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
OlympiAD was a randomized, parallel-group, open-label phase 3 treatment trial comparing olaparib monotherapy with physician's choice chemotherapy in metastatic breast cancer patients with germline BRCA1/2 mutations.
| Feature | OlympiAD |
|---|---|
| Trial name | OlympiAD |
| NCT ID | NCT02000622 |
| Phase | Phase 3 |
| Status | COMPLETED |
| Enrollment | 302 |
| Allocation | RANDOMIZED |
| Design model | PARALLEL |
| Masking | NONE |
| Primary purpose | TREATMENT |
| Lead sponsor | AstraZeneca |
| Sponsor type | INDUSTRY |
| Start | 2014-03-27 |
| Primary completion | 2016-12-09 |
2. Clinical Question
The trial addresses whether olaparib monotherapy produces a different progression-free survival experience than physician's choice chemotherapy in metastatic breast cancer patients with germline BRCA1/2 mutations.
Population
Metastatic breast cancer patients with BRCA1 or BRCA2 gene mutations, as described by the trial conditions and registered primary endpoint population.
Intervention
Olaparib 300 mg bd.
Comparator
Physician's choice chemotherapy.
Primary question
Does olaparib change progression-free survival compared with physician's choice chemotherapy under the prespecified superiority analysis?
3. Trial Design
Olaparib monotherapy
- Olaparib 300 mg bd.
- Randomized treatment arm.
- Serious adverse events affected 37 of 205 patients at risk in the reported safety analysis.
Physician's choice chemotherapy
- Physician's choice chemotherapy.
- Randomized comparator arm.
- Serious adverse events affected 15 of 91 patients at risk in the reported safety analysis.
The registry describes the allocation as randomized and the design model as parallel, with no masking. That combination is statistically important: randomization establishes the framework for comparing outcomes between treatment assignments, while the absence of masking can be relevant when outcomes involve assessments or patient-reported measures.
4. Trial Population and Analysis Populations
The primary PFS analysis used the Full Analysis Set (FAS), consisting of all randomized patients. This is explicitly important for interpreting the primary hazard ratio because the treatment comparison is anchored to randomized assignment rather than restricted to patients who remained on treatment.
| Analysis population | Definition / role |
|---|---|
| Full Analysis Set (FAS) | All randomized patients; used for the primary PFS analysis and the reported secondary time-to-event analyses. |
| QoL analysis subset | Randomized patients with an evaluable baseline EORTC QLQ-C30 assessment and at least one evaluable post-baseline assessment. |
| Safety population represented in SAE data | Olaparib: 37/205 serious adverse events; chemotherapy: 15/91. |
5. Endpoints
| Endpoint | Registry definition / time frame | Type |
|---|---|---|
| Primary PFS | Progression-free Survival (PFS) Using Blinded Independent Central Review (BICR) According to Modified Response Evaluation Criteria In Solid Tumours (RECIST v1.1). Time from randomisation to the earliest of objective radiological progression or death by any cause in the absence of objective progression. Objective radiological progression is defined using RECIST v1.1, as a 20% increase in the sum of the longest diameter of target lesions, or a measurable increase in a non-target lesion, or the appearance of new lesions. | Time-to-event |
| Time to Second Progression or Death (PFS2) | Second progression status reviewed every 8 weeks following the first objective radiological progression as per investigator assessment. Assessed up to a maximum of 30 months. | Time-to-event |
| Overall Survival (OS) | Survival status reviewed every 3 weeks until treatment discontinued, then every 8 weeks. Assessed up to a maximum of 30 months. | Time-to-event |
| Adjusted Mean Change in Global Health Status/QoL Score | EORTC QLQ-C30 assessments performed at baseline then every ~6 weeks until objective radiological disease progression. Assessed up to a maximum of 30 months. | Continuous |
| PFS in Myriad CDx gBRCAm-confirmed patients | Radiological scans performed at baseline then every ~6 weeks up to 24 weeks, then every ~ 12 weeks thereafter until objective radiological disease progression. Assessed up to a maximum of 30 months. | Time-to-event |
| Overall Survival at Final OS | Survival status reviewed every 3 weeks until treatment discontinued, then every 8 weeks. Assessed up to a maximum of 40 months. | Time-to-event |
| Overall Survival at Extended OS | Survival status reviewed every 3 weeks until treatment discontinued, then every 8 weeks until Sep 2017 (final OS DCO), then every 3 months. Assessed up to a maximum of 64 months. | Time-to-event |
| Time to First Subsequent Cancer Therapy or Death (TFST) | Subsequent cancer therapy status reviewed every 8 weeks following study treatment discontinuation. Assessed up to a maximum of 30 months. | Time-to-event |
| Time to Second Subsequent Cancer Therapy or Death (TSST) | Subsequent cancer therapy status reviewed every 8 weeks following study treatment discontinuation. Assessed up to a maximum of 30 months. | Time-to-event |
| TFST at Extended OS | Subsequent cancer therapy status reviewed every 8 weeks following study treatment discontinuation until Sep 2017 (final OS DCO), then every 3 months. Assessed up to a maximum of 64 months. | Time-to-event |
| TSST at Extended OS | Subsequent cancer therapy status reviewed every 8 weeks following study treatment discontinuation until Sep 2017 (final OS DCO), then every 3 months. Assessed up to a maximum of 64 months. | Time-to-event |
6. Statistical Methodology
Primary time-to-event analysis
The registered primary endpoint is progression-free survival, a time-to-event outcome. The formal analysis posted for PFS used a log-rank test in the Full Analysis Set, with a hazard ratio as the effect measure. The analysis text also identifies an intention-to-treat principle and stratification by received prior chemotherapy for metastatic breast cancer (yes/no) and receptor status.
The registry analysis notes that a hazard ratio below 1 favours olaparib. The reported primary HR was 0.58.
Hazard ratio as the effect measure
A hazard ratio compares the estimated instantaneous event rates between the randomized groups within the time-to-event framework. For the primary PFS analysis, the HR of 0.58 corresponds to an estimated hazard in the olaparib group that is approximately 42% lower than the comparator hazard, because 1 − 0.58 = 0.42.
This is a relative time-to-event interpretation. It is not a statement that 42% of patients avoided progression, nor does it directly provide an absolute difference in the probability of progression at a particular time point.
Stratification
The primary analysis was stratified by received prior chemotherapy for metastatic breast cancer (yes/no) and receptor status. Stratification allows the time-to-event comparison to account for these prespecified factors when comparing the randomized groups.
Mixed-effects model for quality of life
The adjusted mean change in global health status/quality of life was analyzed with a mixed-effects model. The registry specifies treatment, visit, and treatment-by-visit interaction, adjusted for baseline global health status/QoL score and baseline score-by-visit interaction.
This is fundamentally different from the survival analysis. The PFS, PFS2, and OS analyses ask about the timing of events, whereas the QoL analysis asks about longitudinal changes in a continuous score across visits.
7. Primary Result: Progression-Free Survival
The registered primary endpoint was PFS assessed by blinded independent central review according to modified RECIST v1.1. The analysis population was the Full Analysis Set consisting of all randomized patients.
Primary PFS hazard ratio
95% CI: 0.43–0.80 · P = 0.0009
Log-rank test · Superiority hypothesis
| Primary endpoint | Olaparib 300 mg bd vs chemotherapy |
|---|---|
| Effect measure | Hazard ratio |
| Estimate | 0.58 |
| 95% confidence interval | 0.43–0.80 |
| P-value | 0.0009 |
| Analysis | Log-rank test |
| Population | Full Analysis Set: all randomized patients |
| Hypothesis | Superiority |
| Stratification | Prior chemotherapy for metastatic breast cancer and receptor status |
The HR of 0.58 indicates a lower estimated instantaneous rate of progression or death in the olaparib group than in the chemotherapy group under the reported time-to-event analysis. Equivalently, the estimated hazard is approximately 42% lower for olaparib relative to chemotherapy, using the simple transformation 1 − HR.
The HR does not mean that 42% of patients benefited, that progression was prevented in 42% of patients, or that each patient's individual risk was reduced by exactly 42%. It is a relative measure of event rates over time.
The 95% CI of 0.43–0.80 describes statistical uncertainty around the estimated HR. It does not describe the range of outcomes for individual patients. Because the entire reported interval is below 1, the estimated direction of the treatment comparison remains below the no-difference value within this confidence interval.
The P-value of 0.0009 addresses evidence against the null hypothesis under the reported testing framework. It does not measure the size of the treatment effect, the probability that the treatment works, or the clinical importance of the HR.
Because this is a time-to-event analysis, the interpretation also depends on the underlying censoring and model framework. The registry reports a log-rank analysis and a hazard ratio, so the HR should not be interpreted as though it were an absolute risk ratio at a fixed time point.
8. Secondary Time-to-Event Results
Time to Second Progression or Death (PFS2)
PFS2 hazard ratio
95% CI: 0.40–0.83 · P = 0.0033
Log-rank test · Full Analysis Set
The registry reports PFS2 as the time to second progression or death, with second progression status reviewed every 8 weeks following the first objective radiological progression. The analysis was performed in all randomized patients using a stratified log-rank test.
The HR of 0.57 means that the estimated instantaneous rate of second progression or death was lower in the olaparib group than in the chemotherapy group under this analysis. The corresponding 95% CI of 0.40–0.83 indicates uncertainty around that relative estimate.
This endpoint is not the same as the primary PFS endpoint. PFS2 incorporates an event after the first progression, so it provides information about the subsequent disease-control pathway rather than simply reproducing the first progression event.
The P-value of 0.0033 is evidence against the null hypothesis under the reported superiority framework. It should not be interpreted as an effect-size measure.
Overall Survival (OS)
Overall survival hazard ratio
95% CI: 0.63–1.29 · P = 0.5665
Log-rank test · Full Analysis Set
The registry reports OS as a secondary time-to-event endpoint, with survival status reviewed every 3 weeks until treatment discontinuation and then every 8 weeks, assessed up to a maximum of 30 months. months in the registry wording.
The reported OS HR of 0.90 is below 1, but the 95% CI of 0.63–1.29 includes 1. The reported P-value of 0.5665 therefore does not provide evidence against the null hypothesis under this superiority analysis.
The HR should not be interpreted as evidence that the treatments have identical survival outcomes. Rather, the reported estimate is accompanied by an interval that includes both values below and above the no-difference value.
The confidence interval is particularly important here because it communicates that the observed estimate of 0.90 is not precise enough to exclude a range of possible relative hazard differences represented by the interval.
PFS in patients confirmed as Myriad CDx gBRCAm
PFS hazard ratio in the specified subgroup
95% CI: 0.41–0.78 · P = 0.0005
Log-rank test · Prespecified secondary analysis
This analysis was restricted to the subset of randomized patients confirmed as Myriad CDx gBRCAm. It used the same general time-to-event framework, with a stratified log-rank analysis and stratification by received prior chemotherapy for metastatic breast cancer and receptor status.
The estimated HR of 0.57 indicates a lower estimated instantaneous rate of PFS events in the olaparib group within this specified subgroup. The 95% CI of 0.41–0.78 quantifies uncertainty around that estimate.
This result should be distinguished from the primary PFS analysis because it uses a restricted analysis population. A subgroup-specific estimate does not automatically establish that the treatment effect differs from the effect in patients outside that subgroup; that question requires an appropriate interaction or heterogeneity analysis.
Overall Survival at Final OS
Final OS hazard ratio
95% CI: 0.66–1.23 · P = 0.5131
Log-rank test · Full Analysis Set
The final OS analysis used the Full Analysis Set and retained the stratified log-rank framework. The reported estimate was 0.90 with a two-sided 95% CI of 0.66–1.23.
The estimate is below 1, but the confidence interval includes 1 and the reported P-value is 0.5131. Thus, within this reported superiority analysis, the data do not exclude the no-difference value.
The distinction between the point estimate and its uncertainty is important: a point estimate alone is not sufficient to characterize the statistical evidence.
Overall Survival at Extended OS
Extended OS hazard ratio
95% CI: 0.67–1.18 · P = 0.4167
Log-rank test · Full Analysis Set
The extended OS analysis followed survival status until September 2017, identified in the registry as the final OS data cutoff. The reported HR was 0.89 with a two-sided 95% CI of 0.67–1.18.
The extended OS estimate remains below 1, but the 95% CI includes 1 and the reported P-value is 0.4167. The result therefore does not establish a statistically significant superiority comparison for OS under the reported analysis.
The successive OS analyses also illustrate why data-cutoff timing matters. Different OS analyses can have different follow-up durations and therefore should be identified by their registry endpoint label rather than silently combined into a single estimate.
9. Quality-of-Life Analysis
Adjusted mean difference in global health status/QoL
95% CI: 2.5–12.4 · P = 0.0035
Mean difference in final values · Mixed-effects model
The adjusted mean change in Global Health Status/Quality of Life was measured using the EORTC QLQ-C30. The analysis population was a subset of the Full Analysis Set consisting of randomized patients with an evaluable baseline assessment and at least one evaluable post-baseline assessment.
| Feature | Reported analysis |
|---|---|
| Outcome | Adjusted Mean Change in Global Health Status/QoL Score from EORTC QLQ-C30 |
| Effect measure | Mean Difference (Final Values) |
| Estimate | 7.5 |
| 95% CI | 2.5–12.4 |
| P-value | 0.0035 |
| Method | Mixed Models Analysis |
| Direction | Mean difference >0 favours olaparib |
| Covariate adjustment | Baseline global health status/QoL score and baseline score-visit interaction |
| Model terms | Treatment, visit, treatment-visit interaction |
The estimated mean difference of 7.5 indicates a higher adjusted final value for the olaparib group relative to chemotherapy under the registry's specified direction, where a positive difference favours olaparib.
The 95% CI of 2.5–12.4 describes uncertainty around the adjusted mean difference. It does not mean that every individual patient's QoL changed by an amount within this interval.
The P-value of 0.0035 addresses the statistical evidence for the comparison under the mixed-model analysis. It does not measure the magnitude or practical importance of the 7.5-point estimate.
The analysis also differs from the primary PFS analysis in its population: it requires evaluable baseline and post-baseline QoL measurements. That restriction should be kept in mind when comparing the QoL result with the FAS-based efficacy results.
10. Other Pre-Specified Time-to-Event Results
Time to First Subsequent Cancer Therapy or Death (TFST)
TFST hazard ratio
95% CI: 0.24–0.47 · P < 0.0001
Supportive analysis to PFS
The registry describes TFST as a supportive analysis to PFS. Subsequent cancer therapy status was reviewed every 8 weeks following study treatment discontinuation. The analysis used the Full Analysis Set and a stratified log-rank test.
The HR of 0.34 corresponds to an estimated hazard of first subsequent cancer therapy or death that is approximately 66% lower in the olaparib group than in the chemotherapy group, using the direct transformation 1 − 0.34.
The endpoint is not identical to PFS. It combines subsequent-treatment initiation and death, so it captures a different part of the patient's treatment pathway.
The 95% CI of 0.24–0.47 quantifies uncertainty around the relative estimate, while the reported P-value of <0.0001 addresses statistical evidence rather than effect magnitude.
Time to Second Subsequent Cancer Therapy or Death (TSST)
TSST hazard ratio
95% CI: 0.38–0.74 · P = 0.0002
Supportive analysis to PFS2
The TSST HR of 0.53 represents a lower estimated hazard of second subsequent cancer therapy or death in the olaparib group. The 95% CI of 0.38–0.74 describes the uncertainty around that relative estimate.
TSST is related to PFS2 but is not interchangeable with it. The registry specifically identifies it as a supportive analysis to PFS2, which provides useful context for understanding its role in the overall statistical evidence.
Extended TFST
TFST at Extended OS
95% CI: 0.27–0.50 · P < 0.0001
Supportive analysis to PFS
The extended TFST analysis produced an HR of 0.36, with a 95% CI of 0.27–0.50. The registry reports a P-value of <0.0001. The endpoint extended subsequent-therapy follow-up until September 2017.
The point estimate is close to the earlier TFST estimate of 0.34, but the analyses have different follow-up definitions. They should therefore be reported separately rather than treated as repeated measurements of exactly the same analysis.
Extended TSST
TSST at Extended OS
95% CI: 0.40–0.72 · P < 0.0001
Supportive analysis to PFS2
The extended TSST HR of 0.54 indicates a lower estimated hazard of second subsequent cancer therapy or death for olaparib relative to chemotherapy under the reported analysis. The 95% CI of 0.40–0.72 gives the corresponding uncertainty interval.
As with the other extended analyses, the September 2017 endpoint definition should be retained when interpreting this result because follow-up duration is part of the meaning of a time-to-event analysis.
11. Secondary Results Summary
| Endpoint | Effect | 95% CI | P-value | Role |
|---|---|---|---|---|
| PFS2 | HR 0.57 | 0.40–0.83 | 0.0033 | Secondary |
| OS | HR 0.90 | 0.63–1.29 | 0.5665 | Secondary |
| Myriad CDx gBRCAm PFS | HR 0.57 | 0.41–0.78 | 0.0005 | Secondary |
| Final OS | HR 0.90 | 0.66–1.23 | 0.5131 | Secondary |
| Extended OS | HR 0.89 | 0.67–1.18 | 0.4167 | Secondary |
| Adjusted QoL | Mean difference 7.5 | 2.5–12.4 | 0.0035 | Secondary |
| TFST | HR 0.34 | 0.24–0.47 | <0.0001 | Other pre-specified |
| TSST | HR 0.53 | 0.38–0.74 | 0.0002 | Other pre-specified |
| Extended TFST | HR 0.36 | 0.27–0.50 | <0.0001 | Other pre-specified |
| Extended TSST | HR 0.54 | 0.40–0.72 | <0.0001 | Other pre-specified |
The table illustrates an important statistical distinction: the trial contains one registered primary endpoint, while the registry also reports multiple secondary and other pre-specified analyses. Their numerical estimates can be compared descriptively, but their roles in the statistical hierarchy are not identical.
12. Serious Adverse Events
The ClinicalTrials.gov record reports serious adverse events by treatment arm as affected patients divided by patients at risk.
| Safety measure | Olaparib 300 mg bd | Chemotherapy |
|---|---|---|
| Serious adverse events | 37/205 | 15/91 |
These are descriptive safety counts reported by the registry data. They should not be converted into a comparative risk estimate here because the ClinicalTrials.gov record does not provide a formal statistical analysis for serious adverse events.
13. Statistical Methods Explained
Why was a log-rank test used for PFS?
PFS is a time-to-event endpoint because each patient contributes a time until progression or death, or a censored observation if the event has not occurred within available follow-up. A log-rank test compares the event-time distributions between randomized groups while using information across the follow-up period rather than reducing every patient to a single fixed-time binary outcome.
What does an HR of 0.58 mean?
An HR of 0.58 means that the estimated instantaneous event rate for progression or death was 0.58 times that of the comparator under the reported model and analysis framework. The simple relative interpretation is a 42% lower estimated hazard. It does not mean that 42% fewer patients experienced an event, because an HR is not an absolute risk difference.
Why is the confidence interval important?
The point estimate is only one part of an inferential result. The 95% CI of 0.43–0.80 for primary PFS describes uncertainty around the estimated HR. A narrower interval generally conveys greater statistical precision than a wider one, although precision and clinical importance are separate questions.
Why doesn't the P-value measure effect size?
The primary P-value of 0.0009 addresses the evidence against the null hypothesis under the reported testing framework. It depends not only on the observed effect but also on statistical variability and information in the analysis. The HR and its confidence interval are therefore needed to describe the size and uncertainty of the treatment comparison.
Why was a mixed-effects model used for QoL?
The QoL endpoint consists of repeated assessments over visits. The registry specifies treatment, visit, treatment-by-visit interaction, baseline QoL adjustment, and baseline score-by-visit interaction. A mixed-effects model is designed for this type of longitudinal structure because observations from the same patient are not treated as independent measurements.
Why was stratification used?
The time-to-event analyses were stratified by received prior chemotherapy for metastatic breast cancer and receptor status. Stratification incorporates these prespecified factors into the comparison so that the reported treatment effect reflects the stratified analysis rather than an unqualified unadjusted comparison.
14. Intention-to-Treat Analysis and Randomization
The primary PFS analysis used the Full Analysis Set consisting of all randomized patients. The registry analysis text also identifies intention-to-treat analysis as a concept in the primary analysis.
The central statistical value of randomization is that treatment assignment is determined independently of patients' subsequent outcomes. An analysis based on randomized assignment therefore preserves the treatment-comparison structure created at the beginning of the trial.
Randomization protects the comparison
Random assignment creates the framework for comparing outcomes between treatment groups without selecting patients for analysis based on their later outcomes.
FAS preserves the randomized framework
Using all randomized patients for the primary PFS analysis avoids redefining the treatment groups according to later treatment exposure or outcome experience.
This does not eliminate every source of uncertainty. It means that the statistical interpretation remains anchored to the randomized comparison specified by the study.
15. Stratified Survival Analysis
The registry specifies two stratification factors for the reported time-to-event analyses: prior chemotherapy for metastatic breast cancer, coded as yes/no, and receptor status.
| Stratification factor | Reported categories | Role in analysis |
|---|---|---|
| Received prior chemotherapy for metastatic breast cancer | Yes / No | Stratified time-to-event analysis |
| Receptor status | Registry identifies receptor status as a stratification factor | Stratified time-to-event analysis |
Stratification is not the same as performing separate primary analyses within every stratum. The registry reports a combined stratified analysis, allowing the treatment comparison to account for the specified factors while estimating an overall treatment effect.
16. Confidence Intervals Across the Trial
| Analysis | Estimate | 95% CI | What the interval communicates |
|---|---|---|---|
| Primary PFS | HR 0.58 | 0.43–0.80 | Uncertainty around the primary relative PFS effect |
| PFS2 | HR 0.57 | 0.40–0.83 | Uncertainty around the second-progression relative effect |
| OS | HR 0.90 | 0.63–1.29 | Uncertainty includes the no-difference value |
| QoL | Mean difference 7.5 | 2.5–12.4 | Uncertainty around the adjusted mean difference |
| Myriad CDx gBRCAm PFS | HR 0.57 | 0.41–0.78 | Uncertainty in the specified molecularly confirmed subgroup |
| Final OS | HR 0.90 | 0.66–1.23 | Uncertainty includes the no-difference value |
| Extended OS | HR 0.89 | 0.67–1.18 | Uncertainty includes the no-difference value |
Confidence intervals are particularly useful when several related analyses are reported. They show whether the point estimate is precise enough to distinguish the estimated effect from the no-difference value under the specified inferential framework.
17. Understanding the Different Time-to-Event Endpoints
OlympiAD reports several related but distinct time-to-event outcomes. Their names should not be treated as interchangeable because each defines a different event process.
PFS
Time from randomisation to the earliest of objective radiological progression or death in the absence of objective progression.
PFS2
Time to second progression or death, with second progression status reviewed after the first objective radiological progression.
TFST
Time to first subsequent cancer therapy or death.
TSST
Time to second subsequent cancer therapy or death.
These endpoints form a conceptual sequence: PFS focuses on the first progression-or-death event; PFS2 extends the disease-course perspective beyond the first progression; TFST and TSST use subsequent cancer therapy as part of the event definition. A lower HR across these different endpoints can therefore provide a consistent descriptive pattern, but each estimate remains tied to its own endpoint definition.
18. Planned Power and Hypothesis Framework
The registry analysis notes that the study was sized to provide 90% power to detect a true treatment effect of PFS hazard ratio 0.653. The primary hypothesis type is listed as superiority.
The observed primary PFS estimate was HR 0.58. The design target and the observed estimate answer different questions: one describes the treatment effect assumed for planning, while the other describes the effect estimated from the posted analysis.
It is important not to reverse-engineer the study design from the observed HR. A planned effect size is a design assumption used for sample-size and power calculations; it is not a claim about what the final treatment effect would necessarily be.
19. Statistical Interpretation of the Primary PFS Result
The primary PFS HR of 0.58 indicates a lower estimated instantaneous rate of progression or death for olaparib compared with chemotherapy under the reported stratified time-to-event analysis.
The HR does not supply an absolute probability of being progression-free at a specified time. No median PFS, Kaplan-Meier survival probability, or event count is provided in the ClinicalTrials.gov record, so none is added here.
The 95% CI of 0.43–0.80 indicates the statistical uncertainty around the HR estimate. The interval remains below 1 throughout, which is consistent with the reported direction of the superiority comparison.
The P-value of 0.0009 provides the reported evidence against the null hypothesis. It should be read alongside the HR and confidence interval rather than used as a standalone measure of treatment magnitude.
The analysis used the Full Analysis Set consisting of all randomized patients. This is important because the interpretation is based on randomized treatment assignment rather than a selected subset defined by treatment completion.
20. Why This Trial Matters Statistically
OlympiAD is a useful teaching example because the registry results bring together several core methods used in clinical-trial statistics without requiring the analyst to treat every endpoint as though it were the same type of outcome.
| Concept | How it appears in OlympiAD |
|---|---|
| Randomization | Randomized allocation to two parallel treatment arms |
| Intention-to-treat principle | Primary PFS analysis in the Full Analysis Set of all randomized patients |
| Time-to-event endpoint | PFS is the registered primary endpoint |
| Log-rank test | Used for the primary PFS and reported secondary time-to-event comparisons |
| Hazard ratio | Primary effect measure for PFS and multiple secondary time-to-event endpoints |
| Confidence interval | 95% two-sided intervals reported around the effect estimates |
| Stratified analysis | Prior metastatic chemotherapy and receptor status used as stratification factors |
| Mixed-effects model | Used for adjusted longitudinal QoL analysis |
| Covariate adjustment | Baseline QoL and baseline score-by-visit interaction incorporated into the QoL model |
| Multiple endpoint roles | Primary, secondary, and other pre-specified analyses are separately identified |
| Safety analysis | Serious adverse events reported by arm with affected and at-risk counts |
21. Important Limitations and Interpretation Issues
- Open-label design: the trial is registered as having no masking. Lack of masking can be particularly relevant for outcomes involving patient reporting or clinical assessment, although the primary PFS endpoint used blinded independent central review.
- Different analysis populations: the primary PFS analysis uses all randomized patients, whereas the QoL analysis requires evaluable baseline and post-baseline assessments.
- Time-to-event interpretation: hazard ratios are relative event-rate measures and should not be substituted for absolute survival probabilities or median event times.
- Secondary endpoint hierarchy: PFS2, OS, molecularly confirmed PFS, QoL, final OS, and extended OS are secondary analyses and should not be described as though they were the registered primary endpoint.
- Supportive analyses: TFST and TSST are identified in the registry as supportive analyses to PFS and PFS2, respectively. Their role differs from that of the primary PFS test.
- Subgroup population: the Myriad CDx gBRCAm PFS result is based on a subset of randomized patients, so its interpretation should remain tied to that population.
- No unreported results: median survival times, subgroup forest plots, detailed baseline characteristics, event counts, and other quantities not present in the ClinicalTrials.gov record is not reconstructed or inferred.
22. What the Hazard Ratio Does — and Does Not — Mean
A PFS HR of 0.58 means that, under the reported time-to-event analysis, the estimated instantaneous rate of progression or death in the olaparib group was 0.58 times the corresponding estimated rate in the chemotherapy group.
Expressed as a simple relative transformation, this corresponds to an approximately 42% lower estimated hazard.
It does not mean that 42% of patients avoided progression, that every patient's risk fell by exactly 42%, or that the absolute probability of progression differed by 42 percentage points.
The 95% CI of 0.43–0.80 describes uncertainty around the estimated HR. It does not describe the distribution of individual treatment effects across patients.
The P-value of 0.0009 quantifies statistical evidence under the reported testing framework. It does not tell the reader how large the treatment effect is. That information comes from the HR and its confidence interval.
23. Related Statistical Concepts
Learn more about the methods used in this trial:
24. Related Statistical Calculators
25. Sources
- ClinicalTrials.gov: NCT02000622 — OlympiAD.
- Linked publication: PubMed record for PMID 28578601.
Continue with the underlying statistical methods
The OlympiAD results connect directly to core methods for time-to-event analysis, hazard ratios, confidence intervals, stratified comparisons, longitudinal modeling, and randomized clinical-trial analysis.
26. Record Summary
OlympiAD provides a compact teaching example of how a randomized phase 3 trial can combine a primary time-to-event endpoint with several secondary efficacy outcomes and a longitudinal quality-of-life analysis. The primary PFS comparison used the Full Analysis Set, a stratified log-rank test, and a hazard ratio of 0.58 with a 95% CI of 0.43–0.80 and P = 0.0009. The registry also reports PFS2, OS, a molecularly confirmed PFS analysis, QoL, final and extended OS, and subsequent-therapy endpoints using methods appropriate to their endpoint structures.
The most useful statistical reading therefore combines the effect measure, confidence interval, P-value, analysis population, stratification variables, and endpoint definition. In particular, the primary PFS result should not be reduced to its P-value alone, while the secondary OS and QoL analyses should not be treated as though they were statistically interchangeable with the primary endpoint.