This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the data reported in the ClinicalTrials.gov record.
1. Trial at a Glance
SOPHIA was a randomized, parallel, open-label phase 3 trial evaluating margetuximab plus physician's choice of chemotherapy versus trastuzumab plus physician's choice of chemotherapy in patients with HER2-positive metastatic breast cancer. The registry reports 624 participants, three arms, and primary time-to-event analyses for progression-free survival and overall survival.
| Feature | SOPHIA |
|---|---|
| Trial name | SOPHIA |
| Phase | Phase 3 |
| Condition | HER-2 Positive Breast Cancer; Metastatic Neoplasm |
| Design | Randomized, parallel, unmasked |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 624 |
| Primary endpoints | Progression-free survival; overall survival; number of patients with Grade 3 or higher infusion-related reactions |
| Primary statistical method reported | Log-rank test |
| Primary effect measure | Hazard ratio |
| Hypothesis type | Superiority |
| ClinicalTrials.gov | NCT02492711 |
2. Clinical Question
The central statistical question was whether margetuximab plus chemotherapy differed from trastuzumab plus chemotherapy with respect to time to disease progression or death and time to death from any cause.
Population
Patients enrolled in the phase 3 SOPHIA study with HER-2 positive breast cancer and metastatic neoplasm.
Intervention
Margetuximab plus physician's choice of chemotherapy.
Comparator
Trastuzumab plus physician's choice of chemotherapy.
Primary question
Does margetuximab plus chemotherapy improve the prespecified time-to-event outcomes relative to trastuzumab plus chemotherapy?
3. Trial Design
Margetuximab Plus Chemotherapy
- Margetuximab, classified in the registry as a biological intervention.
- Physician's choice of chemotherapy.
- Primary efficacy comparison against trastuzumab plus chemotherapy.
Trastuzumab Plus Chemotherapy
- Trastuzumab, classified in the registry as a biological intervention.
- Physician's choice of chemotherapy.
- Comparator for the primary efficacy analyses.
4. Endpoints
The registry lists three primary endpoints. Two are time-to-event outcomes with formal statistical analyses posted; the third concerns Grade 3 or higher infusion-related reactions during Cycle 2 of margetuximab treatment.
| Endpoint | Registry definition | Time frame | Endpoint type |
|---|---|---|---|
| Progression-free Survival (PFS) as Determined by Independent Radiological Review. | PFS is measured from the time of randomization until first documented disease progression or death from any cause, whichever is first. | Tumor assessments are conducted every 6 weeks for the first 24 weeks and then every 24 weeks until progression of cancer, up to 6.5 years. | Time-to-event |
| Overall Survival (OS) Defined as the Number of Days From Randomization to the Date of Death (From Any Cause). | Overall survival is the time from randomization until death from any cause | Throughout the study, average 21 months | Time-to-event |
| Number of Patients With Grade 3 or Higher Infusion Related Reactions | Incidence of Grade 3 or higher infusion-related reactions for patients receiving 60-minute or 30-minute infusions of margetuximab in Cycle 2 of treatment | 22 days | Other / unclear |
5. Statistical Methodology
Primary time-to-event framework
The registry reports the log-rank test as the method for both posted primary efficacy analyses. The corresponding effect measure is the hazard ratio (HR), with two-sided 95% confidence intervals.
The analysis notes specify a stratified log-rank test and a stratified Cox proportional model. Thus, the statistical comparison has two complementary components: the log-rank procedure evaluates the time-to-event distributions between the randomized groups, while the Cox model supplies the hazard-ratio estimate and its confidence interval.
An HR below 1 indicates a lower estimated instantaneous event rate in the margetuximab group relative to the trastuzumab group under the fitted time-to-event model. It is a relative measure, not an absolute probability.
Analysis populations
For both primary PFS and OS, the registry states that the outcome measure was calculated only for the randomized study efficacy population. The 88 participants in the sub-study were not part of this efficacy population.
| Endpoint | Analysis population reported by the registry | Posted method |
|---|---|---|
| Independent-review PFS | Randomized study efficacy population; 88 sub-study participants excluded from the efficacy population | Stratified log-rank test; stratified Cox proportional model |
| OS | Randomized study efficacy population; 88 sub-study participants excluded from the efficacy population | Stratified log-rank test; stratified Cox proportional model |
| Investigator-assessed PFS | Randomized study population; analysis population consisted of patients who had investigator-assessed responses to treatment as of the interim analysis | Stratified log-rank test; stratified Cox proportional model |
Stratified analysis
The registry identifies stratified analysis as an additional concept in all three posted statistical analyses. The ClinicalTrials.gov record does not identify the specific stratification variables, so this page does not infer them.
Interim analysis and alpha spending
The secondary investigator-assessed PFS analysis explicitly includes interim analysis / alpha spending among its statistical concepts. This indicates that the interpretation of that analysis involves an interim-analysis framework rather than treating the reported p-value as if it arose from an isolated single final look.
6. Results: Primary Progression-Free Survival
The registry reports a formal superiority analysis of progression-free survival as determined by independent radiological review. The comparison was margetuximab plus chemotherapy versus trastuzumab plus chemotherapy.
Hazard ratio for progression or death
95% CI: 0.593–0.979 · P = 0.0334
Two-sided confidence interval; stratified log-rank test and stratified Cox proportional model.
| Feature | Primary PFS analysis |
|---|---|
| Endpoint | Progression-free Survival (PFS) as Determined by Independent Radiological Review. |
| Comparison | Margetuximab Plus Chemotherapy vs Trastuzumab Plus Chemotherapy |
| Analysis population | Randomized study efficacy population |
| Effect measure | Hazard ratio |
| Estimate | 0.76 |
| 95% CI | 0.593–0.979 |
| P-value | 0.0334 |
| Hypothesis | Superiority |
The reported HR of 0.76 means that the estimated instantaneous rate of progression or death was approximately 24% lower in the margetuximab-plus-chemotherapy group than in the trastuzumab-plus-chemotherapy group under the fitted hazard model. The arithmetic interpretation follows directly from 1 − 0.76 = 0.24.
The HR does not mean that 24% of patients avoided progression, that each patient had exactly a 24% lower probability of progression, or that median PFS was reduced or increased by 24%. It is a relative time-to-event measure.
The two-sided 95% CI of 0.593–0.979 describes uncertainty around the estimated hazard ratio under the statistical model and sampling framework. It is narrower than a very imprecise estimate would be, but it still spans a range of plausible relative treatment effects rather than identifying one exact effect.
The P-value of 0.0334 addresses the evidence against the null hypothesis in the prespecified statistical comparison. It does not measure the size of the treatment effect, the probability that the treatment works, or the clinical importance of the effect.
Because the analysis is based on a Cox proportional-hazards model, interpretation of one HR also depends on the model's proportional-hazards framework. The ClinicalTrials.gov record does not provide diagnostics for that assumption, so the HR should be interpreted as the model-based summary reported by the trial rather than as a complete description of how the treatment effects behaved at every point in time.
7. Results: Primary Overall Survival
The second formal primary efficacy analysis evaluated overall survival, defined as the number of days from randomization to the date of death from any cause.
Hazard ratio for death
95% CI: 0.774–1.165 · P = 0.6204
Two-sided confidence interval; stratified log-rank test and stratified Cox proportional model.
| Feature | Primary OS analysis |
|---|---|
| Endpoint | Overall Survival (OS) Defined as the Number of Days From Randomization to the Date of Death (From Any Cause). |
| Comparison | Margetuximab Plus Chemotherapy vs Trastuzumab Plus Chemotherapy |
| Analysis population | Randomized study efficacy population |
| Effect measure | Hazard ratio |
| Estimate | 0.950 |
| 95% CI | 0.774–1.165 |
| P-value | 0.6204 |
| Hypothesis | Superiority |
The reported HR of 0.950 is close to 1.00. Under the fitted time-to-event model, the estimated instantaneous rate of death in the margetuximab-plus-chemotherapy group was approximately 95% of that in the trastuzumab-plus-chemotherapy group.
This does not mean that the two treatments have identical survival for every patient. The estimate is subject to statistical uncertainty, and individual patients can experience very different event times.
The two-sided 95% CI of 0.774–1.165 is particularly important here because it includes 1.00. The interval indicates that the observed estimate is compatible with a range of relative hazard values on both sides of the null value under the model and sampling framework.
The P-value of 0.6204 is not a measure of how large or small the treatment effect is. Rather, it quantifies the compatibility of the observed test statistic with the null hypothesis under the specified statistical test. A large p-value does not prove that the treatment effects are identical.
The analysis was designed as a superiority comparison. Therefore, this result should not be reinterpreted as a formal non-inferiority test or as proof of equivalence. A superiority analysis and a non-inferiority analysis answer different statistical questions and use different decision rules.
8. Primary Infusion-Related Reaction Endpoint
The third registered primary endpoint was the number of patients with Grade 3 or higher infusion-related reactions. The registry defines this as the incidence of Grade 3 or higher infusion-related reactions among patients receiving 60-minute or 30-minute infusions of margetuximab in Cycle 2 of treatment, with a time frame of 22 days.
For a binary safety endpoint of this type, an analysis could ordinarily summarize the number and proportion of patients experiencing the event and, when a randomized comparison is intended, compare event risks using an appropriate exact or asymptotic method. The appropriate method depends on the actual analysis population, event counts, and prespecified statistical plan. None of those additional numerical details are reported here for this endpoint.
9. Secondary Endpoint Result: Investigator-Assessed PFS
The registry also reports a formal secondary analysis of progression-free survival as assessed by study investigators. The analysis population consisted of patients who had investigator-assessed responses to treatment as of the interim analysis, and the outcome was calculated only for the randomized study population.
Investigator-assessed PFS hazard ratio
95% CI: 0.556–0.870 · P = 0.0014
Two-sided confidence interval; stratified log-rank test and stratified Cox proportional model.
| Feature | Secondary investigator-assessed PFS analysis |
|---|---|
| Comparison | Margetuximab Plus Chemotherapy vs Trastuzumab Plus Chemotherapy |
| Analysis population | Randomized study population; patients with investigator-assessed responses as of the interim analysis |
| Effect measure | Hazard ratio |
| Estimate | 0.70 |
| 95% CI | 0.556–0.870 |
| P-value | 0.0014 |
| Hypothesis | Superiority |
| Additional statistical concepts | Interim analysis / alpha spending; stratified analysis |
An HR of 0.70 corresponds to an estimated instantaneous rate of progression or death approximately 30% lower in the margetuximab-plus-chemotherapy group under the fitted model.
The estimate does not imply that 30% of patients experienced a benefit, that each patient had a 30% lower probability of progression, or that survival time increased by a fixed 30%. Those interpretations would require different measures.
The 95% CI of 0.556–0.870 quantifies uncertainty around the estimated relative hazard. It remains entirely below 1.00, although the width of the interval demonstrates that the exact magnitude of the treatment effect is not known with certainty.
The P-value of 0.0014 is evidence against the null hypothesis under the reported test. It is not an effect-size metric and should not be read as a probability that the treatment effect is real.
Because this analysis is explicitly associated with an interim analysis / alpha-spending framework, its p-value must be interpreted in the context of the trial's sequential monitoring plan. The ClinicalTrials.gov record does not identify the exact alpha-spending function or boundary, so no additional numerical inference is made here.
10. Comparing the Two PFS Analyses
The registry provides two different PFS analyses: one based on independent radiological review as a primary endpoint and one based on study-investigator assessment as a secondary endpoint. Both use the same broad time-to-event framework but have different assessment sources and analysis roles.
| Feature | Independent-review PFS | Investigator-assessed PFS |
|---|---|---|
| Endpoint role | Primary | Secondary |
| Assessment | Independent radiological review | Study investigators |
| HR | 0.76 | 0.70 |
| 95% CI | 0.593–0.979 | 0.556–0.870 |
| P-value | 0.0334 | 0.0014 |
| Statistical method | Stratified log-rank test; stratified Cox proportional model | Stratified log-rank test; stratified Cox proportional model |
| Interim-analysis concept | Not specified in the ClinicalTrials.gov record | Interim analysis / alpha spending explicitly identified |
The two hazard ratios should not be treated as competing estimates in which one must be selected as the "true" treatment effect. They arise from different PFS assessment frameworks and occupy different roles in the statistical hierarchy. The primary endpoint should retain its prespecified role when describing the main efficacy analysis, while the investigator-assessed result provides a related secondary assessment.
11. Statistical Methods Explained
Why was a log-rank test used?
Both primary efficacy endpoints are time-to-event outcomes. A log-rank test is designed to compare event-time distributions between groups while incorporating information from patients who have not yet experienced the event at the time of analysis. This is important because simply comparing the proportion of patients who have progressed or died would discard information about when events occurred and how long patients were followed without an event.
What does an HR of 0.76 mean?
An HR of 0.76 is a relative model-based comparison of instantaneous event rates. It corresponds to an estimated 24% lower instantaneous rate in the margetuximab group relative to the trastuzumab group under the fitted model. It is not a 24-percentage-point difference in the probability of progression and does not describe an individual's treatment effect.
Why does the confidence interval matter?
A point estimate such as 0.76 is only one estimate from the observed data. The 95% CI of 0.593–0.979 communicates the statistical uncertainty surrounding that estimate. The same principle applies to the OS estimate of 0.950 and its interval of 0.774–1.165. A confidence interval therefore adds information that a point estimate alone cannot provide.
Why doesn't the p-value measure effect size?
The p-value concerns the evidence against a specified null hypothesis under the statistical model. It depends on the effect estimate, its variability, and the amount of information in the analysis. Consequently, a smaller p-value does not automatically represent a larger clinical effect, and a larger p-value does not establish that there is no effect.
Why use a stratified analysis?
The registry identifies the primary analyses as stratified log-rank tests with stratified Cox proportional models. Stratification allows the time-to-event comparison to account for prespecified strata rather than treating all participants as though the relevant stratification structure were absent. The ClinicalTrials.gov record does not identify the specific strata, so this page does not name them.
Why does interim analysis matter?
The secondary investigator-assessed PFS analysis is explicitly associated with interim analysis and alpha spending. When accumulating trial data are examined more than once, the statistical design must account for repeated looks if the overall type I error is to remain controlled according to the prespecified plan. The ClinicalTrials.gov record identifies the concept but do not specify the exact spending function or boundary.
Why does the analysis population matter?
The primary PFS and OS outcomes were calculated only for the randomized study efficacy population, excluding the 88 participants in the sub-study. Analysis populations define which observations contribute to an estimate. Mixing participants from different study components could change the estimand and make a reported result difficult to interpret.
12. Understanding the Hazard Ratio More Deeply
The distance of the HR from 1 describes the relative treatment effect estimated by the model. It does not directly provide a median survival difference, absolute risk difference, or number needed to treat.
PFS HR = 0.76
The estimated instantaneous rate of progression or death was approximately 76% of the comparator rate under the fitted model.
OS HR = 0.950
The estimated instantaneous rate of death was approximately 95% of the comparator rate under the fitted model.
Investigator PFS HR = 0.70
The estimated instantaneous rate of progression or death was approximately 70% of the comparator rate under the fitted model.
HR is not a probability
A hazard ratio should not be interpreted as the probability that a particular patient will experience progression, death, or benefit.
The three HRs illustrate why endpoint definition matters. PFS and OS are different event processes, and independent-review PFS and investigator-assessed PFS use different assessment frameworks. Even though all three analyses use a hazard ratio, their estimates should not be collapsed into one overall statistic.
13. P-Values and Confidence Intervals in SOPHIA
| Analysis | HR | 95% CI | P-value | Null value inside CI? |
|---|---|---|---|---|
| Primary independent-review PFS | 0.76 | 0.593–0.979 | 0.0334 | No |
| Primary OS | 0.950 | 0.774–1.165 | 0.6204 | Yes |
| Secondary investigator-assessed PFS | 0.70 | 0.556–0.870 | 0.0014 | No |
For a hazard ratio, the null value is 1.00. A confidence interval that excludes 1.00 is consistent with evidence against the null hypothesis at the corresponding two-sided confidence level, assuming the interval and test are based on compatible procedures. That statistical relationship should not be confused with clinical importance.
14. Interim Analysis and Alpha Spending
The registry specifically identifies interim analysis / alpha spending as a concept in the secondary investigator-assessed PFS analysis. This is an important feature of clinical-trial statistics because repeated examination of accumulating data can affect the probability of a false-positive result if the design does not account for those repeated looks.
Why alpha spending is used
An alpha-spending framework distributes the allowable type I error across planned analyses or information times rather than treating every interim look as though it were the only analysis.
What is not reported here
The trial data provided do not specify the spending function, information fractions, stopping boundaries, or exact alpha allocated to the interim analysis.
Consequently, the reported p-value of 0.0014 should be read as the registry-reported result within the documented interim-analysis framework, not as a basis for reconstructing an unreported sequential-testing design.
15. Stratified Survival Analysis
Stratification appears explicitly in the statistical-analysis descriptions for all three posted efficacy analyses. The primary analyses are described as stratified log-rank tests with stratified Cox proportional models.
The purpose is to preserve the analysis structure associated with the trial design and avoid ignoring known stratification factors. The ClinicalTrials.gov record does not state which variables define the strata.
A stratified Cox model generally estimates a common treatment hazard ratio across the strata while allowing the baseline hazard to differ between strata. That description explains the statistical role of stratification without assuming unreported details about the SOPHIA protocol.
16. Safety Results
The ClinicalTrials.gov record reports serious adverse events by study arm as affected patients divided by patients at risk. These data are presented exactly as reported in the registry rather than converted into additional derived statistics.
| Study group | Serious adverse events, affected / at risk |
|---|---|
| Margetuximab Plus Chemotherapy | 47 / 264 |
| Trastuzumab Plus Chemotherapy | 51 / 266 |
| Margetuximab Infusion Substudy | 17 / 88 |
The serious-adverse-event counts should be kept distinct from the primary efficacy outcomes. They answer a safety question rather than a time-to-progression or overall-survival question.
17. Analysis Population and the 88-Participant Substudy
The registry makes an explicit distinction between the randomized efficacy population and an 88-participant substudy. This distinction is statistically important because a trial can contain multiple study components with different purposes and populations.
| Population component | Role in the ClinicalTrials.gov record |
|---|---|
| Randomized study efficacy population | Used for primary PFS and OS outcome measures |
| 88-participant substudy | Not part of the efficacy population for primary PFS and OS |
| Infusion substudy safety group | 17 / 88 reported for serious adverse events |
This distinction prevents a common statistical mistake: treating every participant in a clinical trial record as though every participant contributes to every endpoint. The estimand for an endpoint is determined not only by the outcome definition but also by the population included in its analysis.
18. What the PFS Result Does — and Does Not — Mean
The primary independent-review PFS analysis reported an HR of 0.76. Under the fitted time-to-event model, the estimated instantaneous rate of progression or death was approximately 24% lower with margetuximab plus chemotherapy than with trastuzumab plus chemotherapy.
This does not mean that 24% of patients were protected from progression, that PFS time increased by 24%, or that every patient experienced the same relative reduction.
The 95% CI of 0.593–0.979 communicates uncertainty around the HR estimate. The point estimate is not a complete description of the statistical evidence; the interval shows how broad the compatible range of relative hazard values is under the model and sampling framework.
The P-value of 0.0334 evaluates the observed statistical evidence against the relevant null hypothesis under the reported test. It does not quantify the magnitude of benefit or the probability that an individual patient will benefit.
19. What the OS Result Does — and Does Not — Mean
The primary OS analysis reported an HR of 0.950. Under the fitted model, the estimated instantaneous rate of death in the margetuximab-plus-chemotherapy group was approximately 95% of that in the trastuzumab-plus-chemotherapy group.
The estimate does not establish that the treatments are identical. It is an estimate with uncertainty, reflected by the 95% CI of 0.774–1.165.
Because the interval crosses the HR null value of 1.00, the reported data are compatible with relative hazard values below and above 1.00 under the model. The interval therefore provides more information than the point estimate alone.
The registry identifies the hypothesis type as superiority. Therefore, an OS result that does not provide evidence of superiority should not be relabeled as a non-inferiority or equivalence result. Those are separate statistical hypotheses with separate design requirements.
20. Missing Data, Censoring, and Time-to-Event Interpretation
Time-to-event analysis differs from an analysis of a simple binary endpoint because patients can contribute follow-up even when they have not yet experienced the event. A patient who remains progression-free at the time their observation ends is not equivalent to a patient who progressed immediately; the timing of observation contributes information to the analysis.
The ClinicalTrials.gov record defines PFS as time from randomization until first documented progression or death, whichever occurs first, and OS as time from randomization until death from any cause. These definitions imply that time-to-event methods are central to the analysis.
For censored observations, Kaplan-Meier estimation and Cox regression are commonly used because they can incorporate differing follow-up durations. The registry's posted methodology identifies the log-rank and Cox approaches but does not provide a complete censoring or missing-data analysis plan in the ClinicalTrials.gov record.
21. Limitations
- Limited registry detail: the ClinicalTrials.gov record identifies the major statistical methods but do not provide the complete statistical analysis plan.
- Unspecified stratification variables: stratified analysis is reported, but the ClinicalTrials.gov record does not identify the variables used to define the strata.
- Interim-analysis detail: alpha spending is identified for the investigator-assessed PFS analysis, but the exact spending function and stopping boundary are not reported.
- No median survival estimates reported: the ClinicalTrials.gov record contains hazard ratios, confidence intervals, and p-values but not median PFS or OS.
- No absolute efficacy estimates reported: the provided statistical analyses do not include Kaplan-Meier survival probabilities or absolute risk differences.
- Different PFS assessment frameworks: independent radiological review and investigator assessment should not be treated as identical endpoints even though both are analyzed as time-to-event outcomes.
- Analysis-population distinction: the 88 participants in the substudy were excluded from the randomized efficacy population for the primary PFS and OS analyses.
- Safety analysis limitations: serious adverse-event counts are reported by arm, but the ClinicalTrials.gov record does not include a formal comparative statistical analysis of those events.
- Proportional-hazards interpretation: a single Cox HR is a model-based summary and should not automatically be interpreted as a constant relative effect at every time point.
- No non-inferiority interpretation: the registered hypothesis type is superiority, so the OS result should not be interpreted using a non-inferiority margin.
22. Why This Trial Matters Statistically
SOPHIA is a useful teaching example because it combines randomized treatment allocation with multiple time-to-event analyses, stratified log-rank testing, Cox proportional-hazards modeling, two-sided confidence intervals, p-values, an interim-analysis concept, and a distinct infusion-related safety endpoint.
| Concept | How it appears in SOPHIA |
|---|---|
| Randomization | The registry identifies the allocation as randomized. |
| Parallel design | The registry identifies a parallel design model. |
| Time-to-event endpoints | PFS and OS are both primary time-to-event outcomes. |
| Log-rank test | Reported for the primary PFS and OS analyses and the secondary investigator-assessed PFS analysis. |
| Hazard ratio | Used as the reported effect measure for the posted efficacy analyses. |
| Confidence interval | Two-sided 95% CIs accompany all three posted efficacy estimates. |
| Stratified analysis | Identified in the analysis descriptions for PFS and OS. |
| Cox model | Stratified Cox proportional model reported for the efficacy analyses. |
| Interim analysis | Explicitly identified for the investigator-assessed PFS analysis. |
| Alpha spending | Identified as an analysis concept for the secondary investigator-assessed PFS endpoint. |
| Superiority testing | The posted analyses are identified as superiority hypotheses. |
| Analysis populations | The primary efficacy analyses exclude the 88 substudy participants from the randomized efficacy population. |
| Safety endpoint | Grade 3 or higher infusion-related reactions are registered as a primary endpoint. |
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Calculators
25. Sources
- ClinicalTrials.gov: SOPHIA, NCT02492711. Official trial registry record.
- PubMed: PMID 36332179.
- PubMed: PMID 33480963.
Continue through the Clinical Biostats statistical pathway
Use the trial's endpoints and methods as a starting point for deeper study of survival analysis, hazard ratios, confidence intervals, interim analyses, and clinical-trial methodology.
26. Record Summary
SOPHIA provides a clear example of how randomized clinical-trial evidence can be expressed through time-to-event statistics. The ClinicalTrials.gov record reports a primary independent-review PFS hazard ratio of 0.76 with a two-sided 95% CI of 0.593–0.979 and P = 0.0334, alongside a primary OS hazard ratio of 0.950 with a two-sided 95% CI of 0.774–1.165 and P = 0.6204. A secondary investigator-assessed PFS analysis reported an HR of 0.70, 95% CI 0.556–0.870, and P = 0.0014.
The statistical story is not contained in the hazard ratios alone. Interpretation also depends on the endpoint definitions, the randomized efficacy population, the use of stratified log-rank and Cox methods, the interim-analysis and alpha-spending context of the secondary PFS analysis, and the distinction between efficacy and safety endpoints. The registry data also make clear that the 88-participant substudy was excluded from the primary efficacy population.