
Primary endpointsProgression-Free Survival (PFS) and Overall Survival (OS)
Primary endpoint typeTime-to-event
Primary analysis methodLog-rank test
Effect measureHazard ratio
Hypothesis typeSuperiority
StatusActive, not recruiting
Lead sponsorAstellas Pharma Global Development, Inc.
Sponsor typeIndustry

2. Clinical Question

The primary statistical question was whether treatment with enfortumab vedotin plus pembrolizumab produced a different time-to-event profile from standard of care for patients with untreated locally advanced or metastatic urothelial cancer. Both registered primary endpoints were superiority hypotheses.

Population

Participants with untreated locally advanced or metastatic urothelial cancer.

Intervention

Enfortumab vedotin plus pembrolizumab.

Comparator

Standard of care. The registered interventions include cisplatin, carboplatin, and gemcitabine.

Primary question

Does enfortumab vedotin plus pembrolizumab improve progression-free survival and overall survival relative to standard of care?

3. Trial Design

01
Randomize886 participants
02
Two armsIntervention vs standard of care
03
FollowTime-to-event outcomes
04
AssessPFS and OS
05
CompareLog-rank and hazard ratio
INTERVENTION ARM

Enfortumab Vedotin + Pembrolizumab

  • Enfortumab vedotin
  • Pembrolizumab
COMPARATOR ARM

Standard of Care

  • Cisplatin
  • Carboplatin
  • Gemcitabine
Open-label design: the registry classifies EV-302 as having no masking. This means the treatment assignments were not masked in the trial design. Importantly, the primary PFS endpoint was assessed by Blinded Independent Central Review (BICR), providing an independent blinded efficacy assessment despite the overall open-label design.

4. Randomization and Analysis Population

The primary efficacy analyses used the intention-to-treat (ITT) analysis set. The registry defines this population as including all randomized participants, with participants analyzed according to the treatment arm assigned at randomization regardless of the actual treatment received.

Analysis populationDefinition / role
Intention-to-treatAll randomized participants. Participants were analyzed according to the treatment arm assigned at randomization regardless of the actual treatment received.
Primary PFS comparisonEnfortumab vedotin + pembrolizumab versus standard of care.
Primary OS comparisonEnfortumab vedotin + pembrolizumab versus standard of care.

This is an important feature of the statistical analysis. Randomization creates the basis for a causal comparison, and maintaining participants in their randomized groups for the primary efficacy analysis helps preserve that design principle even when actual treatment received differs from assignment.

5. Primary Endpoints

EndpointRegistered definition / time framePrimary analysis
Progression-Free Survival (PFS) Per Response Evaluation Criteria in Solid Tumors (RECIST) Version (v) 1.1 by Blinded Independent Central Review (BICR) From the date of randomization to first documentation of PD or death due to any cause, whichever occurred first. Log-rank test; hazard ratio from a stratified Cox proportional hazards model.
Overall Survival (OS) From randomization to date of death due to any cause or censoring date, whichever occurred first. Log-rank test; hazard ratio from a stratified Cox proportional hazards model.

Progression-Free Survival

The registered PFS definition is a time-to-event endpoint beginning at randomization. The event is the first documentation of disease progression or death due to any cause, whichever occurs first. Disease progression is defined using RECIST v1.1 criteria, with BICR providing the central assessment.

Overall Survival

OS is defined as the time from randomization to death from any cause. When death is not observed, the participant is censored at the date the participant was last known to be alive. The registry states that the Kaplan-Meier method was used for analysis.

6. Statistical Methodology

Kaplan-Meier estimation

Both primary endpoints are time-to-event outcomes. Kaplan-Meier estimation is therefore a natural way to describe the distribution of time until progression or death for PFS and time until death for OS.

Conceptual form
S(t) = ∏ti ≤ t (1 − di/ni)

Here, di is the number of events at time ti and ni is the number of participants at risk immediately before that time. The method allows participants without an observed event to contribute information until censoring.

Log-rank test

The registry reports the log-rank test as the primary comparison method for both PFS and OS. The test compares the observed pattern of events between randomized treatment groups across follow-up rather than comparing only a single time point.

Stratified Cox proportional hazards model

The registry states that the hazard ratio for each primary endpoint was calculated using a stratified Cox proportional hazards model. A Cox model estimates a relative hazard between treatment groups while accounting for the time-to-event structure and censoring.

Interpretation of the hazard ratio
HR < 1  →  lower estimated instantaneous event rate in the treatment group

A hazard ratio is a relative, model-based measure of the event rate over time. It is not the same as a relative risk, an absolute risk difference, a probability of benefit, or a statement that every individual patient experiences the same proportional reduction.

Intention-to-treat analysis

Both primary analyses used the ITT analysis set. This means that treatment assignment at randomization, rather than treatment actually received, defines the groups for the primary efficacy comparison. The approach preserves the treatment contrast created by randomization.

Stratified analysis

The statistical-analysis records identify stratified analysis as part of the primary analyses. The hazard ratios were calculated with stratified Cox proportional hazards models, while the OS p-value was calculated using a stratified log-rank test.

7. Primary Result: Progression-Free Survival

The registry reports a formal statistical analysis of PFS comparing enfortumab vedotin plus pembrolizumab with standard of care in the ITT analysis set.

Hazard ratio for progression or death

0.450

95% CI: 0.377–0.538   ·   P < 0.00001

Two-sided 95% confidence interval; superiority hypothesis.

CharacteristicReported analysis
EndpointPFS per RECIST v1.1 by BICR
PopulationITT analysis set
GroupsEnfortumab Vedotin + Pembrolizumab vs Standard of Care
MethodLog-rank test
Effect measureHazard ratio
Estimate0.450
95% CI0.377–0.538
P-value<0.00001
HypothesisSuperiority
HR modelStratified Cox proportional hazards model
Clinical Biostats interpretation

The reported PFS hazard ratio of 0.450 means that the estimated instantaneous rate of progression or death was about 55% lower in the enfortumab vedotin plus pembrolizumab group than in the standard-of-care group under the fitted time-to-event model. This is a relative hazard interpretation, not a statement that 55% of participants avoided progression or death.

The 95% confidence interval of 0.377–0.538 describes statistical uncertainty around the estimated hazard ratio under the analysis framework. Because the entire interval is below 1, the interval is consistent with a lower estimated hazard in the intervention group.

The p-value of <0.00001 measures the evidence against the null hypothesis under the specified testing framework. It does not measure the size of the treatment effect, the probability that the treatment works, or the clinical importance of the result.

The interpretation also depends on the Cox proportional-hazards framework and on the censoring and event-assessment procedures used for PFS. The registry does not provide enough information here to independently evaluate the proportional-hazards assumption from the underlying event-time data.

8. Primary Result: Overall Survival

The registry also reports a formal ITT analysis of OS using the same randomized comparison. OS is measured from randomization to death from any cause, with censoring when death is not observed.

Hazard ratio for death

0.468

95% CI: 0.376–0.582   ·   P < 0.00001

Two-sided 95% confidence interval; superiority hypothesis.

CharacteristicReported analysis
EndpointOverall Survival
PopulationITT analysis set
GroupsEnfortumab Vedotin + Pembrolizumab vs Standard of Care
MethodLog-rank test
Effect measureHazard ratio
Estimate0.468
95% CI0.376–0.582
P-value<0.00001
HypothesisSuperiority
HR modelStratified Cox proportional hazards model
Reported significance threshold0.01548
Clinical Biostats interpretation

The reported OS hazard ratio of 0.468 means that the estimated instantaneous rate of death was about 53.2% lower in the enfortumab vedotin plus pembrolizumab group than in the standard-of-care group under the fitted Cox model. It does not mean that 53.2% of patients were saved, that individual mortality risk fell by exactly 53.2%, or that the absolute survival difference is 53.2 percentage points.

The 95% confidence interval of 0.376–0.582 expresses the statistical precision of the estimated hazard ratio. It remains below 1 throughout the interval, so the interval is compatible with a lower estimated hazard of death for the intervention group.

The reported p-value of <0.00001 is a measure of evidence against the relevant null hypothesis under the specified stratified testing procedure. It is not an effect-size measure. A smaller p-value does not by itself mean that an effect is clinically larger.

The registry reports a statistical-significance threshold of 0.01548 for the OS analysis. The reported p-value is below that threshold. The interpretation remains subject to the prespecified multiplicity and testing framework; the ClinicalTrials.gov record does not provide enough detail to reconstruct the complete alpha-allocation scheme across every endpoint and analysis.

As with any Cox-model hazard ratio, interpretation also depends on the proportional-hazards assumption. The single HR summarizes a relative event-rate relationship over follow-up; it should not be interpreted as though the same absolute treatment difference necessarily applies at every time point.

9. Reading the Two Primary Results Together

EV-302 reports hazard ratios below 1 for both registered primary endpoints. The PFS analysis estimates a hazard ratio of 0.450, while the OS analysis estimates a hazard ratio of 0.468. Both confidence intervals lie below 1, and both reported p-values are <0.00001.

Primary endpointHazard ratio95% CIP-valueInterpretation of direction
PFS0.4500.377–0.538<0.00001Lower estimated hazard of progression or death with enfortumab vedotin + pembrolizumab
OS0.4680.376–0.582<0.00001Lower estimated hazard of death with enfortumab vedotin + pembrolizumab

The two endpoints answer related but distinct questions. PFS counts progression or death as the event, whereas OS counts death from any cause. A treatment effect on PFS therefore cannot simply be substituted for an OS effect. In this trial, the registry reports formal analyses for both endpoints.

10. Safety: Serious Adverse Events

The ClinicalTrials.gov record reports serious adverse events by randomized arm as affected participants divided by participants at risk.

ArmParticipants with serious adverse eventsParticipants at risk
Enfortumab Vedotin + Pembrolizumab220440
Standard of Care169433

These figures describe the number of participants affected and the corresponding number at risk reported in the ClinicalTrials.gov record. They should be interpreted separately from the efficacy hazard ratios because safety and efficacy use different outcome definitions and analytical objectives.

Safety interpretation: the ClinicalTrials.gov record does not provide a detailed breakdown of individual serious adverse-event categories, severity, timing, treatment relationship, or discontinuations. Those details should not be inferred from the aggregate affected/at-risk counts.

11. Statistical Methods Explained

Why was a log-rank test used?

PFS and OS are time-to-event endpoints, so participants can have different follow-up durations and some may be censored before experiencing the event. The log-rank test is designed to compare survival distributions while using the available event-time information across follow-up rather than reducing the analysis to a single fixed time point.

What does a hazard ratio of 0.450 mean?

A hazard ratio of 0.450 indicates that the fitted model estimates a substantially lower instantaneous event rate in the enfortumab vedotin plus pembrolizumab group than in the standard-of-care group. It does not mean that the intervention reduces every individual's probability of an event by exactly 55%.

Why is the confidence interval important?

The point estimate alone does not describe statistical precision. The 95% confidence interval shows the range of hazard-ratio values compatible with the data and model under the specified confidence procedure. A narrower interval generally indicates greater statistical precision than a wider interval, although precision is not the same as clinical importance.

Why does the p-value not measure effect size?

A p-value addresses evidence against a null hypothesis under a specified statistical model. It does not tell us how large the treatment effect is. The hazard ratio describes relative effect size, while the confidence interval describes uncertainty around that effect estimate.

Why use an intention-to-treat analysis?

Analyzing participants according to randomized assignment preserves the comparison created by randomization. This reduces the risk that post-randomization treatment decisions selectively determine who remains in each efficacy group.

Why was a stratified Cox model used?

The registry specifically reports a stratified Cox proportional hazards model for calculation of the primary hazard ratios. Stratification allows the time-to-event comparison to account for the prespecified stratified analysis structure rather than treating all participants as though the relevant strata had no role in the model.

What does censoring mean for OS?

For OS, a participant who has not been observed to die is not treated as though death never occurs. Instead, the participant contributes follow-up information until the date on which they were last known to be alive, at which point the observation is censored. Kaplan-Meier and Cox methods incorporate these censored observations into the time-to-event analysis.

12. Confidence Intervals and the Meaning of Precision

The two primary hazard-ratio estimates are accompanied by two-sided 95% confidence intervals:

EndpointEstimateLower boundUpper boundInterval width
PFS0.4500.3770.5380.161
OS0.4680.3760.5820.206

The confidence intervals provide more information than the point estimates alone. For PFS, the estimated hazard ratio is 0.450 and the interval extends from 0.377 to 0.538. For OS, the estimate is 0.468 and the interval extends from 0.376 to 0.582.

Neither interval should be interpreted as a range containing the true treatment effect with a 95% probability in this particular trial. The conventional frequentist interpretation concerns the long-run performance of the confidence procedure under repeated sampling. For practical interpretation, the key point is that the interval communicates the uncertainty associated with each estimated hazard ratio.

13. P-Values and Statistical Evidence

Both primary analyses report P < 0.00001. These p-values provide very strong statistical evidence against the relevant null hypothesis under the specified analysis framework.

What the p-value tells us

It quantifies how incompatible the observed test statistic is with the null hypothesis under the specified testing procedure.

What it does not tell us

It does not give the probability that the null hypothesis is true, the probability that treatment benefits an individual, or the size of the treatment effect.

For OS, the ClinicalTrials.gov record also specifies a statistical-significance threshold of 0.01548. The reported OS p-value is below that threshold. The ClinicalTrials.gov record does not provide a corresponding numerical threshold for the PFS analysis, so no additional threshold is inferred here.

14. Time-to-Event Endpoints: Why the Analysis Is More Than a Single Number

PFS and OS are fundamentally longitudinal outcomes. Each participant contributes information through time, and the exact timing of progression, death, or censoring matters. This is why the trial uses survival-analysis methods rather than a simple comparison of proportions.

Core statistical structure
Randomization → follow-up → event or censoring → Kaplan-Meier / log-rank → Cox hazard ratio

The analysis combines the randomized treatment comparison with the timing of events and the handling of incomplete event observation.

For PFS, progression and death are treated as the event definition reported by the registry. For OS, death is the event and participants who remain alive are censored at their last known alive date. These definitions determine what the hazard ratio actually describes.

15. Stratified Analysis and the Hazard Ratio

The statistical analysis records identify both stratified analysis and a stratified Cox proportional hazards model. This is important because the hazard ratio is not simply a ratio obtained by dividing two crude event percentages.

A Cox hazard ratio compares the modeled instantaneous event rates between treatment groups while incorporating the time dimension of the outcome. Stratification modifies the model structure so that the baseline hazard can differ across strata while the treatment effect is estimated within the specified analysis framework.

Educational caution

A hazard ratio is not a direct measure of absolute benefit. For example, HR 0.468 does not mean that the absolute probability of death is reduced by 53.2 percentage points. Absolute survival probabilities at specified times, when available, are needed to answer absolute-risk questions.

16. Proportional-Hazards Assumption

The Cox model used for both primary hazard ratios is a proportional-hazards model. Its standard interpretation assumes that the relative hazard between groups is adequately represented by a common hazard ratio over the relevant follow-up.

If the treatment effect changes materially over time, a single HR can become a less complete description of the underlying survival experience. In that situation, Kaplan-Meier curves, time-specific survival estimates, restricted mean survival time, or time-varying effect models may provide additional information.

Data limitation: the registry-reported EV-302 trial data contain the hazard-ratio analyses but do not provide the underlying event-time dataset or sufficient survival-curve information to independently assess proportional hazards. No such assessment is therefore claimed on this page.

17. What the Primary Results Do — and Do Not — Establish

What is directly reported

The ITT PFS analysis reports HR 0.450 with 95% CI 0.377–0.538 and P < 0.00001. The ITT OS analysis reports HR 0.468 with 95% CI 0.376–0.582 and P < 0.00001.

What the estimates represent

They are relative time-to-event effect measures comparing enfortumab vedotin plus pembrolizumab with standard of care under the reported stratified Cox models.

What cannot be inferred

The hazard ratios do not directly provide absolute survival differences, median survival times, individual treatment benefit probabilities, or the percentage of participants cured.

What requires additional data

Detailed subgroup effects, survival probabilities at selected time points, median PFS or OS, and complete adverse-event category tables require information not reported in the ClinicalTrials.gov record.

18. Limitations

19. Design Topics Not Supported by the Supplied Data

TopicWhat the registry-reported EV-302 data support
Non-inferiority marginNot reported. The registered primary hypotheses are superiority hypotheses.
CrossoverNot reported in the ClinicalTrials.gov record.
Factorial designNot applicable to the reported parallel two-arm design.
MultiplicityOS has a reported statistical-significance threshold of 0.01548; the complete multiplicity strategy is not reported.
Interim analysisNot reported in the ClinicalTrials.gov record.
Missing-data / imputation methodNot reported in the ClinicalTrials.gov record.
Bayesian methodsNot reported. The posted primary methods are frequentist log-rank and stratified Cox analyses.

This distinction matters because a statistical analysis page should not fill gaps in the registry record with assumptions based on how other phase 3 oncology trials are commonly designed. Where the registry-reported EV-302 data do not identify a method, this page does not assign one to the trial.

20. Why This Trial Matters Statistically

EV-302 is a useful teaching example of a modern randomized phase 3 time-to-event analysis because its two registered primary endpoints use the same core statistical framework while measuring different clinical events.

Statistical conceptHow it appears in EV-302
RandomizationThe trial uses randomized allocation with two parallel treatment arms.
Intention-to-treat analysisBoth primary analyses include all randomized participants and preserve treatment assignment at randomization.
Time-to-event endpointBoth PFS and OS are time-to-event primary endpoints.
Kaplan-Meier estimationThe OS registry definition identifies Kaplan-Meier as the analysis method.
Log-rank testThe posted statistical analysis uses a log-rank test for both primary endpoints.
Hazard ratioPFS and OS effects are expressed as hazard ratios.
Confidence intervalBoth primary hazard ratios have two-sided 95% confidence intervals.
Stratified analysisThe primary analysis records identify stratified analysis, with HRs calculated using stratified Cox models.
BICRThe PFS endpoint is assessed by Blinded Independent Central Review.
Superiority testingBoth registered primary analyses use superiority hypotheses.

21. Statistical Concepts in This Trial

Learn more about the methods used in this trial:

22. Related Statistical Calculators

23. Sources

Continue through the Clinical Biostats statistical pathway

Explore the survival-analysis methods and statistical tools that explain how randomized time-to-event trials are designed and analyzed.

24. Record Summary

EV-302 provides a clear example of a randomized phase 3 trial in which both primary endpoints are time-to-event outcomes analyzed in the ITT population. The registry reports log-rank testing and stratified Cox proportional-hazards models for PFS and OS. The PFS analysis reports a hazard ratio of 0.450 with a two-sided 95% CI of 0.377–0.538 and P < 0.00001. The OS analysis reports a hazard ratio of 0.468 with a two-sided 95% CI of 0.376–0.582 and P < 0.00001. The OS analysis record also specifies a statistical-significance threshold of 0.01548.

The statistical interpretation should remain anchored to what those measures actually describe: relative event hazards under specified survival models, with uncertainty quantified by confidence intervals and evidence summarized by hypothesis-test p-values. The analysis does not by itself provide absolute survival probabilities, median survival times, individual treatment benefit probabilities, or subgroup treatment effects unless those quantities are separately reported.

Clinical Biostats methodology: A trial-results page should not merely repeat a trial label or headline result. The goal is to reconstruct the statistical structure of the evidence, explain what each reported estimate means, identify assumptions and limitations, and distinguish reported results from interpretation without filling gaps in the registry record with unsupported numbers.