Study of Pembrolizumab (MK-3475) in Participants With Advanced Urothelial Cancer (MK-3475-052/KEYNOTE-052): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Pembrolizumab (MK-3475) in Participants With Advanced Urothelial Cancer (MK-3475-052/KEYNOTE-052) (NCT02335424), a na phase 2 study. The primary outcome is Objective Response Rate (ORR) Per Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST 1.1) in All Participants.

2. Clinical Question

3. Trial Design

This was a Phase II, single-group clinical trial of pembrolizumab in participants with advanced, unresectable or metastatic urothelial cancer who were ineligible for cisplatin-based therapy. The registry lists enrollment of 374, no masking, and allocation as not applicable. With no randomized control group, the design estimates outcomes within the treated group rather than a randomized difference between treatments.

The intervention description specifies pembrolizumab 200 mg by intravenous infusion on Day 1 of each 21-day cycle for up to 2 years. The registry also describes a second course for eligible participants under specified circumstances, but the available description does not give the complete criteria. With Amendment 4, participants who achieved the study objective or continued until the study ended were to discontinue from this study and enroll in an extension study for continued protocol-defined assessments and treatment.

The trial is listed as completed. Its recorded start was 2015-02-24, and its primary completion date was 2018-06-19. The primary outcome timeframe is stated as up to approximately 80.5 months. The primary outcome population comprised participants who received at least one dose during the initial course, and only initial-course data contributed to that analysis.

The design and registry description do not establish the rationale for the enrollment target, the intended statistical operating characteristics, or the full analysis plan. Those details are not reported in the registry record. Because there was no randomized comparison, outcomes cannot be interpreted as a controlled estimate of the difference between pembrolizumab and another treatment.

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: Results are based on a database cutoff date of 18-February-2022.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All participants who received at least one dose of study treatment during the initial course of treatment

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: The baseline analysis population consisted of all allocated participants.

6. Primary Endpoint

The registry lists the primary outcome as Objective Response Rate (ORR) Per Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST 1.1) in All Participants, measured over Up to approximately 80.5 months.

The primary endpoint was objective response rate under RECIST 1.1. It measures the percentage of participants with a confirmed complete response or partial response. A complete response was defined as disappearance of all target lesions; a partial response was defined as at least a 30% decrease in the sum of diameters of target lesions. Response was assessed by blinded independent central review.

ORR is a binary-response summary: each participant is classified according to whether the specified response occurred, and the endpoint is the proportion classified as responders. It gives a compact description of tumor response in the analysis population. It does not, by itself, describe how long a response lasts, whether disease later progresses, or overall survival. The registry lists duration of response, progression-free survival, and overall survival among the secondary outcomes.

The endpoint’s definition matters for interpretation. The analysis counted participants with missing data as non-responders, and only data generated during the initial treatment course contributed. The stated population was all participants who received at least one dose during that course. The registry record gives the endpoint definition and a 95% confidence interval as its dispersion measure, but does not report the ORR estimate or interval limits. Accordingly, the magnitude and precision of the primary result cannot be described from this record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The methods information should be read alongside the design: this was a single-group Phase II study with no masking and no randomized comparator. The primary outcome was a percentage, and the registry specified a 95% confidence interval as its measure of dispersion. The outcome estimate and interval limits are not reported in the registry record summarized for this analysis.

What does a percentage endpoint represent?

For a binary endpoint such as confirmed response, each participant contributes a response classification under the stated criteria. The response percentage describes the share of the defined analysis population classified as having a response. It is a summary of that population, not a prediction that any particular participant will respond. Here, the protocol definition required a confirmed complete or partial response under RECIST 1.1, assessed by blinded independent central review.

The denominator is important. The primary outcome population was all participants who received at least one dose during the initial course of treatment. This is not necessarily identical to every participant enrolled. The registry separately reports enrollment of 374, but does not give the primary outcome result or its denominator in the information summarized for this analysis. The response definition also explicitly treats missing data as non-response, which affects how participants with incomplete endpoint information enter the binary summary.

What does a confidence interval communicate?

In general, a confidence interval expresses statistical uncertainty around an estimate. It is compatible with a range of values under the assumptions of the method used; it does not establish that all values in that range are equally plausible, and it is not an individual prediction interval. A wider interval generally signals less precision than a narrower interval. The registry identifies a 95% confidence interval for ORR but does not report its endpoints, so the precision of this study’s estimate cannot be judged from the available entry.

The record does not describe the exact confidence-interval method, any covariate adjustment, or other model details for the primary endpoint. Those methods are not reported in the registry record. It would not be appropriate to infer them merely from the fact that ORR is a percentage.

Why does the single-group design matter?

A single-group estimate describes outcomes among participants receiving the study intervention. A randomized comparison, by contrast, is designed to compare groups whose treatment assignment is randomized. Randomization helps balance measured and unmeasured factors on average; without it, differences between a treated group and an external benchmark may reflect differences in participant characteristics, follow-up, assessment, or other factors. No external comparison is specified in the primary outcome description.

The endpoint was assessed by blinded independent central review, while the registry lists masking for the study as none. Central review identifies the stated assessment process for the primary endpoint; it does not turn the trial into a masked or randomized comparison. The record does not explain further how reviewer blinding was implemented.

How should planning assumptions be understood?

The planning illustration assumptions describe a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. They state that the effect to detect is a specified fraction of the observed effect. For time-to-event outcomes, the stated planning quantity is the total number of events; for other outcomes, it is participants per arm. These are planning assumptions, not evidence that this trial enrolled randomized arms: the actual trial is listed as single-group, with allocation not applicable.

A planning illustration is not observed power. It does not show the probability that the completed study detected an effect, and its assumptions do not constitute the trial’s own design justification. The registry record does not report the sample-size justification or enough information to reconstruct a trial-specific planning analysis. No planning output should be interpreted as an observed result.

Why can outcome rates vary?

In general, response percentages are affected by the analysis population, endpoint definition, missing data, and the assessment process. A stricter response definition can classify fewer participants as responders than a broader definition; a different missing-data rule can also change the estimate. These are general statistical considerations, not claims about unreported sensitivity analyses in this study. The record specifies the response criteria and missing-data rule but does not report alternative analyses or their results.

For event-count outcomes, over-dispersion means that observed variation is greater than a simple model would expect. In general, accounting for over-dispersion can affect standard errors and uncertainty estimates. No over-dispersion method is described for this trial’s primary binary endpoint, and no such method should be assumed. Similarly, the registry does not report interim analyses, stopping boundaries, or detailed multiplicity procedures.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Pembrolizumab 200 mg")
n      <- c(370)
events <- c(107)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Pembrolizumab 200 mg  370  107  28.9  24.3  33.8  24.5  33.7  24.5  33.7

11. Multiplicity, Missing Data and Analysis Populations

The registry lists the primary ORR outcome and multiple secondary outcomes, including duration of response, progression-free survival, and overall survival. Some of these are specified for PD-L1 CPS subgroups. The record lists 30 outcome measures and no analysis records. The absence of listed analysis records does not establish which analyses were conducted or what statistical procedures were used.

When several endpoints or subgroup questions are examined, the chance of at least one apparently unusual result can increase if each is tested separately. In general, multiplicity procedures can control or describe this issue, but the registry record does not report whether an adjustment was planned or applied. It also does not report a hierarchy for testing the secondary outcomes. No claim about multiplicity-adjusted inference can therefore be made.

The primary endpoint description states that participants with missing data were considered non-responders. This is a defined handling rule for the ORR analysis, not evidence that there were no missing observations. The number of participants with missing endpoint data and the amount or pattern of missingness are not reported in the registry record summarized for this analysis. The record also does not give sensitivity analyses using alternative missing-data assumptions.

The primary analysis population was participants who received at least one dose during the initial course of treatment. The registry does not label this population as an intention-to-treat population, and the analysis should be described using its stated definition rather than an assumed label. Only data from the initial course contributed to the primary outcome. The record does not report the full statistical analysis plan, interim analyses, or how all secondary populations were defined.

12. 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: cerebrovascular accident (15.4%), diarrhoea (7.7%), cellulitis (7.7%), pneumonia (7.7%), metastases to central nervous system (7.7%).

The registry’s adverse-event entries list a pembrolizumab group with 370 participants at risk, 190 participants with serious adverse events, 342 with other adverse events, and 305 deaths. The second-course pembrolizumab group lists 13 participants at risk, 6 with serious adverse events, 12 with other adverse events, and 8 deaths. These are separate registry entries; the second-course group is not a randomized control group, so the figures do not support a like-for-like safety comparison.

The listed serious adverse-event terms include cerebrovascular accident at 15.4%, and diarrhoea, cellulitis, pneumonia, metastases to central nervous system, and acute kidney injury at 7.7% each. The registry entry provides these term percentages but the summary does not specify the denominator or further event-level details. They should not be interpreted as treatment-attributed risks or as a complete description of all serious events.

Counts of participants affected are not the same as counts of events: a participant may experience more than one event. The registry summary does not provide enough detail to determine event recurrence, timing, duration, severity beyond the serious-event classification, or investigator attribution. The deaths listed are not, by themselves, evidence that treatment caused those deaths.

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

The central statistical question was whether and how often participants in this single-group study had a confirmed response under the specified RECIST 1.1 definition. The registry identifies the population, assessment method, missing-data rule, and confidence-interval measure. Those details clarify what the primary endpoint was intended to summarize.

The numerical response estimate and confidence-interval limits are not reported in the registry record summarized for this analysis. Without those values, the size of the response percentage and its precision cannot be described. Outcome names, study completion status, and the length of the stated assessment timeframe do not supply the missing result.

Even if a response estimate were available, a single-group study would describe the observed outcome in the treated study population rather than establish a randomized causal contrast. Such a result would not, on its own, show how outcomes would differ under another treatment or under no treatment. The study’s lack of a randomized comparator is therefore central to interpretation.

ORR is also distinct from duration of response, progression-free survival, and overall survival. The registry lists those as secondary outcomes, but their numerical results are not reported in the record. A response endpoint alone does not answer questions about how long responses last or about later clinical outcomes.

Safety figures likewise require careful reading. They describe affected participants in the registry’s treatment-group entries, but they do not establish that pembrolizumab caused a particular event or death. The record’s summarized safety information does not provide a randomized comparator or all details needed to characterize attribution and exposure.

Thus, the registry supports a precise description of the study’s question and design, but not a claim about the magnitude of ORR, comparative efficacy, or treatment causation. Those conclusions require results and analysis details that are not reported in the registry record.

14. Why This Trial Matters Statistically

This trial is statistically informative as an example of how a response endpoint is defined and analyzed in a single-group study. The primary endpoint was not simply any reported tumor change: it required a confirmed complete or partial response under RECIST 1.1, assessed by blinded independent central review. The registry also specifies how missing endpoint data were classified and which treated participants formed the analysis population.

These design details show why the endpoint definition and denominator matter when interpreting a percentage. An estimate is meaningful only in relation to its response criteria, assessment process, analysis population, and missing-data rule. A confidence interval would add information about uncertainty, but the estimate and interval limits are not reported in the registry record summarized for this analysis.

The study also illustrates the inferential limits of a single-group design. Outcomes in the treated group can be described, but there is no randomized group against which to estimate a treatment contrast. This distinction prevents descriptive response findings from being treated as proof of comparative benefit or causal effect.

Finally, the registry lists a range of secondary outcomes, including time-to-event measures and PD-L1 CPS subgroup outcomes. Naming those outcomes is not equivalent to reporting their estimates or establishing how multiplicity was handled. The record does not supply the details needed to judge the precision, statistical significance, or multiplicity-adjusted interpretation of those analyses. The trial therefore highlights both the value of a clearly defined endpoint and the importance of complete results and analysis reporting.

15. 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 main limitation for interpreting the primary result is that the ORR estimate and 95% confidence-interval limits are not reported in the registry record summarized for this analysis. Without them, neither the observed response percentage nor its precision can be assessed. The registry provides an endpoint timeframe of up to approximately 80.5 months, but that does not fill the results gap.

The study was single-group, with allocation listed as not applicable and masking listed as none. There was no randomized comparator, so the design does not estimate a randomized difference between pembrolizumab and another treatment. The primary endpoint’s central review process is described, but the record does not give further implementation details about reviewer blinding.

The outcome population was participants who received at least one dose during the initial course, and missing observations were classified as non-response. The number of participants with missing outcome data, the amount and pattern of missingness, and sensitivity analyses are not reported in the registry record. The record also does not provide the complete statistical analysis plan, sample-size justification, interim-analysis details, or multiplicity procedures.

Several secondary outcomes and PD-L1 CPS subgroup outcomes are listed, but their numerical results are not reported in the information summarized for this analysis. The safety entries give counts and selected event terms, but do not provide all details needed to assess causality, recurrent events, timing, or exposure-adjusted risk. These limitations constrain interpretation of both efficacy and safety without implying that unreported analyses were or were not performed.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Study of Pembrolizumab (MK-3475) in Participants With Advanced Urothelial Cancer (MK-3475-052/KEYNOTE-052) — NCT02335424.
• Registry record last updated 2023-03-02; status last verified 2023-02.
• Results first posted 2019-07-05.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT02335424 is a completed Phase II, single-group study of pembrolizumab in participants with advanced, unresectable or metastatic urothelial cancer who were ineligible for cisplatin-based therapy. The registry reports enrollment of 374, no masking, and allocation as not applicable. The primary endpoint was ORR under RECIST 1.1, assessed by blinded independent central review among participants who received at least one dose during the initial course. Missing data were classified as non-response.

The registry lists a 95% confidence interval for the primary endpoint but does not report the ORR estimate or interval limits in the record summarized for this analysis. It also lists secondary duration-of-response, progression-free-survival, and overall-survival outcomes, including PD-L1 CPS subgroup outcomes. The single-group design does not provide a randomized treatment comparison, and the available safety entries do not establish event causation.