Bevacizumab for Recurrent Malignant Glioma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Bevacizumab for Recurrent Malignant Glioma (NCT00271609), a randomized phase 2 study. The primary outcome is Percentage of Participants With Progression Free Survival at 6 Months..

2. Clinical Question

3. Trial Design

This was a phase II, randomized, parallel-group trial with no masking. The registry lists 88 participants enrolled and identifies recurrent high-grade gliomas and malignant gliomas as the conditions studied. The sponsor was the National Cancer Institute. The study start is listed as 2005-12, primary completion as 2011-03, and status as completed.

The posted arms are Anaplastic Glioma (AG) and Glioblastoma Multiforme (GBM). Their descriptions identify types of tumors within each group and specify bevacizumab at 10 mg/kg intravenously every 2 weeks on a 28 day cycle. The description gives a 90-minute duration for the first dose and a 30-minute duration for subsequent doses. Although the registry identifies the allocation as randomized, it does not explain how randomization was conducted or whether it assigned participants between these disease groups. Those details are not reported in the registry record.

The brief summary describes PET before treatment, at the end of the first treatment cycle, and as needed thereafter. MRI was scheduled before the first dose, within 48-96 hours after that dose, and every 4 weeks. The description also includes physical and neurological examinations and research blood collection at MRI visits. It begins to describe quality-of-life assessment, but the full procedure is not available in the registry record.

The primary outcome is the percentage of participants with progression-free survival at 6 months. At the endpoint cutoff, 31 AG participants and 48 GBM participants were evaluable. These endpoint populations differ from the enrollment figure; the reasons for participants not being included in the endpoint evaluation are not reported in the registry record.

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: Minimum washout period of 2-4 weeks required after molecularly targeted or cytotoxic agents.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: at the time of data cutoff for this endpoint, 31 patients in the AG arm and 48 patients in the GBM arm were evaluable for PFS.

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

The primary endpoint is the percentage of participants surviving without disease progression at 6 months after study entry. The registry defines progression as a 25% increase in lesions, clear worsening of any evaluable disease, or the appearance of a new lesion or site on imaging. It also counts failure to return for evaluation because of death or a deteriorating condition.

A progression-free survival endpoint combines time without documented progression with survival status at the specified assessment point. Reporting the percentage at 6 months gives a snapshot of the proportion meeting that definition at that time; it is not the same as reporting overall survival or the duration of response. In general, progression-based endpoints are used to describe disease status over a defined period and can incorporate imaging and clinical assessment. Their interpretation depends on how progression and missing assessments are defined and applied.

The posted estimates are 20.9% for AG and 29.0% for GBM. Their confidence intervals are wide, so each estimate allows considerable uncertainty about the corresponding group percentage. The endpoint is a percentage of participants with an event status, rather than a mean score or a difference in time between groups. The registry does not post a separate secondary outcome result.

7. Results Posted in the Registry

• Glioblastoma Multiforme (GBM): 29.2% (14/48) versus 19.4% (6/31) on control; risk difference 9.8 percentage points (95% CI -9.1 to 28.8), risk ratio 1.51 (0.65–3.50); unadjusted two-proportion z-test p = 0.327.

8. Statistical Interpretation

9. Statistical Methods Explained

The outcome is binary at the stated assessment: each participant is classified according to whether they meet the registry’s definition of progression-free survival at 6 months. The reported group results are percentages with 95% confidence intervals. The record labels the outcome as a percentage of participants with an event and gives progression-free survival as the event definition. This type of summary describes the fraction meeting the endpoint, rather than the average number of months without progression.

What does a confidence interval describe?

In general, a confidence interval communicates uncertainty in an estimate. It gives a range of values compatible with the observed data under the method used to construct it. A 95% confidence interval is not a guarantee that the true value lies in that particular interval, nor does it mean that 95% of participants fall within the interval. It reflects the uncertainty of the estimated percentage. The registry reports 10.3% to 42.5% for AG and 18.0% to 48.0% for GBM.

In general, interval width depends on the amount of information available and the variability of the outcome. Smaller endpoint groups tend to yield less precise estimates than larger groups, all else being equal. Here the endpoint-evaluable groups are 31 in AG and 48 in GBM. The registry does not name the interval-construction method, so the method cannot be independently characterized from the posted record.

How should the group comparison be read?

The registry presents an unadjusted risk difference for GBM compared with AG. A risk difference compares the event proportions between two groups by subtraction; the posted comparison is labelled as a risk difference, and its unadjusted p-value is 0.327. In general, an unadjusted comparison uses the group results without accounting for other participant characteristics. The registry does not report covariate adjustment or a model that accounts for prognostic factors.

A p-value is a measure of how unusual data at least as discrepant as those observed would be under a specified null model and test assumptions. It is not a measure of effect size, clinical importance, or the probability that the null model is correct. A p-value alone also does not show whether the estimates are precise. The posted group confidence intervals provide information about uncertainty in each group estimate, while a confidence interval for the comparison would describe uncertainty in the difference itself. The registry’s unadjusted comparison includes its own estimate and interval, but the analysis method used to derive them is not reported in the registry record.

Why can over-dispersion matter?

In general, over-dispersion occurs when observed variability is greater than a statistical model expects. It can make standard errors too small if the model assumes less variation than the data contain. The primary outcome is a binary percentage, and the registry does not report an over-dispersion assessment or a model requiring that adjustment. Therefore, over-dispersion should not be assumed to have been present or addressed in this trial.

What analysis details are available?

The registry labels allocation as randomized, masking as none, and the model as parallel. It does not provide the randomization method, allocation concealment details, the statistical analysis plan, or an explanation of how the randomized allocation relates to the two disease groups. It also does not report whether analyses were adjusted, whether endpoint assessment was blinded, or whether any interim analyses were performed. Those details are not reported in the registry record.

The primary endpoint description defines how progression and certain failures to return for evaluation are treated. However, the record does not explain how other missing assessments were handled, whether all participants were analyzed according to an assigned group, or how any departures from scheduled assessments were addressed. The posted endpoint denominators therefore identify who was evaluable at the cutoff but do not by themselves describe the full analysis population or missing-data procedure.

The registry lists two outcome measures and no separate analysis entries. It posts no secondary outcome results. The absence of posted details does not show that a method was or was not used; it means that the relevant information is not reported in the registry record. Statistical conclusions should accordingly remain close to the available descriptive estimates and the explicitly labelled unadjusted comparison.

10. Confidence Intervals and Precision

The AG estimate is 20.9%, with a 95% confidence interval of 10.3% to 42.5%. The GBM estimate is 29.0%, with a 95% confidence interval of 18.0% to 48.0%. Both intervals span a broad range of possible group percentages. This means the point estimates should not be treated as highly precise values for the underlying populations.

Confidence intervals are useful because they place estimates in the context of sampling uncertainty. In general, a narrower interval indicates greater precision under the method used, while a wider interval indicates more uncertainty. The intervals for the two groups overlap. Such overlap does not, by itself, determine whether the groups differ: assessment of a difference depends on the comparison estimate and its uncertainty, not merely on visual comparison of the two separate intervals.

The unadjusted comparison has a p-value of 0.327. That value does not quantify the size or importance of a possible difference, and it does not demonstrate equivalence. The registry also posts a comparison estimate and interval, but its calculation method is not reported in the registry record. No adjusted estimate is posted.

Precision is affected by the endpoint-evaluable group sizes, which were 31 for AG and 48 for GBM. The record does not provide enough detail to assess the impact of participants not included in those endpoint groups. As a result, the intervals describe uncertainty in the posted estimates, while uncertainty related to missing endpoint information and analysis choices cannot be fully evaluated from the registry record.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are participants per arm, not total event counts. The calculation is a planning illustration based on the observed comparison, not a report of the trial’s original design assumptions. Those original assumptions are not reported in the registry record.

The listed planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set to a stated fraction of the observed effect. For planning at 100% of the observed effect, the figures are 299 participants per arm for 80% power and 400 participants per arm for 90% power. Planning for 75% of the observed effect gives 513 participants per arm for 80% power and 686 participants per arm for 90% power. Planning for 50% of the observed effect gives 1,111 participants per arm for 80% power and 1,487 participants per arm for 90% power.

In general, power is the probability that a study will identify an effect of the specified size as statistically detectable under the planning assumptions, if that effect is present. The 80% and 90% columns therefore describe different target probabilities for detecting the assumed effect. They are planning targets, not guarantees about a future study’s result. The calculation uses equal allocation and does not establish that the trial itself used those assumptions.

Smaller effects are harder to distinguish from random variation, so planning to detect a smaller fraction of the observed effect requires more participants. These figures are not observed power.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.194; treatment proportion p₁ = 0.292; average p̄ = 0.243.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6062 = 1.1882.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6022 = 0.5069.
4. n = (1.1882 + 0.5069)² / (0.0981)² = 298.4, rounded up to 299 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.193548387096774                          # control proportion
p1 <- 0.291666666666667                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(271609)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 299 (80% power), 400 (90% power)
 75% of observed effect: n per arm = 513 (80% power), 686 (90% power)
 50% of observed effect: n per arm = 1,111 (80% power), 1,487 (90% power)
power.prop.test: 298.4 
Simulated power at n = 299 per arm: 0.800 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures and posts one primary endpoint result. It reports no secondary outcome results. The record does not describe a multiplicity strategy, such as whether or how the chance of false-positive findings across several outcomes or comparisons was controlled. The original statistical analysis plan is not reported in the registry record.

The record identifies an unadjusted comparison between the GBM and AG groups. It does not report adjusted analyses, prespecified subgroup analyses, or a hierarchy of statistical tests. It also does not report interim analyses or stopping rules. Without these details, it is not possible to determine from the registry record how the posted comparison fits into any broader testing plan.

For the primary endpoint, the record gives the numbers evaluable at the cutoff: 31 in AG and 48 in GBM. Its description states that failure to return for evaluation because of death or deteriorating condition is counted in the progression definition. It does not explain how other missing assessments, withdrawals, or deviations from the assessment schedule were handled. The reasons for participants not appearing in the endpoint-evaluable groups are not reported in the registry record.

The registry does not describe an intention-to-treat population, a per-protocol population, or another formal analysis population. It also does not state whether all randomized participants were analyzed in their assigned groups. The enrollment figure and endpoint-evaluable counts should therefore not be treated as interchangeable descriptions of the analysis population.

In general, missing outcome information can affect both the denominator and the estimate, depending on why information is missing and how the analysis handles it. No particular missing-data mechanism or method should be assumed for this trial. The posted estimates are best interpreted alongside the registry’s endpoint-evaluable counts and the acknowledged absence of further analysis details.

14. Safety Results

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

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: thrombosis/thrombus/embolism (12.5%), seizure (12.5%), aspiration (9.4%), hypertension (3.6%), proteinuria (3.6%).

The registry lists serious adverse event information separately for AG and GBM. For AG, 32 participants were at risk for serious adverse events and 11 were affected. For GBM, 56 were at risk and 24 were affected. For other, non-serious adverse events, the corresponding figures are 32 at risk and 32 affected in AG, and 56 at risk and 51 affected in GBM. These are counts, not percentages calculated. Death counts are not reported in the registry record.

The registry also lists serious adverse event terms with percentages: thrombosis/thrombus/embolism, 12.5%; seizure, 12.5%; aspiration, 9.4%; hypertension, 3.6%; proteinuria, 3.6%; and CNS cerebrovascular ischemia, 3.6%. The record does not provide enough context to infer from this list how terms were grouped, whether a participant could contribute to more than one term, or how the listed percentages should be assigned to each arm.

In general, adverse event counts describe events recorded during a study and do not by themselves establish that the study treatment caused them. The registry’s safety figures should be read as reported event information, distinct from the progression-free survival endpoint. No additional safety analysis or causal attribution is reported in the registry record.

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

The posted results describe six-month progression-free survival estimates for two tumor groups treated with bevacizumab as described in the registry. The estimate is 20.9% for AG and 29.0% for GBM. The associated confidence intervals are broad, indicating that the underlying percentages are uncertain. The unadjusted comparison has a p-value of 0.327, which does not establish that the group estimates are identical, nor does it establish a reliable difference between them.

The results do not provide a direct comparison of bevacizumab with placebo, no treatment, or another therapy. Both posted arm descriptions specify bevacizumab, and the groups are defined by tumor type. Consequently, a difference between the group estimates cannot be interpreted as the effect of bevacizumab. The registry’s randomized allocation label does not resolve this issue: the method of randomization and its relationship to the two disease groups are not reported in the registry record.

The endpoint is also specific. It describes whether participants were alive without progression at 6 months under the registry’s definition. It is not an estimate of overall survival, long-term disease control, or quality of life. The record does not post a secondary outcome result, and the brief summary’s description of quality-of-life procedures does not include a posted result.

In general, an estimate should be interpreted in light of its precision, population, endpoint definition, comparison, and analysis. Here, endpoint-evaluable counts are reported, but the reasons for missing endpoint information and its handling are not. Adjustment methods and the original analysis plan are also not reported. The posted statistics therefore support a description of the recorded endpoint results, not a broader causal or clinical conclusion.

16. Why This Trial Matters Statistically

This trial illustrates why a percentage alone is not enough to assess a clinical result. The AG and GBM point estimates are 20.9% and 29.0%, but their confidence intervals extend across broad ranges. Considering the intervals alongside the point estimates makes the uncertainty visible and discourages overly precise interpretation.

It also demonstrates the importance of understanding what groups are being compared. The registry identifies randomized allocation, while the posted arms are disease categories and both describe bevacizumab. Without an explanation of the allocation procedure or a separate control group, the comparison between these groups cannot answer the same question as a randomized comparison of treatments. Labels such as “randomized” do not replace a clear account of the comparator and analysis population.

The endpoint definition matters as well. Progression-free survival at 6 months depends on imaging and clinical criteria and includes failure to return because of death or deteriorating condition. Readers need that definition to understand what the percentage represents and what it does not represent.

Finally, the record shows why missing-data and analysis details are important for interpretation. It reports endpoint-evaluable counts but not the reasons other enrolled participants were not included or the methods used to handle missing information. Statistical interpretation is strongest when estimates, uncertainty, group definitions, and analysis procedures can be considered together. Where some of these details are absent, conclusions should remain appropriately limited.

17. 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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The registry gives limited detail on how the randomized allocation was conducted and how it relates to the AG and GBM groups. Both arm descriptions specify bevacizumab, and no separate control treatment is described. This restricts interpretation of the group comparison as a treatment comparison.

The primary endpoint results use 31 evaluable AG participants and 48 evaluable GBM participants at the cutoff, while the registry lists 88 enrolled. The reasons for the difference between enrollment and endpoint evaluation are not reported in the registry record. The record also does not state how missing assessments other than the specified failure to return because of death or deteriorating condition were handled.

The statistical analysis plan, interval-construction method, covariate adjustment, interim analyses, and multiplicity approach are not reported in the registry record. The posted comparison is unadjusted. No secondary outcome results are posted, and the registry does not report an analysis population definition beyond the endpoint-evaluable counts.

The brief summary describes a quality-of-life procedure but does not give its full details or a result. Safety reporting includes serious and other adverse event counts and a list of serious event terms, but deaths are not reported. Further context for the listed serious-event percentages is not given. These limitations do not invalidate the posted figures, but they narrow the conclusions that can be drawn from them.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Bevacizumab for Recurrent Malignant Glioma — NCT00271609.
• Registry record last updated 2014-04-29; status last verified 2014-04.
• Results first posted 2012-06-08.
• 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.

21. Record Summary

The registry record describes a completed phase II, randomized, parallel-group trial of bevacizumab for recurrent high-grade gliomas, with no masking. It lists 88 participants enrolled and identifies AG and GBM groups. Both arm descriptions specify bevacizumab administration.

The primary endpoint is the percentage with progression-free survival at 6 months. The estimates are 20.9% for AG and 29.0% for GBM, with 95% confidence intervals of 10.3% to 42.5% and 18.0% to 48.0%, respectively. The registry posts an unadjusted comparison with p = 0.327. Endpoint-evaluable counts were 31 for AG and 48 for GBM.

The record does not describe a separate control group, the randomization procedure, the original analysis plan, or the handling of missing endpoint information. No secondary outcome results are posted. These gaps limit interpretation of the comparison and the broader conclusions that can be drawn from the registry results.