This page separates reported registry information from statistical interpretation. The ClinicalTrials.gov record identifies the trial design, registered primary endpoints, and serious adverse-event counts, but does not supply formal statistical-analysis estimates in the posted results.
1. Trial at a Glance
COU-AA-302 was a randomized, parallel-group, quadruple-masked phase 3 treatment trial in prostate cancer. The ClinicalTrials.gov record describes two arms, 1088.0 enrolled participants, two registered primary time-to-event endpoints, and results posted on ClinicalTrials.gov.
| Feature | COU-AA-302 |
|---|---|
| Trial name | COU-AA-302 |
| NCT identifier | NCT00887198 |
| Brief title | Abiraterone Acetate in Asymptomatic or Mildly Symptomatic Patients With Metastatic Castration-Resistant Prostate Cancer |
| Phase | Phase 3 |
| Status | COMPLETED |
| Therapeutic area | Oncology |
| Condition | Prostate Cancer |
| Allocation | RANDOMIZED |
| Design model | PARALLEL |
| Masking | QUADRUPLE |
| Primary purpose | TREATMENT |
| Enrollment | 1088 |
| Interventions | Abiraterone acetate; Placebo; Prednisone |
| Primary endpoints | Overall Survival; Radiographic Progression-free Survival (rPFS) |
| Outcome measures posted | 11 |
| Lead sponsor | Janssen Research & Development, LLC |
| Sponsor type | INDUSTRY |
2. Clinical Question
The registry title identifies the population and intervention of interest: asymptomatic or mildly symptomatic patients with metastatic castration-resistant prostate cancer. The randomized design provides the framework for comparing two treatment groups while maintaining allocation by chance rather than by investigator or participant choice.
Population
Asymptomatic or mildly symptomatic patients with metastatic castration-resistant prostate cancer, as specified by the trial's brief title.
Intervention
Abiraterone acetate with prednisone is represented among the registered interventions.
Comparator
Placebo is represented as the comparator intervention in the ClinicalTrials.gov record.
Primary question
How does randomized treatment assignment relate to the two registered time-to-event outcomes: Overall Survival and Radiographic Progression-free Survival?
The ClinicalTrials.gov record does not provide the complete treatment-administration schedule, dose, baseline disease characteristics, eligibility criteria, or detailed treatment-assignment description.
3. Trial Design
Abiraterone acetate + prednisone
- Abiraterone acetate is listed as a drug intervention.
- Prednisone is listed as a drug intervention.
- The registry-reported serious-adverse-event denominator for this group is 542.
Placebo
- Placebo is listed as a drug intervention.
- The registry-reported serious-adverse-event denominator for this group is 540.
- A separate placebo-to-AA group is also reported for serious adverse events.
The ClinicalTrials.gov record does not specify a randomized allocation ratio, so no allocation ratio is inferred from the enrollment or safety denominators.
4. Trial Timeline and Registry Status
Trial start
The registry data give 2009-04-28 as the trial start date.
Primary completion
The registry data give 2014-03-31 as the primary completion date.
COMPLETED
The ClinicalTrials.gov record identifies the trial as completed and indicates that results have been posted.
5. Primary Endpoints
| Endpoint | Registry time frame | Registry definition | Endpoint type |
|---|---|---|---|
| Overall Survival | From randomization (Day 1) up to end of study (Month 60) | Overall survival is defined as the time from randomization to date of death from any cause. | Time-to-event |
| Radiographic Progression-free Survival (rPFS) | From randomization (Day 1) up to first radiographic progression or cutoff date (Month 18) | The rPFS was defined as the time from randomization to the occurrence of radiographic progression as specified in the registry definition. | Time-to-event |
Overall Survival
OS is a classic time-to-event endpoint. The event is death from any cause, and the analysis begins at randomization rather than at a later treatment milestone.
Radiographic PFS
rPFS is also time-to-event, but the event is radiographic progression rather than death alone. Its registry time frame ends at first radiographic progression or the Month 18 cutoff.
6. Statistical Methodology
The ClinicalTrials.gov record classifies both primary endpoints as time-to-event outcomes. They do not provide a populated statistical-analysis section describing the actual COU-AA-302 analysis model, test statistic, hazard ratio, confidence interval, p-value, stratification factors, censoring rules beyond the registered endpoint definitions, or multiplicity procedure.
That distinction is important. A time-to-event endpoint has a recognizable statistical toolkit, but the existence of that toolkit does not establish that every component was used in this particular trial. The methods below therefore describe the standard statistical framework that would normally be considered for endpoints of these types rather than asserting that these methods were used by COU-AA-302.
Kaplan-Meier estimation
For a time-to-event endpoint, the Kaplan-Meier estimator is commonly used to estimate the probability of remaining event-free over time. It accommodates right-censoring, allowing participants who have not experienced the event by their last evaluable time to contribute information up to that point.
Here, di denotes the number of events at event time ti, while ni is the number at risk immediately before that time.
The value of the Kaplan-Meier approach is that it preserves the timing information in a time-to-event dataset rather than reducing every participant to a simple event/no-event indicator.
Log-rank comparison
A log-rank test is commonly used to compare survival distributions between randomized groups. Conceptually, it compares the observed number of events in each group with the number expected under a null hypothesis of equal survival distributions, while accounting for the changing risk set over time.
The registry-reported COU-AA-302 data do not report whether an ordinary or stratified log-rank test was used, nor do they provide a test statistic or p-value. No specific log-rank result is therefore attributed to this trial on this page.
Cox proportional-hazards model
A Cox proportional-hazards model is commonly used to estimate a hazard ratio comparing two treatment groups. The hazard ratio is a relative measure of the instantaneous event rate, conditional on the model and its assumptions.
A hazard ratio is not the same as a relative risk, an absolute risk difference, a probability of surviving, or the proportion of patients who benefit.
The proportional-hazards assumption is particularly important when interpreting a single Cox hazard ratio. If the relative hazard changes materially over time, one summary hazard ratio may not fully describe the treatment-effect pattern. The ClinicalTrials.gov record does not provide an assessment of proportional hazards, so no conclusion about that assumption is made here.
Intention-to-treat analysis
For randomized trials, an intention-to-treat analysis commonly compares participants according to their randomized assignment. This maintains the treatment comparison created by randomization and is generally important for estimating the effect of assignment to a treatment strategy.
The registry-reported COU-AA-302 data identify the trial as randomized but do not specify the analysis populations used for each endpoint. An ITT analysis is therefore described here as the standard framework rather than as a documented COU-AA-302 analysis.
Censoring
Time-to-event analysis requires explicit rules for what happens when a participant's event time is not observed during the relevant follow-up. Such participants may be censored according to prespecified rules. The exact COU-AA-302 censoring rules are not contained in the registry-reported statistical-analysis field.
7. Planned Analysis: Overall Survival
ClinicalTrials.gov reports Overall Survival as a registered primary endpoint and indicates that results have been posted, but the registry-reported statistical-analysis array contains no formal numerical analysis for this endpoint. Accordingly, this page reports the registered definition and time frame but does not state an OS hazard ratio, confidence interval, p-value, median, event count, or survival probability.
What the endpoint measures
Overall Survival is defined as the time from randomization to the date of death from any cause.
Registry time frame: from randomization (Day 1) up to end of study (Month 60).
How Overall Survival would normally be analyzed
A conventional analysis would use a Kaplan-Meier estimator to describe the survival distribution in each randomized group. A between-group comparison would commonly use a log-rank framework, and a Cox proportional-hazards model could provide a hazard ratio with a confidence interval.
Those methods are statistical options for time-to-event data, not reported COU-AA-302 numerical results in the ClinicalTrials.gov record. Without the populated analysis results, it would be inappropriate to infer an effect estimate from the existence of the endpoint or from the fact that results are posted.
What it means: OS asks how long participants remain alive after randomization, with death from any cause serving as the event.
What it does not mean: OS is not the same as tumor response, radiographic disease control, or time until progression. A participant can remain alive after experiencing progression, and therefore OS and rPFS answer different clinical questions.
Precision: A formal estimate such as a hazard ratio would need a confidence interval to describe statistical uncertainty. No such estimate or interval is present in the ClinicalTrials.gov record.
P-values: A p-value, if reported, would address evidence against a specified null hypothesis; it would not itself quantify the magnitude of an OS difference.
Caution: Interpretation of any Cox-model result would require attention to censoring and the proportional-hazards assumption, as well as the prespecified analysis population and testing framework.
8. Planned Analysis: Radiographic Progression-free Survival
Radiographic Progression-free Survival (rPFS) is the second registered primary endpoint. The registry gives a time frame from randomization (Day 1) up to first radiographic progression or the cutoff date at Month 18.
What the endpoint measures
rPFS is defined as time from randomization to the occurrence of radiographic progression under the registry's prespecified progression criteria.
Registry time frame: from randomization (Day 1) up to first radiographic progression or cutoff date (Month 18).
Why rPFS requires careful event definition
Unlike OS, where the event is death from any cause, rPFS depends on an operational definition of radiographic progression. That makes the exact assessment rule part of the endpoint itself. The registry definition explicitly addresses bone-scan progression and specifies timing and confirmation requirements for one component of the progression definition.
The registry-reported definition states that a participant could be considered to have progressed by bone scan when the first bone scan with ≥2 new lesions compared with baseline was observed at <12 weeks from randomization and was confirmed by a second bone scan taken ≥6 weeks later showing ≥2 additional findings. Because the registry-reported text ends there, the remainder of the registered definition cannot be reproduced reliably from the ClinicalTrials.gov record.
How rPFS would normally be analyzed
For a time-to-event endpoint of this type, Kaplan-Meier estimation would commonly be used to characterize time to progression or censoring, with a between-group comparison using a log-rank framework and a Cox model as a way to estimate a relative hazard.
Again, these are standard analytical approaches rather than documented COU-AA-302 results in the registry-reported statistical-analysis field. No rPFS hazard ratio, confidence interval, p-value, median, or event count is reported here.
What it means: rPFS captures the time until radiographic progression under the trial's defined assessment rules, with the analysis beginning at randomization.
What it does not mean: rPFS is not overall survival. A participant may experience radiographic progression without dying, so the two primary endpoints measure distinct aspects of the disease course.
Precision: Any formal treatment-effect estimate would need an uncertainty interval. The ClinicalTrials.gov record contains no such estimate for rPFS.
P-values: A p-value would describe evidence relative to a null hypothesis; it would not describe the size or clinical importance of the radiographic difference by itself.
Caution: Because rPFS depends on radiographic event definitions, timing, confirmation, and censoring rules, the statistical estimate cannot be separated from the endpoint-assessment framework.
9. Results Available in the ClinicalTrials.gov record
The ClinicalTrials.gov record states Results Posted: Yes and identifies 11 posted outcome measures. However, no formal statistical analyses were posted. The distinction matters: a posted outcome-measure record can establish that results are present in the registry without providing the formal comparative estimates needed to reproduce a statistical treatment-effect analysis from the ClinicalTrials.gov recordset.
| Result component | What the ClinicalTrials.gov record establishes |
|---|---|
| Results status | Yes |
| Outcome measures posted | 11 |
| Formal statistical analyses reported | None; no formal statistical analyses were posted |
| Overall Survival numerical comparison | Not reported in the ClinicalTrials.gov record |
| rPFS numerical comparison | Not reported in the ClinicalTrials.gov record |
| Median survival estimates | Not reported |
| Hazard ratios | Not reported |
| Confidence intervals | Not reported |
| P-values | Not reported |
10. Safety Results
The ClinicalTrials.gov record provides serious adverse events by arm as affected participants divided by the reported at-risk denominator. These are the only safety results posted on ClinicalTrials.gov for the trial.
| Group | Serious adverse events | Affected / at risk |
|---|---|---|
| Abiraterone Acetate + Prednisone (AAP) | Serious adverse events | 215/542 |
| Placebo | Serious adverse events | 156/540 |
| Placebo to AA | Serious adverse events | 40/93 |
The visualization is intentionally based on the affected counts rather than converting the reported fractions into newly calculated percentages. The registry-reported instructions require trial numbers to be reproduced exactly rather than recomputed.
The numerator identifies the number of participants affected by serious adverse events and the denominator identifies the reported at-risk population for that group. These figures describe safety burden; they do not establish whether an observed difference is statistically significant or causally attributable to treatment.
The Placebo to AA category should also be kept analytically distinct from the primary randomized-group comparison. Its presence in the ClinicalTrials.gov record indicates an additional treatment-status group, but the ClinicalTrials.gov record does not provide enough information here to reconstruct the timing, eligibility, or analytical handling of that group.
11. Randomization and Blinding
Randomization and blinding are particularly important statistical design concepts for COU-AA-302 because the registry identifies both features explicitly: allocation was RANDOMIZED and masking was QUADRUPLE.
Why randomization matters
Randomization creates a treatment comparison in which assignment is determined by chance. In a sufficiently large randomized trial, this helps balance prognostic factors between groups in expectation. It also provides the foundation for interpreting a between-group efficacy comparison as an estimate of the effect of treatment assignment rather than an association created by treatment-selection preferences.
Randomization does not guarantee identical baseline characteristics. Rather, it protects the comparison probabilistically through the assignment mechanism. This distinction is important when reading any randomized trial: balance is a property expected from the design, not a promise that every characteristic will be numerically equal.
Why masking matters
Quadruple masking is the registry's description of the trial's masking status. Masking can reduce the possibility that knowledge of treatment assignment influences behavior, assessment, reporting, or other aspects of trial conduct.
For a radiographic endpoint, blinding can be particularly relevant because progression assessment involves classification of whether a prespecified radiographic event has occurred. The ClinicalTrials.gov record does not specify the exact identities of the four masked parties, so they are not inferred here.
12. Statistical Methods Explained
Why are OS and rPFS treated as time-to-event endpoints?
Because both endpoints measure the amount of time from randomization until a defined event occurs. OS uses death from any cause as the event. rPFS uses radiographic progression under the registered progression definition. Time-to-event analysis preserves information about when the event occurred and can accommodate participants whose event has not been observed during follow-up.
Why not simply compare the proportion of participants who experienced an event?
A simple event proportion ignores the timing of events and follow-up duration. Two groups could have the same proportion of events while having very different event timing. Survival methods are designed to use the time dimension and account for censoring.
What would a Kaplan-Meier curve show?
A Kaplan-Meier curve would estimate the probability of remaining event-free over time. For OS, that means remaining alive. For rPFS, it means remaining free of the defined radiographic progression event. Drops in the curve occur when events are observed, while censoring indicates that an individual's event-free follow-up ends without an observed event at that time.
What would a hazard ratio mean?
A hazard ratio compares modeled instantaneous event rates between treatment groups. An HR below 1 would correspond to a lower estimated instantaneous event rate in the treatment group under the model. It would not mean that the same percentage of patients avoid the event, nor would it directly represent an absolute difference in survival probability.
Why does the confidence interval matter?
A confidence interval communicates statistical uncertainty around an estimated treatment effect. A narrow interval generally indicates greater precision than a wide interval, although precision and clinical importance are separate concepts. The registry-reported COU-AA-302 statistical-analysis data contain no confidence intervals, so none are fabricated here.
Why does the p-value not measure treatment-effect size?
A p-value addresses compatibility of the observed data with a specified null hypothesis under a statistical testing framework. It depends on factors including sample size and variability and therefore should not be interpreted as a direct measure of the magnitude or clinical importance of a treatment effect.
13. Interpreting Time-to-Event Evidence Carefully
Relative vs absolute effects
A hazard ratio is relative. A survival probability at a particular time is absolute. A complete analysis benefits from distinguishing these measures rather than treating them as interchangeable.
Event definition
OS and rPFS have different events. The distinction is not cosmetic: changing the event changes the scientific question and the censoring and assessment framework.
Follow-up window
The registered time frames differ: OS extends to the end of study at Month 60, whereas rPFS uses the first radiographic progression or a cutoff at Month 18.
Analysis population
The ClinicalTrials.gov record does not specify the formal efficacy analysis population. It should therefore not be assumed from the enrollment number alone.
These distinctions are central to interpreting randomized oncology trials. An analysis should identify the estimand, event definition, time origin, censoring rules, analysis population, and statistical model before attaching a clinical meaning to a numerical effect estimate.
14. Multiplicity and the Two Primary Endpoints
COU-AA-302 has two registered primary endpoints: Overall Survival and Radiographic Progression-free Survival. The ClinicalTrials.gov record does not describe the trial's formal multiplicity strategy, alpha allocation, endpoint hierarchy, or testing sequence.
| Endpoint | Registered role | Statistical implication |
|---|---|---|
| Overall Survival | Primary endpoint | Requires prespecified treatment comparison and appropriate type-I-error control within the trial's testing framework. |
| Radiographic Progression-free Survival | Primary endpoint | Requires its own prespecified event definition, censoring rules, and treatment comparison. |
When a trial has multiple primary endpoints, a statistically rigorous interpretation depends on how the protocol defines the relationship among them. For example, a design might require both endpoints to meet a criterion, allow either endpoint to establish success under an alpha-allocation scheme, or use a hierarchical testing procedure. The ClinicalTrials.gov record does not identify which framework COU-AA-302 used.
15. Crossover and Treatment Switching
The ClinicalTrials.gov record contains a separate Placebo to AA category with 40/93 serious adverse events. This indicates that the registry data posted on ClinicalTrials.gov for the page include participants identified as having moved from placebo to AA for purposes of the reported safety information.
However, the ClinicalTrials.gov record does not describe the timing, eligibility criteria, reasons, or statistical handling of that treatment transition. It would therefore be inappropriate to infer how crossover or treatment switching affected the primary efficacy analysis.
16. Missing Data, Censoring, and Follow-up
Time-to-event analysis is built around the possibility that not every participant experiences the event during the observation period. A participant may therefore contribute event-free follow-up and then be censored according to prespecified rules.
For COU-AA-302, the registry-reported endpoint definitions establish the time origin and endpoint-specific follow-up windows but do not provide the complete censoring algorithm. This matters because censoring is not merely a technical footnote: it determines which portions of participant follow-up contribute to the estimated event-time distribution.
The ClinicalTrials.gov record also do not specify missing-data imputation procedures. No imputation method is therefore attributed to the trial.
17. What the Registry Data Do and Do Not Establish
| Question | Supported by the ClinicalTrials.gov record? | Interpretation |
|---|---|---|
| Was the trial randomized? | Yes | The allocation field is RANDOMIZED. |
| Was it parallel-group? | Yes | The design model is PARALLEL. |
| Was it quadruple-masked? | Yes | The masking field is QUADRUPLE. |
| How many participants were enrolled? | Yes | 1088.0. |
| How many primary endpoints were registered? | Yes | 2. |
| Are the primary endpoints time-to-event? | Yes | Overall Survival and rPFS. |
| Were results posted? | Yes | The ClinicalTrials.gov record says Yes. |
| Are formal statistical analyses reported here? | No | No formal statistical analyses were posted. |
| Is an OS hazard ratio reported? | No | Not present in the ClinicalTrials.gov record. |
| Is an rPFS hazard ratio reported? | No | Not present in the ClinicalTrials.gov record. |
| Are confidence intervals reported? | No | Not present in the ClinicalTrials.gov record. |
| Are p-values reported? | No | Not present in the ClinicalTrials.gov record. |
| Are serious adverse-event counts reported? | Yes | 215/542, 156/540, and 40/93 for the listed groups. |
18. Limitations and Interpretation Issues
- Empty statistical-analysis field: the posted statistical analyses contains no formal numerical comparisons, so efficacy effect estimates cannot be reconstructed from the ClinicalTrials.gov record.
- Posted results are not equivalent to registry-reported analysis estimates: the record says results are posted and lists 11 outcome measures, but the numerical outcome-analysis details are not included in the ClinicalTrials.gov recordset.
- Incomplete rPFS definition: the registry definition ends partway through the bone-scan progression criteria. The remainder is not inferred.
- Unknown analysis population: the ClinicalTrials.gov record does not identify the exact population used for the primary efficacy analyses.
- Unknown multiplicity strategy: two primary endpoints are identified, but the ClinicalTrials.gov record does not specify how type I error was controlled across them.
- Unknown model specification: the ClinicalTrials.gov record does not establish whether Kaplan-Meier, log-rank, Cox regression, or another specific model was used for the formal trial analysis.
- Unknown censoring rules: the registered time frames are available, but the complete censoring algorithm is not.
- Safety-group interpretation: the separate Placebo to AA category cannot be fully interpreted without information about the timing and circumstances of treatment switching.
- No subgroup results: subgroup estimates and interaction tests are not included in the ClinicalTrials.gov record.
- No baseline table: baseline demographic and disease characteristics are not reported and therefore are not reconstructed.
19. Why This Trial Matters Statistically
COU-AA-302 provides a useful framework for learning how a randomized oncology trial can be translated into a statistical analysis plan even when the available record does not contain the complete numerical analysis.
| Concept | How it appears in COU-AA-302 |
|---|---|
| Randomization | The allocation field identifies a randomized trial. |
| Parallel-group design | The design model is PARALLEL. |
| Blinding | The registry identifies QUADRUPLE masking. |
| Time-to-event analysis | Both registered primary endpoints are classified as time-to-event. |
| Overall Survival | Time from randomization to death from any cause. |
| Radiographic PFS | Time from randomization to first radiographic progression or the Month 18 cutoff. |
| Kaplan-Meier estimation | A standard method for describing time-to-event distributions. |
| Hazard ratio | A standard relative-effect measure that could be used for a two-group time-to-event comparison. |
| Censoring | Essential to time-to-event analysis when the event is not observed during follow-up. |
| Multiplicity | Two primary endpoints make the prespecified type-I-error framework important. |
| Safety analysis | Serious adverse-event counts are posted on ClinicalTrials.gov for three reported groups. |
The key statistical lesson is that an endpoint label alone is not a complete analysis. To understand a randomized time-to-event comparison, the analyst needs the treatment-assignment framework, endpoint definition, time origin, follow-up window, censoring rules, analysis population, effect measure, uncertainty interval, testing procedure, and multiplicity strategy.
20. Statistical Interpretation Without an Effect Estimate
A hazard ratio is often the most recognizable numerical summary of a two-group time-to-event analysis, but it cannot be responsibly reconstructed from enrollment, endpoint names, or safety counts. Without the underlying event and censoring data—or an explicitly reported statistical analysis—there is no valid basis for assigning a numerical HR to COU-AA-302 from the ClinicalTrials.gov record.
The fact that a trial is phase 3 and has posted results does not establish a particular hypothesis-test result. A p-value must come from an identified statistical test and analysis population. The ClinicalTrials.gov record contains no such numerical test result.
The serious-adverse-event counts provide safety information, but they do not substitute for efficacy analysis. Safety and efficacy answer different questions and should not be collapsed into a single unreported benefit-risk statistic.
A rigorous analysis is not defined by the number of numerical estimates displayed. It also requires identifying what the endpoint measures, what the design permits the comparison to mean, what assumptions a proposed model would require, and where the available evidence stops. Keeping those boundaries explicit prevents unsupported conclusions from being presented as trial results.
21. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registry establishes a randomized, parallel, quadruple-masked phase 3 design with two primary time-to-event endpoints. It does not supply the formal numerical efficacy analyses needed to quantify the treatment effect.
Clinical interpretation
The trial addresses treatment in asymptomatic or mildly symptomatic patients with metastatic castration-resistant prostate cancer, with Overall Survival and radiographic progression-free survival specified as the primary outcomes. The ClinicalTrials.gov record does not support an efficacy conclusion beyond those registered questions.
This distinction is important because a clinical question can be well defined even when the numerical answer is not contained in the dataset available for a particular analysis. A statistical reviewer should not turn missing evidence into an implied result.
22. Related Statistical Concepts
Learn more about the methods used in this trial:
23. Related Statistical Calculators
24. Sources
- ClinicalTrials.gov: COU-AA-302 (NCT00887198).
- Linked publication record: PubMed record, PMID 42286315.
- Linked publication record: PubMed record, PMID 34373138.
- Linked publication record: PubMed record, PMID 28939004.
- Linked publication record: PubMed record, PMID 27402060.
- Linked publication record: PubMed record, PMID 26609008.
Continue through the Clinical Biostats statistical pathway
Move from the trial's design and endpoints to deeper tutorials on randomized trials, survival analysis, censoring, hazard ratios, and related statistical methods.
25. Record Summary
COU-AA-302 is a completed randomized, parallel-group, quadruple-masked phase 3 treatment trial in asymptomatic or mildly symptomatic patients with metastatic castration-resistant prostate cancer. The ClinicalTrials.gov record reports 1088.0 enrolled participants, two arms, two registered primary time-to-event endpoints, and 11 posted outcome measures. Overall Survival is defined from randomization to death from any cause through Month 60, while Radiographic Progression-free Survival is defined from randomization to first radiographic progression or the Month 18 cutoff. The ClinicalTrials.gov record reports serious adverse events of 215/542 for abiraterone acetate plus prednisone, 156/540 for placebo, and 40/93 for the Placebo to AA group.
The most important statistical limitation is equally explicit: although the ClinicalTrials.gov record says that results are posted, its the posted statistical analyses contains no formal numerical analyses. Instead, it explains how the registered time-to-event endpoints would normally be analyzed and identifies the design features that determine how such results should be interpreted.