Bevacizumab or Pemetrexed Disodium Alone or In Combination After Induction Therapy in Treating Patients With Advanced Non-Squamous Non-Small Cell Lung Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Bevacizumab or Pemetrexed Disodium Alone or In Combination After Induction Therapy in Treating Patients With Advanced Non-Squamous Non-Small Cell Lung Cancer (NCT01107626), a randomized phase 3 study. The primary outcome is Overall Survival.

2. Clinical Question

3. Trial Design

This is a randomized, open-label, parallel-group phase III study. Its stated focus is maintenance treatment after induction therapy for advanced non-squamous non-small cell lung cancer. The three maintenance strategies are bevacizumab, pemetrexed, and bevacizumab plus pemetrexed. The registry lists enrollment of 1516, but the primary endpoint analysis is described as including patients randomized to maintenance therapy; patients who received induction therapy only and did not take part in the randomized portion are excluded from that analysis. Enrollment should therefore not be read as the number analyzed in each maintenance comparison.

The arm descriptions indicate a shared induction regimen of paclitaxel, carboplatin, and bevacizumab. They describe repeated courses every 21 days for 4 courses, subject to progression or unacceptable toxicity. The maintenance descriptions are truncated in the registry record, so further treatment details are not reported in the registry record. Randomization allocation ratio and the method used to generate or conceal the allocation are not reported in the registry record.

The primary outcome is overall survival. The registry describes assessment every 6 weeks during induction and every 3 cycles during maintenance; after study therapy ends, assessment is every 3 months for 2 years and every 6 months for years 3-5. The start date is 2010-10-25, primary completion is 2019-01-31, and the status is completed. The registry does not report a sample-size justification or further details of the statistical design.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Step 1: Induction Therapy

Reasons for not completing this period:

Step 2: Maintenance Therapy

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized patients

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: All registered patients

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival, measured over Assessed every 6 weeks during induction therapy and every 3 cycles during maintenance therapy; after discontinuation of study therapy, assessed every 3 months for 2 years and every 6 months for years 3-5.

Overall survival measures the time from randomization until death or the date a participant was last known to be alive. It is a time-to-event outcome: the analysis uses both the event and the time at which it occurred, while accounting for participants whose follow-up ends before death is recorded. The registry specifies median overall survival in months and 97.5% confidence intervals.

Overall survival is used because it captures survival duration directly and does not depend on a judgment about whether a tumor has progressed or responded. It is also clinically interpretable as a time-to-event measure. However, a median describes the time by which half of the analyzed group has experienced the event under the survival analysis; it does not describe the outcome for every participant or indicate how survival differs for each individual.

The registry gives the assessment schedule and defines the analysis population as patients randomized to maintenance, excluding those who received induction therapy only and did not enter randomization. It does not report details of censoring rules beyond the definition of the endpoint, nor does it provide an endpoint-specific missing-data procedure. Those details are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is time-to-event data, not a simple count of deaths at one fixed time. Each participant contributes information about how long they were followed and whether death was recorded during that period. Participants still alive at their last known follow-up contribute information up to that point; they are generally treated as censored in survival methods. The registry defines the endpoint using death or the date last known alive, but additional censoring rules are not reported in the registry record.

What does a median survival estimate mean?

A median survival estimate is the time at which the estimated survival curve reaches the halfway point. It summarizes the middle of the survival-time distribution rather than the average survival time or an individual participant's expected outcome. It does not show the full shape of the survival curve. Two groups can have similar medians but different survival patterns at other times, or the reverse. The posted median estimates therefore provide a concise summary, while the log-rank tests use time-to-event information across follow-up.

The registry reports 97.5% confidence intervals for the arm-specific medians. In general, a confidence interval represents the statistical uncertainty around an estimate under the analysis method: intervals that are wider indicate less precision. It is not a range that contains every participant's survival time, and overlap or non-overlap of separate arm-specific intervals is not itself the formal test of a between-arm difference. The registry's reported log-rank tests are the formal comparisons available for the primary endpoint.

How does the log-rank test compare groups?

In general, a log-rank test compares the observed timing of events across groups over follow-up, taking account of the risk sets at different event times. It tests whether the survival experience differs across groups under its assumptions. A p-value is a measure of how compatible the observed data are with the test's null hypothesis; it is not the probability that the null hypothesis is true, nor does it give the size of a treatment effect. The registry reports p-values for bevacizumab versus pemetrexed and bevacizumab versus the combination, but does not report hazard ratios or contrast confidence intervals for these comparisons.

A hazard ratio, if reported, would compare event rates over time under a model and would require assumptions about how those rates relate over follow-up. The record does not include a sponsor-reported hazard ratio in the posted primary results. A median comparison and a hazard ratio answer related but different questions, so one should not be inferred from the other. No recalculated time-to-event contrast is available from the posted summary statistics.

What does randomization contribute?

In general, random allocation aims to make comparison groups similar, on average, with respect to factors known and unknown at the time of assignment. It supports interpretation of between-group differences as treatment contrasts, subject to issues such as departures from the assigned strategy, missing outcome data, and the analysis population. Randomization does not remove statistical uncertainty or guarantee identical group characteristics. Allocation procedures and adherence details beyond the trial's randomized, open-label, parallel design are not reported in the registry record.

Open-label means treatment assignment was not masked. The registry lists no masking. That fact does not, by itself, establish whether outcome assessment or data analysis was blinded; such details are not reported in the registry record. Overall survival is less dependent on subjective assessment than some outcomes, but follow-up and completeness of death information still matter.

Why are the reported tests not the whole interpretation?

Statistical significance and clinical importance are distinct. A p-value can be interpreted only in the context of the effect estimate, its uncertainty, the design, and the analysis plan. Here, the posted primary results include medians, arm-specific confidence intervals, and log-rank p-values, but not an effect estimate with a confidence interval for each comparison. The registry does not identify a minimum clinically important difference, a prespecified significance threshold, or whether any multiplicity adjustment was planned. Consequently, the p-values should be read as the reported tests, not as a complete account of the size or practical importance of the possible differences.

For binary outcomes, general methods compare proportions or model the chance of an event. The response-rate entries are presented as proportions with confidence intervals, and comparison entries include risk-difference estimates and unadjusted p-values. An unadjusted comparison does not account for testing multiple outcomes or contrasts. These secondary results are therefore supportive and exploratory unless the trial's prespecified hierarchy and multiplicity approach are known; those details are not reported in the registry record.

Planning calculations based on observed effects are illustrations of what a future design might require under stated assumptions. They are not observed power and do not describe the power achieved by this completed trial. The registry does not report its original sample-size justification, and no conclusion about the original design's adequacy should be drawn from a planning illustration based on observed results.

10. Analyses Reported by the Sponsor

The primary results reported by the sponsor are arm-specific median overall survival estimates with 97.5% confidence intervals, together with log-rank tests. The reported p-value is 0.12 for bevacizumab versus pemetrexed and 0.28 for bevacizumab versus the combination. The posted results do not include a hazard ratio or a confidence interval for either between-arm effect.

A separate unadjusted recalculation of a time-to-event contrast is not available from the posted summary statistics: the median-time comparison values are missing or not estimable. The sponsor's log-rank analyses therefore remain the available inferential comparisons for this endpoint. They should be interpreted as tests of group survival experience, not as estimates of the magnitude of a difference. The record does not report a direct sponsor analysis comparing pemetrexed with the combination.

The arm-specific medians can be read alongside those tests, but they do not replace them. Nor can the p-values alone establish that the treatments are equivalent or that any observed numerical ordering represents a real difference. The registry record does not flag a discrepancy between sponsor results and a recalculated primary contrast; no such recalculated contrast is posted.

11. Multiplicity, Missing Data and Analysis Populations

The registry reports two primary log-rank comparisons: bevacizumab versus pemetrexed and bevacizumab versus the combination. It does not report whether these comparisons shared a prespecified testing hierarchy or whether the significance level was adjusted for making multiple comparisons. It also does not report an interim analysis plan. These design and analysis details are not reported in the registry record, so the listed p-values should be understood as the reported tests rather than as evidence of a documented multiplicity strategy.

Multiplicity also matters for secondary outcomes. The record lists progression-free survival and response rate as secondary outcomes and reports no sponsor analyses for the response-rate comparisons. A correction for multiple testing, a gatekeeping approach, or another prespecified strategy is not described. Without that information, unadjusted secondary tests have a greater risk of false-positive findings across a collection of comparisons and should not carry the same weight as the primary endpoint.

The primary endpoint description identifies patients randomized to maintenance as the analysis population and excludes patients who received induction therapy only without entering randomization. The population field also says all randomized patients. The detailed endpoint description clarifies the intended primary analysis population, but a more specific analysis-set definition is not reported. The registry does not describe whether the primary analysis followed intention-to-treat principles or how participants who discontinued assigned treatment were handled.

Missing-data procedures, the amount of missing survival follow-up, and reasons for missingness are not reported in the registry record. The endpoint definition refers to the date last known alive, but does not state the full censoring rules. These omissions make it difficult to assess whether incomplete follow-up could affect the estimates. The registry also does not report a protocol-specific plan for missing secondary outcome data, so the completeness and handling of those measures cannot be evaluated from the posted record.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: neutrophil count decreased (27.3%), hypertension (18.9%), white blood cell decreased (14.4%), fatigue (8.4%), lymphocyte count decreased (8.0%).

The safety summaries distinguish the induction group from the three maintenance groups. In the induction group, the registry reports 825 participants with serious adverse events among 1477 at risk and 1178 with other adverse events among 1477 at risk. For maintenance, the corresponding serious-adverse-event counts are 87 among 284 at risk for bevacizumab, 107 among 289 at risk for pemetrexed, and 144 among 286 at risk for the combination. The other-adverse-event counts are 216 among 284, 222 among 289, and 237 among 286, respectively.

These are registry counts, not estimates of a causal effect of one maintenance regimen relative to another. The induction group and maintenance groups represent different treatment periods and populations, so their counts should not be compared as if they were randomized arms observed over identical exposure. The record also reports deaths by group, but does not explain the relationship between those figures and adverse-event attribution.

The leading serious adverse-event terms listed are neutrophil count decreased (27.3%), hypertension (18.9%), white blood cell decreased (14.4%), fatigue (8.4%), lymphocyte count decreased (8.0%), and platelet count decreased (7.4%). The registry does not specify in this summary how these terms were distributed across treatment groups. These figures describe posted safety reporting; they do not by themselves establish causation or comparative risk.

13. What the Result Does, and Does Not, Mean

The results describe overall survival among patients randomized to maintenance after induction therapy. The registry reports arm-specific medians and log-rank tests for two comparisons with bevacizumab. The medians are numerically higher in the pemetrexed and combination arms than in the bevacizumab arm, while the reported tests do not provide evidence of a statistically clear difference at a conventional 0.05 threshold. The posted information does not establish that the strategies have identical effects, and it does not quantify the comparative effect with a hazard ratio or a contrast confidence interval.

These results do not demonstrate that one maintenance regimen is superior, equivalent, or inferior to another. Failure to reject a null hypothesis is not proof of equality. A claim of equivalence or non-inferiority would require an appropriate design, prespecified margin, and analysis, none of which is reported in the registry record. The results also do not answer the comparison between pemetrexed and the combination on overall survival, because that direct test is not reported.

The findings apply to the defined trial setting: advanced non-squamous non-small cell lung cancer treated with the study's induction regimen and then randomized to maintenance. They should not automatically be generalized to other populations, treatment sequences, or care settings. The record does not provide detailed eligibility criteria or enough information to evaluate how closely the trial participants represent other groups.

Secondary response-rate results offer additional descriptive information, but they do not alter the primary endpoint interpretation. The record lacks a reported multiplicity strategy and does not provide all secondary endpoint results. Safety counts describe reported adverse events but do not establish comparative causation. Taken together, the registry supports a careful description of the reported estimates and their uncertainty, not a treatment recommendation.

14. Why This Trial Matters Statistically

This trial is statistically informative because it compares three maintenance strategies within a randomized, parallel-group phase III design after a shared induction regimen. That structure makes the maintenance comparisons distinct from a comparison of the entire treatment course against no treatment. The primary question concerns survival after randomization to maintenance, and the registry explicitly excludes patients who received induction only from that analysis.

The example also illustrates why an arm-specific median is not the same as a comparative treatment effect. The medians offer an interpretable summary of each group, while the log-rank tests compare survival experience over follow-up. Neither the numerical ordering of medians nor the p-values alone provide a precise estimate of how much outcomes differ. A contrast estimate and its confidence interval would give a more direct account of comparative magnitude and uncertainty, but those are not reported for the primary endpoint.

It also highlights the importance of knowing the analysis population and multiplicity plan. The registry's detailed endpoint description clarifies that the primary analysis is restricted to maintenance-randomized patients, while allocation procedures, missing-data methods, and multiplicity handling are not reported. Without those details, readers must separate what the posted results directly show from what would require assumptions about the analysis.

Finally, the secondary and safety summaries illustrate that different outcome types have different interpretations. Response rate is not interchangeable with overall survival, and adverse-event counts across treatment periods do not automatically form a randomized comparison. Reading each estimate in the context of its endpoint, population, and uncertainty is central to sound clinical-trial interpretation.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record does not report a hazard ratio or a confidence interval for either primary between-arm comparison. It gives arm-level medians and confidence intervals, which are useful but do not fully describe the comparative survival effect. A direct primary comparison between pemetrexed and the combination is also not reported.

Several design and analysis details are absent: the randomization ratio and procedure, allocation concealment, the original sample-size justification, a prespecified significance threshold, multiplicity handling, and any interim analysis plan. The maintenance arm descriptions are truncated, limiting detail about treatment delivery. The record does not provide the full eligibility criteria or clarify all aspects of the primary analysis set beyond the statement that induction-only patients were excluded.

Information about missing survival follow-up, censoring rules, treatment discontinuation, adherence, and handling of missing secondary outcomes is not reported in the registry record. These omissions limit assessment of possible bias and the completeness of follow-up. The registry lists progression-free survival as a secondary outcome but does not give numerical results for it in the posted data. Response-rate comparisons are not accompanied by sponsor-reported analyses or a stated multiplicity approach.

The safety summaries are counts across an induction group and maintenance groups, with different at-risk populations and treatment periods. The record does not provide enough detail in these summaries to make a fully adjusted or causal comparison of safety between randomized maintenance strategies. The listed serious adverse-event terms also lack group-specific breakdowns in the record summarized.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Bevacizumab or Pemetrexed Disodium Alone or In Combination After Induction Therapy in Treating Patients With Advanced Non-Squamous Non-Small Cell Lung Cancer — NCT01107626.
• Registry record last updated 2023-07-07; status last verified 2023-06.
• Results first posted 2021-02-23.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT01107626 is a completed randomized, open-label, parallel-group phase III study of maintenance bevacizumab, pemetrexed, or their combination following induction therapy for advanced non-squamous non-small cell lung cancer. The registry lists enrollment of 1516.

The primary endpoint is overall survival from randomization to death or last known alive. The registry reports arm-specific median survival estimates and 97.5% confidence intervals, plus log-rank p-values for bevacizumab compared with pemetrexed and with the combination. The primary analysis is described as including patients randomized to maintenance, not induction-only patients.

Response rate is among the posted secondary results. Safety counts are reported for induction and maintenance groups. Details about multiplicity, missing-data methods, allocation procedures, and the original sample-size justification are not reported in the registry record.