Study of Maintenance Temozolomide Versus Observation in Stable or Responding Stage III/IV Non-Small Cell Lung Cancer Patients (Study P05146): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Maintenance Temozolomide Versus Observation in Stable or Responding Stage III/IV Non-Small Cell Lung Cancer Patients (Study P05146) (NCT00632203), a randomized phase 2 study. The primary outcome is Number of Participants Who Had Brain Metastases.

2. Clinical Question

For the primary endpoint, the posted count was 4 of 22 participants in the temozolomide group and 3 of 23 in the observation group. Those counts describe the endpoint within the evaluable population, rather than every participant enrolled in the trial. The result is therefore best read as a comparison of the recorded brain-metastasis counts over the stated timeframe, alongside the uncertainty in the estimated difference. It does not, by itself, describe the timing of events or answer every question about disease progression, survival, quality of life, or tolerability. Those were among the secondary outcomes listed in the registry record.

3. Trial Design

This was a randomized Phase 2 study with a parallel-group design and no masking reported in the registry record. The registry lists enrollment as 53 and describes the study as terminated. The study compared maintenance temozolomide with observation in people with stable or responding Stage III/IV non-small cell lung cancer after standard treatment. The primary endpoint analysis was restricted to an evaluable population, defined as participants with at least one post-randomization MRI scan.

The temozolomide regimen was described as 75 mg/m^2 orally each day for 21 consecutive days, followed by a 7-day rest period, continuing until progression or up to 6 cycles, whichever occurred first. The observation group received observation. The posted primary endpoint timeframe was up to 12 months, measured from day 1 of cycle 1 of standard first-line systemic chemotherapy. Brain metastases were defined as radiological evidence on MRI.

The primary endpoint results list 22 participants in the temozolomide group and 23 in the observation group, with 4 and 3 participants, respectively, recorded as having brain metastases. The adverse-event record uses group denominators of 26 and 27. These are different endpoint populations as represented in the registry record; the reasons for the differing denominators are not reported in the registry record.

Secondary outcomes included time to radiological CNS progression, time to progression, overall survival, brain metastases at first progression, cancer-related quality of life measured with EORTC questionnaires, and tolerability. The registry lists 6 secondary outcomes and 7 outcome measures. The original sample-size assumptions, detailed procedures for follow-up imaging, and reasons for termination are not reported in the registry record. The registry record also does not state whether any interim analyses were planned or conducted.

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: Evaluable population, defined as a participant who had at least one post-randomization MRI scan

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 Who Had Brain Metastases, measured over Up to 12 months (as measured from day 1 of cycle 1 of standard first-line systemic chemotherapy).

The primary endpoint was the number of participants who had brain metastases. The registry defined brain metastases as radiological evidence on MRI and set the timeframe as up to 12 months from day 1 of cycle 1 of standard first-line systemic chemotherapy. The endpoint population was evaluable: a participant needed at least one post-randomization MRI scan to be included.

This is a binary outcome summarized as a count: each evaluable participant either had brain metastases recorded within the endpoint definition or did not. Comparing the counts between randomized groups gives a direct description of how often the specified finding was recorded in each group. It is not the same as measuring time to brain metastases. The registry listed time to radiological CNS progression separately as a secondary outcome, indicating that timing was addressed as a distinct question.

An MRI-based definition makes the endpoint dependent on radiological assessment. Requiring at least one post-randomization scan also means that participants without such a scan do not meet the stated evaluable-population definition. How missing scans or incomplete endpoint ascertainment were handled beyond that definition is not reported in the registry record.

The observed primary counts were 4 of 22 in the temozolomide group and 3 of 23 in the observation group. These counts should be interpreted with the uncertainty around the between-group estimate in mind. They do not establish why a brain metastasis occurred, whether the timing differed, or what happened for outcomes not included in this endpoint.

7. Results Posted in the Registry

• Temozolomide Treatment: 18.2% (4/22) versus 13.0% (3/23) on control; risk difference 5.1 percentage points (95% CI -16.1 to 26.3), risk ratio 1.39 (0.35–5.53); unadjusted two-proportion z-test p = 0.634.

8. Statistical Interpretation

9. Statistical Methods Explained

What does a risk difference represent?

A risk difference subtracts the event risk in one group from the risk in the other. The recalculated comparison is oriented as temozolomide versus observation. The posted risks were 0.182 for temozolomide and 0.130 for observation, giving a risk difference of 0.051. A positive value under this orientation means the recorded event risk was higher in the temozolomide group in this estimate; a negative value would mean it was lower. This is a plain, unadjusted comparison of the posted group summaries, not an estimate adjusted for other characteristics.

Risk differences use the original outcome scale and are often straightforward to interpret as absolute differences. Their precision depends on the amount of information in the data and the observed event pattern. With relatively few endpoint events, the estimate can be uncertain. The confidence interval, rather than the point estimate alone, indicates how broad that uncertainty is under the recalculation method.

What does the confidence interval tell us?

A confidence interval gives a range of values generated by a statistical procedure that is designed to capture the underlying effect at its stated confidence level across repeated applications. It is not a range containing a known fixed effect with a probability assigned after the data have been observed. The recalculated risk-difference interval extends from -0.161 to 0.263. Because it includes zero, it is compatible with no risk difference under this interval procedure, as well as with differences in either direction.

The interval is not a guarantee that every value within it is equally plausible, nor does it account for limitations not represented in the calculation, such as missing outcome information or choices in the analysis plan. It describes statistical uncertainty conditional on the data and method used. The registry's reported analysis does not include a confidence interval in the record details, so the interval cited belongs to the unadjusted recalculation.

Why can the p-values differ?

The registry names a two-sided Exact Pearson Chi-square Test and reports p = 0.700. The unadjusted recalculation reports p = 0.634. A p-value depends on the test procedure and its assumptions, as well as the observed data. Different procedures can produce different p-values from the same group counts. The registry record does not give further detail about implementation choices, and the recalculation is not a replacement for the sponsor's own analysis.

In general, an exact test evaluates the observed data against a null hypothesis using a probability calculation suited to the discrete patterns possible in the data. A chi-square test assesses whether observed group patterns are compatible with no association, with a test statistic compared to a reference distribution. The named method is two-sided, so departures in either direction contribute to the assessment. A p-value is not an effect size, a measure of clinical importance, or the chance that the result will replicate.

Why does over-dispersion matter for some outcomes?

General statistical note: over-dispersion arises when count variability is greater than a model's assumed distribution allows. If a model underestimates variability, its uncertainty estimates can be too narrow and its tests overly optimistic. It is a consideration for some count-model analyses, but the primary endpoint is presented as a binary participant outcome with group counts, and the registry reports an exact chi-square test. The registry record does not report a count-regression model or an over-dispersion assessment for this endpoint.

What is the distinction between reported and recalculated analyses?

The sponsor's registry result is the reported two-sided Exact Pearson Chi-square Test with p = 0.700. The separate recalculation uses the posted summary counts to make an unadjusted comparison and gives a risk difference of 0.051, a confidence interval from -0.161 to 0.263, and p = 0.634. Such recalculations are useful for interpreting the posted summaries on an absolute scale, but they do not reproduce unreported analysis details or replace the registry's own analysis.

The risk ratio and odds ratio are also listed in the recalculation details, but the absolute risk difference is the most direct measure for describing the event-risk contrast. Any relative measure has its own scale and interpretation. No adjusted covariate analysis, subgroup analysis, or model-based estimate is reported in the registry record details used for this discussion.

What does randomization contribute?

General statistical note: random allocation is intended to make groups comparable, on average, with respect to characteristics that might affect outcomes. It does not ensure exact balance in a particular trial, especially when the amount of information is limited. The registry records a randomized parallel-group design, but detailed information about allocation generation, concealment, and balance of baseline characteristics is not reported in the registry record. Therefore, those implementation features cannot be evaluated from the stated record.

10. Confidence Intervals and Precision

The recalculated risk difference is 0.051 for temozolomide versus observation, with a confidence interval from -0.161 to 0.263. The lower and upper bounds lie on opposite sides of zero. This means the data do not pin down a single direction of difference: the interval includes both a lower event risk under temozolomide and a higher event risk under temozolomide, as well as no difference.

The width of the interval signals substantial uncertainty around the point estimate. The posted primary result contains 4 events among 22 evaluable participants in the temozolomide group and 3 among 23 in the observation group. When an endpoint has few recorded events, small changes in event counts can materially affect an estimate, and statistical precision is limited. This is a general feature of binary outcomes with sparse events, not a claim about any unreported data.

The registry's own reported analysis gives p = 0.700 from a two-sided Exact Pearson Chi-square Test. The recalculated p-value is 0.634. The confidence interval presented with the recalculated estimate is not a confidence interval reported by the sponsor in the registry record. The registry record does not provide an interval for its reported test.

Precision is distinct from whether an estimate is statistically significant under a chosen procedure. A broad interval can leave important effect sizes compatible with the data, even when the point estimate looks small. It also does not show that every value in the interval is equally likely or resolve possible limitations from missing MRI assessments. Those matters are not captured by the interval alone.

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. They are illustrations based on the observed effect, not the original trial design. The calculations assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as a stated fraction of the observed effect. The original design assumptions and sample-size justification are not reported in the registry record.

The 80% power column describes the planning number of participants per arm under which the calculation would have an 80% chance of detecting the specified effect at the stated significance level, if the assumptions hold. The 90% power column uses a higher target chance of detection under those same planning assumptions. For planning at 100% of the observed effect, the figures are 783 participants per arm for 80% power and 1,047 for 90% power. For planning at 75% of the observed effect, they are 1,345 and 1,800. For planning at 50% of the observed effect, they are 2,919 and 3,907.

Smaller effects are harder to distinguish from random variation, so detecting a smaller effect generally requires more participants. The figures are planning illustrations derived from the observed effect and do not describe the trial's achieved power. They 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 = 0.130; treatment proportion p₁ = 0.182; average p̄ = 0.156.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5133 = 1.0061.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5120 = 0.4309.
4. n = (1.0061 + 0.4309)² / (0.0514)² = 782.2, rounded up to 783 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 <- 0.130434782608696                          # control proportion
p1 <- 0.181818181818182                          # 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(632203)
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 = 783 (80% power), 1,047 (90% power)
 75% of observed effect: n per arm = 1,345 (80% power), 1,800 (90% power)
 50% of observed effect: n per arm = 2,919 (80% power), 3,907 (90% power)
power.prop.test: 782.2 
Simulated power at n = 783 per arm: 0.797 (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. Analyses Reported by the Sponsor

The sponsor's own analysis is reported as a two-sided Exact Pearson Chi-square Test, with p = 0.700. The recalculation from the posted group summaries is an unadjusted comparison: it gives a risk difference of 0.051, a confidence interval from -0.161 to 0.263, and p = 0.634. The registry record does not report additional detail about the sponsor's implementation beyond the named test and p-value.

The recalculation makes the absolute group contrast and its uncertainty easier to read, but it is not the sponsor's analysis and should not be treated as a correction to it. The reported p-value and recalculated p-value differ because they come from separately stated analyses. The discrepancy flag is false, so the record does not identify a discrepancy requiring the sponsor's own analysis to take precedence under that condition.

Neither p-value alone gives the size of the effect or its precision. The recalculated risk difference and interval describe the unadjusted contrast, while the registry result records the sponsor's test. The registry record does not provide an adjusted estimate or an interval for the sponsor's reported analysis.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists 6 secondary outcomes in addition to the primary endpoint, and 7 outcome measures overall. The listed secondary topics include radiological CNS progression, time to progression, overall survival, brain metastases at first progression, cancer-related quality of life, and tolerability. Considering multiple outcomes can increase the chance of finding at least one apparently notable result by chance if each is tested independently without accounting for the number of tests. Whether and how multiplicity was addressed is not reported in the registry record. No adjustment method, testing hierarchy, or prespecified prioritization among outcomes is reported in the registry record used.

The registry record reports one analysis for the primary result. It does not report whether interim analyses were planned or conducted, nor does it describe a broader statistical analysis plan. The original sample-size rationale, including any assumptions about the expected effect or statistical power, is not reported in the registry record. Planning illustrations based on the observed effect do not fill that gap.

The primary endpoint population is defined as participants with at least one post-randomization MRI scan. That definition makes MRI availability part of endpoint inclusion. The registry record does not state how many participants lacked a post-randomization scan, whether missing scans were treated in a particular way, or whether any sensitivity analysis addressed missing endpoint information. It also does not describe the handling of other missing data for secondary outcomes.

The primary endpoint counts should therefore be read as applying to the stated evaluable population. The registry record does not provide enough information to assess whether the evaluable participants differed from those without a post-randomization MRI or how such differences might affect interpretation. This limitation is separate from statistical uncertainty in the risk-difference interval.

Details about the randomization procedure, allocation concealment, baseline balance, and handling of protocol deviations are also not reported in the registry record. These omissions do not change the posted counts, but they restrict assessment of the conduct and analysis choices underlying the result.

15. 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: disease progression (7.4%), dyspnoea (7.4%), lymphopenia (3.8%), pyrexia (3.8%), pneumonia (3.8%).

The adverse-event record lists serious adverse events for 8 of 26 participants in the temozolomide group and 4 of 27 in the observation group. Other adverse events are listed for 25 of 26 and 24 of 27 participants, respectively. These figures describe the adverse-event record's group denominators, which differ from the primary endpoint's evaluable-population denominators. The reasons for this difference are not reported in the registry record.

The listed serious-event terms and percentages include disease progression and dyspnoea at 7.4% each, and lymphopenia, pyrexia, pneumonia, and dehydration at 3.8% each. The registry record does not provide additional context in the registry record used about event duration, severity grading beyond the serious classification, or how events were attributed. Death counts are not reported in the registry record.

These figures are descriptive. They do not by themselves establish a causal relationship between an intervention and an event, and comparisons can be affected by the group sizes and the way events were collected. The registry record lists tolerability as a secondary outcome, but these figures alone do not summarize every aspect of tolerability.

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

The primary result is a comparison of MRI-recorded brain metastases in the evaluable population during the stated timeframe. The posted counts were 4 of 22 with temozolomide and 3 of 23 with observation. The unadjusted risk difference is 0.051, with a confidence interval from -0.161 to 0.263. The interval includes zero and extends in both directions, so the estimate is imprecise and does not establish a clear direction of difference.

This result does not show whether the groups differed in the time at which brain metastases developed. Time to radiological CNS progression was listed separately as a secondary outcome, and its results are not reported in the registry record details used. Nor does the primary endpoint answer questions about overall survival, general disease progression, quality of life, brain metastases at first progression, or tolerability. Those were separate outcomes.

The study objective described an interest in whether maintenance temozolomide could prevent or delay brain metastases, but the primary endpoint is a participant count within a defined follow-up period. It does not establish a mechanism or explain why an event occurred. In addition, the endpoint analysis required at least one post-randomization MRI; the registry record does not report how missing scans were handled or how many participants did not meet the evaluable definition.

The registry reports a two-sided Exact Pearson Chi-square Test with p = 0.700, while the simple unadjusted recalculation gives p = 0.634. These values should be understood as results from the respective analyses, not as probabilities that an intervention has or lacks an effect. The record's primary result is best characterized as a sparse, uncertain comparison rather than a definitive answer to the wider clinical question.

The registry record identifies the study as terminated, but the reason for termination is not reported in the registry record. It also does not report the original sample-size justification or results for the listed secondary outcomes in the registry record. Consequently, the primary endpoint alone cannot provide a complete account of the trial's findings or explain the study's status.

17. Why This Trial Matters Statistically

This trial illustrates why a randomized comparison should be interpreted through both the effect estimate and its precision. The primary endpoint gives a direct count of MRI-defined brain metastases in each group, and the unadjusted risk difference expresses the contrast on an absolute scale. But the confidence interval is broad and crosses zero, showing that the posted data do not tightly identify the direction or size of the difference.

It also illustrates the importance of the analysis population. The primary endpoint is based on an evaluable population defined by availability of at least one post-randomization MRI. The adverse-event figures use different group denominators. Readers need to keep those populations distinct rather than assuming every outcome uses the same set of participants. The registry record does not explain the differing denominators or provide missing-scan details, so those aspects cannot be assessed fully.

The comparison between the registry's reported test and the unadjusted recalculation is another useful statistical distinction. A sponsor-reported p-value and a recalculated absolute effect answer related but not identical reporting needs. The p-value alone does not convey the magnitude or uncertainty of the observed difference; the risk difference and interval provide complementary information.

Finally, a trial can list multiple outcomes that address different dimensions of a clinical question. A count of participants with brain metastases is not equivalent to time to progression, survival, quality of life, or tolerability. The registry record's listed secondary outcomes underline the breadth of the study questions, while the lack of their results and multiplicity details limits interpretation beyond the primary endpoint.

18. 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 primary endpoint analysis is restricted to participants with at least one post-randomization MRI. The registry record does not report the number of randomized participants who lacked such a scan, the timing or completeness of imaging, or how missing endpoint information was handled. The counts therefore reflect the stated evaluable population, and the available record does not establish how representative that population was of all randomized participants.

The primary outcome has few recorded events, and the recalculated confidence interval is wide. The point estimate is consequently uncertain. The reported primary endpoint is a count within a follow-up window, not a time-to-event estimate, even though time to radiological CNS progression was listed separately as a secondary outcome.

The registry record lists enrollment as 53, while the primary endpoint groups list 22 and 23 participants and the adverse-event groups list 26 and 27. The record details do not explain the different denominators, so their relationship should not be inferred. The reason for termination is also not reported in the registry record.

Other details that would help evaluate the findings are not reported in the registry record: the original sample-size justification, randomization implementation details, the statistical analysis plan, whether interim analyses occurred, multiplicity procedures, and results for the listed secondary outcomes. The sponsor's result includes a named test and p-value, but no confidence interval or adjusted effect estimate is reported in the record details used. These limitations constrain assessment of precision, missing data, and the broader set of study questions.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Study of Maintenance Temozolomide Versus Observation in Stable or Responding Stage III/IV Non-Small Cell Lung Cancer Patients (Study P05146) — NCT00632203.
• Registry record last updated 2017-06-07; status last verified 2017-05.
• Results first posted 2012-03-26.
• 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.

22. Record Summary

The registry describes NCT00632203 as a randomized Phase 2, parallel-group study with no masking reported. It compared maintenance temozolomide with observation in stable or responding Stage III/IV non-small cell lung cancer. The primary endpoint was the number of evaluable participants with MRI evidence of brain metastases up to 12 months from day 1 of cycle 1 of standard first-line systemic chemotherapy.

The posted primary counts were 4 of 22 in the temozolomide group and 3 of 23 in the observation group. The recalculated risk difference was 0.051, with a confidence interval from -0.161 to 0.263. The registry reports a two-sided Exact Pearson Chi-square Test with p = 0.700. The study status is terminated; the reason is not reported in the registry record.