This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for M7824 Versus Pembrolizumab as a First-line (1L) Treatment in Participants With Programmed Death-ligand 1 (PD-L1) Expressing Advanced Non-small Cell Lung Cancer (NSCLC) (NCT03631706), a randomized phase 3 study. The primary outcome is Progression-Free Survival (PFS) According to Response Evaluation Criteria in Solid Tumors (RECIST Version 1.1) Assessed by Independent Review Committee (IRC).
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03631706 |
| Phase | Phase 3 |
| Condition | Non-small Cell Lung Cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None (open label) |
| Enrollment | 304 |
| Arms in the primary outcome | 2 |
| Lead sponsor | EMD Serono Research & Development Institute, Inc. |
| Results posted | Yes |
2. Clinical Question
Population
The study concerned participants with advanced non-small cell lung cancer (NSCLC) whose tumors expressed PD-L1. The brief summary further describes high PD-L1 tumor expression and no EGFR mutation or ALK translocation. The primary analysis population was the full analysis set, defined as all participants randomized to study intervention. More detailed eligibility criteria are not reported in the registry record.
Intervention
The experimental intervention was M7824, also called bintrafusp alfa in the study summary. The registry identifies it as a drug, but the arm description does not provide dosing, administration schedule, treatment duration, or rules for stopping treatment. Those details are not reported in the registry record.
Comparator
The active comparator was pembrolizumab, also identified as a drug. The study was open label, so participants and clinical teams were not masked to assigned treatment. The registry gives no further arm-level treatment details in the arm descriptions.
Primary question
The primary endpoint was progression-free survival (PFS), assessed by an independent review committee using RECIST version 1.1. It measures time from randomization until documented disease progression or death, whichever occurs first. The central statistical question is how the observed PFS summaries compare between randomized groups and how much uncertainty the record gives for that comparison. The registry reports arm-specific medians and full ranges, but no sponsor-reported hazard ratio or confidence interval for the treatment contrast.
3. Trial Design
This was a phase III, multicenter, randomized, open-label, controlled, parallel-group study comparing M7824 with pembrolizumab as first-line treatment for participants with advanced PD-L1-expressing NSCLC. The registry lists 304 participants as enrolled, with 152 assigned to each arm in the primary endpoint results. The study was completed. Its recorded start date is 2018-10-01, and its primary completion date is 2021-06-07.
The brief summary describes an adaptive design that allowed the option to recruit up to 584 patients based on prespecified rules. The registry record does not explain those rules, whether the option to expand enrollment was used, or what decisions were made under the adaptive design. It also does not report details of any interim analyses, stopping boundaries, or design adaptations. These details should not be inferred from the word “adaptive.”
Random allocation supports comparison of outcomes between groups by reducing systematic differences at assignment, although it does not guarantee identical groups in every measured or unmeasured characteristic. The open-label design means treatment assignment was known to participants and clinical teams. PFS assessment by an independent review committee provides a separate assessment of progression using RECIST version 1.1; the registry does not describe committee procedures or whether the committee was masked to treatment.
The primary population was the full analysis set, defined in the record as all randomized participants. The primary endpoint timeframe is described as extending from randomization until progression or death, assessed approximately up to 746 days. Treatment schedules, follow-up procedures, protocol deviations, and the details of the analysis population for other outcomes are not reported in the registry record.
4. Participant Flow and Accounting
The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.
Overall Study
| Stage | M7824 | Pembrolizumab |
|---|---|---|
| STARTED | 152 | 152 |
| Treated | 151 | 152 |
| COMPLETED | 151 | 152 |
| NOT COMPLETED | 1 | 0 |
Reasons for not completing this period:
| Reason | M7824 | Pembrolizumab |
|---|---|---|
| Randomized, not treated | 1 | 0 |
Analysis population for the primary outcome, as defined in the registry: Full analysis set included all participants who were randomized to study intervention.
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: Full analysis set included all participants who were randomized to study intervention.
| Characteristic | Unit | M7824 | Pembrolizumab | Total |
|---|---|---|---|---|
| Participants analysed | 152 | 152 | 304 | |
| Age, Continuous | Years | 67 (SD 9.56) | 67 (SD 10.17) | 67 (SD 9.85) |
| Sex: Female, Male: Female | Participants | 42 | 36 | 78 |
| Sex: Female, Male: Male | Participants | 110 | 116 | 226 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 13 | 11 | 24 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 138 | 138 | 276 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 1 | 3 | 4 |
| Race/Ethnicity, Customized: American Indian or Alaska Native | Participants | 0 | 0 | 0 |
| Race/Ethnicity, Customized: Asian | Participants | 50 | 55 | 105 |
| Race/Ethnicity, Customized: Native Hawaiian or Other Pacific Islander | Participants | 0 | 0 | 0 |
| Race/Ethnicity, Customized: Black or African American | Participants | 1 | 0 | 1 |
| Race/Ethnicity, Customized: White | Participants | 91 | 79 | 170 |
| Race/Ethnicity, Customized: Unknown or Not Reported | Participants | 7 | 14 | 21 |
| Race/Ethnicity, Customized: Other | Participants | 3 | 4 | 7 |
6. Primary Endpoint
The registry lists the primary outcome as Progression-Free Survival (PFS) According to Response Evaluation Criteria in Solid Tumors (RECIST Version 1.1) Assessed by Independent Review Committee (IRC), measured over Time from randomization of study drug until the first documentation of PD or death, assessed approximately up to 746 days.
| Element | Registry entry |
|---|---|
| Result type | Median |
| Measure of spread | Full Range |
| Unit | months |
| Endpoint classification | Time-to-Event Analysis |
Progression-free survival is a time-to-event endpoint. For each participant, it is the time from randomization until the first documented disease progression or death from any cause, whichever happens first. The registry defines progression using RECIST version 1.1: an increase in the sum of the longest diameters from the smallest recorded value, or the appearance of at least one new lesion. The endpoint was assessed by an independent review committee.
PFS is used to capture how long participants remain alive without documented progression. It combines progression and death into a single outcome, so an event can occur through either route. A median summarizes the time by which half of the observed population has had an event, subject to the endpoint’s analysis and censoring rules. It does not describe every participant’s experience, and it is not an estimate of the treatment difference by itself.
The registry reports a median PFS of 7.0 months for M7824, with a full range from 0.0 to 19.4 months, and 11.1 months for pembrolizumab, with a full range from 0.0 to 18.0 months. These ranges are not confidence intervals. The registry record does not report a hazard ratio, a confidence interval for the treatment contrast, or a sponsor-reported significance test for the primary endpoint. Consequently, the arm-specific medians alone do not quantify the uncertainty of a between-group effect.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry reports median PFS of 7.0 months for M7824 and 11.1 months for pembrolizumab. The recorded full ranges are 0.0 to 19.4 months and 0.0 to 18.0 months, respectively. These are descriptive summaries of the observed time-to-event data. The ranges indicate the recorded endpoints of the observed values; they do not express the precision of the medians or uncertainty about a treatment contrast.
A median is a useful summary when follow-up times and event times are central to the research question, but it does not reveal the shape of the full survival curves. Different patterns over time can yield similar medians, and medians do not show whether a difference is consistent throughout follow-up. Interpreting a time-to-event comparison usually requires a measure of the relative event rate over time, such as a hazard ratio, together with its uncertainty, or another prespecified comparison. Such a sponsor-reported measure is not reported in the registry record.
The registry’s outcome information therefore supports a descriptive statement about the reported medians, but not a complete inferential conclusion about the primary comparison. It gives no primary-endpoint confidence interval or p-value for the treatment contrast. The analysis listing also contains no reported analyses. Without a reported comparative estimate and its precision, the magnitude and statistical uncertainty of the between-group difference cannot be assessed from the primary summaries alone.
A secondary response outcome provides a different kind of information: the reported percentages are 46.7% for M7824 and 51.3% for pembrolizumab, with respective 95% confidence intervals of 38.6% to 55.0% and 43.1% to 59.5%. A recalculated unadjusted risk difference is 0.046, with a 95% confidence interval from -0.066 to 0.158 and an unadjusted p-value of 0.422. This is a secondary, unadjusted calculation, not a substitute for the primary time-to-event analysis. Its interval includes no difference and also allows effects in either direction. No adjustment for testing multiple outcomes is reported.
These results should not be read as proving equivalence or establishing that one intervention is better. Failure to report a treatment-effect estimate is not evidence that the effect is absent. Conversely, descriptive differences between arm-specific medians do not establish that the difference is statistically reliable or clinically important. The registry record does not report a clinical-importance threshold for PFS or the response outcome.
9. Statistical Methods Explained
The primary endpoint is time-to-event data: each participant contributes a time from randomization to progression or death, or a censored observation if the event is not recorded during the relevant follow-up. Censoring means that the event time is not fully observed; it does not mean that the participant experienced no event. The registry describes the endpoint and approximate assessment period but does not report the censoring rules, the number censored, or the analysis procedures used to handle censoring.
What does the median tell us?
The median is the point in the event-time distribution at which half of the population has experienced the event, under the analysis used. It is a compact description of the typical event time, not an estimate of the difference between randomized groups. A median also cannot show all features of the survival experience: the curves might separate early, late, or change relative ordering over time. The registry reports arm-specific medians, but not the survival curves or a sponsor-reported comparative estimate in the analysis listing.
The reported ranges are full ranges, not confidence intervals. A confidence interval instead describes uncertainty around an estimate under a statistical model and analysis. The absence of a primary confidence interval means that the precision of the median treatment contrast is not available in the registry record. It would be inappropriate to treat the range endpoints as bounds on a treatment effect.
Why is a hazard ratio often reported?
A hazard ratio compares the instantaneous event rates between groups over follow-up, subject to the assumptions of the model used to estimate it. It is a relative time-to-event measure and is not the same as a ratio of medians or a difference between medians. The registry record does not report a sponsor hazard ratio for this primary endpoint. No hazard-ratio interpretation should therefore be attached to the two medians.
In general, a hazard ratio is most straightforward to interpret when the relative event rate is reasonably stable over time. If that assumption is not suitable, a single hazard ratio may not capture the whole pattern, and other summaries may be useful. Whether proportional hazards were assessed, what model was specified, and whether an alternative time-to-event analysis was planned are not reported in the registry record. These are general methodological considerations, not claims about the analysis conducted in this study.
How should the response comparison be read?
The response outcome is a percentage, or proportion, of participants meeting the outcome definition. Its confidence interval represents uncertainty around the arm-specific percentage under the calculation used. A risk difference expresses an absolute contrast between proportions. The recalculated unadjusted risk difference is 0.046, with a 95% confidence interval from -0.066 to 0.158. Because this interval includes zero, the posted summary does not establish a difference under that unadjusted comparison. The unadjusted p-value is 0.422. This calculation does not account for multiplicity or replace any sponsor analysis; no sponsor-reported analyses are listed for this secondary outcome.
These recalculations use posted summary statistics and are unadjusted. They should be distinguished from analyses conducted under the study protocol, which are not reported in the registry record. General interpretation of a confidence interval is that it gives a range of estimates reasonably compatible with the data and calculation method; it is not a guarantee that the true value falls within a particular interval.
What does multiplicity change?
When several outcomes or comparisons are examined, the chance of at least one apparently notable result can increase if each is assessed separately without adjustment. A prespecified testing strategy can control that issue, for example by setting an order for hypotheses or adjusting the significance threshold. The registry lists several secondary outcomes, but a multiplicity strategy, hierarchy, or adjusted inference is not reported in the registry record. The secondary response comparison should therefore be treated as exploratory or descriptive unless a separate analysis plan establishes otherwise.
The registry also includes sample-size planning illustrations based on fractions of the observed response effect, using a two-sided significance level of 0.05, equal allocation, and a normal approximation. Such illustrations describe hypothetical planning requirements under their assumptions; they are not observed power, do not establish what the original trial was designed to detect, and do not explain the adaptive enrollment rules. The study’s own sample-size justification and statistical analysis plan are not reported in the registry record.
Why does over-dispersion matter in general?
For outcomes modeled as counts, a simple model may assume variability is tightly linked to the mean. When the observed variation is greater than that assumption allows, over-dispersion can make standard errors too small and confidence intervals too narrow. More flexible models or robust variance methods may be considered in general. The primary outcome is time-to-event, not a count endpoint, and the registry record does not report an over-dispersion analysis or a reason to apply one to the primary comparison.
10. Multiplicity, Missing Data and Analysis Populations
The registry record lists multiple secondary outcomes alongside the primary PFS endpoint. Testing several outcomes can create a multiplicity problem: if each is assessed separately without a planned adjustment, the chance of at least one apparently unusual result can be higher than for a single test. A protocol may address this with a testing sequence, adjusted significance levels, or another strategy. No such strategy is reported in the registry record, so the secondary comparisons should not be treated as if they were protected by a stated confirmatory testing plan.
The record does not report a statistical analysis plan, a sample-size justification, the primary hypothesis test, interim-analysis details, or the rules that governed the adaptive enrollment option. The brief summary says recruitment could increase up to 584 patients under prespecified rules, but the rules and their implementation are not described. The enrollment recorded is 304. The record does not explain whether the planned adaptive option affected recruitment or analysis.
Missing-data procedures are also not reported in the registry record. For a time-to-event endpoint, incomplete follow-up is commonly handled through censoring, but the censoring rules, loss-to-follow-up information, and sensitivity analyses are not described. For the response outcome, the registry does not state how missing assessments or unevaluable participants were handled. No assumptions about these procedures should be inferred from the posted percentages or medians.
The primary population is defined as the full analysis set, including all participants randomized to study intervention. That definition is useful, but the record does not provide a detailed account of exclusions, protocol deviations, or the analysis populations used for each secondary endpoint. The arm-specific safety denominators are also not identical to the primary arm counts, and the registry does not explain the difference. These reporting gaps limit assessment of how robust the reported summaries are to analysis choices and incomplete observations.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| M7824 | 90 | 151 | 59.6% |
| Pembrolizumab | 61 | 152 | 40.1% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| M7824 | 49 of 151 |
| Pembrolizumab | 43 of 152 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 5% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| M7824 | 140 of 151 |
| Pembrolizumab | 135 of 152 |
Most frequent serious adverse event terms
| Event term | Organ system | M7824 | Pembrolizumab |
|---|---|---|---|
| Pneumonia | Infections and infestations | 9 of 151 | 12 of 152 |
| Disease progression | General disorders | 6 of 151 | 7 of 152 |
| Dyspnoea | Respiratory, thoracic and mediastinal disorders | 6 of 151 | 2 of 152 |
| Pneumothorax | Respiratory, thoracic and mediastinal disorders | 5 of 151 | 1 of 152 |
| Anaemia | Blood and lymphatic system disorders | 4 of 151 | 1 of 152 |
| Hypercalcaemia | Metabolism and nutrition disorders | 2 of 151 | 3 of 152 |
| Squamous cell carcinoma of skin | Neoplasms benign, malignant and unspecified (incl cysts and polyps) | 5 of 151 | 0 of 152 |
| Haemoptysis | Respiratory, thoracic and mediastinal disorders | 4 of 151 | 1 of 152 |
| Infusion related reaction | Injury, poisoning and procedural complications | 2 of 151 | 2 of 152 |
| Pleural effusion | Respiratory, thoracic and mediastinal disorders | 1 of 151 | 3 of 152 |
Most frequent non-serious adverse event terms
| Event term | Organ system | M7824 | Pembrolizumab |
|---|---|---|---|
| Pruritus | Skin and subcutaneous tissue disorders | 52 of 151 | 42 of 152 |
| Rash | Skin and subcutaneous tissue disorders | 48 of 151 | 25 of 152 |
| Anaemia | Blood and lymphatic system disorders | 46 of 151 | 18 of 152 |
| Asthenia | General disorders | 26 of 151 | 30 of 152 |
| Fatigue | General disorders | 30 of 151 | 26 of 152 |
| Decreased appetite | Metabolism and nutrition disorders | 25 of 151 | 26 of 152 |
| Diarrhoea | Gastrointestinal disorders | 24 of 151 | 24 of 152 |
| Dyspnoea | Respiratory, thoracic and mediastinal disorders | 29 of 151 | 18 of 152 |
| Cough | Respiratory, thoracic and mediastinal disorders | 23 of 151 | 22 of 152 |
| Pyrexia | General disorders | 28 of 151 | 15 of 152 |
Serious events with the highest arm-level rate: pneumonia (7.9%), disease progression (4.6%), dyspnoea (4.0%), squamous cell carcinoma of skin (3.3%), pneumothorax (3.3%).
The safety record gives counts of participants with serious adverse events and other adverse events, along with counts of deaths. For M7824, the posted serious-event count is 90 among 151 at risk, the other-event count is 140 among 151 at risk, and the death count is 49. For pembrolizumab, the corresponding counts are 61 among 152 at risk, 135 among 152 at risk, and 43 deaths. These are counts as reported; no percentages have been calculated from them.
The safety denominators differ from the primary endpoint arm counts for M7824. The registry does not explain this difference or describe how event records were collected, whether participants could contribute multiple events, or how long each participant was observed for safety. A count of participants with an event is not the same as a count of event episodes. The posted counts also do not, by themselves, establish that an event was caused by treatment.
The registry lists pneumonia as a serious adverse event term at 7.9%, followed by disease progression at 4.6%, dyspnoea at 4.0%, squamous cell carcinoma of skin and pneumothorax at 3.3% each, and anaemia at 2.6%. The record does not provide enough detail alongside these terms to interpret their timing, recurrence, or treatment attribution. Safety counts describe reported events; they do not alone establish comparative risk or causation.
12. What the Result Does, and Does Not, Mean
The registry supports a descriptive account of the recorded trial: it was a randomized, open-label comparison of M7824 and pembrolizumab in advanced PD-L1-expressing NSCLC, and the reported median PFS was 7.0 months in the M7824 arm and 11.1 months in the pembrolizumab arm. Those arm-specific summaries do not, by themselves, determine whether the observed difference is statistically reliable, clinically important, or attributable to treatment. The record does not report a sponsor hazard ratio, a confidence interval for the primary treatment contrast, or a primary-endpoint significance test.
The posted response percentages and their unadjusted comparison offer additional descriptive information, but they are secondary and do not replace the primary time-to-event analysis. The unadjusted confidence interval includes no difference, and the registry does not report a multiplicity plan or sponsor-reported analysis for that outcome. Results from multiple endpoints should not be treated as independent confirmation when the testing strategy is unknown.
The randomized design is a strength for comparing assigned groups, while the open-label design and incomplete reporting of analysis details are relevant when interpreting the record. Independent committee assessment was used for the primary progression endpoint, but the registry does not describe committee masking or procedures. Treatment schedules, handling of missing observations, and the primary comparative analysis are also not reported.
These findings do not establish that either treatment works or does not work. They describe the values recorded in the registry and the limits of what can be inferred from them. The record does not define a threshold for a clinically meaningful PFS difference, and it does not provide the complete comparative evidence needed to judge the primary effect and its precision. Interpretation should remain bounded by the population, outcomes, and analyses actually reported.
13. Why This Trial Matters Statistically
This trial raises a common statistical distinction: an arm-specific outcome summary is not the same as a treatment-effect estimate. Median PFS values describe each group separately, while a comparative estimate needs to express the contrast between groups and its uncertainty. The registry reports the medians and full ranges but does not report a sponsor hazard ratio or confidence interval for the primary comparison. That gap matters because the reader cannot assess the precision of the primary treatment contrast from the medians alone.
The randomized, parallel-group design is suited to a direct comparison of assigned treatments. The use of independent review committee assessment for PFS is also relevant because disease progression is determined using imaging-based criteria. At the same time, the open-label design and the absence of reported details on committee masking, missing-data procedures, analysis methods, and multiplicity limit what can be concluded from the registry record.
The adaptive-design description is another reason to distinguish a study label from a complete analysis plan. The record states that recruitment could increase up to 584 patients under prespecified rules, but it does not report those rules or how the option was used. Without those details, readers cannot assess how adaptation related to enrollment or inference.
More broadly, the record illustrates why results need both estimates and uncertainty, and why secondary outcomes require context about prespecification and multiplicity. The secondary response comparison is unadjusted and does not resolve the missing primary comparative estimate. Clear reporting of the estimand, analysis population, censoring and missing-data rules, and planned testing strategy would be needed to make the statistical evidence easier to evaluate.
14. 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 most direct limitation of the registry record is the absence of a sponsor-reported comparative estimate for the primary PFS endpoint. The record gives arm-specific medians and full ranges, but no hazard ratio, confidence interval for the treatment contrast, or primary significance test. The difference between the medians cannot replace those measures or show how precisely the comparison was estimated.
The registry does not report the statistical analysis plan, sample-size justification, or the adaptive design’s prespecified decision rules. It therefore does not allow assessment of how the analysis was intended to address the primary hypothesis, whether any interim looks were planned, or how the potential enrollment increase was handled. The brief summary notes the option to recruit up to 584 patients, but the registry does not explain its implementation.
Other analysis details are also absent: the censoring rules and counts for PFS, missing-data handling, sensitivity analyses, and the analysis populations for secondary outcomes are not reported in the registry record. The primary population is defined as all randomized participants, but the record does not give a full account of exclusions or protocol deviations. For safety, the M7824 at-risk denominator differs from its primary endpoint arm count, with no explanation in the record.
The study was open label, although PFS was assessed by an independent review committee. The registry does not state whether the committee was masked to assigned treatment or describe review procedures. It also does not provide treatment schedules or detailed follow-up information beyond the approximate assessment period. These limitations concern what can be evaluated from the registry record; they do not establish that the underlying study lacked such procedures or analyses.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: M7824 Versus Pembrolizumab as a First-line (1L) Treatment in Participants With Programmed Death-ligand 1 (PD-L1) Expressing Advanced Non-small Cell Lung Cancer (NSCLC) — NCT03631706.
- Registry record last updated 2025-04-08; status last verified 2025-03.
- Results first posted 2022-06-29.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
NCT03631706 is a completed phase III, multicenter, randomized, open-label, parallel-group comparison of M7824 and pembrolizumab as first-line treatment for advanced PD-L1-expressing NSCLC. The registry records 304 participants enrolled. The primary endpoint is PFS assessed by an independent review committee using RECIST version 1.1, in a full analysis set defined as all randomized participants.
The registry reports median PFS of 7.0 months for M7824 and 11.1 months for pembrolizumab, with full ranges of 0.0 to 19.4 months and 0.0 to 18.0 months, respectively. It does not report a sponsor hazard ratio or confidence interval for the primary treatment contrast. A secondary response outcome has posted arm-level percentages and an unadjusted comparison; other listed secondary outcomes do not have numerical results reported in the registry record. The record also lists safety counts, but provides limited context for interpreting denominators, event timing, and attribution.