A Study of BBI608 Plus Weekly Paclitaxel to Treat Gastric and Gastro-Esophageal Junction Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of BBI608 Plus Weekly Paclitaxel to Treat Gastric and Gastro-Esophageal Junction Cancer (NCT02178956), a randomized phase 3 study. The primary outcome is Overall Survival.

2. Clinical Question

The clinical question is a randomized comparison in previously treated, advanced disease. Because both arms include paclitaxel, the contrast is specifically the treatment strategy with BBI608 versus the strategy with placebo alongside paclitaxel. The registry record does not report the detailed eligibility criteria, the precise operational definition of treatment failure, or other restrictions on who could enter. Those details should not be inferred from the broad population description.

3. Trial Design

This was a completed Phase III, randomized, parallel-group trial. The registry reports enrollment of 714 and identifies the study as quadruple masked. The two posted treatment groups are Napabucasin + Paclitaxel and Placebo + Paclitaxel, with 357 participants in each group in the primary outcome summary. The official title describes the comparison as BBI608 plus weekly paclitaxel versus placebo plus weekly paclitaxel.

The study addressed advanced gastric and gastro-esophageal junction adenocarcinoma after prior first-line platinum- and fluoropyrimidine-based chemotherapy. The primary objective was to assess the effect of orally administered BBI608 plus weekly paclitaxel on overall survival in comparison with placebo plus weekly paclitaxel. The registry lists progression-free survival, objective response rate, disease control rate, and adverse events among the secondary outcomes.

The record gives a start of 2014-10 and primary completion on 2017-09-20. Its status is completed. The arm descriptions themselves are blank, so additional treatment details beyond the intervention names and the weekly paclitaxel description are not reported in the registry record. The record labels masking as quadruple but does not specify which roles were masked or describe how masking was maintained. Allocation procedures beyond randomization, including the method used to generate or conceal the allocation sequence, are not reported in the registry record.

The enrollment figure is a trial-level record; outcome-specific analysis counts can differ. For example, the primary overall-survival summaries list 357 per arm, whereas the binary response outcomes use smaller analysis denominators. The registry does not explain all differences in analysis populations or provide a participant-flow account in the registry record reported for this trial.

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: Study arm was determined at patient randomization to the study, no pre-assignment activities occurred.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival, measured over 36 months.

Overall survival is the primary endpoint. It measures the time from the study's defined starting point until death from any cause. As a time-to-event outcome, it incorporates both whether an event occurred and how long participants remained under observation before an event or censoring. The registry gives the unit as months, identifies the summary as a median, and lists a 36 months timeframe.

The median is a way to describe the middle of the estimated survival-time distribution: it marks the time at which the estimated probability of remaining alive reaches one half. It does not describe every participant's course, and it does not by itself compare the full survival experience between treatment groups. The registry also reports a hazard ratio from a log-rank analysis for the group comparison. That relative measure uses event-time information across follow-up rather than only comparing the two median values.

General statistical context: overall survival is often used because it is a direct, clinically interpretable time-to-event measure. It can still be affected by follow-up duration, censoring, and differences in later care, so its interpretation depends on how the trial defined and analyzed observation time. Specific censoring rules and details about the time origin are not reported in the registry record. The median estimates therefore describe the posted analysis, while the hazard ratio is the sponsor-reported comparison used for the primary treatment contrast.

7. Results Posted in the Registry

• Napabucasin + Paclitaxel vs Placebo + Paclitaxel: hazard ratio 1.01 (95% CI 0.86–1.02), reported p = 0.860 (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

• 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:
  • Napabucasin + Paclitaxel vs Placebo + Paclitaxel: hazard ratio 1.01 (95% CI 0.85–1.19), reported p = 0.903 (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 Napabucasin + Paclitaxel vs Placebo + Paclitaxel and the control. The interval gives the range of effects that remain compatible with the data.

Objective Response Rate

• 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:
  • Napabucasin + Paclitaxel: 5.5% (16/289) versus 6.4% (18/283) on control; risk difference -0.8 percentage points (95% CI -4.7 to 3.1), risk ratio 0.87 (0.45–1.67); unadjusted two-proportion z-test p = 0.677.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Napabucasin + Paclitaxel and the control. The interval gives the range of effects that remain compatible with the data.

Disease Control Rate

• 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:
  • Napabucasin + Paclitaxel: 19.0% (55/289) versus 20.5% (58/283) on control; risk difference -1.5 percentage points (95% CI -8.0 to 5.1), risk ratio 0.93 (0.67–1.29); unadjusted two-proportion z-test p = 0.660.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Napabucasin + Paclitaxel and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcome pattern is broadly consistent with the primary result: the posted summaries do not show a clear separation between the treatment groups. Progression-free survival is another time-to-event measure, while objective response rate and disease control rate are binary outcomes. Each provides a different view of disease course, but none should be treated as interchangeable with overall survival.

The secondary estimates are less conclusive than a single prespecified primary test might suggest because several outcomes are considered. When multiple hypotheses are tested, the chance of at least one apparently unusual result can rise unless a multiplicity strategy is specified. The registry record does not report whether a formal adjustment, testing hierarchy, or other control of multiplicity was planned. The secondary results should therefore be treated as supportive or exploratory rather than as independent confirmatory evidence.

The recalculated binary comparisons are unadjusted and use outcome-specific denominators. Sponsor-reported binary analyses use odds ratios from a Cochran-Mantel-Haenszel method. Because the measures and analysis procedures differ, their findings should not be combined informally into a single overall effect. The registry record does not describe the missing-data rules, stratification variables, or the analysis populations in sufficient detail to make stronger cross-endpoint comparisons.

Adverse events are also listed as a secondary outcome, but safety counts answer a different question from efficacy endpoints. Counts of participants with serious or other adverse events describe reported safety experience within the recorded safety groups; they do not establish that an event was caused by a study drug. The overall pattern across the posted secondary endpoints should be interpreted in context of the primary result and the incomplete information about multiplicity, missing outcomes, and analysis populations.

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

• Progression Free Survival
• Objective Response Rate
• Disease Control Rate
• Number of Patients With Adverse Events

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is time-to-event data, and the registry reports the group comparison using a log-rank method and a hazard ratio. The analysis reports a median for each treatment group, with a 95% confidence interval, as well as the relative hazard estimate. These summaries describe different aspects of the outcome: medians locate a point in each estimated survival distribution, while the hazard ratio summarizes the relative event experience over follow-up.

What does the log-rank comparison assess?

General statistical explanation: the log-rank test compares event-time patterns between groups across follow-up. It accounts for the ordering of events and censoring times, rather than treating all participants as if each had a fully observed event time. The test evaluates whether the observed separation in event patterns is compatible with a null comparison. The registry identifies the method but does not report the detailed implementation, any stratification factors, or the assumptions checked for this analysis.

How should the hazard ratio be read?

A hazard ratio compares the instantaneous event rates between groups during follow-up, conditional on remaining event-free up to that point. A value of 1 corresponds to equal hazards under the model; values on either side represent relative differences according to the order of the comparison. It is not a ratio of median survival times, and it is not a direct estimate of the probability that an individual will experience an event. The primary reported estimate is 1.01. The registry record does not provide enough detail to assess whether proportional hazards was an assumption, how it was assessed, or how departures from it would have been handled.

The registry supplies a sponsor-reported hazard ratio, confidence interval, and p-value. For the time-to-event endpoints, the posted analysis uses those reported hazard ratios rather than estimating a new ratio from the arm-level medians. Medians alone do not contain enough information to reproduce the log-rank analysis or its hazard ratio. A comparison based only on the median values would therefore answer a narrower question and would not be a substitute for the sponsor's event-time analysis.

What do the binary endpoint summaries add?

Objective response rate and disease control rate are posted as counts of participants with the outcome among outcome-specific analysis denominators. A simple unadjusted risk difference compares the observed proportions between groups. Its confidence interval describes uncertainty around that absolute difference. These recalculated summaries are not the same estimand as the sponsor-reported odds ratio from a Cochran-Mantel-Haenszel analysis. An odds ratio compares the odds of the outcome between groups; it is not an absolute percentage-point difference, and its interpretation depends on which outcome and group are used as the reference.

General statistical explanation: the Cochran-Mantel-Haenszel method can combine comparisons across strata, if strata have been defined. The registry identifies the method but does not report the stratification variables in the record details. Therefore, the sponsor's adjusted or stratified odds-ratio analysis cannot be fully reconstructed from the posted arm counts alone. The unadjusted recalculations are useful as descriptive checks, but they should not be treated as replacements for the sponsor's analysis.

The binary outcome displays also include planning illustrations based on the observed differences. Those are hypothetical sample-size calculations, not trial outcomes and not evidence about the probability that the observed result will replicate. Their assumptions are described in the sample-size reading. They should be interpreted as conditional planning quantities rather than as a retrospective assessment of the trial's ability to detect an effect.

Why do event-time analyses involve censoring?

General statistical explanation: a participant is censored when follow-up ends before the event is observed, for example because the observation period ends or the participant is no longer followed. Censoring allows partial follow-up information to contribute to a time-to-event analysis. Valid interpretation depends on the censoring process and the analysis rules. The registry record does not report the detailed censoring definitions or the number censored for the primary outcome.

Randomization is intended to support a fair comparison by balancing measured and unmeasured factors on average, but it does not remove uncertainty or guarantee identical groups on every characteristic. The record identifies random allocation and parallel groups; the sequence-generation and allocation-concealment procedures are not reported. Likewise, the quadruple-masking label does not establish which people were masked or explain how masking was implemented. Those missing details limit assessment of the analysis process, even though the registry does provide the main outcome summaries and the named statistical tests.

11. Confidence Intervals and Precision

The primary overall-survival hazard ratio is 1.01, with a 95% confidence interval from 0.86 to 1.02. The interval includes 1, the value corresponding to no relative hazard difference. Its placement near 1 is consistent with the reported estimate being close to the null comparison, but an interval is not proof that the true effect is exactly 1. The reported p-value is 0.860. The interval and p-value are complementary summaries of the same reported analysis; neither says how likely a treatment is to be superior.

The median overall-survival estimates also have confidence intervals: 6.93 months, with an interval from 6.28 to 7.69, for Napabucasin + Paclitaxel; and 7.36 months, with an interval from 6.64 to 8.15, for Placebo + Paclitaxel. The intervals overlap. Overlap is not itself a formal test of the treatment comparison; the reported log-rank hazard-ratio analysis is the relevant comparison in the registry.

General statistical explanation: a 95% confidence interval is produced by a procedure designed, under its assumptions, to cover the target parameter in 95% of repeated samples. It is not a statement that there is a 95% probability that this particular interval contains the fixed true value. Wider intervals indicate less precision; narrower intervals indicate greater precision, conditional on the model and data. Precision should be assessed alongside the size and clinical meaning of effects, not from the p-value alone.

The registry provides confidence intervals for the primary median and relative hazard summaries, but the record does not report all details needed to assess their construction or the proportional-hazards assumptions. The binary recalculations also have intervals for absolute differences, while sponsor analyses report odds-ratio intervals. These intervals concern different measures and should not be compared as if they were on a common scale.

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 planning figures are conditional illustrations based on the observed effect, not the trial's original sample-size design. For the time-to-event outcomes, values are total events, not participants per arm. For the binary outcomes, values are participants per arm. The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; each calculation considers an effect equal to the stated fraction of the observed effect. The trial's own original design assumptions are not reported in the registry record.

The 80% power column gives the planned event total or per-arm participant count under which the assumed effect would be detected in 80% of repeated trials, if the planning assumptions hold. The 90% power column describes the corresponding target for detection in 90% of repeated trials. These are prospective planning interpretations of hypothetical sample sizes, not descriptions of the completed trial's observed outcomes. This is not observed power.

For the primary time-to-event planning illustration, the total event figures are 317,098 for 80% power and 424,504 for 90% power when planning for 100% of the observed effect. The corresponding figures are 563,729 and 754,673 when planning for 75% of the observed effect, and 1,268,390 and 1,698,014 when planning for 50% of the observed effect. These large totals reflect the very small observed relative effect used as the basis for the illustration; they are not event counts from this trial.

Across these planning rows, a smaller target effect requires more information to distinguish it from the null under the same assumptions. For binary outcomes, the analogous quantities are participants per arm; for time-to-event outcomes, the quantities are total events. A curve of these calculations represents how the planning quantity changes as the assumed effect changes. It does not predict the trial's result or show how much power the completed trial actually achieved.

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 = 1.010, so ln(HR) = 0.0100.
2. d = 4 × (1.96 + 0.842)² / [ln(1.010)]² = 317,097.4, rounded up to 317,098 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 <- 1.01                              # 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(2178956)
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 = 317,098 (80% power), 424,504 (90% power)
 75% of observed effect: events = 563,729 (80% power), 754,673 (90% power)
 50% of observed effect: events = 1,268,390 (80% power), 1,698,014 (90% power)
Simulated power with 317,098 events: 0.797 (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 primary overall-survival comparison in the registry is the sponsor-reported log-rank hazard ratio, 1.01, with a 95% confidence interval from 0.86 to 1.02 and p = 0.860. The posted primary analysis is not a hazard ratio recalculated from the medians. The registry record identifies the analysis method and result but does not provide participant-level event times needed to reproduce that calculation independently.

For the binary response outcomes, unadjusted recalculations from the posted counts give risk differences and associated intervals, while the sponsor reports odds ratios from Cochran-Mantel-Haenszel analyses. These methods use different effect measures and may use different handling of stratification. The registry record does not report the factors used in the sponsor's Cochran-Mantel-Haenszel analyses. The recalculated absolute differences are therefore descriptive checks rather than replacements for the sponsor's analyses.

The posted results do not flag a discrepancy between the sponsor analyses and the recalculations. For the time-to-event outcomes, the sponsor's reported hazard ratios are the estimates used in the statistical summaries; for binary outcomes, the recalculations should be read alongside, not instead of, the sponsor's reported analyses. Differences in estimands and methods matter even when point estimates appear similar. The registry does not provide enough detail to reconcile every aspect of the analysis populations or adjustment procedures.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome, overall survival, and multiple secondary outcomes. The record does not report a multiplicity adjustment, a prespecified hierarchy for testing secondary outcomes, or a strategy for maintaining a family-wise error rate across the outcome set. Without that information, individual secondary p-values should not be interpreted as if each were a standalone confirmatory test. This matters especially when several outcomes and analyses are considered, because repeated testing can produce apparently notable findings by chance.

The record reports outcome-specific analysis denominators for response and disease control, but does not explain why those denominators differ from the primary outcome groups or how participants without an assessable outcome were handled. Missing-data methods, imputation procedures, and sensitivity analyses are not reported in the registry record. For overall survival and progression-free survival, the censoring rules and follow-up completeness are also not reported in the registry record for this trial.

The registry identifies the reported methods but does not give full analysis-population definitions. It is therefore not possible from the record alone to determine whether each endpoint used an intention-to-treat population, a per-protocol population, or another specified set. The handling of protocol deviations and treatment discontinuations is not reported. Nor are subgroup plans or any adjustments to the primary analysis described.

These omissions do not invalidate the posted estimates, but they limit independent evaluation of how the analyses were prespecified and how robust the findings are to missingness or alternative analysis choices. The reported overall-survival comparison remains the primary statistical result in the record. Secondary findings should be read with additional caution because the multiplicity and missing-data procedures are not documented.

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: vomiting (3.9%), diarrhoea (2.8%), abdominal pain (2.2%), dyspnoea (2.0%), pyrexia (1.7%).

The safety record lists serious adverse events affecting 125 of 357 participants in the Napabucasin + Paclitaxel group and 101 of 350 in the Placebo + Paclitaxel group. It also lists other adverse events affecting 352 of 357 and 338 of 350 participants, respectively. The safety denominators differ between groups and do not exactly match the primary outcome groups. The registry record does not explain the denominator difference or provide detailed safety-population rules.

The most frequently listed serious adverse event terms are vomiting (3.9%), diarrhoea (2.8%), abdominal pain (2.2%), dyspnoea (2.0%), pyrexia (1.7%), and nausea (1.7%). The record does not assign those listed terms to a particular treatment group in the registry record. These figures describe reported serious events, not necessarily events judged to be caused by a study treatment.

The adverse-event summary also reports deaths of 286 and 275 in the respective safety groups. The record does not provide enough detail in the registry record to interpret those counts as treatment-related deaths or to reconcile them with the overall-survival analysis. Safety counts are descriptive and should be read with their denominators and event definitions in mind.

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

The registry's primary result is a sponsor-reported overall-survival hazard ratio close to 1, with a confidence interval that includes 1 and a reported log-rank p-value of 0.860. The group median estimates are also similar in scale. Taken together, the posted data do not show a clear difference in overall survival between the two randomized strategies in this trial.

This does not prove that the two strategies have exactly the same effect. A non-significant result is not an equivalence finding, and equivalence or non-inferiority requires an appropriate design, a prespecified margin, and analysis methods. Such a margin is not reported in the registry record. The confidence interval describes the range of relative effects compatible with the analysis under its assumptions; it does not eliminate uncertainty.

The result also does not establish that either treatment is effective or ineffective in general, and it should not be extended automatically to populations or treatment settings outside the trial description. The comparison concerns adults with advanced gastric or gastroesophageal junction adenocarcinoma after prior platinum/fluoropyrimidine-containing treatment, and it compares BBI608 plus paclitaxel with placebo plus paclitaxel.

The secondary outcomes provide additional but not definitive context. They are subject to the limitations of multiple testing and incomplete registry reporting about missing data and analysis populations. The safety counts describe observed reports in the listed groups, but do not establish causation. The record supports a careful description of the posted estimates and their uncertainty, not a treatment recommendation or a claim that no clinically meaningful effect is possible.

18. Why This Trial Matters Statistically

This trial illustrates why a randomized comparison should be interpreted through its prespecified primary endpoint and effect measure rather than by looking only at the arm-specific medians. Overall survival is summarized by medians, but the reported treatment comparison uses a log-rank hazard ratio. The hazard ratio incorporates event-time information across follow-up, while the medians describe one location in each survival distribution. They answer related, not identical, questions.

The posted primary estimate is close to the no-difference value, and its confidence interval includes that value. This makes the uncertainty as important as the point estimate: readers should distinguish a lack of clear evidence for a difference from proof of identical effects. The registry record does not report a clinical equivalence threshold that would allow the result to be framed as demonstrating equivalence.

The trial also highlights how different outcome types call for different statistical summaries. Time-to-event outcomes use hazard ratios and medians; response outcomes are reported as counts and sponsor odds ratios, with unadjusted risk-difference recalculations available as a separate perspective. Because these measures differ, their numerical values cannot be compared directly. Secondary endpoints further require attention to multiplicity and to how outcomes were analyzed.

Finally, the registry has limits as a statistical record. It names the design and principal analyses but does not give all details of allocation, masking roles, missing-data handling, or multiplicity. A disciplined interpretation separates what is reported from what remains undocumented and avoids turning an imprecise or non-significant estimate into a categorical clinical conclusion.

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.

Several trial-specific details needed for a full appraisal are not reported in the registry record. These include the original sample-size justification and design assumptions, detailed allocation and concealment procedures, which people were covered by quadruple masking, and the operational procedures used to maintain masking. The arm descriptions are blank, leaving treatment details beyond the listed interventions and weekly paclitaxel description undocumented in the record details.

For the primary time-to-event analysis, the registry does not report the detailed time origin, censoring rules, event counts, follow-up completeness, or assessment of the proportional-hazards assumption. These omissions constrain independent review of the hazard-ratio analysis. The posted arm-level median estimates do not permit an independent reconstruction of that analysis.

For secondary outcomes, the record does not report a multiplicity plan, analysis-population definitions, missing-data procedures, or the strata used for the sponsor's Cochran-Mantel-Haenszel analyses. The response outcomes have smaller analysis denominators than the primary outcome summaries, and the record does not explain the difference. The safety denominators also differ between the two groups, without a detailed explanation in the recorded details.

These limitations concern the completeness of the statistical record; they do not by themselves show that an analysis was incorrect. They do mean that the posted estimates should be interpreted with the uncertainty and reporting gaps in view. The results apply to the trial population and treatment comparison described in the registry, and generalization beyond that setting is not established by these summaries.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study of BBI608 Plus Weekly Paclitaxel to Treat Gastric and Gastro-Esophageal Junction Cancer — NCT02178956.
• Registry record last updated 2023-11-15; status last verified 2023-11.
• Results first posted 2021-05-10.
• 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

This completed Phase III randomized parallel-group trial compared napabucasin (BBI608) plus weekly paclitaxel with placebo plus weekly paclitaxel in adults with advanced, previously treated gastric or gastro-esophageal junction adenocarcinoma. The registry reports enrollment of 714, quadruple masking, and overall survival as the primary endpoint.

The primary analysis reports median overall survival of 6.93 months and 7.36 months, respectively, with a sponsor-reported hazard ratio of 1.01 (95% CI 0.86 to 1.02; p = 0.860). The posted result does not show a clear overall-survival difference. Secondary outcomes and safety summaries add context, but missing reporting on multiplicity, analysis populations, and missing-data methods limits independent assessment.