Pembrolizumab Plus Epacadostat vs Pembrolizumab Plus Placebo in Metastatic Non-Small Cell Lung Cancer (KEYNOTE-654-05/ECHO-305-05): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pembrolizumab Plus Epacadostat vs Pembrolizumab Plus Placebo in Metastatic Non-Small Cell Lung Cancer (KEYNOTE-654-05/ECHO-305-05) (NCT03322540), a randomized phase 2 study. The primary outcome is Objective Response Rate (ORR) of Pembrolizumab Plus Epacadostat Versus Pembrolizumab Plus Placebo.

2. Clinical Question

3. Trial Design

This was a phase 2, randomized, double-blind, parallel-group study sponsored by Incyte Corporation. The registry record lists quadruple masking. It enrolled 154 participants and is marked completed. The recorded study start date is 2017-12-15, and the primary completion date is 2019-01-10. The stated aim was to evaluate the efficacy and safety of pembrolizumab plus epacadostat compared with pembrolizumab plus placebo as first-line treatment for metastatic NSCLC with high PD-L1 expression.

The two randomized groups were pembrolizumab plus epacadostat and pembrolizumab plus placebo. Each arm had 77 participants in the primary outcome results. The ITT population included all randomized participants, according to the outcome description. The listed pembrolizumab schedule was 200 mg intravenously every three weeks, beginning on Day 1 of each cycle, for up to 35 administrations. Epacadostat or matching placebo was listed as an oral twice-daily treatment. The registry record states that epacadostat and placebo administration were discontinued after implementation of protocol amendment 05; further details about this change are not reported in the registry record.

Randomization supports a between-group comparison by assigning participants to study arms rather than allowing treatment choice to determine assignment. Blinding is intended to limit the influence of treatment knowledge on study conduct and outcome assessment. The record specifically states that the primary response outcome was assessed by blinded independent central review. The registry record does not report the randomization procedure, allocation concealment details, or the original design assumptions for the sample size. It also does not report further details about the masking procedures.

The primary outcome was objective response rate (ORR), defined as the proportion of participants with a confirmed complete response or partial response under RECIST v1.1. The secondary outcomes listed progression-free survival, overall survival, duration of response, adverse events, and discontinuation of study drug due to adverse events. Although these outcomes are named, numerical results for them are not reported in the posted results described for this trial.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: Phase 3 design of the study has been amended soon after it had started to a prospectively randomized phase 2 study. At the time of amendment existing participants were given a choice to move/participate in the new phase 2 study, and some participants who chose to discontinue the study at phase 3 were assigned to "study terminated by sponsor" as the reason for not completing the study in disposition table. The results posted are combined in the prospectively redesigned phase 2 trial.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT population consisted of all randomized participants.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Intent-to-treat (ITT) population consisted of all randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Objective Response Rate (ORR) of Pembrolizumab Plus Epacadostat Versus Pembrolizumab Plus Placebo, measured over Up to approximately 6 months.

Objective response rate is the proportion of participants who have a confirmed complete response or partial response. In this trial, response was defined under Response Evaluation Criteria in Solid Tumors (RECIST) v1.1 and assessed by blinded independent central review. The primary outcome description specifies the ITT population: all randomized participants. Its timeframe was up to approximately 6 months.

ORR is used as a measure of tumor response during a specified assessment period. It focuses on whether participants meet predefined criteria for a complete or partial response; it does not, by itself, describe how long a response lasts, how long participants live, or whether the groups differ in other outcomes. Those are distinct questions, reflected in the registry's secondary outcomes of duration of response, progression-free survival, and overall survival.

In general, a confirmed response requirement means that a response must meet the study's confirmation criteria rather than being treated as established by a single unconfirmed assessment. Independent central review is intended to provide an assessment separate from the treating study team, while blinding helps limit the influence of knowledge about assigned treatment. The record identifies these features but does not report further assessment details, such as the timing of individual scans or the rules for handling unevaluable assessments.

The reported percentages were 32.5% for pembrolizumab plus epacadostat and 39.0% for pembrolizumab plus placebo. Their confidence intervals describe uncertainty around each arm's response rate. The comparative analysis is needed to assess the estimated difference between arms; looking at the two percentages alone does not answer the comparative question with the same precision.

7. Results Posted in the Registry

• Pembrolizumab + Epacadostat: 32.5% (25/77) versus 39.0% (30/77) on control; risk difference -6.5 percentage points (95% CI -21.6 to 8.6), risk ratio 0.83 (0.54–1.28); unadjusted two-proportion z-test p = 0.400.

8. Statistical Interpretation

9. Statistical Methods Explained

Randomization, masking, and the choice of analysis method shape how a clinical comparison should be read. The registry record describes this as a randomized, double-blind, parallel-group phase 2 study with quadruple masking. In general, random allocation is intended to make the groups comparable on average at the start of a trial, including with respect to factors that may not be measured. Masking is intended to reduce the chance that knowledge of treatment assignment influences care, assessment, or reporting. The record does not describe the procedures used to generate or conceal the allocation, or the operational details of masking.

What does the primary analysis estimate?

The endpoint is a binary outcome at the participant level: each participant either met the confirmed complete or partial response definition during the stated assessment period or did not. The arm-specific estimates are therefore percentages of participants with an event, where the event is response. The posted percentages are 32.5% for pembrolizumab plus epacadostat and 39.0% for pembrolizumab plus placebo. The primary outcome population is ITT, defined as all randomized participants, and each arm had 77 participants in the posted results.

The comparative estimate is the difference in response percentages, with the epacadostat arm first and the placebo arm as comparator. A negative estimate indicates a lower observed response percentage in the first arm under that subtraction order. The sponsor reported a difference in percentages of -6.5. The 95% confidence interval, from -21.5 to 8.7, includes zero. In general, zero is the no-difference value for a risk difference. The sponsor's p-value was 0.800. These statistics describe uncertainty and compatibility with a no-difference model; they do not determine the probability that a hypothesis is true.

Why is the sponsor's method important?

The reported analysis used the stratified Miettinen and Nurminen method. The registry record identifies the method as stratified but does not report the stratification factors. In general, this family of methods estimates a difference in proportions and can account for specified strata in the comparison. Because the factors and detailed implementation are not reported, the result should be described using the method label in the registry rather than attributing any particular adjustment to it.

The simple recalculation is explicitly unadjusted and uses the posted summary statistics. It gives a risk difference of -0.065, a 95% confidence interval from -0.216 to 0.086, and p = 0.4. It is not the same analysis as the sponsor's stratified result. Its role is a secondary numerical check, not a replacement for the reported analysis. The registry record does not flag a discrepancy, but the estimates should still be identified by method rather than treated as interchangeable.

How should alternative effect measures be read?

The posted recalculated risk ratio is 0.833, with a 95% confidence interval from 0.544 to 1.277. A risk ratio compares the response proportions as a ratio; the value 1 is the no-difference reference. The posted odds ratio is 0.753, with a 95% confidence interval from 0.389 to 1.459. An odds ratio compares odds rather than proportions, and its no-difference reference is also 1. Both intervals include their respective no-difference values. These measures express the same binary outcome on different scales, so their numerical values should not be read as if they were the same quantity.

In general, odds ratios and risk ratios can differ in magnitude because they use different scales. When the outcome is not rare, an odds ratio should not be interpreted as though it were a risk ratio. Here, the arm-specific percentages are available, so the response proportions and the sponsor's difference in percentages offer a direct description of the observed response rates. The ratios are additional ways of describing the contrast, not separate endpoints.

What does a confidence interval contribute?

A confidence interval gives a range of values compatible with the data and the method at the stated confidence level. In general, a 95% confidence procedure would cover the true parameter in 95% of repeated applications under its assumptions. It is not a statement that there is a 95% probability that this particular interval contains the true value. For the sponsor's difference estimate, the interval spans both negative and positive differences. This indicates substantial uncertainty about the size and direction of the contrast, rather than a precise estimate of no difference.

Confidence intervals also help distinguish the question of statistical compatibility from the question of clinical importance. The registry record does not state a threshold for a clinically important ORR difference. Without such a threshold, the posted interval cannot be classified against a prespecified clinically meaningful margin. The record also does not report a non-inferiority or equivalence margin.

What is not described?

The registry record does not report the original sample-size justification, any interim analyses, a detailed statistical analysis plan, or details of multiplicity control. It does not report how missing or unevaluable response assessments were handled, how the stratification factors were defined, or whether any sensitivity analyses were planned. Those elements should not be inferred from the reported endpoint, method, or p-value. The sample-size figures are planning illustrations based on the observed effect and stated generic assumptions, not evidence about the trial's original design or about observed power.

10. Confidence Intervals and Precision

The arm-specific 95% confidence intervals communicate the precision of the estimated response percentages. For pembrolizumab plus epacadostat, ORR was 32.5%, with an interval from 22.2% to 44.1%. For pembrolizumab plus placebo, ORR was 39.0%, with an interval from 28.0% to 50.8%. Each interval spans a sizeable range of plausible response percentages, so neither arm's estimate should be treated as an exact population value.

For the between-arm comparison, the sponsor's reported difference in percentages was -6.5, with a 95% confidence interval from -21.5 to 8.7. The interval includes zero and extends in both directions. Thus the data are compatible with a lower response percentage in the epacadostat arm, little or no difference, or a higher percentage, within the limits expressed by this interval and analysis. The interval is more informative than the point estimate alone because it conveys the uncertainty around the estimated contrast.

In general, confidence intervals reflect both the amount of information in the data and the assumptions of the statistical method. A wider interval indicates less precision; it does not by itself identify why precision is limited. The trial randomized 154 participants, with 77 per arm in the primary results. The registry record does not state a target difference used to design the trial, so it is not possible to judge the interval against an original design target. It also does not report a clinically important difference threshold.

The recalculated unadjusted difference has a 95% confidence interval from -0.216 to 0.086, while the sponsor's reported interval is expressed as -21.5 to 8.7 percentage points. These are results from different analysis methods and scales of presentation; interpretation should give priority to the sponsor's reported stratified analysis. In general, an interval that includes the no-difference value does not prove that the true difference is zero. It means the data do not distinguish the estimated contrast from no difference at the stated confidence level.

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 for 80% power and participants per arm for 90% power. They are based on the observed effect and the stated planning assumptions: a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set at the stated fraction of the observed effect. These figures are planning illustrations, not a description of the trial's original sample-size calculation. The original design assumptions are not reported in the registry record.

For planning 100% of the observed effect, the figures are 854 participants per arm for 80% power and 1,143 participants per arm for 90% power. For planning 75% of the observed effect, they are 1,534 participants per arm for 80% power and 2,053 participants per arm for 90% power. For planning 50% of the observed effect, they are 3,483 participants per arm for 80% power and 4,663 participants per arm for 90% power.

In plain language, the 80% power column gives the estimated number per arm under which the planning method would detect the chosen effect in 80% of repeated trials if that effect and the assumptions were correct. The 90% power column uses a higher probability of detection, so it requires more participants under these calculations. A smaller target effect is harder to distinguish from random variation, so planning for a smaller fraction of the observed effect requires more participants. The corresponding relationship applies to a sample-size curve: as the effect to detect becomes smaller, the required sample size rises.

These figures are not observed power. They use the trial's observed effect as a planning input and do not show the probability that the completed trial detected a true effect. They also do not establish that the trial was inadequately or adequately designed, since its own design assumptions are not reported.

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 = 0.390; treatment proportion p₁ = 0.325; average p̄ = 0.357.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6776 = 1.3281.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6761 = 0.5690.
4. n = (1.3281 + 0.5690)² / (-0.0649)² = 853.6, rounded up to 854 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 <- 0.38961038961039                          # control proportion
p1 <- 0.324675324675325                          # 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(3322540)
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 = 854 (80% power), 1,143 (90% power)
 75% of observed effect: n per arm = 1,534 (80% power), 2,053 (90% power)
 50% of observed effect: n per arm = 3,483 (80% power), 4,663 (90% power)
power.prop.test: 853.6 
Simulated power at n = 854 per arm: 0.796 (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 registry's reported primary comparison is the sponsor's own analysis: a stratified Miettinen and Nurminen analysis of the difference in percentages. It reports -6.5, with a 95% confidence interval from -21.5 to 8.7 and p = 0.800. That is the primary reported result to use when describing the trial's main comparison.

The unadjusted recalculation from the posted summary statistics gives a risk difference of -0.065, with a 95% confidence interval from -0.216 to 0.086 and p = 0.4. It is a simple reanalysis of the posted counts, not the sponsor's stratified analysis. The methods are different, and the registry record does not provide the stratification factors or additional detail needed to reconstruct the sponsor's procedure independently.

Accordingly, the recalculation can help readers check the broad direction and uncertainty using an unadjusted approach, but it should not replace or overrule the reported analysis. The registry record does not flag a discrepancy. The sponsor's interval includes zero, as does the recalculated interval, but the p-values and interval limits should be quoted with their corresponding methods rather than blended into one result.

14. Multiplicity, Missing Data and Analysis Populations

The registry record lists the primary outcome as ORR and five secondary outcomes: progression-free survival, overall survival, duration of response, number of participants with adverse events, and number of participants who discontinued study drug due to adverse events. Six outcome measures are recorded in total. One analysis is listed for the primary comparison. The posted numerical results give the primary ORR comparison, but numerical results for the secondary efficacy outcomes are not reported in the registry record.

When a trial tests several outcomes, time points, or treatment contrasts, the chance of at least one apparently unusual result can increase if each is treated as an independent confirmatory test. In general, multiplicity procedures are used to control or describe that issue, for example by adjusting significance thresholds or by specifying a hierarchy of hypotheses. The registry record does not report a multiplicity strategy, a testing hierarchy, or whether the secondary outcomes were intended as confirmatory or exploratory. No claim about multiplicity adjustment can therefore be made from the listed outcomes alone.

The primary outcome population is the ITT population, defined as all randomized participants. This is the reported population definition for the response analysis. The record does not report how missing response assessments, participants with unevaluable scans, or other incomplete outcome data were handled. It also does not describe whether any additional analysis populations were used. In general, missing outcome data can affect a binary response estimate if participants with unknown outcomes are treated differently across analyses. The registry entry does not provide enough detail to assess such handling.

The record does not report the detailed statistical analysis plan, interim analyses, or an original sample-size justification. It also does not state a clinically important response-rate difference or an equivalence or non-inferiority margin. These omissions limit assessment of whether the primary comparison was prespecified in the precise form reported, how secondary testing was interpreted, and what difference the trial was designed to detect. They do not change the numerical result that is posted, but they constrain claims about the broader analysis plan.

15. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: malignant neoplasm progression (10.4%), pneumonia (9.1%), pneumonitis (5.3%), anaemia (2.7%), atelectasis (2.7%).

The registry reports serious adverse events for 75 participants at risk in the pembrolizumab plus epacadostat group and 77 at risk in the pembrolizumab plus placebo group. Serious adverse events affected 36 participants in the epacadostat group and 35 in the placebo group; the combined figure was 71 of 152 participants at risk. The at-risk totals differ between arms for this safety summary, so they should not be assumed to match the 77 participants per arm in the primary ORR results.

The listed most frequent serious adverse event terms and percentages were malignant neoplasm progression, 10.4%; pneumonia, 9.1%; pneumonitis, 5.3%; anaemia, 2.7%; atelectasis, 2.7%; and hypokalaemia, 2.6%. Other adverse events affected 69 of 75 participants at risk in the epacadostat group and 67 of 77 at risk in the placebo group, with 136 of 152 overall. The registry also lists 23 deaths in the epacadostat group and 29 in the placebo group, with 52 overall.

These are descriptive counts. The registry record does not establish from these figures alone whether an event was caused by study treatment, and the listed death counts should not be interpreted as treatment-attributable deaths. The record does not report a comparative statistical analysis for these safety figures in the posted results.

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

The primary result describes confirmed objective responses during a period of up to approximately 6 months among participants with metastatic NSCLC expressing high PD-L1. ORR was 32.5% with pembrolizumab plus epacadostat and 39.0% with pembrolizumab plus placebo. The sponsor's reported between-arm difference was -6.5 percentage points, with a 95% confidence interval from -21.5 to 8.7 and p = 0.800. The interval includes no difference and allows uncertainty in either direction.

These findings do not establish that adding epacadostat improves the response rate. They also do not establish that the two regimens are equivalent or that there is no possible difference. A failure to show a statistically distinguishable difference is not the same as demonstrating equivalence. The registry record does not report a non-inferiority or equivalence design, a margin for such a claim, or a clinically important difference threshold.

ORR concerns whether a confirmed complete or partial response occurred under RECIST v1.1. It does not describe the length of time a response lasted, progression-free survival, overall survival, or all aspects of safety. Those are separate outcomes. Although the registry lists several such secondary outcomes, their numerical results are not reported in the record described for this analysis. The response result therefore should not be used as a substitute for those missing outcome estimates.

In general, a trial result applies most directly to the population and conditions studied. This trial focused on first-line treatment of metastatic NSCLC with high PD-L1 expression, with the primary endpoint assessed in the ITT population of all randomized participants. The registry record does not provide further eligibility criteria or the original assumptions used to plan the study. Generalization beyond the registered population, treatment schedules, assessment method, or follow-up timeframe requires evidence not contained in this result.

The best reading is therefore bounded: the posted primary analysis found no clear statistical separation in ORR between the randomized groups, while the confidence interval remains broad enough to include differences in either direction. This is a statement about the precision and results of this comparison, not medical advice or a conclusion about an individual's treatment.

17. Why This Trial Matters Statistically

This trial illustrates why a randomized comparison should be read through both its point estimate and its uncertainty. The two arm-specific response percentages, 32.5% and 39.0%, are descriptive estimates. The sponsor's reported contrast of -6.5 has a 95% confidence interval from -21.5 to 8.7. The interval makes clear that the observed point estimate does not, on its own, pin down the size or even the direction of the underlying difference with precision.

It also illustrates the distinction between an endpoint and the broader clinical question. ORR is a defined, time-bounded response measure, based on confirmed complete or partial responses assessed by blinded independent central review. It is not a complete account of disease control, survival, duration of response, or safety. The registry lists those other outcomes, but their numerical results are not reported in the record used for this comparison.

Method matters too. The sponsor's primary analysis used a stratified Miettinen and Nurminen method, while a simple unadjusted recalculation from posted summary statistics gives a different numerical estimate and p-value. Naming each method prevents an unadjusted check from being mistaken for the reported primary analysis. The record does not give the stratification factors, so readers cannot fully reconstruct the sponsor's procedure from the entry alone.

Finally, the planning illustrations show how sample-size requirements depend on the effect selected for detection and the desired power under stated assumptions. They are not evidence of the trial's original design rationale and are not observed power. Keeping these distinctions visible helps prevent a non-significant comparison, a planning calculation, or a single endpoint from being asked to answer more than it can.

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

The primary comparison is based on 154 enrolled participants, with 77 in each arm in the ORR results. The confidence interval for the sponsor's difference in percentages is broad, extending from -21.5 to 8.7. This limits precision about the magnitude and direction of the response-rate contrast. The registry record does not report the original sample-size justification, so the result cannot be compared with an original target effect or design calculation.

The result is specific to the registered population: participants with metastatic NSCLC expressing high levels of PD-L1 and receiving first-line treatment. Further eligibility criteria are not reported in the registry record. The primary endpoint timeframe was up to approximately 6 months, so this response comparison does not itself describe longer-term outcomes. Numerical results for the listed secondary outcomes of progression-free survival, overall survival, and duration of response are not reported in the registry record.

The sponsor's analysis is identified as stratified, but the registry record does not state the stratification factors or details needed to reproduce the analysis. The unadjusted recalculation is not a substitute for that reported method. The record also does not report missing-data handling, unevaluable response rules, interim analyses, a detailed statistical analysis plan, or a multiplicity strategy. These gaps limit independent assessment of analysis choices and the interpretation of multiple outcomes.

The treatment descriptions state that epacadostat and placebo administration were discontinued after implementation of protocol amendment 05. Further details about the amendment, its timing, and its implications for exposure are not reported in the registry record. The safety at-risk denominators also differ from the equal arm sizes listed for the primary outcome, and the record does not explain those denominators in the information. Safety counts are descriptive and do not by themselves establish treatment causality.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Pembrolizumab Plus Epacadostat vs Pembrolizumab Plus Placebo in Metastatic Non-Small Cell Lung Cancer (KEYNOTE-654-05/ECHO-305-05) — NCT03322540.
• Registry record last updated 2025-08-22; status last verified 2025-08.
• Results first posted 2020-02-05.
• 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

NCT03322540 is a completed phase 2, randomized, double-blind, parallel-group study sponsored by Incyte Corporation. It compared pembrolizumab plus epacadostat with pembrolizumab plus placebo as first-line treatment for metastatic NSCLC expressing high PD-L1. The registry lists 154 enrolled participants and quadruple masking.

The primary endpoint was confirmed objective response rate under RECIST v1.1, assessed by blinded independent central review in the ITT population. ORR was 32.5% with pembrolizumab plus epacadostat and 39.0% with pembrolizumab plus placebo. The sponsor's reported difference in percentages was -6.5, with a 95% confidence interval from -21.5 to 8.7 and p = 0.800.

The registry lists five secondary outcomes, but numerical results for those outcomes are not reported in the registry record. The original sample-size assumptions, multiplicity strategy, missing-data handling, and detailed analysis plan are also not reported.