This page separates reported trial information from statistical interpretation. The ClinicalTrials.gov record identifies three registered primary objective-response endpoints and report that results are posted, but they do not provide formal statistical-analysis estimates, confidence intervals, or p-values for those endpoints.
1. Trial at a Glance
KEYNOTE-052 was a phase 2, single-group, unmasked treatment study evaluating pembrolizumab in participants with advanced urothelial cancer. The ClinicalTrials.gov record reports 374 enrolled participants, one arm, three registered primary endpoints, and posted outcome measures.
| Feature | KEYNOTE-052 |
|---|---|
| Trial name | KEYNOTE-052 |
| Phase | Phase 2 |
| Status | COMPLETED |
| Condition | Urothelial Cancer |
| Design | Single-group |
| Allocation | NA |
| Masking | NONE |
| Primary purpose | TREATMENT |
| Enrollment | 374 |
| Arms | 1 |
| Intervention | Pembrolizumab (biological) |
| Lead sponsor | Merck Sharp & Dohme LLC |
| Sponsor type | INDUSTRY |
| ClinicalTrials.gov | NCT02335424 |
2. Clinical Question
The registry describes a treatment-focused phase 2 study of pembrolizumab in participants with advanced urothelial cancer. Because the study has a single-group design, the registered question is not a randomized comparison between pembrolizumab and a concurrent control arm. The primary efficacy framework instead evaluates objective response rate in the enrolled study population and in two PD-L1-defined populations.
Population
Participants with advanced urothelial cancer enrolled in the study.
Intervention
Pembrolizumab, listed in the registry as a biological intervention.
Comparator
No concurrent comparator arm is specified. The allocation is listed as NA and the design model is SINGLE_GROUP.
Primary question
What objective response rate is observed according to RECIST 1.1 in all participants and in participants with PD-L1 CPS of ≥1% and ≥10%?
3. Trial Design
Pembrolizumab
- Pembrolizumab
- Biological intervention
- Single study arm
- No concurrent comparator arm reported in the ClinicalTrials.gov record
No randomized treatment contrast
- No between-arm hazard ratio can be calculated from the registry-reported design
- No randomized control-group response rate is available in the ClinicalTrials.gov record
- ORR is naturally summarized as a proportion within the treated study population
- External historical comparisons would not be randomized comparisons
4. Endpoints
The registry identifies three primary endpoints. All three are binary objective-response endpoints evaluated through RECIST 1.1. The endpoint time frame is the same in the ClinicalTrials.gov record: up to approximately 80.5 months.
| Primary endpoint | Time frame | Endpoint type | Registry definition |
|---|---|---|---|
| Objective Response Rate (ORR) Per RECIST 1.1 in All Participants | Up to approximately 80.5 months | Binary | ORR was determined in all participants and was defined as the percentage of participants who had a confirmed Complete Response (CR: Disappearance of all target lesions) or a Partial Response (PR: At least a 30% decrease in the sum of diameters of target lesions) per RECIST 1.1 as assessed by Blinded Independent Central Review (BICR). Participants with missing data were considered non-responders. |
| Objective Response Rate (ORR) Per RECIST 1.1 in Participants With PD-L1 CPS of ≥1% | Up to approximately 80.5 months | Binary | ORR was determined in participants who had a PD-L1 CPS of ≥1% as measured by immunohistochemistry assay. ORR was defined as the percentage of participants who had a confirmed Complete Response (CR: Disappearance of all target lesions) or a Partial Response (PR: At least a 30% decrease in the sum of diameters of target lesions) per RECIST 1.1 as assessed by BICR. |
| Objective Response Rate (ORR) Per RECIST 1.1 in Participants With PD-L1 CPS of ≥10% | Up to approximately 80.5 months | Binary | ORR was determined in participants who had a PD-L1 CPS of ≥10% as measured by immunohistochemistry assay. ORR was defined as the percentage of participants who had a confirmed Complete Response (CR: Disappearance of all target lesions) or a Partial Response (PR: At least a 30% decrease in the sum of diameters of target lesions) per RECIST 1.1 as assessed by BICR. |
5. Primary Endpoint Populations
The three primary endpoints form a nested set of efficacy questions: one considers all participants, while the other two restrict the population according to PD-L1 combined positive score. These are distinct endpoint populations rather than randomized treatment groups.
| Population | Endpoint | PD-L1 assessment | Statistical role |
|---|---|---|---|
| All participants | ORR per RECIST 1.1 | Not restricted by CPS in the endpoint definition | Overall response proportion |
| Participants with PD-L1 CPS ≥1% | ORR per RECIST 1.1 | Measured by immunohistochemistry assay | Response proportion in the CPS-defined population |
| Participants with PD-L1 CPS ≥10% | ORR per RECIST 1.1 | Measured by immunohistochemistry assay | Response proportion in the higher CPS-defined population |
Because the ClinicalTrials.gov record does not provide the number of participants meeting the PD-L1 CPS thresholds, those denominators cannot be reconstructed here. The page therefore does not infer response rates or subgroup sizes from the overall enrollment of 374.
6. Results
The ClinicalTrials.gov record states that results have been posted and that 30 outcome measures are posted. However, no formal statistical analyses were posted, and no numerical ORR estimates, confidence intervals, or p-values are included for the three primary endpoints.
| Primary endpoint | Results status | Formal statistical analysis reported? |
|---|---|---|
| ORR in all participants | Results posted: Yes | No |
| ORR in participants with PD-L1 CPS ≥1% | Results posted: Yes | No |
| ORR in participants with PD-L1 CPS ≥10% | Results posted: Yes | No |
A binary ORR endpoint is fundamentally a proportion: the number of participants with a confirmed CR or PR divided by the relevant analysis population. The registry definition also specifies that participants with missing data were considered non-responders for the all-participant endpoint.
Because no numerical ORR estimate is reported here, there is no effect estimate to interpret and no confidence interval or p-value to reproduce. The absence of a formal analysis in the ClinicalTrials.gov record should not be converted into a claim that the endpoint was positive or negative.
How the ORR would normally be summarized
For a single-group binary endpoint, the primary descriptive quantity would ordinarily be the observed response proportion:
A confidence interval can then quantify statistical uncertainty around the observed proportion. The exact interval method should follow the prespecified statistical analysis plan or registry methodology when available.
That calculation is conceptually different from a randomized treatment comparison. In a single-group study, there is no concurrent randomized control rate against which the observed ORR can be directly contrasted.
7. Statistical Methodology
Binary endpoint analysis
Each registered primary endpoint is classified as Binary. The outcome is therefore determined at the participant level: a participant either meets the confirmed CR/PR response definition or does not. The resulting ORR is a proportion rather than a time-to-event measure.
x is the number of participants classified as responders and n is the number in the endpoint's defined analysis population.
RECIST 1.1 classification
The registry defines response according to RECIST 1.1 and requires a confirmed Complete Response or Partial Response. This creates a categorical endpoint from radiologic assessments rather than treating the raw change in lesion diameter as the final primary outcome.
Blinded Independent Central Review
The primary endpoint definition specifies assessment by Blinded Independent Central Review (BICR). Participants with missing data were considered non-responders. Per protocol, only data generated during the initial course of treatment contributed to the analysis.. Central review provides a standardized assessment framework intended to reduce dependence on local investigator assessment for the response endpoint.
Missing data
For the all-participant ORR endpoint, the registry-reported definition explicitly states that participants with missing data were considered non-responders. Statistically, this convention affects the denominator and numerator of the observed response proportion and can therefore affect the resulting estimate.
PD-L1 CPS-defined populations
Two primary endpoints restrict the analysis population according to PD-L1 combined positive score: CPS ≥1% and CPS ≥10%. The registry states that CPS was measured by immunohistochemistry assay. These are population definitions for separate ORR endpoints, not separate treatment arms.
8. Statistical Methods Explained
Why is ORR a binary endpoint?
RECIST 1.1 converts tumor assessment into clinically defined response categories. For this trial's primary endpoint, a participant is counted as a responder when a confirmed CR or PR is documented. Everyone else is not counted as a responder under the registry-reported definition, including participants with missing data for the all-participant endpoint.
Why does a single-group trial not produce a treatment hazard ratio?
A hazard ratio compares event hazards between two groups over time. KEYNOTE-052 is registered as a single-group study with one arm, so the registry-reported design does not provide a concurrent randomized comparator from which a treatment-versus-control hazard ratio could be estimated.
What does BICR contribute to an ORR endpoint?
BICR means the endpoint assessment is based on blinded independent central review rather than simply accepting the treating site's assessment. For a response endpoint, this can provide a more standardized basis for deciding whether RECIST 1.1 criteria for CR or PR have been met.
Why are there three primary ORR endpoints?
The endpoints ask related but distinct questions about response: ORR in all participants, ORR among participants with PD-L1 CPS of ≥1%, and ORR among participants with PD-L1 CPS of ≥10%. The latter two examine response within prespecified biomarker-defined populations.
What does treating missing participants as non-responders do?
It makes the missing-data rule explicit for the all-participant ORR endpoint. A participant with missing response information does not contribute a response to the numerator under this rule. This is important because the treatment of missing observations can materially affect a binary endpoint.
Why should an observed ORR not be interpreted as a causal treatment effect?
In a single-group design, the observed response proportion is not a randomized contrast. Without a concurrent randomized comparator, the study does not separate the observed response proportion from what might have occurred under an alternative treatment or under other sources of outcome variation.
9. Planned Analysis
| Endpoint | Typical analysis | Interpretive target |
|---|---|---|
| ORR in all participants | Estimate the proportion with confirmed CR or PR; calculate an appropriate confidence interval according to the prespecified statistical method. | Magnitude and precision of observed response in the full study population. |
| ORR in PD-L1 CPS ≥1% | Estimate the proportion with confirmed CR or PR among participants meeting the CPS threshold; calculate an appropriate confidence interval according to the prespecified method. | Magnitude and precision of observed response within the CPS-defined population. |
| ORR in PD-L1 CPS ≥10% | Estimate the proportion with confirmed CR or PR among participants meeting the CPS threshold; calculate an appropriate confidence interval according to the prespecified method. | Magnitude and precision of observed response within the higher CPS-defined population. |
The ClinicalTrials.gov record does not report formal comparisons for these endpoints. In particular, they do not provide a prespecified historical-response benchmark, a null response-rate value, a confidence interval, or a p-value in the posted results. Therefore, this page does not introduce such values.
10. Safety Results
the ClinicalTrials.gov record reports serious adverse events by arm. Because this is a single-group study, the registry data identify one principal pembrolizumab group and a separate "Second Course Pembrolizumab" group in the serious-adverse-event field.
| Group | Serious adverse events affected | At risk |
|---|---|---|
| Pembrolizumab | 190 | 370 |
| Second Course Pembrolizumab | 6 | 13 |
The ClinicalTrials.gov record does not provide additional safety estimates, confidence intervals, comparative tests, or definitions of the individual serious adverse-event categories. Accordingly, the reported counts are presented without deriving percentages or adding unreported safety classifications.
The serious-adverse-event field reports affected and at-risk counts rather than a randomized comparison. The denominator also differs between the two listed groups. The counts therefore should be read as descriptive safety information for the reported groups, not as evidence of a between-group treatment effect.
11. Missing Data and Endpoint Classification
Missing-data handling is unusually important for a binary response endpoint because the classification of an observation can change the numerator and therefore the observed ORR. The registry explicitly specifies that participants with missing data were considered non-responders for the all-participant primary ORR endpoint.
Explicit rule
Missing data were considered non-responders for the all-participant ORR definition reported in the registry.
Why it matters
The rule prevents missing observations from being silently omitted from the response classification for that endpoint.
What is not reported
The ClinicalTrials.gov record does not provide the number of missing observations or the resulting numerical ORR.
Interpretive caution
Without the underlying participant-level outcome data, the magnitude of any missing-data effect cannot be calculated from the ClinicalTrials.gov record.
12. PD-L1 Combined Positive Score and Statistical Interpretation
The registry defines two primary endpoints using PD-L1 combined positive score thresholds of ≥1% and ≥10%. PD-L1 status was measured by immunohistochemistry assay.
| PD-L1 definition | Primary endpoint | Interpretive question |
|---|---|---|
| All participants | ORR per RECIST 1.1 | What proportion of the overall study population had a confirmed CR or PR? |
| CPS ≥1% | ORR per RECIST 1.1 | What proportion of participants meeting the CPS threshold had a confirmed CR or PR? |
| CPS ≥10% | ORR per RECIST 1.1 | What proportion of participants meeting the higher CPS threshold had a confirmed CR or PR? |
These endpoints can describe response within different biomarker-defined populations. However, comparing two observed ORRs across biomarker groups is not automatically a formal test of biomarker treatment interaction. In a single-group study there is also no randomized treatment-by-biomarker interaction because treatment assignment is not randomized against a concurrent control.
13. Registry Timeline
Trial start
the ClinicalTrials.gov record records the start date as 2015-02-24.
Primary completion
the ClinicalTrials.gov record records primary completion on 2018-06-19.
Current registry status in the ClinicalTrials.gov record
The trial status is listed as COMPLETED, with results posted and 30 outcome measures posted.
14. What an ORR Estimate Does — and Does Not — Mean
An ORR estimate is the proportion of participants classified as having a confirmed CR or PR under the prespecified RECIST 1.1 framework. It summarizes the frequency of the defined response outcome in the analyzed population.
It does not by itself measure overall survival, progression-free survival, duration of response, or the magnitude of benefit relative to an alternative treatment.
If a confidence interval is calculated for a single-group ORR, it describes statistical uncertainty around the estimated response proportion under the chosen sampling framework. It does not describe the range of responses that an individual participant could experience.
A hypothesis test for a single-group response proportion requires a null hypothesis, such as a prespecified response-rate benchmark. The ClinicalTrials.gov record does not provide such a benchmark or a formal statistical-analysis entry, so no p-value is reproduced here.
15. Single-Group Design: The Key Statistical Issue
The central statistical feature of KEYNOTE-052 is not a randomized treatment contrast but the estimation of response within a single treated population. This changes how the evidence should be interpreted.
| Feature | Randomized comparative trial | KEYNOTE-052 registry-reported design |
|---|---|---|
| Treatment groups | At least two randomized groups | One study arm |
| Allocation | Random assignment | NA |
| Concurrent comparator | Yes | Not reported in the registry-reported design |
| Primary effect measure | Often a between-group contrast | Observed ORR proportion |
| Causal contrast | Supported by randomization | Not created by the single-group design |
| Historical comparison | Usually unnecessary for the randomized contrast | Would require an externally specified benchmark and additional assumptions |
This distinction is important because a large observed response proportion and a large randomized treatment effect are different statistical statements. The first describes what occurred in the study population. The second requires a valid comparison against what would have occurred under another randomized condition.
16. Multiplicity and the Three Primary Endpoints
The registry identifies three primary endpoints. All three concern ORR, but they use different analysis populations. The ClinicalTrials.gov record does not provide a formal multiplicity strategy or a statistical-analysis entry assigning alpha across these endpoints.
| Endpoint | Population | Multiplicity information reported? |
|---|---|---|
| ORR, all participants | All participants | No formal strategy reported |
| ORR, CPS ≥1% | Participants with PD-L1 CPS ≥1% | No formal strategy reported |
| ORR, CPS ≥10% | Participants with PD-L1 CPS ≥10% | No formal strategy reported |
Multiplicity becomes especially relevant when several primary hypotheses are tested against prespecified benchmarks. However, because the ClinicalTrials.gov record does not report formal hypothesis tests, alpha allocation, or a multiplicity procedure, no such procedure is attributed to KEYNOTE-052 here.
17. What the Registry Data Do Not Establish
- No numerical ORR estimates: the ClinicalTrials.gov record does not contain response percentages for any of the three primary endpoints.
- No confidence intervals: no confidence intervals for the primary ORR endpoints are reported.
- No p-values: no formal statistical analyses were posted.
- No formal comparator: the design is listed as SINGLE_GROUP with one arm.
- No median survival: no survival endpoint estimates are contained in the registry-reported primary-endpoint data.
- No subgroup estimates beyond the registered PD-L1-defined endpoint populations: numerical response estimates for those populations are not reported.
- No formal multiplicity method: the ClinicalTrials.gov record does not specify alpha allocation across the three primary endpoints.
- No Bayesian analysis: no Bayesian statistical method is reported in the ClinicalTrials.gov record.
18. Important Limitations and Interpretation Issues
- Single-group design: without a concurrent randomized comparator, observed ORR cannot be interpreted as a randomized treatment effect.
- Historical comparisons: comparison with an external historical response rate would introduce assumptions about comparability of populations, assessment methods, follow-up, and other study characteristics.
- Missing-data rule: the registry-reported definition classifies missing data as non-response for the all-participant ORR endpoint, making the missing-data mechanism relevant to interpretation.
- Biomarker-defined populations: CPS ≥1% and CPS ≥10% define different analysis populations, and their denominators are not reported here.
- Multiplicity: three endpoints are identified as primary, but the ClinicalTrials.gov record does not describe formal alpha allocation or hypothesis-testing procedures.
- Response assessment: ORR depends on RECIST 1.1 classification and BICR assessment, so the endpoint is not interchangeable with an unreviewed measure of tumor-size change.
- Incomplete statistical-analysis information: results are reported as posted, but the registry-reported statistical-analysis field contains no formal analyses or numerical estimates.
- Safety denominators: serious adverse-event counts are provided for two listed groups with different at-risk counts, so the registry-reported figures should not be treated as a direct randomized comparison.
19. Why This Trial Matters Statistically
KEYNOTE-052 is a useful teaching case because it illustrates a different statistical architecture from a conventional randomized comparative trial. The central task is to interpret a binary response endpoint in a single treated population, while keeping biomarker-defined populations and missing-data rules explicit.
| Concept | How it appears in KEYNOTE-052 |
|---|---|
| Single-group design | One study arm and no randomized comparator in the ClinicalTrials.gov record. |
| Binary endpoint | All three registered primary endpoints are classified as Binary. |
| Objective response rate | Response is defined by confirmed CR or PR under RECIST 1.1. |
| Central review | Primary ORR is assessed by Blinded Independent Central Review. |
| Missing data | Participants with missing data are considered non-responders for the all-participant ORR definition. |
| Biomarker populations | Separate primary ORR endpoints are defined for PD-L1 CPS ≥1% and ≥10%. |
| Confidence intervals | Would quantify uncertainty around a single-group response estimate, although none are reported in the provided analysis data. |
| Hypothesis testing | Would require a prespecified benchmark for a single-group ORR test; none is reported in the ClinicalTrials.gov record. |
| Multiplicity | Three primary endpoints are registered, but no formal multiplicity strategy is reported. |
| Safety analysis | Serious adverse-event affected/at-risk counts are reported for the listed groups. |
20. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary endpoints are binary ORR measures in a single-group study. The registry defines the response classification, assessment method, PD-L1 populations, and missing-data rule, but the ClinicalTrials.gov record does not contain numerical estimates or formal tests.
Clinical interpretation
The registry establishes that pembrolizumab was evaluated in participants with advanced urothelial cancer and that objective response was a central efficacy outcome. The ClinicalTrials.gov record does not provide enough numerical efficacy information to characterize the magnitude or durability of response.
21. A Practical Framework for Reading KEYNOTE-052
A statistically disciplined reading of this study can proceed in five steps:
This workflow helps prevent a common interpretive error: treating a within-study response proportion as if it were a randomized comparison. The distinction becomes even more important when the analysis is subdivided by a biomarker such as PD-L1 CPS.
22. Related Tutorials
Learn more about the methods used in this trial:
23. Related Statistical Calculators
24. Sources
- ClinicalTrials.gov: NCT02335424 — KEYNOTE-052.
- Linked publication: PubMed PMID 28967485.
- Linked publication: PubMed PMID 32552471.
- Linked publication: PubMed PMID 37699333.
- Linked publication: PubMed PMID 36494006.
- Linked publication: PubMed PMID 35247908.
Continue through the Clinical Biostats statistical library
Explore tutorials and calculators covering clinical-trial endpoints, binary outcomes, confidence intervals, sample-size methods, and related statistical workflows.
25. Record Summary
KEYNOTE-052 is a completed phase 2, single-group study of pembrolizumab in participants with advanced urothelial cancer. Its three registered primary endpoints are binary objective response rates assessed according to RECIST 1.1 by Blinded Independent Central Review: ORR in all participants, ORR in participants with PD-L1 CPS of ≥1%, and ORR in participants with PD-L1 CPS of ≥10%. The registry defines confirmed CR and PR explicitly and specifies that missing data are considered non-responders for the all-participant endpoint.
The ClinicalTrials.gov record reports 374 enrolled participants, one arm, 30 posted outcome measures, and serious-adverse-event counts for the listed pembrolizumab groups. Results are marked as posted, but the ClinicalTrials.gov record contains no formal efficacy estimates, confidence intervals, or p-values. The appropriate statistical interpretation is therefore to distinguish the registered endpoint definitions and study design from numerical results that are not present in the ClinicalTrials.gov recordset.