Pembrolizumab and Palliative Radiation Therapy in Treating Patients With Metastatic Esophagus, Stomach, or Gastroesophageal Junction Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pembrolizumab and Palliative Radiation Therapy in Treating Patients With Metastatic Esophagus, Stomach, or Gastroesophageal Junction Cancer (NCT02830594), a na phase 2 study. The primary outcome is Changes in Combined Positive Scoring (CPS) in Non-irradiated Sites Assessed by Flow Cytometry.

2. Clinical Question

The trial’s rationale was to study whether the combination could affect immune-related tumor measurements beyond the irradiated site. That rationale is not itself evidence of a treatment effect. The registry describes a biological endpoint, alongside secondary outcomes including treatment-related adverse events, overall response rate, progression-free survival, and overall survival. With no concurrent comparator, changes observed during treatment cannot by themselves separate treatment effects from other explanations, such as disease course or differences in which participants could contribute paired biopsies.

3. Trial Design

This was a phase II, single-group trial sponsored by City of Hope Medical Center. The registry lists enrollment of 14 and a completed status. The start date was 2016-12-13, and the primary completion date was 2022-01-20. Allocation is listed as not applicable, and masking as none. These features describe a study without a randomized control group and without masking.

Participants received pembrolizumab with palliative external beam radiation therapy. The treatment description specifies daily radiation, with the first fraction on day 1 followed by intravenous pembrolizumab over 30 minutes. Cycles were repeated every 3 weeks for up to 35 cycles, in the absence of disease progression or unacceptable toxicity. The registry also lists laboratory biomarker analysis as an intervention, reflecting the study’s measurement of biological outcomes.

A single-group design can describe what happened among people treated under the study regimen, but it cannot provide a randomized estimate of how outcomes would compare with another treatment. Differences between the observed results and a hypothetical alternative may reflect patient characteristics, disease course, or other factors, not just the regimen. The registry’s rationale about immune response and an abscopal effect should therefore be read as the scientific question motivating the study, rather than as a demonstrated finding.

The registry record does not report a protocol-based sample-size justification, a detailed statistical analysis plan, interim analyses, or a comparator group. It also does not report a numerical primary-endpoint estimate. The design and enrollment information alone cannot establish whether the trial was adequately sized to detect a clinically meaningful change.

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

Analysis population for the primary outcome, as defined in the registry: One patient did not undergo any biopsies due to site of metastasis not being visualized under image guidance and 2 patients did not undergo the second biopsy for patient safety. Four patients had posttreatment biopsies without any viable cancer cells but only residual stromal tissue.

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 Changes in Combined Positive Scoring (CPS) in Non-irradiated Sites Assessed by Flow Cytometry, measured over Baseline to 105 weeks.

The primary endpoint is described as change in combined positive scoring (CPS) in non-irradiated sites, measured by flow cytometry from baseline to 105 weeks. CPS is a tumor-related biomarker measurement; the trial used biopsies before and after treatment to assess change at sites that were not irradiated. A non-irradiated site is relevant to the study’s biological question because the investigators were interested in whether an immune-related change might be observed beyond the area receiving radiation.

The registry says multicolor immunofluorescence was used to quantify analogous CPS expression in pre- and posttreatment biopsies. It also describes testing the difference in MIDS expression with a paired t-test. The record’s use of both CPS and MIDS terminology leaves the exact naming of the measured quantity unclear. The endpoint is listed as a mean of a continuous measure, expressed as fold change, with standard error as its dispersion measure.

A paired analysis is appropriate in principle when each participant contributes measurements before and after treatment, because it focuses on within-person change rather than treating the measurements as unrelated. Its interpretation depends on having usable paired measurements and on the assumptions of the chosen test. The numerical mean change, standard error, and test result are not reported in the registry record, so the magnitude and precision of the endpoint cannot be assessed from the record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a continuous biomarker measure expressed as fold change. The registry says the result is summarized as a mean, with standard error as the dispersion measure, and describes a paired t-test for pre- and posttreatment values. No numerical estimate or test result is reported in the registry record. The method can be explained, but its result cannot be reconstructed from the endpoint description alone.

Why use paired measurements?

When a participant has a measurement before treatment and another afterward, the two observations are linked. A paired analysis works with the within-person differences. This can account for stable differences between participants that would otherwise contribute to variation in a comparison of separate group means. The approach requires usable measurements from both time points for the participant included in the paired analysis.

Pairing does not remove all uncertainty or solve missing-data problems. If a posttreatment biopsy is unavailable, that participant may not contribute a complete pair to a straightforward paired test. The registry notes biopsy omissions and biopsies with residual stromal tissue but does not explain how those cases entered the primary analysis. It also does not report whether additional rules were used to determine which tissue measurements counted as evaluable.

What do the mean and standard error convey?

A mean describes the average value of a continuous measure in the analyzed observations. Here, the measure is described as fold change, but the record does not specify the numerical scale or report the mean result. A mean can be influenced by unusually large or small observations, so it should be interpreted alongside information about variability and the distribution of the measurements.

Standard error describes the estimated precision of a mean, rather than the spread of individual participant measurements. A smaller standard error generally indicates a more precise estimate of the mean under the analysis assumptions. It does not describe the full range of participant responses. The standard error for this endpoint is not reported in the registry record, and no confidence interval is reported in the registry entries.

A confidence interval, in general, is a range constructed by a statistical procedure to express uncertainty around an estimate. It is not a guarantee that the true value lies in that particular interval. Without a reported estimate and interval, readers cannot evaluate whether plausible values are tightly clustered or span a broad range. A test result alone would also not convey the magnitude or practical importance of a change.

What does a paired t-test assume?

In general, a paired t-test evaluates whether the average within-person difference is consistent with a value of no average change under the test’s assumptions. The distribution of the paired differences is important; the test is not simply a comparison of two unrelated sets of measurements. The registry names the test but does not report diagnostic checks, transformations, the exact definition of change, or other analysis details. Those aspects of the statistical plan are not reported in the registry record.

Biomarker measures can have uneven distributions or observations that exert substantial influence on a mean. In general, a transformation or a method robust to unusual values may be considered when assumptions are not suitable, but the registry does not say whether such steps were used. It would be inappropriate to infer them from the endpoint name or from the use of a paired t-test.

How should planning calculations be read?

Planning calculations ask how much information a hypothetical study might need to detect a chosen effect under specified assumptions. They are not observed outcomes and are not a measure of the probability that this completed trial detected an effect. The registry describes planning illustrations using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The target effect is a stated fraction of the observed effect. These are illustration assumptions, not a reported justification for the single-group trial’s enrollment.

For time-to-event outcomes, the stated planning approach uses a total number of events; for other outcomes, it gives participants per arm. Such calculations depend on their assumptions and should not be treated as forecasts of what a study will observe. The registry record does not report an interim-analysis plan or a trial-specific sample-size justification. Nor do the planning assumptions create a control group or change the interpretation of this single-group study.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Treatment (Pembrolizumab, RT)")
n   <- c(7)
m   <- c(3.9)
sp  <- c(7)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Treatment (Pembrolizumab, RT)  7  3.900  18.520  -13.228  21.028

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 7 outcome measures and 3 analyses. The available record does not give a complete account of how each analysis maps to each outcome, which outcomes were designated for formal hypothesis testing, or whether a multiplicity adjustment was planned. When many outcomes or tests are examined, the chance of at least one apparently notable result can increase. In general, multiplicity procedures address that issue by controlling or describing the error rates across a set of tests. Whether such a procedure was used is not reported in the registry record.

Missing tissue measurements are directly relevant to the primary endpoint. One patient did not undergo any biopsies because the metastasis site was not visualized under image guidance, and two patients did not undergo the second biopsy for patient safety. Four patients had posttreatment biopsies containing no viable cancer cells, only residual stromal tissue. The record does not explain whether these cases were excluded, included through another measurement approach, or handled in a sensitivity analysis. It also does not describe assumptions about why measurements were unavailable.

In general, an analysis of complete pairs relies on the participants with measurements at both time points. If those participants differ systematically from those without paired observations, the analyzed group may not reflect the full enrolled group. The direction and size of any such effect cannot be inferred from the registry’s biopsy notes.

The registry does not report the primary analysis population, the number of evaluable paired measurements, a missing-data strategy, or sensitivity analyses. It also does not report whether all enrolled participants were included in each secondary outcome analysis. These omissions limit assessment of how much the endpoint result, if reported elsewhere, would depend on participant inclusion and missing-data decisions.

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: atrial fibrillation (14.3%), dysphagia (14.3%), pericardial effusion (7.1%), enterocolitis (7.1%), lung infection (7.1%).

For the treatment group, the registry records 9 participants affected by serious adverse events among 14 at risk. It records 14 participants affected by other adverse events among 14 at risk. These are participant counts, not counts of individual event episodes; one person can experience more than one event. The registry also records 9 deaths. The available information does not establish that the deaths were caused by treatment.

The listed serious adverse event terms include atrial fibrillation and dysphagia, each reported as 14.3%, and pericardial effusion, enterocolitis, lung infection, and sepsis, each reported as 7.1%. These percentages describe the registry’s listed event frequencies. The record does not provide additional detail about timing, duration, severity grading, or attribution for each event term. The figures should not be interpreted as a comparison with another treatment, because the trial had no separate control group.

Adverse-event summaries describe events recorded during the study; they do not by themselves establish whether an event was caused by the regimen. Likewise, the death count does not identify cause of death or establish treatment attribution. The secondary outcome is named as incidence of treatment-related adverse events, but the broader attribution details and numerical analysis are not reported in the registry record.

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

The trial documents an investigation of pembrolizumab given with palliative external beam radiation therapy in metastatic esophageal, gastric, or gastroesophageal junction cancer. Its primary endpoint was a biomarker change in non-irradiated sites, measured using pre- and posttreatment biopsy material. The registry describes the endpoint and a paired t-test, but does not report the numerical primary result. Therefore, the direction, magnitude, and precision of any observed change cannot be determined from the registry record.

The study’s single-group structure is central to what its results could mean. Outcomes among treated participants can describe experience under the regimen, but they do not provide a randomized estimate of the difference between the combination and an alternative. A change after treatment would not, on its own, establish that treatment caused that change. Similarly, a response or survival result, if reported, would need context and a comparison strategy to support claims about relative effects.

The biopsy notes further qualify interpretation. Tissue was unavailable for some participants for the reasons described in the registry, and some posttreatment biopsies had no viable cancer cells. Since the record does not explain how those observations were handled, the composition of the primary analysis population is unclear. This matters because a biopsy-based endpoint can only be directly evaluated when appropriate tissue measurements are available.

The registry’s rationale refers to enhancing anti-tumor T cell response and augmenting an abscopal effect. Those are motivating hypotheses, not conclusions established by the design description. The record’s completion status indicates that the study is listed as completed; it does not imply that every planned outcome has a reported result. Taken together, the information supports describing the trial’s question, regimen, design, and recorded safety figures, while leaving the primary biomarker result and several analysis details unresolved.

14. Why This Trial Matters Statistically

This trial raises a statistical question that is broader than whether a biomarker changed after treatment: how can a change measured in biopsy tissue be interpreted when the study has no control group and not every participant contributes the same tissue measurements? The paired design can focus on within-person change, but it does not establish what would have happened without the combined regimen. The single-group structure therefore limits causal interpretation even if a numerical change were available.

The endpoint also depends on tissue collection and evaluability. The registry records that biopsies could not be obtained in some circumstances and that some posttreatment samples contained residual stromal tissue without viable cancer cells. Those observations make the analysis population and rules for handling tissue availability important statistical details. They are not reported in the registry record.

Another important issue is the distinction between a mean estimate and its precision. The endpoint is described as a mean with standard error, but neither the numerical estimate nor the standard error is reported. Without them, readers cannot assess the size of the biomarker change or how uncertain the estimate is. A named statistical test does not fill that gap.

The registry lists multiple outcome measures and analyses, while the available description does not specify a multiplicity strategy or a complete analysis plan. This makes it difficult to know how formal inferences across outcomes were intended to be controlled or prioritized. These considerations are useful when judging what a completed early-phase trial can contribute: it can define a question and document an approach, but design details and transparent outcome reporting are needed to judge the strength of statistical evidence.

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 limitation is the lack of a reported numerical result for the primary endpoint. The registry describes a mean fold change and standard error and names a paired t-test, but the estimate, its uncertainty, and the test result are not reported in the registry record. This prevents assessment of the observed direction, magnitude, or precision.

The study is listed as a single-group trial with no randomized comparator. That design does not support a direct estimate of the difference between the combined regimen and another treatment. In addition, the primary endpoint depends on paired biopsies. The record notes biopsy unavailability and posttreatment samples without viable cancer cells, but does not report how those cases were included in the analysis or define the analysis population.

The endpoint terminology is not fully consistent: the title refers to CPS, while the description refers to analogous CPS expression and says the difference in MIDS expression was tested. The record does not clarify whether these terms refer to the same measured quantity or how the endpoint was operationalized. The exact scale and definition of fold change are also not reported in the registry record.

Details needed to evaluate the statistical analysis are unavailable, including a complete analysis plan, the handling of missing measurements, diagnostic checks for the paired t-test, and any multiplicity procedure. The record lists secondary outcomes, but their numerical results are not available in the registry record for this analysis. These limitations leave important questions about both the analysis and the findings unresolved.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Pembrolizumab and Palliative Radiation Therapy in Treating Patients With Metastatic Esophagus, Stomach, or Gastroesophageal Junction Cancer — NCT02830594.
• Registry record last updated 2026-11-02; status last verified 2026-10.
• Results first posted 2022-06-30.
• 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

NCT02830594 is a completed phase II, single-group trial sponsored by City of Hope Medical Center. It lists enrollment of 14 and studies pembrolizumab with palliative external beam radiation therapy for metastatic esophageal, gastric, or gastroesophageal junction cancer. The treatment description specifies daily radiation and pembrolizumab, with cycles every 3 weeks for up to 35 cycles.

The primary endpoint concerns change in CPS in non-irradiated sites from baseline to 105 weeks, using biopsy measurements and a paired t-test. The registry records challenges with obtaining or evaluating biopsies. It also lists secondary outcomes including treatment-related adverse events, response rate, progression-free survival, and overall survival. The numerical primary-endpoint result and detailed analysis procedures are not reported in the registry record. With no comparator group, the study’s design describes outcomes under the regimen but does not provide a randomized treatment comparison.