Temozolomide With or Without Veliparib in Treating Patients With Newly Diagnosed Glioblastoma Multiforme: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Temozolomide With or Without Veliparib in Treating Patients With Newly Diagnosed Glioblastoma Multiforme (NCT02152982), a randomized phase 2, phase 3 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

The trial is randomized, double masked and parallel group. Random allocation is intended to make the treatment groups comparable in expectation, while masking can reduce the influence of knowledge about assignment on care, assessment or reporting. The registry record does not describe how allocation was generated or concealed, or which people were masked.

The registry record describes a phase II/III trial and lists an enrollment of 447. The posted overall-survival arm summaries include 224 people in the temozolomide-plus-placebo arm and 223 in the temozolomide-plus-veliparib arm. The record lists the National Cancer Institute as sponsor. Its status is active, not recruiting.

The comparison addresses the addition of veliparib to temozolomide, not temozolomide versus no treatment. The placebo arm helps preserve masking while both arms receive temozolomide. The central statistical question is whether the survival distributions differ, and how precisely the registry's reported estimates describe that difference.

3. Trial Design

This is a randomized, double-masked, parallel-group phase II/III trial. The listed enrollment is 447. One group receives temozolomide with placebo, and the other receives temozolomide with veliparib. The official title identifies the intended setting as newly diagnosed glioblastoma with MGMT promoter hypermethylation. The registry record also lists glioblastoma and gliosarcoma among the conditions.

In both arm descriptions, temozolomide is taken by mouth once daily on days 1–5, and treatment repeats every 28 days for 6 cycles, unless confirmed disease progression or unacceptable toxicity occurs. Placebo or veliparib is taken by mouth twice daily on days 1–7. Laboratory biomarker analysis and quality-of-life assessment are also listed as interventions or study activities; the registry record does not provide their results in the outcome figures.

The primary outcome is overall survival, analyzed as a median time-to-event outcome. The registry description says that each arm's overall-survival distribution will be estimated with the Kaplan-Meier method and compared using a stratified log-rank test. The timeframe listed for the primary outcome is 83 months. The record does not report details of the stratification factors, the primary analysis population, or the handling of censoring and missing outcome data.

Progression-free survival and objective tumor response are among the secondary outcomes. The record also lists interaction with an Optune device and overall adverse-event rates for grade 3 or higher events. Results are posted for only some outcomes; details about the other listed outcomes are not reported in the registry record.

The registry lists a start date of 2014-12-15 and a primary completion date of 2021-12-01. Its current status is active, not recruiting. These record details describe the trial's design and status, but do not establish that every planned analysis or follow-up has been completed.

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.

Overall Study

Reasons for not completing this period:

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.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over 83 months.

Overall survival is the time from the trial's defined starting point until death from any cause. It is a time-to-event outcome: participants who have not had the event by the end of their available follow-up contribute information up to that point, rather than being treated as if their eventual outcome were known. The precise time origin and censoring rules are not reported in the registry record.

The primary result is expressed as a median in months for each arm. The median is the time at which the estimated survival curve reaches the halfway point. It describes the center of the observed time-to-event distribution, but it does not summarize every feature of the curves. Two arms can have similar medians while differing elsewhere in follow-up, or different medians without a constant difference across time.

The registry says the distributions are estimated with the Kaplan-Meier method and compared using a stratified log-rank test. Kaplan-Meier estimation uses the timing of events and accommodates censoring under its assumptions. The log-rank test evaluates evidence of a difference between the survival distributions over follow-up; it is not itself an estimate of the size of a treatment effect.

Overall survival is used because it directly captures time to death and does not require a judgment that a tumor has progressed. Its interpretation still depends on follow-up, the definition of the time origin, censoring and the analysis population. Those details, beyond the method named in the registry, are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

How does the primary survival analysis work?

The Kaplan-Meier method estimates the probability of remaining event-free over time while allowing some participants' follow-up to end before an event is observed. Such observations are censored: they contribute information until their last known follow-up, but their later event status is unknown. In general, Kaplan-Meier estimation relies on assumptions about censoring, including that censoring does not systematically conceal a different event risk. The registry names Kaplan-Meier estimation but does not report its censoring rules or diagnostics.

The primary endpoint is summarized with a median for each arm. The median is read from the estimated survival curve at the point where the estimated probability reaches one half. It is a useful, interpretable summary, but it cannot describe all differences between curves. A median does not tell whether curves separate early or late, whether they cross, or how many participants remain under observation at a particular time. Those details are not reported in the registry record.

The registry names a stratified log-rank test for the primary comparison. In general, a log-rank test compares observed and expected event patterns between groups across follow-up. Stratification means that comparisons are made within designated groups and then combined. This can account for the stratification factors in the design, but the factors themselves are not reported in the registry record. A p-value measures how compatible the test statistic is with a null hypothesis under the test assumptions; it does not measure the probability that the null hypothesis is true, nor does it quantify the size or clinical importance of an effect.

The primary outcome reports 90% confidence intervals for the arm-specific medians. A confidence interval represents the uncertainty associated with an estimate under the method used. The interval does not describe the spread of individual survival times, and its endpoints should not be treated as guaranteed bounds on the true value. The registry does not post a confidence interval for a primary between-arm effect such as a hazard ratio.

What does a hazard ratio mean?

The secondary progression-free survival analysis reports a hazard ratio from a stratified Cox proportional hazards analysis. In general, a hazard is the instantaneous event rate among participants who have not yet experienced the event. A hazard ratio compares those rates between groups over time. It is not a direct comparison of medians, and its interpretation as a single relative effect is most straightforward when the proportional-hazards assumption is reasonable. The registry does not report diagnostics for that assumption or a detailed definition of progression-free survival.

The reported progression-free survival hazard ratio is 1.05, with a 95% confidence interval from 0.86 to 1.29. The interval includes 1, the value corresponding to equal hazards under the model. This interval is compatible with a range of effects, and the result should not be read as proof of no effect. The stratified log-rank method and the Cox model are related ways of analyzing time-to-event information, but they answer their questions through different summaries.

How should a binary outcome be read?

Objective tumor response is posted as participant counts in each arm. A chi-squared test assesses whether the observed counts are compatible with equal response proportions under its assumptions. General limitations include the information lost when time to response or duration of response is not included in a simple count. The registry does not report a detailed response definition or a confidence interval for the sponsor's comparison in the posted analysis.

The posted objective-response counts are 34 of 118 in the temozolomide-plus-placebo arm and 37 of 119 in the temozolomide-plus-veliparib arm. The sponsor's chi-squared analysis reports a p-value of 0.702. A separately recalculated unadjusted p-value is 0.702. These values are close, and the record does not flag a discrepancy. The sponsor's analysis remains the principal reported analysis; a simple recalculation from summary counts does not reproduce every feature of a prespecified analysis.

What are the planning illustrations?

The sample-size illustrations are based on observed effects and are not the trial's original design justification or observed power. For binary outcomes, they express participants per arm for 80% or 90% power under the stated assumptions: a two-sided significance level of 0.05, equal allocation, a normal approximation, and detection of a chosen fraction of the observed effect. The listed fractions are 100%, 75% and 50% of the observed effect. Such calculations depend on the assumed effect: planning for a smaller fraction of an observed effect requires larger illustrative group sizes.

For the time-to-event illustration, the figures are total events, not participants per arm, and use the Schoenfeld approach. The posted planning values are 13,189 and 17,656 total events for 80% and 90% power when planning for 100% of the observed effect; 23,447 and 31,389 for 75%; and 52,755 and 70,624 for 50%. For objective response, the corresponding participants-per-arm figures are 6,341 and 8,489 for 100%; 11,212 and 15,009 for 75%; and 25,087 and 33,584 for 50%. These are planning illustrations, not observed power, and they are not evidence that the trial was designed around these effects.

Why can over-dispersion matter?

General explanation: over-dispersion occurs when outcomes vary more than a model assumes. If ignored, uncertainty estimates can be too small and tests can appear more decisive than warranted. This concept is especially relevant to some count outcomes; it is not a claim that over-dispersion was present in this trial. The primary outcome is time-to-event, and the registry record does not report a dispersion assessment.

More broadly, analyses depend on outcome definitions, allocation, follow-up, censoring, missing data and model assumptions. The registry reports selected methods but not a full statistical analysis plan. That limits the ability to assess whether all analytic choices were specified in advance or how deviations and incomplete data were handled.

10. Analyses Reported by the Sponsor

For overall survival, the registry posts arm-specific median estimates and a sponsor-reported log-rank comparison. No recalculated between-arm effect is presented for this median time endpoint, because the posted arm summaries do not provide the information needed to reconstruct a time-to-event comparison. The sponsor's reported analysis is therefore the relevant comparison to interpret. Its p-value is 0.167, and the record does not flag a discrepancy.

For progression-free survival, the reported hazard ratio is 1.05, with a 95% confidence interval from 0.86 to 1.29. The registry also reports a stratified log-rank p-value of 0.316. The hazard ratio is the sponsor's model-based estimate; it is not replaced by a calculation from the arm medians.

For objective tumor response, an unadjusted recalculation from the posted summary counts gives a p-value of 0.702. The sponsor-reported chi-squared p-value is 0.702. These are close, and the record indicates no discrepancy. The small difference in displayed precision does not change the broad interpretation: the posted analysis does not provide strong evidence of a difference in response proportions.

Recalculations from summary statistics are limited because they may not reproduce stratification, covariate adjustment, missing-data handling or other features of the sponsor's analysis. When a discrepancy is flagged, the sponsor's own analysis takes precedence. No discrepancy is flagged in this record, but the sponsor's reported analyses remain the primary results rather than a substitute for them.

11. Multiplicity, Missing Data and Analysis Populations

The registry identifies overall survival as the primary outcome and lists several secondary outcomes. It does not report how the family of statistical tests was handled, whether a formal testing hierarchy was prespecified, or whether adjustments were made for multiple comparisons. Without that information, secondary p-values should be interpreted cautiously: testing multiple outcomes can increase the chance of at least one apparently notable result occurring by chance.

The record does not describe the analysis population for the primary or secondary outcomes. It does not say whether the analyses used an intention-to-treat population, a per-protocol population, or another defined group. The posted outcome counts differ from the arm summaries for the survival outcomes, but the reasons and participant flow are not described. The registry record does not report how exclusions, treatment discontinuation, crossover or protocol deviations were handled.

Missing outcome data and censoring are important for both survival and response analyses. For overall survival, the registry names Kaplan-Meier estimation and a log-rank test, but does not state the censoring rules, the number at risk over time or how loss to follow-up was assessed. For progression-free survival, details of progression assessment and handling of missing assessments are not reported in the registry record. For objective response, the record does not explain how participants without an evaluable response were treated.

Other statistical-plan details are also not reported in the registry record, including the sample-size justification for the trial, interim analyses, stopping rules, prespecified subgroup analyses, covariate adjustments and sensitivity analyses. Their absence from the posted record does not establish that they were not conducted; it means their methods cannot be assessed from the registry entries available.

Consequently, the posted estimates can be described, but a complete assessment of risk of bias, missing-data sensitivity and whether all analyses followed a prespecified plan is not possible from the registry record alone. The sample-size illustrations are retrospective planning calculations based on observed effects and must not be confused with the trial's original planning assumptions or observed power.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: fatigue (3.8%), seizure (3.2%), generalized muscle weakness (2.8%), thromboembolic event (2.8%), hyperglycemia (2.3%).

The adverse-event record lists serious events affecting 47 participants in the temozolomide-plus-placebo arm, with 213 at risk, and 43 participants in the temozolomide-plus-veliparib arm, with 218 at risk. It also lists other adverse events affecting 207 participants in the placebo arm and 217 in the veliparib arm, with the same respective at-risk counts. Death counts are 175 and 183. These are separate registry counts and should not be interpreted as a complete safety comparison without the outcome definitions and analysis details.

The most frequently listed serious-event terms include fatigue at 3.8%, seizure at 3.2%, generalized muscle weakness at 2.8%, thromboembolic event at 2.8%, hyperglycemia at 2.3% and confusion at 1.9%. The registry record does not explain the denominator or event-counting rules for these percentages, or provide a full event-by-arm breakdown for every term.

Serious adverse events are not the same as all adverse events, and a count of participants affected does not capture severity, duration, recurrence or attribution to treatment. These figures describe what the registry posts; they should not be used alone to conclude that one regimen is safer. The record's listed safety outcome includes grade 3 or higher adverse-event rates, but its detailed definition and complete results are not reported in the registry record.

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

The primary result is a comparison of overall-survival distributions between two groups that both received temozolomide: one with placebo and one with veliparib. The posted median estimates are different, but the primary log-rank analysis does not provide strong statistical evidence of a difference at the conventional 0.05 threshold. The confidence intervals show that the medians are estimated with uncertainty.

This does not show that the treatments are equivalent. A failure to find strong evidence of a difference is not proof that no difference exists. Establishing equivalence or non-inferiority requires a prespecified margin and an analysis designed for that question; those are not reported in the registry record. Nor does the result establish that any observed difference is clinically important. That judgment requires clinical context beyond a p-value or median estimate.

The estimates describe group-level results in the trial's posted record. They do not predict an individual's outcome, establish the effect for every subgroup, or show what would happen in settings outside the trial. The registry record does not report enough information to evaluate subgroup-specific effects or the representativeness of the analyzed population.

Secondary outcomes offer additional perspectives but are not substitutes for the primary endpoint. Their interpretation depends on their definitions, statistical methods and multiplicity considerations. Because the registry does not report a full multiplicity strategy or complete details for all listed secondary outcomes, isolated secondary findings should not be treated as definitive evidence.

Overall, the registry results describe estimates and uncertainty, not a categorical answer about benefit or lack of benefit. They do not establish causality beyond the randomized comparison's assumptions, and they do not provide medical advice. They are best read as the posted evidence for this particular comparison, together with the limitations in reporting and analysis detail.

14. Why This Trial Matters Statistically

This trial illustrates why clinical-trial interpretation should separate an estimate from the test used to compare groups. A median in each arm is an accessible summary, while the log-rank test evaluates event patterns across follow-up. Neither alone gives a complete account: medians omit much of the curve's shape, and a p-value does not measure the size or importance of a difference.

It also shows why uncertainty matters. Confidence intervals around medians and a hazard ratio describe the precision of estimates under their methods. An interval that includes a null value does not demonstrate equivalence; it indicates that the data and model do not sharply distinguish the null from a range of alternatives. Reading the point estimate without its interval would overstate certainty.

Time-to-event endpoints require attention to censoring and follow-up, while response outcomes require clear definitions and an appropriate comparison of proportions. These outcomes can point in different directions without being contradictory because they measure different aspects of disease course. The registry's limited reporting of definitions, analysis populations and missing-data handling makes transparent interpretation particularly important.

Finally, the case highlights the difference between design planning and retrospective calculations. Planning illustrations based on an observed effect are not the original sample-size rationale and are not observed power. Multiple secondary outcomes also require caution, since unadjusted tests can yield chance findings. These are general statistical principles applied to the trial's posted methods and results, not claims about unreported analyses.

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.

Several trial-specific details needed for a fuller appraisal are not reported in the registry record. The primary analysis population, the exact time origin for overall survival, censoring rules, participant flow and numbers at risk over follow-up are not specified. The primary outcome description gives a timeframe of 83 months, but does not provide a detailed account of follow-up completeness or event ascertainment.

The record names a stratified log-rank test but does not identify the stratification factors. It does not post a primary hazard ratio or a confidence interval for a between-arm overall-survival effect. Consequently, interpretation relies on the arm-specific median estimates, their 90% confidence intervals and the reported test, rather than a single effect estimate with an interval for the comparison.

For progression-free survival, the record gives a hazard ratio and confidence interval but does not report the endpoint's detailed event definition, proportional-hazards diagnostics or missing-assessment procedures. For objective tumor response, the definition of response, evaluation schedule and handling of participants without an evaluable assessment are not reported in the registry record. The extent to which results were prespecified or adjusted for multiple testing is also not reported.

Safety reporting includes aggregate counts and selected serious-event terms, but does not provide the complete event-level context needed to compare severity, duration, recurrence or treatment attribution. The record lists other secondary outcomes, including interaction with an Optune device, but detailed results for all listed outcomes are not reported in the information summarized.

These are limitations of what the registry record reports. They do not demonstrate that the trial lacked additional protocols, analyses or data; those details are simply not reported in the registry record.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Temozolomide With or Without Veliparib in Treating Patients With Newly Diagnosed Glioblastoma Multiforme — NCT02152982.
• Registry record last updated 2026-10-27; status last verified 2026-10.
• Results first posted 2023-04-04.
• 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

NCT02152982 is a randomized, double-masked, parallel-group phase II/III trial sponsored by the National Cancer Institute. It compares temozolomide with placebo against temozolomide with veliparib in newly diagnosed glioblastoma with MGMT promoter hypermethylation. The registry lists enrollment of 447 and status active, not recruiting.

Overall survival is the primary outcome. The posted medians are 24.8 months in the placebo arm and 28.1 months in the veliparib arm, with 90% confidence intervals of 22.6 to 27.7 and 24.3 to 33.3 months, respectively. The sponsor-reported log-rank comparison does not provide strong statistical evidence of a difference.

Progression-free survival and objective tumor response are also posted as secondary outcomes. Their findings do not provide strong evidence of a difference between arms. Interpretation is limited by missing registry entries on analysis populations, missing data, outcome definitions and multiplicity. Planning illustrations based on observed effects are not the trial's original design justification or observed power.