Comparing Whole Brain Radiotherapy Using a Technique That Avoids the Hippocampus to Stereotactic Radiosurgery in Patients With Cancer That Has Spread to the Brain and Come Back in Other Areas of the Brain After Earlier Stereotactic Radiosurgery: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Comparing Whole Brain Radiotherapy Using a Technique That Avoids the Hippocampus to Stereotactic Radiosurgery in Patients With Cancer That Has Spread to the Brain and Come Back in Other Areas of the Brain After Earlier Stereotactic Radiosurgery (NCT04588246), a randomized phase 3 study. The primary outcome is Percentage of Participants With Neurologic Death.

2. Clinical Question

3. Trial Design

NCT04588246 was a randomized, parallel-group phase III trial sponsored by NRG Oncology. The registry lists single masking, but which people were masked is not reported in the registry record. Details of the randomization method, allocation concealment, and any stratification are also not reported in the registry record. Enrollment was 19 participants: 10 in the HA-WBRT and memantine arm and 9 in the SRS/fSRS arm.

The study began on 2021-08-05, had a primary completion date of 2024-08-06, and is listed as terminated. Maximum follow-up at study termination was 29.7 months. The record does not report a reason for termination beyond stating that the study closed early. The primary outcome description says its planned analysis would take place after 127 neurologic death events overall. The actual record reports only the number of participants with neurologic death, explaining that the small number of participants due to early study closure meant no statistical testing was done.

The intervention arm combined HA-WBRT and memantine, while the comparator arm received SRS or fSRS. Consequently, the randomized contrast concerns these treatment strategies as assigned, not the isolated effect of memantine or hippocampal avoidance. The registry lists imaging, biospecimen collection, quality-of-life assessment, and questionnaire administration among interventions or study activities. It does not report enough detail to describe how these activities were scheduled or used in the analysis.

Six secondary outcomes are listed: overall survival, intracranial progression-free survival, brain metastasis velocity at subsequent relapse, change from baseline in the M.D. Anderson Symptom Inventory Brain Tumor Module symptom severity score, participants with a grade 3 or higher adverse event, and participants receiving salvage procedures. Although the primary outcome and secondary outcomes are described, no statistical analyses are posted for the secondary outcomes.

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

Analysis population for the primary outcome, as defined in the registry: Randomized participants

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: Randomized participants

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants With Neurologic Death, measured over From baseline to neurologic death or last follow-up. Maximum follow-up at time of study termination was 29.7 months..

The primary endpoint is titled Percentage of Participants With Neurologic Death. Neurologic death is defined in the registry as progressive neurologic decline or new neurologic symptoms or signs at death, irrespective of extracranial disease status, or death from inter-current disease in a participant with severe neurologic dysfunction. The endpoint is intended to capture death attributed to neurologic deterioration in the setting of brain metastases, rather than all deaths regardless of cause.

In general, a cause-specific endpoint can help distinguish deaths related to the disease process of interest from deaths due to other causes. The registry describes non-neurologic deaths as competing risks: such deaths prevent a later neurologic death from being observed. Its planned approach was to estimate neurologic death rates using the cumulative incidence method, treating non-neurologic deaths as competing risks and censoring participants who were alive at analysis.

The timeframe was from baseline to neurologic death or last follow-up, with maximum follow-up at termination of 29.7 months. The intended primary comparison was a one-sided 0.05-level test of the cause-specific hazard ratio in a Cox proportional hazards model, after 127 neurologic death events overall. That planned analysis was not carried out in the posted results. The record gives a count of 3 neurologic deaths in each arm, not estimated cumulative incidence, a hazard ratio, a confidence interval, or a test result. It therefore provides a descriptive endpoint count, rather than an inferential comparison of neurologic death over time.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes a planned time-to-event analysis for neurologic death, but the posted primary result is descriptive. This distinction matters: a protocol-level statistical method can explain how investigators intended to analyze an endpoint without implying that the method was applied to the reported data. The registry explicitly says that no statistical testing was done because the study closed early with a small number of participants.

Why use a time-to-event approach?

In general, time-to-event methods consider both whether an event occurred and the time until it occurred. They can also accommodate participants whose event has not been observed by the end of follow-up. The primary endpoint timeframe runs from baseline to neurologic death or last follow-up, so the timing of events and censoring would be relevant to the planned analysis. The posted record does not include event times or individual follow-up information, so those features cannot be assessed from the reported counts.

The endpoint description specifies cumulative incidence estimation. In general, cumulative incidence is used when another event can prevent the event of interest from happening later. Here, non-neurologic death is identified as a competing risk for neurologic death. Treating non-neurologic deaths as competing risks is different from treating them as though they were ordinary losses to follow-up: the competing event changes the possibility of subsequently observing neurologic death. The record states that participants alive at analysis were otherwise to be censored.

What was the planned arm comparison?

The primary comparison was described as a one-sided 0.05-level test for a cause-specific hazard ratio in a Cox proportional hazards model. In general, a hazard ratio compares the instantaneous event rates between groups among people still at risk, under the model's assumptions. A cause-specific hazard analysis treats the event of interest as the event being modeled and other causes of death as competing events. The registry does not provide a fitted model, estimated hazard ratio, confidence interval, p-value, or assessment of model assumptions. Since no test was done, the planned comparison cannot be interpreted as an observed result.

The registry says the analysis was to occur after 127 neurologic death events overall. This is a planned event threshold, not the number of events observed in the posted results. The record reports 3 participants with neurologic death in each arm and explains that early closure and the small number of participants led to reporting counts only. It does not report a revised analysis plan or an alternative event threshold.

What do uncertainty intervals and tests mean generally?

In general, a confidence interval describes the range of effect estimates compatible with the data and the method's assumptions at a stated confidence level. A statistical test assesses compatibility between observed data and a specified null hypothesis under its assumptions. Neither an interval nor a test result is reported for the primary endpoint. Without them, the precision of any between-arm effect estimate cannot be characterized. The event totals alone do not show whether a difference would be estimated precisely or imprecisely.

The primary endpoint is also described as a percentage, but only counts appear in the posted result. The registry does not provide a cumulative incidence percentage at a defined time, an adjusted or unadjusted risk comparison, or a calculation of the uncertainty around a percentage. No percentage or ratio should be inferred from the counts as a substitute for the planned analysis.

How should planning illustrations be understood?

Any sample-size planning illustrations associated with observed effects are not the trial's original design justification and are not observed power. The described illustration assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation, with the effect to detect set as a stated fraction of the observed effect. For time-to-event outcomes, the illustration uses the total number of events; other outcomes are expressed as participants per arm. Such illustrations are hypothetical planning calculations, not evidence that the trial had a particular probability of detecting an effect. The registry record does not report the original sample-size justification.

For the current primary endpoint, the posted information does not support a recalculated treatment-effect estimate. The record gives counts, but no event-time summary or uncertainty measure. Thus, there is no reported effect estimate whose precision can be recalculated from posted summary statistics. The reported sample sizes and counts should be kept distinct from any planning illustration based on an effect assumption.

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("Arm I (Memantine, HA-WBRT)", "Arm II (SRS/Fractionated Stereotactic Radiosurgery")
n      <- c(10, 9)
events <- c(3, 3)
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  Arm I (Memantine, HA-WBRT)  10  3  30.0  6.7  65.2  10.8  60.3  10.9  61.0
ARM  Arm II (SRS/Fractionated Stereotactic Radiosurgery  9  3  33.3  7.5  70.1  12.1  64.6  12.2  65.2

11. Multiplicity, Missing Data and Analysis Populations

Six secondary outcomes are listed in addition to the primary neurologic death endpoint. They cover overall survival, intracranial progression-free survival, brain metastasis velocity at subsequent relapse, symptom severity change, grade 3 or higher adverse events, and salvage procedures. No statistical analyses are posted for these outcomes. The record does not report a multiplicity adjustment, a hierarchy among the secondary outcomes, or a plan for controlling the chance of false-positive findings across multiple tests. With no secondary test results reported, the effect of multiplicity procedures on interpretation cannot be evaluated.

The primary endpoint description identifies a planned one-sided test at the 0.05 level, but also states that no statistical testing was done. The registry record does not report whether there were interim analyses, monitoring boundaries, or other adjustments to the planned test. These details should not be inferred from the stated endpoint plan.

Missing-data handling beyond censoring participants alive at analysis is not reported in the registry record. The planned endpoint description says non-neurologic deaths were to be treated as competing risks and participants alive at analysis otherwise censored. It does not describe missing outcome assessments, withdrawal handling, or any sensitivity analyses. The record identifies randomized participants as the primary population, but does not state whether the analysis was intended to follow an intention-to-treat principle, a per-protocol approach, or another defined analysis population. Arm-specific exclusions or deviations are not reported.

Accordingly, the available result does not establish the completeness of endpoint ascertainment or how any missing follow-up information was addressed. The event counts should be interpreted as the posted counts, without assuming unreported analysis-population or imputation rules.

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: blood and lymphatic system disorders - other (12.5%), leukocytosis (12.5%), small intestinal obstruction (12.5%), fatigue (12.5%), lung infection (12.5%).

The posted safety figures distinguish participants at risk from participants affected. In the HA-WBRT and memantine arm, 8 participants were at risk for the serious adverse event summary and 4 were affected; 8 were at risk for the other adverse event summary and 6 were affected. In the SRS/fSRS arm, the corresponding figures were 8 at risk and 1 affected for serious adverse events, and 8 at risk and 6 affected for other adverse events. The record also reports 5 deaths in the first arm and 4 in the second.

These are descriptive counts. They do not, by themselves, establish a causal difference in safety between the strategies. The registry does not provide a comparative statistical test or uncertainty interval for these figures. It also does not explain why the safety population at risk is 8 in each arm when the randomized arm sizes are 10 and 9; the reason is not reported in the registry record.

The listed serious adverse event terms each have a posted frequency of 12.5%: blood and lymphatic system disorders - other, leukocytosis, small intestinal obstruction, fatigue, lung infection, and fall. The record does not specify in this listing how these terms map to arms, so they should not be assigned to a particular treatment group. Counts of affected participants and terms are not necessarily interchangeable: a participant may contribute to more than one event term. The summary does not supply enough detail to assess event timing, duration, attribution, or severity beyond the serious-event designation.

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

The record establishes that NCT04588246 was a randomized phase III comparison of HA-WBRT with memantine against SRS/fSRS in participants with distant brain relapse after earlier SRS. It also establishes that the trial terminated and enrolled 19 participants. The primary posted result is 3 neurologic deaths in each arm, without a statistical test or time-to-event estimate.

Those observations do not show that the strategies have the same effect. Equal counts in groups of different sizes, with follow-up extending for different possible lengths, are not equivalent to a formal comparison of the endpoint over time. The planned analysis incorporated competing risks and was intended to compare cause-specific hazards, but the posted results do not include that analysis. No confidence interval or p-value is available to convey uncertainty about an arm difference.

The counts also do not establish that either strategy reduces neurologic death, prevents intracranial progression, or changes survival. Although these and other outcomes are listed, their analyses are not posted. The trial's stated rationale discusses possible effects of hippocampal avoidance and memantine on radiation exposure and neurocognitive side effects, but the primary event counts do not evaluate those mechanisms or establish effects on memory or quality of life.

For safety, the record reports affected-participant counts and deaths, alongside a list of serious adverse event terms. These figures are useful as descriptive observations, but the record does not provide comparative estimates or detail sufficient to attribute events to treatment. They should not be interpreted as demonstrating a safety advantage or disadvantage.

Overall, the posted results are limited: they document enrollment, intervention descriptions, the primary endpoint counts, and safety summaries, while leaving the planned inferential comparison unavailable. Conclusions beyond those descriptive details are not supported by the registry record.

14. Why This Trial Matters Statistically

This trial illustrates the difference between a statistical plan and a statistical result. The registry describes a specific primary endpoint, competing-risk handling, and a planned Cox model comparison. Yet the posted analysis consists only of event counts because the study closed early with a small number of participants. A method written into an endpoint description does not mean its estimates or tests were ultimately produced.

It also demonstrates why a raw event count is not a complete substitute for time-to-event analysis. Neurologic death can be precluded by non-neurologic death, and participants can remain alive at last follow-up. The planned use of cumulative incidence and a cause-specific hazard ratio reflects those features. Without event timing, follow-up information, or the planned model estimate, equal counts cannot answer the comparative question with known precision.

The enrollment of 19 and maximum follow-up of 29.7 months provide important context for the scope of the posted findings. The planned analysis threshold was 127 neurologic death events overall, while the result reports only 3 neurologic deaths in each arm. This makes clear why the record does not present the planned test. It also cautions against reading a lack of inferential results as a null finding.

Finally, the record separates a broad set of planned clinical outcomes from the subset of results actually posted. Six secondary outcomes are listed, but no analyses are available for them. For readers assessing a clinical trial statistically, this distinction helps identify which questions can be answered from the record and which remain unresolved. The primary counts and safety summaries are informative descriptions of the enrolled participants, but they do not deliver a comparative estimate of treatment effect.

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 principal limitation is that the planned primary statistical comparison was not performed. The registry reports counts of neurologic death, but no cumulative incidence estimate, cause-specific hazard ratio, confidence interval, or p-value. The small enrollment and early study closure are stated as the reason that only counts were reported. The record does not provide a reason for termination beyond early closure.

Follow-up information is limited to the stated maximum of 29.7 months. Arm-specific follow-up distributions, event timing, and counts of participants censored alive are not reported. These omissions prevent an assessment of how much observation time contributed to the primary counts or whether follow-up differed between arms. The record also does not report details on outcome adjudication or completeness of neurologic-death classification beyond the endpoint definition.

Six secondary outcomes are named, but no analyses are posted. As a result, there are no reported comparative results for survival, intracranial progression-free survival, relapse velocity, symptom severity, adverse events of grade 3 or higher, or salvage procedures. The safety summaries give affected counts and deaths, but the reason that the safety at-risk group contains 8 participants in each arm is not reported in the registry record. The serious adverse event term list is not assigned to arms in the posted information.

Important design and analysis details are also absent from the registry record, including who was masked, the randomization procedure, allocation concealment, the original sample-size justification, interim analyses, and detailed missing-data procedures. The primary population is identified as randomized participants, but the analysis-population rule is not further described. These limitations constrain both interpretation of the observed counts and assessment of the intended analysis.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Comparing Whole Brain Radiotherapy Using a Technique That Avoids the Hippocampus to Stereotactic Radiosurgery in Patients With Cancer That Has Spread to the Brain and Come Back in Other Areas of the Brain After Earlier Stereotactic Radiosurgery — NCT04588246.
• Registry record last updated 2026-05-20; status last verified 2025-04.
• Results first posted 2025-06-06.
• 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

NCT04588246 is a randomized, parallel-group phase III trial sponsored by NRG Oncology, comparing HA-WBRT with memantine against SRS/fSRS for distant brain relapse after earlier SRS. The record lists single masking, enrollment of 19, a start date of 2021-08-05, a primary completion date of 2024-08-06, and terminated status.

The primary endpoint was neurologic death from baseline to death or last follow-up. The posted counts are 3 in the HA-WBRT and memantine arm and 3 in the SRS/fSRS arm. The planned cause-specific hazard test was not performed because the study closed early with a small number of participants; maximum follow-up at termination was 29.7 months.

Six secondary outcomes are listed, with no analyses posted. Safety summaries report affected-participant counts, deaths, and serious adverse event terms. The record does not provide an inferential treatment comparison for the primary endpoint or the listed secondary outcomes.