This page separates information reported in the trial record from statistical interpretation and educational explanation. The ClinicalTrials.gov record identifies two primary endpoints and state that results have been posted, but the registry-reported statistical-analyses section contains no formal comparative analysis estimates, confidence intervals, or p-values.
1. Trial at a Glance
JAVELIN Merkel 200 was a phase 2, non-randomized, single-group treatment study evaluating avelumab in participants with Merkel cell carcinoma. The registry reports 204 participants, two study parts, no masking, and a treatment-focused primary purpose.
| Feature | JAVELIN Merkel 200 |
|---|---|
| Trial name | JAVELIN Merkel 200 |
| Brief title | Avelumab in Participants With Merkel Cell Carcinoma (JAVELIN Merkel 200) |
| Phase | Phase 2 |
| Condition | Carcinoma, Merkel Cell |
| Allocation | Non-randomized |
| Design model | Single group |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 204 |
| Intervention | Avelumab |
| Status | Completed |
| Start | 2014-07-03 |
| Primary completion | 2019-05-02 |
| Lead sponsor | EMD Serono Research & Development Institute, Inc. |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT02155647 |
2. Clinical Question
The registered clinical question concerns the activity of avelumab in participants with Merkel cell carcinoma. Unlike a conventional randomized comparative trial, this study is registered as non-randomized and single-group, so its primary statistical question is centered on the observed response experience within the study rather than a randomized treatment-versus-control contrast.
Population
Participants with Merkel cell carcinoma.
Intervention
Avelumab.
Comparator
No randomized comparator is registered. The design is non-randomized and single-group.
Primary questions
How many participants achieve a confirmed best overall response in Part A, and what proportion achieve a durable response in Part B?
3. Trial Design
Confirmed best overall response
- Primary endpoint: number of participants with confirmed best overall response.
- Response assessed according to RECIST 1.1.
- Response adjudicated by an Independent Endpoint Review Committee.
- Registered time frame: up to 113 weeks.
Durable response rate
- Primary endpoint: durable response rate.
- Durable response requires an objective response according to RECIST 1.1 with duration of at least 6 months.
- Response determined by an Independent Endpoint Review Committee.
- Registered time frame: up to 161 weeks.
4. Study Population and Enrollment
The registry reports total enrollment of 204 participants. The ClinicalTrials.gov record identifies 88 participants at risk in Part A and 116 participants at risk in Part B, which together account for the full reported enrollment.
| Study component | Participants at risk for reported serious AEs | Intervention |
|---|---|---|
| Part A | 88 | Avelumab |
| Part B | 116 | Avelumab |
| Total | 204 | Avelumab |
The ClinicalTrials.gov record does not provide a baseline demographic table, disease-characteristic table, subgroup estimates, or other population descriptors.
5. Primary Endpoints
The study has two registered primary endpoints. They belong to different statistical categories: Part A is explicitly characterized as a binary endpoint, while Part B is labeled by the ClinicalTrials.gov record as a time-to-event endpoint even though its operational definition expresses durable response as a percentage of participants meeting a duration criterion.
| Endpoint | Registered definition | Time frame | Endpoint type |
|---|---|---|---|
| Part A: Number of Participants With Confirmed Best Overall Response (BOR) as Per RECIST 1.1 | Confirmed BOR was determined according to RECIST 1.1, adjudicated by an Independent Endpoint Review Committee, and defined as the best response of complete response, partial response, stable disease, or progressive disease recorded from the relevant study time through disease progression/recurrence. | Up to 113 weeks | Binary |
| Part B: Durable Response Rate (DRR) | Durable response is defined as an objective response, consisting of confirmed complete response or confirmed partial response according to RECIST 1.1, determined by the Independent Endpoint Review Committee, with a duration of at least 6 months. DRR is determined as the percentage of participants with an objective response meeting the durability criterion. | Up to 161 weeks | Time-to-event |
6. Planned Analysis
The ClinicalTrials.gov record states that results are posted for both primary endpoints, but no formal statistical analyses were posted. Consequently, the ClinicalTrials.gov record does not provide a numerical effect estimate, confidence interval, or p-value for either primary endpoint.
Part A: Confirmed Best Overall Response
What the endpoint measures
Registered time frame: up to 113 weeks
The ClinicalTrials.gov record does not report a formal statistical-analysis estimate, confidence interval, or p-value.
For a single-group binary response endpoint, the natural descriptive analysis would be the proportion of participants meeting the prespecified response definition. Statistical uncertainty would ordinarily be summarized with a confidence interval for that proportion. If the study protocol specified a benchmark response rate, a one-sample hypothesis test could additionally compare the observed response rate with that benchmark.
What the estimate would mean: a response proportion would describe the fraction of evaluable participants whose best recorded response satisfied the specified RECIST 1.1 response definition.
What it would not mean: in a single-group study, the response proportion would not by itself establish how much better avelumab performs than another treatment. Without a randomized comparator or an explicitly defined external benchmark, the estimate is descriptive rather than a randomized comparative treatment effect.
Why the confidence interval matters: a confidence interval would quantify statistical uncertainty around the estimated response proportion. It would not describe the range of responses experienced by individual participants.
Why a p-value would not measure effect size: if a one-sample test were performed against a prespecified benchmark, the p-value would describe evidence against that null hypothesis. It would not quantify the magnitude of response or the probability that the treatment works.
Part B: Durable Response Rate
What the endpoint measures
Registered time frame: up to 161 weeks
The ClinicalTrials.gov record does not report a formal statistical-analysis estimate, confidence interval, or p-value.
The registry defines durable response as a confirmed objective response with a duration of at least 6 months. For a single-group proportion, a natural descriptive analysis would report the percentage of participants satisfying that criterion together with a confidence interval. The duration requirement introduces a time component, because a participant cannot be classified as having a response lasting at least 6 months until sufficient follow-up has accrued.
What the estimate would mean: the DRR would represent the percentage of participants with a confirmed complete or partial response whose response duration reached at least 6 months under the registered definition.
What it would not mean: DRR would not be equivalent to the probability that a randomly selected patient remains progression-free for 6 months, nor would it establish a comparative benefit against another therapy.
Why the confidence interval matters: the precision of a durable-response estimate depends on the number of participants contributing information and the number meeting the response-duration criterion. A confidence interval would make that uncertainty explicit.
Why censoring and follow-up matter: participants need adequate observation to establish that a response lasted at least 6 months. Participants with insufficient follow-up require careful handling according to the prespecified endpoint rules; the ClinicalTrials.gov record does not provide those detailed rules.
7. How the Primary Endpoints Fit Together
The two primary endpoints capture different aspects of tumor response. Confirmed BOR identifies the participant's best response category, whereas DRR adds a minimum duration requirement to an objective response.
| Question | Part A BOR | Part B DRR |
|---|---|---|
| Is an objective response required? | The registered definition includes CR and PR among the possible best responses, as well as SD and PD. | Yes. The response must be confirmed CR or confirmed PR. |
| Is response duration part of the definition? | The registry-reported definition does not impose the same explicit 6-month requirement. | Yes. The objective response must have a duration of at least 6 months. |
| Review mechanism | Independent Endpoint Review Committee | Independent Endpoint Review Committee |
| Registered time frame | Up to 113 weeks | Up to 161 weeks |
| Registry endpoint type | Binary | Time-to-event |
8. Statistical Methodology
Single-group response analysis
The core statistical structure is different from that of a randomized controlled trial. With a single treatment group, the primary efficacy estimand is naturally expressed as an observed response proportion rather than a treatment-versus-control effect.
where x is the number of participants meeting the response definition and n is the relevant analysis population.
The appropriate denominator is important. A response rate can change depending on whether the analysis uses all enrolled participants, all treated participants, or another prespecified evaluable population. The ClinicalTrials.gov record does not provide the detailed analysis-population definitions needed to reconstruct such a denominator beyond the enrollment and safety denominators explicitly reported.
Confidence intervals for a proportion
A confidence interval would complement the response estimate by describing its statistical precision. For a binary endpoint, exact binomial intervals or other prespecified methods can be used. The ClinicalTrials.gov record does not specify the confidence-interval method, so one should not assume a particular interval construction for this trial.
RECIST 1.1 response classification
The primary Part A endpoint is based on confirmed best overall response according to RECIST 1.1, with adjudication by an Independent Endpoint Review Committee. This means the statistical analysis begins with a clinically defined categorical endpoint before any proportion is calculated.
Durability as an additional response condition
Part B adds duration to the response definition. An objective response is not sufficient by itself: the registered definition requires that the response last at least 6 months. This makes follow-up and event or censoring rules especially important to interpretation.
Time-to-event concepts
The ClinicalTrials.gov record classifies the Part B endpoint as time-to-event. In general, time-to-event analyses are designed to account for differing follow-up times and right censoring. Kaplan-Meier estimation is a common approach when the estimand is a time until an event, while response-duration thresholds can also be summarized through prespecified responder definitions. The ClinicalTrials.gov record does not identify a specific statistical model or estimator for DRR.
Here, S(t) represents the probability that an event time T exceeds time t. This concept is useful when response duration is analyzed explicitly as a time-to-event variable.
Independent endpoint review
Both registered primary endpoints identify an Independent Endpoint Review Committee as the mechanism for determining response. Independent review can reduce the influence of knowledge about treatment administration or local investigator assessment, although the registry describes the overall study as unmasked.
9. Statistical Methods Explained
Why is a single-group study different from a randomized trial?
A randomized trial estimates a treatment contrast by comparing outcomes between groups created through random assignment. A single-group study does not contain that internal randomized comparison. Its response estimate therefore describes the observed experience under the intervention rather than an internally randomized causal contrast.
What does confirmed best overall response mean?
Best overall response summarizes the best response category recorded for a participant under the registered RECIST 1.1 framework. The registry definition includes complete response, partial response, stable disease, and progressive disease, with confirmation and adjudication by the Independent Endpoint Review Committee.
Why does durable response require a 6-month threshold?
The 6-month condition separates a response that persists for a prespecified period from an objective response that may be brief. Statistically, this creates a composite condition: the participant must first achieve an objective response and then maintain that response long enough to satisfy the duration criterion.
Why does follow-up matter for durable response?
A participant cannot be confirmed to have a response lasting at least 6 months without sufficient observation. This makes censoring and follow-up particularly relevant. The ClinicalTrials.gov record does not provide the complete censoring algorithm, so no specific handling rule should be attributed to the study beyond the registered endpoint definition.
Why is an Independent Endpoint Review Committee important?
Response classification can involve radiologic interpretation. An independent review committee provides a separate adjudication process for determining the registered response endpoint. This is especially relevant when the endpoint depends on categorical interpretation of imaging assessments.
Why should a response rate not be interpreted as a comparative treatment effect?
Because the study is registered as non-randomized and single-group, an observed response proportion does not isolate the effect of avelumab from differences in patient characteristics, disease course, assessment timing, or other sources of variation that a randomized comparator would help balance.
10. Analysis Population and Denominator Considerations
One of the most important statistical issues for a single-group response endpoint is identifying exactly who belongs in the denominator. The ClinicalTrials.gov record provides overall enrollment of 204 and safety denominators of 88 for Part A and 116 for Part B.
| Quantity | Reported value | Statistical relevance |
|---|---|---|
| Total enrollment | 204 | Overall registered study enrollment |
| Part A safety denominator | 88 | Number at risk for the reported Part A serious-adverse-event count |
| Part B safety denominator | 116 | Number at risk for the reported Part B serious-adverse-event count |
| Formal efficacy-analysis population | Not provided in the ClinicalTrials.gov record | Should not be inferred from the safety denominator alone |
The 88 and 116 denominators are explicitly posted on ClinicalTrials.gov for serious adverse events. They should not automatically be substituted for an efficacy-analysis denominator unless the trial's formal analysis population establishes that equivalence.
11. Safety Results
The ClinicalTrials.gov record reports serious adverse events by study part. These are safety results rather than efficacy outcomes and should be interpreted separately from the primary response endpoints.
| Study part | Intervention | Serious adverse events | Participants at risk |
|---|---|---|---|
| Part A | Avelumab | 48 | 88 |
| Part B | Avelumab | 58 | 116 |
Part A
Serious adverse events affected 48 of 88 participants in the ClinicalTrials.gov record.
Part B
Serious adverse events affected 58 of 116 participants in the ClinicalTrials.gov record.
The ClinicalTrials.gov record does not provide a randomized comparator arm for safety, nor do they provide a formal statistical comparison of serious adverse-event rates. The appropriate presentation is therefore descriptive.
12. What the Registry Does and Does Not Report
| Statistical feature | Information in the ClinicalTrials.gov record | Interpretation |
|---|---|---|
| Randomization | Non-randomized | No randomized treatment comparison |
| Control group | None identified | No internal comparator |
| Masking | None | Study registered as unmasked |
| Primary endpoint results | Results posted = Yes | Formal numerical analyses were not included in the ClinicalTrials.gov record |
| Confidence intervals | Not provided in the ClinicalTrials.gov record | No CI should be reconstructed from absent values |
| P-values | Not provided in the ClinicalTrials.gov record | No hypothesis-test result should be inferred |
| Subgroup analyses | Not provided | No subgroup treatment effects should be reported |
| Baseline characteristics | Not provided | No baseline comparison should be constructed |
| Formal survival model | Not provided | No Cox or other model should be attributed to the registry record |
| Interim analysis | Not provided | No interim alpha-spending framework should be inferred |
| Multiplicity procedure | Not provided | No multiplicity adjustment should be assumed |
| Missing-data method | Not provided | No imputation strategy should be inferred |
| Bayesian methods | Not provided | No Bayesian analysis should be attributed to the trial |
13. Multiplicity, Interim Analysis, and Other Design Topics
Several statistical-design features commonly discussed in clinical-trial analysis are not identified in the ClinicalTrials.gov record. This absence is important: a rigorous analysis should distinguish a documented design feature from a method that would merely be plausible for this type of study.
Multiplicity
Two primary endpoints are registered, but the ClinicalTrials.gov record does not state how type I error was controlled across them.
Interim analysis
No interim-analysis procedure or alpha-spending rule is provided.
Missing data
No imputation method or missing-data estimand is provided.
Bayesian analysis
No Bayesian model, prior distribution, posterior estimate, or Bayesian decision rule is provided.
Similarly, no non-inferiority margin, crossover strategy, factorial structure, or treatment-effect stratification procedure is identified in the ClinicalTrials.gov record. These features should not be added merely because they occur in other clinical-trial designs.
14. How a Formal Statistical Analysis Would Typically Be Structured
The absence of the ClinicalTrials.gov record does not prevent us from explaining the statistical logic that would normally be applied to endpoints of these types. The key is to label these methods as typical analytical approaches, not as undocumented methods of JAVELIN Merkel 200.
| Endpoint / question | Typical analytical approach | What it answers |
|---|---|---|
| Confirmed BOR | Single-group response proportion with an appropriate binomial confidence interval | What fraction of participants met the response definition? |
| Confirmed BOR versus a benchmark | One-sample binomial hypothesis test, if a benchmark was prespecified | Is the observed response rate statistically distinguishable from a specified reference rate? |
| Durable response rate | Proportion satisfying objective-response and duration criteria, with an appropriate confidence interval | What fraction achieved a response lasting at least 6 months? |
| Response duration | Time-to-event methods such as Kaplan-Meier estimation when defined and prespecified | How long do responses persist, accounting for censoring? |
| Safety | Descriptive event counts and proportions | How frequently did serious adverse events occur? |
A formal protocol or statistical analysis plan would be needed to determine the exact estimand, analysis population, confidence-interval construction, handling of missing assessments, censoring rules, and any hypothesis-testing framework.
15. Statistical Interpretation of a Single-Group Trial
In a randomized trial, randomization provides a framework for attributing differences between groups to treatment, subject to the usual assumptions and trial conduct. In a single-group study, there is no concurrent randomized control group. Consequently, an observed response rate describes what happened among participants receiving the intervention but does not by itself identify what would have happened to those same participants under another treatment.
If investigators compare a single-group response rate with a historical or prespecified benchmark, the comparison depends on the validity and relevance of that benchmark. Differences in patient selection, disease characteristics, assessment methods, calendar time, supportive care, and endpoint definitions can affect the comparison. A benchmark is therefore not equivalent to randomization.
Even a large observed response proportion is an estimate from a finite sample. A confidence interval would show how precisely the study estimates the underlying response probability under the chosen statistical framework. Precision and causal identification are separate issues: a narrow interval does not create a randomized control group.
16. Understanding the Two Primary Endpoint Types
| Statistical property | Part A BOR | Part B DRR |
|---|---|---|
| Primary measurement | Best recorded response category | Objective response lasting at least 6 months |
| Binary classification | Yes, as reported in the registry | Expressed as a percentage of participants meeting the durable-response definition |
| Duration requirement | No explicit 6-month requirement in the registry-reported definition | At least 6 months |
| Review | Independent Endpoint Review Committee | Independent Endpoint Review Committee |
| Maximum registered time frame | 113 weeks | 161 weeks |
This distinction is statistically useful because it illustrates how endpoint definitions can change the target of inference. BOR is primarily a categorical tumor-response endpoint, whereas durable response incorporates both response status and persistence.
17. Limitations
- No randomized comparator: the registered design is non-randomized and single-group, so the primary efficacy endpoints do not provide an internally randomized treatment contrast.
- No formal statistical analyses reported: although results are marked as posted, the registry-reported statistical-analysis array is empty. Numerical estimates, confidence intervals, and p-values therefore cannot be responsibly reproduced from the ClinicalTrials.gov record.
- Endpoint-specific denominators: the reported safety denominators should not automatically be treated as the efficacy-analysis population.
- Response adjudication: both primary endpoints depend on RECIST 1.1 assessment and Independent Endpoint Review Committee adjudication.
- Durable-response follow-up: the Part B endpoint requires a response lasting at least 6 months, making adequate follow-up and censoring rules important to the analysis.
- Missing statistical-plan details: the ClinicalTrials.gov record does not specify the confidence-interval method, formal hypothesis-testing framework, multiplicity adjustment, interim analysis, missing-data strategy, or Bayesian methods.
- Limited population detail: baseline characteristics and subgroup results are not included in the ClinicalTrials.gov record.
- Safety interpretation: serious adverse-event counts are descriptive and are not accompanied by a randomized control group or formal comparative safety analysis in the ClinicalTrials.gov record.
18. Why This Trial Matters Statistically
JAVELIN Merkel 200 is a useful teaching example because it illustrates an important class of oncology trial in which the primary statistical evidence is built around response and durability rather than a randomized treatment comparison.
| Concept | How it appears in JAVELIN Merkel 200 |
|---|---|
| Single-group design | The registry identifies the design model as single group. |
| Non-randomized allocation | No randomized assignment is registered. |
| Binary endpoint | Part A confirmed best overall response is classified as binary. |
| Response proportion | A natural descriptive estimand for confirmed response in a single-group study. |
| RECIST 1.1 | Both primary endpoints use RECIST 1.1 response definitions. |
| Independent review | An Independent Endpoint Review Committee determines the primary response endpoints. |
| Durability | Part B requires an objective response lasting at least 6 months. |
| Time-to-event concepts | The registry classifies DRR as a time-to-event endpoint. |
| Confidence intervals | Would be important for communicating precision around single-group response estimates, although none are reported in the ClinicalTrials.gov record. |
| Safety analysis | Serious adverse events are reported descriptively by study part. |
| Analysis-population discipline | Safety denominators should not automatically be substituted for efficacy denominators. |
| Evidence hierarchy | Observed response in a single-group trial should be distinguished from a randomized comparative treatment effect. |
19. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registered primary endpoints quantify confirmed response and durable response within a non-randomized, single-group study. The ClinicalTrials.gov record does not provide the formal numerical analyses needed to quantify their statistical uncertainty or test them against a comparator.
Clinical interpretation
The study was designed to evaluate avelumab in participants with Merkel cell carcinoma, with response assessed using RECIST 1.1 and independent endpoint review. Any clinical interpretation of magnitude or comparative benefit requires the corresponding reported numerical results and their analysis populations.
20. A Practical Reading Guide for the Registry Results
When reviewing the full ClinicalTrials.gov record or associated publications, several questions should be answered before interpreting the response findings.
| Question | Why it matters |
|---|---|
| What was the efficacy analysis population? | The response denominator determines the meaning of the reported percentage. |
| How was confirmation defined operationally? | Confirmed response can depend on repeat assessments and prespecified confirmation rules. |
| How were missing assessments handled? | Missing tumor assessments can affect response classification. |
| How were participants with insufficient follow-up handled for DRR? | A 6-month durability criterion requires adequate observation or prespecified censoring rules. |
| Was a benchmark response rate specified? | A one-sample hypothesis test requires a prespecified reference value if comparative inference is intended. |
| What confidence-interval method was used? | Different binomial interval methods can produce different precision statements. |
| Were multiple primary endpoints subject to a formal type I error strategy? | Multiple confirmatory questions can affect interpretation of p-values. |
21. Longitudinal Trial History
Trial start
The registered study start date was July 3, 2014.
Primary completion
The registered primary completion date was May 2, 2019.
Final registry status
the ClinicalTrials.gov record identifies the study as completed.
22. Sources and Registry Record
- ClinicalTrials.gov: NCT02155647 — JAVELIN Merkel 200.
- PubMed record: PMID 27592805.
- PubMed record: PMID 29347993.
- PubMed record: PMID 29566106.
- PubMed record: PMID 32414862.
- PubMed record: PMID 32472503.
The numerical trial information presented on this page is restricted to the ClinicalTrials.gov record.
23. Related Tutorials
Learn more about the statistical methods used to understand this trial:
24. Related Statistical Calculators
Continue through Clinical Biostats
Explore the statistical concepts behind clinical-trial endpoints, survival analysis, response rates, and quantitative clinical-trial methods.
25. Record Summary
JAVELIN Merkel 200 provides a useful statistical example of a phase 2, non-randomized, single-group oncology study. Its two registered primary endpoints ask related but distinct questions: Part A evaluates confirmed best overall response according to RECIST 1.1, while Part B evaluates durable response requiring an objective response lasting at least 6 months. Both endpoints are adjudicated by an Independent Endpoint Review Committee.
The principal statistical lesson is the distinction between describing an observed response experience and estimating a randomized treatment effect. Because the registry-reported design is single-group and non-randomized, response proportions can describe the study population, but they do not by themselves establish a causal comparison with another treatment. Confidence intervals, analysis-population definitions, follow-up rules, and any prespecified benchmark are therefore essential to a complete interpretation.