An Open-Label Crossover Study to Compare the Relative Bioavailability, Efficacy and Safety of Epanova® and Lovaza® in Men and Women With a History of Pancreatitis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for An Open-Label Crossover Study to Compare the Relative Bioavailability, Efficacy and Safety of Epanova® and Lovaza® in Men and Women With a History of Pancreatitis (NCT02189252), a randomized phase 1 study. The primary outcome is Baseline-adjusted AUC0-24 for Plasma Total EPA + Total DHA.

2. Clinical Question

The study was open-label, so participants and study personnel were not masked according to the registry. Its stated primary objective is pharmacokinetic rather than a clinical outcome comparison. The registry record does not report a separate primary question about preventing pancreatitis or changing symptoms, nor does it provide clinical outcome results for those questions.

3. Trial Design

The registry describes a randomized, open-label crossover study in Phase 1. Enrollment was 30, and the study status is completed. The record gives a start date of 2014-10 and a primary completion date of 2015-07. The sponsor is AstraZeneca.

In a crossover design, participants receive more than one treatment in an assigned order, so treatment comparisons can be made within the study’s crossover structure. The registry lists four sequence labels: Epanova 4 g followed by Lovaza 4 g, Lovaza 4 g followed by Epanova 4 g, Epanova 2 g followed by Lovaza 4 g, and Lovaza 4 g followed by Epanova 2 g. The record does not provide further operational details about sequence allocation, period effects, or the randomization procedure.

The primary endpoint timeframe describes follow-up for up to 12 weeks. Each treatment had a 4-week duration, with a 4-week wash-off period in between. The primary objective was to compare relative bioavailability after a single dose of Epanova at 2 g or 4 g, or Lovaza at 4 g. The listed primary analysis population is the pharmacokinetic population.

The open-label design means masking was absent. For a pharmacokinetic outcome, laboratory measurements may be less directly influenced by expectations than subjective assessments, but the registry does not report measurement procedures or safeguards, so their details cannot be assessed. The record also does not describe the protocol’s sample-size justification, planned interim analyses, or other design features beyond those stated above.

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: No applicable

Period 1 (Visit 4b, Week 4)

Reasons for not completing this period:

Period 2 (Visit6b, Week 12)

Analysis population for the primary outcome, as defined in the registry: PK population

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 Baseline-adjusted AUC0-24 for Plasma Total EPA + Total DHA, measured over participants were followed for the duration of study, up to 12 weeks, each treatment having a 4-week duration and a 4-week wash off period in between..

The primary endpoint is baseline-adjusted area under the plasma concentration-versus-time curve from time 0 to 24 hours after the start of the meal for total EPA plus total DHA. The registry uses the abbreviation AUC0-24 and gives the unit as hr*nmol/mL. In general, an area-under-the-curve measure represents exposure across a defined period, combining information from measured concentrations over time rather than relying on a single concentration reading.

Baseline adjustment accounts for the starting measurement when summarizing the post-dose concentration profile. The endpoint is reported using geometric means, with geometric coefficient of variation as the dispersion measure. In general, geometric summaries are useful when measurements are positive and their variation is better represented on a multiplicative scale. The coefficient of variation describes relative variability around the geometric mean; it is not a confidence interval for the treatment comparison.

The endpoint fits the stated objective of comparing relative bioavailability: it measures plasma exposure to EPA and DHA over the specified post-meal window. It does not directly measure a clinical outcome. The record reports two comparisons against Lovaza 4 g, one for each Epanova dose, with mixed-model geometric mean ratios, confidence intervals, and p-values. The posted primary endpoint is distinct from the separately listed secondary outcomes for EPA and DHA exposure measures and from the study’s broader objectives concerning other fatty acids.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the crossover design contribute?

In general, a crossover study gives participants more than one treatment in a sequence. A comparison can then use information from the same participant under different treatments, which can reduce the impact of stable differences between people. That advantage depends on appropriate handling of treatment sequence and period, and on the assumptions of the analysis. The registry labels four treatment sequences but does not report details of the model terms, carryover assessment, or sequence-specific analysis. The stated 4-week wash-off period is part of the trial description; the record does not provide evidence about whether it was sufficient for every measured analyte.

What does a geometric mean ratio describe?

In general, a geometric mean ratio expresses one treatment’s geometric mean relative to another’s. A ratio is suited to comparing positive measurements when proportional differences are meaningful. The registry reports the contrasts as percentages. For Epanova 4 g versus Lovaza 4 g, the estimated ratio is 160.05%; for Epanova 2 g versus Lovaza 4 g, it is 121.61%. The interpretation must account for uncertainty, not just the point estimate. The registry reports 95% confidence intervals for both comparisons, and each interval includes the equal-exposure benchmark.

How should confidence intervals be read?

In general, a confidence interval gives a range of values that are reasonably compatible with the data and the stated statistical procedure. It is not a guarantee that the true value lies in that particular interval, and it does not assign a probability to the treatment effect being within the interval. A wide interval signals that the estimate is not precise. Here, the intervals run from 64.71% to 395.82% for Epanova 4 g and from 49.17% to 300.76% for Epanova 2 g. Both cover values on either side of equal exposure, so the point estimates should not be read as definitive evidence of greater exposure.

What does a p-value contribute?

In general, a p-value measures how compatible the observed data are with a specified null model, under the assumptions of the analysis. It is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. The reported p-values are 0.270 and 0.637. They do not provide strong evidence against equal exposure under the relevant model. The confidence intervals add useful context by showing the range of effects compatible with the data and emphasizing the limited precision.

Why report geometric coefficient of variation?

In general, a coefficient of variation expresses variability relative to the scale of the measurement. A geometric coefficient of variation is used with geometric summaries and describes relative spread. The registry reports values of 121.7 for Epanova 2 g, 104.6 for Epanova 4 g, and 80.0 for Lovaza 4 g. These figures describe variability in the arm summaries; they do not quantify uncertainty in the between-treatment ratios. Confidence intervals for the ratios address that separate question.

Why can over-dispersion matter?

In general, over-dispersion occurs when variability in the observed data is greater than a statistical model expects. If it is not appropriately handled, standard errors can be too small and confidence intervals too narrow. Pharmacokinetic measurements can vary considerably between observations, which is one reason variability summaries and model-based uncertainty deserve attention. The registry names a mixed-model analysis but does not report enough model detail to assess how its variance structure was specified or whether over-dispersion was considered. The arm-level geometric coefficients of variation indicate relative spread, but they do not establish how the model handled that spread.

What does the mixed-model label tell the reader?

The registry identifies the comparison method as Mixed Models Analysis. In general, mixed models can represent correlated observations and structured sources of variation, which can be relevant when participants contribute observations under multiple treatments or periods. The method name alone does not reveal the exact model, included covariates, transformations, handling of period or sequence, or assumptions used. Those details are not reported in the registry record, so the analysis cannot be independently reconstructed from the method label alone.

How do planning illustrations differ from observed power?

In general, a sample-size planning illustration asks how many participants per arm might be needed under specified assumptions to detect a chosen effect with a chosen power. The registry description of the planning approach states a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and effects defined as fractions of the observed effect. These are planning assumptions, not evidence about the trial’s original design rationale. The illustrations are not observed power: observed power is not a useful substitute for interpreting the estimated effect and its confidence interval. The protocol’s sample-size justification is not reported in the registry record.

10. Analyses Reported by the Sponsor

The registry reports the primary contrasts as Mixed Models Analysis results, with geometric mean ratios, confidence intervals, and p-values. The Epanova 4 g comparison is reported as 160.05%, with a 95% confidence interval of 64.71% to 395.82% and p = 0.270. The Epanova 2 g comparison is reported as 121.61%, with a 95% confidence interval of 49.17% to 300.76% and p = 0.637.

The registry record does not flag a discrepancy between the sponsor analysis and the recalculated results. The sponsor’s model-based analysis is therefore the relevant reported analysis for these comparisons. The arm-level geometric means and variation measures are descriptive summaries; they do not reproduce the crossover model contrasts or their uncertainty. No recalculated contrast values are stated in the registry record, so they cannot be compared numerically with the sponsor’s estimates.

The interpretation remains bounded by the posted endpoint and analysis. The contrasts concern baseline-adjusted plasma exposure in the pharmacokinetic population. They do not establish a difference in clinical outcomes, and the intervals indicate considerable uncertainty around each estimate. The record does not provide enough detail about the model specification to independently verify every analytic choice.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists six secondary outcome names, eight outcome measures, and 16 analyses. The primary endpoint has two reported treatment comparisons. Considering multiple endpoints and comparisons raises a multiplicity question: repeated testing can increase the chance that at least one result appears notable by chance. In general, prespecified adjustment or a clear testing hierarchy can help control that issue, while exploratory analyses may instead be interpreted with caution. The registry record does not report whether multiplicity adjustment was planned or applied, or how the primary and secondary analyses were ordered.

The record identifies the pharmacokinetic population for the primary endpoint, but does not define its inclusion criteria or report how many participants contributed data to each primary comparison. It also does not report how missing measurements, withdrawals, or incomplete treatment sequences were handled. Those details matter in a crossover study because missing observations can affect which within-participant comparisons contribute to an estimate. The enrollment figure should not be treated as the analysis count for every endpoint.

The registry indicates that values were missing or not estimable for a two-arm comparison for another endpoint. It does not identify that endpoint in the brief note, explain why the comparison could not be estimated, or describe how the issue was handled. This note concerns an outcome other than the posted primary endpoint and should not be used to infer that the primary analysis itself was missing or not estimable.

The record does not report interim analyses, a missing-data strategy, or a detailed statistical analysis plan. In general, those items help readers distinguish prespecified analyses from additional comparisons and assess the potential influence of incomplete data. Without them, interpretation should rely on the stated endpoint, reported estimates, confidence intervals, and the limitations in available analysis detail.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports serious adverse events affected as zero in each listed treatment group. It also gives non-serious adverse event counts by groups labelled I and II. For Epanova 2 g, the figures are 2 affected among 3 at risk in group I and 3 among 4 in group II. For Epanova 4 g, they are 3 among 4 in group I and 2 among 2 in group II. For Lovaza 4 g, they are 5 among 8 in group I and 1 among 7 in group II.

These are group-specific counts, not a single person-level total across the crossover study. The record labels the groups I and II but does not explain further how those labels correspond to treatment periods or analysis sets. Because a participant may contribute information under more than one treatment in a crossover design, the group counts should not be added to infer the number of people with an event across the study. The registry record does not report deaths.

Zero reported serious adverse events means none were recorded as affecting participants in the listed groups; it does not establish that serious risks are absent. The record’s serious-event terms list is empty. The counts do not describe event type, timing, severity, or relation to treatment, and the registry excerpt does not provide those details. The small group-specific numbers also limit precision. These safety figures should be read as descriptive reporting, not as a definitive comparison of safety between treatments.

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

The posted results provide estimates of relative plasma exposure to total EPA plus total DHA after the specified treatments. Both Epanova-to-Lovaza point estimates are above the equal-exposure benchmark, but the confidence intervals include that benchmark and extend widely in both directions. The results therefore do not establish a clear difference in the primary pharmacokinetic endpoint for either comparison.

That conclusion is about statistical evidence in the reported analyses, not proof that the treatments have identical exposure. An imprecise estimate can be compatible with a range of underlying differences. The p-values and confidence intervals should be considered together: the p-values do not show strong evidence against equal exposure, while the intervals show that substantial uncertainty remains about the magnitude and direction of the comparison.

The endpoint is a measure of plasma exposure over a defined period, not a direct clinical outcome. The registry record does not report whether the exposure estimates correspond to a difference in pancreatitis events, symptoms, or other health outcomes. The study’s stated purpose and primary outcome are pharmacokinetic. No claim about clinical benefit or lack of benefit follows from these results alone.

The findings also apply to the analysis population and study context recorded for this trial. The record describes a Phase 1 randomized open-label crossover study, with enrollment of 30 and a pharmacokinetic primary analysis population. It does not provide detailed participant characteristics, analysis counts for the primary endpoint, or enough model specification to assess every assumption. The posted safety counts are descriptive and likewise do not establish comparative safety.

In general, a useful statistical reading separates the estimated effect, its precision, and the clinical meaning of the endpoint. Here, the estimated ratios are available, precision is limited by broad confidence intervals, and clinical meaning beyond plasma exposure is not established in the registry record.

14. Why This Trial Matters Statistically

This trial illustrates why a point estimate and its uncertainty need to be read together. The reported geometric mean ratios may draw attention because both estimates are above the equal-exposure benchmark. Yet the confidence intervals extend across that benchmark and cover a broad range of plausible relative exposures. The point estimates alone would give an incomplete account of the evidence.

The crossover structure also matters to interpretation. In general, within-participant comparisons can be informative because they account for stable differences between people. At the same time, sequence, period, and possible carryover considerations can shape the analysis. The record names a mixed-model method and lists treatment sequences, but does not report enough details to evaluate the model specification or these assumptions fully.

The endpoint makes the distinction between pharmacokinetic and clinical evidence clear. AUC0-24 summarizes exposure over time and is suited to the stated relative-bioavailability objective. It does not answer whether treatment changes a clinical outcome. Separating these questions helps prevent a pharmacokinetic comparison from being interpreted as proof of a clinical effect.

Finally, the available safety counts show why denominators and study structure need attention. The registry reports groups labelled I and II, and the crossover design means those entries should not be casually added into a single participant count. The study is a useful example of how uncertainty, analysis population, outcome type, and design all constrain what can be concluded from posted results.

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.

The registry record provides limited detail about the analysis. It names a mixed-model method but does not describe the model specification, covariates, treatment and period terms, treatment-sequence handling, or assumptions. Those omissions limit independent assessment of how the crossover design was represented and whether alternative modeling choices could affect the estimates.

The record does not state how many participants contributed to each primary comparison, how incomplete profiles were handled, or whether withdrawals or missing measurements influenced the analysis. Although enrollment was 30, enrollment is not the same as the number analyzed for a particular endpoint. The pharmacokinetic population is named but not further defined.

Generalizability is also difficult to judge because participant characteristics and detailed eligibility criteria are not reported in the record excerpt. The trial is described as Phase 1 and open-label, and the primary outcome is pharmacokinetic. The registry does not report clinical outcome results that would connect the exposure estimates to pancreatitis or other health outcomes.

The two primary comparisons have broad confidence intervals, indicating substantial uncertainty in the estimated ratios. The record does not report a multiplicity strategy despite listing several outcome measures and analyses. The missing or not-estimable note for another endpoint is not explained, and the record does not state whether interim analyses were planned or performed. These limitations do not invalidate the posted results, but they narrow the conclusions that can be drawn from them.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: An Open-Label Crossover Study to Compare the Relative Bioavailability, Efficacy and Safety of Epanova® and Lovaza® in Men and Women With a History of Pancreatitis — NCT02189252.
• Registry record last updated 2016-06-20; status last verified 2016-05.
• Results first posted 2016-06-20.
• 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

NCT02189252 is a completed, randomized, open-label Phase 1 crossover study sponsored by AstraZeneca. It enrolled 30 and compared Epanova 2 g and 4 g with Lovaza 4 g in men and women with a history of pancreatitis. Severe hypertriglyceridemia is listed as the condition.

The primary endpoint was baseline-adjusted AUC0-24 for plasma total EPA plus total DHA, reported as geometric means in the pharmacokinetic population. The registry reports mixed-model geometric mean ratios of 160.05% for Epanova 4 g versus Lovaza 4 g and 121.61% for Epanova 2 g versus Lovaza 4 g. Both 95% confidence intervals include equal exposure and are wide, so the comparisons do not establish a clear difference in the primary endpoint.

The record also reports group-specific adverse-event counts. Further details about analysis methods, missing data, multiplicity, and participant characteristics are not reported in the registry record.