Efficacy and Safety Study of Pembrolizumab (MK-3475) With or Without Lenvatinib (MK-7902/E7080) in Adults With Programmed Cell Death-Ligand 1 (PD-L1)-Positive Treatment-naïve Nonsmall Cell Lung Cancer (NSCLC) [MK-7902-007/E7080-G000-314/LEAP-007] - China Extension Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Efficacy and Safety Study of Pembrolizumab (MK-3475) With or Without Lenvatinib (MK-7902/E7080) in Adults With Programmed Cell Death-Ligand 1 (PD-L1)-Positive Treatment-naïve Nonsmall Cell Lung Cancer (NSCLC) [MK-7902-007/E7080-G000-314/LEAP-007] - China Extension Study (NCT04676412), a randomized phase 3 study. The primary outcome is Progression-free Survival (PFS) Per Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST 1.1) as Assessed by Blinded Independent Central Review (BICR).

2. Clinical Question

The registry describes a Phase 3 randomized, double-blind trial, with triple masking also recorded. Its official title identifies a parallel-group design. The Chinese extension is associated with the global trial NCT03829332. The posted primary result concerns PFS in the registry-defined population, not necessarily every participant enrolled across the global study. The number enrolled in the extension is listed as 107; the endpoint analyses and safety summaries use their own reported group sizes, so those figures should be read in their stated contexts rather than treated as interchangeable.

3. Trial Design

The registry identifies this as a randomized Phase 3 parallel-group trial with triple masking. The study compared pembrolizumab plus lenvatinib with pembrolizumab plus placebo in treatment-naïve adults with metastatic NSCLC and PD-L1 TPS at least 1%. The extension enrolled participants in China alongside the global study. The registry lists an enrollment of 107 for the extension and marks the study completed. The recorded start date is 2019-10-23 and the primary completion date is 2021-05-19.

Both groups received pembrolizumab by intravenous infusion on Day 1 of each 3-week cycle, for up to 35 administrations. The experimental regimen added oral lenvatinib, while the comparator regimen used oral placebo for lenvatinib. Both oral treatments were scheduled on Days 1-21 of each cycle until progressive disease or unacceptable toxicity. The registry record does not report further details about allocation concealment, randomization strata, adherence, or treatment changes.

The primary endpoint was PFS assessed by blinded independent central review under RECIST 1.1. The posted median PFS analysis lists 42 participants in the pembrolizumab-plus-lenvatinib group and 53 in the pembrolizumab-plus-placebo group. The safety summary uses different group denominators, 49 and 58, respectively. These are different analysis summaries as recorded; the reasons for the differing denominators are not reported in the registry record. The outcome population is described as Chinese participants randomized to the global study or extension who were evaluable for response.

The registry reports 16 outcome measures and 14 analyses. The reported secondary outcomes include objective response, adverse events, treatment discontinuation due to adverse events, and patient-reported quality-of-life and symptom measures. Details of the analysis populations and handling of missing measurements are not reported in the registry record. Although the brief summary lists both PFS and overall survival as primary hypotheses, the result details provide a PFS result; an overall-survival result is not reported in the registry record.

4. Participant Flow and Accounting

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

Before assignment: 107 Chinese participants were randomized and received treatment (global study \[NCT03829332; n=80\] or to the extension portion \[n=27\]).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All Chinese participants who were randomized to the global study (NCT03829332) or to the extension portion that were evaluable for response.

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: All Chinese participants who were randomized to the global study (NCT03829332) or to the extension portion.

6. Primary Endpoint

The registry lists the primary outcome as Progression-free Survival (PFS) Per Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST 1.1) as Assessed by Blinded Independent Central Review (BICR), measured over Up to approximately 18 months.

PFS measures the time from randomization until the first documented disease progression or death from any cause, whichever occurs first. In this trial, progression was assessed by blinded independent central review using RECIST 1.1. The registry describes progression as at least a 20% increase in the sum of target-lesion diameters, together with an absolute increase of at least 5 mm; the appearance of one or more new lesions also counted as progression.

PFS is used in oncology trials because it captures the period during which participants remain alive without documented progression. It combines progression and death into a single time-to-event endpoint, so it does not by itself distinguish which event ended an individual’s PFS. It is also not the same as overall survival, which concerns time to death from any cause. The brief summary names overall survival as another primary hypothesis, but an overall-survival result is not reported in the registry record details for this analysis.

The posted PFS summaries use the Kaplan-Meier product-limit method, which estimates the event-time distribution while accounting for censored observations. A median is the time by which the estimated event-free proportion has reached the midpoint of the distribution. It is a summary of the estimated time-to-event curve, not a statement that every participant experienced progression or death by that time. The registry gives a timeframe of up to approximately 18 months for this endpoint.

As with any endpoint, interpretation depends on how events were assessed, the analysis population, follow-up, and censoring. The use of blinded independent central review provides the stated assessment basis, but additional operational details about review procedures and missing or censored observations are not reported in the registry record.

7. Results Posted in the Registry

• Pembrolizumab + Lenvatinib vs Pembrolizumab + Placebo: hazard ratio 1.18 (95% CI 0.61–2.25), reported p = 0.711 (as reported in the registry).

8. Secondary Efficacy Endpoints

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

Objective Response Rate (ORR) Per RECIST 1.1 as Assessed by BICR

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Pembrolizumab + Lenvatinib: 33.3% (14/42) versus 39.6% (21/53) on control; risk difference -6.3 percentage points (95% CI -25.7 to 13.1), risk ratio 0.84 (0.49–1.45); unadjusted two-proportion z-test p = 0.528.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Pembrolizumab + Lenvatinib and the control. The interval gives the range of effects that remain compatible with the data.

Time to True Deterioration (TTD) in EORTC QLQ-C30 Combined Global Health Status /Quality of Life (Items 29 & 30) Scale Combined Score

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Stratified Log Rank); it was not recalculated.
• Result:
  • Pembrolizumab + Lenvatinib vs Pembrolizumab + Placebo: hazard ratio 1.03 (95% CI 0.47–2.26), reported p = 0.942 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Pembrolizumab + Lenvatinib vs Pembrolizumab + Placebo and the control. The interval gives the range of effects that remain compatible with the data.

Time to True Deterioration (TTD) in EORTC QLQ-LC13 Cough (Item 31) Scale Score

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Stratified Log Rank); it was not recalculated.
• Result:
  • Pembrolizumab + Lenvatinib vs Pembrolizumab + Placebo: hazard ratio 1.01 (95% CI 0.43–2.38), reported p = 0.979 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Pembrolizumab + Lenvatinib vs Pembrolizumab + Placebo and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover response, adverse events, treatment discontinuation, and patient-reported symptoms and quality of life. The record also includes time-to-true-deterioration comparisons for global health status and cough. Across the available summaries, estimates do not establish a clear, precise separation between the treatment groups. Some estimates point in one direction and others in another, while the time-to-event intervals leave substantial uncertainty. This broad pattern is consistent with the imprecision of the primary PFS comparison, but secondary endpoints cannot confirm or overturn the primary result.

Secondary outcomes answer different questions from PFS. Objective response concerns tumor response under its assessment definition, while time to deterioration concerns patient-reported change over time. Adverse-event counts address safety experiences, not disease control. Agreement across distinct outcome types can sometimes add context, but these outcomes differ in measurement, analysis population, follow-up, and susceptibility to missing observations. They should not be treated as interchangeable measures of one underlying effect.

Interpretation also needs to account for multiplicity. When many outcomes and comparisons are examined, the chance of observing an apparently notable result by chance can increase. A secondary p-value is less persuasive if it is unadjusted for the full set of analyses, unless a prespecified testing hierarchy or multiplicity procedure controls that risk. The registry record does not report a multiplicity adjustment or a testing hierarchy for the secondary outcomes. Accordingly, secondary results should be treated as supportive or exploratory rather than as independent confirmatory evidence.

The registry lists 14 posted secondary outcomes, but the record details do not provide complete methods and results for every listed measure. No conclusion about omitted or incompletely described secondary outcomes should be inferred. The overall pattern is best used to frame questions for further evaluation, not to claim a definitive treatment effect or absence of one.

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

• Objective Response Rate (ORR) Per RECIST 1.1 as Assessed by BICR
• Number of Participants Who Experienced an Adverse Event (AE)
• Number of Participants Who Discontinued Study Treatment Due to an AE
• Change From Baseline in European Organization for Research and Treatment (EORTC) Quality of Life Questionnaire-Core 30 (QLQ-C30) Combined Global Health Status / Quality of Life (Items 29 & 30) Scale Combined Score
• Change From Baseline in Cough (EORTC Quality of Life Questionnaire-Lung Cancer Module 13 [QLQ-LC13] Item 31) Score
• Change From Baseline in Chest Pain (EORTC QLQ-LC13 Item 40) Score
• Change From Baseline in Dyspnea (EORTC QLQ-C30 Item 8) Score
• Change From Baseline in Physical Functioning (EORTC QLQ-C30 Items 1-5) Score

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is a time-to-event outcome. The registry states that median PFS was estimated using the product-limit, or Kaplan-Meier, method for censored data. The reported comparison uses a stratified log-rank analysis and expresses the result as a hazard ratio. These methods summarize different aspects of the outcome and should be read together rather than treated as identical measures.

What does the Kaplan-Meier median represent?

In time-to-event data, some participants may not have had a recorded event by the end of their observation period or may have stopped contributing follow-up for another reason. Their event times are censored: the analysis knows that the event had not been recorded up to a point, but does not know the later event time. Kaplan-Meier estimation uses the observed event times and the available follow-up to estimate the probability of remaining event-free over time, while accounting for censoring.

The median is the point on that estimated curve where the estimated event-free proportion reaches its midpoint. It is useful because it gives a time-scale summary that is readily interpretable. It does not show the shape of the whole curve, the number or timing of all events, or how censoring was distributed. A median may also be difficult to estimate if the curve does not reach the midpoint during observed follow-up. The registry reports the median and confidence limits for the combination group and the median and lower confidence limit for the comparator; the comparator’s upper confidence limit is not reported in the registry record.

What does the hazard ratio summarize?

A hazard is the instantaneous rate of an event among participants who remain at risk at a given time. A hazard ratio compares those rates between groups according to the analysis model. The reported estimate of 1.18 compares pembrolizumab plus lenvatinib with pembrolizumab plus placebo. It is a relative measure, not a difference in months and not a direct estimate of the probability that an individual participant will experience an event.

The hazard ratio is often useful for time-to-event comparisons because it uses event timing and follow-up, rather than reducing the data to whether an event happened by a single chosen time. Its interpretation can nevertheless depend on the relationship between the groups’ hazards over time. If the relative hazards change, one overall estimate may conceal that variation. The registry record does not report a proportional-hazards assessment, model diagnostics, or a detailed description of how the stratification factors were selected.

Why use a stratified log-rank analysis?

A log-rank test compares time-to-event experience while accounting for the ordering of events and censoring. Stratification allows comparisons to be made within specified strata and then combined, which can account for designated factors in the analysis. The sponsor’s reported method is a stratified log-rank analysis. The particular strata and any other details of the test are not reported in the registry record. The p-value is evidence about compatibility of the observed data with a null comparison under the analysis; it is not the probability that the null hypothesis is true, and it does not measure clinical importance.

The registry gives p = 0.711 for the primary comparison. A large p-value is not proof that the treatment groups are equivalent. Equivalence requires a defined margin and an analysis designed to assess whether differences lie within it. Such an equivalence objective or margin is not reported in the registry record. The confidence interval is therefore important: it shows the range of relative effects reasonably compatible with the data under the analysis, and it spans values that point in different directions.

How should the binary response result be read?

Objective response rate is a binary outcome: participants are classified according to whether they met the response definition during the assessment period. The posted percentages are 33.3% and 39.6%. The registry also gives a sponsor-reported difference in percentage with a 95% confidence interval and p-value. A difference in percentages is an absolute contrast, unlike a risk ratio or odds ratio, which expresses a relative contrast. The sponsor’s reported analysis uses a stratified Miettinen and Nurminen method. The underlying response definition and further details of the analysis are not reported in the registry record details.

Binary outcome estimates can be sensitive to the number of participants and events in each group. Confidence intervals communicate the uncertainty around the estimate, and an interval spanning no difference indicates that the data do not precisely distinguish a difference from no difference at the stated confidence level. These statistical summaries do not determine whether a difference would be important to patients or clinicians.

Why do analysis population and censoring details matter?

Time-to-event estimates depend on who is included, how events are defined and assessed, and how follow-up ends. The registry describes the primary population as Chinese participants randomized to the global study or extension who were evaluable for response. The exact participant flow into this analysis, reasons for exclusion, follow-up distribution, censoring rules, and missing-data procedures are not reported in the registry record. The primary PFS groups have different reported sizes, and safety summaries use other denominators; these records should not be assumed to describe an identical analysis population.

Blinded independent central review is the stated basis for PFS assessment. Blinding can limit the influence of knowledge of treatment assignment on assessment, while central review applies a common assessment process. The registry does not give further information about disagreements between reviewers, local assessments, or adjudication. These methods details help explain what the estimate represents, while the record’s omissions limit independent assessment of implementation.

11. Confidence Intervals and Precision

A confidence interval expresses uncertainty around an estimated effect under the statistical method used. It is not a guarantee that the true value lies within its endpoints, and it does not show the probability that one treatment is better. A narrower interval indicates greater precision; a wider interval indicates that a broader range of effects remains compatible with the observed data and analysis.

For PFS, the median estimate for pembrolizumab plus lenvatinib is 6.1 months, with a 95% confidence interval from 4.2 to 8.7 months. The comparator median is 10.3 months, with a lower confidence limit of 5.6 months. Its upper confidence limit is not reported in the registry record. The hazard ratio is 1.18, and its 95% confidence interval ranges from 0.61 to 2.25. That interval is wide enough to include materially different relative event-rate patterns, including values below and above 1. The result therefore has limited precision about the direction and magnitude of the relative effect.

The reported p-value, 0.711, should not be used alone to judge precision or importance. The confidence interval gives more information about the range of estimates consistent with the analysis. It does not identify which value is most likely, nor does it account for uncertainties not represented by the statistical model, such as limitations in outcome assessment or missing-data procedures.

Secondary estimates also have broad uncertainty. For example, the sponsor-reported objective-response difference has a 95% confidence interval from -27.4 to 11.9. The registry does not report a confidence-limit upper endpoint for the comparator median PFS. Missing interval limits should not be inferred from a plot or reconstructed from other statistics. Overall, the posted results support cautious description of the observed estimates, not a precise claim about the size or direction of treatment differences.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size illustrations are planning calculations based on the observed effects, not the original trial design. The registry’s original design assumptions and sample-size justification are not reported in the registry record. The illustrations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each row considers an effect equal to 100%, 75%, or 50% of the observed effect. The smaller the effect one wishes to detect, the more participants or events are needed to distinguish it from random variation under the stated assumptions.

For binary outcomes such as objective response, the entries are participants per arm for 80% power or participants per arm for 90% power. The 80% column describes the planning size associated with an 80% chance of detecting the specified effect if the assumptions and effect are correct; the 90% column represents a higher target chance and consequently requires a larger planning size. For time-to-event outcomes, the entries are total events across the groups, not participants per arm. Event totals are used because information in a time-to-event comparison depends substantially on how many events are observed, as well as on follow-up and censoring.

These calculations are planning illustrations, not observed power. They do not show the probability that the completed trial detected its true effect, and they do not demonstrate that the study was adequately powered for its original objectives. The displayed figures should not be read as the number enrolled in the trial or as a reconstruction of its protocol. The curve and the rows communicate how the planning target changes when the effect to detect is scaled down; they do not replace the trial’s own design rationale, which is not reported in the registry record.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

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

Worked calculation for this trial

1. Hazard ratio HR = 1.180, so ln(HR) = 0.1655.
2. d = 4 × (1.96 + 0.842)² / [ln(1.180)]² = 1,146.0, rounded up to 1,147 events in total for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
hr <- 1.18                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(4676412)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 1,147 (80% power), 1,535 (90% power)
 75% of observed effect: events = 2,038 (80% power), 2,728 (90% power)
 50% of observed effect: events = 4,585 (80% power), 6,137 (90% power)
Simulated power with 1,147 events: 0.798 (20000 replicates)

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

14. Analyses Reported by the Sponsor

The sponsor-reported primary comparison is a stratified log-rank hazard ratio of 1.18, with a 95% confidence interval from 0.61 to 2.25 and p = 0.711. The posted analysis details do not identify a discrepancy between the reported result and the calculated display, so the sponsor’s analysis is the relevant result to interpret. It is not appropriate to replace that analysis with a separately recalculated estimate.

The sponsor’s method accounts for the time-to-event nature of PFS and is identified as stratified. The registry does not report the strata or enough analysis detail to reproduce the model independently. The hazard ratio should be interpreted as the sponsor’s relative-effect summary, alongside the posted medians and confidence intervals. It is not a difference in median months, and its uncertainty is substantial.

For objective response, the sponsor reports a difference in percentage of -8.4, with a 95% confidence interval from -27.4 to 11.9 and p = 0.793, using a stratified Miettinen and Nurminen method. That sponsor-reported estimate is distinct from an unadjusted recalculation based on the posted group summaries; the analysis methods and estimands need not match. The sponsor’s analysis takes precedence when describing the trial’s reported inferential result. The unadjusted calculation can be useful as a transparent summary, but it should not be presented as the sponsor’s result or as a correction to it.

Reported p-values and confidence intervals remain conditional on the analysis choices and data available to the sponsor. The registry record does not report all details needed to reconstruct those choices, including full stratification and missing-data specifications.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study evaluates multiple endpoints, time points, comparisons, or subgroups. Each additional statistical test creates another opportunity for a chance finding. A prespecified hierarchy or adjustment procedure can control the overall false-positive risk, but it can also affect how evidence is interpreted. The registry brief summary identifies PFS and overall survival as primary hypotheses, and the record lists numerous secondary outcomes. A multiplicity strategy, testing sequence, or adjusted significance threshold is not reported in the registry record.

Without that information, a nominal p-value for a secondary endpoint should not automatically be treated as confirmatory. The primary PFS comparison can be described using its reported estimate and interval, but the record does not explain how the two stated primary hypotheses were handled together for formal inference. The overall-survival result itself is not reported in the registry record details.

Missing data can affect both time-to-event and patient-reported outcomes. In time-to-event analysis, censoring handles incomplete event-time observation under assumptions about the censoring process. The registry states that the Kaplan-Meier method was used, but the specific censoring rules, follow-up completeness, and reasons for censoring are not reported in the registry record. For patient-reported outcomes, unanswered questionnaires or missed visits can change the available analysis population and may introduce bias if missingness relates to health status or treatment. The record does not report the missing-data strategy for these measures.

The analysis population is described as Chinese participants randomized to the global study or extension who were evaluable for response. Further participant-flow details, exclusions, and the handling of participants without evaluable assessments are not reported in the registry record. The reported group sizes differ across the PFS, response, and safety summaries. This may reflect different outcome populations or assessment availability, but the reason is not reported and should not be assumed.

These gaps do not invalidate the posted summaries, but they limit assessment of robustness. Interpretation should distinguish what the registry reports from what would require protocol, statistical analysis plan, or participant-level information.

16. 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: pneumonia (8.2%), haemoptysis (8.2%), pneumonitis (6.1%), diarrhoea (4.1%), pancreatitis (4.1%).

The safety summary reports serious adverse events in 24 of 49 participants in the lenvatinib-plus-pembrolizumab group and 17 of 58 participants in the placebo-plus-pembrolizumab group. It also lists deaths of 28 and 32 participants, respectively. These are counts as recorded in the safety summary; the registry record does not provide timing, causes, or an attribution assessment for the deaths in the registry record. They should not be assumed to be treatment-related.

The listed serious adverse-event terms include pneumonia and haemoptysis, each reported as 8.2% in the combination group; pneumonitis, 6.1%; and diarrhoea, pancreatitis, and chest discomfort, each 4.1%. These percentages describe the listed terms in the registry summary and do not establish causation. The record also gives counts for participants with other adverse events, but the definitions and severity categories behind that summary are not further described in the registry record.

Safety counts are not direct measures of comparative risk without attention to analysis populations, exposure duration, event definitions, and follow-up. The safety denominators differ from those in the primary endpoint summaries. The registry record does not report exposure-adjusted rates or enough detail to infer why the denominators differ. These summaries describe recorded events; they do not, by themselves, establish that one regimen caused an event or that a regimen is safe for an individual.

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

The registry results describe PFS in the stated China extension population and provide estimates for the two randomized groups. The median PFS estimate was 6.1 months with pembrolizumab plus lenvatinib and 10.3 months with pembrolizumab plus placebo. The sponsor-reported hazard ratio was 1.18, with a broad 95% confidence interval from 0.61 to 2.25 and p = 0.711. The interval leaves considerable uncertainty about the relative event-rate difference. The data do not provide a precise basis for concluding that the groups differ, and the non-significant result does not establish equivalence.

The result is specific to the recorded population, endpoint definition, assessment method, follow-up, and analysis. It is not automatically generalizable to all people with NSCLC, other PD-L1 groups, other treatment settings, or treatment combinations that differ from those described. The registry record does not report enough demographic or clinical detail to assess how closely this extension population resembles other populations.

PFS is not overall survival. Although the brief summary names overall survival as a primary hypothesis, an overall-survival result is not reported in the registry record details. The posted response and patient-reported outcome summaries provide additional context but have their own uncertainty, multiplicity, and missing-data limitations. Safety counts likewise describe recorded events and do not establish causation.

Statistical evidence is one part of evaluating a treatment comparison. Confidence intervals communicate uncertainty in the estimates, while clinical relevance depends on the magnitude and nature of effects, the outcomes that matter to patients, and the quality and completeness of the evidence. Those judgments cannot be settled by a p-value or a median alone. The registry record supports a cautious account of the reported estimates and their limitations, rather than a medical recommendation or a broad claim about treatment benefit.

18. Why This Trial Matters Statistically

This trial illustrates several important principles for reading clinical time-to-event results. First, a median and a hazard ratio are not interchangeable. The medians describe points on estimated PFS curves, while the hazard ratio summarizes a relative event-rate comparison. When the estimates appear to point in different directions, neither should be discarded; each answers a different statistical question, and the full curve and analysis details would help explain their relationship.

Second, precision matters. The primary hazard-ratio interval is broad and includes values on either side of 1. A result with a large p-value is not evidence that the treatments are identical; the interval makes clear that meaningful uncertainty remains. This distinction between failure to show a difference and evidence of equivalence is central to sound interpretation.

Third, the registry distinguishes the extension’s enrollment from outcome-specific analysis groups. The PFS, response, and safety summaries report different group sizes. Recognizing those denominators helps prevent the mistaken assumption that every result uses the same population. The registry does not report the participant flow or the reasons for the differences, so the limitation should remain explicit.

Finally, the record identifies multiple primary hypotheses and a range of secondary outcomes, while not reporting a multiplicity strategy or full missing-data methods. Such details affect how strongly secondary findings can be interpreted. The trial is therefore a useful example of why statistical reading requires attention to endpoint definitions, analysis populations, uncertainty, and reporting completeness—not just a single effect estimate.

19. 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 PFS comparison is imprecise: the hazard-ratio confidence interval ranges from 0.61 to 2.25. The comparator median’s upper confidence limit is not reported in the registry record. This limits how fully the median estimates can be compared using their reported intervals.

The primary population is described as Chinese participants randomized to the global study or extension who were evaluable for response. The registry record does not provide detailed participant flow, the reasons for exclusions, or enough demographic and clinical characteristics to assess representativeness. The reported group sizes also vary between outcome summaries, and the reasons are not reported.

The registry does not report the original sample-size justification, its own design assumptions, or details needed to reproduce the analyses, including the stratification factors, proportional-hazards assessment, and censoring rules. The planning illustrations based on observed effects should not be mistaken for the trial’s original design.

The brief summary names overall survival as a primary hypothesis, but an overall-survival result is not reported in the registry record details. A multiplicity plan, secondary-outcome testing hierarchy, and detailed missing-data procedures are also not reported. These omissions constrain interpretation of secondary results and the robustness of patient-reported outcome comparisons.

Finally, posted safety counts do not include exposure-adjusted rates or sufficient information about event timing and attribution in the registry record. They describe recorded events but cannot alone establish causation or support a complete comparative safety assessment.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Efficacy and Safety Study of Pembrolizumab (MK-3475) With or Without Lenvatinib (MK-7902/E7080) in Adults With Programmed Cell Death-Ligand 1 (PD-L1)-Positive Treatment-naïve Nonsmall Cell Lung Cancer (NSCLC) [MK-7902-007/E7080-G000-314/LEAP-007] - China Extension Study — NCT04676412.
• Registry record last updated 2025-04-04; status last verified 2025-03.
• Results first posted 2022-11-09.
• 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.

23. Record Summary

The registry record describes a completed, randomized Phase 3 parallel-group study with triple masking, comparing pembrolizumab plus lenvatinib with pembrolizumab plus placebo in treatment-naïve adults with metastatic NSCLC and PD-L1 TPS at least 1%. The China extension lists an enrollment of 107 and is associated with the global study NCT03829332.

The primary posted result is PFS assessed by blinded independent central review under RECIST 1.1. Median PFS was 6.1 months for the combination and 10.3 months for the comparator. The sponsor-reported hazard ratio was 1.18, with a 95% confidence interval from 0.61 to 2.25 and p = 0.711. The estimate is uncertain and does not establish equivalence.

The record also includes response, patient-reported, and safety summaries. Important analysis details—including multiplicity, missing-data handling, and the original sample-size rationale—are not reported in the registry record. An overall-survival result is not reported in the registry record.