This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the information contained in the ClinicalTrials.gov record.
1. Trial at a Glance
AUGMENT was a randomized, parallel-group, quadruple-masked phase 3 trial evaluating rituximab plus lenalidomide (R2) versus rituximab plus placebo in relapsed or refractory indolent non-Hodgkin's lymphoma, including follicular lymphoma and marginal zone lymphoma.
| Feature | AUGMENT |
|---|---|
| Trial name | AUGMENT |
| NCT identifier | NCT01938001 |
| Phase | Phase 3 |
| Status | COMPLETED |
| Therapeutic area | Hematology |
| Condition | Lymphoma, Non-Hodgkin |
| Enrollment | 358 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Lead sponsor | Celgene |
| Sponsor type | Industry |
| Study start | November 21, 2013 |
| Primary completion | June 22, 2018 |
2. Clinical Question
The central statistical question was whether adding lenalidomide to rituximab improves progression-free survival compared with rituximab plus placebo in the trial population. The registered primary endpoint was a time-to-event measure assessed by an Independent Review Committee using the 2007 International Working Group Response Criteria.
Population
Patients with relapsed or refractory indolent non-Hodgkin's lymphoma, with the trial description specifying follicular lymphoma and marginal zone lymphoma.
Intervention
Rituximab plus lenalidomide, referred to in the registry analysis as R2.
Comparator
Rituximab plus placebo.
Primary question
Does R2 improve the time from randomization to disease progression or death compared with rituximab plus placebo?
3. Trial Design
Participants were allocated to 2 parallel treatment groups.
The registry identifies the study as quadruple-masked.
The registered primary purpose was treatment.
The posted primary and secondary analyses use superiority hypotheses.
Rituximab + Lenalidomide
- Rituximab
- Lenalidomide
- 176 participants were at risk for the serious-adverse-event summary
- 45 participants were affected by a serious adverse event in that summary
Rituximab + Placebo
- Rituximab
- Placebo
- 180 participants were at risk for the serious-adverse-event summary
- 25 participants were affected by a serious adverse event in that summary
4. Randomization, Stratification, and Analysis Population
The posted analyses identify three stratification factors: previous rituximab treatment, time since last antilymphoma therapy (≤ 2 versus > 2 years), and disease histology (follicular lymphoma versus marginal zone lymphoma). These factors were used in the stratified analyses of the time-to-event endpoints and in the Cochran-Mantel-Haenszel analyses of binary endpoints.
| Analysis population | Definition / role |
|---|---|
| Intention-to-treat | All participants randomized into the trial, regardless of whether they received study treatment or not. |
| Duration of objective response | Participants who achieved an objective response in the ITT population. |
| Duration of complete response | Participants who achieved a complete response in the ITT population. |
5. Endpoints
The registry identifies one primary endpoint and multiple secondary endpoints. The primary endpoint is a time-to-event measure based on IRC assessment. Several secondary endpoints use the same general survival-analysis framework, while response-rate endpoints are binary and use stratified Cochran-Mantel-Haenszel testing.
| Endpoint | Registered definition / time frame | Type |
|---|---|---|
| Primary: Kaplan Meier Estimate of Progression Free Survival Assessed by the Independent Review Committee (IRC) According to the 2007 International Working Group Response Criteria (IWGRC) | From randomization of study drug up to disease progression or death, which occurred first; up to the data cut-off date of 22 June 2018; overall median follow-up time for all participants was 28.30 months (range: 0.1 to 51.3 months). | Time-to-event |
| Overall Survival (OS) | From date of randomization to death due to any cause (Average of 55.71 months and a maximum up to 95.2 months) | Time-to-event |
| Duration of Objective Response | From randomization up to data cut-off date of 22 June 2018; overall median follow-up time for all participants was 28.30 months (range: 0.1 to 51.3 months). | Time-to-event |
| Duration of Complete Response (DOCR) | From randomization up to data cut-off date of 22 June 2018; overall median follow-up time for all participants was 28.30 months (range: 0.1 to 51.3 months). | Time-to-event |
| Event Free Survival | From date of randomization to data cut-off date of 22 June 2018; overall median follow-up time for all participants was 28.30 months (range: 0.1 to 51.3 months). | Time-to-event |
| Time to Next Anti-Lymphoma Treatment (TTNLT) | From date of randomization to date of first documented administration of a new anti-lymphoma treatment (Average of 55.71 months and a maximum up to 95.2 months) | Time-to-event |
| Durable Complete Response Rate (DCCR) | From first dose of investigational product (IP) to data cut-off date of 22 June 2018; the median treatment duration was 11.19 months in the rituximab/lenalidomiade arm and 11.04 months in the rituximab/placebo arm | Binary |
| Percentage of Participants With an Objective Response | From date of first dose to data cut-off date of 22 June 2018; the median treatment duration was 11.19 months in the rituximab/lenalidomide arm and 11.04 months in the rituximab/placebo arm | Binary |
| Percentage of Participants With a Best Response of Complete Response | From date of first dose up to data cut-off date of 22 June 2018; the median treatment duration was 11.19 months in the rituximab/lenalidomide arm and 11.04 months in the rituximab/placebo arm | Binary |
Primary endpoint definition
The registry defines progression-free survival as the time from the date of randomization into the study to the first observation of documented disease progression or death due to any cause, whichever occurred first. PFS was based on IRC review using the modified 2007 International Working Group Response Criteria and FDA censoring rules.
6. Primary Result: Progression-Free Survival
The primary endpoint was analyzed in the ITT population using a log-rank test. The posted analysis also states that the comparison was stratified by previous rituximab treatment, time since last antilymphoma therapy, and disease histology. The hazard ratio and its confidence interval were estimated from a Cox proportional-hazards model adjusting for these stratification factors.
Progression-free survival hazard ratio
95% CI: 0.34–0.62 · P < 0.0001
Two-sided confidence interval · Superiority hypothesis
| Primary endpoint | R2 vs rituximab + placebo |
|---|---|
| Effect measure | Hazard ratio |
| Estimate | 0.46 |
| 95% confidence interval | 0.34–0.62 |
| P-value | < 0.0001 |
| Analysis method | Log-rank test |
| Model for HR and CI | Cox proportional-hazards model adjusted for the three stratification factors |
| Analysis population | ITT |
What the estimate means
A hazard ratio of 0.46 means that the estimated instantaneous rate of progression or death was 46% of the corresponding rate in the rituximab-plus-placebo group under the fitted Cox model. Expressed as a simple relative interpretation, this corresponds to an estimated 54% lower hazard for progression or death.
The estimate is a relative time-to-event measure. It does not mean that 54% of participants avoided progression, that 54% of participants were cured, or that every individual participant experienced the same reduction in risk.
What the confidence interval says
The two-sided 95% confidence interval of 0.34–0.62 describes the statistical uncertainty around the estimated hazard ratio under the analysis framework. It is not a range containing the individual treatment effects experienced by patients.
Why the p-value is different from the effect size
The p-value of < 0.0001 addresses evidence against the null hypothesis within the specified statistical test. It does not measure the magnitude of the treatment effect. The magnitude is described by the hazard ratio and its confidence interval.
Important survival-analysis cautions
The hazard ratio is model-based, and its interpretation depends on the Cox proportional-hazards framework. The primary analysis also uses censoring rules for a time-to-event endpoint, so the estimate reflects both observed events and the handling of participants who were censored. The analysis was performed in the ITT population and adjusted for the three prespecified stratification factors identified in the registry analysis.
7. Secondary Time-to-Event Results
The registry also reports statistical comparisons for overall survival, duration of objective response, duration of complete response, event-free survival, and time to next anti-lymphoma treatment. These analyses illustrate how the same broad survival-analysis framework can be applied to different definitions of an event.
Overall Survival
Overall survival hazard ratio
95% CI: 0.37–0.95
Two-sided confidence interval · Superiority hypothesis
The OS analysis used the ITT population. The registry identifies the comparison as stratified by previous rituximab treatment, time since last antilymphoma therapy, and disease histology. The posted data do not identify a formal statistical method for the OS analysis, so the page does not assign one beyond the registry's reported hazard-ratio result.
The reported hazard ratio of 0.59 corresponds to an estimated death hazard of 59% of that in the comparator group under the reported analysis framework. The 95% CI of 0.37–0.95 describes uncertainty around that estimate. The registry does not provide a p-value for this secondary OS analysis in the ClinicalTrials.gov record, so no p-value is inferred from the confidence interval.
Duration of Objective Response
Hazard ratio for duration of objective response
95% CI: 0.36–0.79 · P = 0.0015
This analysis was restricted to participants who achieved an objective response in the ITT population. The registry reports a log-rank analysis and a hazard ratio of 0.53 with a two-sided 95% CI of 0.36–0.79 and P = 0.0015.
Duration of Complete Response
Hazard ratio for duration of complete response
95% CI: 0.32–1.43 · P = 0.2993
This analysis was restricted to participants who achieved a complete response in the ITT population. The registry reports a log-rank analysis. The confidence interval is relatively wide, spanning values below and above 1, so the point estimate should not be interpreted independently of its uncertainty.
Event-Free Survival
Hazard ratio for event-free survival
95% CI: 0.38–0.67 · P < 0.0001
The event-free-survival analysis used the ITT population and a stratified log-rank test. The hazard ratio and confidence interval were estimated from a Cox proportional-hazards model adjusting for the three stratification factors.
Time to Next Anti-Lymphoma Treatment
Hazard ratio for time to next treatment
95% CI: 0.39–0.71 · P < 0.0001
Time to next anti-lymphoma treatment was defined from randomization to the first documented administration of a new anti-lymphoma treatment. The posted analysis used a stratified log-rank test and the ITT population.
| Secondary endpoint | Method | HR | 95% CI | P-value |
|---|---|---|---|---|
| Overall Survival | Not reported | 0.59 | 0.37–0.95 | Not reported |
| Duration of Objective Response | Log-rank test | 0.53 | 0.36–0.79 | 0.0015 |
| Duration of Complete Response | Log-rank test | 0.68 | 0.32–1.43 | 0.2993 |
| Event-Free Survival | Stratified log-rank test | 0.51 | 0.38–0.67 | < 0.0001 |
| Time to Next Anti-Lymphoma Treatment | Stratified log-rank test | 0.53 | 0.39–0.71 | < 0.0001 |
8. Secondary Binary Endpoint Results
The registry also reports three binary response outcomes. These analyses used the Cochran-Mantel-Haenszel test, stratified by previous rituximab treatment, time since last antilymphoma therapy, and disease histology.
| Endpoint | Method | P-value | Analysis population |
|---|---|---|---|
| Durable Complete Response Rate (DCCR) | Cochran-Mantel-Haenszel test | 0.0006 | ITT |
| Percentage of Participants With an Objective Response | Cochran-Mantel-Haenszel test | < 0.0001 | ITT |
| Percentage of Participants With a Best Response of Complete Response | Cochran-Mantel-Haenszel test | = 0.0010 | ITT |
The Cochran-Mantel-Haenszel framework is useful when a binary outcome is compared across randomized treatment groups while accounting for prespecified strata. Its p-value addresses the evidence for a difference after stratification; it does not itself describe the magnitude of that difference. For magnitude, an effect estimate such as a risk difference, risk ratio, or odds ratio would normally be examined alongside its confidence interval when available.
9. Statistical Methodology
Kaplan-Meier estimation
The primary endpoint is explicitly a Kaplan-Meier estimate of progression-free survival. Kaplan-Meier estimation is designed for time-to-event data in which some participants may be censored before experiencing the event. It estimates the probability of remaining event-free over time using the observed event and censoring information.
where di is the number of events at time ti and ni is the number at risk immediately before that event time.
The Kaplan-Meier method does not require every participant to have an observed event. A participant who remains free of progression or death through the last usable follow-up can contribute information until censoring.
Log-rank test
The primary PFS comparison was reported as a log-rank analysis. The log-rank test compares the observed and expected numbers of events between treatment groups across event times. It is particularly suited to randomized comparisons of time-to-event outcomes.
For AUGMENT, the log-rank framework is paired with a hazard-ratio estimate from a Cox proportional-hazards model. The test and the effect estimate therefore answer related but distinct questions: the test addresses evidence against equality of survival experience, while the hazard ratio quantifies the relative event rate under the fitted model.
Stratified log-rank test
Event-free survival and time to next anti-lymphoma treatment were analyzed using a stratified log-rank test. The registry-reported analysis notes identify the same three stratification factors used for the trial's stratified analyses: previous rituximab treatment, time since last antilymphoma therapy, and disease histology.
Stratification allows the comparison to account for the predefined strata rather than treating all participants as if they belonged to a single homogeneous risk set. This is especially relevant when the randomization or analysis framework incorporates prognostic or design factors.
Cox proportional-hazards model
For the primary PFS endpoint, the registry states that the hazard ratio and its confidence interval were estimated from a Cox proportional-hazards model adjusting for the three stratification factors.
A hazard ratio is a relative time-to-event measure. It is not equivalent to a relative risk, an absolute risk difference, or a percentage of participants who benefit.
Cochran-Mantel-Haenszel test
The three binary response endpoints were analyzed using the Cochran-Mantel-Haenszel test. This method provides a stratified comparison of categorical outcomes while accounting for specified strata.
In AUGMENT, the strata were previous rituximab treatment, time since last antilymphoma therapy, and disease histology. The method therefore complements the stratified survival analyses by applying stratification to binary response outcomes.
10. Statistical Methods Explained
Why was Kaplan-Meier estimation used?
Progression-free survival, overall survival, duration of response, event-free survival, and time to next anti-lymphoma treatment are all time-to-event concepts. Participants can be followed for different lengths of time, and some may be censored before an event occurs. Kaplan-Meier estimation provides a way to use the available follow-up without treating every censored participant as if an event had occurred.
What does a PFS hazard ratio of 0.46 mean?
It means the fitted model estimated the instantaneous rate of progression or death in the R2 group to be 0.46 times that of the rituximab-plus-placebo group. The simple complement, 1 − 0.46 = 0.54, corresponds to a 54% lower estimated hazard. This is not the same as saying that 54% of patients benefited or that time to progression was exactly 54% longer.
Why was the analysis stratified?
The registry identifies three factors used for stratification: previous rituximab treatment, time since last antilymphoma therapy, and disease histology. Stratification incorporates these predefined factors into the comparison, helping the analysis reflect the trial's randomized structure and the characteristics used in the analysis plan.
What is the difference between the log-rank test and the Cox model?
The log-rank test is primarily a hypothesis test comparing time-to-event distributions between groups. The Cox model provides an estimated hazard ratio and confidence interval. In the primary PFS analysis, AUGMENT used the log-rank test for the comparison while estimating the hazard ratio and its confidence interval from a Cox proportional-hazards model adjusted for the stratification factors.
Why use the Cochran-Mantel-Haenszel test for response endpoints?
Response outcomes such as objective response and complete response are binary rather than continuous time-to-event outcomes. A Cochran-Mantel-Haenszel analysis allows the treatment groups to be compared while accounting for the trial's specified strata. Its p-value does not replace the need for an effect estimate when the magnitude of a response difference is being interpreted.
Why is the ITT population important?
The ITT population includes all participants randomized into the trial, regardless of whether they received study treatment. Using ITT preserves the randomized comparison for efficacy analysis and avoids redefining the treatment groups after randomization based on treatment exposure or adherence.
What does a confidence interval add to a hazard ratio?
The hazard ratio provides a point estimate, while the confidence interval describes uncertainty around that estimate. For the primary PFS analysis, the HR was 0.46 and the two-sided 95% CI was 0.34–0.62. The interval shows that the statistical estimate is not known with arbitrary precision. It should be considered alongside the point estimate rather than treated as a range of individual patient outcomes.
11. Safety Results
The ClinicalTrials.gov record provides a serious-adverse-event summary by randomized arm. In the R2 group, 45 of 176 participants were affected; in the rituximab-plus-placebo group, 25 of 180 participants were affected.
| Safety measure | Rituximab + Lenalidomide (R2) | Rituximab + Placebo |
|---|---|---|
| Serious adverse events | 45 / 176 | 25 / 180 |
These are reported as affected participants over participants at risk in the ClinicalTrials.gov record. The data provided do not include a formal statistical comparison, confidence interval, or p-value for this safety measure, so no such comparison is inferred.
12. Stratification and Covariate Adjustment
AUGMENT provides a useful example of how stratification in a randomized trial can carry through into statistical analysis. The posted analyses identify three factors:
| Stratification factor | Categories reported in the registry analysis |
|---|---|
| Previous rituximab treatment | Previous rituximab treatment |
| Time since last antilymphoma therapy | ≤ 2 years; > 2 years |
| Disease histology | FL; MZL |
For the primary PFS analysis, these factors were used in the Cox model that estimated the hazard ratio and confidence interval. They were also used for the stratified survival and Cochran-Mantel-Haenszel analyses reported for secondary endpoints.
Stratification is not subgroup ranking
Using a factor for stratification does not mean the trial was separately powered to establish a different treatment effect within every category.
Adjustment preserves the design
Including the stratification factors in the primary Cox model makes the reported hazard ratio conditional on the analysis framework specified for the randomized comparison.
13. Interpreting the Primary Result Correctly
The primary PFS HR of 0.46 is a relative measure of the estimated event rate. Its most direct interpretation is that the estimated instantaneous rate of progression or death was 46% of the comparator rate under the fitted model.
The registry analysis does not provide treatment-group median PFS, fixed-time PFS probabilities, or an absolute difference in survival probability. Those quantities are therefore not reported or reconstructed on this page.
The two-sided 95% CI of 0.34–0.62 and P-value of < 0.0001 provide information about statistical uncertainty and evidence under the specified analysis. The p-value does not tell us whether the effect is clinically important, nor does it quantify the size of the effect.
A hazard ratio does not establish that every participant benefits to the same degree, does not directly give a probability of individual benefit, and does not by itself describe the absolute difference in survival at a particular time point.
14. Limitations
- Summary-data limitation: The registry analysis provides selected effect estimates and p-values but not all possible treatment-group summaries. In particular, the binary response analyses do not provide response percentages or confidence intervals in the ClinicalTrials.gov record.
- Hazard-ratio interpretation: The primary HR comes from a Cox proportional-hazards model. Its interpretation therefore depends on the model framework and should not automatically be treated as a constant ratio of cumulative risks at every time point.
- Censoring: PFS uses FDA censoring rules according to the registry definition. Censoring is therefore part of the estimand and statistical construction of the reported PFS result.
- Analysis population: The primary efficacy analysis is based on the ITT population, while some duration-of-response endpoints are restricted to participants who achieved the relevant response. These populations answer different questions.
- Secondary endpoints: Several secondary analyses have different endpoint definitions and analysis populations. Their results should not be treated as interchangeable measures of the same outcome.
- OS method reporting: The ClinicalTrials.gov record reports an OS hazard ratio and stratification factors but identify the formal method as not reported. A specific statistical method is therefore not assigned to that analysis beyond what the registry data state.
- Safety comparison: Serious adverse events are reported as affected participants over participants at risk, without a registry-reported formal treatment comparison. No p-value or confidence interval is inferred.
15. Why This Trial Matters Statistically
AUGMENT is a useful teaching case because its registry results connect several core clinical-trial methods within a single randomized phase 3 study. The primary endpoint combines Kaplan-Meier estimation, a log-rank comparison, and Cox-model estimation of a hazard ratio, while secondary response outcomes demonstrate a different approach for binary data.
| Concept | How it appears in AUGMENT |
|---|---|
| Randomization | 358 participants were enrolled in a randomized, parallel-group phase 3 trial with 2 arms. |
| Blinding | The registry identifies the trial as quadruple-masked. |
| ITT analysis | The primary PFS analysis used all randomized participants regardless of whether study treatment was received. |
| Kaplan-Meier estimation | The primary endpoint and multiple secondary endpoints are Kaplan-Meier time-to-event measures. |
| Log-rank testing | The primary PFS comparison used a log-rank test. |
| Stratified log-rank testing | Event-free survival and time to next anti-lymphoma treatment used stratified log-rank tests. |
| Hazard ratio | The primary PFS analysis reported HR 0.46 with a 95% CI of 0.34–0.62. |
| Cox proportional-hazards model | The primary HR and confidence interval were estimated with adjustment for the three stratification factors. |
| Stratified analysis | Previous rituximab treatment, time since last antilymphoma therapy, and disease histology were used as stratification factors. |
| Cochran-Mantel-Haenszel test | DCCR, objective response, and complete response were analyzed using this stratified categorical-data method. |
| Confidence intervals | Two-sided 95% confidence intervals were reported for the time-to-event hazard ratios. |
| P-values | Formal p-values were reported for the primary PFS and several secondary analyses. |
16. Overall Statistical Reading of AUGMENT
The primary PFS result is internally coherent as a time-to-event analysis: the registry defines PFS from randomization to progression or death, identifies the ITT population, reports a log-rank comparison, and describes a Cox model adjusted for the three stratification factors. The resulting hazard ratio is 0.46, with a two-sided 95% CI of 0.34–0.62 and P < 0.0001.
The secondary results extend the statistical picture across several different endpoint definitions. OS has a reported HR of 0.59 with a 95% CI of 0.37–0.95. Duration of objective response, event-free survival, and time to next anti-lymphoma treatment have reported hazard ratios of 0.53, 0.51, and 0.53, respectively. Duration of complete response has a reported HR of 0.68 with a wider 95% CI of 0.32–1.43.
The binary response analyses use a different statistical framework. DCCR, objective response, and best complete response are analyzed using the Cochran-Mantel-Haenszel test, preserving the trial's stratification structure for categorical outcomes.
The most important statistical lesson is that these results should be read as a set of estimands and analyses, not as a collection of interchangeable p-values. PFS, OS, response duration, event-free survival, time to next treatment, and binary response each define a different outcome. The statistical method needs to match that outcome, and interpretation should remain tied to the population, endpoint definition, censoring rules, stratification, and model used for each analysis.
17. Related Tutorials
Learn more about the methods used in this trial:
18. Related Calculators
19. Sources
- ClinicalTrials.gov: AUGMENT, NCT01938001.
- PubMed: PMID 30897038.
- PubMed: PMID 41990300.
Continue with the statistical methods behind AUGMENT
Explore the survival-analysis, stratification, categorical-data, and clinical-trial concepts that appear in the AUGMENT analysis.
20. Record Summary
AUGMENT is a randomized phase 3, quadruple-masked, parallel-group trial with 358 enrolled participants and a primary time-to-event endpoint of progression-free survival assessed by an Independent Review Committee. The primary analysis used the ITT population, a log-rank test, and a Cox proportional-hazards model adjusted for previous rituximab treatment, time since last antilymphoma therapy, and disease histology.
The primary PFS hazard ratio was 0.46 with a two-sided 95% confidence interval of 0.34–0.62 and P < 0.0001. Secondary analyses extended the time-to-event framework to overall survival, duration of response, duration of complete response, event-free survival, and time to next anti-lymphoma treatment, while binary response outcomes were analyzed with the Cochran-Mantel-Haenszel test.