Phase III Study of the Effect of GTx-024 on Muscle Wasting in Patients With Non-Small Cell Lung Cancer (NSCLC): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Phase III Study of the Effect of GTx-024 on Muscle Wasting in Patients With Non-Small Cell Lung Cancer (NSCLC) (NCT01355484), a randomized phase 3 study. The primary outcome is Physical Function.

2. Clinical Question

The question focuses on a thresholded measure of physical function rather than directly on muscle mass. The brief summary also mentions maintaining or gaining lean body mass, but the primary result is the proportion crossing the stated stair-climb power threshold. This distinction matters: the endpoint answers a specific question about the percentage meeting that threshold, not every question about muscle wasting, physical performance, or the other stated study goals.

3. Trial Design

The registry classifies NCT01355484 as a phase III, randomized, double-blind, placebo-controlled study with a parallel-group design. It lists GTx as the sponsor and the status as completed. Enrollment was 321. The registry dates the study start to 2011-07 and primary completion to 2013-05.

The official title places the study in people with non-small cell lung cancer on first-line platinum plus a taxane chemotherapy. The trial compared GTx-024 with placebo, with each arm description stating that participants received their assigned treatment for the duration of the trial. The record does not provide further detail about the chemotherapy regimen, randomization procedure, allocation concealment, or who was masked. The double-blind designation is recorded, but the specific roles kept unaware of assignment are not reported in the registry record.

The primary endpoint was assessed at Day 84 in subjects included in the Full Analysis Set. The posted arm counts for this result are 160.0 for GTx-024 and 161.0 for placebo. These counts describe the outcome population as posted; the registry record does not explain how the Full Analysis Set was defined or how participants without a Day 84 assessment were handled.

Two outcome measures are listed in the record, but no secondary outcomes are posted in the results information. The registry lists zero analyses. The trial's original statistical analysis plan, including any prespecified model, adjustment strategy, interim analyses, or sample-size justification, is not reported in the registry record. The planning calculations associated with the posted comparison are illustrations based on the observed effect, not a description of the original trial design.

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: Subjects included in the Full Analysis Set

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: Subjects included in the Full Analysis Set

6. Primary Endpoint

The registry lists the primary outcome as Physical Function, measured over Day 84.

The primary endpoint is the percentage of subjects at Day 84 whose stair-climb power change from baseline was at least 10%. It is a binary, threshold-based outcome: each participant either met the stated threshold or did not, and the arm result is the percentage meeting it. The endpoint is assigned to subjects included in the Full Analysis Set.

Stair-climb power is a measure of physical function. In general, functional measures can complement measures of body composition because they describe performance rather than muscle quantity alone. That is general context, not a claim about how this trial's investigators selected or validated the endpoint. The registry record does not report the threshold's justification, the assessment procedure, or whether other thresholds were considered.

The measure is not a direct estimate of lean body mass. Although the brief summary mentions maintaining or gaining lean body mass as a study aim, this primary result addresses the proportion meeting a specified change in stair-climb power. It also does not describe the full distribution of individual changes: participants just below and just above the threshold are classified differently, while the amount of change within either category is not conveyed by the percentage alone.

The threshold and assessment time make the endpoint interpretable as a specific Day 84 response proportion. They also constrain what can be concluded: the posted percentages do not answer questions about later function, other functional tasks, or lean body mass unless separate results are available. Those additional results are not reported in the registry record summarized for this endpoint.

7. Results Posted in the Registry

• GTx-024: 29.4% (47/160) versus 24.2% (39/161) on control; risk difference 5.2 percentage points (95% CI -4.5 to 14.8), risk ratio 1.21 (0.84–1.74); unadjusted two-proportion z-test p = 0.297.

8. Statistical Interpretation

9. Statistical Methods Explained

What does a percentage with an event mean?

For this endpoint, a participant counts as meeting the outcome if stair-climb power change from baseline is at least 10% at Day 84. The arm percentage is the proportion of participants in that arm who meet the condition, expressed as a percentage. This kind of binary outcome compresses each person's measurement into one of two categories. It is straightforward to interpret, but it does not preserve how far each participant's change was from the cutoff.

The registry reports 29.4% for GTx-024 and 24.2% for placebo. The related counts in the comparison information are 47 and 39, with outcome denominators of 160.0 and 161.0, respectively. The record identifies the primary population as the Full Analysis Set. It does not explain whether all participants in that set had observed Day 84 measurements or how missing measurements were treated.

What is a risk difference?

A risk difference subtracts one arm's event proportion from the other arm's event proportion. Its scale is an absolute difference in proportions, rather than a relative comparison. The posted estimate is 0.051, and its 95% confidence interval is -0.045 to 0.148. The direction of the estimate indicates which arm had the higher observed proportion under the comparison's stated ordering; the interval communicates how uncertain that contrast is. Because the interval crosses zero, it includes both directions of the contrast.

Other comparative measures, such as a risk ratio or an odds ratio, use different scales. Ratios compare relative event frequencies, while the odds ratio compares odds rather than probabilities. Their interpretation differs from the absolute scale of a risk difference. The registry record does not identify an adjusted model for these comparisons, and the posted numbers should not be taken as evidence that an adjustment was made.

What does a confidence interval convey?

A confidence interval is an uncertainty range around an estimate, calculated using a statistical procedure. In general, a narrower interval indicates greater precision under the procedure's assumptions, while a wider interval indicates less precision. A confidence interval does not guarantee that the true value lies in the particular interval, and it does not show the range of individual participant outcomes. It concerns uncertainty in the estimated quantity.

The posted arm intervals describe each arm percentage; the interval for the risk difference describes the comparative estimate. For a treatment comparison, the latter is the direct interval for the contrast. It is useful to consider its endpoints and width, rather than relying only on a point estimate or a p-value.

What does the p-value say?

The unadjusted p-value is 0.297. In general, a p-value is calculated under a null hypothesis and a statistical model, and represents how often results at least as incompatible with that null would arise if the model and null were true. It is not a measure of effect magnitude, clinical importance, or the probability that a result is due to chance. The registry does not report the test used, its assumptions, or the planned significance level for the trial's primary analysis. Consequently, no unreported testing choices should be inferred from the posted p-value.

The p-value and confidence interval are related but answer different presentation needs: the p-value gives a test statistic's tail probability under a null model, while the interval shows a range of estimates compatible with the data and method. Neither should be interpreted without considering the estimate, its scale, and the underlying design.

Why do the analysis population and missing-data method matter?

The primary result is described for subjects included in the Full Analysis Set. In general, analysis populations define which randomized participants contribute to an estimate, and different definitions can affect interpretation. The registry does not provide the Full Analysis Set definition, details on exclusions, or a missing-data strategy for Day 84. It is therefore not possible to determine from the posted record whether missing endpoint values were excluded, imputed, or handled another way.

For a binary endpoint, the denominator and the treatment of participants without an observed outcome are especially important because the percentage depends on who is counted. The posted arm counts and percentages should be read alongside the registry's stated population, without assuming that every randomized participant contributed a measured Day 84 result.

Why can over-dispersion matter in general?

For binary data, a simple statistical model often assumes that observations are independent and that the event probability is sufficiently similar across participants within the group being analyzed. Over-dispersion describes variation greater than a model expects. When present, it can make standard errors too small and intervals too narrow if the model does not account for it. This is general statistical context; the registry record does not report a model or diagnostics for over-dispersion in this trial.

Similarly, randomization supports a fair comparison in expectation, but it does not guarantee identical participant characteristics in every realized sample. The registry identifies the trial as randomized, but does not provide baseline characteristics or details needed to assess balance from the posted primary result alone.

10. Confidence Intervals and Precision

The arm-specific estimates are 29.4% for GTx-024, with a 95% confidence interval from 22.4% to 37.1%, and 24.2% for placebo, with a 95% confidence interval from 17.8% to 31.6%. These intervals show uncertainty around each estimated percentage. They are not ranges containing a stated proportion of individual participants, and they do not indicate how much each participant's stair-climb power changed.

For the comparison, the risk difference is 0.051, with a 95% confidence interval from -0.045 to 0.148. This interval is the most direct precision summary for the between-arm contrast. It includes zero and extends to values on either side, so the estimate does not precisely establish the direction or magnitude of the difference. The interval's breadth is important alongside the point estimate: a single estimate can give a deceptively definite impression if its uncertainty is not considered.

Confidence intervals are generally affected by the amount of information in the data, the outcome frequency, and the statistical method. Those are general principles. The registry record does not describe the interval calculation method, any adjustments, or whether the reported interval was prespecified. It also does not report a sensitivity analysis for the primary result. The intervals therefore communicate the uncertainty under the calculation used for the posted results, while leaving those methodological details unspecified.

Precision is distinct from clinical importance. The registry does not report a threshold for a clinically important between-arm difference, so the statistical interval cannot establish whether any value in its range would be meaningful to patients or clinicians. It is also important not to interpret the arm intervals in isolation as a formal test of the arm-to-arm contrast; the risk-difference interval directly summarizes that contrast.

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 sample-size figures are planning illustrations based on the observed effect, not the trial's original design justification. Each value is the number of participants per arm for the stated power level; these are not total event counts. 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 trial's own original design assumptions are not reported in the registry record.

The rows consider planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. For 100% of the observed effect, the illustration gives 1,160 participants per arm for 80% power and 1,552 participants per arm for 90% power. For 75% of the observed effect, it gives 2,030 participants per arm for 80% power and 2,718 participants per arm for 90% power. For 50% of the observed effect, it gives 4,496 participants per arm for 80% power and 6,018 participants per arm for 90% power.

In plain terms, the 80% power column gives a planning sample size intended to yield an 80% chance of detecting the specified effect under the stated assumptions, if that effect and assumptions hold. The 90% power column gives the corresponding planning target for a higher chance of detection. Smaller effects require more participants to distinguish them from random variation under otherwise unchanged assumptions. These figures are not observed power.

A sample-size curve, when read alongside such a table, expresses how the required planning size changes across assumed effect sizes under the chosen calculation. It should be interpreted as a planning aid, not as evidence about what the trial was designed to detect. The calculations use the observed effect as a reference, and the registry does not report whether the original trial used these assumptions or targets.

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 = 0.242; treatment proportion p₁ = 0.294; average p̄ = 0.268.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6264 = 1.2277.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6253 = 0.5263.
4. n = (1.2277 + 0.5263)² / (0.0515)² = 1,159.3, rounded up to 1,160 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 <- 0.242236024844721                          # control proportion
p1 <- 0.29375                          # 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(1355484)
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 = 1,160 (80% power), 1,552 (90% power)
 75% of observed effect: n per arm = 2,030 (80% power), 2,718 (90% power)
 50% of observed effect: n per arm = 4,496 (80% power), 6,018 (90% power)
power.prop.test: 1159.3 
Simulated power at n = 1,160 per arm: 0.804 (20000 replicates)

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

13. Multiplicity, Missing Data and Analysis Populations

The registry record lists two outcome measures, but no secondary outcome results are posted in the results information summarized for this study. It also lists zero analyses. The record does not describe a multiplicity strategy, such as how the chance of false-positive findings across multiple endpoints, time points, or comparisons was addressed. It is therefore not possible to determine whether a hierarchy, adjustment, or other prespecified approach governed interpretation of multiple tests.

Multiplicity matters in general because testing more than one hypothesis can increase the chance that at least one result appears statistically unusual even when none of the underlying effects is real. Adjustment methods can control that risk, but may affect precision and interpretation. This is general statistical context; no particular approach should be attributed to this trial without a registry report of it.

The primary population is identified as the Full Analysis Set, but its definition is not reported in the registry record. The posted endpoint counts are 160.0 for GTx-024 and 161.0 for placebo. The record does not state how those counts relate to all randomized participants, whether exclusions occurred, or whether participants with missing Day 84 outcomes were included in another way.

Missing-data handling can affect a binary percentage if the probability of having an observed outcome differs across participants or arms. In general, exclusion, imputation, and other approaches rely on different assumptions. The registry does not report which approach was used, how many primary outcomes were missing, or any sensitivity analyses. This limits assessment of how robust the posted percentages are to missing endpoint data.

The record also does not report subgroup analyses, adjusted estimates, or the complete statistical analysis plan. These omissions do not establish that such work was or was not conducted; they mean that it is not reported in the registry record. Interpretation should remain tied to the posted primary result and avoid treating unreported analyses as confirmatory evidence.

14. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 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 (11.9%), pneumonia (3.8%), dehydration (3.7%), anaemia (3.1%), febrile neutropenia (2.5%).

The serious adverse event figures describe the number of participants affected and the number at risk in each arm: 56 of 160 participants in the GTx-024 arm and 60 of 161 participants in the placebo arm. These counts are not, by themselves, a comparison adjusted for follow-up time, event severity, or other participant characteristics. The registry record does not provide further analytic detail for these figures.

The record lists Disease Progression (11.9%), Pneumonia (3.8%), Dehydration (3.7%), Anaemia (3.1%), Febrile Neutropenia (2.5%), and Respiratory Failure (2.5%) among the top serious terms. The record does not specify in this listing how these terms are distributed between the arms, so they should not be assigned to one arm from the percentages alone. Death counts are not reported in the registry record.

Other adverse events are also counted: 141 of 160 participants in the GTx-024 arm and 144 of 161 participants in the placebo arm were affected. These are distinct from the serious adverse event counts. Counts and listed terms describe reported events; they do not establish that a drug caused an event or permit a complete assessment of comparative safety without fuller definitions, timing, and analysis details.

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

The posted primary result describes the percentage of participants meeting a specific stair-climb power threshold at Day 84. The percentage was 29.4% in the GTx-024 arm and 24.2% in the placebo arm. The reported risk difference was 0.051, with a 95% confidence interval from -0.045 to 0.148, and the unadjusted p-value was 0.297. Together, these figures describe the observed result and its statistical uncertainty.

The estimate should not be read as a definitive statement about the effect of GTx-024. The confidence interval includes values on both sides of zero, and the registry record does not provide the trial's full statistical analysis plan, a clinically important difference threshold, or sensitivity analyses. The posted result therefore does not establish how the finding would change under alternative assumptions or methods.

The endpoint is a thresholded physical-function outcome, not a direct measure of lean body mass. It does not summarize every aspect of physical performance, and it does not describe longer-term outcomes beyond the reported Day 84 assessment. The brief summary mentions maintaining or gaining lean body mass, but the primary result cannot answer that separate question.

The result also does not establish a causal explanation for individual outcomes or determine why participants did or did not meet the threshold. Randomization is an important design feature for comparing groups, but the available registry information does not include the full details needed to evaluate all aspects of conduct and analysis. Safety figures should likewise be read as reported counts, not as proof of causation.

In general, statistical evidence is strongest when the design, endpoint, analysis population, missing-data handling, and uncertainty are all considered together. For this study, the posted record supports a narrow description of the primary outcome and its estimated contrast. It does not, on its own, resolve broader questions about muscle wasting, clinical relevance, or the balance of benefits and harms.

16. Why This Trial Matters Statistically

This trial illustrates why an endpoint's definition matters. The study's broad stated interest includes muscle wasting and lean body mass, while the posted primary result concerns a threshold in stair-climb power. A statistical comparison can only answer the question encoded by its outcome measure. Keeping the distinction clear prevents a functional result from being mistaken for a direct measurement of muscle quantity.

It also demonstrates the value of reporting a contrast with a confidence interval. The arm percentages are easy to recognize, but the risk-difference estimate and its interval address the comparative question directly. The interval includes zero and spans both negative and positive values, making precision an essential part of interpretation rather than an afterthought. The p-value alone would not communicate the same range of plausible contrasts.

Trial design labels provide useful context: this was a randomized, double-blind, placebo-controlled phase III parallel-group study, with enrollment of 321. Yet those labels do not replace details about the analysis population, missing data, or statistical methods. The Full Analysis Set is named, but its definition is not reported. The record lists zero analyses, and secondary outcome results are not posted. These gaps limit how completely readers can connect the design to the reported estimate.

The planning illustrations also show a general principle: sample-size requirements depend on the effect planners aim to detect. Smaller target effects require larger planned samples under the same assumptions. Because the figures are based on the observed effect and are not the trial's own design rationale, they should not be confused with the study's original planning process or with evidence that the trial was adequately powered for a particular effect.

For clinical researchers and biostatisticians, the central lesson is to keep the estimand, uncertainty, and reporting boundaries aligned. The posted result is informative about one Day 84 binary endpoint, but its interpretation should not be expanded beyond what that endpoint and the registry record support.

17. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The posted primary result has several limits specific to this trial record. The endpoint is a thresholded stair-climb power change of at least 10% from baseline, assessed at Day 84. It does not show the distribution of individual changes, describe the magnitude of changes among participants who met or missed the threshold, or directly measure lean body mass. The registry record does not report why the threshold was chosen or how the assessment was conducted.

The primary population is called the Full Analysis Set, but the record does not define that set. It also does not report how missing Day 84 outcomes were handled, how many were missing, or whether sensitivity analyses were performed. These details matter for understanding the denominators and the assumptions underlying the percentages.

The statistical methods behind the posted intervals and p-value are not described. The record does not report the trial's analysis plan, adjustment strategy, interim analyses, subgroup analyses, or sample-size justification. It lists zero analyses and no posted secondary results. As a result, the primary comparison cannot be placed in the context of a complete set of prespecified analyses from the registry record alone.

The study is described as randomized and double-blind, but details of the randomization procedure, allocation concealment, and masking roles are not reported. The record also does not give further participant characteristics or the specific chemotherapy regimen beyond the official title. These omissions restrict assessment of generalizability and implementation details.

Finally, the serious adverse event counts are reported without a fuller comparative analysis, and death counts are not reported in the registry record. The named serious terms are not broken down by arm in the listing. These limitations mean that the posted efficacy endpoint and safety counts should be treated as a partial account rather than a complete evaluation of all study outcomes.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Phase III Study of the Effect of GTx-024 on Muscle Wasting in Patients With Non-Small Cell Lung Cancer (NSCLC) — NCT01355484.
• Registry record last updated 2016-03-03; status last verified 2016-02.
• Results first posted 2016-03-03.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

21. Record Summary

NCT01355484 is a completed phase III randomized, double-blind, placebo-controlled parallel-group study sponsored by GTx. It enrolled 321 participants and compared GTx-024 with placebo in people with non-small cell lung cancer, with the official title referring to first-line platinum plus a taxane chemotherapy.

The primary outcome was the percentage of subjects in the Full Analysis Set with stair-climb power change of at least 10% from baseline at Day 84. The posted percentages were 29.4% for GTx-024 and 24.2% for placebo. The risk difference was 0.051, with a 95% confidence interval from -0.045 to 0.148; the unadjusted p-value was 0.297.

The registry record lists two outcome measures, no posted secondary outcome results, and zero analyses. It does not report the original statistical plan, missing-data methods, or sample-size justification. Serious adverse event counts were 56 of 160 participants for GTx-024 and 60 of 161 for placebo; deaths are not reported in the registry record.