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
IMpassion030 was a randomized, open-label, parallel-group phase 3 trial evaluating atezolizumab in combination with adjuvant anthracycline/taxane-based chemotherapy versus chemotherapy alone in patients with operable triple-negative breast cancer. The registry reports 2,199 randomized participants and one primary time-to-event endpoint: invasive disease-free survival.
| Feature | IMpassion030 |
|---|---|
| Phase | Phase 3 |
| Condition | Triple Negative Breast Cancer |
| Design | Randomized, parallel-group, open-label |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 2,199 |
| Primary endpoint | Invasive Disease-Free Survival (iDFS) |
| Primary endpoint type | Time-to-event |
| Primary analysis | Log-rank test |
| Effect measure | Hazard ratio |
| Hypothesis type | Superiority |
| Status | Terminated |
| Start | 2018-08-02 |
| Primary completion | 2023-08-14 |
| Lead sponsor | Hoffmann-La Roche |
2. Clinical Question
The primary clinical question was whether adding atezolizumab to adjuvant anthracycline/taxane-based chemotherapy changes invasive disease-free survival compared with chemotherapy alone in patients with operable triple-negative breast cancer.
Population
Patients with operable triple-negative breast cancer.
Intervention
Atezolizumab in combination with adjuvant anthracycline/taxane-based chemotherapy.
Comparator
Adjuvant anthracycline/taxane-based chemotherapy alone.
Primary question
Does the atezolizumab-containing regimen improve invasive disease-free survival relative to chemotherapy alone?
3. Trial Design
Atezolizumab and chemotherapy
- Atezolizumab
- Paclitaxel
- Dose-dense doxorubicin or dose-dense epirubicin
- Cyclophosphamide
Chemotherapy
- Paclitaxel
- Dose-dense doxorubicin or dose-dense epirubicin
- Cyclophosphamide
The registry describes the study as randomized and parallel-group, with no masking. The primary purpose was treatment. These features are important for interpreting the primary efficacy comparison: randomization establishes the treatment groups before outcome follow-up, while the open-label design means treatment assignment was not masked.
Enrollment and trial timeline
Study start
The registry records the start of IMpassion030 on August 2, 2018.
Primary completion
The registry records primary completion on August 14, 2023.
4. Primary Endpoint
| Endpoint | Registry definition | Analysis |
|---|---|---|
| Invasive Disease-Free Survival (iDFS) | From randomization until the occurrence of an iDFS event or death from any cause, whichever occurred earlier (up to 5 years). | Kaplan-Meier estimates and log-rank comparison; hazard ratio as the effect measure. |
What counted as an iDFS event?
The registry defines iDFS as the time from randomization until the first occurrence of any of the following: ipsilateral invasive breast tumor recurrence; ipsilateral local-regional invasive breast cancer recurrence; ipsilateral second primary invasive breast cancer; contralateral invasive breast cancer; distant recurrence; or death attributable to any cause, including breast cancer, non-breast cancer, or an unknown cause.
The definition therefore combines several clinically meaningful recurrence events with death into one time-to-event endpoint. Participants with no event at the time of analysis, or with no post-baseline information, were censored for the Kaplan-Meier analysis.
5. Statistical Methodology
Intention-to-treat analysis
The primary analysis population was the ITT population, defined in the registry as all randomized participants, whether or not the assigned study treatment was received. The comparison was between the randomized chemotherapy group and the randomized atezolizumab-plus-chemotherapy group.
The ITT principle preserves the comparison created by randomization. It is especially important for a superiority trial because excluding participants after randomization can introduce post-randomization selection into the efficacy comparison.
Kaplan-Meier estimation
The registry states that iDFS was analyzed using Kaplan-Meier estimates. Kaplan-Meier estimation describes the probability of remaining event-free over time while allowing participants who have not experienced an event by their last available observation to contribute information until censoring.
Here, di is the number of events at time ti and ni is the number at risk immediately before that event time.
Log-rank test
The formal comparison reported in the registry used a log-rank test. For a time-to-event endpoint, the log-rank test evaluates whether the observed pattern of events over follow-up differs between randomized treatment groups.
The test is based on event information across follow-up rather than simply comparing the proportion of participants with an event at one fixed time point. That makes it appropriate for an endpoint such as iDFS, where both the timing of events and censoring matter.
Hazard ratio
The registry reports a hazard ratio as the effect measure. The primary comparison is expressed as chemotherapy relative to atezolizumab plus chemotherapy.
Primary iDFS hazard ratio
95% CI: 0.87–1.42 · P = 0.3846
Two-sided confidence interval; superiority hypothesis.
Because the comparison is specified as Chemotherapy vs Atezolizumab and Chemotherapy, an HR above 1 corresponds to a higher estimated instantaneous rate of an iDFS event in the chemotherapy group relative to the atezolizumab-containing group. Conversely, an HR below 1 would favor the atezolizumab-containing group under this direction of comparison.
Stratified analysis
The registry identifies the following stratification factors for the primary analysis: axillary nodal status, surgery (breast conserving versus mastectomy), and tumor PD-L1 status.
| Stratification factor | Categories reported by the registry |
|---|---|
| Axillary nodal status | Used as a stratification factor |
| Surgery | Breast conserving vs mastectomy |
| Tumor PD-L1 status | Used as a stratification factor |
Stratification incorporates these prespecified factors into the time-to-event comparison. Conceptually, this allows the randomized comparison to account for the stratified design rather than treating the study as though these factors had played no role in randomization or analysis.
6. Primary Result: Invasive Disease-Free Survival
The registry reports one formal statistical analysis for the primary endpoint. The analysis used the ITT population and compared chemotherapy with atezolizumab plus chemotherapy using a log-rank test. The reported effect measure was a hazard ratio.
| Measure | Reported result |
|---|---|
| Endpoint | Invasive Disease-Free Survival (iDFS) |
| Analysis population | ITT population |
| Groups compared | Chemotherapy vs Atezolizumab and Chemotherapy |
| Method | Log-rank test |
| Hazard ratio | 1.11 |
| 95% confidence interval | 0.87–1.42 |
| P-value | 0.3846 |
| Hypothesis | Superiority |
| Time frame | From randomization until the occurrence of an iDFS event or death from any cause, whichever occurred earlier (up to 5 years) |
The reported HR of 1.11 means that, under the direction of the registry's comparison, the estimated instantaneous rate of an iDFS event or death was 11% higher in the chemotherapy group than in the atezolizumab-plus-chemotherapy group. Equivalently, because the reciprocal of 1.11 is approximately 0.90, the estimate is compatible with an effect that is close to no relative difference between the groups.
The HR should not be interpreted as an 11% difference in the proportion of patients who experienced an event, nor as an 11% difference in absolute iDFS. It is a relative time-to-event measure that incorporates the timing of events over follow-up.
The 95% CI of 0.87–1.42 indicates substantial statistical uncertainty around the point estimate. Importantly, the interval includes 1.00, the value corresponding to equal hazards under the hazard-ratio framework. The interval therefore includes both a possible relative reduction and a possible relative increase in the instantaneous event rate under the model.
The P-value of 0.3846 measures the compatibility of the observed test statistic with the null hypothesis under the specified statistical framework. It does not measure the magnitude of the treatment effect, the probability that either treatment is beneficial, or the probability that the null hypothesis is true.
Because this was a time-to-event analysis, interpretation also depends on censoring and on the assumptions underlying the hazard-ratio representation. A single hazard ratio is most straightforward when the relative hazards are reasonably stable over time; the registry does not provide additional information here about proportional-hazards diagnostics.
Why the confidence interval is as important as the point estimate
The HR of 1.11 alone gives only one estimate. The interval from 0.87 to 1.42 communicates how much uncertainty surrounds that estimate. It spans values below and above 1, so the data represented by the reported interval are compatible with different directions and magnitudes of relative treatment effect.
The graphic is a conceptual display of the reported numerical range rather than a forest plot. It emphasizes the central statistical feature of the result: the confidence interval crosses the null value of 1.
7. Serious Adverse Events
The registry reports serious adverse events by randomized treatment arm using affected participants over participants at risk.
| Treatment group | Serious adverse events | At risk |
|---|---|---|
| Chemotherapy | 173 | 1,084 |
| Atezolizumab and Chemotherapy | 280 | 1,093 |
Chemotherapy
173 participants with serious adverse events among 1,084 participants at risk.
Atezolizumab and chemotherapy
280 participants with serious adverse events among 1,093 participants at risk.
These safety figures are descriptive counts by treatment group. They should not be interpreted as an adjusted comparison of safety risk because the registry information reported here does not provide a formal inferential analysis for these serious adverse events.
8. Statistical Methods Explained
Why was a time-to-event endpoint used?
iDFS records not only whether an event occurred, but also when the first qualifying event occurred. That structure preserves more information than a simple binary endpoint at a single time point. Participants who remain event-free can contribute follow-up until censoring, allowing the analysis to use different lengths of observed follow-up.
What does the hazard ratio of 1.11 mean?
With the registry's comparison direction of chemotherapy versus atezolizumab plus chemotherapy, an HR of 1.11 represents an estimated instantaneous event rate that is 1.11 times as high in the chemotherapy group. Put differently, the estimate is above 1, but its interpretation must be considered together with its confidence interval and P-value rather than treated as an isolated measure of treatment effect.
Why is 1 the key reference value for a hazard ratio?
A hazard ratio of 1 represents equal instantaneous event rates between the two comparison groups under the model. Values below 1 favor the denominator group in the registry's comparison, while values above 1 favor the numerator group. The reported 95% CI of 0.87–1.42 crosses this reference value.
What does the log-rank test contribute?
The log-rank test provides a formal comparison of the time-to-event experience between the randomized groups. Rather than comparing only one fixed-time survival proportion, it evaluates the pattern of observed events across follow-up while accounting for participants who remain at risk at different times.
Why use the ITT population?
The ITT population retains all randomized participants in their assigned groups. This maintains the treatment comparison established by randomization and avoids redefining the randomized groups based on treatment receipt after allocation.
Why does censoring matter for Kaplan-Meier estimation?
A participant who has not experienced an iDFS event by the time they are last observed does not automatically count as an event-free participant forever. Instead, their observation is censored at the appropriate point. Kaplan-Meier estimation uses their available follow-up while recognizing that their later outcome is unknown.
Why were stratification factors used?
The primary analysis incorporated axillary nodal status, surgery type, and tumor PD-L1 status as stratification factors. Stratified analysis is particularly relevant when randomization is organized around prespecified factors because it retains the structure of the randomized design in the treatment comparison.
9. Interpreting the Superiority Framework
The registry identifies the hypothesis type as superiority. A superiority analysis asks whether the randomized treatment groups differ in the prespecified direction and according to the endpoint and statistical testing framework.
For the hazard-ratio representation, the conventional null reference is HR = 1. The reported HR of 1.11 is accompanied by a two-sided 95% confidence interval of 0.87–1.42 and a P-value of 0.3846.
The distinction between a point estimate and a statistical test is important. An HR of 1.11 does not by itself establish a treatment difference. The inferential result comes from the combination of the prespecified analysis method, the observed test statistic, the confidence interval, and the P-value.
10. Reading the iDFS Endpoint Correctly
What iDFS measures
Time from randomization to the first qualifying invasive recurrence, contralateral invasive breast cancer, distant recurrence, or death from any cause.
What an iDFS event does not require
An iDFS event is not limited to death. Several recurrence categories qualify as events before death.
Why timing matters
A participant experiencing an event earlier contributes a different time-to-event observation from a participant experiencing an event later.
Why censoring matters
Participants without an observed event by the analysis point contribute follow-up through their censoring time.
This endpoint structure is why a time-to-event method is more informative than simply calculating the fraction of participants who experienced an iDFS event. The registry's analysis explicitly uses Kaplan-Meier estimates and a log-rank comparison.
11. Analysis Population and Treatment Assignment
The primary statistical analysis included all randomized participants in the ITT population, whether or not the assigned study treatment was received.
| Element | Registry description |
|---|---|
| Analysis population | ITT |
| Population definition | All randomized participants, whether or not assigned study treatment was received |
| Comparison | Chemotherapy vs Atezolizumab and Chemotherapy |
| Primary method | Log-rank test |
| Effect measure | Hazard ratio |
| Hypothesis type | Superiority |
This distinction is important when reading a randomized trial. The treatment effect is defined by randomized assignment rather than by whether a participant completed the intended treatment course. Consequently, treatment discontinuation or other post-randomization events do not automatically remove a participant from the primary ITT efficacy population.
12. Precision, P-values, and Effect Size
The reported effect estimate is HR 1.11. This is the primary quantitative description of the relative treatment comparison on the time-to-event scale.
The 95% CI of 0.87–1.42 describes uncertainty around the hazard-ratio estimate. Its width shows that the point estimate should not be interpreted without considering the range of compatible values represented by the interval.
The P-value of 0.3846 is associated with the reported log-rank analysis. A P-value is not an effect-size measure and should not be interpreted as the probability that the treatment works or does not work.
These three quantities answer different questions. The hazard ratio describes the estimated relative effect, the confidence interval describes uncertainty around that estimate, and the P-value describes the statistical evidence against the null hypothesis under the specified testing framework.
13. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary iDFS analysis reported an HR of 1.11 with a two-sided 95% CI of 0.87–1.42 and P = 0.3846 using a log-rank test in the ITT population.
Clinical interpretation
The statistical result describes the randomized comparison of iDFS between the two treatment strategies. The registry does not provide additional numerical outcome measures here that would support a broader clinical-effect summary.
Keeping these statements separate is important. Statistical evidence describes the observed randomized comparison under a defined model and testing procedure. Clinical interpretation requires considering the endpoint definition, magnitude and precision of the estimate, safety information, follow-up, and the broader clinical context.
14. Important Limitations and Interpretation Issues
- Confidence-interval width: the 95% CI extends from 0.87 to 1.42, so the point estimate should not be interpreted without acknowledging the uncertainty around it.
- Hazard-ratio interpretation: an HR is a relative time-to-event measure, not an absolute risk difference or a percentage of patients who benefit.
- Proportional-hazards assumption: a single hazard ratio summarizes relative event rates over follow-up. The registry information does not report a separate assessment of proportional hazards.
- Censoring: Kaplan-Meier estimation depends on the handling of participants who have not experienced an event by the time they leave observation or reach the analysis cutoff.
- Open-label design: treatment assignment was not masked. This is a design feature that should be distinguished from the randomized nature of the efficacy comparison.
- Composite endpoint: iDFS includes several recurrence categories as well as death from any cause, so an iDFS event is not synonymous with mortality.
- Safety and efficacy populations: the primary efficacy analysis is explicitly based on ITT participants. Safety information is reported separately by treatment group.
- Multiplicity: the ClinicalTrials.gov record identifies one registered primary endpoint and one posted statistical analysis. No additional multiplicity framework is reported here.
15. Why This Trial Matters Statistically
IMpassion030 is a useful teaching example because the trial connects randomization, a composite time-to-event endpoint, Kaplan-Meier estimation, log-rank testing, hazard-ratio interpretation, stratified analysis, and the ITT principle within one primary analysis.
| Concept | How it appears in IMpassion030 |
|---|---|
| Randomization | Participants were randomized between two parallel treatment groups. |
| ITT analysis | The primary analysis included all randomized participants, whether or not assigned treatment was received. |
| Time-to-event endpoint | iDFS was measured from randomization until the first qualifying event or death. |
| Kaplan-Meier estimation | Used to estimate iDFS over time while incorporating censoring. |
| Log-rank test | Used for the formal comparison between randomized groups. |
| Hazard ratio | Used as the primary effect measure. |
| Confidence interval | The HR was reported with a two-sided 95% CI of 0.87–1.42. |
| P-value | The primary analysis reported P = 0.3846. |
| Stratification | Axillary nodal status, surgery, and tumor PD-L1 status were used as stratification factors. |
| Safety analysis | Serious adverse events were reported by treatment group using affected participants and participants at risk. |
16. A Practical Statistical Reading of the Primary Result
A useful way to read the primary result is to move from the endpoint to the estimate, and then from the estimate to its uncertainty.
The final step is the formal statistical test: the reported log-rank P-value was 0.3846. Reading the result this way prevents the P-value from becoming the entire interpretation. The endpoint definition, analysis population, treatment comparison, effect estimate, and confidence interval all carry information that is lost if the result is reduced to a single significance label.
17. What the Hazard Ratio Does — and Does Not — Mean
With chemotherapy as the numerator of the comparison, an HR of 1.11 represents an estimated instantaneous iDFS-event rate 1.11 times that of the atezolizumab-plus-chemotherapy group, under the time-to-event model represented by the reported analysis.
It does not mean that 11% more participants experienced an iDFS event, that 11% of participants benefited or were harmed, or that every participant had an 11% difference in event risk.
A hazard ratio is relative. Absolute event probabilities at clinically relevant time points would provide a different perspective on the magnitude of an effect. The registry information reported here does not provide those additional iDFS time-point estimates.
18. Related Statistical Concepts
Learn more about the methods used in this trial:
19. Related Statistical Calculators
20. Sources
- ClinicalTrials.gov: IMpassion030, NCT03498716.
- PubMed: Publication indexed under PMID 39883436.
- PubMed: Publication indexed under PMID 32450725.
Continue through Clinical Biostats
Connect this trial's endpoints and statistical methods to deeper biostatistics tutorials and practical statistical tools.
21. Record Summary
IMpassion030 provides a clear example of a randomized phase 3 time-to-event analysis. The primary endpoint, invasive disease-free survival, combines recurrence and death events and is analyzed from randomization using Kaplan-Meier estimation and a log-rank comparison. The primary analysis uses the ITT population and reports a hazard ratio of 1.11 with a two-sided 95% confidence interval of 0.87–1.42 and a P-value of 0.3846. The registry also reports serious adverse events by treatment group, with 173 of 1,084 participants affected in the chemotherapy group and 280 of 1,093 in the atezolizumab-and-chemotherapy group.
From a statistical perspective, the central lesson is that the primary result should be read as a complete set of linked quantities: the endpoint definition, randomized analysis population, comparison direction, time-to-event method, hazard ratio, confidence interval, and P-value. Each describes a different part of the evidence.