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
DESTINY-Breast09 is a randomized, parallel-group, phase 3 trial evaluating trastuzumab deruxtecan with or without pertuzumab against a taxane, trastuzumab and pertuzumab regimen in patients with HER2-positive metastatic breast cancer.
| Feature | DESTINY-Breast09 |
|---|---|
| Trial | DESTINY-Breast09 |
| ClinicalTrials.gov identifier | NCT04784715 |
| Phase | Phase 3 |
| Status | Active, not recruiting |
| Enrollment | 1157 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Number of arms | 3 |
| Lead sponsor | AstraZeneca |
| Sponsor type | Industry |
| Trial start | 2021-04-26 |
| Primary completion date | 2026-11-11 |
2. Clinical Question
The clinical question is whether treatment strategies based on trastuzumab deruxtecan, either with or without pertuzumab, should be compared with taxane, trastuzumab and pertuzumab in patients with HER2-positive metastatic breast cancer, with progression-free survival serving as the registered primary endpoint.
Population
Patients with HER2-positive metastatic breast cancer.
Intervention strategies
Trastuzumab deruxtecan with pertuzumab, or trastuzumab deruxtecan without pertuzumab.
Comparator
Taxane, trastuzumab and pertuzumab.
Primary question
How does progression-free survival differ among the randomized treatment strategies?
3. Trial Design
Trastuzumab deruxtecan + pertuzumab
- Trastuzumab deruxtecan
- Pertuzumab
Trastuzumab deruxtecan
- Trastuzumab deruxtecan
Taxane + trastuzumab + pertuzumab
- Taxane
- Trastuzumab
- Pertuzumab
The registry identifies five intervention types in the study: trastuzumab deruxtecan, placebo, taxane, pertuzumab and trastuzumab. The registered study title describes the principal treatment comparison as trastuzumab deruxtecan with or without pertuzumab versus taxane, trastuzumab and pertuzumab.
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Progression Free Survival (PFS) by Blinded Independent Central Review (BICR) assessment | Defined as time from date of randomisation until the date of objective radiological disease progression according to Blinded Independent Central Review (BICR) using RECIST 1.1 or death by any cause. | Until progression or death, assessed up to approximately 60 months |
The registered primary endpoint is therefore a time-to-event endpoint. The event is the first of two possible outcomes: objective radiological disease progression according to BICR using RECIST 1.1, or death from any cause.
Why the endpoint combines progression and death
A progression-free survival endpoint treats either documented disease progression or death as an event. This avoids treating death as if it were simply an administrative observation after which progression could no longer be assessed. Statistically, both events terminate the period of being progression-free.
Why BICR matters
Blinded Independent Central Review provides a centralized radiologic assessment intended to reduce the influence of treatment assignment on the determination of objective disease progression. In an unmasked trial, this distinction is especially important because local clinical assessment and treatment knowledge can otherwise interact with progression assessment.
Why RECIST 1.1 matters
RECIST 1.1 provides a structured framework for determining objective radiological progression. Using a defined response-assessment framework makes the event definition more reproducible than an unrestricted clinical judgment about whether a patient's cancer has worsened.
5. Statistical Methodology
Because the primary endpoint is PFS, the natural statistical framework is survival analysis. The key feature is that not every participant necessarily experiences progression or death during the observation period. Participants without an event at their last evaluable assessment contribute follow-up information up to that point and are then censored according to the prespecified analysis rules.
Kaplan-Meier estimation
A Kaplan-Meier estimator is commonly used to describe the distribution of PFS over time. It estimates the probability of remaining event-free at successive event times while accommodating right-censored observations.
Here, di is the number of events at time ti, while ni is the number of participants at risk immediately before that event time.
Hazard ratios
For a randomized multi-arm PFS comparison, a Cox proportional-hazards model is a standard method for estimating a hazard ratio between treatment groups. A hazard ratio below 1 indicates a lower estimated instantaneous event rate in the numerator treatment group relative to the reference group, under the model.
A hazard ratio is a relative time-to-event measure. It is not a probability of progression, a median PFS, or a percentage of patients who benefit.
Log-rank comparison
A log-rank test is commonly used to compare survival distributions between randomized groups. For a three-arm study, the analysis may involve an overall comparison and prespecified pairwise comparisons, depending on the statistical analysis plan.
Confidence intervals
A confidence interval around a treatment-effect estimate describes statistical uncertainty around that estimate under the specified model and sampling framework. It should be interpreted together with the effect estimate rather than as a substitute for it.
Censoring
Censoring is fundamental to PFS analysis. A participant who has neither documented progression nor death by the relevant observation point may contribute information up to that point without being counted as an event. The validity of the resulting survival estimate depends on the prespecified censoring rules and the assumptions underlying those rules.
6. Planned Analysis
The registry says that PFS will be assessed by Blinded Independent Central Review using RECIST 1.1, from randomisation until objective radiological disease progression or death by any cause, with assessment continuing up to approximately 60 months.
For an endpoint defined this way, a standard analysis would describe PFS using Kaplan-Meier methods and compare randomized treatment groups using a time-to-event model such as a Cox proportional-hazards model, accompanied by an appropriate confidence interval for the treatment effect. A log-rank test is also commonly used for comparing survival distributions.
Because this is a three-arm randomized study, the prespecified statistical analysis plan determines how the overall comparison and individual treatment comparisons are handled. In particular, multiple treatment comparisons can create a multiplicity problem if each is tested independently at the same nominal significance level.
The registry's primary endpoint time frame extends until progression or death and specifies assessment up to approximately 60 months. That means the eventual PFS analysis is intended to incorporate follow-up over a substantial period rather than relying only on an early post-randomization snapshot.
What will be measured
Time from randomisation to objective radiological disease progression according to BICR using RECIST 1.1 or death from any cause.
How it is typically summarized
Kaplan-Meier estimates describe PFS over time, while a time-to-event model can quantify the relative treatment effect between randomized groups.
Why censoring matters
Participants without a qualifying event contribute follow-up information until their applicable censoring time.
Why the three arms matter
The comparison structure must distinguish the two trastuzumab deruxtecan strategies from the taxane, trastuzumab and pertuzumab comparator while controlling the overall type I error as specified by the study plan.
7. Statistical Methods Explained
Why is PFS a time-to-event endpoint?
PFS is not simply a yes-or-no outcome because the timing of progression or death matters. Two participants can both experience progression, yet the participant whose event occurs later contributes a longer period without the event. Survival methods preserve that timing information.
Why use Kaplan-Meier estimation?
Clinical trials often end observation before every participant has progressed or died. Kaplan-Meier estimation allows participants who remain event-free at their last evaluable observation to contribute information without incorrectly treating them as if their event occurred at that time.
What does a PFS hazard ratio mean?
A PFS hazard ratio compares the estimated instantaneous rate of progression or death between two treatment groups. For example, a hazard ratio below 1 would indicate a lower estimated event rate in the treatment group relative to the reference group under the fitted model. It does not mean that the same percentage of patients will avoid progression.
Why use Blinded Independent Central Review?
Progression can involve judgment about radiologic findings. Central review provides an independent assessment using the prespecified RECIST 1.1 framework. Blinding the central review helps reduce the possibility that knowledge of treatment assignment influences the classification of progression.
Why does death count as progression in PFS?
The registry definition treats death from any cause as an event. This prevents a participant who dies before documented radiological progression from being treated as if they remained progression-free indefinitely.
Why is multiplicity important in a three-arm trial?
With three randomized groups, several pairwise treatment comparisons are possible. Testing multiple hypotheses without appropriate control of the overall type I error can increase the chance of a false-positive finding. The prespecified analysis plan therefore matters not only for the individual statistical tests but also for how those tests are organized as a family.
Why does the unmasked design matter statistically?
The trial is registered as having no masking. Because treatment assignment is not masked, objective progression assessment becomes an important design consideration. BICR using RECIST 1.1 provides a more standardized endpoint assessment that can reduce the influence of treatment knowledge on radiologic progression classification.
8. Interpreting a Future PFS Result
When a formal PFS result is posted, the most useful statistical reading will require several pieces of information rather than a single number.
| Statistic | What it answers | What it does not answer |
|---|---|---|
| Median PFS | When the estimated survival curve reaches 50% event-free probability | Whether every patient experiences a similar duration |
| Hazard ratio | Relative event rate between treatment groups under the fitted model | Absolute probability of progression at a specific time |
| 95% confidence interval | Uncertainty around the estimated treatment effect | The range of outcomes for individual patients |
| p-value | Evidence against a specified null hypothesis under the statistical model and testing framework | The size or clinical importance of the treatment effect |
| Kaplan-Meier estimate | Estimated probability of remaining event-free at a specified time | A guarantee that an individual patient will remain progression-free until that time |
This distinction is particularly important for PFS. A hazard ratio summarizes relative event rates across follow-up, whereas a Kaplan-Meier estimate at a specified time provides an absolute time-specific perspective. Median PFS provides yet another summary of the same underlying time-to-event distribution.
A future treatment-effect estimate should be interpreted together with its confidence interval, the analysis population, the censoring rules, the number and timing of events, and the multiplicity framework. A numerical hazard ratio by itself is not a complete statistical description of a three-arm randomized trial.
9. Analysis Population and Randomization
Randomization is the central design feature that permits treatment groups to be compared as randomized rather than selected according to subsequent treatment exposure. The registry identifies DESTINY-Breast09 as a randomized, parallel-group study with 1157 enrolled participants and three arms.
For the eventual primary efficacy analysis, the analysis population should be read directly from the prespecified statistical analysis plan. In randomized trials, an intention-to-treat analysis is commonly used for the primary treatment comparison because it preserves the randomized assignment, while other analysis populations may be used for sensitivity or safety analyses.
10. Missing Data and Censoring in PFS
PFS analysis is different from an ordinary continuous or binary endpoint because incomplete follow-up can be handled through time-to-event methods. A participant who has not experienced progression or death by the end of available observation may be censored rather than classified as an event.
The distinction is important: censoring is not the same as missingness. Censoring represents incomplete observation of the event time, whereas missing assessments can affect whether and when progression can be established. The prespecified analysis rules determine how these situations are handled.
Event
Objective radiological disease progression by BICR using RECIST 1.1, or death from any cause.
Censored observation
A participant who has not experienced a qualifying event contributes observed follow-up up to the applicable censoring point.
The registry's endpoint definition does not provide detailed censoring conventions for every possible assessment scenario. Those rules belong in the protocol or statistical analysis plan when the detailed analysis is reviewed.
11. Multiplicity in a Three-Arm Trial
A three-arm randomized study creates a broader hypothesis-testing structure than a two-arm trial. If the three groups are labelled A, B and C, possible pairwise comparisons include A versus B, A versus C, and B versus C.
The study's prespecified statistical strategy determines which comparisons constitute confirmatory hypotheses and how the overall type I error is controlled.
For DESTINY-Breast09, the clinical question naturally involves comparing the two trastuzumab deruxtecan strategies with the taxane, trastuzumab and pertuzumab strategy. If both experimental strategies are formally tested against the same comparator, multiplicity becomes an important component of the interpretation.
The fact that a trial has multiple arms does not mean that every possible pairwise comparison must be treated as a primary hypothesis. The hierarchy and alpha allocation should come from the prespecified statistical analysis plan rather than being reconstructed from eventual numerical results.
12. Proportional-Hazards Assumption
If a Cox proportional-hazards model is used for PFS, one important consideration is the proportional-hazards assumption. This assumption concerns the relative hazard between treatment groups over time.
A hazard ratio is easiest to interpret as a single summary when the relative hazards are reasonably stable over the follow-up period. If the treatment effect changes materially over time, the single hazard ratio can become a less complete description of the treatment-effect pattern.
A future PFS hazard ratio should therefore be read alongside the Kaplan-Meier curves and clinically relevant time-specific PFS estimates. The hazard ratio summarizes the fitted model; it does not display the full shape of the survival distributions.
13. Interim Analysis and Follow-Up
The registry specifies that the primary PFS endpoint will be assessed until progression or death, with assessment continuing up to approximately 60 months. The registered primary completion date is 2026-11-11, while the study is currently classified as active, not recruiting.
Trial start
The registered study start date is 2021-04-26.
Active, not recruiting
The ClinicalTrials.gov record currently classifies the study as active, not recruiting.
Primary completion
The registry lists 2026-11-11 as the primary completion date.
Primary endpoint assessment window
PFS is registered for assessment until progression or death, up to approximately 60 months.
The available registry information does not post interim statistical results or identify a numerical interim PFS analysis. Consequently, the appropriate interpretation at this stage is based on the registered design and endpoint definition rather than an observed treatment effect.
14. Safety and Secondary Endpoints
The ClinicalTrials.gov information available for this analysis identifies the primary endpoint as PFS by BICR assessment. It does not provide posted statistical analyses or numerical safety results by treatment arm.
The absence of a posted numerical safety analysis means that serious adverse events, treatment-emergent adverse events, discontinuations due to adverse events, and other safety measures cannot be quantitatively compared here without additional registry results.
15. What the PFS Endpoint Does — and Does Not — Mean
PFS measures the time from randomisation until the first qualifying event: objective radiological disease progression according to BICR using RECIST 1.1, or death from any cause.
PFS is not identical to overall survival, tumor response rate, duration of response, or quality of life. A difference in PFS should not automatically be translated into an identical difference in any of those other outcomes.
Because the trial is unmasked, centralized blinded assessment of radiological progression provides an important methodological safeguard for the primary endpoint.
The registry specifies follow-up for progression or death up to approximately 60 months. Time-to-event results should therefore always be interpreted in relation to the follow-up period and the amount of information accumulated by the analysis cutoff.
16. Important Limitations and Interpretation Issues
- No posted numerical efficacy analysis: the ClinicalTrials.gov record available for this analysis does not provide a numerical PFS estimate, hazard ratio, confidence interval, p-value, or event count.
- Three-arm structure: multiple possible treatment comparisons require a clearly prespecified testing hierarchy and multiplicity strategy for confirmatory interpretation.
- Unmasked study: treatment assignment is not masked, making standardized independent central radiologic assessment particularly relevant to the primary endpoint.
- Time-to-event censoring: participants may contribute incomplete follow-up, so PFS estimates depend on appropriate censoring and event-classification rules.
- RECIST-based assessment: progression depends on objective radiological criteria under RECIST 1.1 rather than an unrestricted clinical assessment.
- Proportional-hazards assumption: if a Cox model is used, the adequacy of summarizing the treatment effect with a single hazard ratio should be considered in relation to the observed survival curves.
- Follow-up maturity: the registered primary completion date is 2026-11-11, and the endpoint is assessed up to approximately 60 months. The maturity of any particular analysis depends on its data cutoff and accumulated events.
- Registry versus statistical analysis plan: detailed rules for multiplicity, censoring, analysis populations, and model specifications should be taken from the final prespecified statistical analysis plan when available.
17. Why This Trial Matters Statistically
DESTINY-Breast09 is a useful teaching example because its design combines a randomized three-arm comparison with a clinically important time-to-event endpoint and centralized radiologic assessment.
| Concept | How it appears in DESTINY-Breast09 |
|---|---|
| Randomization | The study uses randomized allocation. |
| Three-arm design | Three parallel treatment strategies are included. |
| Time-to-event analysis | The primary endpoint is progression-free survival. |
| Blinded central review | PFS progression is assessed by BICR. |
| RECIST 1.1 | Objective radiological progression is determined using RECIST 1.1. |
| Censoring | PFS analysis must account for participants who do not experience progression or death during observed follow-up. |
| Hazard ratio | A standard time-to-event treatment-effect measure for randomized PFS comparisons. |
| Kaplan-Meier estimation | A standard method for describing PFS over time. |
| Multiplicity | Multiple treatment groups create potential multiple-comparison issues. |
| Unmasked design | The study is registered as having no masking, increasing the importance of standardized endpoint assessment. |
| Long follow-up | The primary endpoint is assessed until progression or death, up to approximately 60 months. |
18. Statistical Interpretation: A Framework for the Eventual Results
When the primary analysis is posted, a rigorous interpretation should begin with the exact analysis population and data cutoff. The next step is to examine the Kaplan-Meier estimate of PFS, the number and timing of events, and the treatment-effect estimate with its confidence interval.
For a three-arm trial, the interpretation should then identify exactly which comparison the reported effect represents. A hazard ratio comparing trastuzumab deruxtecan plus pertuzumab with the comparator answers a different question from a hazard ratio comparing trastuzumab deruxtecan alone with the comparator.
Relative effect
The hazard ratio summarizes the relative rate of progression or death between two specified treatment groups.
Absolute effect
Kaplan-Meier PFS estimates at specified time points describe the absolute probability of remaining progression-free at those times.
Precision
The confidence interval describes uncertainty around the estimated treatment effect and should be considered alongside its point estimate.
Statistical evidence
The p-value, if reported, must be interpreted according to the prespecified hypothesis, testing hierarchy, and multiplicity framework.
19. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary endpoint is a randomized comparison of time to objective radiological progression or death, assessed by BICR using RECIST 1.1. The eventual statistical interpretation depends on the observed event data, model, confidence interval, and prespecified testing framework.
Clinical interpretation
The clinical meaning of any future PFS difference should consider the magnitude and durability of the effect, the pattern of progression, the treatment strategies being compared, and the broader benefit-risk evidence when those results are available.
Keeping these two levels separate is important. Statistical evidence describes uncertainty and the strength of the observed treatment comparison; clinical interpretation asks what that evidence means in the context of the disease and treatment strategies.
20. Registry Timeline and Study Status
| Study feature | Registry information |
|---|---|
| Start date | 2021-04-26 |
| Status | Active, not recruiting |
| Primary completion date | 2026-11-11 |
| Enrollment | 1157 |
| Phase | 3 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
The status and completion information are important when interpreting the absence of numerical results. A trial can have a defined primary endpoint and substantial enrollment while the registry record still does not contain a posted statistical analysis.
21. Sources
- ClinicalTrials.gov: NCT04784715 — DESTINY-Breast09.
- PubMed: PubMed record associated with DESTINY-Breast09.
Continue through the Clinical Biostats clinical-trial library
Explore additional randomized clinical trials and statistical analyses across oncology and other therapeutic areas.
22. Record Summary
DESTINY-Breast09 is a randomized phase 3, three-arm, parallel-group trial enrolling 1157 participants with HER2-positive metastatic breast cancer. The study compares trastuzumab deruxtecan with or without pertuzumab against taxane, trastuzumab and pertuzumab, with no masking and a primary purpose of treatment.
The registered primary endpoint is progression-free survival by Blinded Independent Central Review, defined as the time from randomisation to objective radiological disease progression according to RECIST 1.1 or death from any cause, assessed until progression or death for up to approximately 60 months.
Statistically, this endpoint calls for time-to-event methods that account for censoring and the timing of events. Kaplan-Meier estimation can describe PFS over time, while a Cox model and log-rank framework are standard approaches for comparative analysis. Because the study has three randomized arms, the interpretation of any eventual pairwise treatment effects also depends on the prespecified multiplicity and hypothesis-testing strategy.
The ClinicalTrials.gov record currently contains no posted formal statistical analyses for the primary endpoint. The appropriate statistical focus is therefore the registered design, endpoint definition, analysis framework, and the principles that will govern interpretation when mature numerical results are posted.