← Clinical Trials
Metastatic Breast Cancer Phase 3 Completed NCT03262935

TULIP: Complete Statistical Analysis of SYD985 in HER2-positive Metastatic Breast Cancer

An independent statistical review of the randomized phase 3 TULIP trial comparing (vic-)trastuzumab duocarmazine with physician's choice in participants with HER2-positive locally advanced or metastatic breast cancer.

Trial start: 15 December 2017  ·  Primary completion: 31 March 2021  ·  Sponsor: Byondis B.V.
Scope of this record

This page separates reported trial results from statistical interpretation. Numerical results are restricted to the information reported in the ClinicalTrials.gov record. ClinicalTrials.gov provides the official trial registry record.

1. Trial at a Glance

TULIP was a randomized, parallel-group, open-label phase 3 trial evaluating (vic-)trastuzumab duocarmazine versus physician's choice in participants with HER2-positive locally advanced or metastatic breast cancer. The primary endpoint was progression-free survival, analyzed using a stratified log-rank test with a stratified Cox regression model for the hazard ratio.

437
Enrolled
2 treatment arms
0.6401
Primary PFS HR
95% CI 0.4885–0.8389
0.002
PFS P-value
Two-sided
3
Stratification factors
Used in primary Cox analysis
FeatureTULIP
PhasePhase 3
ConditionMetastatic breast cancer
PopulationParticipants with HER2-positive locally advanced or metastatic breast cancer
DesignRandomized, parallel-group, open-label
AllocationRandomized
Primary purposeTreatment
Primary endpointProgression-free survival
Primary hypothesisSuperiority
Enrollment437
Lead sponsorByondis B.V.
ClinicalTrials.govNCT03262935

2. Clinical Question

The central statistical question was whether (vic-)trastuzumab duocarmazine improved progression-free survival compared with physician's choice in participants with HER2-positive locally advanced or metastatic breast cancer.

Population

Participants with HER2-positive locally advanced or metastatic breast cancer.

Intervention

(Vic-)trastuzumab duocarmazine.

Comparator

Physician's choice.

Primary question

Does (vic-)trastuzumab duocarmazine improve progression-free survival relative to physician's choice under a superiority framework?

3. Trial Design

01
Randomize437 participants
02
Two armsActive treatment vs physician's choice
03
FollowTime-to-event outcomes
04
AssessPFS and secondary outcomes
05
AnalyzeStratified statistical methods
Design model
Parallel-group randomized trial with two treatment arms.
Masking
None. The trial was open-label.
Primary purpose
Treatment.
Hypothesis
Superiority of (vic-)trastuzumab duocarmazine versus physician's choice.
ARM 1

(Vic-)trastuzumab duocarmazine

  • Intervention: (vic-)trastuzumab duocarmazine.
  • Serious adverse events were reported for 53 of 288 participants at risk.
ARM 2

Physician's choice

  • Comparator: physician's choice.
  • Serious adverse events were reported for 12 of 137 participants at risk.

4. Randomization and Stratified Analysis

The primary progression-free survival analysis used the full-analysis set (FAS), comprising all randomized patients analyzed according to the treatment group and strata to which they were assigned. This is an important distinction: the efficacy comparison remains anchored to randomized treatment assignment rather than being restricted to patients who completed treatment.

The primary analysis incorporated the stratification factors assigned at randomization:

Stratification factorCategories used in the analysis
World regionEurope, Singapore, and North America
Number of prior treatment lines for locally advanced or metastatic breast cancer1 to 2; >2, excluding hormone therapy
Prior treatment with pertuzumabYes; no

Stratification is useful when important prognostic factors are known before randomization. Instead of treating all participants as if they came from one homogeneous risk population, the analysis compares treatment groups while accounting for the prespecified strata.

Statistical distinction: stratification does not change the randomized treatment comparison into a matched observational analysis. Treatment assignment remains randomized; the stratified analysis incorporates the randomization strata into the inferential procedure.

5. Endpoints

EndpointRegistry definition / time frameAnalysis method
Progression Free Survival Time from the date of randomization to the date of first documented disease progression by central assessment according to Response Evaluation Criteria In Solid Tumors (RECIST) v1.1 or death due to any cause, whichever occurred earlier. Time frame: baseline until primary analysis data cut-off date of 31March2021. Stratified log-rank test; stratified Cox regression for HR
Overall Survival Time frame: baseline until final Overall Survival analysis data cut-off date of 30June2022. Stratified log-rank test; stratified Cox regression for HR
Investigator Assessed Progression Free Survival Time frame: baseline until primary analysis data cut-off date of 31March2021. Stratified log-rank test; stratified Cox regression for HR
Objective Response Rate Percentage of patients. Time frame: baseline until primary analysis data cut-off date of 31March2021. Cochran-Mantel-Haenszel test
Patient Reported Outcomes for Health Related Quality of Life Scores on a scale. Time frame: baseline until primary analysis data cut-off date of 31March2021. MMRM

The registry therefore contains one primary endpoint—progression-free survival—and four posted secondary outcome measures. The statistical methods correspond to the measurement structure of those endpoints: time-to-event methods for survival outcomes, a stratified categorical-data method for objective response rate, and a repeated-measures model for longitudinal quality-of-life scores.

6. Statistical Methodology

Primary progression-free survival analysis

The primary endpoint was analyzed in the full-analysis set, comprising all randomized patients. The treatment groups were compared using a two-sided stratified log-rank test. A stratified Cox regression analysis estimated the hazard ratio and its 95% confidence interval.

Primary analysis framework
Treatment comparison → stratified log-rank test → hazard ratio from stratified Cox regression

The log-rank test addresses evidence of a difference between time-to-event distributions, while the Cox model provides an estimated relative hazard together with a confidence interval.

Secondary overall survival analysis

Overall survival was also analyzed using a stratified log-rank test and stratified Cox regression. The final overall survival analysis used a data cutoff of 30June2022. The same three randomization stratification factors were incorporated into the reported Cox analysis.

Investigator-assessed progression-free survival

Investigator-assessed progression-free survival was analyzed as a secondary time-to-event endpoint. The registry reports a two-sided stratified log-rank comparison and a stratified Cox regression estimate of the hazard ratio.

Objective response rate

Objective response rate is a percentage-of-patients endpoint rather than a time-to-event endpoint. The registry reports use of the Cochran-Mantel-Haenszel test, with strata based on the baseline stratification factors, to compare the treatment groups at a two-sided 5% significance level.

Patient-reported quality of life

The change from baseline in the global health status/quality-of-life scale transformed score was analyzed using a mixed model for repeated measures (MMRM). This is a longitudinal method designed to use repeated observations rather than reducing each participant's follow-up to a single post-baseline value.

7. Primary Result: Progression-Free Survival

The registry reports a formal primary analysis of progression-free survival using the full-analysis set. The treatment groups were compared with a stratified log-rank test, and the hazard ratio was estimated using a stratified Cox regression model.

Hazard ratio for progression or death

0.6401

95% CI: 0.4885–0.8389   ·   P = 0.002

Superiority hypothesis; two-sided 95% confidence interval.

Primary endpointComparisonEstimate95% CIP-value
Progression Free Survival (Vic-)Trastuzumab Duocarmazine vs Physician's Choice HR 0.6401 0.4885–0.8389 =0.002
Clinical Biostats interpretation

The estimated hazard ratio of 0.6401 means that, under the stratified Cox model, the estimated instantaneous hazard of the PFS event was about 64% of the corresponding hazard in the physician's-choice group. Expressed as a relative complement, this corresponds to an estimated 35.99% lower hazard under the fitted model.

The hazard ratio does not mean that 35.99% of participants avoided progression, that 35.99% of participants were cured, or that every participant experienced the same proportional reduction in risk. It is a model-based relative measure of the event hazard over follow-up.

The 95% confidence interval of 0.4885 to 0.8389 describes statistical uncertainty around the estimated hazard ratio. It does not describe the range of treatment effects experienced by individual participants.

The P = 0.002 value addresses evidence against the null hypothesis under the specified testing framework. It is not a measure of the magnitude of the treatment effect and does not tell us how clinically important an effect is.

Because the analysis is based on a Cox proportional-hazards model, interpretation of a single hazard ratio is most straightforward when the proportional-hazards assumption is reasonably appropriate. The ClinicalTrials.gov record does not provide a diagnostic assessment of that assumption, so the HR should not be interpreted as a literal constant risk ratio at every time point.

What the primary PFS result establishes statistically

The reported two-sided P-value of 0.002 provides evidence of a difference between the randomized treatment groups under the registered superiority analysis. The effect estimate is below 1, and the reported 95% confidence interval lies below 1. Together, these quantities provide both an estimate of relative treatment effect and a measure of its statistical precision.

Importantly, the PFS definition is clinically specific: the event was the first documented disease progression by central assessment according to RECIST v1.1 or death from any cause, whichever occurred earlier. Thus, the endpoint combines radiographic disease progression and death into one time-to-event outcome.

8. Secondary Result: Overall Survival

Overall survival was a secondary time-to-event endpoint with a final analysis data cutoff of 30June2022. The registry reports a full-analysis-set analysis using a stratified log-rank test and stratified Cox regression.

Hazard ratio for overall survival

0.868

95% CI: 0.676–1.1145   ·   P = 0.236

Superiority hypothesis; two-sided 95% confidence interval.

Secondary endpointComparisonEstimate95% CIP-value
Overall Survival (Vic-)Trastuzumab Duocarmazine vs Physician's Choice HR 0.868 0.676–1.1145 =0.236
Clinical Biostats interpretation

The estimated OS hazard ratio of 0.868 corresponds to an estimated instantaneous hazard of death approximately 86.8% of that in the physician's-choice group under the stratified Cox model. The corresponding relative complement is approximately a 13.2% lower estimated hazard.

This does not mean that mortality was reduced by exactly 13.2% for every patient, nor does it provide an absolute difference in survival probability. The HR is a relative time-to-event measure.

The 95% confidence interval, 0.676 to 1.1145, spans 1.0. Thus the interval is compatible with a range of relative hazard differences that includes no difference under the model.

The P = 0.236 value is not evidence about the size or clinical importance of the estimated HR. It describes the statistical evidence against the null hypothesis under the reported superiority testing framework. A nonsignificant P-value should not be translated into proof that the treatments are equivalent.

The OS analysis also has a different data cutoff from the primary PFS analysis. These results should therefore not be treated as though they arose from one common time point.

9. Secondary Result: Investigator-Assessed Progression-Free Survival

Investigator-assessed progression-free survival was analyzed through the same primary-analysis data cutoff of 31March2021. The registry reports a two-sided stratified log-rank comparison and a stratified Cox regression estimate.

Hazard ratio for investigator-assessed PFS

0.5995

95% CI: 0.4666–0.7703   ·   P < 0.001

Superiority hypothesis; two-sided comparison.

Secondary endpointEstimate95% CIP-value
Investigator Assessed Progression Free Survival HR 0.5995 0.4666–0.7703 <0.001
Clinical Biostats interpretation

An HR of 0.5995 means that the estimated instantaneous hazard of the investigator-assessed PFS event was approximately 60% of the corresponding hazard in the physician's-choice group under the stratified Cox model. The relative complement is approximately a 40.05% lower estimated hazard.

The estimate should not be interpreted as a 40.05% absolute improvement in progression-free survival, and it does not mean that 40.05% of patients necessarily benefited.

The 95% CI of 0.4666 to 0.7703 provides a range of values expressing uncertainty around the estimated relative hazard. Its location below 1 is consistent with the direction of the reported treatment effect.

The P < 0.001 result indicates strong statistical evidence against the null hypothesis under the reported test. It does not quantify the clinical magnitude of the effect.

Because this endpoint is investigator assessed rather than the centrally assessed primary PFS endpoint, it is useful as a separate outcome rather than being silently substituted for the primary endpoint.

10. Secondary Result: Objective Response Rate

Objective response rate was analyzed as a percentage-of-patients endpoint. The registry reports use of a Cochran-Mantel-Haenszel test, with strata based on the baseline stratification factors, to compare the treatment groups at a two-sided 5% significance level.

Objective response rate comparison

P = 0.732

Cochran-Mantel-Haenszel test; two-sided 5% significance level.

Secondary endpointMethodP-valueAnalysis framework
Objective Response Rate Cochran-Mantel-Haenszel test =0.732 Strata based on baseline stratification factors
Clinical Biostats interpretation

The registry's reported P = 0.732 does not provide evidence of a statistically detectable difference in objective response rate under the specified Cochran-Mantel-Haenszel comparison.

The ClinicalTrials.gov record does not report the response percentages, response counts, or a confidence interval for the response-rate comparison. Therefore, the P-value should not be converted into an estimated treatment effect or interpreted as evidence that the response rates are identical.

The choice of a stratified Cochran-Mantel-Haenszel test is appropriate to the endpoint structure: response is categorical, while the baseline stratification factors can be incorporated into the treatment comparison rather than ignored.

11. Secondary Result: Patient-Reported Health-Related Quality of Life

The patient-reported outcome analysis evaluated the change from baseline in the global health status/quality-of-life scale transformed score using an MMRM approach.

Quality-of-life analysis

P = 0.473

MMRM analysis of change from baseline in the global health status/QoL scale transformed score.

Clinical Biostats interpretation

The reported P = 0.473 does not provide statistical evidence of a treatment-group difference under the reported MMRM analysis.

The ClinicalTrials.gov record does not provide an estimated between-group change, confidence interval, or detailed repeated-measures parameter estimates. The P-value therefore should not be used to infer the size of any difference or to claim that the treatment groups had identical quality-of-life trajectories.

MMRM is particularly useful for longitudinal outcomes because repeated observations from the same participant are correlated. The model is designed to account for this within-participant structure rather than treating every observation as independent.

12. Safety Results

The ClinicalTrials.gov record reports serious adverse events by treatment arm. These are presented as affected participants over the corresponding number at risk.

Safety measure(Vic-)Trastuzumab DuocarmazinePhysician's Choice
Serious adverse events 53/288 12/137
Serious adverse events · reported affected / at risk
(Vic-)Trastuzumab Duocarmazine
53/288
Physician's Choice
12/137

The ClinicalTrials.gov record does not provide a complete adverse-event table, severity breakdown, treatment-relatedness classification, or discontinuation analysis. Those quantities are therefore not inferred from the serious-adverse-event counts.

Safety interpretation: serious adverse events and efficacy endpoints answer different questions. A serious-adverse-event count should not be combined mathematically with a hazard ratio or P-value to produce a single overall treatment judgment. The safety and efficacy evidence should be considered as separate statistical domains.

13. Statistical Methods Explained

Why was a stratified log-rank test used for progression-free survival?

Progression-free survival is a time-to-event endpoint, so participants may have different follow-up times and may be censored before experiencing progression or death. The log-rank test compares the event-time distributions between randomized treatment groups while accounting for the timing of events. In this trial, the comparison was stratified according to the factors used at randomization.

What does an HR of 0.6401 mean?

An HR of 0.6401 means that the estimated instantaneous event hazard under the fitted Cox model was 0.6401 times that in the comparison group. The complementary percentage, approximately 35.99%, describes the relative reduction implied by the point estimate. It is not an absolute reduction in the probability of progression or death.

Why was a Cox regression model used?

The Cox model provides a way to estimate a relative hazard while allowing the baseline hazard to remain unspecified. In TULIP, the registry reports a stratified Cox regression analysis, allowing the primary PFS hazard ratio to account for the prespecified randomization strata.

What does the 95% confidence interval tell us?

A 95% confidence interval quantifies statistical uncertainty around an estimated parameter under the specified inferential framework. For the primary PFS HR, the interval was 0.4885–0.8389. The interval should not be interpreted as saying that 95% of individual patients experience an HR somewhere inside those limits.

Why was the Cochran-Mantel-Haenszel test used for objective response rate?

Objective response rate is a categorical outcome. The Cochran-Mantel-Haenszel approach allows treatment groups to be compared while accounting for prespecified strata. This can be preferable to an unstratified comparison when the trial's randomization and analysis plan explicitly incorporate baseline stratification factors.

Why was MMRM used for quality-of-life measurements?

Quality-of-life scores can be measured repeatedly over time. Observations from the same participant are correlated, so ordinary methods that assume complete independence of all observations are not appropriate. MMRM provides a framework for modeling repeated measurements and estimating treatment-group differences in longitudinal change.

Why should the P-value not be treated as the effect size?

A P-value describes the strength of statistical evidence against a specified null hypothesis under a particular model and sampling framework. It does not tell us how large the treatment effect is. The estimated HR and its confidence interval provide effect-size and precision information, while the P-value provides a different piece of statistical evidence.

14. Confidence Intervals and Precision

The TULIP results illustrate why an effect estimate should be read together with its confidence interval rather than in isolation.

EndpointPoint estimate95% confidence intervalWhat the interval adds
Primary PFS HR 0.6401 0.4885–0.8389 Shows the statistical uncertainty around the estimated relative hazard.
Overall Survival HR 0.868 0.676–1.1145 Shows greater uncertainty around the direction and magnitude of the secondary OS estimate.
Investigator-assessed PFS HR 0.5995 0.4666–0.7703 Shows uncertainty around the investigator-assessed relative hazard.

Comparing the width and location of these intervals is more informative than simply sorting P-values. The primary PFS estimate has a confidence interval entirely below 1, whereas the OS interval extends above 1. These are descriptive properties of the reported estimates and uncertainty intervals, not substitute measures of clinical importance.

15. Time-to-Event Endpoints and Censoring

Both progression-free survival and overall survival are time-to-event outcomes. The key statistical feature is that not every participant necessarily experiences the event during the observation period. A participant can therefore contribute follow-up information without having an observed event.

Conceptual survival function
S(t) = P(T > t)

The survival function describes the probability that the event time exceeds a specified time t. Kaplan-Meier estimation is commonly used to estimate this function when observations can be right censored.

For the registered primary endpoint, the event definition is itself composite: either the first documented disease progression by central RECIST v1.1 assessment or death from any cause, whichever occurs earlier. This means that death can produce a PFS event even when radiographic progression has not previously been documented.

Interpretation caution: the ClinicalTrials.gov record reports the statistical method and hazard ratio but do not provide the underlying Kaplan-Meier event counts, censoring table, or reconstructed survival curve. No survival probabilities or median PFS values are therefore inferred here.

16. Stratification and the Cochran-Mantel-Haenszel Framework

Three baseline factors were incorporated into the primary stratified analysis: world region, number of prior treatment lines, and prior pertuzumab treatment. These factors were assigned at randomization and then used in the stratified Cox analysis.

The same baseline stratification factors were also used for the Cochran-Mantel-Haenszel comparison of objective response rate. This creates a coherent statistical structure across the trial: the factors used to organize randomization are also represented in the corresponding stratified analyses.

Why stratify?

Stratification can account for important baseline factors that may influence outcome and can improve alignment between the randomization scheme and the inferential analysis.

What stratification does not do

It does not turn the study into an observational matched comparison. The fundamental treatment contrast remains based on randomized assignment.

17. Primary vs Secondary Evidence

The trial's statistical evidence should be organized according to endpoint role rather than treating every reported P-value as equally confirmatory.

EndpointRoleMethodReported result
Progression Free Survival Primary Stratified log-rank; stratified Cox regression HR 0.6401; 95% CI 0.4885–0.8389; P = 0.002
Overall Survival Secondary Stratified log-rank; stratified Cox regression HR 0.868; 95% CI 0.676–1.1145; P = 0.236
Investigator Assessed Progression Free Survival Secondary Stratified log-rank; stratified Cox regression HR 0.5995; 95% CI 0.4666–0.7703; P < 0.001
Objective Response Rate Secondary Cochran-Mantel-Haenszel test P = 0.732
Patient Reported Outcomes for Health Related Quality of Life Secondary MMRM P = 0.473

This hierarchy matters. The primary PFS analysis is the central confirmatory result identified in the ClinicalTrials.gov record. Secondary endpoints provide additional evidence about survival, investigator-assessed disease control, response, and patient-reported outcomes, but they should retain their registered endpoint role when interpreting the overall evidence.

18. How the Results Fit Together

The reported results are not all pointing to the same statistical quantity. PFS and investigator-assessed PFS are time-to-event endpoints summarized with hazard ratios. Overall survival is another time-to-event endpoint but uses a later data cutoff. Objective response rate is a categorical percentage outcome, while health-related quality of life is longitudinal and analyzed with MMRM.

PFS

Primary endpoint with HR 0.6401, 95% CI 0.4885–0.8389, and P = 0.002.

Overall survival

Secondary endpoint with HR 0.868, 95% CI 0.676–1.1145, and P = 0.236 at the final OS cutoff.

Investigator-assessed PFS

Secondary time-to-event endpoint with HR 0.5995, 95% CI 0.4666–0.7703, and P < 0.001.

Response and quality of life

ORR was analyzed with the Cochran-Mantel-Haenszel test, while quality-of-life change was analyzed with MMRM.

A statistically coherent reading therefore avoids collapsing these outcomes into a single number. The primary endpoint supplies the main superiority test; the secondary analyses describe additional dimensions of treatment effect and patient experience.

19. Limitations and Interpretation Issues

20. Why This Trial Matters Statistically

TULIP is a useful teaching case because the registry data bring several core clinical-trial methods together in one randomized phase 3 analysis: a time-to-event primary endpoint, stratified randomization factors, log-rank testing, Cox regression, a categorical response analysis, and a repeated-measures quality-of-life analysis.

ConceptHow it appears in TULIP
RandomizationParticipants were randomized to two parallel treatment arms.
Full-analysis setThe primary PFS analysis included all randomized patients according to assigned treatment group and strata.
Time-to-event endpointPFS was defined from randomization to first documented progression or death, whichever occurred earlier.
Stratified log-rank testUsed for the primary PFS comparison and the reported secondary survival analyses.
Hazard ratioUsed to quantify relative treatment effects for PFS, OS, and investigator-assessed PFS.
Confidence interval95% two-sided CIs were reported for the hazard-ratio analyses.
Cox regressionStratified Cox regression estimated the hazard ratios for time-to-event endpoints.
Cochran-Mantel-Haenszel testUsed for the stratified objective response rate comparison.
MMRMUsed to analyze change from baseline in the global health status/QoL scale transformed score.
Multiple endpoint typesThe trial combines survival, categorical response, and longitudinal patient-reported outcomes, each requiring a method matched to its data structure.

21. Statistical Methods Explained: A Deeper Walkthrough

Why does the primary PFS definition include both progression and death?

The registered PFS definition treats either first documented disease progression or death as the event, whichever occurs earlier. This creates a composite time-to-event endpoint that captures both radiographic disease worsening and death without requiring a participant to have a documented progression before death.

Why use a full-analysis set?

Using the full-analysis set keeps the primary efficacy comparison tied to randomized assignment. This reduces the risk that post-randomization treatment adherence or discontinuation changes the population being compared and thereby weakens the protection provided by randomization.

Why is the PFS HR different from the investigator-assessed PFS HR?

The registry identifies them as distinct endpoints: the primary PFS endpoint uses progression by central assessment according to RECIST v1.1, whereas the secondary endpoint is investigator assessed. Different assessment mechanisms can produce different event classifications and therefore different time-to-event estimates.

Why is an HR not the same as a median survival difference?

A hazard ratio compares event hazards under a statistical model over follow-up. A median is a time point at which an estimated survival function reaches 0.5. These are different summaries and cannot be converted into one another without additional information about the survival distributions.

Why can a confidence interval be more informative than a P-value alone?

The P-value addresses statistical evidence against a null hypothesis, while the confidence interval provides information about the range of effect estimates compatible with the statistical model and data. For example, the primary PFS HR of 0.6401 is accompanied by a 95% CI of 0.4885–0.8389, allowing the reader to assess both direction and precision.

What does the OS result mean when its confidence interval crosses 1?

The OS estimate was HR 0.868 with a 95% CI of 0.676–1.1145 and P = 0.236. The interval includes 1.0, so the reported interval includes a no-difference value under the hazard-ratio scale. This should not be converted into a claim that the treatments are equivalent; it indicates that the registry-reported analysis did not establish a statistically detectable difference under its reported superiority framework.

22. Related Tutorials

Learn more about the methods used in this trial:

23. Related Calculators

24. Sources

Continue with the statistical methods behind TULIP

Explore tutorials on survival analysis, stratified testing, confidence intervals, repeated-measures models, and related clinical-trial methods.

25. Record Summary

TULIP provides a compact example of how a randomized phase 3 trial can use different statistical methods for different clinical questions. The primary endpoint, progression-free survival, was analyzed in the full-analysis set using a stratified log-rank test and stratified Cox regression, producing an HR of 0.6401 with a two-sided 95% CI of 0.4885–0.8389 and P = 0.002. Secondary analyses included overall survival, investigator-assessed progression-free survival, objective response rate, and patient-reported health-related quality of life, with methods matched to their respective data structures.

The overall survival analysis reported an HR of 0.868 with a 95% CI of 0.676–1.1145 and P = 0.236. Investigator-assessed PFS reported an HR of 0.5995 with a 95% CI of 0.4666–0.7703 and P < 0.001. Objective response rate was compared with a Cochran-Mantel-Haenszel test with P = 0.732, while the MMRM analysis of change from baseline in the global health status/QoL scale transformed score reported P = 0.473.

Clinical Biostats methodology: The statistical interpretation of a clinical trial should distinguish the endpoint definition, analysis population, effect measure, confidence interval, P-value, and data cutoff. Keeping those components separate makes it possible to understand what each result actually establishes without extending the registry data beyond what was reported.