Diffuse Gastric and Esophagogastric Junction Cancer S-1 Trial: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Diffuse Gastric and Esophagogastric Junction Cancer S-1 Trial (NCT01285557), a randomized phase 3 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

3. Trial Design

This was a randomized, open-label, multicenter, parallel-group phase 3 study sponsored by Taiho Oncology, Inc. It compared S-1 plus cisplatin with 5-FU plus cisplatin in participants with metastatic diffuse gastric cancer, including cancer of the gastro-esophageal junction, previously untreated with chemotherapy. The registry lists enrollment of 361 participants. The study is marked terminated; the reason for termination is not reported in the registry record.

The registry lists the study start as 2011-04-14 and primary completion as 2014-08-15. The overall-survival cutoff is described as 15 August 2014, approximately 40 months. The primary analysis used the ITT population, described as all randomized participants. The arm sizes for that outcome were 239 for S-1 plus cisplatin and 122 for 5-FU plus cisplatin. The registry does not report the original allocation ratio or the rationale for the difference in the recorded arm sizes.

Because the study was open-label, participants and study personnel were not masked to treatment assignment. The registry reports no masking. Open-label status is relevant when considering outcomes that involve clinical judgment or participant reporting; overall survival is based on death and censoring records, although the registry does not describe procedures for outcome assessment. The study included safety outcomes and several time-to-event or response outcomes in addition to overall survival. The full protocol, original sample-size justification, and details of the study's planned statistical analysis 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.

Before assignment: Randomization was stratified by the histologic subtype (adenocarcinoma, diffuse type or signet ring cell adenocarcinoma); extent of metastasis (1 versus more than 1 metastatic site); Eastern Cooperative Oncology Group (ECOG) performance status (0 or 1) and region (North America, Western Europe, Eastern Europe, Rest of world). Cut-off dates for all clinical data collection was 07 March 2014 and for overall survival analysis was 15 August 2014.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Analysis was performed on the ITT population that included all randomized participants.

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: The Intent-to-Treat (ITT) population included all randomized participants including those not dosed according to their randomization.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over From the date of randomization until disease progression or death, cut-off date: 15 August 2014 (approximately 40 months).

Overall survival (OS) measures the time from randomization to death. In the registry definition, participants who did not die were censored at the date they were last known to be alive. The primary analysis was performed in the ITT population, which included all randomized participants. The registry describes the analysis as using the Kaplan-Meier method.

OS is used in trials because it captures the time until death from any cause, rather than relying on a judgment about whether a tumor has progressed or responded. General statistical context: time-to-event methods can include participants with different lengths of follow-up by accounting for censoring, rather than treating every participant as if their outcome were fully observed at a common end date. Censoring does not establish what happened after the last known follow-up; it records that the event time was not observed during the available follow-up.

The registry's primary-outcome timeframe says “until disease progression or death,” while its endpoint definition specifies time from randomization to death. The analysis description defines OS by death, and that definition is the basis for interpreting the reported OS medians and hazard ratio. The registry does not report additional detail resolving the difference in the timeframe wording.

7. Results Posted in the Registry

• S-1+Cisplatin vs 5FU+Cisplatin: hazard ratio 0.99 (95% CI 0.76–1.28), reported p = 0.931 (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:
  • S-1+Cisplatin vs 5FU+Cisplatin: hazard ratio 0.86 (95% CI 0.65–1.14), reported p = 0.304 (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 S-1+Cisplatin vs 5FU+Cisplatin and the control. The interval gives the range of effects that remain compatible with the data.

Time to Treatment Failure (TTF)

• 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:
  • S-1+Cisplatin vs 5FU+Cisplatin: hazard ratio 0.84 (95% CI 0.66–1.08), reported p = 0.168 (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 S-1+Cisplatin vs 5FU+Cisplatin and the control. The interval gives the range of effects that remain compatible with the data.

Number of Participants With TEAEs With Severity Greater Than or Equal to (>=) Grade 3

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • 5FU+Cisplatin: 66.1% (78/118) versus 68.3% (157/230) on control; risk difference -2.2 percentage points (95% CI -12.6 to 8.3), risk ratio 0.97 (0.83–1.13); unadjusted two-proportion z-test p = 0.684.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 5FU+Cisplatin and the control. The interval gives the range of effects that remain compatible with the data.

The posted secondary results broadly align with the uncertainty in the primary OS comparison: the reported PFS and TTF hazard-ratio intervals include 1, and neither secondary comparison provides clear evidence of a difference between treatment groups. They measure outcomes distinct from overall survival, however, and their estimates should not be treated as additional versions of the primary endpoint. Their event definitions and analysis cutoffs differ from the OS endpoint.

The registry lists additional secondary outcomes, including response rate, duration of response, time to tumor response, and treatment-emergent adverse events. Results for several of these listed outcomes are not reported in the registry record. Counts are posted for participants with TEAEs of Grade 3 or higher, but no sponsor-reported comparative analysis is recorded for that outcome. This means that the broad secondary pattern is only partly described by the posted numerical results.

Secondary outcomes generally carry less weight than the primary endpoint in interpretation because they are additional questions, often with different definitions and analyses. General statistical context: testing multiple outcomes creates more opportunities for apparently notable findings to occur by chance unless the analysis plan addresses multiplicity. The registry does not report whether a multiplicity adjustment was planned or applied. Accordingly, the secondary results are best treated as supportive descriptive context, not as a way to override the primary OS comparison. The registry also does not report whether secondary analyses were prespecified in a detailed statistical analysis plan.

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

• Progression-free Survival (PFS)
• Time to Treatment Failure (TTF)
• Number of Participants With Treatment-emergent Adverse Events (TEAEs) and Treatment-emergent Serious Adverse Events (TESAE)
• Number of Participants With TEAEs With Severity Greater Than or Equal to (>=) Grade 3
• Overall Response Rate (ORR): Percentage of Participants With Overall Response
• Duration of Response (DR)
• Time to Tumor Response (TTR)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is a time-to-event outcome, and the registry states that overall survival was analyzed with the Kaplan-Meier method. The reported comparison between treatment groups used an unstratified log-rank test and is expressed as a hazard ratio. General statistical context: these methods are suited to outcomes where participants may have different follow-up times and some may be censored before an event is observed.

What does the Kaplan-Meier median describe?

The Kaplan-Meier method estimates the probability of remaining event-free over follow-up while accounting for censoring. For overall survival, the event is death. The median is the estimated time by which half of the population is estimated to have experienced the event. It is not the arithmetic average of participants' observed follow-up times, and it does not mean that each participant experiences the event at that time. The median estimates reported for the two groups are 7.5 months and 6.6 months, each with its own 95% confidence interval.

Censoring means that a participant's event time is not observed within the available follow-up; for OS, the registry says participants not known to have died were censored at the date last known to be alive. General statistical context: valid interpretation relies on appropriate censoring and follow-up information. Further details about follow-up completeness and missing outcome data are not reported in the registry record.

How should the hazard ratio be read?

The sponsor-reported hazard ratio for S-1 plus cisplatin versus 5-FU plus cisplatin is 0.99. General statistical context: a hazard ratio compares the event rates over time between groups, conditional on participants remaining at risk. A value near 1 indicates that the estimated relative event rates are similar in the fitted comparison. It is not a ratio of the two median survival times, and it does not directly state how many months a participant might gain or lose.

The 95% confidence interval for the OS hazard ratio is 0.76 to 1.28, and the reported p-value is 0.931. The interval includes 1, the value corresponding to no relative difference in hazards. The unstratified log-rank method is the reported comparison; stratification factors or covariate adjustments are not reported in the registry record. General statistical context: a log-rank comparison evaluates evidence of differences in event-time distributions across groups, while a hazard ratio gives a relative-effect summary. The registry does not report additional modeling assumptions or diagnostics.

What do the secondary time-to-event analyses add?

Progression-free survival (PFS) and time to treatment failure (TTF) are also reported as median time-to-event outcomes. Their comparisons use log-rank methods and report hazard ratios. The registry gives a PFS hazard ratio of 0.86, with a 95% confidence interval from 0.65 to 1.14 and p = 0.304. For TTF, the reported hazard ratio is 0.84, with a 95% confidence interval from 0.66 to 1.08 and p = 0.168. Both intervals include 1. These estimates concern their own endpoint definitions and follow-up cutoffs, not OS alone.

General statistical context: a progression-free endpoint typically concerns progression or death, whereas a treatment-failure endpoint depends on its specified treatment-failure definition. The registry gives the PFS timeframe as randomization until disease progression or death, and the TTF timeframe as randomization until disease progression. The registry does not provide further detail about event adjudication or the handling of other reasons for stopping treatment.

How should the binary safety comparison be understood?

The registry posts counts of participants with TEAEs of Grade 3 or higher: 157 of 230 for S-1 plus cisplatin and 78 of 118 for 5-FU plus cisplatin. The accompanying comparison is an unadjusted recalculation from these summary counts, not a sponsor-reported analysis. It gives a risk difference of -0.022, with an interval from -0.126 to 0.083, and an unadjusted p-value of 0.684. General statistical context: a risk difference expresses the difference between group event proportions on the probability scale; an interval spanning zero does not establish a clear difference. The registry does not report a sponsor analysis for this endpoint.

The reported analyses should be distinguished from planning illustrations. The hazard ratios and p-values for OS, PFS, and TTF are sponsor-reported results. The binary comparison is an unadjusted recalculation based on posted counts. General statistical context: a recalculation from summary data can be useful for orientation but does not replace an analysis specified or conducted by the sponsor. The registry does not report enough detail to assess whether the recalculated binary method matches any prespecified analysis.

11. Confidence Intervals and Precision

Confidence intervals communicate the range of estimates compatible with the data under the analysis and its assumptions. They are not guarantees that the true value lies inside the interval, and their width indicates how precisely an effect has been estimated. General statistical context: a 95% confidence interval is generated by a procedure intended to cover the true parameter in 95% of repeated samples under its assumptions; it is not a statement that there is a 95% probability that this particular interval contains the true value.

For OS, the hazard-ratio estimate is 0.99 and its 95% confidence interval is 0.76 to 1.28. This interval includes 1 and allows for relative effects on either side of no difference. The reported p-value, 0.931, is also consistent with little evidence against the no-difference reference under the reported test. Neither statistic proves equivalence. The registry does not report an equivalence or non-inferiority design or margin.

The arm-specific OS median confidence intervals are 6.7 to 9.3 months for S-1 plus cisplatin and 5.7 to 8.1 months for 5-FU plus cisplatin. These intervals quantify precision for each median separately; overlap or non-overlap of arm-specific intervals is not a substitute for the reported comparative test. For PFS and TTF, the hazard-ratio intervals are also broad enough to include 1: 0.65 to 1.14 for PFS and 0.66 to 1.08 for TTF. Those outcomes therefore do not yield a clear between-group difference in the reported comparisons.

The binary recalculation for Grade 3 or higher TEAEs has a risk-difference interval from -0.126 to 0.083. It spans zero, so the posted summary counts do not give a precise directional distinction under that unadjusted comparison. The registry does not report confidence intervals for every listed safety outcome. Precision also depends on follow-up and the number of observed events; event counts for the primary OS comparison are not reported in the registry record.

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 figures are planning illustrations based on the observed effect, not the trial's original design calculation. The original sample-size justification and design assumptions are not reported in the registry record. The illustrations assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They consider detecting the stated fraction of the observed effect. For time-to-event outcomes, the figures are total events required. For the binary outcome, the figures are participants per arm for 80% power or 90% power.

In plain language, the 80% power column gives an illustrative number of events or participants under the stated assumptions for a design with an 80% chance of detecting the specified effect if that effect were the true one. The 90% power column uses a higher chance of detection under those same assumptions and therefore requires a larger planning target. These are design-planning concepts; the figures are not observed power.

The rows use 100%, 75%, and 50% of the observed effect. A smaller effect is harder to distinguish from random variation, so the illustrations require more events or participants to detect it at the stated power. The time-to-event totals should not be mistaken for people per treatment arm, while the binary figures are participants per arm and not a combined total. These calculations do not establish what enrollment the trial needed or was intended to achieve, and they do not explain the recorded enrollment of 361. The assumptions used by the original trial designers are not reported in the registry record.

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 = 0.990, so ln(HR) = -0.0101.
2. d = 4 × (1.96 + 0.842)² / [ln(0.990)]² = 310,818.3, rounded up to 310,819 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 <- 0.99                              # 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(1285557)
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 = 310,819 (80% power), 416,098 (90% power)
 75% of observed effect: events = 552,566 (80% power), 739,729 (90% power)
 50% of observed effect: events = 1,243,274 (80% power), 1,664,390 (90% power)
Simulated power with 310,819 events: 0.798 (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. Group-Sequential Design Analysis

Many time-to-event trials, especially in oncology, stop or report early when interim data are overwhelming, so a single fixed-sample number understates the real design problem. The calculation below is an illustrative group-sequential design built from this trial's posted control median and hazard ratio. The registry record does not state the planned number of looks or the spending function, so the design is a planning illustration, not a reconstruction of the protocol. Assumed: one-sided α = 0.025, 90% power, three equally spaced looks on information, Lan-DeMets O'Brien-Fleming spending, 24 months of accrual, 60 months in total and a 1% monthly dropout hazard.

Under these assumptions the design needs about 463,862 participants in total, with the boundaries below.

Fixed design versus group-sequential alternatives

Group-sequential designs cost a small increase in maximum sample size and events compared with the fixed design, in exchange for the chance to stop early for overwhelming efficacy. O'Brien-Fleming spending keeps early boundaries very conservative and the final boundary close to the fixed-design value, whereas Pocock spending spends more alpha early.

Sensitivity to a smaller true effect

Planning for exactly the observed effect is optimistic, because observed effects in published trials tend to be somewhat larger than the true effect. The table repeats the main design (three looks, O'Brien-Fleming spending, 90% power) for smaller true hazard ratios.

For context, this trial enrolled 361 participants.

Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993. Agreement between the two packages confirms the boundary calculation.

R code and output

suppressWarnings(suppressMessages({ library(gsDesign); library(rpact) }))
med_control <- 6.6          # control-arm median, months (converted from the posted unit)
hr          <- 0.99          # hazard ratio used for planning
lambdaC     <- log(2) / med_control
eta         <- 0.01               # dropout hazard per month (assumption)
acc_months  <- 24                 # accrual duration (assumption)
study_months<- 60                 # total study duration (assumption)

design <- function(k, sfu, beta, h = hr) gsSurv(k = k, test.type = 1, alpha = 0.025, beta = beta, sfu = sfu,
  lambdaC = lambdaC, hr = h, hr0 = 1, eta = eta, gamma = 1, R = acc_months, T = study_months, minfup = study_months - acc_months)

# Main design: 3 looks, Lan-DeMets O'Brien-Fleming spending, one-sided alpha 0.025, 90% power
x <- design(3, sfLDOF, 0.10)
cat(sprintf("N_TOTAL|%s\n", format(ceiling(sum(x$gamma * x$R)), big.mark = ",")))
for (i in 1:3) cat(sprintf("GS|%d|%.1f|%.0f|%.3f|%.5f|%.5f\n", i, x$T[i], x$n.I[i], x$upper$bound[i], pnorm(-x$upper$bound[i]), cumsum(x$upper$spend)[i]))
cat("\n"); print(gsBoundSummary(x))

# Comparison of designs
cmp <- list(list("Fixed design, one analysis", 1, sfLDOF), list("Two looks, O'Brien-Fleming", 2, sfLDOF),
            list("Three looks, O'Brien-Fleming", 3, sfLDOF), list("Three looks, Pocock", 3, sfLDPocock))
cat("\n")
for (b in c(0.20, 0.10)) for (d in cmp) {
  y <- design(d[[2]], d[[3]], b)
  cat(sprintf("DES|%s|%d|%.0f|%s|%.1f\n", d[[1]], round(100 * (1 - b)), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ","), tail(y$T, 1)))
}

# Sensitivity: what if the true effect is smaller than the observed one? (3 looks, O'Brien-Fleming, 90% power)
cat("\n")
for (fr in c(1, 0.75, 0.5)) {
  y <- design(3, sfLDOF, 0.10, exp(log(hr) * fr))
  cat(sprintf("SENS|%.2f|%.3f|%.0f|%s\n", fr, exp(log(hr) * fr), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ",")))
}

# Independent cross-check of the boundaries with rpact
r <- getDesignGroupSequential(kMax = 3, alpha = 0.025, beta = 0.10, sided = 1, typeOfDesign = "asOF", informationRates = x$timing)
cat(sprintf("\nCross-check, upper boundaries: gsDesign %s | rpact %s\n", paste(round(x$upper$bound, 3), collapse = ", "), paste(round(r$criticalValues, 3), collapse = ", ")))

Method: LachinFoulkes 
       Analysis               Value Efficacy
      IA 1: 33%                   Z   3.7103
      N: 292778         p (1-sided)   0.0001
 Events: 140337        ~HR at bound   0.9804
      Month: 15    P(Cross) if HR=1   0.0001
                P(Cross) if HR=0.99   0.0338
      IA 2: 67%                   Z   2.5114
      N: 463862         p (1-sided)   0.0060
 Events: 280673        ~HR at bound   0.9906
      Month: 24    P(Cross) if HR=1   0.0060
                P(Cross) if HR=0.99   0.5603
          Final                   Z   1.9930
      N: 463862         p (1-sided)   0.0231
 Events: 421010        ~HR at bound   0.9939
      Month: 60    P(Cross) if HR=1   0.0250
                P(Cross) if HR=0.99   0.9000



Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993

15. Analyses Reported by the Sponsor

For OS, the sponsor-reported analysis is an unstratified log-rank comparison with a hazard ratio of 0.99, a 95% confidence interval from 0.76 to 1.28, and p = 0.931. The corresponding arm-specific median estimates are 7.5 months and 6.6 months. The reported hazard ratio and its uncertainty are the appropriate primary comparative results to emphasize; comparing medians alone would not reproduce the sponsor's time-to-event analysis.

The registry also reports sponsor analyses for PFS and TTF using log-rank methods. Their reported hazard ratios are 0.86 and 0.84, respectively, with confidence intervals that include 1. The recalculated binary safety comparison for Grade 3 or higher TEAEs is different in kind: it is an unadjusted analysis derived from posted group counts, and the registry does not report a sponsor analysis for that outcome. It should not be presented as the sponsor's result.

The registry record reports no discrepancy between the sponsor's analysis and the posted recalculation. The sponsor's reported time-to-event analyses therefore remain the primary source for those comparisons. The recalculated binary estimate is descriptive support for reading the posted counts, not evidence that the sponsor conducted the same test or used the same analysis plan.

16. Multiplicity, Missing Data and Analysis Populations

The registry record identifies overall survival as the primary outcome and lists multiple secondary outcomes, including PFS, TTF, response outcomes, and safety outcomes. It reports results for some of these outcomes, but does not report whether multiplicity was addressed through a hierarchy, adjustment, or another prespecified strategy. The statistical analysis plan and the original protocol are not reported in the registry record.

General statistical context: when several outcomes or comparisons are tested, the chance of at least one apparently unusual result can increase. Adjustments or prespecified testing sequences can help control that issue, but the registry record does not establish whether either approach was used. Unadjusted secondary p-values should therefore be interpreted cautiously and should not be treated as independent confirmation of the primary result.

The primary OS population is specified as the ITT population, including all randomized participants. The registry says that participants who did not die were censored at the date last known to be alive. Further information about missing observations, loss to follow-up, sensitivity analyses, or the completeness of vital-status ascertainment is not reported in the registry record. Analysis populations for the secondary outcomes are not reported in the registry record.

The registry reports an unstratified log-rank method for OS and log-rank methods for PFS and TTF. It does not report interim analyses, stopping rules, or whether any analyses were conducted before the final reported cutoff. The reason the study is marked terminated is not reported in the registry record. No interpretation of the termination or of unreported analysis procedures should be inferred from the available record.

17. 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: anaemia (4.8%), gastric haemorrhage (3.4%), vomiting (3.4%), dehydration (3.4%), hypokalaemia (3.4%).

The registry posts serious adverse-event counts for 63 of 230 participants at risk in the S-1 plus cisplatin group and 31 of 118 at risk in the 5-FU plus cisplatin group. The listed serious-event terms and reported percentages include anaemia at 4.8%, and gastric haemorrhage, vomiting, dehydration, and hypokalaemia each at 3.4%; renal failure acute is listed at 2.2%. These figures describe reported serious events and should not be confused with the Grade 3 or higher TEAE counts, which are a separate outcome.

For the Grade 3 or higher TEAE outcome, the posted counts are 157 of 230 for S-1 plus cisplatin and 78 of 118 for 5-FU plus cisplatin. The registry timeframe runs from the first dose through 30 days after the last study medication, with a maximum duration of 35.7 months. These are descriptive counts. The registry does not report a sponsor comparative analysis for this endpoint, and the unadjusted recalculation should not be read as a comprehensive safety assessment.

General statistical context: event counts do not alone describe duration, recurrence, severity distribution, or causal attribution. The registry's listed serious-event terms are not a complete account of all possible harms. Further safety details and any planned comparisons are not reported in the registry record.

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

The posted primary analysis estimates overall survival in the ITT population and compares the randomized treatment groups. The median estimates are 7.5 months for S-1 plus cisplatin and 6.6 months for 5-FU plus cisplatin. The sponsor-reported hazard ratio is 0.99, with a 95% confidence interval from 0.76 to 1.28 and p = 0.931. These results do not establish a clear difference in overall survival between the groups.

That finding is not proof that the treatments are equivalent or that their effects are exactly the same. The confidence interval includes 1 and also allows for possible effects in either direction. The registry does not report an equivalence margin, a non-inferiority margin, or an analysis designed to demonstrate equivalence. The result should therefore be read as an imprecise estimate that does not show a clear difference under the reported analysis, rather than as evidence of identical outcomes.

The result also does not determine what would happen to an individual participant, nor does it establish the comparative effects in patient subgroups. The registry does not report subgroup analyses. The secondary time-to-event estimates address other outcomes and likewise do not provide a clear between-group distinction in their posted comparisons. Safety counts describe recorded events but do not provide a complete account of all possible adverse effects or a definitive comparative safety conclusion.

Finally, the registry marks the study as terminated, but does not report the reason. It would be inappropriate to attribute that status to the survival result or to any other specific cause. The interpretation is limited to the posted trial design, endpoint definitions, analyses, and follow-up information.

19. Why This Trial Matters Statistically

This trial illustrates why a time-to-event comparison should be read through both its effect estimate and its uncertainty. The OS medians give a useful description of each group's estimated survival distribution, while the hazard ratio and its confidence interval provide the sponsor-reported comparison over follow-up. The median estimates alone do not establish a treatment difference, and the comparative hazard ratio should not be mistaken for a ratio of medians.

The result also highlights the distinction between absence of clear evidence for a difference and evidence of equivalence. The OS hazard-ratio interval crosses 1, and the reported p-value is 0.931; these do not prove that the groups have identical outcomes. Without a reported equivalence or non-inferiority design and margin, equivalence is not the question answered by this result.

The record gives a randomized, open-label, multicenter phase 3 design and an ITT primary population, but does not report the original sample-size justification, detailed statistical plan, multiplicity strategy, or interim-analysis procedures. Those omissions matter when assessing the intended sensitivity of the study and the status of its secondary analyses. The planning illustrations based on observed effects cannot substitute for the trial's original design rationale.

Finally, the combination of a primary survival endpoint and several secondary outcomes shows the importance of distinguishing confirmatory evidence from supportive description. The reported secondary comparisons do not supply a clear contrasting signal, while several listed outcomes lack posted results. Statistically, the record supports a careful, bounded reading: characterize the estimates and their precision, distinguish sponsor analyses from recalculations, and avoid conclusions that the registry does not establish.

20. 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 marks the study as terminated, but the reason is not reported in the registry record. The record provides a primary completion date and an OS cutoff, but detailed follow-up completeness and the number of deaths in each group are not reported. The primary-outcome timeframe wording refers to progression or death, whereas the endpoint definition specifies death; the analysis description is consistent with OS defined by death, but the registry does not explain the wording difference.

The trial was open-label. The registry does not describe safeguards for outcome assessment or provide further detail on the assessment of progression and response. This is particularly relevant to outcomes involving clinical assessment, although OS is defined by death. The registry also does not provide the full protocol, detailed statistical analysis plan, original sample-size justification, interim-analysis plan, or multiplicity strategy.

Some listed secondary outcomes do not have results reported in the registry record. The posted safety counts cover specified groups and timeframes; they do not alone capture all dimensions of safety or establish causality. The registry does not report sponsor comparative analyses for the Grade 3 or higher TEAE counts. Any unadjusted recalculation from posted summary counts is consequently limited to that summary information.

The record does not report subgroup analyses or detailed sensitivity analyses for OS. It also does not report further details about missing data beyond the stated OS censoring rule. These limitations constrain interpretation of the precision, robustness, and generalizability of the posted findings.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Diffuse Gastric and Esophagogastric Junction Cancer S-1 Trial — NCT01285557.
• Registry record last updated 2024-09-19; status last verified 2024-08.
• Results first posted 2022-05-17.
• 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.

24. Record Summary

NCT01285557 is a randomized, open-label, multicenter, parallel-group phase 3 study of S-1 plus cisplatin compared with 5-FU plus cisplatin in previously untreated metastatic diffuse gastric cancer, including gastro-esophageal junction cancer. The registry lists 361 participants enrolled and marks the study as terminated. The reason for termination is not reported in the registry record.

Overall survival was the primary endpoint, analyzed in the ITT population. The sponsor-reported unstratified log-rank comparison gave a hazard ratio of 0.99, a 95% confidence interval from 0.76 to 1.28, and p = 0.931. The posted result does not establish a clear difference in OS between the randomized groups. The registry also posts PFS and TTF comparisons and selected safety counts; several listed secondary outcomes have no results reported in the registry record.

The registry does not report the original sample-size justification, a detailed statistical analysis plan, multiplicity procedures, or interim analyses. The posted planning illustrations are not the trial's original design rationale.