This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical results on this page are restricted to the trial data posted on ClinicalTrials.gov for PACIFIC-6.
1. Trial at a Glance
PACIFIC-6 was a completed phase 2, non-randomized, single-group study of durvalumab after sequential chemoradiation in patients with unresectable stage III non-small cell lung cancer. The registry reports 117 participants and two ECOG performance-status groups, with safety analyses based on patients who received at least one dose of investigational product.
| Feature | PACIFIC-6 |
|---|---|
| Trial name | PACIFIC-6 |
| ClinicalTrials.gov identifier | NCT03693300 |
| Phase | Phase 2 |
| Status | Completed |
| Population | Patients with unresectable stage III non-small cell lung cancer |
| Allocation | Non-randomized |
| Design model | Single group |
| Masking | None |
| Primary purpose | Treatment |
| Intervention | Durvalumab |
| Enrollment | 117 |
| Lead sponsor | AstraZeneca |
| Primary endpoint | Number of Patients With Grade 3 and Grade 4 Treatment-related Adverse Events (TRAEs) |
| Results posted | Yes |
| Statistical analyses posted | 6 |
2. Clinical Question
The registered study question centers on the safety and tolerability of durvalumab after sequential chemoradiation in patients with unresectable stage III non-small cell lung cancer. The primary registered endpoint counts patients with Grade 3 and Grade 4 treatment-related adverse events during the first 6 months.
Population
Patients with unresectable stage III non-small cell lung cancer enrolled in the PACIFIC-6 study.
Intervention
Durvalumab after sequential chemoradiation.
Comparator
There was no randomized comparator. The study was non-randomized and used a single-group design.
Primary question
What proportion of treated participants experienced Grade 3 or Grade 4 treatment-related adverse events within the registered 6-month time frame?
3. Trial Design
Durvalumab
- Durvalumab intervention
- 114 participants were at risk for serious adverse events
- Primary safety analyses report proportions and exact 95% confidence intervals
Durvalumab
- Durvalumab intervention
- 3 participants were at risk for serious adverse events
- Primary safety analyses report proportions and exact 95% confidence intervals
The registry describes PACIFIC-6 as non-randomized, with a single-group design and no masking. The two reported groups are distinguished by ECOG performance status rather than by randomized treatment assignment.
4. Endpoints
| Endpoint | Registered definition / time frame | Statistical treatment |
|---|---|---|
| Number of Patients With Grade 3 and Grade 4 Treatment-related Adverse Events (TRAEs) | Up to 6 months. Safety and tolerability of Durvalumab as defined by Grade 3 and Grade 4 TRAEs following IV infusion administration was assessed. | Binary endpoint; proportion (%) with two-sided 95% Clopper-Pearson confidence interval |
The registry lists one primary endpoint and reports six statistical analyses for that endpoint. The analyses are presented for the ECOG PS 0 or 1 group, the ECOG PS 2 group, and the total durvalumab population, with two related six-month safety definitions represented in the posted analyses.
Primary safety endpoint
The endpoint is binary at the participant level: whether a participant experienced the specified Grade 3 or Grade 4 treatment-related adverse event outcome during the registered period.
Time frame
The registered primary endpoint uses a time frame of up to 6 months. Some posted analyses further describe events with onset within 6 months of the first dose.
5. Analysis Population
The statistical analyses use the safety analysis set, defined in the registry analysis text as all patients who received at least one dose of investigational product, whether partially or in full.
| Population / group | Role in the posted analysis |
|---|---|
| Safety analysis set | All patients who received at least one dose of investigational product, partial or in full. |
| Durvalumab ECOG PS 0 or 1 | Primary safety analysis group. |
| Durvalumab ECOG PS 2 | Primary safety analysis group. |
| Durvalumab Total | Total population analysis. |
6. Statistical Methodology
Binary endpoint analysis
The primary safety outcome is binary: a participant either contributes to the event category being estimated or does not. For a binary endpoint, the natural descriptive quantity is the observed proportion of participants with the event.
Here, x represents the number of participants meeting the endpoint definition and n represents the participants in the relevant analysis population. The registry reports the resulting proportion directly rather than providing event counts for each posted statistical analysis.
Exact / Clopper-Pearson binomial confidence intervals
The registry identifies Exact / Clopper-Pearson binomial methodology for the posted analyses. This is an exact confidence-interval procedure for a binomial proportion. It is especially useful when sample sizes are small or when the observed proportion is near 0 or 1, where simple normal approximations can perform poorly.
The reported interval expresses statistical uncertainty around the estimated proportion. It does not describe the range of adverse-event experiences that an individual participant might have.
Two-sided confidence intervals
All six posted analyses use two-sided 95% confidence intervals. A two-sided interval allows uncertainty on both sides of the estimated proportion and does not assume in advance that the true event rate must be above or below the observed estimate.
No comparative effect measure
The registry describes the effect measure as a proportion (%). There is no randomized control group in the design from which to estimate a treatment hazard ratio, risk ratio, risk difference, or odds ratio. The central statistical task is therefore estimation of the safety-event probability and its uncertainty.
7. Primary Results
The registry reports six statistical analyses for the primary safety endpoint. The first set evaluates Grade 3 or Grade 4 possibly related adverse events during the six-month period. The second set evaluates Grade 3 or Grade 4 possibly related adverse events with onset within 6 months of the first dose.
7.1 Grade 3 and Grade 4 treatment-related adverse events
| Analysis group | Estimate | 95% CI | Method |
|---|---|---|---|
| Durvalumab ECOG PS 0 or 1 | 6.1% | 2.50–12.24 | Exact / Clopper-Pearson binomial |
| Durvalumab ECOG PS 2 | 0% | 0.00–70.76 | Exact / Clopper-Pearson binomial |
| Durvalumab Total | 6.0% | 2.44–11.94 | Exact / Clopper-Pearson binomial |
Durvalumab total
95% CI: 2.44–11.94 · Two-sided exact binomial interval
Primary endpoint: Number of Patients With Grade 3 and Grade 4 Treatment-related Adverse Events (TRAEs), up to 6 months.
The reported 6.0% estimate means that the observed proportion for the specified Grade 3 and Grade 4 treatment-related adverse-event endpoint in the total durvalumab analysis was 6.0%.
It does not mean that durvalumab caused a 6.0% risk in every patient, nor does it establish a comparative increase or decrease in risk because the study was not randomized and had no comparator arm.
The 95% CI of 2.44–11.94 communicates uncertainty around the estimated proportion. Its width is important: the point estimate alone gives a much narrower impression of certainty than the interval does.
No p-value is reported for this analysis. A p-value would address compatibility with a specified null hypothesis; it would not measure the size of the observed safety-event proportion. For a single-arm safety estimate, the proportion and its confidence interval are the more directly relevant descriptive quantities.
The use of an exact binomial interval is particularly relevant because the analysis includes a small ECOG PS 2 group. The very wide interval for that group illustrates why a point estimate of 0% should not be interpreted as evidence that the underlying event probability is exactly zero.
7.2 Grade 3 and Grade 4 adverse events with onset within 6 months
| Analysis group | Estimate | 95% CI | Method |
|---|---|---|---|
| Durvalumab ECOG PS 0 or 1 | 4.4% | 1.44–9.94 | Exact / Clopper-Pearson binomial |
| Durvalumab ECOG PS 2 | 0% | 0.00–70.76 | Exact / Clopper-Pearson binomial |
| Durvalumab Total | 4.3% | 1.40–9.69 | Exact / Clopper-Pearson binomial |
Durvalumab total
95% CI: 1.40–9.69 · Two-sided exact binomial interval
Posted analysis: Grade 3 or Grade 4 possibly related adverse events with onset date within 6 months of the first dose.
The 4.3% estimate is the reported proportion for the total durvalumab population under the analysis definition requiring onset within 6 months of the first dose.
This estimate is not a hazard ratio, relative risk, or causal treatment effect. Because PACIFIC-6 is a single-group, non-randomized study, the analysis describes the observed safety experience rather than a difference attributable to durvalumab relative to a concurrent randomized comparator.
The 95% CI of 1.40–9.69 indicates meaningful uncertainty around the 4.3% estimate. The interval is not a prediction interval for individual patients and should not be read as saying that future individual event probabilities must fall inside those numerical limits.
Again, no p-value is reported. Statistical significance testing would require a prespecified null value or comparator. The reported confidence interval instead provides a direct description of estimation precision.
The small ECOG PS 2 group is particularly important when interpreting its 0% estimate and 0.00–70.76 confidence interval. An observed zero does not prove that the underlying event probability is zero; exact binomial intervals make that uncertainty explicit.
8. Comparing the Two Posted Safety Definitions
The registry provides two related sets of analyses. One set reports the Grade 3 and Grade 4 treatment-related adverse-event proportion during the primary six-month framework, while another specifically describes Grade 3 and Grade 4 possibly related adverse events with onset within 6 months of the first dose.
| Analysis definition | ECOG PS 0 or 1 | ECOG PS 2 | Total |
|---|---|---|---|
| Grade 3 and Grade 4 treatment-related adverse events | 6.1% (95% CI 2.50–12.24) | 0% (95% CI 0.00–70.76) | 6.0% (95% CI 2.44–11.94) |
| Grade 3 and Grade 4 possibly related adverse events with onset within 6 months of first dose | 4.4% (95% CI 1.44–9.94) | 0% (95% CI 0.00–70.76) | 4.3% (95% CI 1.40–9.69) |
The two sets of estimates should not be treated as if they were independent randomized comparisons. They are analyses of the same overall study population under related safety definitions. The registry does not provide a multiplicity framework for treating these analyses as separate confirmatory hypothesis tests.
9. Serious Adverse Events
The ClinicalTrials.gov record also report serious adverse events by arm/group using affected participants over participants at risk.
| Group | Affected / at risk |
|---|---|
| Durvalumab ECOG PS 0 or 1 | 32/114 |
| Durvalumab ECOG PS 2 | 0/3 |
| Durvalumab Total | 32/117 |
These serious-adverse-event data are distinct from the primary Grade 3 and Grade 4 treatment-related adverse-event analyses. Severity grade and seriousness are different safety concepts, and the ClinicalTrials.gov record does not provide a formal statistical comparison between the serious-adverse-event groups.
10. Statistical Methods Explained
Why was an exact binomial method used?
The primary outcome is a binary participant-level endpoint summarized as a proportion. The exact or Clopper-Pearson method provides a confidence interval directly from the binomial distribution rather than relying on a large-sample normal approximation. This is particularly useful when one analysis group is small or the observed event rate is close to zero.
What does a 6.0% estimate mean?
It is the reported proportion for the total durvalumab analysis under the specified primary safety endpoint. It is a descriptive estimate of the observed frequency of the defined event category. It is not a percentage reduction, hazard ratio, relative risk, or probability that durvalumab caused an event in a particular patient.
Why is the confidence interval important?
The point estimate is only one summary of the data. The 95% CI of 2.44–11.94 around the 6.0% total-population estimate shows the statistical uncertainty surrounding that estimate. The interval is substantially wider than the point estimate itself, so reporting only 6.0% would understate the uncertainty visible in the analysis.
Why is the ECOG PS 2 interval so wide?
The registry reports an ECOG PS 2 group with 3 participants at risk for serious adverse events. When a binary endpoint is evaluated in a very small group, the possible underlying event probabilities remain uncertain even when no events are observed. The reported 0.00–70.76 exact confidence interval illustrates this directly.
Why is there no hazard ratio?
A hazard ratio is generally used for a time-to-event comparison between groups. PACIFIC-6 is described as non-randomized and single-group, and the posted primary analysis is a binary safety proportion. The ClinicalTrials.gov record therefore support proportion estimates with exact confidence intervals rather than a comparative hazard ratio.
Why is a p-value not the main result here?
The registry does not report a p-value for the primary safety analyses. A p-value requires a specified null hypothesis. For a single-group safety endpoint, a confidence interval around the observed proportion is directly informative about estimation uncertainty without requiring a comparison against an arbitrary null value.
Does the confidence interval prove that the true rate lies inside it?
No. A 95% confidence interval is a property of the statistical procedure, not a statement that there is a 95% probability that one fixed unknown parameter is inside this particular interval. In repeated sampling under the relevant assumptions, the procedure would capture the underlying binomial proportion at the stated confidence level in the long-run sense associated with the method.
11. Understanding a Zero Estimate
ECOG PS 2 estimate
95% CI: 0.00–70.76
Exact / Clopper-Pearson binomial confidence interval
The ECOG PS 2 group illustrates one of the most important lessons in clinical-trial safety statistics: zero observed events is not the same as zero underlying risk.
If no participants in a small group meet a binary endpoint, the observed proportion is exactly 0%. But the data can still be compatible with a broad range of underlying probabilities. The corresponding 95% exact confidence interval reported here extends from 0.00 to 70.76.
What the 0% tells us
No event was observed in the analysis represented by the reported 0% estimate.
What the 0% does not tell us
It does not establish that the underlying event probability is exactly zero.
What the upper limit tells us
The 70.76 upper confidence limit shows how uncertain the underlying probability remains in this small group.
Why exact methods help
The Clopper-Pearson interval remains applicable when event counts are very small, including an observed count of zero.
12. Precision and Sample Size
The PACIFIC-6 results provide a useful demonstration of how precision depends on the amount of information available. The total population contains 117 participants, whereas the ECOG PS 2 group contains only 3 participants at risk for serious adverse events.
| Group | Participants at risk for serious adverse events | Primary safety estimate | 95% CI |
|---|---|---|---|
| Durvalumab ECOG PS 0 or 1 | 114 | 6.1% | 2.50–12.24 |
| Durvalumab ECOG PS 2 | 3 | 0% | 0.00–70.76 |
| Durvalumab Total | 117 | 6.0% | 2.44–11.94 |
The contrast is primarily about information quantity. The ECOG PS 2 estimate is 0%, but its interval is extremely wide. The larger ECOG PS 0 or 1 group produces a more informative estimate, although its interval still demonstrates uncertainty around the 6.1% point estimate.
13. Interpreting the Safety Analysis Without a Comparator
Randomization is not simply a procedural feature; it determines what kinds of causal comparisons a trial can make. PACIFIC-6 is non-randomized and single-group, so the primary analysis is fundamentally an estimation problem rather than a randomized comparative-effectiveness problem.
| Question | What PACIFIC-6 data support | What the ClinicalTrials.gov record does not establish |
|---|---|---|
| How frequent was the specified safety event? | Observed proportions with 95% exact confidence intervals. | — |
| How precise is the estimated proportion? | Clopper-Pearson confidence intervals quantify statistical uncertainty. | — |
| Was durvalumab safer than a comparator? | Not answered by a single-group design. | A randomized comparative treatment effect. |
| Is the ECOG PS 2 event rate exactly zero? | The observed estimate is 0%. | An underlying event probability of exactly zero. |
This distinction is important when reading safety evidence. A single-arm study can characterize the frequency and uncertainty of adverse events, but it cannot use randomization to balance both measured and unmeasured prognostic factors between competing treatment groups.
14. Limitations
- Non-randomized design: PACIFIC-6 is described as non-randomized and single-group, so its primary safety estimates do not provide a randomized treatment comparison.
- No comparator arm: the ClinicalTrials.gov record does not support a treatment-effect estimate such as a risk ratio, risk difference, odds ratio, or hazard ratio.
- Small ECOG PS 2 group: the reported serious-adverse-event denominator is 3 participants, producing a very wide exact confidence interval for the corresponding primary safety estimate.
- Safety analysis population: the posted analyses use patients who received at least one dose of investigational product, which is appropriate for exposure-based safety assessment but differs conceptually from an intention-to-treat efficacy analysis.
- Endpoint definitions: the registry supplies related but distinct six-month safety analyses. The 6.0% and 4.3% total-population estimates should not be treated as if they were competing randomized treatment effects.
- No formal p-values reported: the statistical analyses posted on ClinicalTrials.gov provide proportions and confidence intervals but no p-values or hypothesis-test results.
- No time-to-event analysis for the primary endpoint: the registry-reported analysis is a binary proportion analysis rather than a survival analysis. Consequently, proportional-hazards assumptions are not relevant to the posted primary method.
- Limited detail in the ClinicalTrials.gov record: the ClinicalTrials.gov record does not include baseline characteristics, subgroup efficacy analyses, median survival, or a detailed missing-data/imputation strategy.
15. Missing Data and Analysis Interpretation
The ClinicalTrials.gov record specifies the safety analysis population but do not provide a detailed missing-data or imputation strategy for the primary endpoint. It is therefore not appropriate to infer an imputation method or assume that a particular missing-data mechanism was used.
This is an important distinction between statistical education and reconstruction of an unavailable statistical analysis plan. A method should be attributed to the trial only when the trial data support that attribution.
16. Multiplicity, Interim Analysis, Stratification, and Bayesian Methods
The registry-reported PACIFIC-6 trial data do not report a multiplicity-adjustment strategy, an interim-analysis procedure, a stratified primary analysis, or a Bayesian method for the posted statistical analyses.
| Design topic | Supported by the ClinicalTrials.gov record? | Interpretation |
|---|---|---|
| Multiplicity adjustment | Not reported | No multiplicity procedure should be inferred. |
| Interim analysis | Not reported | No interim-monitoring rule should be inferred. |
| Stratification | Not reported for the primary statistical method | The analyses posted on ClinicalTrials.gov are reported by ECOG performance-status group, but that is not evidence of a stratified inferential model. |
| Bayesian analysis | Not reported | The registry-reported method is exact frequentist binomial estimation. |
| Crossover | Not reported | No crossover adjustment should be inferred. |
| Non-inferiority margin | Not reported | The study data registry-reported do not describe a non-inferiority hypothesis. |
These omissions are deliberate. A statistical analysis page should distinguish between methods documented for a particular trial and methods that might be common elsewhere in clinical research.
17. Why This Trial Matters Statistically
PACIFIC-6 is a useful teaching case because its primary statistical problem is deliberately narrower than the complex comparative survival analyses seen in many oncology trials. The key task is estimating a safety proportion and communicating uncertainty appropriately in both a larger overall population and a very small performance-status subgroup.
| Concept | How it appears in PACIFIC-6 |
|---|---|
| Single-group design | The study is non-randomized with a single-group design. |
| Binary endpoint | The primary endpoint counts participants with Grade 3 and Grade 4 treatment-related adverse events. |
| Proportion | The reported effect measure is a proportion (%). |
| Exact confidence interval | The registry uses the Exact / Clopper-Pearson binomial method. |
| Small-sample inference | The ECOG PS 2 group has a very small denominator, making precision a central interpretive issue. |
| Zero-event interpretation | An observed estimate of 0% is accompanied by a very wide 95% confidence interval. |
| Safety analysis population | The safety analysis set includes all patients who received at least one dose of investigational product. |
| Absolute rather than comparative interpretation | The results describe observed safety-event proportions rather than randomized treatment effects. |
18. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
19. Related Statistical Calculators
20. Sources
- ClinicalTrials.gov: PACIFIC-6, NCT03693300.
- Linked publication: PubMed record for PMID 35961520.
Continue through Clinical Biostats
Explore statistical tutorials and calculators that explain the estimation methods used in clinical-trial analysis.
21. Record Summary
PACIFIC-6 provides a focused example of single-group safety estimation in oncology. The study was a completed phase 2, non-randomized, single-group trial with 117 participants evaluating durvalumab after sequential chemoradiation in unresectable stage III non-small cell lung cancer. Its registered primary endpoint was the number of patients with Grade 3 and Grade 4 treatment-related adverse events up to 6 months. The posted statistical analyses use the safety analysis set and report proportions with two-sided 95% exact Clopper-Pearson confidence intervals.
The central statistical lesson is that a safety proportion should be interpreted together with its confidence interval and analysis population. The total-population estimate was 6.0% with a 95% CI of 2.44–11.94 for the primary analysis, while the ECOG PS 2 group had an observed estimate of 0% with a much wider 95% CI of 0.00–70.76. The latter result demonstrates why an observed zero event rate, particularly in a very small group, should not be interpreted as evidence of zero underlying risk.