Avelumab in Third-Line Gastric Cancer (JAVELIN Gastric 300): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Avelumab in Third-Line Gastric Cancer (JAVELIN Gastric 300) (NCT02625623), a randomized phase 3 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

The registry record describes this as a phase III superiority trial. Its primary endpoint was overall survival, a time-to-event measure that captures how long participants remained alive after randomization. The posted comparison is between two treatment strategies, each delivered with BSC, rather than between a treatment and no care.

The posted results do not show a clear separation in overall survival: the median estimates were close, and the confidence interval for the reported hazard ratio includes the value representing no relative difference in hazard. This describes the estimates and their uncertainty; it does not establish that the strategies are equivalent. The record does not report a formal equivalence or non-inferiority objective.

3. Trial Design

JAVELIN Gastric 300 was a phase III, randomized, open-label, parallel-group study. The registry lists an enrollment of 371 participants. The primary analysis included 371 randomized participants: 186 assigned to physician-choice chemotherapy plus BSC and 185 assigned to avelumab plus BSC. The registry status is completed.

The trial compared avelumab plus BSC with BSC plus physician-choice chemotherapy. The physician selected chemotherapy from a prespecified list; the listed interventions include paclitaxel and irinotecan. The description states that paclitaxel could be given on Days 1, 8, and 15 of a 4-week treatment cycle until confirmed progressive disease or unacceptable toxicity. The irinotecan description is incomplete in the registry record, so its full regimen is not reported there. The avelumab arm description specifies a 1-hour intravenous infusion once every 2-week treatment cycle until confirmed progressive disease or unacceptable toxicity. The BSC description says that it was intended to maximize quality of life, but the full definition is not reported in the registry record.

Because the trial was open-label, participants and investigators were not masked to treatment assignment. Randomization helps create comparable groups at the start of a trial, while open-label conduct can affect care, reporting, or assessment after assignment. The registry does not report further details about the randomization process, allocation concealment, or steps taken to limit possible effects of open-label assessment.

The registry lists overall survival as the primary outcome and progression-free survival, response, and quality-of-life measures among the secondary outcomes. The stated primary and secondary time-to-event timeframe runs from randomization up to 627 days. The original sample-size justification and the trial’s original statistical design assumptions are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS included all participants who were randomized to study treatment.

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.

Population: Full analysis set (FAS) included all participants who were randomized to study treatment.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over From randomization up to 627 days.

Overall survival is the time from randomization to death from any cause. In this trial, participants who were alive at the time of analysis or lost to follow-up had their survival time censored at the date of last contact. Censoring means that the exact time of death was not observed during the available follow-up; it does not mean that the participant died at the censoring date.

The registry says overall survival was estimated using the Kaplan-Meier method. This method uses observed event times and censoring information to estimate the survival experience over follow-up. The posted summary is the median survival time: the point at which the estimated survival curve reaches the halfway mark. A median is a summary of the curve, not a claim that each participant lived for that duration.

Overall survival is used in clinical trials because death is an objectively defined event and the measure summarizes survival time across the randomized groups. The hazard ratio provides a relative comparison of the event rate over the observed follow-up under the analysis model; it is not a difference in median survival time. The trial’s primary question was whether avelumab plus BSC was superior to physician-choice chemotherapy plus BSC on overall survival. The registry record does not report further detail about the original analysis plan or assumptions supporting this endpoint.

7. Results Posted in the Registry

• Physician Choice Chemotherapy + Best Supportive Care (BSC) vs Avelumab + BSC: hazard ratio 1.11 (95% CI 0.88–1.41), reported p = 0.808 (as reported in the registry).

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.

Progression Free Survival (PFS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Physician Choice Chemotherapy + Best Supportive Care (BSC) vs Avelumab + BSC: hazard ratio 1.73 (95% CI 1.36–2.21), reported p = 1.000 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Physician Choice Chemotherapy + Best Supportive Care (BSC) vs Avelumab + BSC and the control. The interval gives the range of effects that remain compatible with the data.

Objective Response Rate (ORR)

• 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:
  • Avelumab + SC: 2.2% (4/185) versus 4.3% (8/186) on control; risk difference -2.1 percentage points (95% CI -5.7 to 1.5), risk ratio 0.50 (0.15–1.64); unadjusted two-proportion z-test p = 0.244.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Avelumab + SC and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline in European Quality of Life 5-dimensions (EQ-5D-5L) Health Outcome Questionnaire Through Composite Index Score at End of Treatment (EOT)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Avelumab + BSC: mean -0.14 versus -0.10 on control; difference -0.04 (95% CI -0.11 to 0.02), standardized effect d = -0.20; z-test p = 0.211.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Avelumab + BSC and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover several domains: time to progression or death, tumor response, and patient-reported quality of life. The available posted estimates do not form a consistent body of evidence favoring avelumab across those domains. They should be viewed as additional context rather than as a replacement for the primary overall-survival comparison.

The secondary results do not overturn the primary finding. Overall survival was the primary endpoint and the reported analysis did not demonstrate superiority. Secondary endpoints may help describe other aspects of the treatment comparison, but they answer distinct questions and may have different analysis populations, methods, and precision. A finding on one secondary measure cannot establish an overall-survival effect.

The registry lists several quality-of-life outcomes, including the EQ-5D-5L visual analogue scale, EORTC QLQ-C30 global health status, and EORTC QLQ-STO22 scores. Results for these listed measures are not reported in the registry record. The posted EQ-5D-5L composite-index change is therefore only part of the quality-of-life information named in the record, and it cannot stand in for the unreported measures.

Multiple secondary outcomes create a multiplicity concern: when many outcomes are tested, the chance of at least one apparently notable result arising by chance can increase if testing is not addressed. The registry record does not report whether or how multiplicity was controlled. For that reason, unadjusted secondary results generally carry less confirmatory weight than a prespecified primary analysis with a reported method. This caution does not make the secondary results irrelevant; it limits how strongly they can support a broad conclusion.

The registry lists eight outcome measures and three analyses, but does not give a complete account of the planned analysis hierarchy. The broader pattern should thus be read descriptively, with attention to what is and is not posted, rather than as a collection of independent confirmatory tests.

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

• Progression Free Survival (PFS)
• Best Overall Response (BOR)
• Objective Response Rate (ORR)
• Change From Baseline in European Quality of Life 5-dimensions (EQ-5D-5L) Health Outcome Questionnaire Through Composite Index Score at End of Treatment (EOT)
• Change From Baseline in European Quality of Life 5-dimensions (EQ-5D-5L) Health Outcome Questionnaire Through Visual Analogue Scale (VAS) at End of Treatment (EOT)
• Change From Baseline in European Organization for the Research and Treatment of Cancer Quality of Life (EORTC QLQ-C30) Global Health Status Scale Score at End of Treatment (EOT)
• Change From Baseline in European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire-Stomach Cancer Specific (EORTC QLQ-STO22) Questionnaire Scores at End of Treatment (EOT)

9. Statistical Interpretation

10. Statistical Methods Explained

How should the survival summaries be read?

The primary outcome is time to death from any cause, measured from randomization. The Kaplan-Meier method estimates survival over time while accounting for participants whose outcome time is censored. A participant censored at last contact contributes information up to that contact; the method does not assume that death occurred then. The median is read from the estimated survival curve, while the confidence interval describes uncertainty around the median estimate.

The registry reports a hazard ratio from a Log Rank analysis. A hazard ratio compares the event hazards between the two groups over follow-up, as summarized by the analysis. It is not a ratio of the two median survival times and does not directly state how many months participants gained or lost. The reported comparison is physician-choice chemotherapy plus BSC versus avelumab plus BSC, so the numerical estimate must be read with that group ordering in mind.

What does a confidence interval contribute?

A confidence interval gives a range of values compatible with the observed data under the statistical method and its assumptions. It is useful for judging both the direction and precision of an estimate. For a hazard ratio, a confidence interval that includes 1 does not rule out no relative difference in hazard. That does not demonstrate equivalence; it means the data and method do not distinguish the estimate from that reference value at the stated interval level.

The p-value summarizes how unusual the data, or more extreme data, would be under a specified null hypothesis and analysis procedure. It is not the probability that the null hypothesis is true, nor is it a measure of clinical importance. The primary result’s p-value is reported as 0.808. Read alongside the interval and the estimated hazard ratio, it does not support the trial’s stated superiority objective.

Why does the analysis population matter?

The primary analysis population is the full analysis set, which the registry defines as all participants randomized to study treatment. The posted arm sizes for the primary outcome are 186 and 185. Using all randomized participants is consistent with preserving the comparison created by randomization, although the registry record does not give further detail about deviations from assigned treatment or the handling of particular analysis cases.

Outcome-specific denominators can differ. For example, the quality-of-life composite index results use smaller groups than the primary survival analysis. That difference means the quality-of-life summary describes participants with data in that analysis, rather than automatically representing every randomized participant. The registry record does not report the reasons for the difference or the missing-data methods for those quality-of-life results.

How do the secondary analyses differ?

Progression-free survival is also a time-to-event outcome and is summarized with medians, confidence intervals, and a reported hazard ratio. Objective response rate is a binary outcome: participants are classified according to whether they met the response definition. Its estimate is expressed as a percentage, and the registry also reports a Cochran-Mantel-Haenszel odds ratio. The EQ-5D-5L composite index is a continuous measure; the posted values describe mean change from baseline and standard deviations.

These summaries use different scales and methods, so their numerical estimates are not interchangeable. A hazard ratio is not comparable in magnitude with a percentage, odds ratio, or mean change. Each endpoint also has its own uncertainty and analysis population. The posted tables should therefore be read endpoint by endpoint, with attention to the type of measure, group order, and confidence interval where one is available.

What is not described?

The registry record does not report the full statistical analysis plan, the trial’s original sample-size justification, interim analyses, or a multiplicity strategy. It also does not provide a complete account of missing-data methods for the secondary outcomes. The overall-survival description does state how survival time was censored for participants alive at analysis or lost to follow-up, but that does not specify how missing information was addressed for every other endpoint.

Without those details, it is not possible to determine from the registry alone whether all planned analyses are represented, how secondary testing was controlled, or how sensitive results are to alternate analysis choices. The numerical results remain informative as posted, but their interpretation should stay within the scope of the reported methods.

11. Confidence Intervals and Precision

The overall-survival median estimate for physician-choice chemotherapy plus BSC is 5.0 months, with a 95% confidence interval from 4.5 to 6.3 months. For avelumab plus BSC, the median is 4.6 months, with a 95% confidence interval from 3.6 to 5.7 months. The intervals indicate uncertainty around each group’s estimated median; they are not ranges containing a stated proportion of individual participants’ survival times.

For the reported hazard ratio, the estimate is 1.11 and the 95% confidence interval is 0.88 to 1.41. The interval includes 1, so the posted data do not rule out no relative difference in hazard. It also spans values on either side of 1, leaving uncertainty about the direction and size of the relative effect. The p-value of 0.808 is consistent with the absence of evidence for superiority in the reported test, but it does not quantify the probability that the treatment strategies are the same.

Confidence intervals are most useful when considered with the endpoint, model, and analysis population. They communicate statistical uncertainty, not the range of outcomes that every participant might experience. The registry does not report a prespecified threshold for a clinically important difference in overall survival, so the interval cannot be judged against such a threshold from this record alone.

The posted precision also depends on the available events and follow-up, as well as assumptions of the analysis. The registry gives the primary timeframe as from randomization up to 627 days, but does not report further detail about the maturity of follow-up or the original precision target. The interval is therefore the clearest posted summary of uncertainty for the primary estimate; it should not be treated as proof of either benefit or no effect.

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 section is a planning illustration based on the observed effects, not a report of the trial’s original design calculation. For the time-to-event outcomes, the displayed quantities are total events, not participants per arm. For the other outcomes, the figures are participants per arm. Read the row labels as planning for the stated fraction of the observed effect, and the columns as the number of events or participants associated with 80% or 90% power under the listed assumptions.

In plain terms, 80% power means that a study designed under those assumptions would have an 80% chance of detecting the specified effect at the stated significance level if that effect were the true one. The 90% power column represents a higher chance of detecting that effect, and generally requires a larger study. These are planning concepts, not guarantees about a particular trial.

The illustration assumes a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. For time-to-event outcomes, it uses the Schoenfeld method and reports total events. For the other outcomes, it reports participants per arm. The effect to detect is the stated fraction of the observed effect. Smaller effects require more participants or events to distinguish them from chance variation under the same planning framework.

These figures are not observed power. They should not be read as a retrospective measure of whether the completed study was adequately powered, and they do not reveal the sample-size assumptions used for the trial itself. The original sample-size justification and design assumptions are not reported in the registry record. Use the table and curve to understand how planning requirements change across the illustrated effect sizes, not as a substitute for the trial’s own design documentation.

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

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 1.110, so ln(HR) = 0.1044.
2. d = 4 × (1.96 + 0.842)² / [ln(1.110)]² = 2,882.7, rounded up to 2,883 events in total 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
hr <- 1.11                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(2625623)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 2,883 (80% power), 3,860 (90% power)
 75% of observed effect: events = 5,125 (80% power), 6,861 (90% power)
 50% of observed effect: events = 11,531 (80% power), 15,437 (90% power)
Simulated power with 2,883 events: 0.792 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Analyses Reported by the Sponsor

The sponsor’s reported primary analysis is a Log Rank analysis with a hazard ratio of 1.11, a 95% confidence interval from 0.88 to 1.41, and a p-value of 0.808. Those are also the values used for the posted primary comparison. The registry does not flag a discrepancy between the sponsor’s primary analysis and the recalculated presentation, so the primary interpretation rests on the same reported survival comparison.

For progression-free survival, the registry also reports a Log Rank hazard ratio and p-value. For objective response rate, the sponsor’s analysis uses a Cochran-Mantel-Haenszel odds ratio, whereas the unadjusted recalculation uses a risk difference from the posted group summaries. Those are different effect measures and analysis approaches. Their p-values should not be treated as if they came from the same test, and the sponsor’s analysis is the appropriate account of the sponsor-reported inferential result.

The posted EQ-5D-5L composite-index summary has no sponsor-reported comparative analysis listed. A simple unadjusted recalculation from the group summaries is therefore not a replacement for an analysis specified by the trial. The registry record does not report further model details for that outcome. In general, recalculated summaries help describe the posted data, while the sponsor’s analysis reflects the trial’s reported method where one is available.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study examines several outcomes, time points, or subgroups. Without an appropriate strategy, the chance of observing at least one result that appears statistically notable by chance can rise as more analyses are considered. This matters in JAVELIN Gastric 300 because the registry lists multiple secondary outcomes, including progression-free survival, best overall response, objective response rate, and several quality-of-life measures.

The registry record does not report a multiplicity adjustment, testing hierarchy, or other approach for controlling the interpretation of multiple analyses. It is therefore not possible to determine from this record whether secondary tests were intended to be confirmatory or exploratory, or whether their reported p-values account for the number of outcomes assessed. The primary endpoint remains the clearest stated basis for the trial’s superiority question.

Missing-data handling is also only partly described. For overall survival, the registry says that participants alive at analysis or lost to follow-up were censored at last contact. This describes how their observed follow-up contributes to the survival analysis. The record does not report the broader missing-data procedures, or details such as sensitivity analyses, for the secondary outcomes.

The quality-of-life composite-index groups contain fewer participants than the randomized groups. The registry record does not explain the reasons for the reduced analysis counts or specify whether missing questionnaires were addressed through imputation, complete-case analysis, or another method. Results for several listed quality-of-life outcomes are not reported in the registry record.

The full statistical analysis plan, interim analyses, and the original sample-size justification are also not reported in the registry record. These omissions limit assessment of whether all planned analyses are shown and how robust results might be to alternative assumptions. They do not alter the numerical results posted, but they define important limits on what can be concluded from the public record.

16. 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: disease progression (21.2%), general physical health deterioration (6.0%), abdominal pain (3.3%), vomiting (2.7%), pneumonia (2.7%).

The registry lists serious adverse event counts for 177 participants at risk in the physician-choice chemotherapy plus BSC group and 184 at risk in the avelumab plus BSC group. It reports 81 participants affected in the first group and 90 in the second. These are counts, not a comparative statistical test, and the record excerpt does not provide a confidence interval or adjusted comparison for them.

The record also lists other adverse event counts: 150 among 177 at risk in the physician-choice group and 146 among 184 at risk in the avelumab group. Death counts are listed as 131 and 142, respectively. The registry excerpt does not establish that these deaths were caused by treatment or by adverse events, so the death counts should not be interpreted as treatment-related deaths.

The most frequently listed serious-event terms include disease progression, general physical health deterioration, abdominal pain, vomiting, pneumonia, and anaemia. Their percentages are registry-listed event summaries; the record does not provide enough detail to infer causality or compare risks statistically. These figures describe reported safety information, not a complete benefit-risk assessment.

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

The primary result describes the overall-survival experience observed in the randomized groups during the follow-up represented in the registry. The medians were 5.0 months and 4.6 months, and the reported hazard ratio was 1.11 with a 95% confidence interval from 0.88 to 1.41. The reported p-value was 0.808. Taken together, these findings do not demonstrate the trial’s stated objective of superiority for avelumab plus BSC on overall survival.

That conclusion is narrower than saying the treatments are the same. The confidence interval allows a range of relative effects, and the trial was not described in the registry as an equivalence or non-inferiority study. A non-significant superiority test cannot prove no effect, just as an estimated difference cannot by itself prove a meaningful benefit.

The results also do not establish that either strategy is appropriate for an individual. The registry reports group summaries from a trial population and does not give individualized predictions. The estimates apply to the participants and treatment setting represented in this study, and the posted record does not establish that the same results would apply in other settings or populations.

Secondary outcomes provide a broader description but do not change the primary interpretation. They involve different measures and, in some cases, different analysis counts. The registry does not report results for every listed quality-of-life endpoint or describe a multiplicity strategy. Those limits make it inappropriate to select isolated secondary findings as proof of an overall treatment effect.

In short, the posted evidence does not support a claim of overall-survival superiority for avelumab plus BSC in this trial. It also does not support a definitive claim of equivalence or no effect. The confidence interval, the reported test, the endpoints’ different meanings, and gaps in the registry’s methods description all matter when interpreting what the study does and does not establish.

18. Why This Trial Matters Statistically

This trial illustrates why statistical interpretation should not rely on a single point estimate. The median overall-survival estimates are close, but the hazard-ratio confidence interval is broad enough to include values on either side of 1. The p-value alone does not describe this uncertainty; the estimate and interval are needed to understand what effects remain compatible with the posted data.

It also shows why the effect measure must be matched to the endpoint. A median survival time, hazard ratio, response percentage, odds ratio, and mean quality-of-life change answer different questions. They cannot be compared as if they were all measurements on one scale. Reading the group ordering and analysis method alongside each result avoids misleading comparisons.

The trial’s randomized design supports a comparative interpretation, while its open-label conduct and incomplete public reporting leave questions about assessment and analysis details. The record does not report the original sample-size assumptions, interim analyses, multiplicity approach, or full missing-data methods. These are not minor details when assessing the precision and confirmatory status of secondary findings.

Finally, planning illustrations based on observed effects should not be confused with trial design evidence or observed power. They can show how planning requirements change when a target effect is smaller, but they do not explain the completed study’s original design. Taken together, the record is a useful example of reading results with attention to endpoint, method, uncertainty, and the limits of what the registry documents.

19. 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 provides an incomplete description of some treatment and analysis details. The physician-choice arm includes paclitaxel or irinotecan, but the posted irinotecan regimen description ends before the full regimen is stated. The BSC description is also incomplete. The registry does not report the full protocol or a complete statistical analysis plan.

The study was open-label. The record does not report details of allocation concealment, the randomization process, or any measures used to reduce the possibility that knowledge of assignment affected assessment or care. This matters especially for outcomes involving investigator assessment or participant-reported information, although the registry does not quantify any such effect.

The public record does not report the trial’s original sample-size justification, interim analyses, or multiplicity strategy. It also does not fully describe missing-data methods for secondary outcomes. Some quality-of-life results are based on fewer participants than the randomized groups, and results for several listed quality-of-life outcomes are not reported in the registry record.

Follow-up for the posted outcomes is stated as from randomization up to 627 days, but further detail about follow-up distribution and data maturity is not reported. The safety summary lists counts and selected serious-event terms without a full causal assessment or comparative uncertainty measures in this record. These reporting limits constrain how comprehensively the trial can be appraised from registry results alone.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Avelumab in Third-Line Gastric Cancer (JAVELIN Gastric 300) — NCT02625623.
• Registry record last updated 2020-11-24; status last verified 2020-10.
• Results first posted 2018-10-15.
• 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.

23. Record Summary

JAVELIN Gastric 300 was a completed, randomized, open-label, phase III parallel-group trial with an enrollment of 371. It compared avelumab plus best supportive care with physician-choice chemotherapy plus best supportive care in third-line unresectable, recurrent, locally advanced or metastatic gastric or gastroesophageal junction adenocarcinoma.

The primary endpoint was overall survival, measured from randomization to death from any cause and estimated with the Kaplan-Meier method. The reported hazard ratio was 1.11, with a 95% confidence interval from 0.88 to 1.41 and a Log Rank p-value of 0.808. The posted analysis did not demonstrate superiority on overall survival.

The registry lists secondary survival, response, and quality-of-life outcomes. Some secondary results are posted, while results for several listed measures and important analysis details are not reported in the registry record. The findings should be interpreted with attention to uncertainty, endpoint differences, and those reporting limits.