Phase II Study of Short Course FOLFOX Chemotherapy With Either Nivolumab or Nivolumab + Radiation in the First Line Treatment of Metastatic or Unresectable Gastroesophageal Cancers (BMS Protocol CA209-76L): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Phase II Study of Short Course FOLFOX Chemotherapy With Either Nivolumab or Nivolumab + Radiation in the First Line Treatment of Metastatic or Unresectable Gastroesophageal Cancers (BMS Protocol CA209-76L) (NCT04021108), a randomized phase 2 study. The primary outcome is Number of Patients With 12-month Progression Free Survival in Each Arm.

2. Clinical Question

The statistical question is whether the observed 12-month progression-free survival counts distinguish the randomized strategies in this study population. The posted comparison is expressed as a risk difference, with Cohort 2 compared with Cohort 1. Its point estimate is -0.155, and its confidence interval extends from -0.342 to 0.032. The unadjusted p-value is 0.109. These figures describe the observed data and their uncertainty; they do not establish that the strategies have identical effects or that a difference is present.

The study is described as enrolling participants who had not progressed on initial FOLFOX and nivolumab. Therefore, the randomized comparison pertains to that selected post-initial-treatment population, rather than to every person beginning treatment for metastatic or unresectable disease. The registry record does not report further detail about the population used for the posted endpoint beyond the arm counts and endpoint description.

3. Trial Design

This is a randomized, phase II, parallel-group study sponsored by Weill Medical College of Cornell University. The registry lists an enrollment of 80, an allocation of randomized, and masking of none. The trial title describes a comparison of short-course FOLFOX chemotherapy with either nivolumab or nivolumab plus radiation in first-line treatment of metastatic or unresectable gastroesophageal cancers. The condition listed is gastroesophageal adenocarcinoma.

The trial summary describes an initial period of FOLFOX and nivolumab. Participants who did not have disease progression were eligible for the randomized comparison; the arm descriptions state that participants who were responding would receive FOLFOX plus nivolumab for one additional month and then be randomized to Cohort 1 or Cohort 2. The registry record does not give complete arm descriptions in the available text, so treatment details beyond the strategies named in the title and endpoint description are not reported in the registry record.

The posted primary endpoint counts are 9 participants in Cohort 1 and 4 in Cohort 2 who achieved 12-month progression-free survival. The reported arm sizes for this endpoint are 33 and 34, respectively. Those counts are the basis for the posted unadjusted comparison. They are not the same as the registry's enrollment figure: the record does not explain the difference between the enrollment listing and the numbers contributing to this endpoint.

The registry lists the study as active, not recruiting. The listed study start is 2019-07-22, and the listed primary completion date is 2025-06-11. These registry entries describe the study record; they do not, by themselves, specify when the posted results were analyzed or whether all follow-up is complete. The registry record does not report the original design assumptions, detailed analysis populations, or a statistical analysis plan. It also does not report masking beyond the listing that masking was none.

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: Of the 80 enrolled patients, 13 were not randomized, leaving 67 randomized patients. Reasons for non-randomization included progressive disease (n=7), non-evaluable disease (n=5), and withdrawal of consent (n=1).

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 Number of Patients With 12-month Progression Free Survival in Each Arm, measured over 12 months.

The primary endpoint counts participants in each randomized arm who were without disease progression at 12 months. It is a binary outcome at a specified time: each participant is classified according to whether the endpoint was achieved, and the reported result is the number meeting that definition in each arm. The registry describes the measure as the number of patients without disease progression at 12 months.

The posted results give 9 of 33 for Cohort 1 and 4 of 34 for Cohort 2. The comparison summarizes the difference between the observed proportions, with Cohort 2 contrasted against Cohort 1. A fixed-time endpoint can make an arm-level comparison easier to interpret than a count of events accumulated over varying follow-up. It focuses attention on the proportion meeting the specified status at the stated time point.

As a general statistical point, a binary endpoint discards some detail about when progression occurred and may not capture the full course of a participant's disease. Interpretation also depends on how the endpoint handles events such as death, incomplete follow-up, or unavailable assessments. The registry record does not report those operational details, including the precise handling of missing 12-month status. The endpoint should therefore be read as the registry-defined count of participants without progression at 12 months, rather than as a complete description of time-to-progression experience.

Progression-free survival is used to summarize whether disease progression has occurred by a clinically specified follow-up point. This result is distinct from overall survival, which is listed as a secondary endpoint and concerns survival at one year. Neither endpoint alone describes every benefit or harm relevant to treatment decisions.

7. Results Posted in the Registry

• Cohort 2: 11.8% (4/34) versus 27.3% (9/33) on control; risk difference -15.5 percentage points (95% CI -34.2 to 3.2), risk ratio 0.43 (0.15–1.27); unadjusted two-proportion z-test p = 0.109.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Amongst All Randomized Subjects, Overall Survival, as Measured by the Rate of Survival in Patients, at 1-year

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Cohort 2:: 64.7% (22/34) versus 72.7% (24/33) on control; risk difference -8.0 percentage points (95% CI -30.1 to 14.1), risk ratio 0.89 (0.64–1.23); unadjusted two-proportion z-test p = 0.479.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Cohort 2: and the control. The interval gives the range of effects that remain compatible with the data.

The registry lists secondary outcomes covering 12-month progression-free survival, one-year overall survival, and significant grade 3 and grade 4 toxicity attributed to immunotherapy. The posted numerical secondary comparison available in the record is for one-year overall survival. Numerical results for the other listed secondary outcomes are not reported in the registry record. The safety summary contains adverse-event counts, but those counts are not a substitute for a reported comparative result on the specified toxicity endpoint.

The overall survival comparison is directionally consistent with the primary point estimate in that Cohort 2 has a lower observed value than Cohort 1. Its confidence interval is broad and includes no difference as well as values in either direction. This overall pattern does not independently establish a treatment effect, and overall survival does not measure the same outcome as progression-free survival. The endpoint estimates should be interpreted separately rather than combined into a single conclusion.

Secondary outcomes often provide context about other aspects of the study, but they usually carry less confirmatory weight than a prespecified primary outcome. Multiple comparisons create more opportunities for chance findings, particularly when many outcomes are tested. An unadjusted secondary p-value does not account for that multiplicity. In general, multiplicity control requires a specified approach to the family of tests; such an approach is not reported in the registry record.

The registry record does not report a complete set of numerical secondary analyses, an adjustment plan, or the analysis population for each outcome. This limits the ability to assess whether the secondary findings were prespecified, how missing observations were handled, and how to compare their evidential strength. Secondary results can help describe the overall pattern, but they should not be treated as separate proof that confirms the primary result or resolves its uncertainty.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Amongst All Randomized Subjects, the Number of Subjects That Achieved 12-month Progression Free Survival
• Amongst All Randomized Subjects, Overall Survival, as Measured by the Rate of Survival in Patients, at 1-year
• Occurrence of Significant Toxicity, as Measured by Rate of Grade 3 and Grade 4 Adverse Events (Combined) Attributable to Immunotherapy Per the Study Cohort

9. Statistical Interpretation

10. Statistical Methods Explained

What does the primary comparison estimate?

The primary outcome is binary: participants either met the registry's definition of being without disease progression at 12 months or did not. The arm summaries report the number meeting the endpoint and the number included in each arm. The comparison is a risk difference, which expresses the difference between the two observed proportions. The direction is Cohort 2 compared with Cohort 1, so a negative estimate means the observed proportion was lower in Cohort 2. The posted estimate is -0.155.

As a general statistical explanation, a risk difference is an absolute contrast, not a relative one. It describes how far apart the observed arm proportions are on the probability scale. It differs from a risk ratio, which compares proportions by division, and an odds ratio, which compares odds. The registry record also lists relative measures in the analysis output, but the principal interpretation follows the posted risk difference and its confidence interval.

What does a confidence interval express?

A confidence interval gives a range of values compatible with the data under the method used to calculate it. The primary interval, from -0.342 to 0.032, includes zero, the value corresponding to no risk difference. It also includes negative values, which would represent a lower proportion in Cohort 2, and a positive value, which would represent a higher proportion in Cohort 2. This range indicates that the point estimate alone should not be treated as a precise measure of the underlying difference.

In general, a confidence interval is not a range that assigns a probability to the true effect after observing the data. It is a procedure with repeated-sampling properties under its assumptions. Its width is informative: a wider interval conveys less precision than a narrower one. The registry record does not state the specific interval-construction method or provide an adjusted interval.

How should the p-value be read?

The unadjusted p-value for the primary comparison is 0.109. In general, a p-value measures how unusual data at least as incompatible with a stated null model would be if that model and the test assumptions held. It is not the probability that the null is true, and it does not measure the size or clinical importance of an effect. A p-value should be read alongside the estimate, interval, design, and outcome definition. The registry record does not report a multiplicity-adjusted p-value or a formal analysis plan.

The term unadjusted matters because multiple outcomes or comparisons can raise the chance of finding a nominally notable result when many tests are conducted. Adjustment methods can address specified families of tests, but the registry record does not report whether a multiplicity strategy was planned. No such adjustment should be assumed. The posted p-value is a simple, unadjusted comparison.

Why do modest counts affect interpretation?

For a binary outcome, the amount of information depends in part on how many participants contribute and how many meet the endpoint. Here, the posted primary counts are 9 and 4, with arm denominators of 33 and 34. With limited counts, moving a small number of outcome classifications can materially affect an estimated proportion. This general property of binary data helps explain why an estimate can have substantial uncertainty even when the point estimate appears notable.

The analysis compares the endpoint as posted; the record does not detail whether every randomized participant had complete 12-month follow-up, how censoring was handled, or how missing status was classified. These details matter because outcome availability can influence a binary endpoint at a fixed time. They are not reported in the registry record, so the numerical comparison should not be read as proof that all randomized participants had the same duration or completeness of observation.

What is the role of randomization and parallel groups?

Random allocation is intended, in general, to make treatment groups comparable in expectation, reducing systematic differences in baseline characteristics. A parallel design assigns participants to one study strategy and compares outcomes between groups. The registry describes allocation as randomized and the model as parallel. It lists masking as none, meaning the record does not indicate blinded assignment. Lack of masking can matter for outcomes involving judgment or behavior, though the record does not describe how endpoint assessment was conducted.

Randomization does not guarantee that groups will be identical in a study of this size, and it does not resolve limitations in outcome definitions, missing data, follow-up, or analysis reporting. The registry record does not give baseline characteristics or a detailed accounting of participants through the study stages. It is therefore not possible to assess balance or attrition in the registry record recorded.

How should the other measures be distinguished?

The secondary overall survival measure is reported as a percentage at one year, with a risk-difference comparison. It is related to but distinct from progression-free survival: the latter concerns progression status at 12 months, whereas the former concerns survival at one year. A participant can contribute differently to these endpoints, and one should not substitute one outcome for the other.

The safety summary reports participants affected by serious adverse events and other adverse events, as well as deaths. These are descriptive counts. They do not, by themselves, provide an exposure-adjusted rate, establish causal attribution, or show how long participants were observed. The registry record does not report detailed definitions or a comparative statistical analysis for these safety summaries.

Finally, the posted sample-size illustrations use a normal approximation, equal allocation, and a two-sided significance level of 0.05. They scale the observed effect to specified fractions for planning illustrations. Those are general planning calculations, not the trial's original design basis. The record does not report the trial's original sample-size justification or analysis assumptions.

11. Confidence Intervals and Precision

The primary risk-difference estimate is -0.155, with a confidence interval from -0.342 to 0.032. Because the interval includes zero, the result is compatible with no difference under the interval's method. It also includes negative and positive differences, so it permits effects in either direction. The point estimate should not be read without this range: the interval shows that the comparison is imprecise.

The interval's breadth is important. The observed counts are 9 of 33 in Cohort 1 and 4 of 34 in Cohort 2. Binary outcomes based on relatively small counts can yield considerable sampling uncertainty. In general, greater information tends to produce more precise estimates, while limited information tends to leave a wider range of plausible values. The record does not report the exact interval method or an adjusted estimate.

The overall survival comparison is also uncertain. Its risk difference is -0.080, with a confidence interval from -0.301 to 0.141. This range includes zero and effects in either direction. The interval does not demonstrate equivalence: showing that differences are sufficiently small would require a prespecified equivalence or non-inferiority framework and its margin, neither of which is reported in the registry record.

Confidence intervals are not guarantees about the true value, and they do not capture every source of uncertainty. They reflect sampling uncertainty under a chosen analysis method. Unreported details about follow-up, missing outcome status, endpoint handling, and analysis populations can affect how fully the posted intervals characterize uncertainty. The registry record does not report those details. The appropriate reading is therefore that the posted estimates are informative descriptions of the observed comparison, but not highly precise or complete statements about the underlying treatment effects.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size illustrations describe participants per arm for a future two-group comparison, not total events. The endpoint is binary, so these figures are participants per arm. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as the stated fraction of the observed effect. The original trial design assumptions are not reported in the registry record.

For planning for 100% of the observed effect, the illustration gives 102 participants per arm for 80% power and 136 participants per arm for 90% power. For planning for 75% of the observed effect, it gives 195 participants per arm for 80% power and 260 participants per arm for 90% power. For planning for 50% of the observed effect, it gives 467 participants per arm for 80% power and 625 participants per arm for 90% power.

In plain terms, the 80% power column indicates the approximate number of participants per arm that the stated calculation would require for an 80% chance of detecting the specified effect under its assumptions, if that effect were the true one. The 90% column uses a higher target chance and therefore requires more participants for the same assumed effect. A smaller effect is harder to distinguish from random variation, so the illustrations require more participants as the assumed effect shrinks.

These calculations use the observed effect as their starting point, which can be unstable when the original study has limited information. They are planning illustrations, not observed power. They do not show the probability that the trial's conclusion is correct, and they are not evidence that the trial was designed with these assumptions.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.273; treatment proportion p₁ = 0.118; average p̄ = 0.195.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5605 = 1.0986.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5497 = 0.4626.
4. n = (1.0986 + 0.4626)² / (-0.1551)² = 101.3, rounded up to 102 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.272727272727273                          # control proportion
p1 <- 0.117647058823529                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(4021108)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 102 (80% power), 136 (90% power)
 75% of observed effect: n per arm = 195 (80% power), 260 (90% power)
 50% of observed effect: n per arm = 467 (80% power), 625 (90% power)
power.prop.test: 101.3 
Simulated power at n = 102 per arm: 0.808 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Multiplicity, Missing Data and Analysis Populations

The registry record does not report how the study addressed multiplicity across its primary and secondary outcomes, or whether any formal hierarchy or adjustment procedure was specified. Multiplicity arises when several outcomes or comparisons are examined: even if each test is conducted at a conventional threshold, the chance of at least one apparently unusual result can increase as the number of tests grows. As a general statistical principle, interpretation of unadjusted results should account for how many analyses were planned and reported.

The primary comparison is presented with an unadjusted p-value. The one-year overall survival comparison is also unadjusted. The record lists other secondary outcomes but does not report numerical results for all of them. Without a reported analysis plan and a complete account of tested outcomes, it is not possible to determine the intended testing family or whether selective reporting affected the apparent pattern. The unadjusted values should therefore be read as descriptive evidence rather than as multiplicity-controlled confirmatory tests.

Missing-data handling is also not reported in the registry record. For a fixed-time endpoint, the approach to participants without an observed status at the target time can affect the denominator, the estimated proportion, and the uncertainty. The record does not explain how incomplete follow-up, unavailable assessments, or other missing observations were treated. It also does not report whether participants were analyzed according to randomized assignment, according to treatment received, or under another analysis population definition.

The registry lists a randomized parallel design, but the reported endpoint arm counts do not by themselves establish how participants moved through screening, initial treatment, eligibility, randomization, and follow-up. The record does not report a detailed participant flow or baseline characteristics. It is therefore not possible to evaluate attrition, assess balance between randomized groups, or determine whether the posted analyses include all randomized participants.

These omissions do not invalidate the posted counts, but they limit how completely the result can be assessed. A fuller interpretation would require definitions of analysis populations, missing-data procedures, outcome assessment, multiplicity handling, and the analysis plan. Those details are not reported in the registry record.

15. 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: neutrophil count decreased (24.2%), hyperglycemia (9.1%), alanine aminotransferase increased (6.1%), aspartate aminotransferase increased (6.1%), covid-19 (6.1%).

The adverse-event summary reports that 18 participants among 33 at risk in Cohort 1 and 18 among 34 at risk in Cohort 2 were affected in the serious-event category. In the other adverse-event category, the counts are 22 among 33 and 23 among 34, respectively. The summary also reports 23 deaths in Cohort 1 and 28 in Cohort 2. These figures are descriptive counts; the registry record does not provide enough detail to interpret them as treatment-caused events or as rates adjusted for exposure or follow-up.

The registry lists neutrophil count decreased as a leading serious adverse-event term at 24.2%. Other listed terms are hyperglycemia at 9.1%, alanine aminotransferase increased at 6.1%, aspartate aminotransferase increased at 6.1%, COVID-19 at 6.1%, and diarrhea at 6.1%. A term's reported frequency does not establish its cause, severity for every participant, duration, or relationship to a specific study arm. The record does not report a comparative statistical analysis for these terms or define the attribution process in detail.

Safety counts should be read alongside the study's treatment sequence and follow-up, but that detail is not fully reported in the registry record. The summaries alone cannot establish comparative safety.

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

The posted results show the numbers meeting the 12-month progression-free survival endpoint in each randomized cohort and give an unadjusted comparison between them. The point estimate is in the direction of a lower observed proportion in Cohort 2, but its confidence interval includes no difference and extends to values on either side. The result is therefore uncertain and does not establish a precise comparative effect. The one-year overall survival comparison is likewise imprecise.

These findings do not show that the treatment strategies are equivalent. A non-significant result is not proof of no effect, and the confidence intervals allow a range of possible differences. Conversely, the point estimate alone is not evidence that one strategy produces a true advantage. The registry record does not report a non-inferiority or equivalence framework, nor does it specify a clinical threshold for a meaningful difference.

The study's randomized comparison applies to participants who reached randomization after the initial FOLFOX and nivolumab period without disease progression, as described in the registry. It does not directly estimate outcomes for everyone beginning first-line therapy, including people who did not reach randomization. Generalizing beyond the randomized population requires caution.

The primary endpoint is progression-free status at 12 months, not a complete account of disease course, quality of life, symptoms, or overall survival. The registry record lists other outcomes, but numerical results for some are not reported. The adverse-event summaries provide counts and terms, but they do not establish causal attribution or a complete comparative safety profile.

In general, trial results are interpreted by combining the design, endpoint definition, estimates, precision, missing-data methods, and the complete analysis plan. Several of those details are not reported in the registry record. The posted data are best understood as an uncertain phase II comparison in a selected randomized population, rather than as a definitive conclusion about treatment effects or a basis for medical advice.

17. Why This Trial Matters Statistically

This study is statistically notable for examining a randomized comparison after an initial treatment period, rather than comparing strategies from the start of therapy. That sequence defines the population to which the primary result applies: participants who had not progressed on initial FOLFOX and nivolumab and proceeded to randomization. Keeping that distinction clear helps avoid applying the comparison to a broader population than the trial actually randomized.

The primary endpoint uses a fixed-time binary measure of progression-free status. Such measures can be straightforward to communicate, but they compress follow-up into whether a participant met the endpoint at a specified time. The arm counts and risk difference therefore need to be read with attention to follow-up completeness and the handling of unavailable status, details that are not reported in the registry record.

The posted confidence interval spans no difference and effects in both directions. This makes precision central to interpretation: the point estimate is not enough to settle the comparison. The sample-size illustrations also show how planning requirements rise when the target effect is smaller, under their stated assumptions. Those illustrations are not the study's original design rationale and should not be mistaken for evidence about the trial's achieved power.

The record's incomplete reporting of analysis populations, missing-data procedures, multiplicity, and some secondary outcomes limits the strength of conclusions that can be drawn. Statistical significance alone would not resolve these questions, and the unadjusted secondary comparisons carry less weight when the testing strategy is not reported. The trial is therefore a useful example of why effect estimates, intervals, design details, and reporting completeness all matter when interpreting clinical research.

18. 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 registry record does not give complete arm descriptions. The title and endpoint description distinguish nivolumab alone from nivolumab with radiation, but the abbreviated cohort descriptions do not establish which named cohort received which strategy. The primary counts are therefore reported by cohort rather than assigned to a specific treatment strategy.

The record describes initial FOLFOX and nivolumab before randomization, with randomization among participants whose disease had not progressed. This means the posted comparison is conditional on reaching that stage. The record does not provide a detailed participant flow, the number screened or excluded, or reasons participants did not proceed to randomization. The general enrollment listing and endpoint arm counts differ, and the reason is not reported in the registry record.

The endpoint is defined as the number without progression at 12 months, but operational details are not reported. The record does not explain how death, incomplete follow-up, or missing assessment was handled, nor does it describe endpoint adjudication. Analysis populations and baseline characteristics are also not reported, limiting assessment of attrition, comparability, and the precise population represented by the estimates.

The study is listed as unmasked. The registry record does not report whether outcome assessors were masked or how potential assessment bias was mitigated. It also does not report the original sample-size justification, interim analyses, a statistical analysis plan, or multiplicity procedures. Numerical results are not reported for all listed secondary outcomes, and the adverse-event summary does not establish causation or provide exposure-adjusted rates.

These limitations restrict interpretation of both the efficacy and safety summaries. They do not change the posted numbers, but they leave important questions about the analysis and follow-up unresolved.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Phase II Study of Short Course FOLFOX Chemotherapy With Either Nivolumab or Nivolumab + Radiation in the First Line Treatment of Metastatic or Unresectable Gastroesophageal Cancers (BMS Protocol CA209-76L) — NCT04021108.
• Registry record last updated 2026-08-21; status last verified 2026-07.
• Results first posted 2026-06-24.
• 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.

22. Record Summary

NCT04021108 is a randomized phase II parallel-group study of treatment strategies involving nivolumab, with or without radiation, after initial FOLFOX and nivolumab in metastatic or unresectable gastroesophageal adenocarcinoma. The registry lists no masking, enrollment of 80, and an active, not recruiting status.

The primary endpoint is the number of participants with 12-month progression-free survival. The posted counts are 9 of 33 in Cohort 1 and 4 of 34 in Cohort 2. The risk difference for Cohort 2 compared with Cohort 1 is -0.155, with a confidence interval from -0.342 to 0.032 and an unadjusted p-value of 0.109.

The posted one-year overall survival comparison is also imprecise. The registry record does not report several important analysis details, including missing-data handling, analysis populations, and multiplicity procedures; numerical results for some listed secondary outcomes are also not reported. These limitations call for cautious interpretation of the observed comparisons.