A Study of the Effectiveness and Safety of Nivolumab Compared to Bevacizumab and of Nivolumab With or Without Ipilimumab in Glioblastoma Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of the Effectiveness and Safety of Nivolumab Compared to Bevacizumab and of Nivolumab With or Without Ipilimumab in Glioblastoma Patients (NCT02017717), a randomized phase 3 study. The primary outcome is Percentage of Participants With Drug-Related Adverse Events Leading to Discontinuation by Worst CTC Grade for All Treated Participants in Cohorts 1, 1b, 1c and 1d Who Permanently Discontinued Study Medication Prior to Completing Four Doses.

2. Clinical Question

3. Trial Design

The registry record classifies NCT02017717 as a phase 3, randomized, open-label, parallel study of recurrent glioblastoma. It lists Bristol-Myers Squibb as the sponsor and reports enrollment of 529. The study status is completed. The recorded study start is 2014-02-07, and primary completion is recorded as 2019-06-17.

The official title describes a randomized phase 3 study of nivolumab versus bevacizumab, alongside multiple phase 1 safety cohorts of nivolumab or nivolumab with ipilimumab across different lines of glioblastoma. The registry classifies the study phase as PHASE3, while the official title describes those additional phase 1 cohorts. The record’s brief summary likewise distinguishes the comparison of nivolumab alone with bevacizumab from evaluation of the safety and tolerability of nivolumab alone or with ipilimumab.

The listed interventions are nivolumab, bevacizumab and ipilimumab. Cohort descriptions include nivolumab alone, nivolumab plus ipilimumab, and the Cohort 2 bevacizumab arm. Cohorts 1c and 1d also have descriptions involving radiation therapy, with or without temozolomide. The registry lists secondary outcomes including overall survival, progression-free survival and objective response rate for specified cohorts.

The record identifies no masking, consistent with its open-label designation. It does not report the randomization ratio, sequence-generation method, allocation concealment, site count, or the detailed schedule of assessments in the registry record for this analysis. The primary safety endpoint is not the main randomized comparison described in the official title: it applies to treated participants from Cohorts 1, 1b, 1c and 1d who permanently stopped study medication before four doses. That distinction matters when interpreting the endpoint population and when considering whether its results can be compared across treatment arms.

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.

Before assignment: 529 participants participated in the study. Of these 529 participants, 448 were randomized in the pre-treatment phase while 81 were directly treated in a non-randomized fashion. Of the 448 participants randomized, 423 moved to the treatment phase. The total number of treated participants was 504 (81 directly treated, 423 randomized and treated). GBM = Glioblastoma MGMT = O6-methylguanin-DNA-methyltransferase

Only the first 6 of 9 groups are shown.

Before Treatment Period

Reasons for not completing this period:

Treatment Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All Treated Participants in Cohorts 1, 1b, 1c and 1d Who Permanently Discontinued Study Medication Prior to Completing Four Doses

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: Cohorts 1b and 1c/1d (Part A) baseline characteristics are in regards to Treated Participants, while Cohorts 1, 1c/1d (Part B) and 2 baseline characteristics are in regards to Randomized Participants.

Only the first 6 of 10 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants With Drug-Related Adverse Events Leading to Discontinuation by Worst CTC Grade for All Treated Participants in Cohorts 1, 1b, 1c and 1d Who Permanently Discontinued Study Medication Prior to Completing Four Doses, measured over Includes events reported between first dose and 30 days after last dose of study therapy (up to 3 doses, up to approximately 2 months).

The primary endpoint is the percentage of participants with a drug-related adverse event leading to discontinuation, classified by worst CTC grade. It covers treated participants in Cohorts 1, 1b, 1c and 1d who permanently discontinued study medication before completing four doses. The record states that adverse events were graded using the National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.0, and identifies MedDRA Version 24.1.

The specified observation window runs from the first dose through 30 days after the last dose of study therapy. The description limits treatment exposure to up to 3 doses and approximately 2 months. The endpoint therefore focuses on early treatment discontinuation associated with drug-related toxicity, rather than on all adverse events during any possible follow-up or on the study’s efficacy outcomes.

In general, recording the worst grade gives a way to describe the most severe reported toxicity in the endpoint population. Grade categories help distinguish degrees of severity, while the discontinuation condition makes the endpoint relevant to treatment stopping. However, the measure does not by itself describe the full range of symptoms, their duration, or whether participants experienced more than one event. Nor does it establish a causal comparison between treatments unless results are reported for comparable groups under a suitable analysis. Values by grade and a single summary comparison are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

In general, an endpoint defined as a percentage is a proportion: the number of participants meeting a stated event definition relative to the number in the endpoint population. Its interpretation depends on both parts of that definition. For this endpoint, the event is a drug-related adverse event leading to permanent discontinuation before completion of four doses, and the population is restricted to treated participants in Cohorts 1, 1b, 1c and 1d who met that discontinuation condition. The registry record does not report the grade-specific endpoint values or the corresponding denominators.

Why does the denominator matter?

A percentage is meaningful only when its denominator is clear. The study-wide enrollment of 529 describes the overall enrollment record, not the number of participants in this endpoint population. Using overall enrollment as the denominator would change the endpoint definition. The record does not report how many participants in the specified cohorts permanently discontinued before completing four doses, nor does it report grade-specific denominators. These quantities cannot be inferred from the serious adverse-event counts.

In general, a confidence interval expresses the range of values compatible with the observed data under a stated statistical procedure. It helps convey precision, rather than only giving a point estimate. The registry record does not report a confidence interval for this endpoint in the information. Without the endpoint counts, denominators and an uncertainty estimate, the precision of any grade-specific percentage cannot be assessed.

What does the grading add?

The endpoint classifies events by worst CTC grade, using the National Cancer Institute CTCAE version 4.0. In general, standardized severity grades make safety reporting more structured and can distinguish levels of severity. A worst-grade summary emphasizes the most severe grade recorded for a participant under the endpoint definition; it does not describe every event, the timing of each event, or the total burden of repeated events. The registry record does not report the category-level values needed to examine the distribution across grades.

The record identifies a window from first dose through 30 days after the last dose of study therapy, with exposure described as up to 3 doses and approximately 2 months. In general, a defined observation window helps specify which events count, but differences in time observed can affect the opportunity to record events. The record does not provide participant-level exposure durations or additional details that would allow a duration-adjusted analysis.

Why distinguish description from comparison?

In general, a descriptive percentage answers how often an event was recorded within a defined group. A comparative analysis asks whether outcomes differ between groups and must account for the study design and the populations being compared. The primary endpoint population spans several cohorts and is selected by early permanent discontinuation. No single comparative estimate for this endpoint is reported in the registry record, and the record states that the primary outcome has several categories or time points rather than one reduced comparison. It would therefore be inappropriate to present a treatment contrast as though it were an analysis reported by the registry.

The overall study includes randomized and nonrandomized components in its description: a randomized phase 3 comparison and multiple phase 1 safety cohorts. In general, randomization can support a fair comparison when treatment groups are analyzed as assigned and other design features are respected. The registry entries do not specify how the endpoint was analyzed in relation to randomization, whether participants were analyzed as treated or as assigned, or whether any adjustment was used. The primary endpoint’s listed population is based on treatment received and discontinuation status, so it should not automatically be treated as the randomized comparison population.

What can be said about statistical testing?

The record does not report a statistical test, model, effect estimate, confidence interval, or hypothesis for the primary endpoint in the information. It also does not report a sample-size justification, interim-analysis plan, or detailed statistical analysis plan. In general, a p-value addresses compatibility of data with a specified null model; it does not measure the size or clinical importance of an effect. No p-value for this endpoint is reported in the registry record, so no significance claim can be made.

The registry lists 11 outcome measures and 5 analyses. Those counts alone do not explain which outcomes were tested together, which analyses were confirmatory, or how multiple comparisons were handled. In general, testing many outcomes or subgroups can increase the chance of chance findings if multiplicity is not addressed. The method for handling multiplicity is not reported in the registry record.

For planning, a sample-size illustration based on an observed effect is not an observed-power result and does not establish the study’s original design assumptions. The record’s stated illustration assumptions are a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect set as a fraction of the observed effect. No illustration values are reported in the registry record for this analysis, and those assumptions should not be attributed to the trial’s original design.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 11 outcome measures and 5 analyses, but those counts do not specify a multiplicity strategy. The record does not report whether outcomes were organized into a prespecified testing hierarchy, whether alpha was divided among tests, or whether any other adjustment for multiple comparisons was used. In general, multiplicity matters when several outcomes, time points, or subgroup comparisons are evaluated because the chance of at least one chance-positive finding can rise as testing expands. No conclusion about the study’s control of that risk can be drawn from the outcome and analysis counts alone.

The primary endpoint itself has several categories or time points and is not reduced to one comparison in the registry record. The endpoint values by grade, any formal test, and any corresponding confidence intervals are not reported in the registry record. The analysis plan identifying which category or time point was primary for inference is also not reported in the registry record.

The endpoint population is described as treated participants in Cohorts 1, 1b, 1c and 1d who permanently discontinued before completing four doses. The number of eligible participants, handling of participants with incomplete adverse-event information, and method for missing data are not reported in the registry record. The registry entries also do not specify whether the primary endpoint used an intention-to-treat, per-protocol, or other analysis population. These omissions limit assessment of both the denominator and potential selection effects.

In general, missing-data methods matter when participants lack outcome information or when follow-up ends at different times. The defined adverse-event window helps set the endpoint’s scope, but the record does not report whether all participants had complete observation through the specified period or how incomplete observation was addressed. It is therefore not possible to evaluate the sensitivity of the reported endpoint to missing information from the registry entries available.

11. 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: malignant neoplasm progression (40.0%), hypothyroidism (20.0%), colitis (20.0%), diarrhoea (20.0%), alanine aminotransferase increased (20.0%).

The registry reports serious adverse-event counts for listed groups using numbers at risk and numbers affected. For Cohort 1: Arm N1+I3, the serious-event figures are 10 at risk and 8 affected; for Cohort 1: Arm N3, they are 10 at risk and 7 affected. For Cohort 1b: Arm N3+I1, the figures are 20 at risk and 15 affected. Part A Cohort 1c: Arm N3+RT+TMZ lists 31 at risk and 24 affected, while Part A Cohort 1d: Arm N3+RT lists 30 at risk and 19 affected.

For Part B Cohort 1c: Arm N3+RT+TMZ, the serious-event figures are 28 at risk and 16 affected; for Part B Cohort 1d: Arm N3+RT, they are 28 at risk and 19 affected. Cohort 2: Arm N3 lists 182 at risk and 117 affected, and Cohort 2: Arm B lists 165 at risk and 94 affected. These are descriptive counts for the listed groups, not the primary endpoint values.

The registry also lists malignant neoplasm progression at 40.0%, and hypothyroidism, colitis, diarrhoea, alanine aminotransferase increased, and confusional state at 20.0% each among the top serious-event terms. The denominator and the exact scope of these term percentages are not reported in the registry record. Do not interpret them as percentages of the whole study or as comparative risks. The deaths counts listed by group are not the same measure as serious adverse events and should not be used to infer a cause of death.

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

The primary endpoint is designed to describe drug-related adverse events that led to permanent discontinuation before four doses among a defined set of treated participants. Its focus is therefore early discontinuation-associated toxicity, categorized by worst CTC grade. It is not an overall measure of every adverse event, all treatment exposure, survival, progression, or response.

The registry record does not report the grade-specific endpoint values or a single comparison for this outcome. As a result, the available information does not establish the percentage of the endpoint population with such events, how those events were distributed by grade, or whether the rate differed between treatments or cohorts. The listed serious adverse-event counts concern a separate safety summary and do not answer those questions.

In general, a registry outcome is best interpreted according to its defined population, event criteria and observation window. Here, the population is restricted to participants in Cohorts 1, 1b, 1c and 1d who permanently discontinued study medication before completing four doses; the window runs from first dose through 30 days after the last dose, with treatment exposure limited to up to 3 doses and approximately 2 months. The study-wide enrollment figure of 529 should not be treated as the denominator for this endpoint.

The study’s broad structure also matters. The official title describes a randomized phase 3 comparison of nivolumab and bevacizumab as well as multiple phase 1 safety cohorts. The primary endpoint is assigned to the specified safety cohorts, not described as the principal efficacy comparison. The registry record does not report a primary-endpoint estimate or uncertainty interval that would support a comparative conclusion. The endpoint information therefore cannot establish that one treatment is safer, more effective, or less effective than another.

In general, absence of a reported estimate is not evidence that no event occurred or that groups were equivalent. It means the registry record does not provide the information needed to quantify this endpoint or evaluate its precision. Conclusions should remain limited to the study structure and descriptive safety information that the record actually reports.

13. Why This Trial Matters Statistically

This study is statistically notable for combining a randomized comparison with additional safety cohorts. The registry describes a phase 3 randomized, open-label, parallel study, while the official title also identifies multiple phase 1 safety cohorts. That structure means that outcome populations and comparisons must be kept distinct: a result from one cohort cannot automatically be treated as a result from the randomized comparison or generalized across all listed arms.

The primary endpoint illustrates why the population definition is central to safety interpretation. It concerns treated participants in Cohorts 1, 1b, 1c and 1d who permanently discontinued before completing four doses, and it classifies drug-related discontinuation events by worst CTC grade. This is a focused endpoint rather than a broad estimate for all enrolled participants. The registry’s overall enrollment of 529 does not supply its denominator.

The record also highlights the difference between an outcome’s description and a reportable estimate. The endpoint definition and observation window are described, but the grade-specific values and a single comparison are not reported. Without those quantities, a reader cannot assess the endpoint’s magnitude or precision. That distinction is important in registry-based interpretation: detailed outcome wording does not itself provide evidence about the observed frequency or between-group contrast.

The listed serious adverse-event counts provide additional descriptive information across cohorts, but they answer a different question from the primary endpoint. Keeping those measures separate avoids conflating serious events with events that were drug-related and caused early discontinuation. More generally, clear reporting of denominators, analysis populations, uncertainty and multiplicity methods would help readers judge how much weight to place on safety comparisons. Those details are not reported in the registry record for the primary endpoint.

14. 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 principal limitation for interpreting the primary endpoint is that its category-level values are not reported in the registry record. The endpoint has multiple worst-grade categories or time points and is not reduced to a single comparison. Without the values, denominators and uncertainty estimates, its frequency, grade distribution and precision cannot be assessed.

The endpoint population is selected by treatment exposure and early permanent discontinuation: treated participants in Cohorts 1, 1b, 1c and 1d who stopped medication before completing four doses. The number meeting this definition is not reported in the registry record. The overall enrollment of 529 is not an appropriate substitute for that endpoint-specific denominator. Because the endpoint is also not described as the Cohort 2 randomized comparison, it should not be interpreted as a direct estimate of the nivolumab-versus-bevacizumab contrast.

The record does not report the statistical analysis method, confidence intervals, missing-data approach, multiplicity procedure, or detailed analysis population rules for the primary endpoint. It does not report a sample-size justification, interim-analysis plan, or the specific hypothesis for this endpoint. These gaps prevent assessment of the inferential framework and the robustness of any comparison.

The serious adverse-event summaries have their own limits. Although group-level numbers at risk and affected are listed, the selected term percentages have no reported denominator in the record. They should not be read as directly comparable risks without additional context. The registry also reports deaths by group, but those counts alone do not identify causes or establish relationships to treatment.

Finally, the study includes different cohorts, treatment combinations and settings. Results from these groups may not be interchangeable, and the registry entries do not provide enough detail to assess comparability across them. These limitations call for a descriptive, record-specific interpretation rather than a broad conclusion about comparative safety or efficacy.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: A Study of the Effectiveness and Safety of Nivolumab Compared to Bevacizumab and of Nivolumab With or Without Ipilimumab in Glioblastoma Patients — NCT02017717.
• Registry record last updated 2025-04-04; status last verified 2025-03.
• Results first posted 2022-06-29.
• 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.

18. Record Summary

NCT02017717 is recorded as a completed, randomized, open-label, parallel phase 3 study in recurrent glioblastoma, sponsored by Bristol-Myers Squibb. The registry lists enrollment of 529, with a study start of 2014-02-07 and primary completion on 2019-06-17. The official title describes a randomized nivolumab-versus-bevacizumab comparison and multiple phase 1 safety cohorts involving nivolumab alone or with ipilimumab.

The primary endpoint is the percentage of treated participants in Cohorts 1, 1b, 1c and 1d who had a drug-related adverse event leading to discontinuation before completing four doses, classified by worst CTC grade. The registry record does not report the grade-specific endpoint values or a single comparative estimate. It does list serious adverse-event counts by group and selected serious-event terms, which are distinct from the primary endpoint. Details of the primary endpoint’s statistical analysis, missing-data handling and multiplicity approach are not reported in the registry record.