Study of Anti-PD-L1 in Combination With Chemo(Radio)Therapy for Oesophageal Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Anti-PD-L1 in Combination With Chemo(Radio)Therapy for Oesophageal Cancer (NCT02735239), a non-randomized phase 1, phase 2 study. The primary outcome is Number of Subjects Reporting Treatment Emergent Adverse Events (TEAEs).

2. Clinical Question

The study asks whether durvalumab can be administered in combination with chemotherapy or neoadjuvant chemo(radio)therapy in people with metastatic or locally advanced oesophageal cancer and in people with operable oesophageal cancer. Its stated primary question concerns safety: how many participants report treatment-emergent adverse events (TEAEs) after receiving study treatment.

The study has distinct cohorts rather than a single treatment group. Cohorts A1, A2 and B involve metastatic or locally advanced disease and combine durvalumab with oxaliplatin/capecitabine chemotherapy; A2 and B also include tremelimumab. Cohorts C, C-FLOT and D/D2 involve operable disease and pair durvalumab with different neoadjuvant chemotherapy or chemo(radio)therapy approaches before surgery. These cohorts address related but not identical clinical settings and treatment combinations.

There is no randomized control group. The comparison is therefore not a like-for-like test of one intervention against another: the cohorts differ in disease setting and treatment regimen, and allocation is non-randomized. Any differences in the observed safety counts across cohorts should be read as descriptive differences, not as estimates of the causal effect of adding or changing a drug.

The registry also lists secondary questions about response at Cycle 6, progression-free survival, overall survival, one-year survival in operable disease, and response before surgery. The posted record does not include analyses or results for those outcomes. The available quantitative safety information is organized by cohort and includes serious and other adverse-event counts, along with selected serious-event terms.

3. Trial Design

This is an open-label Phase 1/2 study with parallel cohorts and non-randomized allocation. The registry lists an enrollment of 73 and describes the study as completed. The study start date is 2016-06-24 and the primary completion date is 2022-06-16. Those dates describe the registry record; they do not establish the duration of follow-up for each participant.

The design has a Phase 1 safety run-in and a Phase 2 expansion. The run-in included Cohorts A1 and A2; the expansion included Cohorts B, C, C-FLOT and D/D2. Participants in the metastatic or locally advanced cohorts received durvalumab with chemotherapy, with tremelimumab included in A2 and B. Operable-disease cohorts paired durvalumab with chemotherapy or chemo(radio)therapy before surgery. The different cohorts therefore reflect different treatment strategies and populations rather than randomized study arms intended for direct comparison.

The brief study description says immunotherapy was to be given for a four-week period before standard chemo(radio)therapy began, with durvalumab continuing after chemotherapy started. Cohort schedules varied. For example, Cohort A1 planned durvalumab for up to 11 doses and chemotherapy beginning on the day of the third durvalumab dose. The detailed cohort descriptions also specify distinct neoadjuvant regimens and schedules. These are planned treatment descriptions and should not be mistaken for evidence that every participant received every planned dose.

The primary endpoint timeframe is listed as up to 1 year. The record defines TEAEs as events that occurred or worsened after the first study-treatment dose, with adverse-event reporting from signed informed consent through 110 days after the last dose. Toxicity grading used NCI CTCAE version 4.03. Dose-limiting toxicities were evaluated in Cohorts A1, A2 and B during the first 10 weeks. A more detailed protocol, statistical analysis plan, and sample-size justification are not reported in the registry record.

4. Participant Flow and Accounting

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

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All subjects who received at least one dose of study treatment.

5. Baseline Characteristics

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

Population: All subjects who received at least one dose of study treatment.

Only the first 6 of 7 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Number of Subjects Reporting Treatment Emergent Adverse Events (TEAEs), measured over up to 1 year.

The primary endpoint is the number of participants reporting treatment-emergent adverse events. A TEAE is an adverse event that occurs or worsens after the first dose of study treatment. The registry defines the safety population as all participants who received at least one dose. Adverse events were assessed using clinical laboratory tests, vital signs, physical examinations and other medically indicated assessments, including participant interviews. Toxicity was graded according to NCI CTCAE version 4.03.

A participant count is a descriptive safety measure: it records how many people in the defined population experienced an event, rather than how many separate event episodes occurred. This kind of endpoint is useful in an early-phase study because safety and tolerability are central questions when treatment combinations are being evaluated. It can indicate how frequently events were observed in the treated cohorts, while still leaving important details—such as event severity, timing, duration, attribution and recurrence—to other summaries.

The record gives the primary endpoint title and definition, but the posted adverse-event summaries distinguish serious adverse events and other adverse events. They do not provide a single aggregate count explicitly identified as the primary TEAE result across the full safety population. The serious and other event counts should therefore not be treated as interchangeable with a reported overall TEAE count. The record does not report a statistical comparison or an inferential estimate for the primary endpoint.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical interpretation begins with the outcome definition and the population to which it applies. For the primary safety endpoint, the registry identifies participants who received at least one dose of study treatment. A count of affected participants is a direct summary for that population: it answers how many people had at least one event of the specified kind. It does not by itself describe the number of event episodes, the time until an event, or how long an event lasted.

Why are the cohorts not a controlled comparison?

Random allocation helps balance measured and unmeasured participant characteristics across treatment groups, on average. This study is non-randomized, and the cohorts correspond to different clinical settings and regimens. A comparison of their observed counts can therefore reflect differences in participants, disease context, treatment components, observation time or other factors, in addition to any treatment-related differences. The registry explicitly cautions that the arms are not a like-for-like control comparison. The counts should be read descriptively rather than as a causal treatment contrast.

What does a count tell the reader?

A participant count is often intuitive, but it has to be interpreted with its denominator. The serious-event summaries pair the number affected with the number at risk in each cohort. The other-event summaries likewise show the number affected and at risk. These counts communicate the observed burden in each group, but do not provide uncertainty intervals. A confidence interval, in general, is a range of values compatible with the data under a specified statistical model and procedure. It helps communicate uncertainty around an estimate; no such intervals are reported for these safety counts in the registry record.

When event data are sparse or group sizes are small, estimates can be unstable: a small change in the number of participants with events may materially change the apparent pattern. This is a general limitation of small-group summaries, not a calculation of uncertainty for these cohorts. The registry record does not report model-based estimates or inferential tests that would quantify uncertainty for the observed safety differences.

How should event terminology be read?

TEAEs are defined by their timing relative to treatment: they occurred or worsened after the first dose. Serious adverse events are a narrower category than all adverse events, while the record separately reports other adverse events. The listed counts for those categories should not be collapsed into a single total unless the underlying definitions and overlap are known. The record does not report whether individuals could appear in both categories in the summaries, or provide a participant-level dataset that would resolve overlap. Event grading followed NCI CTCAE version 4.03, but the summaries do not give the grade distribution for the events listed.

The selected serious-event terms are presented with percentages. A percentage is meaningful only alongside its denominator and definition—such as whether it represents affected participants or event occurrences, and which population is included. Those details are not reported for the listed percentages in the registry record. The terms and percentages can be quoted as recorded, but should not be expanded into assumptions about incidence within a particular cohort.

What can sample-size planning illustrate?

General sample-size planning asks how much information a future study would need to detect a specified effect under stated assumptions. The registry describes planning illustrations based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and the effect expressed as a stated fraction of the observed effect. For time-to-event outcomes, the described planning approach uses a total number of events; for other outcomes, it expresses participants per arm for a specified power target. These are planning illustrations, not observed power and not proof that the study was designed with those assumptions.

The registry record does not report the trial's own sample-size justification. It also reports no analyses, and the numerical outputs of the planning illustrations are not included in the registry record for this analysis. No required sample size or detectability claim can therefore be stated from this record. In general, a planning calculation depends on the assumed effect, variability or event rate, allocation, significance level and target power. If those inputs do not match the eventual study conditions, the planning result may not describe the study's actual precision.

What does the absence of analyses imply?

The record lists seven outcome measures and five secondary outcome descriptions, but indicates no analyses. That distinction matters: registering an outcome identifies an intended measurement, but does not provide an estimate, confidence interval, p-value or conclusion. In particular, the listed response and survival outcomes cannot be interpreted without their reported analysis results, definitions of analysis populations, follow-up information and uncertainty estimates. The statistical methods used for those outcomes are not reported in the registry record beyond the endpoint descriptions, including the mention of Kaplan-Meier estimation for median survival outcomes.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary safety outcome and several secondary outcomes involving response, progression-free survival, overall survival and survival or response in operable disease. When a study examines multiple outcomes, the chance of at least one apparently notable result can increase if each is tested separately. General approaches to address this include prespecifying a hierarchy, adjusting significance thresholds, or treating secondary analyses as exploratory. The registry record does not report whether any multiplicity strategy was planned or used.

The record reports no analyses. It therefore provides no inferential results to assess for multiplicity, and no adjusted or unadjusted p-values are reported. The statistical analysis plan, including any prespecified hierarchy or interim analyses, is not reported in the registry record. No interim-analysis procedure should be inferred from the Phase 1 safety run-in: the description of dose-limiting-toxicity monitoring is not itself evidence of a formal interim efficacy analysis.

Missing-data procedures and analysis populations for the secondary outcomes are also not reported in the registry record. The primary safety population is defined as participants receiving at least one dose, but that definition does not explain how participants with incomplete follow-up, missing event assessments or treatment discontinuation were handled. The record does not specify censoring rules for progression-free or overall survival, the handling of missing response assessments, or whether all enrolled participants were included in each outcome analysis.

These omissions limit interpretation of both the intended analysis and the completeness of the posted outcome evidence. They do not establish that missing data or multiplicity problems occurred; they mean the registry record does not describe how those issues were addressed.

11. 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: vomiting (40.0%), infection (22.2%), abdominal pain upper (20.0%), dysphagia (20.0%), nausea (20.0%).

The serious adverse-event summaries report the number affected and the number at risk by cohort. The counts are 7 of 12 in A1, 4 of 5 in A2, 12 of 21 in B, 7 of 11 in C, 6 of 9 in C-FLOT and 10 of 15 in D/D2. These are cohort-level descriptions, not estimates from a randomized comparison. The other adverse-event summaries list every participant at risk in each cohort as affected.

The registry also lists deaths by cohort, but the summary does not state whether these deaths were attributed to treatment or provide their causes or follow-up context. They cannot be equated with treatment-related deaths. Serious-event counts do not indicate how many events each participant experienced, when they occurred, or their grade unless those details are separately reported. Although CTCAE version 4.03 was used to grade toxicity, the cohort summaries do not provide a grade-by-grade breakdown.

Selected serious-event terms are listed with percentages: vomiting, infection, abdominal pain upper, dysphagia, nausea and pneumonia. The record does not specify the denominator for those percentages in the information available. They should therefore be treated as the registry's reported descriptors, not recalculated or assigned to a particular cohort. The counts are useful as a safety description, but cannot establish comparative risk or causation.

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

The available results describe safety events observed in treated cohorts. They show that serious adverse events were recorded in each cohort, while the other adverse-event counts list all participants at risk as affected. The registry also provides selected serious-event terms and cohort-level death counts. These details can help readers understand what was recorded during the study, but they are not sufficient to compare treatment strategies reliably.

Because allocation was non-randomized, the cohorts are not like-for-like controls. Participants differed by disease setting and treatment regimen, and the record does not provide adjusted comparisons. The safety counts cannot establish whether a component of a combination caused an event, whether one regimen was safer than another, or whether the observed differences reflect underlying group characteristics or follow-up.

The endpoint is defined as the count of participants reporting TEAEs, but the posted summary does not provide one aggregate TEAE count explicitly identified as the primary endpoint result. The serious and other adverse-event counts give related safety information, not a substitute for an explicitly reported overall primary outcome result. The registry record also contains no analyses for the listed response and survival outcomes. Thus, no statistical conclusion about response or survival can be drawn from the posted record.

These limitations do not mean that the study's clinical observations are unimportant. They define what those observations can support: descriptive reporting of events within the listed cohorts. They do not support claims that a treatment works or does not work, nor a causal explanation of the safety counts. More detailed outcome reporting, uncertainty estimates, analysis methods and population definitions would be needed for stronger statistical interpretation.

13. Why This Trial Matters Statistically

This trial is statistically notable because it combines an early-phase safety focus with several distinct treatment settings and regimens. Its Phase 1 safety run-in and Phase 2 expansion reflect a staged study structure, while its non-randomized parallel cohorts mean that the observed groups do not provide an internally randomized comparison. That design distinction is central when reading any differences between cohort counts.

The study also illustrates why endpoint definitions and analysis populations matter. The primary endpoint is a participant count for TEAEs among people who received at least one dose, whereas the posted safety summaries distinguish serious and other events. Without an explicit aggregate primary-endpoint result and detailed analysis methods, the reader must keep those summaries conceptually separate.

The registry lists response and survival outcomes alongside safety outcomes, but reports no analyses. This contrast is useful for statistical literacy: an outcome being named does not mean an estimate or inferential result has been posted. Readers need effect estimates, uncertainty measures, population definitions and analytic methods to assess how strongly an outcome is supported.

Finally, the cohort-level counts make the descriptive evidence visible without resolving causal questions. In a non-randomized study with different disease settings and regimens, descriptive safety information can characterize what was observed, but it cannot by itself identify which treatment component produced an outcome. The record's limits are therefore part of the statistical interpretation, not a reason to overstate or dismiss the observations.

14. 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 principal trial-specific limitation is the non-randomized design. The cohorts are not like-for-like controls, and their treatment regimens and disease settings differ. Consequently, their safety counts should not be interpreted as controlled estimates of relative risk or treatment effect. The registry does not report adjustment methods that might address measured differences between cohorts.

The group sizes reported for safety are limited, and the record gives no confidence intervals or statistical comparisons for the event counts. The summaries do not describe event timing, duration, recurrence, grade distribution, or participant-level overlap between serious and other adverse-event categories. The listed serious-event percentages have no stated denominator in the information available. Death counts are presented without causes, attribution or cohort-specific follow-up context.

The record defines a primary TEAE endpoint, but does not report an overall primary-endpoint count in the posted summary. It reports no analyses for the response and survival outcomes, and does not provide details on missing-data handling, censoring, analysis populations beyond the primary safety population, or multiplicity. A full protocol and statistical analysis plan are not reported in the registry record. The stated planning approach does not replace a trial-specific sample-size justification, which is also not reported.

These limitations restrict inference from the registry record. They do not show that a particular analysis was incorrectly conducted or that a particular event was caused by treatment; those details are not reported.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Study of Anti-PD-L1 in Combination With Chemo(Radio)Therapy for Oesophageal Cancer — NCT02735239.
• Registry record last updated 2024-03-22; status last verified 2023-09.
• Results first posted 2024-03-22.
• 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.

18. Record Summary

The registry describes a completed, open-label, non-randomized Phase 1/2 study of durvalumab-based combinations for metastatic or locally advanced and operable oesophageal cancer. Enrollment is listed as 73. The study included a safety run-in and an expansion, with parallel cohorts using differing chemotherapy or chemo(radio)therapy regimens; some metastatic or locally advanced cohorts also included tremelimumab.

The primary endpoint is the number of participants reporting TEAEs, with the safety population defined as participants who received at least one dose. Posted safety summaries provide serious and other adverse-event counts by cohort, selected serious-event terms and deaths by cohort. The registry lists response and survival outcomes but reports no analyses. It does not report a trial-specific sample-size justification, a statistical analysis plan, or details of missing-data and multiplicity procedures. The evidence in the registry record is therefore descriptive and does not establish comparative treatment effects.