Postoperative Chemoradiation or Chemotherapy After Preoperative Chemotherapy for Gastric Cancers: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Postoperative Chemoradiation or Chemotherapy After Preoperative Chemotherapy for Gastric Cancers (NCT03515941), a non-randomized early/phase 1 study. The primary outcome is Number of Participants Who Complete the Recommended Therapy From Each Arm.

2. Clinical Question

The study also planned to evaluate recurrence and explore whether changes in a pre-specified gene panel after preoperative chemotherapy were associated with lymph node involvement at surgery. The registry lists median time to recurrence as a secondary outcome. It does not report the recurrence results or the gene-expression results.

Because treatment assignment follows lymph node status, the central question is feasibility of a status-based treatment approach, not whether chemoradiation is superior to chemotherapy. The two treatment groups differ by the feature used to assign therapy, and may differ in other ways as well. A comparison of their outcomes cannot isolate a treatment effect from those differences. The registry reports enrollment of 6, and the study status is terminated. It does not report why the study stopped or give an outcome count for the primary feasibility question.

3. Trial Design

The registry describes an open-label, stratified, two-arm pilot study with a parallel design and non-randomized allocation. Participants were to receive the same initial preoperative chemotherapy and surgical resection. After surgery, treatment assignment depended on whether lymph nodes showed cancer: node-negative patients were assigned postoperative chemotherapy, while node-positive patients were assigned postoperative chemoradiation.

The listed postoperative chemotherapy options include CAPEOX, using oxaliplatin and capecitabine, or FOLFOX, using oxaliplatin, leucovorin, and 5-fluorouracil. The chemoradiation arm description lists chemotherapy with capecitabine or 5-fluorouracil and leucovorin, followed by chemoradiation. The planned radiation course is described as 5 weeks with 45 Gy in 1.8 Gy/fr. These are registry descriptions of the regimens; the record does not report what each enrolled participant actually received or whether treatment modifications occurred.

The primary outcome is the number of participants completing the recommended therapy in each arm, with follow-up from assigned therapy up to 17 weeks. The secondary outcome listed is median time to recurrence. The stated secondary objectives also include evaluating recurrence and exploring associations between gene-expression changes and nodal status. The record does not report results for these outcomes.

The study is listed as early phase 1, with enrollment of 6 and status terminated. The start date is 2018-06-22 and primary completion is listed as 2020-03-17. The reason for termination is not reported in the registry record. Allocation was not randomized, so the treatment arms are not like-for-like control groups: their defining difference is also the basis for assignment. The registry record does not report a statistical analysis plan, a sample-size justification, or details of any interim analyses.

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

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 Number of Participants Who Complete the Recommended Therapy From Each Arm, measured over From date of assigned therapy up to 17 weeks.

The primary endpoint counts participants who complete the recommended therapy in each treatment arm. Completion is assessed from the date of assigned therapy up to 17 weeks. The registry describes this as a feasibility outcome: it is intended to address whether patients can enroll in and receive the postoperative treatment recommended according to surgical lymph node status.

A completion count can describe whether the planned treatment pathway was carried through in this study. It does not, by itself, measure tumor response, recurrence, survival, or comparative treatment efficacy. Nor does a count alone explain why a participant did or did not complete treatment. The registry record does not report reasons for non-completion, the observed completion counts, or an analysis of this endpoint.

In general, feasibility outcomes are useful in pilot work because they can indicate whether a proposed study process is workable enough to evaluate further. Their interpretation depends on the outcome definition and on complete reporting of the counts and relevant denominators. Here, the registry supplies the endpoint definition and timeframe but not the observed endpoint results. It is therefore not possible to assess completion rates or the precision of an estimate from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The design determines what statistical comparisons can mean. In a randomized trial, random allocation is used to make treatment groups comparable in expectation. This study is listed as non-randomized: postoperative therapy was assigned according to lymph node status at surgery. Therefore, any comparison between the arms is also a comparison between groups defined by nodal status. Statistical adjustment cannot automatically remove this structural difference, and the registry record does not report an adjusted analysis.

What does a count endpoint tell us?

The primary endpoint is a count of participants who complete the recommended therapy in each arm. A count is directly interpretable as the number meeting the endpoint, but its meaning depends on the number assessed and on how completion and missing status are handled. The registry gives the timeframe as up to 17 weeks from assigned therapy but does not report observed completion counts, denominators for this endpoint, or rules for participants without an observed completion status. No completion proportion or uncertainty interval can be assessed from the registry record.

In general, an interval estimate describes the range of values reasonably compatible with observed data under a statistical model. Greater precision is typically associated with narrower intervals, while sparse data often yield substantial uncertainty. An interval cannot be interpreted without knowing what quantity it estimates and what method was used. The registry record does not report an interval estimate or an analysis of the primary endpoint.

Why does allocation matter?

Because lymph node status determines treatment assignment, arm-level outcomes are not a clean estimate of what would happen if the same patients received one treatment rather than the other. Nodal status may be related to prognosis and to other clinical features, but further participant characteristics are not reported in the registry record. The observed groups also have little enrollment, making estimates especially sensitive to individual observations. This is a limitation of interpretation, not something that a significance test can resolve.

In general, a p-value measures how compatible observed data are with a specified null model and its assumptions. It does not establish causation, quantify clinical importance, or correct a non-randomized design. The registry reports 0 analyses, and no p-value or inferential comparison for the efficacy or feasibility endpoints is reported. A formal treatment-effect claim would not be justified by the design description alone.

What can safety counts show?

Safety counts describe the number of participants recorded with events in each arm, alongside the number at risk. They can identify events reported during the study, but sparse counts provide limited precision. An event count does not describe its severity, timing, duration, or attribution unless those details are reported. The registry supplies selected serious and other adverse-event counts, but does not provide detailed event-level narratives in the registry record for this analysis.

In general, when events occur more often than expected under a simple binomial model, over-dispersion may matter: the variation in outcomes can exceed what that model assumes. That possibility is especially important to consider in clinical data with differing participant risks or repeated events. The registry record does not report a model for adverse events, whether counts are participant-level or event-level for all listed terms, or any adjustment for differences between arms. The counts should therefore be read descriptively rather than as evidence of a reliable between-arm safety difference.

What is the role of planning calculations?

Any sample-size illustrations associated with the posted results are described as planning calculations based on an observed effect, with two-sided significance level 0.05, equal allocation, and a normal approximation. They are not the trial's design justification and are not observed power. A planning illustration asks how many participants per arm might be needed under stated assumptions to detect a specified fraction of an effect. It does not show that an effect is real, that the study had a particular probability of success, or that the assumptions match a future study.

For time-to-event outcomes, the stated illustration method uses a total number of events; other illustrations use participants per arm. These are general distinctions in planning calculations, not results from this trial. The record does not report an analysis of recurrence or the observed effect estimate needed to assess a trial-specific planning scenario. The registry also does not report the protocol's sample-size rationale, the statistical model for recurrence, or a multiplicity procedure. Such details should not be inferred from the existence of a planned outcome or from the study's phase designation.

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("Arm 1: Adjuvant Chemotherapy", "Arm 2: Adjuvant Chemoradiation")
n      <- c(4, 2)
events <- c(3, 2)
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  Arm 1: Adjuvant Chemotherapy  4  3  75.0  19.4  99.4  30.1  95.4  28.4  94.7
ARM  Arm 2: Adjuvant Chemoradiation  2  2  100.0  15.8  100.0  34.2  100.0  29.2  99.2

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary feasibility outcome and secondary objectives involving recurrence and gene-expression changes. It also lists median time to recurrence as a secondary outcome. When a study assesses several outcomes or several hypotheses, multiplicity can increase the chance of at least one apparently unusual finding if each is tested separately. In general, a prespecified testing hierarchy or adjustment can help control that issue, but the appropriate approach depends on the study's aims and analysis plan.

The registry record does not report a multiplicity procedure, a hierarchy of statistical tests, or results of inferential analyses. It is therefore not possible to tell whether any adjustment was planned or used. The record's listing of outcomes alone does not establish that each was formally tested.

Information about missing data and analysis populations is also not reported in the registry record. The primary endpoint description does not say how participants with unknown completion status would be handled. The record does not specify whether the endpoint was analyzed in all enrolled participants, only those who began assigned treatment, or another defined population. It also does not report missingness for recurrence follow-up or gene-expression measurements.

In general, missing outcomes can affect interpretation when the probability of missingness is related to participant characteristics or outcomes. The direction and size of any effect depend on the data and assumptions, so they cannot be reconstructed without reporting. The small enrollment and terminated status make clear reporting of participant flow and outcome availability particularly important, but those details are not reported in the registry record. No missing-data method, sensitivity analysis, or analysis-population definition can be inferred.

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: gastrointestinal pain (50.0%).

The registry lists serious adverse-event counts separately for the two treatment arms. In the adjuvant chemotherapy arm, 0 participants with serious adverse events are listed among 4 at risk. In the adjuvant chemoradiation arm, 1 participant with a serious adverse event is listed among 2 at risk. Deaths are listed as 0 in both arms.

For other adverse events, the registry lists 3 participants affected among 4 at risk in the chemotherapy arm and 2 among 2 at risk in the chemoradiation arm. These descriptive counts are not a randomized comparison: treatment assignment depended on nodal status. The small groups also mean that each observation has substantial influence on how the counts appear. They should not be treated as precise estimates of relative safety.

Gastrointestinal pain is listed among the top serious terms at 50.0%. The denominator and further context for this percentage are not reported in the registry record, so its interpretation is limited. The record does not provide detailed information about event timing, duration, severity beyond the serious classification, or attribution to treatment. The figures describe what was recorded; they do not establish that treatment caused an event or support a broader safety conclusion.

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

The record documents a pilot approach in which postoperative treatment was selected according to lymph node status after preoperative chemotherapy and surgery. Its stated primary aim was feasibility: counting participants who completed their recommended therapy. This makes the study question different from a randomized efficacy comparison. The assignment strategy and the non-randomized design are central to interpreting any arm-level result.

The record reports enrollment of 6 and a terminated status, but does not report the primary completion counts. It also does not report the secondary recurrence results or findings from the planned gene-expression exploration. Consequently, the posted information does not establish whether the feasibility aim was met, what recurrence outcomes occurred, or whether gene-expression changes were associated with nodal status.

The adverse-event counts offer limited descriptive information, including serious-event counts and deaths by arm. With the small groups and status-based treatment assignment, those counts cannot provide a dependable comparative estimate of safety. The listed percentage for gastrointestinal pain also lacks a reported denominator and context in the registry record.

In general, a pilot study may help characterize whether a study process can be carried out and what information should be reported in later research. It does not necessarily provide precise estimates or a definitive treatment comparison. Here, the registry record leaves key outcome and analysis details unavailable, so the conclusions must be correspondingly narrow. The absence of reported efficacy results is not evidence that either treatment was effective or ineffective; it means the registry record does not support such a conclusion.

14. Why This Trial Matters Statistically

This study is statistically notable for making treatment assignment part of the clinical pathway under evaluation. Patients were to receive postoperative chemotherapy or chemoradiation according to lymph node status after preoperative treatment and surgery. That design can address feasibility of a status-guided approach, but it does not create a randomized comparison of the two postoperative treatments.

The primary endpoint is a practical completion count rather than a measure of comparative efficacy. That focus is consistent with a pilot objective: before interpreting a larger outcome study, it can be useful to know whether participants can proceed through the recommended treatment pathway. However, the registry record does not report the completion counts, so the stated feasibility question cannot be answered from the posted results.

The study also illustrates why small enrollment and complete reporting matter. When a study enrolls 6, individual observations can strongly affect descriptive summaries. Clear counts, denominators, follow-up information, and explanations of missing outcomes would be needed to understand the precision and completeness of the findings. Those primary and secondary outcome details are not reported in the registry record.

Finally, the study's recurrence and gene-expression objectives point to distinct kinds of analysis: time-to-event description and exploratory association, respectively. Their methods and results are not reported. The statistical lesson is not a conclusion about the treatments, but the importance of matching claims to design and available results. A non-randomized, status-assigned pilot can inform feasibility if those outcomes are reported; it cannot by itself establish a treatment effect between the arms.

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 trial-specific limitation is the non-randomized assignment. Surgical lymph node status determined which postoperative treatment was recommended, so the arms are inherently different on a clinically relevant feature. An outcome contrast between arms would not isolate the effect of treatment. The registry record does not report an adjusted analysis that might address measured differences, and adjustment would not guarantee removal of unmeasured confounding.

Enrollment is listed as 6, and the study is terminated. Sparse data limit the precision and stability of descriptive estimates, especially for outcomes reported by arm. The registry record does not state why the study was terminated, and the reason should not be inferred from its status.

The primary endpoint is defined, but the observed completion counts are not reported. The record also lacks reported results for recurrence and the gene-expression objective. Without these outcomes, the feasibility aim and secondary objectives cannot be evaluated from the registry information. Reasons for non-completion, participant flow, and outcome availability are not reported either.

The safety information is limited to selected counts and a listed serious-event term. Details about event timing, duration, attribution, and the context for the reported gastrointestinal pain percentage are not reported in the registry record. The record also does not report a statistical analysis plan, handling of missing data, analysis populations, multiplicity procedures, or the sample-size rationale. These omissions constrain both statistical interpretation and assessment of how the analyses were intended to address the study objectives.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Postoperative Chemoradiation or Chemotherapy After Preoperative Chemotherapy for Gastric Cancers — NCT03515941.
• Registry record last updated 2020-10-12; status last verified 2020-10.
• Results first posted 2020-10-12.
• 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 an open-label, non-randomized, parallel pilot study of postoperative chemotherapy or chemoradiation assigned according to lymph node status after preoperative chemotherapy and surgery. It lists early phase 1, enrollment of 6, and terminated status. The primary outcome is the number completing recommended therapy within the stated timeframe of up to 17 weeks after assignment.

The registry does not report the primary completion counts, recurrence results, or gene-expression findings. It lists selected adverse-event counts by arm and reports 0 deaths in both arms. Because assignment was based on nodal status and the study enrollment was small, the arms are not a like-for-like treatment comparison. The posted record supports a description of the planned feasibility question and selected safety counts, but not a conclusion about comparative treatment effects.