This page separates registry-reported information from statistical interpretation. The ClinicalTrials.gov record identifies the primary endpoint and report that results are posted, but do not provide a formal statistical-analysis record for that endpoint.
1. Trial at a Glance
DESTINY-Breast03 was a randomized, parallel-group phase 3 trial comparing trastuzumab deruxtecan (T-DXd) with ado-trastuzumab emtansine (T-DM1) in participants with HER2-positive, unresectable and/or metastatic breast cancer previously treated with trastuzumab and taxane. The ClinicalTrials.gov record reports an enrollment of 524 participants across two treatment arms.
| Feature | DESTINY-Breast03 |
|---|---|
| Trial name | DESTINY-Breast03 |
| NCT identifier | NCT03529110 |
| Phase | Phase 3 |
| Condition | Breast Cancer |
| Population | HER2-positive, unresectable and/or metastatic breast cancer previously treated with trastuzumab and taxane |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 524.0 |
| Interventions | Trastuzumab deruxtecan (T-DXd); ado-trastuzumab emtansine (T-DM1) |
| Results posted | Yes |
| Outcome measures posted | 5 |
| Trial status | Active, not recruiting |
| Lead sponsor | Daiichi Sankyo |
2. Clinical Question
The trial addresses a treatment-comparison question in participants with HER2-positive, unresectable and/or metastatic breast cancer who had previously been treated with trastuzumab and taxane. The randomized comparison is between trastuzumab deruxtecan and ado-trastuzumab emtansine.
Population
Participants with HER2-positive, unresectable and/or metastatic breast cancer previously treated with trastuzumab and taxane.
Intervention
Trastuzumab deruxtecan (T-DXd).
Comparator
Ado-trastuzumab emtansine (T-DM1).
Primary question
How does the time to progression or death compare between the randomized T-DXd and T-DM1 treatment groups?
3. Trial Design
Trastuzumab deruxtecan
- Trastuzumab deruxtecan (T-DXd)
- Drug intervention
Ado-trastuzumab emtansine
- Ado-trastuzumab emtansine (T-DM1)
- Drug intervention
Randomization is the key design feature for the primary comparison. Because treatment assignment is randomized, the intended statistical comparison is between outcomes according to the assigned treatment groups rather than a comparison constructed from observed treatment selection.
4. Endpoints
The ClinicalTrials.gov record identifies one registered primary endpoint. It is a time-to-event endpoint based on blinded independent central review (BICR).
| Endpoint | Registry definition | Time frame | Endpoint type |
|---|---|---|---|
| Progression-Free Survival (PFS) Based on Blinded Independent Central Review (BICR) in Participants With HER2-Positive, Unresectable and/or Metastatic Breast Cancer Previously Treated With Trastuzumab and Taxane | Progression-free survival (PFS) by BICR was defined as the time from the date of enrollment to the earlier of the dates of the first objective documentation of disease progression (as per RECIST v1.1) or death due to any cause. Progressive disease was defined as at least a 20% increase in the sum of diameters of target lesions. | Up to 33 months (data cut-off) | Time-to-event |
What the primary endpoint measures
The endpoint combines two possible events: objective disease progression or death from any cause. The analysis therefore does not require every participant to experience progression before contributing an event. Death is itself a qualifying event, occurring before or instead of documented progression.
The time origin is the date of enrollment. The event time is the earlier of objective progression or death. Participants who have not experienced either event by the relevant end of observation are ordinarily handled as censored observations in a time-to-event analysis.
Progression is based on objective documentation according to RECIST v1.1, with progressive disease defined in the ClinicalTrials.gov record as at least a 20% increase in the sum of diameters of target lesions.
5. Statistical Methodology
The ClinicalTrials.gov record classifies the primary endpoint as a time-to-event outcome and define PFS using a time origin, an event rule, and a data-cutoff period. They do not provide a formal the posted statistical analyses entry for the endpoint. Therefore, no trial-specific hazard ratio, confidence interval, p-value, median PFS, event count, or treatment-effect estimate is reproduced here.
Typical analysis for a randomized PFS endpoint
For a randomized trial with this endpoint structure, the standard analytical framework is survival analysis. A Kaplan-Meier estimator can describe the distribution of PFS over time, while a comparison between randomized groups can be made with a log-rank test. A Cox proportional-hazards model is commonly used to estimate a hazard ratio and its confidence interval when the model assumptions are appropriate.
The survival function gives the probability that the event time exceeds a specified time t. For PFS, the event is the first qualifying progression or death defined by the endpoint.
Why censoring is part of the analysis
Not every participant will necessarily have a documented progression or death by the end of the available observation period. A time-to-event analysis can use the participant's observed follow-up up to censoring rather than simply excluding that participant. The exact censoring rules, however, are not provided in the ClinicalTrials.gov record and therefore should not be inferred as trial-specific rules here.
Why BICR matters statistically
The primary endpoint is explicitly based on blinded independent central review. Central review can provide a standardized assessment of radiologic progression across treatment groups. Because the registry specifically identifies BICR as part of the endpoint definition, progression is not simply treated as an investigator's unreviewed assessment in the endpoint description in the ClinicalTrials.gov record.
6. Results
What is known from the registry data
| Results item | Supplied registry information |
|---|---|
| Results posted | Yes |
| Outcome measures posted | 5 |
| Primary endpoint results | Results are reported as posted, but no formal statistical analysis is reported in the ClinicalTrials.gov record |
| Primary endpoint | PFS based on BICR |
| Primary endpoint time frame | Up to 33 months (data cut-off) |
| Formal HR | Not present in the registry-reported statistical-analyses data |
| Formal 95% CI | Not present in the registry-reported statistical-analyses data |
| Formal p-value | Not present in the registry-reported statistical-analyses data |
This distinction is important. A statement that results are posted is not equivalent to having a complete statistical-analysis table available in the registry extract. A rigorous educational analysis should preserve that distinction rather than reconstructing a numerical treatment effect from external knowledge.
The primary endpoint is a time-to-event comparison of progression or death. Without a reported treatment-effect estimate and its uncertainty, the available data do not support a numerical statement about the magnitude of the difference between T-DXd and T-DM1.
The endpoint definition does tell us what a reported effect would represent: it would describe the relative timing of progression or death between randomized groups over the specified analysis period. It would not by itself represent the percentage of participants cured, the percentage experiencing tumor shrinkage, or an absolute difference in survival probability at a particular time.
Similarly, without a confidence interval, the precision of any numerical treatment effect cannot be assessed from the ClinicalTrials.gov record. Without a p-value, statistical evidence against a null hypothesis cannot be quantified from this extract.
7. Statistical Methods Explained
Why is PFS a time-to-event endpoint?
PFS incorporates both whether an event occurs and when it occurs. Two participants can both experience progression, but the participant whose progression occurs later has a different PFS time. A conventional binary endpoint would discard much of that timing information.
Why does the endpoint include death?
The registry definition specifies the earlier of objective disease progression or death due to any cause. Including death prevents a participant who dies before documented progression from being treated as though they remained progression-free simply because radiologic progression was not observed first.
Why use Kaplan-Meier estimation?
Kaplan-Meier estimation is designed for time-to-event data with censoring. It provides an estimate of the event-free probability over time while retaining information from participants whose follow-up ends before the event is observed.
What would a hazard ratio mean for this trial?
If a Cox model were used, the hazard ratio would compare the estimated instantaneous event rates between T-DXd and T-DM1 under the model. A hazard ratio below 1 would indicate a lower estimated hazard in the first group relative to the second. It would not mean that the same percentage reduction applies to every individual participant or to every time-specific absolute risk.
Why is the confidence interval important?
A confidence interval communicates uncertainty around an estimated treatment effect. A narrow interval indicates greater statistical precision than a wide interval, all else equal. The interval does not describe the range of effects experienced by individual patients.
Why does BICR matter for progression?
Progression is based on an assessment of objective disease status. The registry specifically defines this primary endpoint using blinded independent central review, which provides a standardized review framework for the radiologic component of progression assessment.
Why should results not be inferred from the trial's design?
Randomization establishes the framework for a causal treatment comparison, but it does not tell us what the observed treatment effect was. The magnitude, uncertainty, and statistical evidence for the comparison require the actual outcome data and the prespecified analysis. Those numerical analysis results are not contained in the registry-reported statistical-analysis field.
8. Analysis Population and Interpretation
The ClinicalTrials.gov record establishes randomized allocation and an enrollment of 524.0 participants, but they do not provide separate analysis-population counts or detailed definitions for efficacy and safety populations.
| Analysis component | What the ClinicalTrials.gov record establishes | What is not reported |
|---|---|---|
| Randomized population | Randomized phase 3 trial with enrollment of 524.0 | Detailed participant counts by treatment arm |
| Efficacy analysis | Primary PFS endpoint based on BICR | Formal analysis population and treatment-effect estimates |
| Safety analysis | Serious adverse events are reported by arm | Full safety-analysis population definition |
| Subgroups | Not specified in the ClinicalTrials.gov record | Subgroup estimates and interaction analyses |
For a randomized trial, analysis according to assigned treatment is commonly used for the primary efficacy comparison because it preserves the benefits of randomization. However, the ClinicalTrials.gov record does not explicitly identify the trial's exact efficacy analysis population, so that convention should not be presented as a trial-specific registry fact.
9. Safety
The ClinicalTrials.gov record includes serious adverse events by treatment arm. These counts are presented as affected participants divided by participants at risk.
| Treatment arm | Serious adverse events | Affected / at risk |
|---|---|---|
| Trastuzumab deruxtecan (T-DXd) | 49 | 49/257 |
| Ado-trastuzumab emtansine (T-DM1) | 47 | 47/261 |
The reported serious-adverse-event counts should be interpreted in the context of their corresponding at-risk denominators. The ClinicalTrials.gov record does not provide a formal statistical comparison, confidence interval, p-value, exposure-adjusted incidence rate, or detailed adverse-event breakdown.
The raw counts are close in absolute terms: 49 affected participants in the T-DXd group and 47 in the T-DM1 group. However, counts alone are not sufficient for a formal comparison because the denominators, exposure duration, follow-up, event definitions, and statistical analysis all matter.
The reported values are 49/257 and 47/261. They should not be converted into a new set of percentages or a risk ratio here because the task requires reported numbers to be preserved exactly and the ClinicalTrials.gov record does not report a formal comparative safety analysis.
10. Time Frame and Data Cut-Off
The primary endpoint has a registry time frame of up to 33 months (data cut-off). This is an analysis-period descriptor, not a statement that every participant was followed for 33 months or that the median follow-up was 33 months.
Trial start
The registry profile gives August 9, 2018 as the study start date.
Primary completion
The registry profile gives May 21, 2021 as the primary completion date.
Primary endpoint data cut-off
The registered primary PFS endpoint is defined over a period of up to 33 months at the data cut-off.
Active, not recruiting
the ClinicalTrials.gov record lists the study status as ACTIVE_NOT_RECRUITING.
11. What a Formal PFS Analysis Would Report
For this trial, the primary endpoint is sufficiently structured to identify the principal statistical outputs that would normally be useful in a complete time-to-event analysis, even though those outputs are not reported in the ClinicalTrials.gov record.
| Output | Purpose | Status in the ClinicalTrials.gov record |
|---|---|---|
| Kaplan-Meier PFS curve | Describes progression-free probability over time | Not reported |
| Median PFS | Time at which the estimated PFS probability reaches 50%, when estimable | Not reported |
| Hazard ratio | Relative time-to-event effect under a fitted survival model | Not reported |
| 95% confidence interval | Quantifies statistical uncertainty around the estimated effect | Not reported |
| p-value | Quantifies evidence against a specified null hypothesis under the chosen test | Not reported |
| Event counts | Shows the number of qualifying progression/death events contributing to analysis | Not reported |
This separation is important for reproducible statistical reporting. The existence of a posted outcome does not justify filling missing statistical outputs with values obtained from memory or from an unrelated source. The numerical result should be tied to the exact analysis population and data cutoff under which it was calculated.
12. Understanding the PFS Endpoint in Detail
Time origin
The registry definition starts the PFS clock at the date of enrollment.
First event
The endpoint uses the earlier of objective disease progression or death due to any cause.
Progression rule
Progressive disease is defined as at least a 20% increase in the sum of diameters of target lesions.
Assessment framework
The primary endpoint is based on blinded independent central review.
Why the event definition matters
Changing the event definition can change the statistical estimand. Here, progression and death are combined into a single endpoint, so an analysis of PFS is answering a question about the time until either of those outcomes occurs, not simply the time until radiologic progression.
Why RECIST v1.1 matters
The registry explicitly identifies RECIST v1.1 as the framework for objective disease progression. The registry-reported definition also gives the quantitative progression rule: at least a 20% increase in the sum of diameters of target lesions.
13. Randomization as the Core Statistical Design
Randomization is the principal statistical design concept identified in the trial's statistical concepts. Participants are assigned to treatment groups by a randomized process rather than by a choice based on clinical characteristics.
The purpose is to make the treatment groups comparable in expectation with respect to prognostic factors, allowing differences in outcomes to be attributed to treatment assignment under the assumptions of the randomized trial.
Randomization does not eliminate statistical uncertainty. Even a randomized sample can show chance differences between groups, which is why the treatment effect must be accompanied by appropriate uncertainty measures in a complete analysis.
14. Interpreting a Hypothetical Hazard Ratio
The registry data do not supply a hazard ratio for DESTINY-Breast03 in the provided statistical-analysis field. It is therefore inappropriate to insert a numerical HR into this page merely because a hazard ratio would ordinarily be expected for a PFS analysis.
If a fitted Cox model produced an HR below 1 for T-DXd versus T-DM1, the statistical interpretation would be that the estimated instantaneous rate of PFS events was lower in the T-DXd group under that model. If the HR were above 1, the estimated event rate would be higher. An HR of exactly 1 would correspond to equal estimated hazards under the model.
Those statements explain how the statistic would be interpreted; they do not imply that any particular numerical HR was observed in DESTINY-Breast03.
15. Confidence Intervals and P-Values
The ClinicalTrials.gov record contains neither a confidence interval nor a p-value for the primary endpoint. These omissions are analytically important because an effect estimate without a measure of uncertainty is incomplete, and a p-value without its effect estimate does not describe the magnitude of a treatment difference.
Confidence interval
A confidence interval describes statistical uncertainty around an estimated parameter under the specified inferential framework. It should be read together with the point estimate.
P-value
A p-value addresses evidence against a specified null hypothesis. It is not a measure of treatment magnitude, clinical importance, or the probability that a treatment works.
For this trial, the absence of those quantities in the ClinicalTrials.gov record means that no numerical inference about statistical significance or precision should be made from the registry profile alone.
16. Limitations
- Missing formal statistical-analysis results: no formal statistical analyses were posted, despite the ClinicalTrials.gov record stating that results are posted.
- No numerical efficacy estimates reported: the available data do not contain a hazard ratio, confidence interval, p-value, median PFS, event count, or time-specific PFS estimate.
- No arm-specific enrollment counts reported: the total enrollment is 524.0, but the registry-reported profile does not give the randomized sample size for each treatment arm.
- No detailed analysis populations reported: the ClinicalTrials.gov record do not identify separate efficacy, safety, per-protocol, or other analysis populations beyond the serious-adverse-event denominators.
- No subgroup results reported: the ClinicalTrials.gov record do not include subgroup-specific treatment effects or interaction tests.
- No multiplicity information reported: the data do not describe an alpha allocation, multiplicity adjustment, or hierarchical testing strategy.
- No interim-analysis information reported: the registry-reported profile does not describe interim looks, stopping boundaries, or alpha spending.
- No missing-data or imputation rules reported: the registry extract does not describe censoring rules, imputation procedures, or other missing-data handling details.
- Safety scope is limited: serious adverse events by arm are reported, but a complete safety profile is not included in the ClinicalTrials.gov record.
- Data-cutoff interpretation: the primary endpoint time frame is given as up to 33 months at the data cut-off, which should not be interpreted as a participant-level follow-up summary.
17. Why This Trial Matters Statistically
DESTINY-Breast03 is a useful teaching case because its registry structure illustrates several fundamental concepts in randomized clinical-trial statistics without requiring unsupported numerical reconstruction.
| Concept | How it appears in DESTINY-Breast03 |
|---|---|
| Randomization | The trial uses randomized allocation. |
| Parallel-group design | The trial has two parallel treatment arms. |
| Time-to-event analysis | The registered primary endpoint is PFS. |
| Composite event definition | The PFS event is the earlier of objective progression or death. |
| Independent review | The primary PFS endpoint is based on blinded independent central review. |
| RECIST v1.1 | Objective disease progression is assessed according to RECIST v1.1. |
| Censoring | Time-to-event methodology ordinarily accommodates participants without an observed event by the end of follow-up. |
| Hazard ratio | A Cox-model HR would be a standard relative treatment-effect measure for a PFS endpoint, although no trial-specific HR is reported here. |
| Confidence interval | A formal treatment-effect estimate would ordinarily be accompanied by an uncertainty interval, which is not reported here. |
| Safety comparison | Serious adverse events are reported as affected participants over participants at risk in each arm. |
18. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The trial provides a randomized comparison with a prespecified time-to-event primary endpoint. The endpoint is defined using BICR, RECIST v1.1, and the earlier occurrence of progression or death.
Clinical interpretation
The clinical question concerns outcomes for participants receiving T-DXd compared with T-DM1. The ClinicalTrials.gov record establishes the treatment comparison and endpoint but do not provide enough numerical efficacy information to quantify the difference.
Keeping these two levels separate is important. The statistical design determines how an observed difference should be estimated and tested. The clinical interpretation then considers what that estimated difference means in the context of the population, treatment alternatives, follow-up, and safety profile.
19. Safety Interpretation: Serious Adverse Events
The serious-adverse-event data provide two arm-specific affected/at-risk counts:
Serious adverse events
Trastuzumab deruxtecan (T-DXd) vs ado-trastuzumab emtansine (T-DM1)
These figures describe the number of participants affected and the corresponding number at risk in the ClinicalTrials.gov record. They are not, by themselves, a formal hypothesis test.
A safety comparison should distinguish counts from rates and from time-to-first-event or exposure-adjusted analyses. The ClinicalTrials.gov record does not specify which of those approaches was used for the serious-adverse-event comparison.
Consequently, the safest interpretation is descriptive: serious adverse events affected 49 of 257 participants in the T-DXd group and 47 of 261 participants in the T-DM1 group in the ClinicalTrials.gov record.
20. What Is Not Reported in the Supplied Trial Data
The reference architecture for a full clinical-trial analysis often includes additional sections such as median survival, subgroup forest plots, detailed baseline characteristics, interim-analysis boundaries, multiplicity procedures, crossover, or long-term follow-up. Those sections are not included here when the registry-reported DESTINY-Breast03 data do not support them.
| Potential analysis component | Supplied data status |
|---|---|
| Median PFS | Not reported |
| PFS hazard ratio | Not reported |
| PFS confidence interval | Not reported |
| PFS p-value | Not reported |
| Arm-specific randomized enrollment | Not reported |
| Subgroup treatment effects | Not reported |
| Forest plot estimates | Not reported |
| Baseline characteristic table | Not reported |
| Multiplicity strategy | Not reported |
| Interim-analysis strategy | Not reported |
| Missing-data/imputation strategy | Not reported |
| Crossover information | Not reported |
| Bayesian analysis | Not reported |
This is a deliberate evidence boundary. A complete statistical page is more useful when it clearly distinguishes what the registry actually reports from what a statistician would ordinarily expect to see in a full statistical analysis plan or publication.
21. Planned Analysis
Because the ClinicalTrials.gov record identifies a posted outcome but do not provide a formal statistical-analysis entry, the most defensible analysis description is methodological rather than numerical.
Primary PFS analysis
The registered endpoint would ordinarily be analyzed as a time-to-event outcome. The Kaplan-Meier method can estimate the PFS distribution, with participants who have not experienced progression or death by the end of observation contributing censored follow-up according to the prespecified censoring rules.
Treatment comparison
A randomized comparison of PFS would ordinarily use a survival-comparison procedure such as a log-rank test and could estimate a treatment hazard ratio with a Cox proportional-hazards model. The exact model specification, covariate adjustment, stratification factors, and testing procedure are not reported in the ClinicalTrials.gov record.
Interpretation
The primary estimand concerns the time from enrollment to the earlier of objective disease progression or death. Any numerical treatment effect should therefore be interpreted as a time-to-event treatment comparison rather than as a simple proportion of participants experiencing progression.
22. Related Tutorials
Learn more about the methods used in this trial:
23. Related Statistical Calculators
24. Sources
- ClinicalTrials.gov: NCT03529110 — DESTINY-Breast03.
- Linked publication, PMID 38979893: PubMed record.
- Linked publication, PMID 38825627: PubMed record.
- Linked publication, PMID 38442809: PubMed record.
- Linked publication, PMID 38059424: PubMed record.
- Linked publication, PMID 37207306: PubMed record.
Continue through the Clinical Biostats statistical tutorials
Connect randomized-trial design, time-to-event endpoints, survival models, and clinical-trial calculations to deeper statistical methods.
25. Record Summary
DESTINY-Breast03 is a randomized phase 3, parallel-group treatment trial with 524.0 enrolled participants and two interventions: trastuzumab deruxtecan (T-DXd) and ado-trastuzumab emtansine (T-DM1). Its registered primary endpoint is progression-free survival based on blinded independent central review, defined as the time from enrollment to the earlier of objective disease progression according to RECIST v1.1 or death due to any cause, with the registry specifying a time frame of up to 33 months at the data cut-off.
The principal statistical lesson is the distinction between an endpoint being defined and results being numerically analyzable. The registry profile states that results are posted and that five outcome measures are posted, but the registry-reported statistical-analysis section contains no formal analyses. Consequently, this page explains the appropriate time-to-event framework without introducing unsupported hazard ratios, confidence intervals, p-values, median PFS values, event counts, subgroup estimates, or other efficacy results.
The ClinicalTrials.gov record adds a descriptive comparison of serious adverse events: 49/257 participants in the T-DXd group and 47/261 participants in the T-DM1 group. These figures should remain descriptive because the ClinicalTrials.gov record do not include a formal statistical safety comparison or the additional exposure and follow-up information required for a more detailed analysis.