Efficacy and Safety of AZD4547 Versus Paclitaxel in Patients With Advanced Gastric or Gastro-oesophageal Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Efficacy and Safety of AZD4547 Versus Paclitaxel in Patients With Advanced Gastric or Gastro-oesophageal Cancer (NCT01457846), a randomized phase 2 study. The primary outcome is Median Progression Free Survival.

2. Clinical Question

The comparison concerns a selected population whose tumours had a specified FGFR2 finding, not all people with advanced gastric or gastro-oesophageal cancer. The study was randomized and parallel-group, but masking was absent. Consequently, treatment allocation was known during the study. The registry record describes the trial purpose as assessing efficacy, safety and tolerability. It does not report the rationale for the selected tumour marker, nor does it provide enough detail to assess how the marker was measured or how eligibility was confirmed.

3. Trial Design

This was a randomized, open-label, parallel-group phase II study of AZD4547 monotherapy versus paclitaxel. It enrolled patients with advanced gastric adenocarcinoma, including specified lower-oesophageal and gastro-oesophageal junction adenocarcinomas, with FGFR2 polysomy or gene amplification. The registry lists enrollment as 960. That is the recorded enrollment figure, not the number contributing to each posted outcome: the primary endpoint results list 41 in the AZD4547 group and 30 in the paclitaxel group.

The registry lists the study status as terminated. The recorded start is 2011-11, and primary completion is 2013-08. The reason for termination and the relationship between the enrollment figure and the outcome populations are not reported in the registry record. The record does not state whether the enrollment figure reflects an intended target, actual enrollment at termination, or another registry convention, so it should not be treated as the number analyzed for each endpoint.

The AZD4547 regimen was oral tablets at 80mg twice daily, following a two-week-on, one-week-off schedule. Paclitaxel was an 80mg/m² one-hour infusion on days 1, 8 and 15 of a 28-day cycle, with the maximum number of cycles determined by local practice. The record does not describe allocation concealment, the randomization sequence, stratification factors, crossover, or treatment adherence. It also does not report the protocol’s sample-size justification or planned enrollment rationale.

The primary outcome was median progression-free survival in the full analysis set. Secondary outcomes included overall survival, objective response rate, change in target-lesion size at week 8, and the proportion without progressive disease at 8 weeks. Objective response rate is listed as a secondary outcome, but its result is not among the posted secondary results described in the registry record. The record reports serious and other adverse-event counts for groups at risk, but not a complete account of the safety analysis methods.

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: Patients ≥ 25 years with locally advanced or metastatic gastric adenocarcinoma that had FGFR2 polysomy or FGFR2 gene amplification and whose disease had progressed during or after 1st line therapy. Patients whose disease had progressed within 6 months following adjuvant or neo-adjuvant therapy could be included at the discretion of the investigator

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set

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 Median Progression Free Survival, measured over Tumour size assessed at week 8 (±1 week) and then every 8 weeks (±1 week).

Progression-free survival measures the time from randomisation until objective disease progression or death before documented progression. The registry defines progression using RECIST version 1.1. This endpoint combines tumour progression and death into a time-to-event outcome, allowing follow-up to continue until an event or the end of observation. Median progression-free survival is the time by which half of the analyzed group has had an event, subject to the details of the survival estimation method.

Time-to-event outcomes are used in trials because they retain information about when events occur rather than reducing follow-up to a simple event count. They can also accommodate participants whose event has not yet been observed by the data cutoff, provided the analysis appropriately handles censoring. The registry record does not report the censoring rules, survival estimator, or analysis cutoff details needed to assess those aspects for this trial. Tumour assessments were scheduled at week 8 (±1 week) and every 8 weeks (±1 week) thereafter, so progression timing is informed by scheduled assessments as well as the event definition.

The posted medians are 1.8 months for AZD4547 and 3.5 months for paclitaxel. Medians alone do not describe the full survival distributions or the uncertainty around either estimate. The record does not provide confidence intervals for the medians. A separate Cox regression result is reported as a hazard ratio of 1.57 with an 80% confidence interval from 1.12 to 2.21 and p = 0.958. The reference-group coding and details needed to reconcile the reported test with the interval are not reported in the registry record. These omissions limit a complete interpretation of the primary endpoint estimate.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Progression-free survival is a time-to-event measure. Each participant contributes follow-up from randomisation until progression or death, if observed. A participant without a recorded event by the end of available observation may be censored, meaning that the event time is not known beyond the last relevant observation. This is a general explanation; the registry record does not report the exact censoring rules used in this trial.

What does the median represent?

The median is a point on the estimated event-time distribution: by that time, half of the group is estimated to have experienced the event. It is not an average duration, and it does not describe how the entire progression-free survival curves compare. A median can be estimated even when some participants have incomplete follow-up, but its precision depends on the amount and pattern of follow-up and events. Confidence intervals for the primary medians are not reported in the registry record.

How should a hazard ratio be read?

A Cox regression hazard ratio compares the instantaneous event rates between groups over time, subject to the model’s assumptions. It is not a ratio of median survival times and does not directly state how much longer an individual participant will remain progression-free. The primary reported estimate is 1.57, with an 80% confidence interval from 1.12 to 2.21. The group coding and model details are not reported in the registry record, so the direction cannot be assigned confidently from the estimate alone.

A confidence interval describes a range of values compatible with the data and model at the stated confidence level. It conveys uncertainty, not a guarantee that the true effect lies in the interval. The registry reports an 80% interval for the Cox estimate, rather than a more conventional higher-confidence interval. The reported p-value is 0.958. Since the record does not identify the test or explain the relationship between the p-value and the interval, both statistics should be quoted as reported without assuming they form a coherent conventional test-and-interval pair.

Why can counts and percentages tell a different story?

Binary outcomes classify participants according to whether an event or condition was observed. For those outcomes, the recalculated risk difference compares the observed proportions between groups. Its confidence interval describes uncertainty in that unadjusted difference under the calculation method. It does not account for any adjustment or modeling used in the trial analysis. For a binary outcome, a confidence interval spanning zero means that the recalculated data are compatible with differences in either direction at that interval level; it is not proof that the groups are equivalent.

The overall-survival outcome is posted as the number who had died at the data cutoff. A count at a cutoff does not itself incorporate the timing of deaths or differing follow-up. The registry also reports a Cox analysis for overall survival, which uses time-to-event methods. The registry record does not supply enough detail to evaluate how follow-up and censoring were handled for either presentation.

What does a mean difference convey?

The week-8 target-lesion outcome is reported as a mean percentage change from baseline, with standard deviations. A mean difference compares the two group means, while the standard deviations describe spread among participant measurements. In this outcome, the recalculated difference is unadjusted and uses the posted summary statistics. General cautions apply: a mean can be sensitive to unusually large or small measurements, and the interpretation depends on how target lesions and missing week-8 assessments were handled. Those details are not reported in the registry record.

What are the planning illustrations?

Planning illustrations estimate participants per arm under stated assumptions and are not observed power. The stated assumptions are a two-sided significance level of 0.05, equal allocation, a normal approximation, and an effect equal to a stated fraction of the observed effect. The time-to-event illustration gives total events, while the other illustrations give participants per arm. These calculations are hypothetical planning aids based on the observed effect; they are not the trial’s original sample-size calculation. The registry record does not report the original sample-size justification.

Illustrations based on the observed effect are especially uncertain because the observed effect may be unstable, particularly when outcome groups are small. Planning for a smaller fraction of the observed effect requires larger proposed sample sizes in the listed examples. These scenarios do not establish that the trial had adequate power, and they do not quantify the probability that a treatment effect is true. They simply show how planning requirements change under the stated assumptions.

Why does over-dispersion matter in general?

For count outcomes, over-dispersion means that variability exceeds what a simple count model assumes. Ignoring it can make uncertainty estimates too narrow. This is a general statistical consideration, not a reported feature of this trial’s primary analysis. The primary endpoint is time-to-event, and the registry record does not report an over-dispersion assessment.

10. Analyses Reported by the Sponsor

The registry reports a sponsor-associated Cox regression result for progression-free survival: hazard ratio 1.57, 80% confidence interval 1.12 to 2.21, and p = 0.958. The record does not report a separate recalculated primary comparison. The median estimates are 1.8 months for AZD4547 and 3.5 months for paclitaxel, but confidence intervals for those medians are not reported.

The group reference and coding for the hazard ratio, the exact test represented by the p-value, and the analysis details needed to relate the estimate, interval, and p-value are not reported in the registry record. In particular, the interval excludes 1 while the reported p-value is 0.958; the record offers no explanation for that apparent inconsistency. It is not possible to determine from the registry entry whether this reflects different tests, a transcription issue, or another analytical detail.

For secondary outcomes, unadjusted recalculations are available from the posted summary statistics for selected comparisons. They are distinct from the trial’s reported analyses and do not supersede or reproduce the sponsor’s modeling. The overall-survival entry includes a reported Cox result, while the other two secondary outcomes have no reported analysis in the registry record. No discrepancy is flagged for the posted results. The record does not provide enough methodological detail to independently validate the sponsor’s analyses.

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists several secondary outcomes in addition to the primary progression-free survival endpoint. It does not report a multiplicity strategy, such as a hierarchy of hypotheses or adjustment of significance thresholds across outcomes. When multiple outcomes are examined, the chance of at least one apparently unusual result increases in general. Accordingly, unadjusted secondary p-values are exploratory context rather than a controlled family of confirmatory tests. The record does not state whether the original protocol or statistical analysis plan specified a different approach.

Missing-data methods are not reported in the registry record. The posted outcome group sizes differ across endpoints: the primary analysis lists 41 and 30, the overall-survival entry lists 38 and 30, the week-8 lesion-size entry lists 36 and 26, and the 8-week progression status entry lists 41 and 30. These are outcome-specific analysis counts, not a basis for inferring why participants are absent from a particular analysis. Reasons for missing outcomes, evaluability criteria, imputation methods, and sensitivity analyses are not reported.

The registry identifies the primary population as the full analysis set, but does not define the set or explain how it relates to randomized participants. It does not report whether analyses followed intention-to-treat principles, how protocol deviations were handled, or whether the analysis population was determined before results were examined. The recorded enrollment is 960, but the reason that the endpoint populations are smaller is not reported in the registry record.

Interim analyses, stopping rules, and any adjustment for repeated looks at accumulating data are also not reported. The study status is terminated, but the reason and any effect on follow-up or analysis are not reported. These gaps mean that the posted estimates can be described, but the extent of potential selection, missing-data, and multiplicity effects cannot be assessed from the registry entry alone.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

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 (7.4%), biliary tract infection (3.7%), pulmonary embolism (3.7%), asthenia (3.7%), dysphagia (3.7%).

The registry reports serious adverse events in groups at risk of 40 for AZD4547 and 27 for paclitaxel. Serious adverse events affected 8 and 6 participants, respectively. Other adverse events affected 39 of 40 and 26 of 27, respectively. These are group-specific counts; they should not be interpreted as a complete comparison of safety without considering follow-up, exposure, event definitions, and the analysis method. Those details are not reported in the registry record.

The listed serious-event terms include vomiting at 7.4%, and biliary tract infection, pulmonary embolism, asthenia, dysphagia, and lower respiratory tract infection at 3.7% each. The record does not connect these term percentages to a specific treatment group in the displayed information, and they should not be assigned to one. The number of deaths attributed to adverse events is not reported.

Counts of participants with events do not convey event severity beyond the serious-event classification, duration, recurrence, or causal attribution. Nor do they account for differences in treatment exposure. The registry’s figures are descriptive and do not, by themselves, establish comparative safety or explain why the study was terminated.

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

The registry reports median progression-free survival of 1.8 months in the AZD4547 group and 3.5 months in the paclitaxel group. It also reports a Cox hazard ratio of 1.57, an 80% confidence interval from 1.12 to 2.21, and p = 0.958. These values describe the posted primary result, but the group coding and the apparent inconsistency between the interval and p-value are not explained in the registry record. The data therefore do not support a fully resolved interpretation of the reported model result from this entry alone.

The results do not establish what would happen for an individual patient, nor do they answer the same question for patients outside the trial’s selected population. The trial concerned advanced gastric or specified lower-oesophageal and gastro-oesophageal junction adenocarcinoma with FGFR2 polysomy or gene amplification. The registry record does not report enough information about the tumor-marker assessment, missing data, or analysis populations to assess all potential limits on generalization.

Secondary results add descriptive information but involve different endpoints, assessment times, and analysis populations. Their unadjusted recalculations are not substitutes for the primary analysis or for a prespecified confirmatory testing strategy. The record does not report a multiplicity plan, missing-data methods, the original sample-size justification, or the reason for termination. Safety counts likewise require context about exposure and ascertainment that is not reported.

These results should be understood as registry-recorded estimates with notable uncertainty about their analytical context. They do not by themselves prove that either treatment is effective or ineffective, establish equivalence, or determine the cause of the study’s termination. A careful interpretation distinguishes the observed summaries from the conclusions that would require additional protocol and analysis information.

14. Why This Trial Matters Statistically

This trial illustrates why statistical interpretation depends on more than comparing point estimates. The primary outcome is a time-to-event measure, for which medians and hazard ratios summarize different features of follow-up. The registry gives median progression-free survival in each group and a Cox hazard ratio, but it does not provide confidence intervals for the medians or enough model detail to establish the hazard-ratio direction. It also reports a p-value that appears inconsistent with the stated interval. Readers need to distinguish what is numerically posted from what can be confidently interpreted.

The outcome populations also matter. The registry lists enrollment as 960, while the primary endpoint results list 41 and 30. Other endpoints use different analysis counts. Without a description of participant flow, missing data, and the full analysis set, the relationship between these numbers cannot be reconstructed. The record’s terminated status adds context, but the reason for termination and its analytical consequences are not stated.

Finally, secondary outcomes and planning illustrations serve different purposes. Unadjusted re-calculations can help readers understand posted summary statistics, but they do not replace a prespecified analysis or address multiplicity. Sample-size illustrations based on observed effects are hypothetical planning exercises, not observed power and not evidence that the original study was designed around those assumptions. This case therefore highlights the value of complete reporting of estimands, analysis populations, missing-data methods, multiplicity, and statistical plans alongside outcome estimates.

15. 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 limitations affect interpretation of the registry record. Although the record gives enrollment as 960, the primary endpoint results list 41 in the AZD4547 group and 30 in the paclitaxel group. Endpoint populations also vary across secondary outcomes. Participant flow, reasons for missing outcome data, and the relationship between enrollment and analysis counts are not reported. The full analysis set is named but not defined.

The study was open-label, with no masking. The record does not report whether outcome assessment was blinded or how potential assessment effects were addressed. Progression was defined by RECIST version 1.1 and tumour assessments were scheduled at week 8 (±1 week) and then every 8 weeks (±1 week), but the assessment and censoring rules are not described in detail.

The reported primary Cox result has a hazard ratio of 1.57, an 80% confidence interval from 1.12 to 2.21, and p = 0.958. The group coding and explanation for the apparent mismatch between interval and p-value are not reported. Confidence intervals for the group medians are also absent. The record does not report a sample-size justification, statistical analysis plan, multiplicity strategy, missing-data methods, interim analyses, or stopping rules.

The registry status is terminated, while the reason for termination and its effect on follow-up are not reported. Treatment duration and exposure summaries, adverse-event attribution, and a detailed safety analysis are also unavailable. These gaps do not invalidate the posted numbers, but they restrict assessment of their precision, completeness, and relevance beyond the recorded study population.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Efficacy and Safety of AZD4547 Versus Paclitaxel in Patients With Advanced Gastric or Gastro-oesophageal Cancer — NCT01457846.
• Registry record last updated 2017-03-07; status last verified 2017-01.
• Results first posted 2017-03-07.
• 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.

19. Record Summary

NCT01457846 is a randomized, open-label, parallel-group phase II study comparing AZD4547 monotherapy with paclitaxel in advanced gastric or specified lower-oesophageal and gastro-oesophageal junction adenocarcinoma with FGFR2 polysomy or gene amplification. The registry lists enrollment as 960, start as 2011-11, primary completion as 2013-08, and status as terminated.

The primary endpoint was median progression-free survival in the full analysis set. Posted medians were 1.8 months for AZD4547 and 3.5 months for paclitaxel. The reported Cox estimate was a hazard ratio of 1.57, with an 80% confidence interval from 1.12 to 2.21 and p = 0.958; important analysis details, including the group coding and explanation of the apparent interval–p-value mismatch, are not reported.

Secondary outcomes cover overall survival, week-8 lesion-size change, and progression status at 8 weeks. Safety figures include serious and other adverse events. The registry does not report the reason for termination, missing-data methods, multiplicity strategy, or original sample-size justification.