← Clinical Trials
Metastatic Castration-Resistant Prostate Cancer Phase 3 Randomized NCT02987543

PROfound: Complete Statistical Analysis of Olaparib in Metastatic Castration-Resistant Prostate Cancer

An independent statistical analysis of the randomized phase 3 PROfound trial comparing olaparib with investigators' choice of enzalutamide or abiraterone acetate in men with metastatic castration-resistant prostate cancer.

Study start: 2017-02-06  ·  Primary completion: 2019-06-04  ·  Enrollment: 387
Scope of this record

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

PROfound was a randomized, parallel, open-label phase 3 treatment trial with 387 participants evaluating olaparib against investigators' choice of a non-hormonal therapy represented in the registry intervention data by enzalutamide or abiraterone acetate. The registry reports one primary time-to-event endpoint and four additional outcome analyses, using Cox proportional-hazards and logistic regression methods.

387
Enrollment
Participants
2
Arms
Parallel design
0.34
Primary rPFS HR
95% CI 0.25–0.47
<0.0001
Primary P-value
Superiority analysis
FeaturePROfound
Trial namePROfound
NCT IDNCT02987543
PhasePhase 3
ConditionMetastatic Castration-resistant Prostate Cancer
PopulationMen with metastatic castration-resistant prostate cancer
AllocationRandomized
Design modelParallel
MaskingNone
Primary purposeTreatment
Enrollment387
Arms2
StatusCompleted
Lead sponsorAstraZeneca
Sponsor typeIndustry
Study start2017-02-06
Primary completion2019-06-04
Results postedYes

2. Clinical Question

The central statistical question is whether olaparib produces a superior outcome compared with investigators' choice of NHA in men with metastatic castration-resistant prostate cancer, using the registered primary time-to-event endpoint of radiological progression-free survival by blinded independent central review in Cohort A.

Population

Men with metastatic castration-resistant prostate cancer, as specified by the trial's brief title and condition.

Intervention

Olaparib 300 mg bd, as named in the statistical analysis groups.

Comparator

Investigators' choice of NHA. The registry intervention data identify enzalutamide and abiraterone acetate as comparator drugs.

Primary question

Does olaparib improve radiological progression-free survival compared with investigators' choice of NHA in Cohort A?

3. Trial Design

01
Randomize 387 participants
02
Parallel arms 2 treatment groups
03
Treatment Olaparib vs investigators' choice of NHA
04
Assess Time-to-event and response outcomes
05
Analyze Cox and logistic regression
Allocation
Randomized allocation in a parallel design.
Masking
None. The registry classifies the study as unmasked.
Primary purpose
Treatment.
Hypothesis
Superiority.
TREATMENT GROUP

Cohort A Olaparib 300 mg bd

  • Olaparib
  • 300 mg bd
  • Primary rPFS comparison in Cohort A
  • Also included in the reported Cohort A+B safety population
COMPARATOR GROUP

Cohort A Investigators' Choice of NHA

  • Investigators' choice of NHA
  • Registry intervention data identify enzalutamide and abiraterone acetate
  • Primary rPFS comparison in Cohort A
  • Also included in the reported Cohort A+B safety population
Important design distinction: the registry's primary analysis is specifically identified as a Cohort A comparison, while one secondary rPFS analysis is labelled Cohort A+B. The analysis populations and cohort labels should therefore be retained rather than treating all reported estimates as if they came from one identical population.

4. Endpoints

The registry identifies one primary endpoint and reports four additional statistical analyses. The primary endpoint is a time-to-event measure; the secondary analyses include two additional time-to-event outcomes, one binary response outcome, and overall survival.

RoleEndpointTime frameType
Primary Radiological Progression Free Survival (rPFS) by Blinded Independent Central Review (BICR) - Cohort A Only Tumor assessments every 8 weeks from randomisation until radiographic progression assessed by BICR (median duration of treatment of 7 and 4 months for Olaparib and Investigators Choice of NHA respectively). Time-to-event
Secondary Confirmed Objective Response Rate (ORR) by Blinded Independent Central Review (BICR) - Cohort A Only Tumor assessments every 8 weeks from randomisation until radiographic progression assessed by BICR (median duration of treatment of 7 and 4 months for Olaparib and Investigators Choice of NHA respectively). Binary
Secondary Radiological Progression Free Survival (rPFS) by Blinded Independent Central Review (BICR) - Cohort A+B Tumor assessments every 8 weeks from randomisation until radiographic progression assessed by BICR (median duration of treatment of 7 and 4 months for Olaparib and Investigators Choice of NHA respectively). Time-to-event
Secondary Time to Pain Progression - Cohort A Only Every 4 weeks from randomisation (for 7 consecutive days) throughout the study (median duration of treatment of 7 and 4 months for Olaparib and Investigators Choice of NHA respectively). Time-to-event
Secondary Overall Survival (OS) - Cohort A Only Approximately 35 months after the first patient was randomised. Time-to-event

Primary endpoint definition

The registry definition states that rPFS is the time from randomisation until the date of objective radiological disease progression or death by any cause in the absence of progression, regardless of whether the patient withdrew from randomised therapy or received another anti-cancer therapy prior to progression. The registry specifies RECIST 1.1 for soft tissue and PCWG-3 for bone, with the registry-reported definition continuing into its progression criteria.

5. Statistical Methodology

Cox proportional-hazards model

The registry reports a Cox regression for the primary rPFS analysis and for the secondary rPFS, time-to-pain-progression, and OS analyses. Cox regression is appropriate for time-to-event data because it models the relative hazard between groups while accommodating different follow-up times and right-censored observations.

Conceptual Cox model
h(t | X) = h0(t) exp(βX)

The hazard ratio is obtained from the regression coefficient. An HR below 1 indicates a lower estimated instantaneous event hazard for the first group relative to the comparator under the fitted model.

Hazard ratio

The primary rPFS analysis uses the hazard ratio as its effect measure. The same effect measure is used for the two other reported rPFS/time-to-event analyses and for OS.

Hazard-ratio interpretation
HR = exp(β)

An HR of 0.34, for example, corresponds to an estimated hazard approximately 34% as large as that in the comparator group under the Cox model. Equivalently, 1 − 0.34 = 0.66, so the estimated hazard is 66% lower. This is a relative hazard interpretation, not a statement that 66% of patients are protected or that survival probability is reduced by exactly 66% at every time point.

Logistic regression

Confirmed objective response rate was analyzed using logistic regression in participants evaluable for response. The reported effect measure was an odds ratio.

Odds-ratio interpretation
OR = odds of response in treatment group ÷ odds of response in comparator group

An odds ratio above 1 indicates higher estimated odds of the binary response outcome in the first group. The odds ratio is not the same as a risk ratio or a difference in response probabilities.

Intention-to-treat principle

The registry identifies intention-to-treat analysis as an analysis concept for the reported time-to-event analyses. In an intention-to-treat framework, randomized participants are analyzed according to their assigned group rather than being reclassified based on later treatment exposure.

Analysis populations

The primary rPFS analysis used the Full Analysis Set. The secondary ORR analysis used participants evaluable for response. The secondary Cohort A+B rPFS and time-to-pain-progression analyses used the Full Analysis Set, while the OS analysis used the Full analysis set.

AnalysisPopulationMethodEffect measure
Primary rPFS, Cohort AFull Analysis SetCox proportional-hazards modelHazard ratio
Confirmed ORR, Cohort AEvaluable for responseLogistic regressionOdds ratio
rPFS, Cohort A+BFull Analysis SetCox proportional-hazards modelHazard ratio
Time to Pain Progression, Cohort AFull Analysis SetCox proportional-hazards modelHazard ratio
OS, Cohort AFull analysis setCox proportional-hazards modelHazard ratio

6. Primary Result: Radiological Progression-Free Survival

The primary endpoint was radiological progression-free survival by blinded independent central review in Cohort A. The registry reports a superiority comparison using a Cox proportional-hazards model in the Full Analysis Set.

Primary rPFS hazard ratio

0.34

95% CI: 0.25–0.47   ·   P < 0.0001

Cohort A: olaparib 300 mg bd vs investigators' choice of NHA

Primary endpointOlaparib groupComparatorEffect estimateP-value
rPFS by BICR, Cohort A Only Cohort A Olaparib 300 mg bd Cohort A Investigators' Choice of NHA HR 0.34 (95% CI 0.25–0.47) <0.0001
Clinical Biostats interpretation

What the estimate means: an HR of 0.34 means that, under the reported Cox model, the estimated instantaneous rate of the rPFS event was approximately 34% of the comparator hazard. Put another way, 1 − 0.34 = 0.66, so the estimated hazard was 66% lower for the olaparib group relative to investigators' choice of NHA.

What it does not mean: it does not mean that 66% of participants avoided progression, that each participant experienced a 66% reduction in risk, or that the probability of progression was exactly 66% lower at every time point.

What the confidence interval says: the two-sided 95% CI of 0.25–0.47 describes statistical uncertainty around the estimated hazard ratio under the model and sampling framework. It does not describe the range of individual patient effects.

Why the p-value is different from effect size: P < 0.0001 addresses evidence against the null hypothesis under the specified statistical framework. It does not measure the magnitude or clinical importance of the treatment effect. The HR and its confidence interval provide the effect-size information.

Important modeling caution: interpretation of a single Cox HR depends on the proportional-hazards assumption. The ClinicalTrials.gov record does not report a formal assessment of that assumption, so the HR should be understood as the reported model-based summary rather than as a statement that the relative hazard must be identical at every time point.

Population caution: the primary analysis is explicitly limited to Cohort A and the Full Analysis Set. The primary result should not be silently substituted for the separate Cohort A+B secondary analysis.

7. Secondary Results

Confirmed Objective Response Rate

Confirmed objective response rate by blinded independent central review in Cohort A was analyzed among participants evaluable for response using logistic regression. The reported effect measure was an odds ratio.

Odds ratio for confirmed objective response

20.86

95% CI: 4.18–379.18   ·   P < 0.0001

Cohort A: olaparib 300 mg bd vs investigators' choice of NHA

Clinical Biostats interpretation

An OR of 20.86 means that the estimated odds of confirmed objective response were 20.86 times the comparator odds in the reported logistic-regression analysis. Because this is an odds ratio, it should not be described as meaning that the response probability was 20.86 times higher.

The 95% CI of 4.18–379.18 is very wide. That width indicates substantial statistical uncertainty around the estimated odds ratio, even though the interval is entirely above 1. The p-value of <0.0001 addresses the evidence against the null value of 1; it does not make the wide confidence interval narrow or precisely quantify the magnitude of benefit.

The analysis population is also different from the Full Analysis Set used for the primary rPFS analysis: the ORR analysis used participants evaluable for response. That distinction matters when comparing estimates across endpoints.

Radiological Progression-Free Survival: Cohort A+B

The registry also reports a secondary rPFS analysis labelled Cohort A+B. It used the Full Analysis Set and a Cox proportional-hazards model.

Cohort A+B rPFS hazard ratio

0.49

95% CI: 0.38–0.63   ·   P < 0.0001

Groups compared: Cohort A Olaparib 300 mg bd vs Cohort A Investigators' Choice of NHA

Clinical Biostats interpretation

The reported HR of 0.49 corresponds to an estimated hazard approximately 49% as large as the comparator hazard under the Cox model, or an estimated 51% lower hazard. The 95% CI of 0.38–0.63 describes uncertainty around that estimate.

The result is statistically distinct from the primary Cohort A estimate of 0.34 because it represents a different registered analysis labelled Cohort A+B. The two estimates should therefore not be treated as duplicate measurements of exactly the same endpoint population.

The p-value of <0.0001 is evidence against the null hypothesis in the reported analysis; it is not a measure of the size of the hazard reduction.

Time to Pain Progression

Time to pain progression in Cohort A was analyzed using a Cox proportional-hazards model in the Full Analysis Set. The registry describes assessments every 4 weeks from randomisation, for 7 consecutive days, throughout the study.

Hazard ratio for time to pain progression

0.44

95% CI: 0.22–0.91   ·   P = 0.0192

Cohort A: olaparib 300 mg bd vs investigators' choice of NHA

Clinical Biostats interpretation

An HR of 0.44 corresponds to an estimated instantaneous event hazard approximately 44% as large as the comparator hazard under the Cox model, or an estimated 56% lower hazard.

The two-sided 95% CI of 0.22–0.91 indicates uncertainty around the estimate. Because the interval extends substantially below 1 and its upper endpoint is 0.91, the estimate is compatible with a range of relative effects rather than a single fixed treatment effect.

The p-value of 0.0192 describes evidence against the null hypothesis in this reported analysis. It should not be interpreted as a 1.92% probability that the null hypothesis is true, nor as a measure of the magnitude of the treatment effect.

As with other Cox-model results, the interpretation is conditional on the model and its assumptions, including the proportional-hazards framework.

Overall Survival

Overall survival in Cohort A was analyzed approximately 35 months after the first patient was randomised. The Full analysis set was used with a Cox proportional-hazards model.

Overall survival hazard ratio

0.69

95% CI: 0.50–0.97   ·   P = 0.0175

Cohort A: olaparib 300 mg bd vs investigators' choice of NHA

Clinical Biostats interpretation

An HR of 0.69 means that the estimated instantaneous hazard of death was approximately 69% of the comparator hazard under the reported Cox model, corresponding to an estimated 31% lower hazard.

The 95% CI of 0.50–0.97 indicates uncertainty around the estimated HR. It does not mean that individual treatment effects fall between 0.50 and 0.97, and it does not directly provide an interval for survival probabilities.

The p-value of 0.0175 provides evidence against the null hypothesis in the reported analysis, but it is not a measure of effect size. The HR and confidence interval are the appropriate reported quantities for describing the relative treatment effect and its statistical precision.

Because OS is a time-to-event endpoint, censoring and the timing of events are integral to the analysis. The ClinicalTrials.gov record does not provide enough information to reconstruct a Kaplan-Meier curve or describe median OS, so neither is added here.

8. Results Summary

EndpointAnalysis populationMethodEffect95% CIP-value
Primary rPFS, Cohort A Full Analysis Set Cox proportional-hazards model HR 0.34 0.25–0.47 <0.0001
Confirmed ORR, Cohort A Evaluable for response Logistic regression OR 20.86 4.18–379.18 <0.0001
rPFS, Cohort A+B Full Analysis Set Cox proportional-hazards model HR 0.49 0.38–0.63 <0.0001
Time to Pain Progression, Cohort A Full Analysis Set Cox proportional-hazards model HR 0.44 0.22–0.91 0.0192
OS, Cohort A Full analysis set Cox proportional-hazards model HR 0.69 0.50–0.97 0.0175

Across the five posted statistical analyses, the registry reports hazard ratios below 1 for each time-to-event comparison and an odds ratio above 1 for confirmed objective response. These estimates use different endpoints and, for ORR, a different analysis population, so they should be interpreted separately rather than combined into one overall numerical measure.

9. Understanding the Time-to-Event Endpoints

PROfound's primary endpoint and three of its four secondary analyses are time-to-event outcomes. These outcomes are not analyzed like a simple comparison of proportions because participants can have different follow-up times and some participants may not experience the event during observation.

rPFS

The primary endpoint combines radiological progression and death into a time-to-event outcome, with radiological assessment by blinded independent central review.

Time to pain progression

This is another time-to-event endpoint, with registry-specified assessments every 4 weeks from randomisation for 7 consecutive days.

Overall survival

OS is a time-to-event endpoint for death from any cause. The registry reports its Cohort A analysis approximately 35 months after the first patient was randomised.

Right censoring

Time-to-event methods allow information from participants who have not yet experienced the event at the end of their observable follow-up to contribute up to their censoring time.

Why a hazard ratio is not a median difference

A hazard ratio summarizes relative instantaneous event hazards under the Cox model. A median time-to-event estimate answers a different question: the time at which the estimated event-free probability reaches 50%. The registry-reported PROfound data provide hazard ratios and confidence intervals but do not provide median rPFS, median time to pain progression, or median OS values. Those quantities are therefore not inferred from the hazard ratios.

10. Why the Primary and Secondary rPFS Analyses Are Not Interchangeable

The primary endpoint is explicitly labelled Cohort A Only, while the secondary rPFS analysis is labelled Cohort A+B. Even though both use BICR, rPFS, the Full Analysis Set, and Cox proportional-hazards modeling, their cohort labels are different.

FeaturePrimary rPFSSecondary rPFS
Cohort labelCohort A OnlyCohort A+B
Analysis populationFull Analysis SetFull Analysis Set
MethodCox proportional-hazards modelCox proportional-hazards model
Effect measureHazard ratioHazard ratio
Estimate0.340.49
95% CI0.25–0.470.38–0.63
P-value<0.0001<0.0001

The appropriate statistical reading is therefore to preserve the endpoint and cohort definitions attached to each estimate. Similar methodology does not make two analyses identical.

11. Safety Results

The ClinicalTrials.gov record reports serious adverse events by arm for the combined Cohort A+B population. The reported measure is the number affected divided by the number at risk.

Safety populationAffectedAt risk
Cohort A+B Olaparib 300 mg bd 94 256
Cohort A+B Investigators' Choice of NHA 39 130

Olaparib

Serious adverse events were reported in 94 of 256 participants at risk in the Cohort A+B olaparib 300 mg bd population.

Investigators' choice of NHA

Serious adverse events were reported in 39 of 130 participants at risk in the Cohort A+B investigators' choice of NHA population.

Safety interpretation: the ClinicalTrials.gov record provides affected and at-risk counts for serious adverse events, but they do not provide a statistical comparison, confidence interval, or p-value for this safety measure. No comparative safety effect estimate is therefore calculated or implied here.

12. Statistical Methods Explained

Why was a Cox proportional-hazards model used?

The registry classifies rPFS, time to pain progression, and OS as time-to-event endpoints and reports Cox regression for each. The Cox model is designed to use both event occurrence and event timing while allowing censored observations to contribute information up to their follow-up time.

What does an HR of 0.34 mean?

An HR of 0.34 means that the fitted Cox model estimates the instantaneous event hazard in the olaparib group at approximately 34% of the comparator hazard. It does not mean that 34% of participants experienced the event, nor that the probability of an event was 34% at every point in time.

Why is the confidence interval important?

The confidence interval shows the statistical uncertainty around an effect estimate. For the primary rPFS HR of 0.34, the 95% CI is 0.25–0.47. The interval is therefore part of the result, not an optional supplement to the point estimate.

Why does the p-value not measure effect size?

The p-value evaluates the compatibility of the observed data with a null hypothesis under the specified statistical model. It does not tell us how large the treatment effect is. The effect size is described by the HR or OR, while the confidence interval describes uncertainty around that estimate.

Why is the OR for response so different from the HRs?

The confirmed ORR analysis is binary: a participant is classified according to the response outcome. Its effect measure is therefore an odds ratio. The rPFS and OS analyses incorporate time until an event and use hazard ratios. Odds ratios and hazard ratios have different definitions and should not be numerically compared as though they were the same quantity.

Why is the ORR confidence interval so wide?

The reported OR confidence interval is 4.18–379.18, substantially wider than the intervals around the reported hazard ratios. A wide interval indicates considerable statistical uncertainty about the magnitude of the estimated odds ratio. The point estimate alone should not be used to conceal that uncertainty.

Why does the Cohort A+B rPFS result need separate interpretation?

The primary rPFS analysis is labelled Cohort A Only, whereas the secondary rPFS analysis is labelled Cohort A+B. Even though both analyses use Cox regression and the Full Analysis Set, their registered cohort definitions differ. Keeping those labels attached to the results prevents accidental mixing of estimates from different analysis populations.

13. P-values, Confidence Intervals, and Statistical Evidence

The five posted statistical analyses all include an effect estimate, a two-sided 95% confidence interval, and a p-value. This provides three complementary pieces of information.

ComponentWhat it tells usWhat it does not tell us
Point estimateThe estimated relative treatment effect under the specified model.The exact effect for every participant.
95% confidence intervalStatistical uncertainty around the point estimate.The range of individual patient responses.
P-valueEvidence against the specified null hypothesis under the analysis framework.The probability that the null hypothesis is true or the size of the treatment effect.

For example, the primary rPFS result combines an HR of 0.34, a 95% CI of 0.25–0.47, and P < 0.0001. These numbers answer related but different statistical questions and should be interpreted together.

14. Superiority Testing

The registry identifies the hypothesis type for the posted analyses as superiority. This means the statistical question is whether the treatment groups differ in the direction specified by the superiority hypothesis rather than whether a new treatment merely remains within a predefined non-inferiority margin.

Primary rPFS

The superiority analysis reports HR 0.34 with a two-sided 95% CI of 0.25–0.47 and P < 0.0001.

ORR

The superiority analysis reports OR 20.86 with a two-sided 95% CI of 4.18–379.18 and P < 0.0001.

Time to pain progression

The superiority analysis reports HR 0.44 with a two-sided 95% CI of 0.22–0.91 and P = 0.0192.

OS

The superiority analysis reports HR 0.69 with a two-sided 95% CI of 0.50–0.97 and P = 0.0175.

No non-inferiority margin is reported in the ClinicalTrials.gov record. The statistical analyses are explicitly identified as superiority analyses, so a non-inferiority margin is not introduced or inferred for this page.

15. Multiplicity, Interim Analysis, and Other Design Features

The ClinicalTrials.gov record identifies the hypothesis type as superiority and provide the five posted statistical analyses, but they do not report an alpha-allocation scheme, multiplicity adjustment procedure, interim-analysis schedule, stopping boundary, or information fraction.

Design topicWhat the ClinicalTrials.gov record supports
MultiplicityThe registry reports five statistical analyses, but the ClinicalTrials.gov record does not specify a multiplicity-adjustment procedure or alpha allocation.
Interim analysisNo interim-analysis schedule or stopping boundary is reported in the ClinicalTrials.gov record.
Alpha spendingNo alpha-spending method is reported.
Non-inferiority marginNot applicable to the reported hypothesis type; analyses are identified as superiority.
Factorial designThe registry classifies the design model as parallel, not factorial.
CrossoverNo crossover information is reported.
Missing-data/imputation strategyNo imputation strategy is reported.
Bayesian methodsNo Bayesian method is reported; the normalized methods are Cox proportional-hazards model and logistic regression.
Stratification factorsNo stratification factors are reported in the ClinicalTrials.gov record.
Statistical discipline: absence of a reported method in the registry extract is not evidence that the underlying protocol or statistical analysis plan lacked that method. It means only that the method is not available in the ClinicalTrials.gov record.

16. What the Hazard Ratios Do — and Do Not — Mean

Primary rPFS

The HR of 0.34 is a relative measure of the modeled instantaneous event hazard. It corresponds to an estimated 66% lower hazard for the olaparib group relative to the comparator under the reported Cox model.

It does not mean that 66% of participants were progression-free, that 66% of participants benefited, or that every participant experienced the same proportional reduction in risk.

Overall survival

The OS HR of 0.69 corresponds to an estimated 31% lower instantaneous hazard of death under the reported Cox model. It is not a 31-percentage-point increase in survival probability and does not provide a median survival time.

Confidence intervals

The primary rPFS CI of 0.25–0.47, the Cohort A+B rPFS CI of 0.38–0.63, the time-to-pain-progression CI of 0.22–0.91, and the OS CI of 0.50–0.97 quantify uncertainty around four different Cox-model estimates.

17. Comparing the Effect Measures

The trial provides both hazard ratios and an odds ratio. This is useful statistically because it demonstrates why effect measures must be matched to endpoint structure.

Effect measureEndpointInterpretation
Hazard ratiorPFS, time to pain progression, OSRelative instantaneous event hazard under a Cox model.
Odds ratioConfirmed ORRRatio of the odds of response between groups under logistic regression.

An HR and an OR cannot be placed on the same numerical scale. For example, the OR of 20.86 for confirmed ORR should not be described as a larger or smaller treatment effect than an HR of 0.34 simply because the numbers have different magnitudes. They measure different quantities.

18. Limitations

19. Why This Trial Matters Statistically

PROfound is a useful teaching case because the ClinicalTrials.gov record connects several core clinical-trial methods within one randomized oncology study: a time-to-event primary endpoint, Cox proportional-hazards modeling, an odds-ratio analysis of a binary response endpoint, intention-to-treat analysis concepts, separate analysis populations, and cohort-specific efficacy analyses.

ConceptHow it appears in PROfound
RandomizationThe study is classified as randomized.
Parallel designThe design model is parallel with 2 arms.
Time-to-event endpointThe registered primary endpoint is rPFS, and the registry also reports time to pain progression and OS.
Blinded independent central reviewThe primary rPFS endpoint and confirmed ORR are assessed by BICR.
Cox proportional-hazards modelUsed for the primary rPFS, secondary rPFS, time to pain progression, and OS analyses.
Hazard ratioReported for all four time-to-event analyses.
Logistic regressionUsed for confirmed objective response rate.
Odds ratioReported as the effect measure for confirmed ORR.
Intention-to-treat analysisIdentified as an analysis concept for the reported time-to-event analyses.
Analysis populationsFull Analysis Set, Full analysis set, and evaluable-for-response populations are explicitly identified.
SuperiorityThe registry identifies superiority as the hypothesis type.
Safety by armSerious adverse events are reported as affected/at-risk counts for Cohort A+B.

20. Statistical Methods in Context

The most important lesson from the PROfound statistical record is that a clinical-trial result is not a single number. The primary rPFS result combines an endpoint definition, a cohort, an analysis population, a regression model, an effect measure, a confidence interval, and a p-value. Each component answers a different part of the statistical question.

Endpoint

Defines what event is being studied and when follow-up begins and ends.

Population

Defines which participants contribute to the analysis.

Model

Defines how the observed data are translated into an estimated treatment effect.

Effect measure

Communicates the relative magnitude of the estimated treatment difference.

This is why the statement "the HR was 0.34" is incomplete without identifying that it refers to radiological progression-free survival by BICR in Cohort A, analyzed in the Full Analysis Set using a Cox proportional-hazards model.

21. Statistical Concepts in This Trial

Learn more about the methods used in this trial:

22. Related Statistical Calculators

23. Sources

Continue exploring clinical-trial statistics

Connect the PROfound endpoints and statistical methods to focused tutorials, calculators, and additional clinical-trial analyses.

24. Record Summary

PROfound provides a compact example of several fundamental principles in clinical-trial statistics. The trial was randomized, parallel, open-label, phase 3, and designed for treatment. Its primary endpoint was a time-to-event measure of radiological progression-free survival by blinded independent central review in Cohort A. The reported primary analysis used the Full Analysis Set and a Cox proportional-hazards model, producing an HR of 0.34 with a two-sided 95% CI of 0.25–0.47 and P < 0.0001.

The secondary analyses illustrate why endpoint-specific statistical interpretation matters. Confirmed ORR used logistic regression and an odds ratio of 20.86, while rPFS in Cohort A+B, time to pain progression in Cohort A, and OS in Cohort A used Cox regression with HRs of 0.49, 0.44, and 0.69, respectively. These values cannot be treated as interchangeable because they describe different endpoints, cohort definitions, and, for ORR, a different analysis population.

The ClinicalTrials.gov record reports serious adverse events in 94/256 participants for Cohort A+B olaparib 300 mg bd and 39/130 for Cohort A+B investigators' choice of NHA. No formal comparative safety statistic is reported. Likewise, the registry extract does not provide enough information to reconstruct median survival times, Kaplan-Meier curves, stratification factors, multiplicity procedures, interim boundaries, or missing-data methods, so those elements are not inferred.

Clinical Biostats methodology: A rigorous trial-results page should preserve the distinction between reported evidence and statistical interpretation. For PROfound, that means keeping cohort labels, analysis populations, endpoint definitions, effect measures, confidence intervals, and p-values attached to the exact analyses from which they were reported.