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
AVAGAST was a randomized, double-blind, parallel phase 3 trial evaluating bevacizumab in combination with capecitabine and cisplatin as first-line therapy in patients with advanced gastric cancer. The registry identifies overall survival as the single primary endpoint and defines it as a time-to-event outcome.
| Feature | AVAGAST |
|---|---|
| Trial name | AVAGAST |
| ClinicalTrials.gov identifier | NCT00548548 |
| Brief title | A Study of Bevacizumab in Combination With Capecitabine and Cisplatin as First-line Therapy in Patients With Advanced Gastric Cancer |
| Phase | Phase 3 |
| Status | Completed |
| Condition | Adenocarcinoma |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 774 |
| Lead sponsor | Genentech, Inc. |
| Sponsor type | Industry |
| Results posted | Yes |
2. Clinical Question
The registry describes a phase 3 treatment trial of bevacizumab in combination with capecitabine and cisplatin as first-line therapy in patients with advanced gastric cancer. The primary statistical question is expressed through overall survival, measured from randomization until death.
Population
Patients with advanced gastric cancer and the registered condition of adenocarcinoma.
Intervention
Bevacizumab in combination with capecitabine and cisplatin as the registered first-line treatment approach.
Comparator
The trial registry identifies placebo as an intervention and records two randomized treatment arms.
Primary question
How does the randomized treatment comparison relate to overall survival, defined as time from randomization to death?
3. Trial Design
Bevacizumab combination
- Bevacizumab
- Capecitabine
- Cisplatin
Placebo-containing arm
- Placebo
- Capecitabine
- Cisplatin
The registry also lists 5-fluorouracil as an intervention. The ClinicalTrials.gov record does not provide additional arm-specific dosing, administration schedules, or treatment-sequence details, so those features are not inferred here.
4. Endpoints
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Overall Survival | The primary efficacy endpoint for this study was overall survival (time to death), defined as the time between randomization and the date of death irrespective of the cause of death. Patients for whom no death was captured on the clinical database were censored at the most recent date they were known to be alive. Median survival was estimated by the Kaplan-Meier method. | From randomization until death, up to 26 months | Time-to-event |
ClinicalTrials.gov lists 1 primary endpoint and identifies the primary endpoint type as time-to-event. The ClinicalTrials.gov record does not provide additional registered endpoint definitions beyond overall survival.
5. Statistical Methodology
Overall survival as a time-to-event endpoint
Overall survival is not analyzed simply by counting how many patients died. The endpoint incorporates when death occurred and accommodates patients whose death was not captured during the available observation period.
The registry defines overall survival as the time between randomization and the date of death irrespective of cause. If no death was captured on the clinical database, the patient was censored at the most recent date they were known to be alive.
The event is death from any cause. A patient without a captured death contributes follow-up until the most recent date they were known to be alive and is then censored.
Kaplan-Meier estimation
The registry explicitly states that median survival was estimated by the Kaplan-Meier method. Kaplan-Meier estimation is appropriate for time-to-event data because it incorporates event times while allowing observations to be right-censored.
Here, di represents the number of events at an event time and ni represents the number at risk immediately before that time.
What the ClinicalTrials.gov record does not establish
The ClinicalTrials.gov record does not report a formal statistical comparison, hazard ratio, confidence interval, p-value, log-rank test, Cox proportional-hazards model, stratification factors, interim-analysis procedure, multiplicity adjustment, or missing-data/imputation strategy.
Those methods should therefore not be attributed to AVAGAST on the basis of this registry extract alone. For this page, the confirmed analysis method is Kaplan-Meier estimation of median survival as specified in the registered endpoint definition.
6. Statistical Methods Explained
Why is overall survival a time-to-event endpoint?
Because the outcome contains both an event indicator and a time component. Two patients who both die are not statistically equivalent if one dies earlier than the other. The analysis therefore uses the timing of events rather than only the final event count.
Why is Kaplan-Meier estimation appropriate?
The registry definition explicitly uses Kaplan-Meier estimation for median survival. Kaplan-Meier methods allow patients who have not experienced the event by the end of their observed follow-up to contribute information without treating them as if they had experienced the event.
What does censoring mean in AVAGAST?
Under the registered definition, a patient for whom no death was captured on the clinical database was censored at the most recent date they were known to be alive. Censoring means that the available observation period ends without a captured death event; it does not mean that the patient died at that time.
What does "up to 26 months" mean?
The registered time frame is "From randomization until death, up to 26 months." This describes the registry-defined observation window for the endpoint. It should not be interpreted as a statement that every participant was followed for exactly 26 months.
Why can the median survival be estimated from a Kaplan-Meier curve?
The median survival is the time at which the estimated survival probability reaches 0.50, when that point is estimable. Kaplan-Meier estimation provides a survival function that accounts for the timing of deaths and censoring rather than requiring complete follow-up for every participant.
Why should a hazard ratio not be added when the registry does not report one?
A hazard ratio would be a model-based estimate requiring a specified comparison and statistical model. Although hazard ratios are commonly used for randomized time-to-event trials, the registry-reported AVAGAST data do not report one. Adding an estimate from another source would violate the scope of this registry-based analysis.
7. Planned Analysis
ClinicalTrials.gov reports results for the overall-survival endpoint, but the ClinicalTrials.gov record contains no formal statistical analyses with estimates, confidence intervals, or p-values.
What would normally be reported for this endpoint?
For a randomized phase 3 trial with overall survival as the primary time-to-event endpoint, a complete statistical report would typically describe the estimated survival curves, median survival where estimable, a comparison between randomized groups, an effect measure such as a hazard ratio with a confidence interval, and an associated hypothesis test when prespecified by the trial's statistical analysis plan.
Those elements are methodological expectations rather than AVAGAST results in the ClinicalTrials.gov record. The distinction is important: a method can be appropriate for an endpoint without being evidence that the trial actually used that method or produced a particular result.
8. Safety
The ClinicalTrials.gov record reports serious adverse events by randomized arm as affected patients over patients at risk. These figures provide an arm-level safety count but do not include event-specific definitions, severity categories, exposure durations, or formal comparative statistical analyses.
| Safety measure | Bevacizumab | Placebo |
|---|---|---|
| Serious adverse events | 134 / 386 | 137 / 381 |
The numerator is the number of affected patients and the denominator is the number at risk as reported in the ClinicalTrials.gov record. The registry extract does not provide a formal confidence interval or p-value for the difference, so no inferential conclusion about a treatment effect on serious adverse events is drawn here.
9. Clinical Biostats Interpretation of the Primary Endpoint
Overall survival measures the time from randomization to death irrespective of the cause of death. This makes it a direct time-to-event endpoint: both the occurrence of death and its timing contribute to the analysis.
The ClinicalTrials.gov record establishes that overall survival was the single primary endpoint, that it was defined from randomization until death, that patients without a captured death were censored at their most recent known-alive date, and that median survival was estimated by Kaplan-Meier methodology.
The ClinicalTrials.gov record does not establish a numerical treatment effect, confidence interval, p-value, or formal between-arm statistical test. Therefore, no statement about the magnitude or statistical significance of a treatment difference is made from the registry record.
If a formal effect estimate were reported, its confidence interval would describe statistical uncertainty around that estimate under the specified model and sampling framework. A confidence interval would not describe the range of individual patient outcomes.
A p-value addresses evidence against a specified null hypothesis under the statistical model. It does not describe how large or clinically important an effect is. Effect estimates and confidence intervals are needed to characterize magnitude and precision.
10. Limitations
- No formal statistical analyses reported: no formal statistical analyses were posted, so numerical efficacy estimates, confidence intervals, and p-values cannot be reported from the registry record.
- Registry scope: the ClinicalTrials.gov record identifies the primary endpoint and its definition but does not provide the full statistical analysis plan.
- Limited safety detail: serious adverse events are reported only as affected patients over patients at risk by arm. Event-level descriptions and formal comparisons are not reported.
- No subgroup analysis data: the ClinicalTrials.gov recordset does not provide subgroup estimates or interaction tests.
- No baseline table: baseline demographic or disease-characteristic values are not included in the ClinicalTrials.gov record.
- No additional endpoint definitions: the registry extract identifies one primary endpoint and does not provide definitions for secondary endpoints.
- No reported proportional-hazards analysis: because no Cox model or hazard ratio is reported, proportional-hazards assumptions cannot be evaluated from the registry record.
- No reported interim-analysis strategy: the ClinicalTrials.gov record does not establish whether an interim efficacy analysis, alpha spending, or another sequential-testing procedure was used.
11. Why This Trial Matters Statistically
AVAGAST is a useful statistical teaching example because the ClinicalTrials.gov record illustrates a central distinction in clinical-trial analysis: an endpoint definition is not the same thing as a reported treatment effect.
| Concept | How it appears in AVAGAST |
|---|---|
| Randomization | The trial is registered as randomized with 2 arms. |
| Blinding | The trial is registered as double masked. |
| Parallel design | The registered design model is parallel. |
| Time-to-event analysis | Overall survival is the primary endpoint type. |
| Kaplan-Meier estimation | The registry explicitly states that median survival was estimated by Kaplan-Meier methodology. |
| Censoring | Patients without a captured death were censored at the most recent date they were known to be alive. |
| Primary endpoint discipline | One primary endpoint is registered: overall survival. |
| Safety analysis | Serious adverse events are reported as affected patients over patients at risk by arm. |
| Evidence versus inference | The ClinicalTrials.gov record contains no formal statistical-analysis estimates, confidence intervals, or p-values. |
The most important statistical lesson is therefore methodological restraint.
12. Trial Timeline
Trial start
The registry lists September 2007 as the trial start.
Primary completion
The registry lists November 2009 as the primary completion date.
Registry status
The ClinicalTrials.gov record lists the trial status as completed.
13. Statistical Interpretation of Censoring
Censoring is one of the most important concepts in this trial's primary endpoint. The registered definition states that patients for whom no death was captured on the clinical database were censored at the most recent date they were known to be alive.
What is observed
For a patient who dies during the observation period, the analysis uses the time from randomization to the recorded date of death.
What is censored
For a patient without a captured death, the analysis uses the patient's observed follow-up through the most recent date they were known to be alive.
Why this matters
Simply excluding patients without a recorded death would discard useful follow-up information. Time-to-event methods preserve that information through censoring.
What censoring is not
A censoring date is not a death date. It marks the end of the observed event-free period under the registry definition.
14. Randomization and Blinding
The trial is registered as randomized and double masked. Randomization is important because it provides the basis for comparing outcomes between treatment assignments while reducing systematic differences in prognostic characteristics in expectation.
Double masking is likewise a design feature intended to reduce the influence of knowledge of treatment assignment. The ClinicalTrials.gov record does not provide further details about exactly which participants or study personnel were masked, so no additional masking structure is inferred.
The strength of a randomized comparison comes from the treatment assignment mechanism, while the time-to-event analysis addresses the timing of outcomes and censoring.
15. Analysis Population and Denominators
The registry-reported enrollment figure is 774. The serious-adverse-event data use separate arm-level denominators of 386 for bevacizumab and 381 for placebo.
| Data element | Value |
|---|---|
| Total enrollment | 774 |
| Number of randomized arms | 2 |
| Serious AE denominator, bevacizumab | 386 |
| Serious AE denominator, placebo | 381 |
| Serious AE affected, bevacizumab | 134 |
| Serious AE affected, placebo | 137 |
These denominators should not automatically be treated as interchangeable analysis populations. The ClinicalTrials.gov record does not state why the safety denominators differ from one another or from total enrollment, and therefore no additional population definition is assigned to them.
16. Secondary and Other Outcomes
The ClinicalTrials.gov record states that 8 outcome measures were posted. However, the ClinicalTrials.gov record does not supply their individual definitions, time frames, estimates, or statistical analyses.
17. Multiplicity, Interim Analysis, and Advanced Methods
The ClinicalTrials.gov record does not report a non-inferiority margin, crossover procedure, factorial structure, multiplicity strategy, interim-analysis plan, missing-data or imputation method, stratification factors, or Bayesian methodology.
| Design topic | Status in reported AVAGAST data |
|---|---|
| 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 parallel |
| Multiplicity adjustment | Not reported in the ClinicalTrials.gov record |
| Interim analysis | Not reported in the ClinicalTrials.gov record |
| Missing-data / imputation method | Not reported in the ClinicalTrials.gov record |
| Stratification | Not reported in the ClinicalTrials.gov record |
| Bayesian methods | Not reported in the ClinicalTrials.gov record |
This distinction is important in statistical review. The absence of a method in the registry extract is not evidence that the method was not used in the full protocol or statistical analysis plan; it means only that the method is not established by the ClinicalTrials.gov record.
18. What a Complete Time-to-Event Result Would Contain
For an educational statistical analysis of a randomized overall-survival endpoint, several layers of information are useful. The registry-reported AVAGAST record provides the endpoint definition and Kaplan-Meier method but not the numerical comparative analysis.
Survival function
A Kaplan-Meier curve describes the estimated probability of remaining alive over time.
Median survival
The registry explicitly states that median survival was estimated by Kaplan-Meier methodology.
Relative effect
A formal report may include a hazard ratio, but no AVAGAST hazard ratio is reported in the ClinicalTrials.gov record for this page.
Precision
A confidence interval would quantify uncertainty around a reported effect estimate, but no AVAGAST confidence interval is reported here.
Keeping these components separate prevents a common statistical error: treating a single summary measure as though it completely describes the survival experience of both randomized groups.
19. Important Statistical Interpretation Issues
Hazard ratios require a model
A hazard ratio is not simply a ratio of the numbers of deaths. It is an effect measure produced by a time-to-event model. Because the ClinicalTrials.gov record does not report a Cox model or another hazard-ratio analysis, a hazard ratio is not presented for AVAGAST here.
Kaplan-Meier curves are not ordinary proportions
A Kaplan-Meier survival estimate at a particular time incorporates the event and censoring history up to that point. It should not be interpreted as a simple percentage calculated by dividing deaths by enrollment.
Median survival is not the same as mean survival
The median is a percentile-based measure of the survival distribution. It is not an arithmetic average of observed survival times and can remain estimable even when follow-up is incomplete.
Censoring requires assumptions
Time-to-event methods rely on assumptions about the censoring process. The ClinicalTrials.gov record specifies the censoring rule but do not provide enough information to evaluate the assumptions empirically.
20. Why the Absence of a Reported P-Value Matters
The ClinicalTrials.gov record does not contain a p-value for the primary endpoint. It would therefore be inappropriate to describe AVAGAST as statistically significant or nonsignificant based on the registry record alone.
A p-value is meaningful only in relation to a defined statistical hypothesis, analysis population, test statistic, and testing framework. Without a reported analysis, inserting a p-value from memory or from an unspecified external source would make the page appear more complete while making its evidentiary basis less transparent.
The same principle applies to hazard ratios, confidence intervals, median survival estimates, subgroup results, and other numerical outcomes: they should be reported only when they are contained in the registry-reported trial evidence for this page.
21. Why This Trial Matters for Clinical-Trial Statistics
AVAGAST illustrates several foundational principles that recur across phase 3 clinical trials.
| Statistical principle | AVAGAST example |
|---|---|
| Define the estimand through the endpoint | Overall survival is explicitly defined from randomization to death. |
| Preserve timing information | The endpoint is analyzed as a time-to-event outcome rather than a simple binary outcome. |
| Account for incomplete observation | Patients without captured death are censored at the most recent date known alive. |
| Use an appropriate estimator | Kaplan-Meier methodology is explicitly identified for median survival. |
| Separate descriptive from inferential statistics | The ClinicalTrials.gov record provides endpoint methodology but no formal comparative estimates. |
| Respect analysis populations | Safety denominators are separately registry-reported and should not automatically be treated as identical to enrollment. |
| Do not infer unreported methods | Stratification, Cox modeling, multiplicity, interim analysis, and Bayesian methods are not established by the ClinicalTrials.gov record. |
22. Related Tutorials
Learn more about the methods used in this trial:
The trial's statistical concepts contains two explicit tutorial concepts: Blinding and Randomization. Additional tutorial concepts are not added here because they are not specified in the trial data.
23. Related Calculators
24. Sources
- ClinicalTrials.gov: NCT00548548 — AVAGAST.
- Linked publication: PubMed PMID 24942210.
- Linked publication: PubMed PMID 21844504.
Continue through Clinical Biostats
Use the related tutorials and statistical tools to deepen your understanding of randomized clinical-trial design and time-to-event analysis.
25. Record Summary
AVAGAST is registered as a completed, randomized, double-masked, parallel phase 3 trial with 774 enrolled participants and 2 arms. Its single registered primary endpoint is overall survival, defined as the time from randomization to death irrespective of cause, with patients without a captured death censored at their most recent date known alive. The registry specifies Kaplan-Meier estimation for median survival.
The ClinicalTrials.gov record also reports serious adverse events by arm: 134 of 386 patients for bevacizumab and 137 of 381 patients for placebo. Because no formal statistical analyses were posted, this page deliberately does not add hazard ratios, confidence intervals, p-values, subgroup findings, or other numerical efficacy results that are not present in the ClinicalTrials.gov record.