Perioperative Chemo and Pembrolizumab in Gastric Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Perioperative Chemo and Pembrolizumab in Gastric Cancer (NCT02918162), a na phase 2 study. The primary outcome is Pathologic Complete Response (pCR) Rate.

2. Clinical Question

The study asks whether perioperative chemotherapy combined with pembrolizumab can be associated with pathologic complete response (pCR) in people with locoregional gastric or gastroesophageal (GE) junction adenocarcinoma. The registry describes the treatment as intended to address micrometastatic disease and also describes paired tissue and serum samples collected before and after treatment to explore immune effects and potential predictors of response.

The study is therefore descriptive for the treatment group enrolled. Without a control group, an observed pCR result, even if reported, would not by itself distinguish the effects of the combination from other explanations such as patient selection, disease characteristics, or the contribution of chemotherapy. The registry record also lists overall survival, disease-free survival, and overall response rate as secondary outcomes, but does not report their results.

3. Trial Design

The registry describes a phase II, non-randomized, multi-site, open-label study with a single-group model. Allocation is recorded as not applicable, and masking is none. The trial enrolled 49 participants and is marked completed. Its recorded start date is 2017-11-25, and its primary completion date is 2023-02-28.

All study subjects were assigned the described combination regimen; there is no separate control arm. The arm description says subjects receive standard-of-care chemotherapy for 3 cycles before surgery and 3 cycles following surgery, in combination with pembrolizumab, with an additional pembrolizumab cycle to make 4 total in the pre-operative period. It also states that subjects complete 12 months of maintenance treatment, but the registry wording is truncated at that point. The full maintenance regimen is therefore not reported in the registry record.

The primary endpoint is a participant count: subjects with absence of viable tumor in the surgical resection specimen, as determined by local pathology review. The stated timeframe is up to 34 months. Secondary outcomes are overall survival, disease-free survival, and overall response rate. Although the registry lists four outcome measures, it does not provide posted outcome results or analyses for these outcomes.

Open-label means that masking is not described; it does not mean that outcome assessment is necessarily free from bias. Local pathology review is named for pCR, but central review, blinding of pathologists, and adjudication procedures are not reported in the registry record. The criteria used to define the secondary outcomes and their analysis time origins are also not reported.

The single-group design can describe outcomes among participants treated in the study, but it cannot provide a randomized estimate of the treatment effect relative to a control. The registry record does not report a protocol-level explanation for the design, a sample-size justification, or details about how the multiple study sites were managed analytically.

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: 49 participants signed a consent form, 10 of which were screen failures and 3 of which withdrew consent prior to completion of screening.

Overall Study

Reasons for not completing this period:

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 Pathologic Complete Response (pCR) Rate, measured over Up to 34 months.

Pathologic complete response (pCR) is the primary endpoint. The registry defines it as the number of subjects with absence of viable tumor on the surgical resection specimen, as determined by local pathology review. Its unit is participants, and its stated timeframe is up to 34 months. The record names the measure but does not report the observed number of participants meeting the definition or a pCR rate.

In general, a pathology-based endpoint can capture residual tumor status in tissue removed during surgery, rather than relying only on changes measured by imaging or symptoms. A count of participants meeting a prespecified definition can be summarized as a proportion when an appropriate denominator is specified. Interpreting that proportion requires knowing which participants were eligible for assessment, whether all enrolled participants underwent surgery, and how missing or unevaluable specimens were handled. These details are not reported in the registry record.

The endpoint is used in this study to assess tumor response following the perioperative treatment strategy. Because there is no comparison group and no posted pCR result, the registry does not establish the frequency of pCR in the study or whether the regimen changed that frequency. The local-review method is stated, but pathology criteria, review procedures, and any independent confirmation are not reported.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies the primary outcome as a count of participants with pCR. This is a binary participant-level outcome for each assessed person: the surgical specimen either meets the stated definition of absence of viable tumor or it does not. In general, binary outcomes can be reported as counts and proportions, provided the analysis denominator and handling of missing assessments are clear. The registry does not provide the pCR count, denominator, or an uncertainty estimate, so no pCR proportion can be interpreted.

What does a confidence interval communicate?

In general, a confidence interval describes uncertainty around an estimate under a specified statistical model and analysis procedure. It is not a guarantee that the true value lies within the interval, nor does it replace clear reporting of the population and endpoint being analyzed. For a participant proportion, the interval depends on the observed count, denominator, and method used. Since the registry record contains no pCR result or interval, precision for the primary endpoint is not assessable.

What does the single-group design allow?

A single-group study can describe outcomes in its enrolled treatment group. It does not directly estimate a randomized contrast with a control group. Differences between an observed outcome and outcomes in another study or an external benchmark can reflect differences in eligibility, care, assessment, follow-up, or other characteristics, as well as treatment. No external comparator or formal comparative analysis is reported in the registry record. The allocation entry is not applicable, consistent with the record’s single-group model.

How should the sample-size assumptions be read?

The registry includes a planning illustration based on an observed effect, rather than the trial’s own design. Its stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is a stated fraction of the observed effect. For time-to-event outcomes, the illustration uses a total number of events; for other outcomes, it uses participants per arm. These are general planning assumptions, not observed trial results, and the illustration is not observed power. No numeric planning outputs or numeric observed effect are reported in the registry record. The record also does not report the trial’s sample-size justification or establish that its design used these illustration assumptions.

What do the outcome types imply?

The primary endpoint is a count, while the named secondary outcomes include time-to-event outcomes—overall survival and disease-free survival—and overall response rate. In general, time-to-event analysis must account for differing follow-up and censoring, where a participant’s event time is not observed within the available follow-up. Interpretation depends on event definitions, the time origin, follow-up completeness, and analysis methods. These details and the secondary outcome results are not reported in the registry record. Overall response rate is generally a binary response outcome summarized for a defined population, but the response criteria and analysis denominator are not reported.

For count outcomes, general statistical models may need to account for over-dispersion: variability greater than a simple model assumes. Over-dispersion matters because an inadequate variance model can make uncertainty appear too small. It is not possible to determine whether this issue applies to any analysis in this study because the registry lists no analyses and does not report an analysis plan or outcome data. The primary pCR definition is a participant count, not a repeated event count; the record does not describe a model for it.

What analysis details are missing?

The registry record does not report the prespecified statistical plan, hypothesis-testing procedures, confidence-interval methods, covariate adjustment, interim analyses, or rules for stopping. It does not state whether analyses were planned for all enrolled participants, only participants who underwent surgery, or another analysis population. It also does not describe missing-data methods, endpoint adjudication, or sensitivity analyses. These omissions prevent assessment of whether the analysis choices were prespecified and how robust any result would be to alternative assumptions.

The record gives a primary endpoint timeframe of up to 34 months, but it does not clarify the time origin or specify the assessment schedule. General interpretation of a pathology endpoint depends on specimen availability and consistent assessment. Interpretation of survival outcomes depends on follow-up and censoring information. Neither set of details can be reconstructed from the outcome names alone.

Finally, the registry lists 0 analyses. That does not demonstrate that no analyses were conducted outside the registry; it means the registry record does not contain analysis entries to evaluate. Without posted outcome estimates, denominators, and methods, readers cannot verify precision, reproduce an analysis, or assess how missing observations were treated.

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(34)
events <- c(7)
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  34  7  20.6  8.7  37.9  10.3  36.8  10.4  36.9

11. Multiplicity, Missing Data and Analysis Populations

The registry names a primary endpoint and several secondary outcomes, but does not report a multiplicity strategy. In general, when multiple outcomes or statistical tests are examined, the chance of at least one apparently notable result can increase. A prespecified hierarchy, adjustment procedure, or clearly exploratory interpretation can help readers understand that issue. None is reported in the registry record, and no hypothesis-test results are posted.

The record does not report missing-data counts or methods for handling missing outcomes. For pCR, it does not state how participants without an evaluable surgical specimen were treated, whether the analysis denominator was all enrolled participants or a subset, or whether any imputation or sensitivity analysis was used. For survival outcomes, censoring rules and completeness of follow-up are not reported. Missingness can affect both the estimate and its interpretation, particularly if participants with unavailable outcomes differ from those assessed.

Analysis populations are also not described. The registry does not say whether the primary endpoint was analyzed in an intention-to-treat population, a treated population, or another defined group. No exclusions, deviations, or analysis denominators are given. The enrollment of 49 is not enough to identify the population contributing to any outcome.

The registry lists 0 analyses. The protocol’s statistical plan, interim analyses, and approach to multiplicity are not reported in the registry record. As a result, neither the analysis population nor the handling of missing information and multiple outcomes can be assessed from the posted record.

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 serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: diarrhea (17.1%), dehydration (5.7%), febrile neutropenia (5.7%), generalized muscle weakness (2.9%), hypokalemia (2.9%).

The registry lists a Pembrolizumab group with 35 participants at risk for serious adverse events and 9 participants affected. For other adverse events, it lists 35 at risk and 35 affected. It also records 8 deaths. These are group-level counts as recorded in the registry; the record does not provide the event collection period or explain whether each count refers to participants or repeated events beyond the labels shown.

The listed serious adverse-event terms and percentages are diarrhea, 17.1%; dehydration, 5.7%; febrile neutropenia, 5.7%; generalized muscle weakness, 2.9%; hypokalemia, 2.9%; and hyperglycemia, 2.9%. These entries describe the reported safety terms and their percentages, not a comparison with another group. The registry does not provide the corresponding event counts for each term, details on overlap among participants, or a control-group safety profile.

These figures do not establish that treatment caused an event or death. The record does not report attribution rules, severity grading details, exposure-adjusted rates, timing, or the causes of death. With no concurrent control arm, the counts cannot be used as a randomized comparison of safety between treatment strategies.

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

The registry record describes a phase II study of perioperative chemotherapy and pembrolizumab in gastric or GE junction adenocarcinoma, with pCR as the primary endpoint. It defines pCR using the surgical resection specimen and local pathology review. It also identifies overall survival, disease-free survival, and overall response rate as secondary outcomes. These details establish what the study was intended to examine.

The record does not report the primary endpoint result or results for the secondary outcomes. It therefore does not establish how often pCR occurred, how precisely an outcome was estimated, or whether the observed outcomes differed from any comparator. There is no control group described, and no posted statistical analyses. No efficacy conclusion can be drawn from the outcome definitions or the completed status alone.

In general, a clinical trial’s observed outcome is not automatically a causal treatment effect. Causal interpretation is stronger when treatment groups are made comparable by design and analyzed as planned. This study is single-group and open-label, so an outcome estimate would describe the enrolled group rather than provide a randomized treatment contrast. The registry record does not report details that would support adjustment for differences from an external comparison.

The safety entries do provide selected counts for the Pembrolizumab group, including serious adverse events, other adverse events, and deaths. They describe recorded events in that group, but do not establish causality or comparative safety. Important context—including timing, attribution, and follow-up—is not reported in the registry record.

The appropriate reading is limited but clear: the registry documents the design, treatment description, endpoint definitions, enrollment, and selected safety information, while leaving outcome results and analysis details unreported. The sample-size illustration is a planning exercise under stated assumptions, not evidence of observed efficacy, precision, or power in this trial.

14. Why This Trial Matters Statistically

This trial is statistically informative as an example of the difference between a study question and a reported result. The registry gives a specific primary endpoint—absence of viable tumor in the resection specimen—but does not post the outcome count or rate. A precise endpoint definition is useful, yet it cannot substitute for an estimate, denominator, and measure of uncertainty.

The design also makes the limits of interpretation important. A single-group study can characterize outcomes in the participants who received the regimen, but it lacks a concurrent randomized control for estimating a comparative effect. This distinction matters whenever a treatment outcome might otherwise be interpreted as evidence that treatment changed the outcome. The registry’s record does not support that inference.

The trial also has a mix of endpoint types: a pathology-based participant count, survival outcomes, and response rate. Each requires different reporting context. A pathology count needs a clear assessment denominator and missing-specimen handling; survival outcomes require definitions of time origin, events, follow-up, and censoring; a response rate requires response criteria and an analysis population. Those details and results are not reported in the registry record.

Finally, the safety information illustrates why event counts need context. Counts and listed terms can describe what the registry recorded, but without timing, attribution, and a comparator they do not establish causation or relative risk. Statistical interpretation is therefore shaped as much by the unavailable methods and outcomes as by the study’s stated design.

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 interpretation is that the registry record does not report results for the primary endpoint or the named secondary outcomes. It provides no pCR count, denominator, rate, confidence interval, or analysis. Consequently, the magnitude and precision of the primary outcome cannot be assessed.

The single-group, non-randomized design has no concurrent control arm. It can describe outcomes among enrolled participants but cannot isolate a treatment effect from other explanations. Open-label conduct may also be relevant to outcome assessment, although the record specifies local pathology review for pCR and does not report whether assessors were blinded or whether pathology was independently confirmed.

Further limitations arise from missing methodological details. Eligibility criteria, analysis populations, endpoint assessment schedules, handling of missing outcomes, and statistical methods are not reported in the registry record. The primary endpoint has a timeframe of up to 34 months, but the time origin and assessment process are not stated. The secondary survival outcomes lack reported definitions and follow-up procedures. The maintenance-treatment description is truncated, so its full regimen is not clear.

The safety record includes selected counts but does not report event timing, attribution, detailed collection methods, or a comparator. Deaths are listed without reported causes. The registry’s sample-size illustration is planning information based on an observed effect, but its numerical outputs and the trial’s own sample-size justification are not reported. These limitations restrict both efficacy and safety interpretation.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Perioperative Chemo and Pembrolizumab in Gastric Cancer — NCT02918162.
• Registry record last updated 2024-02-15; status last verified 2024-01.
• Results first posted 2024-02-15.
• 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 NCT02918162, a completed phase II, open-label, single-group study of pembrolizumab with standard-of-care chemotherapy for gastric or GE junction adenocarcinoma. Enrollment was 49. The recorded start date is 2017-11-25, and the primary completion date is 2023-02-28.

The primary endpoint is the count of participants with no viable tumor in the surgical resection specimen, determined by local pathology review, with a timeframe of up to 34 months. Secondary outcomes include overall survival, disease-free survival, and overall response rate. The registry record does not report outcome results or statistical analyses.

The safety record lists 35 participants at risk and 9 affected by serious adverse events in the Pembrolizumab group, 35 at risk and 35 affected by other adverse events, and 8 deaths. These counts do not establish causality. The design has no concurrent control group, and the record does not report the analysis populations, missing-data methods, or full statistical plan.