S1201: Combination Chemo for Patients W/Advanced or Metastatic Esophageal, Gastric, or Gastroesophageal Junction Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for S1201: Combination Chemo for Patients W/Advanced or Metastatic Esophageal, Gastric, or Gastroesophageal Junction Cancer (NCT01498289), a randomized phase 2 study. The primary outcome is Progression-free Survival (PFS) in High-ERCC1 Patients.

2. Clinical Question

The trial asked whether outcomes differed between two combination chemotherapy regimens for people with advanced or metastatic esophageal, gastric, or gastroesophageal junction cancer. Its primary endpoint focused more narrowly on progression-free survival among eligible and analyzable participants with ERCC1 level >= 1.7. The registry describes the study as prospectively evaluating treatment based on ERCC1 and identifies the primary outcome as progression-free survival in the high-ERCC1 group.

The primary result is a comparison of time-to-event outcomes, reported as a hazard ratio with a confidence interval and p-value. The registry also lists overall response rate, variation in progression-free survival by ERCC1, and grade 3 through 5 adverse events related to study drugs as secondary outcomes. Numerical results for those secondary outcomes are not reported in the registry record.

3. Trial Design

This was a randomized, open-label, parallel-group phase II trial sponsored by SWOG Cancer Research Network. The registry lists an enrollment of 213 participants and marks the study completed. The recorded study dates are 2012-02 for start and 2018-03 for primary completion.

Participants were assigned to one of two regimens. Arm I received oxaliplatin and leucovorin calcium by intravenous infusion over 2 hours on day 1, followed by fluorouracil by intravenous infusion over 46-48 hours on days 1-2. Courses repeated every 14 days in the absence of disease progression or unacceptable toxicity. Arm II received irinotecan hydrochloride over 90 minutes and docetaxel over 30-60 minutes on days 1 and 8; courses repeated every 21 days in the absence of disease progression or unacceptable toxicity.

The primary endpoint was progression-free survival in eligible and analyzable participants with ERCC1 level >= 1.7, with the timeframe stated as up to 3 years after registration. The overall enrollment and the primary analysis population are different quantities: the posted primary result includes 13 participants in Arm I and 15 in Arm II. The registry does not explain the reasons for this difference, so exclusions, eligibility failures, or missing outcome data should not be inferred from the counts alone.

Randomization and a parallel-group structure support a comparison between the assigned regimens, while the lack of masking means participants and study personnel were not masked to treatment assignment. The registry record does not report the randomization procedure, allocation ratio, stratification factors, or detailed eligibility criteria. It also does not report the trial's original sample-size justification or the assumptions used in its design.

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: 213 participants were randomized, but six were ineligible and 5 were excluded from analysis due to death or withdrawal prior to randomization. Therefore 202 were deemed eligible and analyzable for the primary analysis.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Eligible and analyzable participants with ERCC1 level \>= 1.7

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: 213 participants were randomized, but six were ineligible and 5 were excluded from analysis due to death or withdrawal prior to randomization. Therefore 202 were deemed eligible and analyzable for the primary analysis.

6. Primary Endpoint

The registry lists the primary outcome as Progression-free Survival (PFS) in High-ERCC1 Patients, measured over Up to 3 years after registration.

Progression-free survival is defined in the registry as the time from protocol registration until disease progression or death, whichever occurs first. It is a time-to-event endpoint: each participant contributes information about how long they remain alive without recorded progression, subject to the trial's endpoint and follow-up procedures.

In general, a time-to-event endpoint can incorporate both whether an event occurred and when it occurred. It can also account for participants whose event time is not observed during the available follow-up, using censoring methods when those methods are specified and applied. The registry record does not describe the censoring rules or how missing outcome information was handled.

The primary endpoint is reported as a median in months for each arm, together with a 95% confidence interval. Arm I FOLFOX had a reported median of 4.7 months, with a 95% confidence interval from 2.1 to 8.7 months. Arm II IT had a reported median of 5.3 months, with a 95% confidence interval from 0.9 to 11.5 months. The comparison between arms is reported separately as a Cox hazard ratio.

Using progression-free survival allows the trial to assess the timing of progression or death rather than only recording whether a participant had an event by one fixed time. The endpoint definition combines progression and death, so it does not by itself distinguish which type of event occurred. The registry record does not give separate counts or timing for progression and death within this primary analysis.

7. Results Posted in the Registry

• Arm I FOLFOX vs Arm II IT: hazard ratio 0.91 (95% CI 0.41–2.05), reported p = 0.830 (as reported in the registry).

8. Statistical Interpretation

9. Statistical Methods Explained

The primary comparison uses a Cox proportional hazards regression analysis. In general, Cox regression is a method for comparing time-to-event outcomes while accounting for the timing of events and, under the analysis approach, participants whose event time is censored. It estimates a hazard ratio, which compares the instantaneous event rates between groups among participants who have not yet had the event. The registry reports a hazard ratio of 0.91 for Arm I FOLFOX versus Arm II IT, with a 95% confidence interval from 0.41 to 2.05 and p = 0.83.

What does a hazard ratio represent?

In general, the hazard is the rate at which the event occurs at a particular time among participants still at risk. A hazard ratio compares that rate between groups over the follow-up period, under the model's assumptions. A value of 1 corresponds to equal hazards; values below or above 1 indicate a relative difference according to which arm is in the numerator and which is the reference. The registry gives the comparison order but does not provide further detail about model coding or covariate adjustment. The reported estimate should therefore be described as the registry's comparison, not reinterpreted using an unstated model specification.

The Cox model is commonly associated with a proportional-hazards assumption: the relative hazard between groups is treated as constant over time. This is a general feature of the method, not a claim that the assumption was tested or met in this trial. Whether proportional hazards were assessed, and what diagnostics were used, are not reported in the registry record. If the relative hazard changes over time, one hazard ratio may not capture the full pattern of group differences.

How do the medians relate to the hazard ratio?

The registry reports median progression-free survival separately for each arm and a Cox hazard ratio for the comparison. These statistics answer related but distinct questions. A median identifies a point in the time-to-event distribution, while the hazard ratio summarizes a relative event-rate comparison over time under the Cox model. In general, the two arm medians alone are not enough to reconstruct a hazard ratio or the full pattern of event times. The registry's reported comparison is therefore the appropriate source for the hazard ratio rather than a calculation from the medians.

In general, confidence intervals express uncertainty around an estimate under the statistical model and method used. The reported interval for the hazard ratio, 0.41 to 2.05, includes 1 and is wide. The corresponding p-value, 0.83, is the registry's reported result for the Cox comparison. A p-value is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. It describes how compatible the observed data are with a specified null model under the test's assumptions.

What is the role of the primary analysis population?

The primary endpoint is defined for eligible and analyzable participants with ERCC1 level >= 1.7. The arm summaries list 13 participants in Arm I and 15 in Arm II. In general, limiting an analysis to a subgroup can make its result relevant specifically to that subgroup, but the result should not be automatically generalized to participants outside it. The registry does not report the full analysis-population algorithm, the count screened or excluded, or how missing ERCC1 measurements were treated.

Randomization is intended in general to support comparable treatment groups at the point of assignment. However, the registry record does not provide the randomization method, allocation ratio, or any stratification variables. It also does not describe whether the Cox regression was adjusted for baseline factors. Those details cannot be assumed from the presence of randomization or the name of the regression method.

For time-to-event analyses, the treatment of censoring and follow-up is important because event times may not be observed for every participant. General Cox methods rely on assumptions about censoring and model structure. The endpoint timeframe is listed as up to 3 years after registration, but the registry record does not specify censoring rules, follow-up completeness, the proportional-hazards assessment, or sensitivity analyses. These omissions limit how fully the model can be evaluated from the reported summary.

The registry identifies three secondary outcomes: overall response rate, progression-free survival variation by ERCC1, and grade 3 through 5 adverse events related to study drugs. It lists six outcome measures and eight analyses overall. The detailed numerical results and the relationship between all planned and reported analyses are not included in the registry record for this account. The statistical analysis plan, including any multiplicity procedure, is not reported in the registry record. Consequently, the available primary p-value should not be treated as if it accounts for every possible outcome or analysis.

10. Confidence Intervals and Precision

The 95% confidence interval for the primary hazard ratio is 0.41 to 2.05, around the reported estimate of 0.91. In general, a confidence interval gives a range of estimates compatible with the data and the statistical method at the stated confidence level. It conveys uncertainty in the estimate; it is not a range containing a known, fixed percentage of individual participant outcomes.

This interval is broad and crosses 1, the value corresponding to equal hazards in the general interpretation of a hazard ratio. It therefore leaves substantial uncertainty about the direction and magnitude of the relative hazard. The point estimate alone should not be read as a precise estimate of a treatment difference. The reported p-value is 0.83, and the interval and p-value together do not provide evidence of a clear difference in this primary comparison.

The arm-specific median estimates are also uncertain. The 4.7-month median in Arm I has a 95% confidence interval from 2.1 to 8.7 months; the 5.3-month median in Arm II has a 95% confidence interval from 0.9 to 11.5 months. These intervals indicate that the medians are not precisely estimated in the analyzed groups. In general, a median confidence interval concerns uncertainty about the population median, not the range of times experienced by individual participants.

The primary result is based on 13 participants in Arm I and 15 in Arm II. The registry enrollment is 213, but that larger number is not the sample size for this high-ERCC1 primary analysis. The registry does not report event counts for the primary comparison, so the amount of event information behind the hazard ratio cannot be assessed from the stated arm counts alone. It also does not report the censoring distribution or the number of participants still under observation at particular follow-up times.

Precision is therefore limited both by the breadth of the reported intervals and by the small subgroup counts in the primary result. The registry record does not report further uncertainty analyses or model diagnostics that could clarify the stability of the estimates.

11. 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 planning figures are illustrations based on the observed hazard-ratio effect, not a report of the trial's original design calculation. Because the outcome is time-to-event, the figures are stated as total events across the trial, not participants per arm. The calculation assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each scenario considers detecting the stated fraction of the observed effect.

The 80% power column means the total event count estimated to give an 80% chance of detecting the specified effect under those planning assumptions, if that effect is present and the assumptions hold. The 90% power column expresses the corresponding planning target for a 90% chance. These probabilities describe the operating characteristics of a hypothetical design under its assumptions; they are not observed power.

For 100% of the observed effect, the planning figures are 3,530 total events for 80% power and 4,726 total events for 90% power. For 75% of the observed effect, they are 6,276 and 8,401 total events, respectively. For 50% of the observed effect, they are 14,120 and 18,902 total events, respectively. Smaller effects require more event information, so the estimated event totals rise as the target effect becomes smaller.

These figures should not be mistaken for the trial's enrollment or for observed event counts. The registry's enrollment is 213, while the planning illustrations use total events as their unit. The trial's own original sample-size assumptions and justification are not reported in the registry record. The illustrations also do not establish that the trial was designed to detect the observed effect or any of the listed fractions.

12. 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 = 0.910, so ln(HR) = -0.0943.
2. d = 4 × (1.96 + 0.842)² / [ln(0.910)]² = 3,529.8, rounded up to 3,530 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 <- 0.91                              # 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(1498289)
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 = 3,530 (80% power), 4,726 (90% power)
 75% of observed effect: events = 6,276 (80% power), 8,401 (90% power)
 50% of observed effect: events = 14,120 (80% power), 18,902 (90% power)
Simulated power with 3,530 events: 0.794 (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.

13. 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 5,128 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 213 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 <- 4.7          # control-arm median, months (converted from the posted unit)
hr          <- 0.91          # 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: 2948         p (1-sided)   0.0001
 Events: 1594        ~HR at bound   0.8304
    Month: 14    P(Cross) if HR=1   0.0001
              P(Cross) if HR=0.91   0.0338
    IA 2: 67%                   Z   2.5114
      N: 4784         p (1-sided)   0.0060
 Events: 3188        ~HR at bound   0.9149
    Month: 22    P(Cross) if HR=1   0.0060
              P(Cross) if HR=0.91   0.5603
        Final                   Z   1.9930
      N: 5128         p (1-sided)   0.0231
 Events: 4781        ~HR at bound   0.9440
    Month: 60    P(Cross) if HR=1   0.0250
              P(Cross) if HR=0.91   0.9000



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

14. Analyses Reported by the Sponsor

The registry reports the sponsor's primary comparison as a Cox proportional hazards result: hazard ratio 0.91 for Arm I FOLFOX versus Arm II IT, 95% confidence interval 0.41 to 2.05, and p = 0.83. The posted comparison and the analysis summary give the same estimate, interval, and p-value; no discrepancy is indicated between them.

The reported sponsor analysis is the primary basis for interpreting the hazard ratio. The arm-specific medians are descriptive summaries and are not a substitute for the Cox comparison. In general, recalculating a time-to-event comparison from arm-level medians alone would not reproduce the Cox analysis, because the regression uses event-time information and the model's analysis procedures.

The registry does not report enough detail to reproduce the Cox model independently, including its full specification, covariate handling, censoring rules, or diagnostics. The results should therefore be read as the sponsor's reported analysis, with its uncertainty represented by the stated confidence interval. That interval is broad and crosses 1, so the result does not establish a clear difference between the arms.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists several outcomes beyond the primary progression-free survival endpoint: overall response rate, progression-free survival variation by ERCC1, and grade 3 through 5 adverse events related to study drugs. It also records six outcome measures and eight analyses. The registry record does not report whether the analyses were designated as confirmatory or exploratory, or how the outcomes and analyses were related to a prespecified testing sequence.

In general, testing multiple outcomes or conducting multiple analyses can increase the chance of obtaining a nominally small p-value by chance. Statistical plans may address this through adjustments, a hierarchy of tests, or other procedures. Whether any such approach was used for this trial is not reported in the registry record. The p-value for the primary Cox comparison should therefore be understood as reported, without assuming an adjustment for the wider set of outcomes or analyses.

The registry identifies the primary population as eligible and analyzable participants with ERCC1 level >= 1.7 and gives arm counts of 13 and 15. The specific rules for eligibility, analyzability, and inclusion in the analysis are not reported in the registry record. The record also does not give counts screened, excluded, or lacking a usable ERCC1 result, nor does it describe how participants were handled if their endpoint data were incomplete.

Missing-data methods matter for both subgroup membership and time-to-event outcomes. In general, censoring rules determine how participants without an observed event are represented, while missing-data procedures address information that is unavailable or unusable. The registry record does not report the censoring rules, reasons for censoring, missing-data methods, or sensitivity analyses. It is not possible to determine from the reported summaries whether missingness could have affected the estimates.

The registry does not state whether the primary analysis followed an intention-to-treat population, a per-protocol population, or another analysis set. It also does not report a detailed statistical analysis plan, interim analyses, or procedures for handling deviations from assigned treatment. These details should not be inferred from the randomized design alone.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: sepsis (6.1%), neoplasms benign, malignant and unspecified - other (5.1%), death nos (4.1%), multi-organ failure (2.0%), respiratory failure (2.0%).

The adverse-event summary lists 91 participants at risk in Arm I FOLFOX, of whom 7 were affected by the serious adverse-event category recorded in the summary. For Arm II IT, 98 participants were at risk and 17 were affected. The same summary lists other adverse events for 88 of 91 participants in Arm I and 95 of 98 in Arm II. It records deaths of 89 in Arm I and 94 in Arm II.

The registry also lists serious adverse-event terms with percentages: sepsis, 6.1%; neoplasms benign, malignant and unspecified - Other, 5.1%; death NOS, 4.1%; multi-organ failure, 2.0%; respiratory failure, 2.0%; and oral mucositis, 1.1%. The listed terms do not establish, by themselves, how many participants experienced each term, whether a participant could appear under multiple terms, or how the percentages were calculated. Arm-specific attribution for these term percentages is not reported in the registry record.

These figures describe adverse-event reporting and are not a direct comparison of treatment benefit. The registry record does not provide further detail about event timing, duration, severity distribution, or the context of the recorded deaths. The listed secondary outcome concerning grade 3 through 5 adverse events related to study drugs does not have numerical results reported in the registry record.

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

The primary result describes progression-free survival in eligible and analyzable participants with ERCC1 level >= 1.7. The reported medians were 4.7 months for Arm I FOLFOX and 5.3 months for Arm II IT. The sponsor's Cox comparison gave a hazard ratio of 0.91, with a 95% confidence interval from 0.41 to 2.05 and p = 0.83. Together, these estimates indicate considerable uncertainty in the relative comparison.

The result does not establish that one regimen is superior, and it does not establish that the regimens are equivalent. A nonsignificant p-value is not proof that there is no difference. The confidence interval includes 1 and spans values on either side, so the reported data are compatible with a range of relative-hazard estimates. The interval is also broad, limiting precision.

The result applies to the stated high-ERCC1 eligible and analyzable subgroup. It should not automatically be extended to all 213 enrolled participants, to participants with other ERCC1 levels, or to every person with one of the listed cancer types. The registry does not report the full screening and analysis flow needed to explain how the primary subgroup was formed.

The trial's randomized design supports a comparative question, but the primary analysis population is small relative to the overall enrollment. The record does not provide event counts or enough methodological detail to assess all assumptions behind the Cox model. Nor does it report numerical results for the listed secondary outcomes in the registry record for this analysis.

These statistical results alone do not determine treatment decisions. They describe the estimates, uncertainty, and population reported in the registry; they do not establish a clinical recommendation or a definitive conclusion about the relative effects of the regimens.

18. Why This Trial Matters Statistically

This trial raises a useful statistical question about how a randomized study's overall enrollment relates to a primary result reported in a biomarker-defined subgroup. Although the registry enrollment is 213, the high-ERCC1 primary analysis includes 13 participants in Arm I and 15 in Arm II. Those counts define the immediate evidence base for interpreting the reported primary estimates; the larger enrollment should not be substituted for them.

The result also illustrates why a point estimate should be read alongside its confidence interval. The reported hazard ratio is 0.91, but its 95% confidence interval extends from 0.41 to 2.05. The interval includes the value corresponding to equal hazards and is broad, so the point estimate alone gives an incomplete impression of what the data support. The reported p-value of 0.83 does not remove this uncertainty or demonstrate equivalence.

There is also a distinction between descriptive medians and a time-to-event model comparison. The arm medians summarize the estimated event-time distributions at a particular point, whereas the Cox hazard ratio compares hazards under a model. Neither statistic should be treated as a complete account of the other.

Finally, the planning illustrations show how event requirements depend on the effect targeted: smaller target effects require more events under the stated assumptions. Those illustrations are separate from the trial's original design rationale, which is not reported in the registry record. Keeping these planning quantities distinct from the observed trial evidence helps prevent a design illustration from being mistaken for an achieved result.

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 main limitation for interpretation is the small primary analysis population: 13 participants in Arm I and 15 in Arm II. The registry enrollment of 213 does not describe the number contributing to this specific high-ERCC1 result. The registry record does not report a participant flow explaining how the primary analysis population was reached, so the reasons for the difference cannot be determined from these registry entries.

The confidence interval for the hazard ratio is broad, and the arm-specific median intervals are also wide. This limits precision. The record does not include primary endpoint event counts, detailed follow-up information, or the distribution of censoring, which makes it difficult to assess how much event information supports the estimates.

The registry describes the primary population by ERCC1 level but does not give further details about assay procedures, the handling of missing biomarker measurements, or the number of participants assessed for the threshold. It also does not report the complete eligibility and analysis rules. These omissions limit assessment of how representative the analyzed subgroup is of the full enrolled population.

Further methodological details are not reported in the registry record, including the Cox model specification, covariate adjustment, proportional-hazards checks, missing-data procedures, sensitivity analyses, and interim analyses. The statistical analysis plan and any multiplicity approach are also not reported. The numerical results for the listed secondary outcomes are not reported in the registry record for this account, limiting interpretation beyond the primary endpoint and the adverse-event summaries.

The trial is listed as open-label. The registry record does not describe measures used to reduce potential influence of knowledge of treatment assignment on outcome assessment. These limitations do not invalidate the posted results, but they define what can be concluded from the recorded information.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: S1201: Combination Chemo for Patients W/Advanced or Metastatic Esophageal, Gastric, or Gastroesophageal Junction Cancer — NCT01498289.
• Registry record last updated 2019-11-07; status last verified 2019-10.
• Results first posted 2019-08-29.
• 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

NCT01498289 is a completed randomized, open-label, parallel-group phase II trial sponsored by SWOG Cancer Research Network. The registry lists an enrollment of 213 and study dates from 2012-02 to 2018-03. The comparison was between FOLFOX and irinotecan plus docetaxel for advanced or metastatic esophageal, gastric, or gastroesophageal junction cancer.

The primary endpoint was progression-free survival among eligible and analyzable participants with ERCC1 level >= 1.7. The reported medians were 4.7 months for Arm I and 5.3 months for Arm II. The Cox hazard ratio was 0.91, with a 95% confidence interval from 0.41 to 2.05 and p = 0.83. The interval indicates substantial uncertainty in the comparison.

The registry lists secondary outcomes and adverse-event summaries, but numerical results for the listed secondary outcomes are not reported in the registry record. The original sample-size justification, detailed analysis plan, and several methods needed to assess the primary analysis are also not reported in the registry record.