Pembrolizumab in Treating Patients With Recurrent Glioblastoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pembrolizumab in Treating Patients With Recurrent Glioblastoma (NCT02337686), a na phase 2 study. The primary outcome is Number of Participants With Progression Free Survival at 6 Months.

2. Clinical Question

3. Trial Design

This is a phase II, single-group pharmacodynamic study sponsored by M.D. Anderson Cancer Center. The registry lists recurrent glioblastoma and recurrent gliosarcoma as the conditions, and pembrolizumab, laboratory biomarker analysis, pharmacological study, and conventional surgery among the interventions. Enrollment is recorded as 18. The study began on 2015-04-28, and the primary completion date is 2017-05-24. Its status is listed as active, not recruiting.

The treatment description combines pembrolizumab with surgery rather than evaluating pembrolizumab in a randomized comparison. Participants were scheduled to receive pembrolizumab intravenously over 30 minutes on day -21 and day -1, then undergo surgery on day 0. After two to three weeks or recovery from surgery, pembrolizumab continued every three weeks; courses were to repeat every 42 days for up to 24 months in the absence of disease progression. The recorded description ends mid-sentence, so further treatment conditions are not reported in the registry record.

Because there is one group and no control group, outcomes describe participants enrolled in this treatment program. They do not directly estimate a contrast between treatment strategies. Allocation is listed as not applicable and masking as none. Further details about eligibility criteria, site count, recruitment procedures, and protocol deviations are not reported in the registry record. The record identifies five outcome measures and lists no analyses. These design features matter when interpreting any count: without a concurrent comparison group, changes or outcomes cannot be attributed to pembrolizumab or surgery separately from other influences.

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: A total of 18 participants were consented; 3 screen failures.

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 Number of Participants With Progression Free Survival at 6 Months, measured over At 6 months.

The primary endpoint is the number of participants with progression-free survival at six months. It is a count, not a mean score or a direct measure of how long each participant remained progression-free. The registry defines progression-free survival as the time from study enrollment until the first occurrence of disease progression, relapse, or death due to disease. Response is assessed using Response Assessment in Neuro-Oncology criteria. Participants who are alive without progression or relapse are censored at their last contact.

As a general statistical concept, a fixed-time progression-free-survival count gives a snapshot at a stated follow-up point. It can be easy to communicate, but its interpretation depends on follow-up being adequate and on clear rules for determining progression, relapse, death, and censoring. Censoring means that the exact event time is not observed during the available follow-up; it does not mean that the participant had no event afterward.

The endpoint is relevant to a study of recurrent disease because it captures whether progression, relapse, or disease-related death has occurred by the specified time. However, the registry record does not report the observed six-month count, the number evaluable at that time, or an estimate with a confidence interval. The primary endpoint’s result therefore cannot be quantified from the record, and no conclusion about the magnitude or precision of the six-month outcome can be drawn.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical method that can be interpreted directly is the endpoint definition. Progression-free survival starts at study enrollment and ends at the first recorded disease progression, relapse, or death due to disease. The registry specifies censoring for participants alive without progression or relapse: their follow-up is censored at the last contact. This definition identifies the event and the follow-up origin, but it does not report the statistical estimator or analysis procedure used to turn individual follow-up into a result.

As a general concept, censoring is necessary when a participant’s event time is not observed before follow-up ends. Such a participant contributes information up to the last known contact, but their eventual event status after that point is unknown. A fixed-time count such as the number progression-free at six months depends on follow-up and event ascertainment around that time. The registry record does not report how participants lost to follow-up, incomplete assessments, or uncertain progression dates were handled beyond the stated censoring rule.

The record names Response Assessment in Neuro-Oncology criteria for evaluating response. This tells readers which response framework is identified, but it does not provide operational details about review procedures, independent assessment, adjudication, or the timing of assessments. Those details are not reported in the registry record. The record also lists overall survival and response rate as secondary outcomes, but does not give their analysis definitions or numerical results in the information available.

General statistical interpretation distinguishes a point estimate from its uncertainty. A point estimate summarizes an observed quantity, while a confidence interval describes a range of values compatible with the data under the method’s assumptions. No primary endpoint estimate or confidence interval is reported in the registry record, so neither the observed six-month result nor its statistical precision can be assessed. Likewise, no p-value or hypothesis-test result for the primary endpoint is reported. The record’s listing of zero analyses means that no analysis is listed there; it does not establish what analyses may have been performed outside the registry record.

The study’s single-group structure is central to interpretation. In general, an estimate from one group can describe outcomes in that group, but a treatment effect is a contrast between outcomes under different treatment strategies. Without a concurrent comparison group, the record cannot directly estimate such a contrast. The study is not described as randomized, and its allocation is recorded as not applicable. The single-group findings should therefore not be interpreted as a controlled comparison or as isolating the contribution of pembrolizumab from that of surgery.

The sample-size assumptions described in the record are planning assumptions for an illustration, not the trial’s own design. The illustration specifies a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect to detect defined as a stated fraction of the observed effect. For time-to-event tables it uses a total number of events; for other outcomes it gives participants per arm. These assumptions do not describe this trial’s single-group design and should not be mistaken for an actual allocation, observed power, or evidence about the trial’s planning rationale. A sample-size calculation is a planning exercise; it cannot replace reported outcome estimates or their uncertainty.

In general, over-dispersion means that variability is greater than a simple statistical model expects. It can affect standard errors and the reliability of intervals or tests if a model assumes less variability than the data contain. The registry record does not report a statistical model or indicate that over-dispersion was assessed, so this general issue cannot be tied to a reported analysis for this study. Similarly, adjustment for prognostic factors, handling of multiple endpoints, interim analyses, and prespecified statistical tests are not reported in the registry record. Their absence from the record should not be filled by assuming standard choices.

For the safety information, the registry reports counts of participants affected and lists serious-event terms with percentages. In general, participant counts summarize how many people had an event, whereas event counts can include repeated events in the same person; these are not interchangeable. The record does not provide enough detail to reconstruct event-level totals or determine whether terms refer to unique participants or recurrent events. It also does not report a comparison group, event attribution, or the observation period for the death count in the registry record. These limits constrain interpretation of safety frequency and causality.

Finally, the record identifies five outcome measures but reports no analyses. A named outcome is not the same as a result, and an endpoint definition is not a statistical analysis plan. The choices of analysis population, missing-data procedure, confidence-interval method, and multiplicity adjustment are not reported in the registry record. Without those details and the primary endpoint’s observed data, readers cannot independently evaluate the analysis choices or quantify uncertainty.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Treatment (Pembrolizumab, Surgery)")
n      <- c(15)
events <- c(15)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Treatment (Pembrolizumab, Surgery)  15  15  100.0  78.2  100.0  79.6  100.0  79.4  99.8

11. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several outcomes or statistical tests are considered. In general, examining multiple outcomes can increase the chance of finding an apparently notable result by chance if tests are treated as independent opportunities for significance. A prespecified hierarchy or adjustment can address that concern, but the appropriate approach depends on the study’s aims and analysis plan. The registry lists five outcome measures, including progression, overall survival, response rate, and adverse events, but multiplicity procedures and a testing hierarchy are not reported in the registry record. The record also lists zero analyses, so no multiplicity-adjusted result can be interpreted from it.

The primary endpoint description states that participants alive without progression or relapse are censored at last contact. Other missing-data rules are not reported in the registry record. In particular, the record does not specify how missing six-month assessments, incomplete follow-up, or unavailable response evaluations were handled, or whether sensitivity analyses were conducted. General statistical practice treats missingness and censoring as distinct issues: censoring is part of the stated time-to-event definition, while missing-data methods concern absent measurements or unavailable outcome information.

The analysis population is not reported in the registry record. It is therefore unclear from this record which enrolled participants were included in each endpoint analysis, or how exclusions and deviations were handled. Although 18 participants were enrolled, the safety summary identifies 15 participants at risk; the record does not explain the relationship between those counts. Neither count should be assumed to be the analysis denominator for the primary endpoint. These gaps prevent a full assessment of how missing outcome information, eligibility for analysis, or multiple outcomes may affect the reported findings.

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 1% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hydrocephalus (33.3%), muscle weakness left-sided (20.0%), edema cerebral (6.7%), wound dehiscence (6.7%), neoplasms benign malignant and unspecified (incl and polyps (other) (6.7%).

The registry’s safety summary identifies 15 participants at risk in the treatment group. Five are recorded as affected by serious adverse events, 12 by other adverse events, and 14 deaths are listed. These counts are separate categories in the record; they should not be combined or assumed to represent mutually exclusive groups. The time period for these counts and the relationship between the deaths and treatment are not reported in the registry record.

The record lists hydrocephalus at 33.3%, muscle weakness left-sided at 20.0%, and edema cerebral, wound dehiscence, neoplasms benign malignant and unspecified (incl and polyps (other), and enterocolitis infections each at 6.7% among the top serious-event terms. These are registry-listed percentages. The record does not provide event-specific timing, severity details beyond the serious-event classification, or causal attribution. A listed adverse event does not establish that treatment caused it.

As a general concept, safety summaries describe events observed and recorded in a study; without a comparison group, they cannot show whether event frequency is higher or lower than under another treatment. The safety figures should therefore be read as descriptive counts, not as comparative risk estimates. The record does not explain why the safety denominator is 15 while enrollment is 18.

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

The record describes a phase II, single-group study of pembrolizumab and surgery for recurrent glioblastoma or recurrent gliosarcoma. It specifies a six-month progression-free-survival endpoint, defines progression-free survival, and gives a censoring rule for participants who are alive without progression or relapse. It also provides descriptive safety counts and a list of serious-event terms. These details help explain what the study set out to examine and how some outcomes were defined.

The record does not report the observed number of participants with progression-free survival at six months. It also does not give a numerical estimate or confidence interval for that endpoint. Thus, the endpoint’s observed magnitude and precision cannot be assessed from the information recorded. Naming an endpoint or enrolling participants does not establish what the endpoint result was.

Because there is no separate control group, the study cannot directly compare the regimen with another treatment strategy. Even if an outcome count were available, an outcome observed in one group would not by itself show what would have happened to comparable participants under an alternative strategy. The combined treatment description also means the design does not isolate the effect of pembrolizumab from surgery.

The safety counts show what the registry lists for the group at risk, but the record does not report whether events were attributed to treatment or when the death count occurred. These counts are not a causal assessment. The record also does not report analysis methods, missing-data procedures beyond the endpoint’s stated censoring rule, or multiplicity handling. Consequently, it does not support a full evaluation of statistical uncertainty or analysis robustness.

In general, a trial record is most informative when design, outcome definitions, analysis methods, and numerical results can be considered together. Here, the record supports a careful account of design and endpoint meaning, while leaving key outcome and analysis details unreported. The findings should not be expanded beyond those documented details.

14. Why This Trial Matters Statistically

This study illustrates why design and result availability matter when reading clinical-trial statistics. Its single-group design is not a randomized comparison, so the core statistical question is descriptive: what outcomes occurred in the enrolled group? A treatment contrast requires outcomes under a comparator strategy, and no such group is listed. Recognizing that distinction prevents a single-group outcome from being mistaken for an estimate of comparative treatment effect.

The primary endpoint also illustrates the difference between a well-defined question and a reported answer. The registry defines progression-free survival from enrollment to progression, relapse, or disease-related death and states how some participants are censored. Yet the observed six-month count and its uncertainty are not reported in the registry record. Without those results, the endpoint definition cannot establish the observed outcome or its precision.

Safety reporting raises a related issue: denominators and event descriptions need context. The record lists 15 participants at risk for safety, while enrollment is 18, and it does not explain the relationship between these figures. It reports participant counts and serious-event terms, but not attribution or event timing for the deaths. Statistical reading requires preserving those distinctions instead of inferring causes or assuming that every count uses the same population.

Finally, the record lists multiple outcomes but no analyses. Multiplicity handling, missing-data procedures, and analysis populations are not reported. These omissions matter because analysis choices can affect how results are estimated and interpreted. The study is therefore a useful example of the limits of what can be concluded from a registry description when key numerical results and statistical methods are absent.

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 most important limitation for interpreting this study record is that the primary endpoint’s observed result is not reported in the registry record. The six-month progression-free-survival count, an estimate of progression-free survival, and an interval expressing uncertainty are not available in the registry record described. The record therefore does not support a quantitative assessment of the primary outcome.

The single-group design has no like-for-like control comparison. This limits causal interpretation and prevents a direct estimate of how outcomes differ from another treatment strategy. The intervention combines pembrolizumab and surgery, so the design also does not separate the contributions of those components. Masking is listed as none, and the record does not report independent outcome review or adjudication procedures.

The registry lists 18 enrolled participants, while the safety group has 15 participants at risk. The reason for the difference is not reported in the registry record. The record also does not report the primary endpoint analysis population, missing-data methods beyond its stated censoring rule, or how incomplete assessments were handled. The meaning and observation period of the 14 recorded deaths are not explained in the registry record.

Further limitations include the absence of reported statistical tests, confidence intervals, adjustment for multiple outcomes, interim analyses, and a statistical analysis plan. These details are not reported in the registry record and should not be inferred. The treatment description itself ends mid-sentence, limiting the available account of treatment conditions. Together, these gaps make it possible to describe the study’s stated design and outcomes, but not to assess the primary result’s precision, robustness, or causal meaning.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Pembrolizumab in Treating Patients With Recurrent Glioblastoma — NCT02337686.
• Registry record last updated 2026-09-03; status last verified 2026-08.
• Results first posted 2023-01-11.
• 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

NCT02337686 is a phase II, single-group study titled “Pembrolizumab in Treating Patients With Recurrent Glioblastoma,” with the official title “Pharmacodynamic Study of Pembrolizumab in Patients With Recurrent Glioblastoma.” M.D. Anderson Cancer Center is the sponsor. The listed conditions are recurrent glioblastoma and recurrent gliosarcoma. Enrollment is 18; the start date is 2015-04-28, the primary completion date is 2017-05-24, and the status is active, not recruiting.

The treatment combines pembrolizumab with surgery. The primary outcome is the number of participants with progression-free survival at six months, but its observed result is not reported in the registry record. Five outcome measures are listed, and no analyses are listed. Safety reporting identifies 15 participants at risk, five affected by serious adverse events, 12 affected by other adverse events, and 14 deaths; causal attribution and the death observation period are not reported in the registry record.