Study of Pembrolizumab in Subjects With Advanced Gastric or Gastroesophageal Junction Adenocarcinoma Who Progressed After First-Line Therapy With Platinum and Fluoropyrimidine: Integration of Molecular Subtypes Through Integrative Genomic Analysis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Pembrolizumab in Subjects With Advanced Gastric or Gastroesophageal Junction Adenocarcinoma Who Progressed After First-Line Therapy With Platinum and Fluoropyrimidine: Integration of Molecular Subtypes Through Integrative Genomic Analysis (NCT02589496), a na phase 2 study. The primary outcome is RR.

2. Clinical Question

The central statistical question is therefore whether the study documents response in this treated group, and with what precision. A response-rate estimate describes the frequency of observed responses in the study population; it does not by itself establish how outcomes would compare with a different treatment. Because the design has no control arm, any observed response could not be contrasted within this study against a concurrent untreated or alternative-treatment group. The record's response question is specified for up to 2 years, but it does not provide the details needed to determine the timing, completeness, or analysis population for a primary response estimate.

3. Trial Design

This is a phase II, single-group study sponsored by Samsung Medical Center. The registry lists an enrollment of 80 and a completed status. The study began on 2016-03-26, and primary completion is recorded as 2021-12. The record describes it as single-center and open-label, with no masking and no randomized allocation.

The brief summary describes an initial cohort of approximately 60 subjects and an expansion cohort of 20 additional patients with MSI-high gastric cancer. It says that the expansion followed response analysis in the original cohort. The arm description identifies Cohort A as gastric cancer patients and Cohort B as MSI-high gastric cancer patients; both received pembrolizumab 200 mg every 3 weeks. The registry enrollment figure and the cohort descriptions are reported as separate details; further participant-flow information is not reported in the registry record.

Enrollment was described as proceeding without regard to PD-L1 expression status, although an evaluable specimen for PD-L1 status had to be available and confirmed before enrollment. The summary says participants were evaluated every 6 weeks, with a stated allowance of +/- 7 days, for the first six months after the first drug administration, and every 12 weeks, also with an allowance of +/- 7 days, thereafter until radiologic progression was documented by computed tomography or magnetic resonance imaging.

The trial has no concurrent control arm. Its open-label, single-group structure means the recorded response and safety counts describe people who received pembrolizumab; they do not provide a randomized treatment contrast. Details of the analysis plan, including how response evaluations were handled in the primary analysis, 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

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.

6. Primary Endpoint

The registry lists the primary outcome as RR, measured over up to 2 years.

The primary endpoint is response rate, abbreviated RR in the registry record, assessed per mRECIST in people with advanced gastric or gastroesophageal junction adenocarcinoma who had progressed after one previous line of therapy. The stated timeframe is up to 2 years. The record describes the outcome as a count of participants, but it does not provide the response count or the corresponding response-rate estimate in the detailed outcome information available for this analysis.

In general, response rate summarizes how many people in a defined analysis population meet prespecified criteria for tumor response. It is useful in a phase II study because it provides a direct measure of observed tumor response rather than relying only on a time-to-event outcome. Interpreting it requires knowing which participants were included, how response was defined and confirmed, how missing or unevaluable assessments were handled, and the precision around the estimate. Those details and an uncertainty interval are not reported in the registry record.

The stated objective says pembrolizumab increases RR, but the study has no comparison group. The endpoint can describe response among treated participants if reported; without a comparator, it cannot by itself quantify a treatment difference. The registry's brief summary separately reports a response narrative for five MSI-high patients in the original cohort, but that does not substitute for a complete primary endpoint result and its analysis details.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical design is a single-group study. In general, this means the outcome is observed in a treated group without a concurrent control group assigned to a different treatment. Such a design can describe outcomes in the enrolled participants, but it cannot separate the effect of treatment from other explanations using a randomized comparison. The registry lists no randomized allocation and no masking, and describes the study as open-label.

What does a response-rate count represent?

In general, a binary response endpoint classifies each participant according to whether the specified response criterion was met. The response count is the number classified as responders in the analysis population. A response rate relates that count to the number of participants included in the analysis. The registry record gives the primary endpoint as a count of participants and identifies mRECIST, but does not report a primary response count, rate, denominator, or confidence interval in the detailed outcome information available for this analysis.

A confidence interval, in general, describes a range of values compatible with the observed data under the statistical method used. It communicates uncertainty around an estimate; a wider interval indicates less precision than a narrower one. No confidence interval for the primary endpoint is reported in the registry record, so its precision cannot be evaluated from the available outcome details. A p-value would address compatibility with a specified statistical null model, but no primary p-value or analysis result is reported either.

Why does the absence of a control group matter?

In general, a controlled randomized comparison is designed to estimate a treatment contrast while reducing systematic differences between groups. Here, every described cohort received pembrolizumab. A response count, if fully reported, would describe the observed response experience in those participants; it would not estimate how those same participants would have fared under a different treatment. The stated objective uses the language of an increase in response rate, but the registry design does not include a concurrent comparator for estimating such an increase.

The open-label design also means treatment and outcome assessment were not masked according to the record. The record says response was assessed per mRECIST but does not report who assessed scans, whether assessments were independently reviewed, or how disagreements were resolved. Those procedures can matter when a binary endpoint depends on interpreting imaging assessments.

How should the cohort expansion be understood?

The summary says five MSI-high patients in the original cohort all demonstrated response, and describes that as a 100% response rate. It then describes enrolling 20 more MSI-high gastric cancer patients in Cohort B. In general, when a cohort is expanded after observing an early response pattern, interpretation depends on whether the expansion and its analysis were prespecified, how outcomes are pooled or separated, and how uncertainty is handled. Those statistical-plan details are not reported in the registry record. The subgroup statement is descriptive and does not establish a comparison with another group.

The study summary describes imaging evaluations every 6 weeks, with an allowance of +/- 7 days, during the first six months, and every 12 weeks, with an allowance of +/- 7 days, thereafter until radiologic progression. In general, assessment timing can affect whether and when a response is observed. The record does not describe the rules for confirmation, handling missed scans, or determining response status for participants without an evaluable assessment.

What do the planning assumptions mean?

The record describes planning illustrations based on an observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; it says the effect to detect is a stated fraction of the observed effect. It also distinguishes event totals for time-to-event calculations from participants per arm for other calculations. These are planning assumptions, not observed trial outcomes. The detailed outcome information does not include the observed primary effect or numerical planning estimates, and the actual trial is single-group rather than a two-arm comparison. The registry record does not report that these planning illustrations were the trial's own design justification, and the sample-size justification is not reported in the registry record.

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")
n      <- c(61)
events <- c(15)
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  61  15  24.6  14.5  37.3  15.5  36.7  15.5  36.7

11. Multiplicity, Missing Data and Analysis Populations

The registry record does not report a statistical plan for multiplicity, including whether any adjustment was intended across endpoints, cohorts, or subgroup analyses. The primary outcome is response rate per mRECIST, while the summary also discusses MSI-high patients and the expansion cohort. Because the relationship between these analyses and any formal testing strategy is not described, the subgroup narrative should be interpreted as descriptive rather than as a confirmed multiplicity-controlled finding.

Missing-data methods are not reported in the registry record. The summary requires an evaluable specimen for PD-L1 status before enrollment and describes repeated imaging follow-up, but it does not explain how missing specimens, unevaluable scans, missed visits, early discontinuation, or incomplete follow-up were handled in the response analysis. It also does not state whether participants without an evaluable response assessment were counted as nonresponders or excluded from a particular analysis.

The primary analysis population is not reported in the registry record. The record does not specify whether the response endpoint was analyzed among all enrolled participants, those who received treatment, or another defined population. It gives an enrollment of 80, but the response denominator for the primary endpoint is not available in the detailed outcome information. The absence of a reported analysis population and missing-data approach limits assessment of how robust any response estimate would be to exclusions or incomplete assessments.

No interim-analysis plan, stopping rule, or other formal monitoring approach is reported in the registry record. The summary's account that the expansion followed response analysis does not supply those statistical details. Without a specified plan, it is not possible to determine from the record how the timing of that decision related to formal hypothesis testing or uncertainty assessment.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry safety figures list 61 participants at risk for serious adverse events, with 0 affected. For other adverse events, 61 participants are listed at risk and 25 affected. The record also lists 25 deaths. These are counts, not estimates of comparative risk; the study has no control group.

The record does not report the timing of these events, event severity details for the other adverse events, whether events were considered related to pembrolizumab, or the causes and timing of deaths. It also does not explain how the safety population was defined or whether the observation period was the same for everyone. Therefore, the counts should not be interpreted as establishing whether treatment caused or did not cause an event, or as directly comparable with findings from another study with different participants or follow-up.

The absence of participants recorded as having serious adverse events in this entry does not establish that serious events are impossible or that all relevant safety information is captured. It is a registry count with limited accompanying context. The figures describe what is recorded, while further interpretation would require event definitions, exposure and follow-up details, and attribution information that are not reported in the registry record.

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

The study addresses a clinically focused response question in advanced gastric or gastroesophageal junction adenocarcinoma after prior platinum- and fluoropyrimidine-containing therapy. It also describes an expansion limited to MSI-high gastric cancer. The registry record identifies pembrolizumab as the treatment and gives the intended response endpoint as RR per mRECIST.

The brief summary reports that all 5 MSI-high patients in the original cohort demonstrated response, and labels this a 100% response rate. This statement is a descriptive subgroup finding in the summary. The detailed outcome information does not give a complete primary response result, its denominator, a confidence interval, or analysis methods. The finding therefore cannot be read as a precise estimate for the full study population or as a comparison against another treatment.

Because this is a single-group, open-label study, the record does not establish a causal treatment difference. In general, an observed outcome in a treated group may reflect treatment and other factors, and a concurrent randomized comparison is not available to distinguish them within this study. The result also cannot establish how outcomes compare across patient groups defined by molecular features unless an appropriately described analysis supports that conclusion.

The safety entry reports the recorded counts, but it does not establish whether events were caused by treatment. The record does not include the contextual details needed to interpret the deaths, event timing, or attribution. These limitations do not erase the descriptive information; they define what can and cannot be concluded from it.

In practical statistical terms, the record is informative about the design and the questions researchers intended to study, but incomplete for judging the size and precision of the primary endpoint result. No treatment recommendation follows from these registry entries alone.

14. Why This Trial Matters Statistically

This trial is statistically notable because its intended response question is paired with a single-group design. Such studies can document outcomes in a selected treated population and can be useful in an early phase of evaluation. Their central limitation is that an observed response rate has no concurrent control rate for comparison. The design therefore does not estimate a randomized treatment contrast.

The record also describes a shift from a broadly enrolled original cohort to a cohort focused on MSI-high gastric cancer after a response observation in five patients. Statistically, this makes clarity about prespecification, analysis populations, and uncertainty especially important. The registry record does not report these details. The summary's subgroup statement is consequently best read as an observation motivating further study, not as a fully characterized inferential result.

Another important issue is the distinction between a stated endpoint and a reported result. The registry identifies response rate per mRECIST as primary, but the detailed outcome information available for this analysis does not give a response count or precision estimate. Without them, readers cannot evaluate the primary result's magnitude or uncertainty. This is a reporting limitation, not evidence that the endpoint was or was not met.

The safety counts illustrate a similar point: counts alone do not describe attribution, timing, or comparative risk. Clear denominators, follow-up, event definitions, and analysis methods are needed to place both efficacy and safety observations in context. The study's statistical importance lies partly in showing why design and reporting details matter when interpreting an uncontrolled phase II response study.

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 primary response-rate result is not reported in the detailed outcome information available for this analysis. The registry record does not provide a primary response count, estimate, confidence interval, or test result. That prevents assessment of the endpoint's magnitude and precision.

The single-group, open-label design offers no randomized comparator. Accordingly, the record cannot establish a treatment contrast, and it is not a like-for-like comparison with a control arm. The registry does not report details of independent imaging review, response confirmation, or how assessment disagreements were resolved.

The summary describes an initial cohort of approximately 60 subjects and an expansion cohort of 20 MSI-high gastric cancer patients, while the registry lists enrollment of 80. Participant flow and the relationship between these descriptions are not reported in the registry record. The summary's account of the original-cohort MSI-high response observation does not include response assessment details or an uncertainty interval. The statistical plan for the expansion and any pooling of cohorts are also not reported.

Missing-data methods, the primary analysis population, and handling of unevaluable assessments are not reported in the registry record. The safety figures lack event timing, attribution, and causes of death. Sample-size justification and any interim-analysis or multiplicity plan are not reported. Together, these omissions limit independent assessment of the completeness, precision, and inferential meaning of the reported findings.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Study of Pembrolizumab in Subjects With Advanced Gastric or Gastroesophageal Junction Adenocarcinoma Who Progressed After First-Line Therapy With Platinum and Fluoropyrimidine: Integration of Molecular Subtypes Through Integrative Genomic Analysis — NCT02589496.
• Registry record last updated 2023-05-09; status last verified 2022-07.
• Results first posted 2023-05-09.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The registry describes a completed phase II, single-arm, single-center, open-label study sponsored by Samsung Medical Center. It lists enrollment of 80 and pembrolizumab 200 mg every 3 weeks. The population was advanced gastric or gastroesophageal junction adenocarcinoma after one previous line of platinum- and fluoropyrimidine-containing therapy; the expansion cohort was described as MSI-high gastric cancer.

The primary outcome was response rate per mRECIST, with a timeframe of up to 2 years. The detailed outcome information available for this analysis does not report a primary response estimate or statistical analysis. The brief summary reports a response observation in five MSI-high patients in the original cohort and describes the subsequent expansion. Safety counts include 0 serious adverse events among 61 at risk, 25 participants with other adverse events among 61 at risk, and 25 deaths. Attribution and further event context are not reported in the registry record.