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
AEGEAN is a randomized, double-blind, parallel-group phase 3 study of neoadjuvant/adjuvant durvalumab in patients with resectable non-small cell lung cancer. The registry lists 825 participants and two study arms, with pathological complete response and event-free survival as the primary endpoints.
| Feature | AEGEAN |
|---|---|
| Study acronym | AEGEAN |
| ClinicalTrials.gov identifier | NCT03800134 |
| Official title | A Study of Neoadjuvant/Adjuvant Durvalumab for the Treatment of Patients With Resectable Non-small Cell Lung Cancer |
| Phase | Phase 3 |
| Status | Active, not recruiting |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 825 |
| Number of arms | 2 |
| Condition | Non-Small Cell Lung Cancer |
| Lead sponsor | AstraZeneca |
| Sponsor type | Industry |
2. Clinical Question
The central statistical question is whether a treatment strategy incorporating neoadjuvant/adjuvant durvalumab produces a difference in pathological complete response and event-free survival among patients with resectable non-small cell lung cancer.
Population
Patients with resectable non-small cell lung cancer.
Intervention
Neoadjuvant/adjuvant durvalumab, studied alongside the chemotherapy agents and surgery listed in the registry.
Comparator
The registry lists placebo as an intervention in the two-arm study.
Primary question
How does the randomized treatment comparison affect pathological complete response and event-free survival?
3. Trial Design
Durvalumab-containing strategy
- Durvalumab
- Carboplatin or cisplatin
- Pemetrexed
- Paclitaxel
- Gemcitabine
- Surgery
Placebo-containing strategy
- Placebo
- Carboplatin or cisplatin
- Pemetrexed
- Paclitaxel
- Gemcitabine
- Surgery
4. Trial Timing and Registry Status
Study start
The registry lists 2018-12-06 as the study start date.
Primary completion
The registry lists 2022-11-10 as the primary completion date.
Active, not recruiting
The ClinicalTrials.gov record currently lists the study status as active, not recruiting.
The distinction between study completion and registry status is important. A trial may have completed its primary data collection while continuing to be followed or maintained in the registry. Here, the record separately reports a primary completion date and an active, not recruiting status.
5. Primary Endpoints
| Endpoint | Registry definition | Time frame | Statistical type |
|---|---|---|---|
| Pathological Complete Response (pCR) in modified intent-to-treat (mITT) population | Defined as the lack of any viable tumour cells after complete evaluation in the resected lung cancer specimen and all sampled regional lymph nodes. | Up to approximately 15 weeks after randomization | Binary pathological endpoint |
| Event-Free Survival (EFS) in modified intent to treat (mITT) population | An event is defined as documented RECIST 1.1 local or distant recurrence of lung cancer; death due to any cause; disease progression that precludes surgery or discovered upon attempting surgery that prevents completion of surgery. | Up to 5.5 years after first patient randomized. | Time-to-event endpoint |
Why the two primary endpoints are statistically different
pCR and EFS do not measure the same quantity. pCR is a binary outcome assessed at a defined postoperative point: each participant is classified according to whether the prespecified pathological definition is met. EFS is a time-to-event outcome: participants are followed until an event occurs or their event time is censored.
This distinction determines the appropriate statistical framework. A comparison of pCR is naturally expressed through proportions and an effect measure for binary outcomes, whereas EFS is naturally analyzed with methods that account for different follow-up times and censoring.
6. Modified Intent-to-Treat Analysis
Both registered primary endpoints specify the modified intent-to-treat (mITT) population. The use of mITT is consequential because it defines which randomized participants contribute to the primary efficacy analyses.
For a modified intent-to-treat analysis, the precise inclusion criteria determine which randomized participants enter the primary analysis set. The registry identifies the mITT population for both primary endpoints but does not provide its detailed construction in the information reported here.
That distinction should be kept separate from the broader intention-to-treat principle. In a pure ITT analysis, all randomized participants are ordinarily retained according to randomized assignment. An mITT population can apply additional prespecified criteria. The exact criteria matter because excluding participants after randomization can affect the population to which the primary estimate applies.
7. Planned Analysis
No formal statistical analyses are posted to ClinicalTrials.gov for the registered primary endpoints. The registry nevertheless defines what will be measured and the populations and time frames in which the endpoints are assessed.
Pathological Complete Response
pCR is a binary endpoint. Each participant meeting the registry definition of lack of any viable tumour cells after complete evaluation in the resected lung cancer specimen and all sampled regional lymph nodes would be classified as having a pathological complete response.
A treatment comparison for a binary endpoint would typically begin with the proportion of participants achieving pCR in each randomized group. Appropriate effect measures can include a risk difference, risk ratio, or odds ratio, with a corresponding confidence interval. The exact confirmatory model and hypothesis-testing procedure should follow the prespecified statistical analysis plan.
A higher pCR proportion would indicate that a greater fraction of the analyzed population met the prespecified pathological definition. It would not, by itself, establish a corresponding magnitude of improvement in long-term event-free survival. pCR is measured at approximately 15 weeks after randomization, whereas EFS extends over years.
Event-Free Survival
EFS is a time-to-event endpoint. The registry defines an event broadly: documented RECIST 1.1 local or distant recurrence, death from any cause, disease progression that precludes surgery, or disease progression discovered during attempted surgery that prevents completion of surgery.
For an endpoint of this type, the standard analytical framework would generally include Kaplan-Meier estimation to describe the event-free survival distribution and a treatment comparison based on the prespecified survival-analysis method. A Cox proportional-hazards model may be used to estimate a hazard ratio when specified by the statistical analysis plan.
Here, S(t) represents the probability of remaining free of the defined event beyond time t. Because participants may have different lengths of observed follow-up, survival-analysis methods allow censored observations to contribute information up to their censoring time.
A hazard ratio below 1 would represent a lower estimated instantaneous event rate in the durvalumab-containing group relative to the comparator under the fitted model. It would not mean that the same percentage of participants avoided an event, nor would it directly describe an absolute difference in event-free survival at a particular time point.
8. Statistical Methodology
Binary endpoint analysis for pCR
pCR is defined as a yes/no pathological outcome. The basic data structure is therefore a two-by-two treatment-by-response table: randomized treatment strategy by pCR status.
| pCR | No pCR | |
|---|---|---|
| Durvalumab-containing strategy | Count to be observed | Count to be observed |
| Placebo-containing strategy | Count to be observed | Count to be observed |
The key statistical issue is not simply whether one observed proportion is numerically larger than the other. A confirmatory analysis also needs an effect estimate, uncertainty interval, and prespecified hypothesis-testing framework. The mITT population is particularly important because it determines the denominator for the primary pCR comparison.
Kaplan-Meier estimation for EFS
Kaplan-Meier estimation is appropriate for time-to-event data because participants can remain event-free at the end of their observed follow-up. Those participants are censored rather than treated as if they experienced the event at the end of follow-up.
where di is the number of events at time ti and ni is the number at risk immediately before that time.
The resulting curve estimates the probability of remaining event-free over time. For AEGEAN, the event definition is especially important because EFS includes several clinically distinct ways of reaching the endpoint.
Hazard ratios and Cox models
If a Cox proportional-hazards model is prespecified for EFS, the hazard ratio summarizes the relative instantaneous event rate between randomized treatment strategies under that model.
The hazard ratio is a relative model-based measure. It is not an absolute risk difference, a risk ratio at a fixed time point, or the percentage of participants who benefit.
The proportional-hazards assumption also matters. If the relative hazard changes materially over time, a single Cox hazard ratio may compress a more complicated treatment-effect pattern into one summary number. Time-specific EFS estimates can therefore provide important complementary information.
Censoring
EFS analysis requires explicit rules for determining when a participant's follow-up ends without an observed event. Such observations are censored. The validity of a survival analysis depends in part on the relationship between the censoring process and the underlying event process.
For AEGEAN, the registry defines the events but does not provide detailed censoring or missing-data rules in the information reported here. Those rules are important components of the full statistical analysis plan.
9. Statistical Methods Explained
Why is pCR analyzed differently from EFS?
pCR is a binary endpoint assessed within a defined period after randomization, whereas EFS records the timing of clinically defined events. Binary analysis focuses on proportions and between-group effect measures; survival analysis must additionally account for censoring and unequal follow-up.
Why does the mITT population matter?
The primary endpoints are explicitly defined in the mITT population. Therefore, the denominator and inclusion rules used for the primary analysis are part of the estimand being reported. If participants are excluded according to prespecified criteria, the resulting treatment effect describes that defined mITT population rather than automatically describing every randomized participant.
Why use Kaplan-Meier estimation for EFS?
Not every participant will experience recurrence, progression, or death during the same observation period. Kaplan-Meier estimation allows participants who remain event-free at their last observation to contribute information without incorrectly treating them as having experienced an event.
What does an EFS hazard ratio measure?
A hazard ratio compares the estimated instantaneous event rates between treatment strategies over the modeled follow-up. An HR below 1 would indicate a lower estimated event rate in the treatment group, but it would not directly tell us the absolute probability of remaining event-free at a particular time.
Why is the EFS event definition important?
The AEGEAN registry defines EFS using several event types: local or distant recurrence, death from any cause, progression preventing surgery, and progression discovered during attempted surgery that prevents completion. The statistical endpoint therefore represents a composite time-to-first-event outcome rather than recurrence alone.
Why should pCR and EFS not be treated as interchangeable?
pCR is an early pathological measure and EFS is a longitudinal clinical outcome. A difference in pCR describes what happened to the resected specimen at the specified assessment point; an EFS analysis describes the subsequent timing of defined clinical events. Their statistical interpretations are therefore distinct even when both are primary endpoints.
10. Interpreting Confidence Intervals and P-values
No numerical treatment-effect estimates, confidence intervals, or p-values are posted in the ClinicalTrials.gov statistical-analysis fields for the primary endpoints. When such results are available, the confidence interval should be interpreted alongside the point estimate rather than as a substitute for it.
Confidence interval
A confidence interval describes statistical uncertainty around an estimated treatment effect under the specified analysis framework. It does not describe the range of outcomes that individual patients will experience.
P-value
A p-value addresses evidence against a specified null hypothesis under the statistical model. It does not measure the magnitude or clinical importance of a treatment effect.
Absolute effect
For pCR, the absolute difference in response proportions can provide an immediately interpretable measure of between-group separation.
Relative effect
For EFS, a hazard ratio can summarize relative event rates, but time-specific survival estimates are needed to understand absolute event-free probabilities.
11. Endpoint-Specific Interpretation
Pathological Complete Response
The pCR endpoint asks whether the resected lung cancer specimen and all sampled regional lymph nodes contain no viable tumour cells after complete evaluation. Statistically, it is therefore a categorical endpoint with a clinically defined success criterion.
The approximately 15-week time frame is important. It fixes the period in which the pathological outcome is evaluated and separates this endpoint from the much longer EFS follow-up.
A statistically supported difference in pCR would establish a difference in the proportion of participants meeting the specified pathological criterion. It would not by itself quantify the long-term effect on EFS or establish that every participant with pCR will remain event-free.
Event-Free Survival
EFS captures events occurring after randomization over a period extending up to 5.5 years after the first patient was randomized. The composite definition means that an event can arise from recurrence, death, or failure to complete surgery because of disease progression.
A statistically supported difference in EFS would describe separation between the randomized groups with respect to the prespecified composite event process. The magnitude should be interpreted using both relative measures such as a hazard ratio and absolute time-specific EFS estimates.
12. Statistical Considerations for a Two-Endpoint Trial
AEGEAN has two registered primary endpoints. This creates an important multiplicity question: the confirmatory testing strategy must specify how the two primary endpoints are handled while controlling the relevant type I error rate.
The registry information reported here identifies both pCR and EFS as primary endpoints but does not provide the alpha-allocation or multiplicity procedure. Consequently, a future numerical result for either endpoint should be interpreted in the context of the prespecified testing hierarchy or multiplicity strategy rather than by looking at an isolated nominal p-value.
| Endpoint | Primary role | Core statistical structure | Key interpretation issue |
|---|---|---|---|
| pCR | Primary | Binary outcome | Definition of response and mITT denominator |
| EFS | Primary | Time-to-event outcome | Event definition, censoring, and time-to-event modeling |
Multiplicity is not a technical footnote. When multiple primary hypotheses are evaluated, the probability of at least one false-positive conclusion can increase unless the analysis plan accounts for the multiple testing structure. The appropriate interpretation therefore depends on the prespecified confirmatory framework.
13. Surgical and Composite-Endpoint Considerations
The EFS definition includes events specifically related to the ability to proceed through surgery. This makes the endpoint different from a conventional recurrence-free survival definition in which recurrence and death may be the principal components.
Including disease progression that precludes surgery can capture clinically important treatment-pathway failure earlier than waiting for a later recurrence event. At the same time, the statistical interpretation must respect the fact that several event mechanisms are combined into one endpoint.
Composite endpoint
EFS combines multiple prespecified event types into a single time-to-first-event endpoint.
Event hierarchy
Participants are counted when the first qualifying EFS event occurs, regardless of which qualifying event type occurs first.
Clinical interpretation
The treatment comparison concerns the composite endpoint as defined, not recurrence alone.
Statistical interpretation
Different component events may have different clinical meanings, so a composite result should be interpreted together with its component definitions.
14. Limitations
- No posted statistical results: the ClinicalTrials.gov record does not contain posted statistical analyses for the registered primary endpoints.
- mITT definition: the registry identifies the mITT population but does not provide the detailed construction criteria in the information reported here.
- Multiplicity: the registry identifies two primary endpoints but does not report the multiplicity-control procedure in the available trial information.
- Censoring rules: the event definitions are specified, but detailed censoring and missing-data rules are not reported in the available registry information.
- Composite EFS endpoint: EFS combines recurrence, death, and surgery-related progression events, which are clinically distinct event types.
- Hazard-ratio assumptions: if a Cox proportional-hazards model is used, its assumptions must be evaluated rather than treating a single hazard ratio as a complete description of the event-free survival experience.
- Endpoint timing: pCR and EFS operate on very different time scales, so evidence for one endpoint should not automatically be substituted for evidence for the other.
- Registry detail: the ClinicalTrials.gov record provides the core design and endpoint definitions but not every statistical-analysis-plan detail needed to reproduce a complete confirmatory analysis.
15. Why This Trial Matters Statistically
AEGEAN is a useful statistical teaching case because it connects an early binary pathological endpoint with a long-term time-to-event endpoint within a randomized, double-masked phase 3 design.
| Concept | How it appears in AEGEAN |
|---|---|
| Randomization | The study is registered as randomized with two arms. |
| Parallel-group design | The two randomized strategies are evaluated in a parallel design. |
| Double masking | The registry identifies the study as double-masked. |
| Modified intent-to-treat | Both primary endpoints are defined in the mITT population. |
| Binary endpoint | pCR is defined as a pathological yes/no outcome. |
| Time-to-event analysis | EFS follows participants over time until a defined event or censoring. |
| Composite endpoint | EFS combines recurrence, death, and progression preventing or discovered during attempted surgery. |
| Kaplan-Meier | A standard framework for estimating EFS over time. |
| Hazard ratio | A potential relative summary for the EFS treatment comparison when a Cox model is prespecified. |
| Multiplicity | Two primary endpoints require a prespecified confirmatory testing strategy. |
| Different endpoint horizons | pCR is assessed up to approximately 15 weeks after randomization, whereas EFS extends up to 5.5 years after first patient randomized. |
16. Statistical Concepts in Practice
The AEGEAN design illustrates an important principle in clinical-trial statistics: the scientific question determines the endpoint, and the endpoint determines the statistical framework.
For pCR, the central quantity is the probability of meeting a defined pathological criterion. For EFS, the central quantity is the distribution of time until one of several clinically defined events occurs. These require different estimands, different data structures, and different approaches to uncertainty.
A sound treatment comparison requires all six pieces. A numerical difference without its population, endpoint definition, effect measure, and uncertainty is incomplete evidence.
17. Sources
- ClinicalTrials.gov: AEGEAN, NCT03800134.
- PubMed: PubMed record.
- PubMed: PubMed record.
Continue through the Clinical Biostats knowledge graph
Clinical trial endpoints connect naturally to deeper statistical methods, interpretation frameworks, and practical analysis workflows.
18. Record Summary
AEGEAN is a randomized, double-masked, parallel-group phase 3 study enrolling 825 participants with resectable non-small cell lung cancer. The registry identifies pathological complete response and event-free survival as the two primary endpoints, both analyzed in the modified intent-to-treat population.
The statistical structure is deliberately different for the two endpoints. pCR is a binary pathological outcome assessed up to approximately 15 weeks after randomization, while EFS is a time-to-event outcome followed for up to 5.5 years after first patient randomized. The EFS definition includes local or distant recurrence, death from any cause, and disease progression that prevents or is discovered during attempted surgery and prevents completion of surgery.
For statistical interpretation, the most important principles are therefore the definition of the mITT population, the binary nature of pCR, the time-to-event structure of EFS, the composite nature of the EFS endpoint, censoring, and the confirmatory testing strategy for two primary endpoints. The ClinicalTrials.gov record does not currently contain posted statistical analyses for these primary endpoints, so numerical treatment-effect estimates and inferential results are not available from the registry record.