Radiation Therapy Combined With Chemotherapy in Treating Patients With Anaplastic Astrocytoma or Mixed Gliomas: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Radiation Therapy Combined With Chemotherapy in Treating Patients With Anaplastic Astrocytoma or Mixed Gliomas (NCT00004259), a randomized phase 3 study. The primary outcome is (Phase III) Overall Survival (OS).

2. Clinical Question

The trial was a randomized Phase III study. It compared treatment strategies that shared radiation therapy but used different chemotherapy regimens. The registry describes temozolomide as the study drug in one group and BCNU or CCNU in the comparison group. The purpose was to assess the relative outcomes of these assigned strategies, not to compare chemotherapy with no chemotherapy.

The primary outcome was time to death from any cause. The registry also reports time to tumor progression and outcomes by MGMT status, as well as toxicity. These additional measures address different questions: survival, disease progression, outcome patterns within biomarker groups, and harms. Their interpretations should therefore remain specific to each outcome. The registry record does not report enough detail to establish whether the secondary analyses were adjusted for multiple comparisons or how all secondary endpoints were defined and analyzed.

3. Trial Design

This was a randomized, open-label, parallel-group Phase III study sponsored by the Radiation Therapy Oncology Group. The registry reports enrollment of 230 and a completed status. The study opened in 2000-06, with primary completion recorded as 2015-02-01. The trial title and summary describe patients with anaplastic astrocytoma or mixed gliomas; the official title specifies anaplastic astrocytoma and mixed anaplastic oligoastrocytoma with astrocytoma dominance.

The principal comparison was radiation therapy with temozolomide versus radiation therapy with BCNU or CCNU. In the temozolomide group, radiation therapy was given for 6 weeks concurrently with treatment, followed by temozolomide 200mg/m2 for twelve 28-day cycles. In the BCNU/CCNU group, radiation therapy was given for 6 weeks concurrently with chemotherapy, followed by BCNU 80mg/m2 or CCNU 130 mg/m2 for six 8-week cycles. The registry also lists pilot arms that combined radiation therapy, temozolomide, and BCNU. Per the protocol description in the registry, those pilot arms were not included in the Phase III analyses.

The posted Phase III primary outcome results include 97 patients in the temozolomide group and 99 in the BCNU/CCNU group. These outcome-analysis group sizes do not replace the overall enrollment figure: the registry separately identifies pilot arms and says they were excluded from Phase III analyses. The record does not report the randomization procedure, concealment of allocation, original sample-size justification, or detailed protocol for the Phase III analysis. Masking is recorded as none. That makes knowledge of assigned treatment possible for participants and study personnel, though the record does not describe how this may have affected outcome assessment.

The overall survival analysis used the Kaplan-Meier method, with patients last known to be alive censored at their last contact. The registry says analysis was to occur after 155 deaths had been reported, estimated at 5.5 years from study opening. That description is part of the stated outcome timeframe; it does not establish which interim analyses, if any, were performed. Such details 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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Eligible randomized patients

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: Eligible patients

6. Primary Endpoint

The registry lists the primary outcome as (Phase III) Overall Survival (OS), measured over From randomization to date of death. Patients are followed until death. Analysis occurs after 155 deaths have been reported, estimated at 5.5 years from the study opening..

The primary endpoint was overall survival, defined as the time from randomization to death from any cause. It measures how long patients remain alive after assignment to a treatment group, regardless of the cause of death. The registry reports that survival was estimated using the Kaplan-Meier method and that patients last known to be alive were censored at the date of last contact.

Overall survival is a time-to-event outcome: the analysis uses both the timing of deaths and the follow-up available for patients who had not died by their last contact. Kaplan-Meier estimation accommodates this censoring under assumptions about how censoring relates to the chance of the event. A median survival estimate is the time by which the estimated survival curve reaches the halfway point. It describes the middle of the estimated survival distribution; it does not mean every patient lived for that duration.

The registry reports median overall survival in years for each Phase III group, with 95% confidence intervals, and a hazard ratio for the group comparison. The hazard ratio is a relative summary of event rates over follow-up under the analysis model and should not be read as a direct difference in median survival or as an individual patient's prognosis. The reported primary comparison used a log-rank method. The registry record does not report the full model specification or checks of its assumptions.

Overall survival is used because it is a direct, clinically meaningful time-to-event outcome and does not require deciding in every case whether a tumor has progressed. Its interpretation still depends on follow-up, censoring, and the analysis population. The registry does not provide detailed information about completeness of follow-up or the handling of missing endpoint data beyond its statement about censoring at last contact.

7. Results Posted in the Registry

• Radiation Therapy + Temozolomide (TMZ) vs RT + BCNU/CCNU: hazard ratio 0.94 (95% CI 0.67–1.32), reported p = 0.360 (as reported in the registry).

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

(Phase III) Time to Tumor Progression (TTP)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Gray's test); it was not recalculated.
• Result:
  • Radiation Therapy + Temozolomide (TMZ) vs RT + BCNU/CCNU: hazard ratio 0.80 (95% CI 0.55–1.16), reported p = 0.460 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Radiation Therapy + Temozolomide (TMZ) vs RT + BCNU/CCNU and the control. The interval gives the range of effects that remain compatible with the data.

(Phase III) Survival Time by MGMT Status

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Methylated MGMT vs Unmthylated MGMT: hazard ratio 1.78 (95% CI 0.93–3.40), reported p = 0.080 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Methylated MGMT vs Unmthylated MGMT and the control. The interval gives the range of effects that remain compatible with the data.

(Phase III) Progression-free Survival by MGMT Status

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Methylated MGMT vs Unmthylated MGMT: hazard ratio 1.29 (95% CI 0.70–2.35), reported p = 0.410 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Methylated MGMT vs Unmthylated MGMT and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes extend the primary survival question in several directions: tumor progression, serious toxicity, and survival or progression-free survival by MGMT status. Together, they offer additional descriptions of disease course and subgroup patterns, but they do not change the primary endpoint's role in the trial's main comparison.

The posted secondary results do not tell a single uniform story that can be collapsed into a conclusion about treatment. Time to tumor progression is a distinct endpoint from overall survival, and MGMT subgroup analyses concern a different comparison from the randomized treatment groups. Subgroup results should not be treated as evidence that treatment effects vary by biomarker status unless a treatment-by-subgroup interaction has been assessed. Such an interaction result is not reported in the registry record.

Secondary tests are generally less conclusive when many outcomes or subgroup comparisons are examined without adjustment. A chance finding becomes more likely as the number of tests increases. The registry lists multiple secondary analyses, but does not report a multiplicity strategy, a hierarchy of confirmatory endpoints, or which results were prespecified as confirmatory. Therefore, unadjusted secondary p-values carry less weight than the primary analysis and should be read as exploratory evidence rather than independent confirmation.

The registry record also describes the time-to-tumor-progression timeframe in terms of death, which does not fully clarify the operational definition of that endpoint. The record does not report enough detail to resolve that description. This uncertainty reinforces the need to distinguish the registry's stated endpoint label from a fully specified analysis definition.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• (Phase III) Time to Tumor Progression (TTP)
• (Phase III) Number of Patients With Grade 3 or Higher Toxicity
• (Phase III) Survival Time by MGMT Status
• (Phase III) Progression-free Survival by MGMT Status

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is a time-to-event outcome. Each patient's observed information consists of the time from randomization to death, or the time to last known contact if death had not been recorded by then. The latter observation is censored: it contributes information up to the last contact without being counted as a death at that time. The registry says this censoring rule was used for overall survival.

What does Kaplan-Meier estimation do?

As a general statistical explanation, the Kaplan-Meier method estimates the probability of remaining free of the event over time while allowing follow-up to end at different times for different participants. At each event time, the estimate uses the number still under observation and at risk immediately before that event. Censored observations contribute while they are followed, then leave the risk set. This approach relies on censoring being sufficiently non-informative: after accounting for relevant factors, those lost to observation should not have systematically different event prospects from those remaining under observation. The registry record does not report an assessment of that assumption.

A median from a Kaplan-Meier curve is the time where the estimated probability of remaining alive reaches the median point. The confidence interval expresses uncertainty in that estimate. It is not an interval containing a fixed fraction of individual survival times, and it does not imply that every patient in a group had survival within its limits. Medians summarize one feature of the curve; different curves can have similar medians but differ at other times.

What does the hazard ratio summarize?

For a time-to-event comparison, a hazard is the instantaneous event rate among patients who have not yet experienced the event and remain under observation. A hazard ratio compares that rate between groups according to the analysis method. The sponsor-reported primary hazard ratio is 0.94 for radiation therapy plus temozolomide versus radiation therapy plus BCNU/CCNU. A value near 1 is consistent with similar hazards under the model, but the estimate alone does not show whether hazards were proportional over time. The registry reports the hazard ratio and log-rank test, but the full statistical model, its assumptions, and diagnostic checks are not reported in the registry record.

The hazard ratio is not a ratio of median survival times, nor is it a direct measure of the number of additional years an individual might live. It is a group-level relative summary. If the treatment contrast changes over follow-up, a single hazard ratio may be difficult to interpret as a constant effect. General statistical practice therefore treats the model assumptions as part of interpretation, rather than assuming them from the reported estimate.

How does the log-rank test contribute?

The log-rank test compares event experience between groups over follow-up, using the event timing and risk sets. The registry identifies it as the method for the primary hazard-ratio analysis. A p-value measures how unusual the observed test result, or a more extreme one, would be under the test's null hypothesis and assumptions. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The reported p-value of 0.36 should therefore be read alongside the hazard ratio and its confidence interval.

The registry also identifies Gray's test for the time-to-tumor-progression comparison. As a general explanation, Gray's test is used for comparing cumulative incidence when competing events may prevent the event of interest from occurring. The registry does not explain how competing events were defined or handled in the endpoint record, so the test's detailed application cannot be reconstructed from the available description.

Why can over-dispersion matter?

As a general concept, over-dispersion occurs when observed variability exceeds what a statistical model expects. It can make standard errors too small and intervals too narrow if the model does not accommodate that extra variability. The posted primary outcome is a time-to-event analysis, and the registry record does not describe a count model or report an over-dispersion assessment for this endpoint. This general issue should not be mistaken for a reported problem in this trial.

The record describes a Kaplan-Meier estimate, a log-rank comparison, and the sponsor-reported hazard ratio. It does not report details of covariate adjustment, stratification, missing-data sensitivity analyses, or model diagnostics. Those omissions limit independent assessment of how robust the reported estimates are to alternative assumptions. The numbers should be understood as results of the stated analysis, with those methods details left unspecified.

11. Confidence Intervals and Precision

A confidence interval describes the range of estimates reasonably compatible with the observed data under the method's assumptions. It communicates uncertainty in the estimate; it is not a guarantee that the true value lies within its limits. A 95% confidence interval is produced by a procedure designed to cover the fixed underlying quantity in 95% of repeated applications under the model. It does not mean there is a 95% probability that this particular interval contains the true effect.

For overall survival, the reported median is 3.9 years in the temozolomide group, with a 95% confidence interval of 3.0 to 7.0 years. In the BCNU/CCNU group, the median is 3.8 years, with a 95% confidence interval of 2.2 to 7.0 years. The interval limits communicate uncertainty in each median separately. Their overlap is not itself a formal test of the between-group difference.

The primary hazard ratio is 0.94, with a 95% confidence interval of 0.67 to 1.32. This interval includes 1 and permits a range of possible relative hazard effects. Its breadth means the estimate does not pin down the comparison precisely. The p-value of 0.36 is consistent with the same broad reading: the data do not strongly distinguish the observed result from the null under the reported test. Neither statistic proves that the groups are equal.

Precision depends on the amount and timing of observed event information, as well as the analysis method. For time-to-event outcomes, follow-up and the number of events are important; the registry states that the analysis was to occur after 155 deaths had been reported. It does not report a complete account of event accrual, censoring patterns, or the original precision target. Thus the displayed confidence interval is the most direct posted indication of uncertainty, while the registry record leaves some factors affecting that precision undescribed.

12. 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 illustrations based on the observed effect, not a report of the trial's original sample-size design. For the time-to-event comparisons, figures refer to total events, not participants per arm. The 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 the stated fraction of the observed effect. The original design assumptions are not reported in the registry record.

The 80% power column means the illustrated design would have an 80% chance of detecting the specified effect under those assumptions, if that effect were the true one. The 90% power column describes a higher chance of detecting that same specified effect, again under the assumptions. Greater power generally requires more information, which for these time-to-event illustrations means more events. The figures are not participants per arm because the planning calculation is expressed in total events.

Rows for smaller fractions of the observed effect require more events than rows for the full observed effect. This is a general feature of statistical planning: smaller effects are harder to distinguish from random variation, so more information is needed to detect them at a chosen significance level and power. The planning figures are conditional on the observed estimate being a useful basis for future planning; they do not establish that the underlying effect is known precisely.

These planning calculations are not observed power. They do not estimate the probability that the completed trial's result was correct or that a treatment works. They are illustrations of the event information a hypothetical study would need under stated assumptions. Because the record does not report the trial's original design assumptions, the planning illustration should not be taken as a reconstruction of the study's intended sample size or stopping plan.

13. 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

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 0.940, so ln(HR) = -0.0619.
2. d = 4 × (1.96 + 0.842)² / [ln(0.940)]² = 8,200.3, rounded up to 8,201 events in total 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
hr <- 0.94                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(4259)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 8,201 (80% power), 10,978 (90% power)
 75% of observed effect: events = 14,579 (80% power), 19,517 (90% power)
 50% of observed effect: events = 32,802 (80% power), 43,912 (90% power)
Simulated power with 8,201 events: 0.796 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Group-Sequential Design Analysis

Many time-to-event trials, especially in oncology, stop or report early when interim data are overwhelming, so a single fixed-sample number understates the real design problem. The calculation below is an illustrative group-sequential design built from this trial's posted control median and hazard ratio. The registry record does not state the planned number of looks or the spending function, so the design is a planning illustration, not a reconstruction of the protocol. Assumed: one-sided α = 0.025, 90% power, three equally spaced looks on information, Lan-DeMets O'Brien-Fleming spending, 24 months of accrual, 60 months in total and a 1% monthly dropout hazard.

Under these assumptions the design needs about 27,001 participants in total, with the boundaries below.

Fixed design versus group-sequential alternatives

Group-sequential designs cost a small increase in maximum sample size and events compared with the fixed design, in exchange for the chance to stop early for overwhelming efficacy. O'Brien-Fleming spending keeps early boundaries very conservative and the final boundary close to the fixed-design value, whereas Pocock spending spends more alpha early.

Sensitivity to a smaller true effect

Planning for exactly the observed effect is optimistic, because observed effects in published trials tend to be somewhat larger than the true effect. The table repeats the main design (three looks, O'Brien-Fleming spending, 90% power) for smaller true hazard ratios.

For context, this trial enrolled 230 participants.

Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993. Agreement between the two packages confirms the boundary calculation.

R code and output

suppressWarnings(suppressMessages({ library(gsDesign); library(rpact) }))
med_control <- 45.6          # control-arm median, months (converted from the posted unit)
hr          <- 0.94          # hazard ratio used for planning
lambdaC     <- log(2) / med_control
eta         <- 0.01               # dropout hazard per month (assumption)
acc_months  <- 24                 # accrual duration (assumption)
study_months<- 60                 # total study duration (assumption)

design <- function(k, sfu, beta, h = hr) gsSurv(k = k, test.type = 1, alpha = 0.025, beta = beta, sfu = sfu,
  lambdaC = lambdaC, hr = h, hr0 = 1, eta = eta, gamma = 1, R = acc_months, T = study_months, minfup = study_months - acc_months)

# Main design: 3 looks, Lan-DeMets O'Brien-Fleming spending, one-sided alpha 0.025, 90% power
x <- design(3, sfLDOF, 0.10)
cat(sprintf("N_TOTAL|%s\n", format(ceiling(sum(x$gamma * x$R)), big.mark = ",")))
for (i in 1:3) cat(sprintf("GS|%d|%.1f|%.0f|%.3f|%.5f|%.5f\n", i, x$T[i], x$n.I[i], x$upper$bound[i], pnorm(-x$upper$bound[i]), cumsum(x$upper$spend)[i]))
cat("\n"); print(gsBoundSummary(x))

# Comparison of designs
cmp <- list(list("Fixed design, one analysis", 1, sfLDOF), list("Two looks, O'Brien-Fleming", 2, sfLDOF),
            list("Three looks, O'Brien-Fleming", 3, sfLDOF), list("Three looks, Pocock", 3, sfLDPocock))
cat("\n")
for (b in c(0.20, 0.10)) for (d in cmp) {
  y <- design(d[[2]], d[[3]], b)
  cat(sprintf("DES|%s|%d|%.0f|%s|%.1f\n", d[[1]], round(100 * (1 - b)), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ","), tail(y$T, 1)))
}

# Sensitivity: what if the true effect is smaller than the observed one? (3 looks, O'Brien-Fleming, 90% power)
cat("\n")
for (fr in c(1, 0.75, 0.5)) {
  y <- design(3, sfLDOF, 0.10, exp(log(hr) * fr))
  cat(sprintf("SENS|%.2f|%.3f|%.0f|%s\n", fr, exp(log(hr) * fr), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ",")))
}

# Independent cross-check of the boundaries with rpact
r <- getDesignGroupSequential(kMax = 3, alpha = 0.025, beta = 0.10, sided = 1, typeOfDesign = "asOF", informationRates = x$timing)
cat(sprintf("\nCross-check, upper boundaries: gsDesign %s | rpact %s\n", paste(round(x$upper$bound, 3), collapse = ", "), paste(round(r$criticalValues, 3), collapse = ", ")))

Method: LachinFoulkes 
      Analysis               Value Efficacy
     IA 1: 33%                   Z   3.7103
      N: 26038         p (1-sided)   0.0001
  Events: 3703        ~HR at bound   0.8852
     Month: 23    P(Cross) if HR=1   0.0001
               P(Cross) if HR=0.94   0.0338
     IA 2: 67%                   Z   2.5114
      N: 27002         p (1-sided)   0.0060
  Events: 7406        ~HR at bound   0.9433
     Month: 38    P(Cross) if HR=1   0.0060
               P(Cross) if HR=0.94   0.5603
         Final                   Z   1.9930
      N: 27002         p (1-sided)   0.0231
 Events: 11109        ~HR at bound   0.9629
     Month: 60    P(Cross) if HR=1   0.0250
               P(Cross) if HR=0.94   0.9000



Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993

15. Analyses Reported by the Sponsor

The registry reports the sponsor's primary analysis as a log-rank comparison, with a hazard ratio of 0.94, a 95% confidence interval of 0.67 to 1.32, and p = 0.36. It also reports Kaplan-Meier median overall survival estimates for each Phase III group. These are the study's posted inferential results and should be the starting point for interpretation.

Any recalculation based on posted summary statistics is a simple, unadjusted re-calculation. It cannot reproduce participant-level time-to-event analysis, account for censoring in the same way, or recover adjustments and modeling choices that are not described. A recalculated estimate can therefore differ from the sponsor's analysis without showing that the sponsor's analysis is wrong. The registry record does not report a discrepancy between the posted sponsor analysis and the calculated comparison.

The hazard ratio and p-value are explicitly reported, so they are not merely estimates reconstructed from the medians. The median estimates and their confidence intervals provide complementary summaries but do not determine the hazard ratio. A comparison of medians cannot substitute for a log-rank analysis of event timing.

The record does not provide enough detail to reproduce the full sponsor analysis independently, including participant-level event times and complete model specifications. Accordingly, interpret the reported result as the sponsor's analysis as recorded, and treat summary-statistic recalculations as supplementary rather than as a replacement for it.

16. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study examines several outcomes, subgroups, or comparisons. If each is tested separately without a plan to control the overall chance of false-positive findings, the collection of tests can produce apparently notable results by chance. This is why a primary endpoint is important: it identifies the main question, while secondary analyses add context and generate additional evidence.

The registry records overall survival as the primary outcome and lists secondary outcomes including time to tumor progression, toxicity, survival by MGMT status, and progression-free survival by MGMT status. It does not report whether statistical testing for these outcomes was adjusted for multiplicity, whether a testing hierarchy was used, or whether particular secondary endpoints were intended to be confirmatory. Their p-values therefore should not be interpreted as though each came from a single isolated test.

For missing data, the primary endpoint description states that patients last known to be alive were censored at the date of last contact. That is a specific rule for incomplete survival follow-up, not a complete missing-data plan. The registry does not report the amount or pattern of loss to follow-up, reasons for missing observations, sensitivity analyses, or how missing biomarker information affected subgroup inclusion.

The primary population is described as eligible randomized patients, and the record says pilot arms were excluded from Phase III analyses. It does not report a detailed analysis-population definition, such as how protocol deviations or eligibility issues were handled, or whether analyses followed an intention-to-treat principle. The posted group counts identify the reported comparison but do not answer those methodological questions.

These reporting gaps do not invalidate the results, but they limit independent appraisal of confirmatory status, missing-data sensitivity, and the population to which the estimates apply. The registry record should not be used to infer an adjustment procedure, interim-analysis plan, or statistical strategy that it does not describe.

17. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: platelet count decreased (69.2%), neutropenia (43.4%), leukopenia nos (30.8%), hemoglobin decreased (20.0%), platelet transfusion (15.4%).

The safety record reports serious adverse event counts for the Phase III groups among those at risk: 43 of 96 in the radiation-therapy-plus-temozolomide group and 72 of 99 in the RT + BCNU/CCNU group. The record also reports other adverse events affecting 94 of 96 and 95 of 99, respectively. These are descriptive counts from the registry and should be read with their reported at-risk denominators.

The record lists serious-event terms including platelet count decreased, neutropenia, leukopenia NOS, hemoglobin decreased, platelet transfusion, and infection with grade 3 or 4 neutropenia. It does not provide a complete account of the severity, duration, timing, recurrence, or attribution for every event in the listed terms. Nor does the record establish that every listed term applies equally to both Phase III arms.

Safety counts are not directly comparable without understanding the reporting period, ascertainment rules, and patient exposure. The registry record does not report enough detail on these matters to support a detailed comparative safety analysis from the listed counts alone. Pilot-arm safety figures are reported separately and should not be combined with the Phase III comparison.

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

The posted primary result describes an overall-survival comparison between two radiation-plus-chemotherapy strategies. The median estimates are 3.9 years and 3.8 years, while the sponsor-reported hazard ratio is 0.94 with a 95% confidence interval of 0.67 to 1.32 and p = 0.36. Taken together, these statistics do not provide a precise estimate that rules out a range of relative effects.

A non-significant p-value does not prove that the treatments are equal, and it does not show that either strategy is ineffective. Establishing equivalence would require a prespecified equivalence margin and an analysis designed to assess it; the registry does not report such a margin. Likewise, an absence of strong evidence against the null is not proof that no clinically meaningful difference exists.

The hazard ratio is a group-level summary of event rates, not a prediction for an individual patient. The medians are summaries of estimated survival distributions, not guarantees about how long an individual will live. Confidence intervals express uncertainty under the analysis assumptions. Their interpretation depends on censoring, follow-up, and the statistical method, some details of which are not reported in the registry record.

The secondary outcomes add information about progression, biomarker-defined groups, and adverse events, but they have different definitions and uncertainties. Their results should not be used to overturn the primary finding or to claim a treatment effect specific to MGMT status without an appropriate interaction analysis. Multiplicity handling is not reported, which limits the evidential weight of secondary p-values.

The results apply to the Phase III comparison as recorded, not automatically to every regimen in the study record. The pilot arms were excluded from Phase III analyses. The registry also does not provide the original sample-size rationale, full protocol analysis plan, or complete detail on missing data. Those limits mean the posted results can be described and interpreted, but should not be extended beyond what the record supports.

19. Why This Trial Matters Statistically

This trial illustrates why time-to-event results should be read through several related summaries. Median survival gives an interpretable point on the estimated survival curve, while a hazard ratio summarizes relative event rates across follow-up and a log-rank test assesses differences in event experience. These measures answer related but non-identical questions. Similar medians do not by themselves establish that the full curves are similar, and a hazard ratio does not translate directly into a difference in median time.

The confidence interval is essential to the interpretation. The primary hazard-ratio interval spans values below and above 1, leaving meaningful uncertainty about the direction and magnitude of the relative effect. A p-value alone would conceal that range. The result demonstrates why statistical non-significance should not be translated into equivalence, especially when no equivalence margin is reported.

The registry also makes an important distinction between the overall study enrollment and the Phase III analysis population. Pilot arms were not included in the Phase III analyses, so the primary comparison concerns only the two randomized treatment strategies reported for that phase. Keeping those populations separate prevents an inappropriate pooling of regimens that were not part of the main comparison.

Finally, the secondary endpoints show how multiplicity and subgroup interpretation matter. Additional outcomes can illuminate progression, toxicity, and biomarker patterns, yet without a reported testing strategy their results are less definitive than the primary comparison. This is a useful statistical lesson: the strength of a conclusion depends not only on the numerical estimate, but also on the endpoint's role, analysis population, assumptions, precision, and prespecified context.

20. 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 registry record leaves important design and analysis details unspecified. It does not report the original sample-size justification, allocation-concealment procedure, full statistical analysis plan, interim-analysis details, or whether any covariate adjustments were used. These omissions make it difficult to assess the planned precision and to reproduce the analysis fully.

The primary result is based on time-to-event methods, but the record does not describe follow-up completeness, censoring patterns, model diagnostics, or sensitivity analyses. It says patients last known to be alive were censored at last contact, but does not report whether censoring could have been related to prognosis. The proportional-hazards assumption is not addressed in the record.

There are also limitations in the outcome descriptions. The time-to-tumor-progression timeframe refers to death, leaving the operational definition unclear. The MGMT subgroup results do not include a reported treatment-by-MGMT interaction. Without that test, subgroup estimates cannot establish that treatment effects differ by MGMT status. The record does not explain how missing biomarker data were handled.

Safety reporting is descriptive and does not supply all information needed to compare adverse events by exposure, timing, duration, or attribution. The record's adverse-event summaries should not be treated as a complete comparative safety analysis. Pilot arms are listed in the study record but excluded from the Phase III efficacy analyses, and their safety data should remain separate from the main comparison.

These limitations concern the scope of inference from the registry record. They do not imply that the trial analyses were improperly conducted; rather, the details needed to evaluate those issues are not reported in the registry record.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Radiation Therapy Combined With Chemotherapy in Treating Patients With Anaplastic Astrocytoma or Mixed Gliomas — NCT00004259.
• Registry record last updated 2019-09-12; status last verified 2019-08.
• Results first posted 2017-11-13.
• 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.

24. Record Summary

NCT00004259 is a completed randomized Phase III study sponsored by the Radiation Therapy Oncology Group. It compared radiation therapy with temozolomide against radiation therapy with BCNU or CCNU in patients with anaplastic astrocytoma or mixed anaplastic oligoastrocytoma. The record reports enrollment of 230 and notes that pilot arms were excluded from Phase III analyses.

The primary endpoint was overall survival from randomization to death from any cause, estimated by Kaplan-Meier methods. The sponsor reported a hazard ratio of 0.94, a 95% confidence interval of 0.67 to 1.32, and p = 0.36 by the log-rank method. Median estimates were 3.9 years and 3.8 years in the two Phase III groups.

Secondary results address tumor progression and MGMT status, alongside safety reporting. Multiplicity procedures, the original sample-size justification, and several analysis details are not reported in the registry record. The posted estimates should therefore be interpreted with their stated uncertainty and limitations.