This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial facts on this page are limited to the ClinicalTrials.gov record. The ClinicalTrials.gov record contains posted outcome measures but no formal statistical analyses.
1. Trial at a Glance
DESTINY-Breast12 is a phase 3, single-group study evaluating trastuzumab deruxtecan in participants with or without brain metastasis who have previously treated advanced or metastatic HER2-positive breast cancer. The registry reports two primary endpoints: objective response rate in participants without brain metastasis at baseline and progression-free survival rate at 12 months in participants with brain metastasis at baseline.
| Feature | DESTINY-Breast12 |
|---|---|
| Phase | Phase 3 |
| Condition | Breast Cancer |
| Population | Participants with or without brain metastasis who have previously treated advanced or metastatic HER2 positive breast cancer |
| Intervention | Trastuzumab Deruxtecan (T-DXd) |
| Design | Single-group |
| Allocation | NA |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 506 |
| Lead sponsor | AstraZeneca |
| Results posted | Yes |
| ClinicalTrials.gov | NCT04739761 |
2. Clinical Question
The clinical question is centered on the activity of trastuzumab deruxtecan in participants with previously treated advanced or metastatic HER2-positive breast cancer, including participants with brain metastasis at baseline. Because the registry identifies the study as a single-group design with one intervention, the statistical question differs from that of a randomized comparative trial.
Population
Participants with or without brain metastasis who have previously treated advanced or metastatic HER2 positive breast cancer.
Intervention
Trastuzumab deruxtecan (T-DXd).
Comparator
No concurrent comparator arm is identified. The registered design is single-group with one arm.
Primary questions
What is the objective response rate among participants without brain metastasis at baseline, and what proportion of participants with brain metastasis at baseline remain alive and free of disease progression at 12 months?
3. Trial Design
Statistically, the single-group structure is central to interpreting the trial. There is no randomized control group against which an observed response proportion or progression-free survival curve can be directly contrasted. Consequently, measures such as a between-arm hazard ratio, treatment-effect difference, or comparative p-value are not naturally generated by the registered design itself.
4. Trial Timeline
Study begins
The registered trial start date is 2021-06-22.
Primary completion
The registered primary completion date is 2024-02-08.
Active, not recruiting
The ClinicalTrials.gov record lists the study status as ACTIVE_NOT_RECRUITING.
5. Endpoints
The registry identifies two primary endpoints, one binary and one time-to-event. Their definitions are important because they determine what statistical summaries are appropriate.
| Primary endpoint | Time frame | Definition | Endpoint type |
|---|---|---|---|
| Objective Response Rate (ORR) in Cohort 1 (Participants Without Brain Metastasis at Baseline) | From first dose (Day 1) to progression of disease (up to 2 years 7 months) | The ORR is the percentage of participants who have a confirmed complete response (CR) or confirmed partial response (PR), as determined by independent central review (ICR) per Response Evaluation Criteria in Solid Tumors (RECIST) 1.1. | Binary |
| Progression-free Survival (PFS) Rate at 12 Months in Cohort 2 (Participants With Brain Metastasis at Baseline) | At 12 months | The PFS rate is the percentage of participants alive and free of disease progression at 12 months, estimated by the Kaplan-Meier method. | Time-to-event |
Why the endpoint definitions matter
The first primary endpoint converts each participant's response assessment into a binary outcome: confirmed complete response or confirmed partial response versus not meeting either response category. The second primary endpoint is fundamentally different. Progression-free survival is a time-to-event variable, and the 12-month rate must account for participants who have not yet experienced progression or death when follow-up ends.
The use of independent central review for ORR is also statistically meaningful. It specifies that the response classification is based on an independent review process rather than simply an investigator's assessment, and the registry specifies RECIST 1.1 as the response framework.
6. Statistical Methodology
Objective response rate
For a binary endpoint such as ORR, the fundamental descriptive quantity is the proportion of evaluable participants achieving the specified response definition. Conceptually, if R participants have a confirmed complete or partial response among N participants in the relevant analysis population, the observed response proportion is:
The registry definition requires a confirmed complete response or confirmed partial response by independent central review according to RECIST 1.1.
For a formal statistical presentation, an ORR estimate would ordinarily be accompanied by a confidence interval for the underlying response probability. The ClinicalTrials.gov record does not provide the observed number of responders, the analyzed denominator for this endpoint, a confidence interval, or a formal statistical analysis entry. Those quantities therefore cannot be reconstructed from the ClinicalTrials.gov record without introducing data from outside the requested source.
Kaplan-Meier estimation of 12-month PFS
The second primary endpoint is explicitly defined as a Kaplan-Meier estimate. Kaplan-Meier methodology is designed for time-to-event data and accounts for right censoring. A participant who remains alive and free of disease progression at the end of observed follow-up contributes information up to the time at which follow-up ends.
Here, di represents the number of events at event time ti, while ni represents the number at risk immediately before that event time.
For this trial, the relevant summary is the estimated percentage alive and free of disease progression at 12 months in Cohort 2. The registry supplies the method but does not supply the resulting 12-month PFS estimate in the provided statistical-analysis data.
No randomized treatment-effect model
Because the registry-reported design is single-group with one arm, there is no randomized treatment-versus-control contrast. A Cox model could be used descriptively to model time-to-event relationships if covariates or comparative groups were available, but a treatment hazard ratio comparing T-DXd with a control treatment cannot be produced from this single-group design alone.
7. Results
Results are posted for both registered primary endpoints, but no formal statistical analyses were posted. The registry data therefore identify the endpoints and their methods without providing the numerical endpoint estimates, confidence intervals, or p-values needed for a complete numerical efficacy analysis.
Primary Endpoint 1: Objective Response Rate
Registered analysis
Cohort 1: participants without brain metastasis at baseline
Confirmed complete response or confirmed partial response by independent central review according to RECIST 1.1.
The ORR endpoint asks a straightforward binary question: what percentage of participants in the specified cohort achieved a confirmed complete or partial response? Because the ClinicalTrials.gov record does not provide the number of responders or the analyzed denominator, no numerical ORR can be reported here without importing information from another source.
An ORR estimate would describe the frequency of response in the treated cohort. It would not by itself establish a causal treatment effect, because there is no randomized comparator arm in the registry-reported design. It would also not mean that the remaining participants experienced disease progression; participants who did not meet the confirmed CR or PR definition form a broader category that can include other response or disease-status outcomes.
A confidence interval, if reported, would quantify statistical uncertainty around the estimated response probability. It would not describe the range of responses that individual participants might experience. A p-value, if a formal hypothesis test had been prespecified against a reference response rate, would address evidence against that reference hypothesis rather than measuring the magnitude of the observed response.
Primary Endpoint 2: 12-Month Progression-Free Survival Rate
Registered analysis
Cohort 2: participants with brain metastasis at baseline
Percentage alive and free of disease progression at 12 months, estimated by the Kaplan-Meier method.
The 12-month PFS rate is a time-to-event estimate rather than a simple proportion of participants who happened to be observed for 12 months. Kaplan-Meier estimation accounts for censoring and uses the available event-time information to estimate the probability of remaining alive and free of progression at the specified time point.
The ClinicalTrials.gov record does not provide the resulting 12-month PFS estimate, its confidence interval, or a formal p-value. Therefore, no numerical treatment-effect claim can be made from the ClinicalTrials.gov record.
Importantly, a 12-month PFS rate is not the same as median PFS and is not a hazard ratio. A hazard ratio would require a comparative time-to-event framework, whereas this registered endpoint is a single-cohort Kaplan-Meier estimate.
8. Safety Results
The ClinicalTrials.gov record reports serious adverse events by cohort. These are presented as affected participants over participants at risk. Because the study is single-group, the two rows represent different clinically defined cohorts rather than randomized treatment arms.
| Cohort | Intervention | Serious adverse events | Interpretation of denominator |
|---|---|---|---|
| Cohort 1 | Trastuzumab Deruxtecan (T-DXd) | 46/241 | 46 affected participants among 241 at risk |
| Cohort 2 | Trastuzumab Deruxtecan (T-DXd) | 62/263 | 62 affected participants among 263 at risk |
These safety counts should not be interpreted as a randomized comparison because both cohorts received the same intervention and were defined according to brain-metastasis status at baseline. Differences between the cohort-level counts or proportions would therefore not constitute evidence that T-DXd causes different serious-adverse-event rates between randomized treatments.
The notation 46/241, for example, means that 46 participants were affected among 241 participants at risk in Cohort 1 according to the registry summary. It is a descriptive safety count. Without a comparator arm or additional adjustment framework, it should not be converted into a comparative treatment-effect claim.
9. Statistical Methods Explained
Why is ORR a binary endpoint?
The registered ORR definition classifies participants according to whether they achieved a confirmed complete response or confirmed partial response. That creates a binary analysis target: response meeting the specified definition versus not meeting it. The natural descriptive summary is therefore a proportion.
Why does independent central review matter?
The registry specifies that CR and PR are determined by independent central review using RECIST 1.1. Central review creates a prespecified assessment framework for classifying tumor response and separates the endpoint determination from relying solely on the investigator's assessment.
Why is Kaplan-Meier used for 12-month PFS?
PFS includes a time component: participants are followed until disease progression or death, while some participants may reach the end of observation without either event. Kaplan-Meier estimation is designed to use information from both observed events and right-censored follow-up when estimating the event-free probability over time.
Why is a 12-month PFS rate not the same as median PFS?
A 12-month PFS rate asks for the estimated probability of remaining alive and progression-free at one particular time point. Median PFS asks for the time by which the estimated PFS probability has fallen to 50%. These are different summaries and cannot be substituted for one another.
Why is there no treatment hazard ratio in this study?
A treatment hazard ratio is inherently comparative: it describes the relative hazard between two groups under a specified model. The registry design is single-group with one intervention arm. Without a randomized or otherwise defined comparison group, a treatment-versus-control hazard ratio is not available from this design.
What would a confidence interval add to ORR?
A confidence interval around an ORR estimate would communicate sampling uncertainty around the observed response probability. A narrow interval would indicate greater statistical precision than a wide interval under the specified confidence procedure. The interval would still describe uncertainty in the population response estimate, not the range of outcomes for individual participants.
What would a p-value mean if one were reported?
A p-value would need to be tied to a specific prespecified null hypothesis. For a single-group binary endpoint, that could involve comparison with a predefined reference response rate; for a time-to-event endpoint, a formal test could compare an estimate with a reference value under an explicitly defined hypothesis. A p-value does not measure effect size and cannot substitute for an effect estimate and confidence interval.
10. Cohort Structure and Statistical Interpretation
The trial's two primary endpoints are deliberately tied to different baseline brain-metastasis cohorts. This structure has an important statistical consequence: the endpoint populations are not interchangeable.
| Component | Cohort 1 | Cohort 2 |
|---|---|---|
| Baseline brain metastasis status | Without brain metastasis at baseline | With brain metastasis at baseline |
| Primary endpoint | Objective Response Rate | Progression-free Survival Rate at 12 Months |
| Endpoint type | Binary | Time-to-event |
| Registered method | Confirmed CR or PR by ICR using RECIST 1.1 | Kaplan-Meier method |
| Time frame | From first dose (Day 1) to progression of disease, up to 2 years 7 months | At 12 months |
Because the primary endpoints differ between the cohorts, the study should not be reduced to a single overall efficacy number. Each endpoint answers a different question in a defined population. This is also why the absence of a numerical formal analysis in the registry-reported statistical-analysis field is important: the appropriate statistical summary depends on the endpoint and cohort being analyzed.
11. What This Design Can and Cannot Establish
What it can describe
The observed frequency of confirmed CR or PR in Cohort 1 and the estimated probability of being alive and progression-free at 12 months in Cohort 2.
What it cannot directly compare
The study design does not provide a randomized T-DXd-versus-control treatment comparison.
What ORR does not establish
ORR does not by itself establish duration of benefit, progression-free survival, or overall survival.
What 12-month PFS does not establish
A 12-month PFS rate does not by itself provide a median PFS or a treatment hazard ratio.
This distinction is particularly important when reading results from a single-group oncology study. A strong observed outcome can be clinically informative while still requiring caution about causal attribution, because the design does not create the same randomized counterfactual as a two-arm controlled trial.
12. Registry Results and Missing Formal Analyses
The ClinicalTrials.gov record states that results are posted and identifies 27 posted outcome measures. It also identifies both primary endpoints as having results posted. However, no formal statistical analyses were posted.
| Information category | ClinicalTrials.gov record |
|---|---|
| Results posted | Yes |
| Outcome measures posted | 27 |
| Primary endpoints | 2 |
| Statistical analyses reported | None |
| Formal numerical estimates posted on ClinicalTrials.gov for the primary endpoints | Not provided in the ClinicalTrials.gov record |
| Linked publication | PMID 39271844 |
13. Multiplicity, Stratification, and Other Design Topics
The ClinicalTrials.gov record does not report a non-inferiority margin, crossover, factorial design, interim-analysis strategy, missing-data or imputation method, stratification factors, or Bayesian analysis. Those topics are therefore not assigned a trial-specific method or interpretation here.
| Design topic | What the ClinicalTrials.gov record supports |
|---|---|
| Non-inferiority margin | Not reported in the ClinicalTrials.gov record. |
| Crossover | Not reported in the ClinicalTrials.gov record. |
| Factorial design | Not reported; the registered design model is SINGLE_GROUP. |
| Interim analysis | Not reported in the ClinicalTrials.gov record. |
| Missing-data / imputation strategy | Not reported in the ClinicalTrials.gov record. |
| Stratification factors | Not reported in the ClinicalTrials.gov record. |
| Bayesian methods | Not reported in the ClinicalTrials.gov record. |
The absence of a reported method in the ClinicalTrials.gov record should not be interpreted as evidence that no such method was used in the underlying trial. It means only that the requested source material does not provide enough information to describe it responsibly.
14. Statistical Interpretation of the Primary Endpoints
For ORR, the principal estimand is the response probability in the specified cohort under the registered response definition. The useful statistical presentation would normally include the estimated proportion and an appropriate confidence interval. Without the observed numerator and denominator for the efficacy endpoint in the ClinicalTrials.gov record, those quantities cannot be calculated here.
For 12-month PFS, the principal estimand is the Kaplan-Meier probability of being alive and free of disease progression at 12 months in the specified cohort. The useful statistical presentation would normally include the point estimate and confidence interval, together with enough information to understand censoring and event definitions. Those numerical results are not reported here.
Neither primary endpoint, as registered in this single-group design, directly estimates a randomized treatment effect. The appropriate interpretation is therefore descriptive and cohort-specific unless an explicitly defined external comparator or historical benchmark is introduced from another source.
15. Limitations
- Single-group design: the registry identifies one intervention arm and no concurrent comparator, limiting direct causal comparison with another treatment.
- Different endpoint structures: ORR is a binary response endpoint, whereas 12-month PFS is a time-to-event estimate. They should not be combined into a single efficacy measure.
- Cohort-specific interpretation: the two primary endpoints correspond to different baseline brain-metastasis populations.
- Missing numerical efficacy results in the ClinicalTrials.gov record: the registry identifies results as posted but no formal statistical analyses were posted or numerical primary-endpoint estimates.
- No confidence intervals reported: precision around the primary efficacy estimates cannot be assessed from the provided statistical-analysis data.
- No formal p-values reported: statistical significance cannot be evaluated from the ClinicalTrials.gov record, and no p-value should be inferred from the endpoint definitions.
- Safety is descriptive: serious adverse-event counts by cohort do not constitute randomized safety comparisons.
- Unreported methods: the ClinicalTrials.gov record does not specify stratification, missing-data handling, interim monitoring, multiplicity procedures, or Bayesian methods.
- External comparison: comparisons with historical response or survival benchmarks would introduce information not contained in the ClinicalTrials.gov record and require additional assumptions.
16. Why This Trial Matters Statistically
DESTINY-Breast12 is a useful teaching example because it illustrates how statistical interpretation changes when an oncology study uses a single-group design and cohort-specific endpoints rather than a randomized control group.
| Statistical concept | How it appears in DESTINY-Breast12 |
|---|---|
| Single-group design | One registered intervention arm with trastuzumab deruxtecan. |
| Binary endpoint | Objective response rate in participants without brain metastasis at baseline. |
| RECIST 1.1 | Used for the registered ORR definition. |
| Independent central review | Determines confirmed CR and PR for the ORR endpoint. |
| Time-to-event analysis | 12-month PFS rate in participants with brain metastasis at baseline. |
| Kaplan-Meier estimation | Specified method for the 12-month PFS rate. |
| Cohort-specific estimands | Different primary endpoints are assigned to the two brain-metastasis cohorts. |
| Confidence intervals | Would ordinarily accompany numerical endpoint estimates, but are not reported in the provided analysis data. |
| Comparative inference | Limited because the registry-reported design has no randomized comparator arm. |
| Safety analysis | Serious adverse events are reported descriptively by cohort. |
The most important statistical lesson is that the analysis method must follow the design. A Kaplan-Meier estimate can answer a well-defined time-to-event question without a control group. An ORR can be estimated in a single cohort. But neither automatically becomes a comparative treatment effect merely because the observed outcome is favorable or unfavorable.
17. Planned Analysis and Results Interpretation Framework
Although the registry states that results are posted, the registry-reported formal statistical-analysis data do not contain numerical analyses. The appropriate educational framework is therefore to identify what would ordinarily be evaluated for each endpoint without asserting an outcome.
| Endpoint | Typical analysis | Useful reported quantities | What the ClinicalTrials.gov record contains |
|---|---|---|---|
| ORR in Cohort 1 | Estimate the proportion with confirmed CR or PR; report an appropriate confidence interval. | ORR estimate, confidence interval, and analysis population. | Endpoint definition and time frame; no numerical efficacy estimate reported. |
| 12-month PFS rate in Cohort 2 | Kaplan-Meier estimation of the probability of being alive and progression-free at 12 months. | 12-month PFS estimate, confidence interval, event/censoring information. | Endpoint definition and Kaplan-Meier method; no numerical efficacy estimate reported. |
This distinction preserves an important statistical principle: a method can be specified even when its numerical output is unavailable in the data being used to construct an analysis page.
18. Sources
- ClinicalTrials.gov: NCT04739761.
- Linked publication: PubMed record for PMID 39271844: PMID 39271844.
Continue through the Clinical Biostats statistical pathway
Explore the statistical methods behind binary response endpoints, Kaplan-Meier estimation, confidence intervals, and clinical-trial interpretation.
19. Related Tutorials
Learn more about the methods used in this trial:
20. Related Calculators
Useful statistical tools for the endpoint types represented in this study:
21. Record Summary
DESTINY-Breast12 is a phase 3, single-group study enrolling 506 participants to evaluate trastuzumab deruxtecan in previously treated advanced or metastatic HER2-positive breast cancer among participants with or without brain metastasis. Its statistical structure is defined by two primary endpoints: a binary objective response rate endpoint in participants without brain metastasis at baseline and a 12-month Kaplan-Meier progression-free survival endpoint in participants with brain metastasis at baseline.
The most important interpretive feature is the absence of a randomized comparator in the registry-reported design. Consequently, the primary statistical questions concern estimation within defined cohorts rather than direct treatment-versus-control comparison. ORR is naturally summarized as a response proportion, while 12-month PFS is naturally estimated using Kaplan-Meier methodology.