This page separates reported trial results from statistical interpretation. Numerical results are restricted to the ClinicalTrials.gov record and the linked publication record. Where the registry does not provide a requested detail, it is not inferred from outside sources.
1. Trial at a Glance
PACE is a randomized, open-label, parallel phase 2 trial in metastatic breast cancer. The trial enrolled 220 participants across 3 arms and evaluates treatment involving palbociclib, fulvestrant, and avelumab. The posted primary efficacy analysis compares progression-free survival between fulvestrant alone and fulvestrant with palbociclib.
| Feature | PACE |
|---|---|
| Trial name | PACE |
| Brief title | Palbociclib After CDK and Endocrine Therapy (PACE) |
| Phase | Phase 2 |
| Status | Active, not recruiting |
| Condition | Metastatic Breast Cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 220 |
| Arms | 3 |
| Lead sponsor | Dana-Farber Cancer Institute |
| Start | 2017-09-05 |
| Primary completion | 2022-12-31 |
| Results posted | Yes |
| ClinicalTrials.gov | NCT03147287 |
2. Clinical Question
The PACE trial addresses a treatment question in participants with metastatic breast cancer after CDK and endocrine therapy. The registered trial evaluates palbociclib, fulvestrant, and avelumab across 3 randomized arms; the posted primary statistical comparison evaluates progression-free survival between fulvestrant alone and fulvestrant with palbociclib.
Population
Participants with metastatic breast cancer enrolled in the PACE trial after CDK and endocrine therapy, as reflected in the trial's brief title and condition.
Intervention of the primary comparison
Fulvestrant with palbociclib.
Comparator of the primary comparison
Fulvestrant.
Primary question
Does fulvestrant with palbociclib produce a superior progression-free survival distribution compared with fulvestrant alone?
3. Trial Design
The registry identifies the allocation as randomized and the design model as parallel. Masking is listed as none, so this is an open-label design rather than a blinded comparison.
Registered treatment arms
Fulvestrant
- Fulvestrant is listed as a study intervention.
- This arm is the comparator in the posted primary PFS analysis.
Fulvestrant With Palbociclib
- Fulvestrant with palbociclib.
- This arm is the treatment group in the posted primary PFS analysis.
Fulvestrant With Palbociclib and Avelumab
- Fulvestrant with palbociclib and avelumab.
- The registry lists avelumab as an intervention.
4. Endpoints
| Endpoint | Registry definition / time frame | Posted analysis |
|---|---|---|
| Primary: Progression-Free Survival (PFS), According to RECIST v1.1 Criteria (Investigator Assessment) | The time from randomization to the earlier of progression or death due to any cause. Participants alive without disease progression were censored at date of last disease assessment. Tumor were assessed every 8 weeks until 24 weeks, then every 12 weeks. The median PFS follow-up time at database lock was 12.1 months (range 1 day to 50.1 months). | Log-rank test; hazard ratio from Cox proportional-hazards model |
| Secondary: Overall Response Rate, According to RECIST v1.1 Criteria (Investigator Assessment) | Tumor were assessed every 8 weeks until 24 weeks, then every 12 weeks. The median follow-up time for overall response rate at database lock was 12.1 months (range 1 day to 50.1 months). | Fisher exact test |
The registry reports 1 primary endpoint and 3 posted outcome measures. Two statistical analyses are posted: 1 primary-endpoint analysis and 1 secondary analysis.
5. Primary Result: Progression-Free Survival
The posted primary analysis compares the PFS distributions of fulvestrant alone with fulvestrant plus palbociclib. The registry specifies a one-sided log-rank test with an alpha level of 0.05 for the primary objective. The hazard ratio is estimated from a Cox proportional-hazards model, with fulvestrant plus palbociclib divided by fulvestrant.
Hazard ratio for progression or death
Two-sided 90% CI: 0.79–1.55 · P = 0.62
Comparison: Fulvestrant With Palbociclib / Fulvestrant
| Primary PFS analysis | Reported result |
|---|---|
| Endpoint | Progression-Free Survival, according to RECIST v1.1 criteria, investigator assessment |
| Comparison | Fulvestrant vs Fulvestrant With Palbociclib |
| Analysis method | Log-rank test |
| Effect measure | Hazard ratio |
| Hazard ratio | 1.11 |
| Confidence interval | Two-sided 90% CI: 0.79–1.55 |
| P-value | 0.62 |
| Hypothesis type | Superiority |
| Testing framework | One-sided alpha level of 0.05 |
The reported hazard ratio of 1.11 is defined in the registry analysis as the hazard for the fulvestrant-plus-palbociclib group divided by the hazard for the fulvestrant group. Under that convention, an HR below 1 would indicate a lower estimated hazard of a PFS event with fulvestrant plus palbociclib. An HR of 1.11 therefore corresponds to an estimated hazard that is 11% higher in the numerator group relative to the comparator under the fitted model.
That statement is about the estimated relative hazard; it does not mean that 11% more participants progressed, that every participant had an 11% higher probability of progression, or that the difference in individual outcomes was exactly 11%.
The two-sided 90% confidence interval of 0.79–1.55 describes uncertainty around the estimated hazard ratio under the model and analysis framework. Because the interval includes 1, the estimate is compatible with hazard ratios below and above the null value. The interval is about precision of the estimated relative hazard, not a range containing the treatment effect for individual patients.
The P = 0.62 value is evidence from the prespecified one-sided superiority test; it is not a measure of the magnitude of the treatment effect. A p-value should not be interpreted as the probability that the treatment works, nor as a percentage reduction or increase in risk.
Finally, the hazard ratio is model-based. Its interpretation depends on the time-to-event analysis and the Cox proportional-hazards framework. Censoring and the proportional-hazards assumption are therefore relevant to interpretation. The registry also identifies stratified analysis as an analysis concept, but the ClinicalTrials.gov record does not provide the complete stratification variables or the full Cox model specification, so those details are not reconstructed here.
What the primary test was actually asking
The analysis notes define the primary hypotheses as:
The registry describes a one-sided alpha level of 0.05 and a log-rank comparison of the PFS distributions. The exact labels “1” and “2” are retained from the registry analysis note rather than assigning new arm numbering on this page.
This is important because the primary hypothesis was directional: the trial was testing superiority rather than simply asking whether the two PFS distributions were different in either direction.
6. Secondary Result: Overall Response Rate
Overall Response Rate, according to RECIST v1.1 criteria with investigator assessment, is listed as a secondary outcome measure. The posted statistical analysis compares fulvestrant with fulvestrant plus palbociclib using an unstratified Fisher exact test.
Fisher exact comparison
Comparison: Fulvestrant vs Fulvestrant With Palbociclib
Analysis: Fisher exact test
| Secondary analysis | Reported result |
|---|---|
| Outcome | Overall Response Rate, according to RECIST v1.1 criteria, investigator assessment |
| Outcome unit | Percentage of participants analyzed |
| Comparison | Fulvestrant vs Fulvestrant With Palbociclib |
| Method | Fisher exact test |
| P-value | 1.000 |
| Hypothesis type | Superiority |
The Fisher exact test is appropriate for comparing a binary response outcome between two treatment groups, particularly when exact inference is desired. The registry reports P = 1.000 for this comparison.
The registry-reported statistical analysis does not provide the response counts, response percentages, confidence intervals, or an effect estimate. Consequently, this page does not infer or reconstruct those quantities. The reported p-value alone does not tell us the absolute difference in response rate or the precision of that difference.
Likewise, the p-value does not establish that the response rates are identical. It reports the result of the specified statistical comparison; effect size and uncertainty require the underlying response counts or corresponding estimates.
7. Statistical Methodology
Kaplan-Meier estimation for PFS
Progression-free survival is a time-to-event endpoint. The registry defines PFS as the time from randomization to the earlier of progression or death due to any cause, with participants alive without progression censored at their last disease assessment. A Kaplan-Meier estimator is the standard nonparametric method for describing such a time-to-event distribution.
Here, di represents the number of events at an event time and ni represents the number at risk immediately beforehand. The estimator naturally accommodates right-censored observations.
The ClinicalTrials.gov record does not provide a Kaplan-Meier table or individual event and censoring times. Therefore, this page does not attempt to reconstruct a survival curve.
Log-rank test
The primary PFS comparison uses a log-rank test. Conceptually, the log-rank test compares the observed and expected numbers of events between treatment groups across the observed event times. It uses the ordering of event times rather than reducing the entire follow-up period to a single binary outcome.
For PACE, the registry specifies a one-sided alpha level of 0.05 for the primary superiority objective. This directional testing framework is important when interpreting the reported P = 0.62.
Cox proportional-hazards model
The hazard ratio was estimated from a Cox proportional-hazards model. The registry analysis notes specify the direction as F+P / F, meaning fulvestrant with palbociclib divided by fulvestrant.
Under this direction, HR < 1 indicates a reduced estimated hazard of a PFS event with fulvestrant plus palbociclib, while HR > 1 indicates a higher estimated hazard under the fitted model.
Fisher exact test
The secondary overall-response comparison uses Fisher's exact test. Unlike PFS, response is a binary endpoint: participants are classified according to whether they meet the prespecified response definition. Fisher's exact test evaluates the association between treatment group and the binary response outcome without relying on the large-sample approximation used by some other categorical tests.
Stratified analysis
The registry-reported analysis metadata identifies stratified analysis as a concept associated with the posted PFS analysis. However, the ClinicalTrials.gov record does not identify the complete stratification variables or provide enough information to reproduce the stratified analysis exactly. The page therefore distinguishes the documented presence of stratified analysis from any unsupported reconstruction of its implementation.
8. Statistical Methods Explained
Why was a log-rank test used for PFS?
PFS is defined as time from randomization to progression or death, with censoring for participants who remain alive without progression at their last disease assessment. A log-rank test is designed for comparing time-to-event distributions while retaining information about when events occur during follow-up. A simple comparison of the proportion of participants who progressed would discard much of that timing information.
What does an HR of 1.11 mean?
Because the registry defines the Cox estimate as fulvestrant with palbociclib divided by fulvestrant, an HR of 1.11 means the estimated instantaneous event hazard was 1.11 times the comparator hazard under the fitted model. Equivalently, it is an estimated 11% higher hazard under that model. It is not an 11-percentage-point difference in progression probability and does not mean that every participant experienced an 11% increase in risk.
Why is the confidence interval two-sided when the primary test is one-sided?
The registry reports a one-sided alpha level of 0.05 for the primary superiority test while reporting a two-sided 90% confidence interval for the hazard ratio. These are related but distinct pieces of the statistical presentation. The test addresses the prespecified directional hypothesis, while the two-sided interval describes uncertainty on both sides of the estimated hazard ratio.
Why does P = 0.62 not tell us the size of the treatment effect?
A p-value measures the compatibility of the observed data with the null hypothesis under the specified testing framework. It is not an effect-size statistic. The PFS effect size is represented by the hazard ratio of 1.11, while the two-sided 90% CI of 0.79–1.55 describes uncertainty around that estimate.
Why use Fisher exact test for overall response?
Overall response is a binary outcome. Fisher exact testing compares the allocation of responders and nonresponders between the two treatment groups using an exact method. The registry-reported analysis reports P = 1.000 but does not provide the underlying response counts, so the magnitude of the response-rate difference cannot be reconstructed from the ClinicalTrials.gov record alone.
What does censoring mean in this PFS analysis?
Participants who were alive without disease progression were censored at the date of their last disease assessment. Censoring means that the participant contributes information up to the last time their event-free status is known; it does not mean that the participant is counted as having progressed at that time. The validity of a time-to-event analysis depends on the assumptions governing censoring and follow-up.
9. One-Sided Testing and the PFS Result
The PACE primary analysis is especially useful for understanding the distinction between a directional hypothesis and the numerical value of an estimated effect. The registry specifies a one-sided alpha level of 0.05 and a superiority hypothesis.
Directional question
The primary objective was framed around whether the PFS distribution with fulvestrant plus palbociclib was superior to the comparator distribution.
Observed estimate
The reported hazard ratio was 1.11, with the numerator defined as fulvestrant plus palbociclib.
Uncertainty
The two-sided 90% confidence interval was 0.79–1.55, spanning the null value of 1.
Test result
The reported p-value was 0.62 under the specified superiority testing framework.
These quantities should be read together. The hazard ratio gives the estimated relative effect, the confidence interval gives a measure of uncertainty around that estimate, and the p-value reports the result of the specified hypothesis test. None of the three quantities alone provides a complete description of the evidence.
10. Confidence Intervals: What the PFS Interval Tells Us
Reported two-sided 90% confidence interval
Point estimate: HR 1.11
The interval extends below and above the null hazard ratio of 1. This means the registry-reported estimate is accompanied by uncertainty that includes both lower and higher relative hazards under the Cox model.
The confidence interval is a statement about statistical uncertainty under the specified model and sampling framework. It should not be interpreted as a range of outcomes that individual participants could experience. Nor should its width be translated directly into a clinical judgment without considering the endpoint, study design, population, and magnitude of effects that would matter clinically.
The confidence interval also explains why reporting only the point estimate would be inadequate. An HR of 1.11 by itself gives a single estimate. The interval shows how much uncertainty surrounds that estimate in the reported analysis.
11. Safety Results
The ClinicalTrials.gov record reports serious adverse events by arm as affected participants over participants at risk. These are the only arm-specific safety counts provided in the ClinicalTrials.gov record.
| Arm | Serious adverse events | Affected / at risk |
|---|---|---|
| Fulvestrant | Serious adverse events | 1 / 53 |
| Fulvestrant With Palbociclib | Serious adverse events | 46 / 110 |
| Fulvestrant With Palbociclib and Avelumab | Serious adverse events | 37 / 53 |
These counts should be interpreted as reported arm-level safety summaries. The ClinicalTrials.gov record does not provide definitions for individual serious adverse events, exposure duration, severity distributions, attribution, or confidence intervals. Accordingly, this page does not calculate additional comparative safety measures from the counts.
12. What Is and Is Not Reported in the Registry Data
The ClinicalTrials.gov record contains enough information to describe the principal PFS and response analyses, but it does not support reconstruction of every statistical feature that might normally appear in a full clinical-trial report.
| Topic | What the ClinicalTrials.gov record supports |
|---|---|
| Primary endpoint | 1 registered primary time-to-event endpoint: PFS |
| Primary statistical analysis | Log-rank test with hazard ratio from a Cox proportional-hazards model |
| Primary confidence interval | Two-sided 90% CI: 0.79–1.55 |
| Primary p-value | 0.62 |
| Secondary statistical analysis | Fisher exact test for overall response rate |
| Stratified analysis | Identified as an analysis concept; complete strata are not reported |
| Kaplan-Meier results | Endpoint type supports Kaplan-Meier estimation, but no KM estimates or curve data are reported |
| Median PFS | Not provided in the ClinicalTrials.gov record |
| Subgroup results | Not provided in the ClinicalTrials.gov record |
| Formal response estimate | Not provided; only Fisher exact P = 1.000 is reported |
| Interim analysis | Not identified in the ClinicalTrials.gov record |
| Multiplicity procedure | Not identified in the ClinicalTrials.gov record |
| Missing-data / imputation method | Not identified in the ClinicalTrials.gov record |
| Bayesian methods | Not identified in the ClinicalTrials.gov record |
| Non-inferiority margin | Not applicable to the reported superiority hypothesis and not identified in the ClinicalTrials.gov record |
| Crossover | Not identified in the ClinicalTrials.gov record |
| Factorial design | Not identified; the registered design model is parallel |
13. Analysis Populations and Interpretation
The ClinicalTrials.gov record does not provide detailed definitions of the analysis populations used for the posted PFS or response analyses. That limitation matters because treatment-effect estimates can depend on whether participants are analyzed according to randomized assignment, treatment received, or another prespecified population.
For the primary PFS result, the safest interpretation is therefore to retain the registry's stated comparison and statistical method rather than assigning an unsupported label such as intention-to-treat or per-protocol to the posted analysis.
14. Stratified Analysis
The statistical analysis metadata identifies stratified analysis as a concept associated with the primary PFS analysis. Stratification can be used to account for prespecified factors when comparing time-to-event outcomes, particularly when those factors were incorporated into the randomization or analysis plan.
For a stratified log-rank test, evidence from the individual strata is combined rather than treating each stratum as an entirely separate hypothesis test. A stratified Cox model can similarly estimate a common treatment hazard ratio while allowing the underlying baseline hazard to differ across strata.
The exact strata used in the PACE analysis are not provided in the ClinicalTrials.gov record, so no specific stratification factors are assigned here.
This is a useful example of why registry metadata and a complete statistical analysis plan are not interchangeable. The registry identifies that stratified analysis was part of the analysis concept, but the available data do not expose every implementation detail.
15. Limitations
- Limited numerical reporting: the ClinicalTrials.gov record provides the primary hazard ratio, confidence interval, and p-value, but does not provide median PFS, event counts, or Kaplan-Meier estimates.
- Incomplete response reporting: the secondary Fisher exact analysis reports P = 1.000 but does not provide response counts, response percentages, or a confidence interval.
- Three-arm design with a two-arm posted primary comparison: the registry's primary analysis compares fulvestrant with fulvestrant plus palbociclib. The third arm cannot be analytically characterized from that comparison alone.
- Open-label design: masking is listed as none. Investigator-assessed RECIST outcomes can therefore be considered in the context of the absence of masking, although the ClinicalTrials.gov record does not quantify any resulting bias.
- Incomplete analysis-population information: the ClinicalTrials.gov record does not define the exact population used for the posted statistical analyses.
- Stratification details: stratified analysis is identified, but the ClinicalTrials.gov record does not provide the complete set of stratification factors or the exact model specification.
- Proportional-hazards assumption: the Cox model supplies a single hazard ratio, but the ClinicalTrials.gov record does not provide diagnostics for the proportional-hazards assumption.
- Censoring: PFS includes right-censoring at the last disease assessment for participants alive without progression. Interpretation therefore depends on the assumptions underlying the censoring process.
- Multiplicity: the ClinicalTrials.gov record does not specify a multiplicity-adjustment procedure across the 3 arms or the posted outcome measures. No such procedure is inferred here.
- Missing data: no imputation or missing-data strategy is identified in the ClinicalTrials.gov record.
- Bayesian analysis: no Bayesian method is identified in the ClinicalTrials.gov record.
- Non-inferiority: the reported primary hypothesis is superiority, so a non-inferiority margin is not part of the registry-reported primary analysis.
16. Why This Trial Matters Statistically
PACE is a useful teaching example because its posted analysis brings together several core clinical-trial methods without requiring a large collection of different endpoint types. The primary endpoint is time-to-event, while the secondary response endpoint is binary. That makes the trial a compact illustration of how the statistical method changes with the structure of the outcome.
| Concept | How it appears in PACE |
|---|---|
| Randomization | The trial is randomized with a parallel design and 3 arms. |
| Time-to-event endpoint | PFS is defined from randomization to progression or death. |
| Kaplan-Meier estimation | The PFS endpoint is naturally suited to Kaplan-Meier estimation, although no curve or numerical KM estimates are reported. |
| Log-rank test | The posted primary comparison uses a log-rank test. |
| Hazard ratio | The primary effect measure is an HR of 1.11. |
| Confidence interval | The HR is accompanied by a two-sided 90% CI of 0.79–1.55. |
| One-sided testing | The primary objective uses a one-sided alpha level of 0.05 for a superiority hypothesis. |
| Stratified analysis | Stratified analysis is identified in the primary analysis metadata. |
| Fisher exact test | The secondary overall-response comparison uses Fisher exact testing. |
| Binary endpoint | Overall response rate is analyzed as a percentage of participants analyzed. |
| Safety by arm | Serious adverse events are reported as affected participants over participants at risk for each arm. |
17. A Deeper Reading of the PFS Result
The most important statistical lesson from the PFS analysis is that a treatment effect should not be reduced to a single p-value. The PACE analysis reports three complementary pieces of information: a hazard ratio, a confidence interval, and a hypothesis-test result.
Effect estimate
HR 1.11 is the model-based estimate of the relative hazard, with fulvestrant plus palbociclib in the numerator.
Uncertainty
90% CI 0.79–1.55 shows uncertainty around the estimated hazard ratio and includes the null value of 1.
Hypothesis test
P = 0.62 is the reported result of the one-sided superiority test using alpha 0.05.
Endpoint meaning
PFS incorporates both progression and death and also incorporates censoring for participants who remain alive without progression at their last assessment.
These are different statistical objects. The HR describes the estimated relative treatment effect, the confidence interval describes uncertainty around that estimate, and the p-value addresses the prespecified null hypothesis. Keeping those roles separate prevents several common interpretation errors.
18. Primary Endpoint: What PFS Actually Measures
The registered PFS definition is time from randomization to the earlier of progression or death due to any cause. This composite event definition means that death can produce a PFS event even if documented tumor progression has not occurred first.
Participants who were alive without disease progression were censored at their last disease assessment. Thus, PFS analysis uses more information than a simple yes/no determination of whether a participant eventually progressed.
A PFS hazard ratio should therefore be understood as a relative comparison of the event process over follow-up, not as a direct statement about the percentage of participants who are progression-free at a particular calendar time. To make an absolute interpretation at a particular time point, a survival estimate such as a Kaplan-Meier PFS probability would be needed.
The ClinicalTrials.gov record does not report such time-specific PFS probabilities or median PFS, so this page does not create them.
19. Overall Response Rate: A Different Statistical Question
Overall response rate and PFS have different data structures. ORR is a binary outcome based on whether participants meet the response criterion, whereas PFS records the timing of progression or death and can include censoring.
| Feature | PFS | Overall Response Rate |
|---|---|---|
| Endpoint type | Time-to-event | Binary |
| Posted method | Log-rank test | Fisher exact test |
| Effect measure posted | Hazard ratio | No effect estimate reported |
| Reported p-value | 0.62 | 1.000 |
| Confidence interval reported | Two-sided 90% CI: 0.79–1.55 | Not reported |
This distinction illustrates a broader principle in clinical biostatistics: the analysis should follow the structure of the endpoint. Applying a time-to-event method to a binary response endpoint, or treating a hazard ratio as if it were a response-rate ratio, would change the question being answered.
20. Serious Adverse Events Across the Three Arms
The three-arm structure becomes particularly visible in the safety data. Unlike the posted primary efficacy analysis, which compares 2 arms, the registry-reported serious-adverse-event summary contains a value for each of the 3 registered arms.
The bar lengths above are visual representations of the reported affected counts relative to the largest affected count; the underlying reported quantities are shown explicitly as affected / at risk. The ClinicalTrials.gov record does not provide enough information to determine whether differences in serious-adverse-event burden are statistically tested or clinically attributable to individual interventions.
21. Trial Timeline
Trial start
The registered start date for PACE is 2017-09-05.
Primary completion
The registered primary completion date is 2022-12-31.
Active, not recruiting
The ClinicalTrials.gov record identifies the trial status as ACTIVE_NOT_RECRUITING.
Statistical analyses available
The registry reports results and 2 statistical analyses, including 1 primary-endpoint analysis and 1 secondary analysis.
22. What the PACE Analysis Does Not Establish
Several statements would go beyond the registry-reported evidence. The reported PFS hazard ratio does not establish a treatment effect for the third arm, does not provide median PFS, and does not quantify the absolute difference in PFS at a particular time point. Likewise, the secondary P = 1.000 does not provide the response-rate difference.
This distinction is particularly important for registry-based statistical education. A complete analysis should report what is actually available, identify the appropriate statistical framework, and resist filling missing quantities with estimates taken from memory or from unrelated sources.
23. Limitations of the Statistical Evidence
Incomplete survival summaries
No median PFS, event counts, or time-specific Kaplan-Meier estimates are included in the ClinicalTrials.gov record.
Incomplete response summaries
The Fisher exact analysis provides P = 1.000 but not the underlying response counts or effect estimate.
Third-arm interpretation
The primary statistical analysis registry-reported is a 2-arm comparison even though the randomized trial contains 3 arms.
Model details
Stratified analysis is identified, but complete stratification and Cox-model details are not reported.
These limitations do not make the reported analysis unusable. They define what can be concluded from the registry information without importing additional numerical results.
24. Why This Trial Is a Useful Statistical Case Study
PACE provides a compact example of how a clinical trial can require multiple statistical tools because its endpoints have different structures. The primary PFS endpoint requires survival-analysis methods, while the response endpoint uses a categorical-data method.
| Statistical lesson | PACE example |
|---|---|
| Match the method to the endpoint | PFS uses log-rank testing; overall response uses Fisher exact testing. |
| Separate effect size from hypothesis testing | HR 1.11 and P = 0.62 answer different statistical questions. |
| Report uncertainty | The PFS HR includes a two-sided 90% CI of 0.79–1.55. |
| Respect test direction | The primary PFS hypothesis is specified as a one-sided superiority test. |
| Understand censoring | PFS censors participants alive without progression at their last disease assessment. |
| Distinguish randomized arms from posted comparisons | The trial has 3 arms, while the registry-reported primary analysis compares 2 of them. |
| Avoid unsupported reconstruction | Missing median PFS, response estimates, and subgroup results are not inferred. |
25. Related Tutorials
Learn more about the methods used in this trial:
26. Related Statistical Calculators
27. Sources
- ClinicalTrials.gov: PACE, NCT03147287. Trial design, endpoint definitions, posted statistical analyses, and arm-level serious adverse-event counts used for this page.
- PubMed: PMID 38513188. Linked publication record associated with the trial.
Continue with the underlying statistical methods
PACE connects directly to core clinical-trial concepts including survival analysis, hazard ratios, log-rank testing, confidence intervals, one-sided hypothesis testing, and Fisher exact testing.
28. Record Summary
PACE is a randomized phase 2, parallel, open-label trial with 220 participants and 3 registered treatment arms. The registry-reported primary analysis evaluates progression-free survival between fulvestrant and fulvestrant with palbociclib using a log-rank test under a one-sided superiority framework with alpha 0.05. The corresponding hazard ratio, defined as fulvestrant with palbociclib divided by fulvestrant, is 1.11, with a two-sided 90% confidence interval of 0.79–1.55 and a reported P = 0.62.
The secondary overall-response analysis uses an unstratified Fisher exact test and reports P = 1.000, without supplying the underlying response counts or effect estimate in the ClinicalTrials.gov record. Serious adverse events are reported for all 3 arms as affected participants over participants at risk: 1/53 for fulvestrant, 46/110 for fulvestrant with palbociclib, and 37/53 for fulvestrant with palbociclib and avelumab.
The main statistical lesson is methodological: PFS requires a time-to-event framework that accounts for event timing and censoring, whereas overall response is a binary endpoint analyzed with Fisher's exact test. The hazard ratio, confidence interval, and p-value should be interpreted as complementary pieces of evidence rather than interchangeable measures of treatment effect.