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
PALOMA-2 was a randomized, parallel-group, quadruple-masked phase 3 trial evaluating palbociclib plus letrozole versus placebo plus letrozole for first-line treatment of postmenopausal women with ER+/HER2- advanced breast cancer. The registry reports 666 participants, two treatment arms, one registered primary endpoint, and 20 statistical analyses.
| Feature | PALOMA-2 |
|---|---|
| Phase | Phase 3 |
| Condition | Breast Neoplasms |
| Population | Postmenopausal women with ER+/HER2- advanced breast cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 666 |
| Primary endpoint | Progression-Free Survival (PFS) as Assessed by the Investigator |
| Primary endpoint type | Time-to-event |
| Hypothesis type | Superiority |
| Trial status | Completed |
| Start | 2013-02-22 |
| Primary completion | 2016-02-26 |
| Lead sponsor | Pfizer |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT01740427 |
2. Clinical Question
The central statistical question was whether adding palbociclib to letrozole changes progression-free survival compared with letrozole alone in the registered population. The primary comparison was specified as a superiority comparison between palbociclib plus letrozole and placebo plus letrozole.
Population
Postmenopausal women with ER+/HER2- advanced breast cancer, as described by the registered study title.
Intervention
PD-0332991 (palbociclib) plus letrozole.
Comparator
Placebo plus letrozole.
Primary question
Does palbociclib plus letrozole improve investigator-assessed PFS relative to placebo plus letrozole?
3. Trial Design
Palbociclib Plus Letrozole
- PD-0332991 (palbociclib)
- Letrozole
Placebo Plus Letrozole
- Placebo
- Letrozole
The registry identifies the trial as randomized, parallel-group, quadruple-masked, and treatment-focused. These design features are important statistically because randomization establishes the basis for a treatment comparison, parallel assignment preserves that comparison across two contemporaneous groups, and masking can reduce the influence of treatment knowledge on trial conduct and assessment.
4. Endpoints
| Endpoint | Registry definition / time frame | Type |
|---|---|---|
| Progression-Free Survival (PFS) as Assessed by the Investigator | From randomization date to date of first documentation of progression or death, up to approximately 2.5 years. PFS is defined as the time from the date of randomization to the date of the first documentation of objective tumor progression as per RECIST v1.1 or death due to any cause in the absence of documented PD, whichever occurs first. | Time-to-event |
| Objective Response as Assessed by the Investigator | From randomization until end of treatment, up to approximately 2.5 years. | Binary |
| Objective Response: Participants With Measurable Disease at Baseline as Assessed by the Investigator | From randomization until end of treatment, up to approximately 2.5 years. | Binary |
| Disease Control (DC)/Clinical Benefit Response (CBR) | From randomization until end of treatment, up to approximately 2.5 years. | Binary |
| PFS by Tumor Tissue Biomarkers Status | From randomization until end of treatment, up to approximately 24 months. Biomarker categories include genes such as CCND1 and CDKN2A, proteins such as Ki67 and pRb, and RNA expression such as cdk4 and cdk6. | Time-to-event |
| Change From Baseline Between Treatment Comparison in EQ-5D Index | From baseline up to 2.5 years. | Continuous |
| Change From Baseline Between Treatment Comparison in FACT-B | From baseline up to 2.5 years. | Continuous |
| Overall Survival (OS): Primary Analysis | From date of randomization until death due to any cause or censored, assessed up to a data cut-off date of 15-Nov-2021. | Time-to-event |
| Overall Survival (OS): Final Analysis | From date of randomization until death due to any cause or censored, with final analysis through study completion. | Time-to-event |
5. Statistical Methodology
Intention-to-treat analysis
The primary PFS analysis used the ITT population or full analysis set. The registry describes this population as including all participants who were randomized, with study drug assignment designated according to initial randomization, regardless of whether participants received study medication or received a different drug from the assigned treatment.
This is a central feature of randomized trial analysis. Once treatment assignment is randomized, analyzing participants according to that initial assignment helps preserve the comparability created by randomization. It also avoids redefining the treatment groups after randomization based on adherence or treatment received.
Stratified log-rank test
The primary PFS comparison used a stratified log-rank test. The registry identifies the comparison as palbociclib plus letrozole versus placebo plus letrozole and specifies a superiority hypothesis.
The log-rank framework compares the observed and expected numbers of events between randomized groups across the follow-up period. Stratification allows the comparison to account for prespecified strata rather than treating every participant as if they came from one homogeneous risk set.
Hazard ratio
The primary effect measure was the hazard ratio. The reported estimate was 0.576, with a two-sided 95% confidence interval of 0.463 to 0.718.
An HR of 0.576 corresponds to an estimated hazard that is 57.6% of the comparator hazard under the fitted time-to-event comparison, or a 42.4% lower estimated hazard relative to the comparator. It is not a statement that 42.4% of participants avoided progression, nor does it describe an individual patient's probability of remaining progression-free.
Fisher exact test
Binary secondary endpoints were analyzed with Fisher exact tests. The registry reports odds ratios as the effect measure for objective response, objective response among participants with measurable disease at baseline, and disease control/clinical benefit response.
Mixed-effects model
Patient-reported outcomes were analyzed with repeated-measures mixed-effects models. For the EQ-5D analysis, the model included an intercept, treatment, time, treatment-by-time, and baseline as a covariate. The FACT-B analysis used the same stated repeated-measures modeling structure.
Covariate adjustment
The EQ-5D and FACT-B analyses incorporated baseline as a covariate. This differs from the primary PFS analysis, where the registered method is a stratified log-rank test. The analysis method therefore follows the structure of the endpoint: time-to-event data are treated as survival data, binary response data as categorical data, and repeated patient-reported measurements as longitudinal data.
6. Results: Primary Endpoint
Progression-Free Survival as Assessed by the Investigator
The registered primary endpoint was PFS from randomization to first documentation of objective tumor progression according to RECIST v1.1 or death from any cause in the absence of documented progression, whichever occurred first. The primary analysis was performed using the ITT population and a stratified log-rank test.
Primary PFS treatment effect
95% CI: 0.463–0.718 · P < 0.000001
Two-sided superiority hypothesis · Palbociclib plus letrozole vs placebo plus letrozole
| Primary endpoint | Palbociclib + Letrozole vs Placebo + Letrozole |
|---|---|
| Endpoint | Progression-Free Survival (PFS) as Assessed by the Investigator |
| Analysis population | ITT population / full analysis set |
| Analysis method | Stratified log-rank test |
| Effect measure | Hazard ratio |
| Estimate | 0.576 |
| 95% CI | 0.463–0.718 |
| P-value | <0.000001 |
| Hypothesis | Superiority |
The estimated hazard ratio of 0.576 is below 1. Under the time-to-event model, this corresponds to an estimated hazard approximately up to 10.51 years) 42.4% lower for palbociclib plus letrozole relative to placebo plus letrozole.
The hazard ratio does not mean that 42.4% of participants were protected from progression or death. It is a relative measure of the instantaneous event rate represented by the time-to-event analysis. It also does not directly tell us the absolute difference in the probability of being progression-free at any particular time.
The two-sided 95% confidence interval of 0.463 to 0.718 describes uncertainty around the estimated hazard ratio. Because the entire interval is below 1, the registry's reported estimate is consistently on the side of a lower event hazard for the palbociclib-plus-letrozole group within this confidence interval.
The P-value of <0.000001 addresses the statistical evidence against the null hypothesis used for the superiority comparison. It does not measure the magnitude of the treatment effect, the probability that the treatment works, or the probability that the null hypothesis is true.
As with any Cox-type time-to-event interpretation, a hazard ratio is most straightforward when the proportional-hazards assumption is reasonable. The registry result should therefore be interpreted as a model-based relative treatment effect rather than as a universal constant risk reduction for every participant throughout follow-up. Censoring and the timing of progression or death are also intrinsic to the PFS analysis.
7. Secondary Efficacy Results
Objective Response
Objective response
95% CI: 1.008–2.030 · P = 0.0224
Fisher exact test · two-sided superiority comparison
| Feature | Reported analysis |
|---|---|
| Endpoint | Objective Response as Assessed by the Investigator |
| Time frame | From randomization until end of treatment, up to approximately 2.5 years |
| Analysis population | ITT population / full analysis set |
| Method | Fisher exact test |
| Effect measure | Odds ratio |
| Estimate | 1.428 |
| 95% CI | 1.008–2.030 |
| P-value | 0.0224 |
| Stratification | By disease site (visceral vs non-visceral) per randomization |
An odds ratio of 1.428 means that the estimated odds of the binary response outcome were 1.428 times those in the comparator group under the reported analysis.
An odds ratio is not the same as a risk ratio or a percentage-point difference in response probability. Without the underlying arm-specific response proportions in the ClinicalTrials.gov record, the odds ratio should not be converted into an absolute response difference.
The 95% confidence interval, 1.008 to 2.030, is relatively close to the null value of 1 at its lower boundary. The interval therefore conveys substantially more uncertainty about the magnitude of the response association than the point estimate alone suggests.
The P-value of 0.0224 measures evidence against the relevant null hypothesis under the Fisher exact framework. It does not quantify the size or clinical importance of the response effect. It also should not be interpreted without considering that this was a secondary endpoint rather than the single registered primary endpoint.
Objective Response Among Participants With Measurable Disease
Response in measurable disease population
95% CI: 1.080–2.347 · P = 0.0090
Fisher exact test · two-sided superiority comparison
This analysis was restricted to participants who had measurable disease at baseline. The reported odds ratio was 1.594, with a two-sided 95% confidence interval of 1.080 to 2.347 and a P-value of 0.0090. The analysis was stratified by disease site, visceral versus non-visceral, per randomization.
The estimate indicates higher estimated odds of objective response in the palbociclib-plus-letrozole group within the measurable-disease analysis population. The confidence interval remains above 1, although its width indicates that the precise magnitude of the odds ratio is uncertain.
The analysis population is important: this result should not be generalized automatically to every randomized participant because the registry specifically defines the population as participants with measurable disease at baseline.
Disease Control / Clinical Benefit Response
Disease control / clinical benefit response
95% CI: 1.619–3.722 · P < 0.0001
Fisher exact test · two-sided superiority comparison
The reported odds ratio for disease control/clinical benefit response was 2.451, with a 95% confidence interval of 1.619 to 3.722 and P-value <0.0001. The analysis used the ITT population and was stratified by disease site.
An odds ratio of 2.451 indicates that the estimated odds of disease control/clinical benefit response were approximately 2.451 times those in the comparator group under the reported analysis.
The confidence interval is entirely above 1, indicating that the estimated association is on the treatment-favoring side throughout the reported 95% interval. Nevertheless, the odds ratio remains a relative measure; it does not provide the absolute probability of disease control in either group.
8. Biomarker-Defined PFS Analyses
The registry contains multiple secondary PFS analyses according to tumor tissue biomarker status. These analyses are particularly useful statistically because they illustrate how a single broad endpoint can be examined across biologically defined subgroups. They should, however, be distinguished from the single registered primary PFS endpoint.
| Biomarker-defined analysis | HR | 95% CI | P-value |
|---|---|---|---|
| ER positive | 0.571 | 0.443–0.737 | <0.0001 |
| ER negative | 0.405 | 0.218–0.751 | 0.0030 |
| Rb positive | 0.531 | 0.416–0.680 | <0.0001 |
| Rb negative | 0.675 | 0.308–1.481 | 0.3237 |
| Cyclin D1 positive | 0.555 | 0.437–0.705 | <0.0001 |
| Cyclin D1 negative | 0.997 | 0.287–3.461 | 0.9964 |
| p16 positive | 0.518 | 0.400–0.670 | <0.0001 |
| p16 negative | 0.731 | 0.392–1.364 | 0.3221 |
| p16 HScore <175 | 0.581 | 0.455–0.742 | <0.0001 |
| p16 HScore ≥175 | 0.255 | 0.100–0.650 | 0.0022 |
| Ki67 ≤20% | 0.530 | 0.379–0.742 | 0.0002 |
| Ki67 >20% | 0.569 | 0.409–0.791 | 0.0007 |
The registry's individual analyses are described as unstratified log-rank tests and also identify unstratified Cox proportional-hazards models in the analysis notes. The primary PFS analyses are reported as stratified log-rank tests. The biomarker-specific Cox models provide the hazard-ratio framework for these analyses.
9. Patient-Reported Outcomes
EQ-5D Index
Change from baseline: EQ-5D
95% CI: -0.004–0.051 · P = 0.0925
Repeated-measures mixed-effects model
The analysis used a Patient Reported Outcome Analysis Set consisting of a subset of ITT participants who had both baseline and at least one follow-up PRO assessment. The repeated-measures mixed-effects model included an intercept, treatment, time, treatment-by-time, and baseline as a covariate.
The estimated final-value mean difference was 0.023. The 95% confidence interval, -0.004 to 0.051, spans zero. The P-value was 0.0925.
This result concerns the modeled mean difference in the EQ-5D outcome, not PFS or OS. The mixed-effects approach is appropriate to the longitudinal structure because each participant can contribute repeated measurements over time and the model explicitly includes time and treatment-by-time.
The P-value does not indicate the magnitude of the difference. The estimate and confidence interval are the more direct measures of the estimated difference and its precision.
FACT-B
Change from baseline: FACT-B
95% CI: -2.63–1.98 · P = 0.7822
Repeated-measures mixed-effects model
The reported mean difference in final values was -0.325, with a 95% confidence interval from -2.63 to 1.98 and a P-value of 0.7822. The analysis used the same repeated-measures mixed-effects framework with treatment, time, treatment-by-time, and baseline as a covariate.
The confidence interval spans zero and includes both negative and positive differences. That means the reported estimate is compatible with a range of possible treatment differences under the model. The P-value of 0.7822 provides little evidence against a null mean difference, but it does not prove that the two groups are identical.
10. Overall Survival Results
Although PFS is the registered primary endpoint, the registry also reports two OS analyses: a primary OS analysis with a data cut-off date of 15-Nov-2021 and a final OS analysis through study completion.
Overall Survival: Primary Analysis
Primary OS analysis
95% CI: 0.777–1.177 · P = 0.337750
Stratified log-rank test · two-sided superiority comparison
| Feature | Primary OS analysis |
|---|---|
| Time frame | From randomization until death due to any cause or censored; assessed up to data cut-off 15-Nov-2021 |
| Population | ITT population / full analysis set |
| Method | Stratified log-rank test |
| Effect measure | Hazard ratio |
| Estimate | 0.956 |
| 95% CI | 0.777–1.177 |
| P-value | 0.337750 |
The estimated OS hazard ratio of 0.956 is close to 1. Under the reported analysis, the estimated instantaneous hazard of death was approximately 95.6% of the comparator hazard.
The 95% confidence interval of 0.777 to 1.177 crosses 1. Thus, the interval includes values corresponding to a lower hazard, approximately equal hazard, and a higher hazard for the palbociclib-plus-letrozole group relative to the comparator.
The P-value of 0.337750 does not measure the size of the observed OS difference. It quantifies the statistical evidence against the null hypothesis within the specified test framework. It should not be converted into a probability that one treatment is effective or ineffective.
Overall Survival: Final Analysis
Final OS analysis
95% CI: 0.755–1.124 · P = 0.208706
Stratified log-rank test · two-sided superiority comparison
| Feature | Final OS analysis |
|---|---|
| Time frame | From randomization until death due to any cause or censored; final analysis through study completion |
| Population | ITT population / full analysis set |
| Method | Stratified log-rank test |
| Effect measure | Hazard ratio |
| Estimate | 0.921 |
| 95% CI | 0.755–1.124 |
| P-value | 0.208706 |
The final OS estimate of 0.921 corresponds to an estimated hazard approximately 7.9% lower in the palbociclib-plus-letrozole group relative to placebo plus letrozole under the reported model.
The 95% confidence interval, 0.755 to 1.124, crosses 1 and is therefore compatible with a range of relative hazards on both sides of the null value. The interval is the key description of statistical precision; the point estimate alone should not be treated as the complete result.
The P-value of 0.208706 should be interpreted as a test statistic under the specified superiority framework rather than as a measure of treatment effect magnitude. The registry does not supply median OS values in the ClinicalTrials.gov record, so no median survival estimate is presented here.
11. Safety
The ClinicalTrials.gov record reports serious adverse events by randomized treatment arm. The denominator is the number at risk in each arm for this safety summary.
| Safety measure | Palbociclib Plus Letrozole | Placebo Plus Letrozole |
|---|---|---|
| Serious adverse events | 125 / 444 | 38 / 222 |
The ClinicalTrials.gov record does not provide a formal comparative statistical analysis for serious adverse events. Accordingly, these figures are presented descriptively rather than as a treatment-effect estimate. The denominators also differ from the overall enrollment of 666, so the safety population represented by this specific registry field should not be assumed to be identical to the randomized population.
12. Statistical Methods Explained
Why was a stratified log-rank test used for the primary PFS endpoint?
PFS is a time-to-event endpoint because each participant has a time from randomization until progression or death, or a censoring time if the event is not observed. The log-rank test compares the survival experience of the treatment groups across follow-up. Stratification allows the comparison to respect the trial's stratified analysis framework rather than collapsing all participants into a single unstratified comparison.
What does a hazard ratio of 0.576 mean?
It is a relative measure of the event hazard under the reported time-to-event analysis. An HR of 0.576 corresponds to an estimated hazard 57.6% of the comparator hazard, or approximately a 42.4% lower estimated hazard. It does not mean that 42.4% of participants experienced benefit or that each participant's individual risk fell by exactly 42.4%.
Why does the confidence interval matter?
A point estimate is only one estimate from the observed data. The 95% confidence interval of 0.463–0.718 communicates the statistical uncertainty around the primary PFS hazard-ratio estimate. A narrower interval would indicate greater precision; a wider interval would indicate less precision. The interval should be considered alongside, rather than replaced by, the P-value.
Why is the odds ratio not the same as a risk ratio?
For binary endpoints, the odds of response are defined as the probability of response divided by the probability of non-response. The odds ratio compares those odds between treatment groups. A risk ratio instead compares probabilities directly. The two measures can differ substantially, particularly when the outcome is not rare.
Why was Fisher exact testing used for response endpoints?
The registry reports Fisher exact tests for the binary response outcomes. Fisher's exact test evaluates the treatment-by-response association without relying on the large-sample approximation used by some other categorical-data tests. The associated odds ratio provides an effect-size description, while the P-value addresses evidence against the relevant null hypothesis.
Why use a mixed-effects model for EQ-5D and FACT-B?
These outcomes are longitudinal: participants can have baseline and follow-up measurements. A repeated-measures mixed-effects model can represent treatment, time, and treatment-by-time effects while incorporating baseline as a covariate. This is structurally different from a single time-point comparison because it uses the repeated measurement framework specified in the registry.
Why is ITT important in a randomized trial?
ITT analysis preserves the original randomized groups. If participants are reclassified after randomization according to treatment actually received, the treatment groups can become less comparable because post-randomization behavior is no longer controlled by randomization. The PALOMA-2 primary PFS analysis was based on the ITT population/full analysis set described in the registry.
13. Understanding the Primary PFS Result
What the estimate says
HR 0.576 indicates a lower estimated event hazard for palbociclib plus letrozole than for placebo plus letrozole under the reported PFS analysis.
What it does not say
It is not a percentage of patients who benefited, an absolute risk difference, or a guarantee that every patient's risk changed by the same proportion.
What the CI says
The 95% CI of 0.463–0.718 describes uncertainty around the estimated relative hazard and remains below 1 throughout the interval.
What the P-value says
P < 0.000001 indicates very strong statistical evidence against the null hypothesis in the specified superiority test. It does not measure effect size.
The primary PFS result is therefore best understood as a combination of three pieces of information: the effect estimate, the uncertainty interval, and the hypothesis-test result. None of these pieces alone completely describes the evidence.
14. Analysis Populations and Why They Matter
| Population | Role in registry-reported analyses |
|---|---|
| ITT population / full analysis set | Primary PFS, response, disease control, biomarker PFS, and OS analyses. Includes randomized participants according to initial treatment assignment. |
| Participants with measurable disease at baseline | Objective response analysis restricted to participants with measurable disease. |
| PRO Analysis Set | Subset of ITT participants with both baseline and at least one follow-up patient-reported outcome assessment. |
The distinction between these populations is not merely administrative. The primary PFS and OS analyses retain the randomized comparison, whereas the PRO analysis set is defined by availability of baseline and follow-up measurements. That difference affects the population to which the corresponding estimate applies.
15. Stratified Analysis
The registry identifies stratified analysis as an important concept for PALOMA-2. For objective response and disease control, the analysis notes specify stratification by disease site, specifically visceral versus non-visceral, per randomization.
Stratification can improve the alignment between the analysis and the randomization scheme and can reduce the influence of imbalances in important baseline factors. It does not mean that the treatment effect is estimated separately and independently within every stratum for purposes of the primary conclusion.
For the response analyses, disease site was explicitly identified as the stratification factor. For the primary PFS analysis, the registry reports a stratified log-rank test but the ClinicalTrials.gov record does not provide the specific stratification factors for that primary analysis. No additional stratification factors are therefore introduced here.
16. Confidence Intervals and P-values Across Endpoints
| Endpoint | Effect | 95% CI | P-value |
|---|---|---|---|
| Primary PFS | HR 0.576 | 0.463–0.718 | <0.000001 |
| Objective response | OR 1.428 | 1.008–2.030 | 0.0224 |
| Objective response, measurable disease | OR 1.594 | 1.080–2.347 | 0.0090 |
| Disease control / CBR | OR 2.451 | 1.619–3.722 | <0.0001 |
| EQ-5D | Mean difference 0.023 | -0.004–0.051 | 0.0925 |
| FACT-B | Mean difference -0.325 | -2.63–1.98 | 0.7822 |
| OS primary analysis | HR 0.956 | 0.777–1.177 | 0.337750 |
| OS final analysis | HR 0.921 | 0.755–1.124 | 0.208706 |
These results demonstrate why effect measures should be interpreted within their endpoint type. Hazard ratios describe time-to-event comparisons, odds ratios describe relative odds for binary outcomes, and mean differences describe differences in continuous outcomes. A P-value from one endpoint cannot be directly compared with the P-value from another as though they represented the same quantity.
17. Limitations
- Registry-level reporting: the ClinicalTrials.gov record contains selected statistical analyses rather than every element of a full statistical analysis plan. Conclusions should therefore remain limited to the information reported here.
- Missing absolute survival summaries: the ClinicalTrials.gov record does not provide median PFS or median OS values, so they are not reported.
- Hazard-ratio interpretation: a single HR is a model-based relative measure and should not be interpreted as a constant individual-level risk reduction without considering the underlying survival structure and proportional-hazards assumption.
- Secondary endpoints: response, disease control, biomarker analyses, PROs, and OS are distinct from the single registered primary PFS endpoint. Their results should not be treated as interchangeable confirmatory evidence.
- Multiple biomarker analyses: the registry reports many biomarker-defined PFS analyses. The ClinicalTrials.gov record does not provide a multiplicity adjustment across this collection of analyses, so individual P-values should not automatically be interpreted as if they were each independent confirmatory tests.
- Subgroup precision: several biomarker estimates have wide confidence intervals, illustrating the additional uncertainty created by subgroup analyses.
- Different analysis populations: the measurable-disease response analysis and PRO analyses apply to populations that differ from the overall ITT population.
- Safety denominators: the registry-reported serious-adverse-event summary uses 444 and 222 as denominators, which differ from the overall enrollment of 666. The specific safety population represented by these figures should therefore not be assumed to equal the randomized population.
18. Why This Trial Matters Statistically
PALOMA-2 is a useful statistical teaching case because it connects a randomized superiority trial with several common methods used across modern clinical research. Its registry record includes a primary time-to-event analysis, binary response analyses, longitudinal patient-reported outcomes, biomarker-defined survival analyses, and mature overall-survival analyses.
| Concept | How it appears in PALOMA-2 |
|---|---|
| Randomization | Participants were randomized to two parallel treatment arms. |
| Blinding | The registry identifies quadruple masking. |
| Intention-to-treat analysis | The primary PFS and OS analyses use the ITT population / full analysis set. |
| Time-to-event endpoint | PFS and OS are analyzed from randomization to an event or censoring. |
| Stratified log-rank test | Used for the primary PFS analysis and reported for OS analyses. |
| Hazard ratio | Used to express the relative PFS and OS treatment effect. |
| Confidence interval | Quantifies uncertainty around HR, OR, and mean-difference estimates. |
| Fisher exact test | Used for binary objective-response and disease-control analyses. |
| Odds ratio | Used as the effect measure for binary response endpoints. |
| Mixed-effects model | Used for repeated EQ-5D and FACT-B measurements. |
| Covariate adjustment | Baseline was included as a covariate in the repeated-measures PRO models. |
| Biomarker subgroup analysis | PFS was examined across multiple tumor tissue biomarker categories. |
19. Related Tutorials
Learn more about the methods used in this trial:
20. Related Statistical Calculators
21. Sources
- ClinicalTrials.gov: PALOMA-2, NCT01740427.
- Linked publication: PubMed record for PMID 38951507.
- Linked publication: PubMed record for PMID 38861871.
- Linked publication: PubMed record for PMID 38252901.
- Linked publication: PubMed record for PMID 38107828.
- Linked publication: PubMed record for PMID 36463643.
Continue studying the statistical methods
The PALOMA-2 analysis connects randomized trial design with survival analysis, binary endpoint methods, confidence intervals, and longitudinal mixed-effects models.
22. Record Summary
PALOMA-2 provides a compact example of how different clinical-trial endpoints require different statistical tools. The registered primary endpoint was investigator-assessed PFS, analyzed in the ITT population using a stratified log-rank test, with a reported hazard ratio of 0.576 (95% CI 0.463–0.718; P < 0.000001). Secondary binary outcomes used Fisher exact testing and odds ratios, while EQ-5D and FACT-B used repeated-measures mixed-effects models with baseline covariate adjustment. OS was also analyzed using stratified log-rank methods, with a primary-analysis HR of 0.956 and a final-analysis HR of 0.921.