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
e3 is a randomized, double-blind, phase 3 parallel-group trial evaluating hormone therapy with or without everolimus in patients with breast cancer. The registry identifies invasive disease-free survival (IDFS) as the primary endpoint and reports additional time-to-event analyses for overall survival (OS) and distant recurrence-free survival (DRFS).
| Feature | e3 |
|---|---|
| Trial name | e3 |
| ClinicalTrials.gov identifier | NCT01674140 |
| Brief title | S1207 Hormone Therapy With or Without Everolimus in Treating Patients With Breast Cancer |
| Phase | Phase 3 |
| Condition | Breast Cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 1939 |
| Primary endpoint | Invasive Disease-Free Survival (IDFS) |
| Primary endpoint type | Time-to-event |
| Primary analysis method | Log-rank test |
| Effect measure | Hazard ratio |
| Hypothesis type | Superiority |
2. Clinical Question
The central statistical question is whether adding everolimus to hormone therapy changes the time to invasive disease recurrence, second invasive primary cancer, or death compared with hormone therapy without everolimus.
Population
Patients with breast cancer enrolled in the phase 3 randomized e3 trial.
Intervention
Everolimus administered as part of the randomized treatment comparison alongside hormone therapy.
Comparator
Placebo in the control arm, with hormone therapy.
Primary question
Does the everolimus-containing strategy improve invasive disease-free survival relative to placebo?
The registered intervention list includes anastrozole, everolimus, exemestane, goserelin acetate, leuprolide acetate, letrozole, tamoxifen citrate, and placebo. The registry therefore describes the treatment strategy at the intervention level rather than reducing the trial to a single hormone-therapy agent.
3. Trial Design
Hormone therapy + placebo
- Placebo is the registered non-active comparator.
- Hormone therapy interventions are included in the registered intervention list.
Hormone therapy + everolimus
- Everolimus is the registered experimental drug.
- Hormone therapy interventions are included in the registered intervention list.
The registry reports 1939 participants enrolled overall. The posted primary and secondary statistical analyses use an eligible and evaluable population of 1792 participants, with 896 participants per arm.
4. Trial Timeline
Trial start
The registered study start date was September 12, 2013.
Primary completion
The registered primary completion date was December 1, 2022.
Active, not recruiting
The ClinicalTrials.gov record identifies the study as active, not recruiting.
5. Registered Primary Endpoint
| Endpoint | Time frame | Definition | Type |
|---|---|---|---|
| Invasive Disease-Free Survival (IDFS) | Up to 5 years post registration | Time from date of registration to date of first invasive recurrence (local, regional or distant), second invasive primary cancer (breast or not), or death due to any cause. Patients last known to be alive who have not experienced recurrence or second primary cancer are censored at their last contact date. | Time-to-event |
The registry states that the primary results were presented as a 5-year IDFS estimate. This is an important distinction from simply treating IDFS as a binary five-year outcome: the underlying endpoint is defined as a time-to-event variable, with censoring for participants who remain alive without recurrence or a second primary cancer at their last contact.
What is actually being counted as an IDFS event?
The composite endpoint includes several clinically distinct event types: invasive recurrence, a second invasive primary cancer, or death from any cause. Consequently, the IDFS event is not limited to recurrence of the original breast cancer. Death also qualifies as an event even when it is not preceded by a documented invasive recurrence.
6. Secondary Time-to-Event Endpoints
| Endpoint | Time frame | Type | Analysis method |
|---|---|---|---|
| Overall Survival (OS) | 5 years after last accrual | Time-to-event | Log-rank test; hazard ratio |
| Distant Recurrence-Free Survival (DRFS) | 5 years after last accrual | Time-to-event | Log-rank test; hazard ratio |
OS and DRFS extend the statistical evaluation beyond the primary IDFS endpoint. Their reported analyses use the same broad survival-analysis framework: comparison between randomized groups with a log-rank test and treatment effect summarized by a hazard ratio.
7. Statistical Methodology
Kaplan-Meier estimation
The registered endpoints are time-to-event outcomes. For this type of endpoint, Kaplan-Meier estimation is used to describe the estimated probability of remaining event-free over time while accounting for right-censoring.
Here, di represents the number of events at an event time and ni represents the number of participants at risk immediately before that time.
The ClinicalTrials.gov record does not report median IDFS, median OS, median DRFS, or Kaplan-Meier curve coordinates. Those quantities therefore are not necessary to interpret the posted hazard-ratio results.
Log-rank test
The registry identifies the log-rank test as the analysis method for the posted time-to-event comparisons. The test compares the observed pattern of events between treatment groups over follow-up rather than comparing only a single time-point proportion.
The resulting P-value addresses the evidence against the null comparison under the statistical testing framework. It does not quantify the magnitude of the treatment effect.
Hazard ratio
The posted analyses summarize treatment effects using hazard ratios. A hazard ratio compares the estimated instantaneous event rate between groups over the analyzed time-to-event follow-up.
For the analyses here, the hazard ratio is reported as Everolimus relative to Placebo. An HR below 1 therefore corresponds to a lower estimated hazard in the everolimus group relative to the placebo group.
Two-sided 95% confidence intervals
The posted analyses provide two-sided 95% confidence intervals for the hazard ratios. The interval describes statistical uncertainty around the estimated relative hazard under the model and sampling framework.
A confidence interval that crosses 1.00 includes the value corresponding to no hazard-ratio difference. That is directly relevant to interpretation of the IDFS, OS, and DRFS estimates reported for this trial.
8. Analysis Population
| Population | Participants | Role |
|---|---|---|
| Eligible and evaluable | 1792 | Posted efficacy analyses |
| Placebo | 896 | Analysis group |
| Everolimus | 896 | Analysis group |
| Overall enrolled | 1939 | Trial enrollment |
The distinction between enrollment and the analysis population matters. The trial enrolled 1939 participants, while the posted statistical analyses explicitly identify 1792 participants as eligible and evaluable. The primary analysis therefore should not be described as if all 1939 enrolled participants were necessarily included in the efficacy analysis.
9. Primary Result: Invasive Disease-Free Survival
The primary endpoint was invasive disease-free survival up to 5 years post registration. The registry reports a formal log-rank analysis in the 1792 eligible and evaluable participants, with 896 participants per arm.
Invasive Disease-Free Survival
95% CI: 0.77–1.14 · P = 0.52
Two-sided 95% confidence interval · Superiority hypothesis
An IDFS hazard ratio of 0.94 means that the estimated instantaneous rate of an IDFS event in the everolimus group was approximately 6% lower than in the placebo group, based on the reported hazard ratio. This is a relative time-to-event measure, not a 6-percentage-point reduction in the probability of an event.
The 95% CI of 0.77–1.14 indicates substantial uncertainty around the estimate. Because the interval includes 1.00, the range of effects compatible with the statistical uncertainty includes both a lower hazard and a higher hazard for the everolimus group relative to placebo.
The P-value of 0.52 addresses the evidence against the null hypothesis under the reported superiority testing framework. It does not measure the size of the effect, the probability that the treatment works, or the probability that the null hypothesis is true.
The endpoint is subject to right-censoring, and the hazard-ratio interpretation is based on a time-to-event model. A single hazard ratio should therefore not be interpreted as if it were an absolute five-year risk difference.
How to read HR 0.94 in context
An HR of 0.94 is numerically close to 1.00. The corresponding confidence interval extends from 0.77 to 1.14, so the statistical uncertainty is considerably wider than the point estimate alone suggests. The result therefore should be understood as an estimated relative hazard together with its uncertainty, rather than as a precise 6% reduction.
The registry's primary hypothesis type is superiority. For a superiority analysis, an effect estimate below 1 does not by itself establish superiority. The evidence is interpreted jointly through the prespecified statistical test, its P-value, and the confidence interval.
10. Secondary Result: Overall Survival
Overall survival was analyzed as a secondary time-to-event endpoint over 5 years after last accrual. The posted analysis includes the same 1792 eligible and evaluable participants, with 896 per arm.
Overall Survival
95% CI: 0.75–1.26 · P = 0.84
Two-sided 95% confidence interval · Superiority hypothesis
An OS hazard ratio of 0.97 is close to 1.00, corresponding to an estimated instantaneous death hazard approximately 3% lower in the everolimus group relative to placebo under the reported analysis.
The 95% CI of 0.75–1.26 includes 1.00 and spans both directions of possible relative effect. The interval therefore communicates considerably more than the point estimate alone: the data are compatible with a range of relative hazards around the null value.
The P-value of 0.84 is a measure of compatibility with the null hypothesis under the stated testing framework. It is not a measure of clinical importance and should not be interpreted as an 84% probability that there is no treatment effect.
Because OS is a time-to-event endpoint, censoring and the underlying event-time distribution remain important to interpretation. The registry does not report median OS or time-specific survival estimates here.
11. Secondary Result: Distant Recurrence-Free Survival
Distant recurrence-free survival was also analyzed over 5 years after last accrual. The registry reports a log-rank analysis with the treatment effect summarized by a hazard ratio.
Distant Recurrence-Free Survival
95% CI: 0.74–1.10 · P = 0.32
Two-sided 95% confidence interval · Superiority hypothesis
An HR of 0.90 corresponds to an estimated instantaneous distant-recurrence event rate approximately 10% lower in the everolimus group than in the placebo group under the reported model.
The 95% CI of 0.74–1.10 includes 1.00. The point estimate therefore should not be interpreted in isolation as evidence of a definitive reduction in distant recurrence.
The P-value of 0.32 does not quantify the size of the observed effect. It describes the statistical evidence against the null comparison under the stated testing framework.
As with IDFS and OS, this endpoint is a time-to-event outcome. The hazard ratio should not be translated directly into an absolute five-year probability without the underlying survival estimates.
12. Comparing the Three Posted Hazard Ratios
| Endpoint | Hazard ratio | 95% CI | P-value | Role |
|---|---|---|---|---|
| Invasive Disease-Free Survival | 0.94 | 0.77–1.14 | 0.52 | Primary |
| Overall Survival | 0.97 | 0.75–1.26 | 0.84 | Secondary |
| Distant Recurrence-Free Survival | 0.90 | 0.74–1.10 | 0.32 | Secondary |
All three posted point estimates are below 1.00, but their confidence intervals all include 1.00. The primary IDFS estimate is 0.94, the OS estimate is 0.97, and the DRFS estimate is 0.90. These are separate endpoints with different clinical event definitions and therefore should not be treated as interchangeable measures of the same outcome.
13. Statistical Methods Explained
Why was a log-rank test used?
Each registered efficacy endpoint is a time-to-event outcome. The log-rank test is designed to compare event-time distributions between treatment groups while using information across the follow-up period. This is more appropriate for survival data than a simple comparison of the percentage of participants who have experienced an event at one arbitrarily selected time.
What does a hazard ratio of 0.94 mean?
A hazard ratio of 0.94 indicates an estimated instantaneous event rate that is 0.94 times the corresponding rate in the comparator group under the reported analysis. Expressed descriptively, 1 − 0.94 = 0.06, so the point estimate corresponds to approximately a 6% lower estimated hazard. That does not mean a 6% absolute reduction in five-year IDFS events.
Why does the confidence interval include 1.00?
For a hazard ratio, 1.00 is the value corresponding to equal hazards between the groups. The IDFS confidence interval of 0.77–1.14 includes 1.00, as do the OS interval of 0.75–1.26 and the DRFS interval of 0.74–1.10. This means that the statistical uncertainty around each point estimate includes the no-difference value.
Why does the P-value not measure treatment-effect size?
The P-value reflects the statistical evidence against a specified null hypothesis under the testing framework. It depends on both the observed effect and the amount of information available. A P-value cannot by itself tell the reader whether an effect is clinically large, small, or precisely estimated.
Why is censoring important?
Participants do not necessarily experience an event during observation. For IDFS, the registry explicitly states that patients last known to be alive without recurrence or a second primary cancer are censored at their last contact date. Survival methods use this partial follow-up information rather than simply discarding it, while relying on assumptions about the relationship between censoring and the event process.
What is the difference between a hazard ratio and a five-year risk?
A hazard ratio is a relative measure comparing event rates over time. A five-year risk or survival estimate is an absolute probability at a specified time point. The two measures answer different questions. An HR of 0.94 cannot be converted into a five-year survival percentage without additional information about the underlying event-time distributions.
Why does the analysis population matter?
The registry identifies 1792 eligible and evaluable participants, with 896 in each analysis group. The hazard ratios therefore describe the posted analysis population rather than automatically applying to every one of the 1939 enrolled participants. Clearly identifying the population is essential when interpreting any treatment effect estimate.
14. Confidence Intervals and Statistical Precision
| Endpoint | Point estimate | Lower 95% CI | Upper 95% CI | Includes 1.00? |
|---|---|---|---|---|
| IDFS | 0.94 | 0.77 | 1.14 | Yes |
| OS | 0.97 | 0.75 | 1.26 | Yes |
| DRFS | 0.90 | 0.74 | 1.10 | Yes |
The widths of these intervals are informative. The IDFS interval spans 0.77 to 1.14, the OS interval spans 0.75 to 1.26, and the DRFS interval spans 0.74 to 1.10. The point estimate is only one part of the result; the confidence interval shows how much uncertainty surrounds that estimate.
A confidence interval can be relatively narrow while still including the null value, or relatively wide while excluding it. Statistical precision describes the uncertainty around an estimate, whereas statistical significance is determined by the prespecified testing framework. These concepts overlap but are not interchangeable.
15. P-values and Superiority Testing
The registry identifies the hypothesis type as superiority for each posted analysis. The P-values are 0.52 for IDFS, 0.84 for OS, and 0.32 for DRFS.
| Endpoint | Hypothesis type | Two-sided? | P-value |
|---|---|---|---|
| IDFS | Superiority | Yes | 0.52 |
| OS | Superiority | Yes | 0.84 |
| DRFS | Superiority | Yes | 0.32 |
In a superiority framework, the null comparison corresponds to no treatment difference. For a hazard ratio, that null is represented by 1.00. The two-sided confidence intervals and the P-values provide complementary descriptions of the evidence surrounding the treatment comparison.
16. Safety: Serious Adverse Events
The registry reports serious adverse events by treatment arm using affected participants divided by participants at risk.
| Treatment arm | Affected / at risk | Reported measure |
|---|---|---|
| Placebo | 77 / 881 | Serious adverse events |
| Everolimus | 180 / 874 | Serious adverse events |
The denominators differ slightly between the two safety summaries: 881 participants are reported as at risk in the placebo group and 874 in the everolimus group. That distinction matters when describing safety frequencies, because the numerator alone does not represent the incidence rate or proportion.
The registry does not provide additional serious-adverse-event detail in the information summarized here, such as event-level categories or exposure-adjusted rates. The reported 77/881 and 180/874 values should therefore be treated as the registry's arm-level serious-adverse-event summary rather than expanded into unreported event classifications.
17. Statistical Interpretation of the Primary Endpoint
Effect estimate
The IDFS hazard ratio is 0.94. The point estimate corresponds to approximately a 6% lower estimated hazard in the everolimus group relative to placebo.
Precision
The 95% CI is 0.77–1.14, indicating uncertainty that extends across the no-difference value of 1.00.
Hypothesis test
The two-sided P-value is 0.52 under the reported superiority framework.
Clinical endpoint
IDFS is a composite time-to-event endpoint incorporating invasive recurrence, second invasive primary cancer, or death.
These four components should be read together. Reporting only the hazard ratio would omit uncertainty. Reporting only the P-value would omit effect size. Reporting only the confidence interval would omit the prespecified hypothesis-testing result. And reporting any of these without the endpoint definition could obscure what constitutes an event.
18. Why the Three Endpoints Should Not Be Combined Into One Number
IDFS, OS, and DRFS overlap conceptually but are not identical. IDFS includes invasive recurrence, second invasive primary cancer, and death. OS focuses on death from any cause. DRFS focuses on distant recurrence-free time. Consequently, an event can contribute differently to each endpoint.
This distinction is particularly important for statistical interpretation. A treatment effect on IDFS does not automatically imply the same effect on OS, because the endpoints contain different event definitions. Similarly, DRFS isolates distant recurrence from the broader collection of events captured by IDFS.
| Endpoint | What the result describes | Posted HR |
|---|---|---|
| IDFS | First invasive recurrence, second invasive primary cancer, or death | 0.94 |
| OS | Overall survival | 0.97 |
| DRFS | Distant recurrence-free survival | 0.90 |
19. Important Statistical Limitations
- Analysis population: the posted efficacy analyses use 1792 eligible and evaluable participants rather than the full enrollment of 1939.
- Composite endpoint: IDFS combines invasive recurrence, second invasive primary cancer, and death, so the overall hazard ratio does not identify which component drives the result.
- Right-censoring: participants who remain event-free at last contact are censored, making the interpretation dependent on the survival-analysis framework and censoring assumptions.
- Hazard-ratio interpretation: a hazard ratio is a relative time-to-event measure and is not equivalent to a relative risk, absolute risk difference, or five-year survival probability.
- Proportional-hazards consideration: the meaning of a single hazard ratio is most straightforward when the relative hazards are reasonably represented by a common ratio over time. The registry summary does not report a formal assessment of the proportional-hazards assumption.
- Confidence intervals: all three posted 95% confidence intervals include 1.00, so the point estimates should not be interpreted without their uncertainty intervals.
- Multiple endpoints: IDFS is the registered primary endpoint, while OS and DRFS are secondary endpoints. Their P-values should therefore be interpreted according to their distinct roles in the trial rather than treated as interchangeable confirmatory tests.
- Unreported details: the registry information summarized here does not provide subgroup results, median survival times, Kaplan-Meier estimates at specific time points, or a detailed missing-data/imputation strategy.
20. What the Registry Does and Does Not Report
| Statistical feature | Reported | Educational implication |
|---|---|---|
| Primary endpoint | IDFS | Primary statistical interpretation centers on a time-to-event endpoint. |
| Analysis method | Log-rank test | Group comparison uses survival-analysis methodology. |
| Effect measure | Hazard ratio | Treatment effects are expressed on a relative event-rate scale. |
| 95% confidence interval | Yes | Precision can be assessed directly from the posted results. |
| Formal primary analysis | Yes | The registry contains a formal statistical analysis for IDFS. |
| Subgroup estimates | Not reported in the summarized record | No subgroup treatment-effect conclusions should be added to the analysis. |
| Median survival | Not reported in the summarized record | Median IDFS, OS, and DRFS should not be substituted for the reported hazard ratios. |
| Detailed missing-data strategy | Not reported in the summarized record | No specific imputation procedure can be attributed to the registry. |
21. Why This Trial Matters Statistically
The e3 trial is a useful example of how a randomized clinical trial can be interpreted when the central outcomes are time-to-event endpoints. The statistical story is not contained in a single P-value. It is built from the design, endpoint definition, analysis population, log-rank comparison, hazard ratio, confidence interval, and the distinction between primary and secondary endpoints.
| Concept | How it appears in e3 |
|---|---|
| Randomization | Participants were randomized in a two-arm parallel-group phase 3 design. |
| Blinding | The registry identifies the study as double-blind. |
| Time-to-event analysis | IDFS, OS, and DRFS are analyzed as time-to-event outcomes. |
| Log-rank test | The registry identifies log-rank testing for the posted analyses. |
| Hazard ratio | All three posted statistical analyses use hazard ratio as the effect measure. |
| Confidence interval | Two-sided 95% confidence intervals are reported for each posted hazard ratio. |
| Superiority testing | The posted analyses use a superiority hypothesis type. |
| Analysis population | The efficacy analyses include 1792 eligible and evaluable participants, 896 per arm. |
| Composite endpoint | IDFS combines invasive recurrence, second invasive primary cancer, and death. |
| Safety analysis | Serious adverse events are reported by randomized treatment arm using affected and at-risk counts. |
The trial also illustrates why statistical interpretation should remain closely tied to endpoint definitions. The same hazard-ratio framework is applied to three outcomes, but each outcome has a different clinical meaning. A reader who focuses only on the numerical HRs could miss that distinction.
22. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary IDFS analysis produced an HR of 0.94 with a two-sided 95% CI of 0.77–1.14 and P = 0.52. The secondary OS and DRFS analyses produced HRs of 0.97 and 0.90, respectively, with confidence intervals that also include 1.00.
Clinical interpretation
The endpoints measure different aspects of disease control and survival. Their results should therefore be discussed separately rather than compressed into a single overall treatment-effect statement.
The distinction is important because statistical evidence and clinical meaning are related but not identical. The hazard ratios quantify relative time-to-event comparisons. The endpoint definitions determine what those events represent. The confidence intervals communicate uncertainty, while the P-values describe evidence under the corresponding hypothesis-testing framework.
23. Sources
- ClinicalTrials.gov: NCT01674140 — e3 / S1207 Hormone Therapy With or Without Everolimus in Treating Patients With Breast Cancer.
- Linked publication: PubMed record for PMID 39621968.
- Linked publication: PubMed record for PMID 38833643.
Continue with the statistical methods
Explore the survival-analysis concepts that underlie randomized time-to-event trials, including hazard ratios, confidence intervals, Kaplan-Meier estimation, and log-rank testing.
24. Related Tutorials
Learn more about the methods used in this trial:
25. Related Calculators
26. Record Summary
The e3 trial provides a focused example of statistical analysis for a randomized phase 3 oncology study with a time-to-event primary endpoint. The primary IDFS analysis used the eligible and evaluable population of 1792 participants, with 896 per arm, and reported a hazard ratio of 0.94 with a two-sided 95% CI of 0.77–1.14 and P = 0.52. Secondary analyses reported an OS HR of 0.97 (95% CI 0.75–1.26; P = 0.84) and a DRFS HR of 0.90 (95% CI 0.74–1.10; P = 0.32).
The most informative way to read these results is to combine the effect estimates, confidence intervals, P-values, analysis population, and endpoint definitions. The primary endpoint is a composite time-to-event outcome, while OS and DRFS represent distinct secondary outcomes. The registry's use of log-rank testing and hazard ratios makes this trial a useful example of how survival-analysis methods translate randomized clinical-trial data into interpretable relative treatment effects.