A Study to Evaluate ABP 938 and Aflibercept (Eylea®) in Participants With Chorioretinal Vascular Disease (CVD): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Evaluate ABP 938 and Aflibercept (Eylea®) in Participants With Chorioretinal Vascular Disease (CVD) (NCT05704725), a randomized phase 3 study. The primary outcome is Percentage of IVT Injections Successfully Administered.

2. Clinical Question

This is a question about the administration process and syringe use, rather than a comparison of longer-term clinical outcomes. The primary endpoint records the specialist’s yes-or-no answer to a defined question about safe and effective delivery of the prescribed dose. The registry reports a percentage and a Clopper-Pearson 95% confidence interval for each arm.

The study also lists ocular treatment-emergent adverse events, serious ocular treatment-emergent adverse events, and non-ocular serious treatment-emergent adverse events as secondary outcomes. Those outcomes provide a safety context, but the posted primary result addresses injection administration. The registry record does not report a formal statistical test comparing the arms on the primary endpoint or a prespecified margin for judging similarity.

3. Trial Design

This completed Phase 3 study used randomized allocation, an open-label approach, and a parallel two-arm design. The registry lists an enrollment of 49 and study dates from 2023-01-23 to primary completion on 2023-02-17. Amgen is listed as the sponsor. The population is described as participants with chorioretinal vascular disease.

Participants were assigned in a stated 2:1 ratio to a single intravitreal injection of ABP 938 in a prefilled syringe or a single injection of aflibercept in a prefilled syringe. The registry’s primary objective was to assess whether retina specialists could successfully administer the prescribed dose using each syringe. The study was open label, so the record does not describe masking of participants or study personnel.

The posted primary endpoint results include 32 participants in the ABP 938 arm and 16 in the aflibercept arm. The endpoint population is described as FAS, with only participants who had available data included. These arm-specific counts are the numbers associated with the posted endpoint results; the registry record does not explain how they relate to the overall enrollment or describe reasons for any unavailable endpoint data.

The primary endpoint was assessed on the day of the intravitreal injection, identified as study day 1. The record lists three secondary safety outcomes, but the available results do not provide separate arm-level estimates for each of those outcomes. Further details about eligibility criteria, site procedures, or additional study conduct 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.

Before assignment: Participants were randomized in a ratio of 2:1 to receive either a single intravitreal (IVT) injection of ABP 938 in a pre-filled syringe (PFS) or a single injection of aflibercept in a PFS.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS (only participants with available data are included.)

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: Full Analysis Set (FAS): All randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of IVT Injections Successfully Administered, measured over Day of IVT injection (study day 1).

The primary endpoint is the percentage of intravitreal injections successfully administered on the day of injection. Success was defined by the administering retinal specialist answering yes to whether the prefilled syringe allowed safe and effective administration of the prescribed dose. The endpoint therefore captures the clinician’s assessment of the injection process, not a measure of visual function, disease progression, or longer-term treatment response.

This kind of endpoint is relevant to a study whose stated objective is to evaluate administration using a prefilled syringe. A binary yes-or-no definition makes it possible to report the proportion of recorded injections meeting the stated criterion. The registry reports that proportion with a Clopper-Pearson 95% confidence interval in each arm.

The endpoint is limited to the judgment recorded for the injection. It does not, by itself, establish how well the treatment controls the underlying disease or whether one product has a clinical advantage. Nor does the posted description explain how the specialist’s judgment was standardized across clinicians. The endpoint’s timeframe is the day of the injection, identified as study day 1.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a binary observation for each injection: the specialist answered yes or did not answer yes to the question defining successful administration. The registry reports the percentage meeting the success definition separately for each randomized arm. A percentage is useful because it expresses the observed frequency relative to the endpoint data available in that arm. The arm-level counts associated with the posted results are 32 for ABP 938 and 16 for aflibercept.

What does a confidence interval express?

In general, a confidence interval describes sampling uncertainty around an estimate, using a specified statistical procedure. If the study were repeated under comparable conditions, intervals formed using the same method would vary from sample to sample. A 95% confidence level refers to the long-run behavior of that procedure; it is not a statement that there is a 95% probability that the particular interval contains the fixed underlying value.

The registry identifies the intervals as Clopper-Pearson 95% confidence intervals. In general, the Clopper-Pearson method is an exact interval procedure for a binomial proportion. It can be used when the outcome is recorded as success or failure, as in this endpoint. The posted intervals are 89.1% to 100.0% for ABP 938 and 79.4% to 100.0% for aflibercept. The registry does not report an additional model-based adjustment for these arm-level estimates.

Why are the intervals different in width?

In general, estimates based on fewer observations tend to have greater sampling uncertainty, all else being equal. The posted endpoint count is smaller for aflibercept than for ABP 938, and its interval is wider. This is a description of precision, not proof that the treatments differ. Interval width also depends on the observed result and the interval method. Because the estimates are bounded proportions, confidence limits cannot extend beyond the allowable percentage scale.

Is there a formal comparison between arms?

The registry record reports arm-specific percentages and confidence intervals, but no analyses are posted for a two-arm comparison. It does not report a p-value, a comparative confidence interval, an adjusted model, or a decision rule for equivalence or noninferiority. Comparing the arm-specific intervals informally is not a substitute for a prespecified between-arm analysis. In general, overlap or non-overlap between separate confidence intervals does not by itself answer every question about a treatment contrast.

The record’s sample-size assumptions describe planning illustrations based on a stated fraction of the observed effect, with a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. These assumptions concern illustrative planning, not the trial’s own design justification. They should not be interpreted as observed power. The registry record does not report the rationale used to choose the study enrollment or a statistical plan for the primary analysis.

How should the endpoint definition affect interpretation?

The endpoint depends on a specialist’s answer to a specific yes-or-no question about safe and effective administration of the prescribed dose. That definition makes the event being counted explicit, but it also means the result reflects the specialist’s recorded assessment. The registry record does not report details on assessment training, inter-rater agreement, or procedures for resolving uncertain answers. As a general matter, subjective or judgment-based endpoints can be influenced by how consistently the question is applied.

The study is open label. In general, awareness of the assigned product can affect assessments that require judgment, although the record does not establish that such influence occurred in this study. Random allocation can help balance prognostic factors across arms on average, but it does not remove the possibility of measurement differences or sampling uncertainty. The registry does not report an analysis addressing these considerations.

The endpoint is measured on the day of injection, so the statistical result concerns administration at that time. It does not estimate a longer-term clinical outcome. The confidence intervals quantify uncertainty for the measured proportion, not uncertainty about outcomes the endpoint was not designed to measure. The distinction matters when interpreting an apparently uniform observed result: a high observed percentage on this endpoint is not a statistical estimate of every aspect of safety, effectiveness, or usability.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists three secondary safety outcomes: participants with ocular treatment-emergent adverse events in the study eye, participants with serious ocular treatment-emergent adverse events in the study eye, and participants with non-ocular serious treatment-emergent adverse events. The record does not report a multiplicity adjustment, a hierarchy among these outcomes, or a plan for controlling the overall chance of false-positive findings across multiple analyses. It also reports no posted analyses for a two-arm comparison.

Multiplicity is a general statistical concern when several hypotheses are tested. The more questions tested, the more opportunity there can be for a chance-positive finding if each is judged separately. Whether and how that concern should be addressed depends on the study’s objectives and analysis plan; that plan is not reported in the registry record. The listed secondary outcomes can therefore be described as outcomes of interest, but the available record does not support claims about confirmatory testing across them.

The primary endpoint population is described as FAS, including only participants with available data. This indicates that the posted endpoint calculation does not include participants without available endpoint data. The number of missing observations, reasons for missingness, and method for handling missing data are not reported in the registry record. It is therefore not possible to assess from this record whether unavailable data could affect the estimated percentages.

The registry record also does not report a protocol-based sample-size justification or the detailed statistical analysis plan. Its planning assumptions are illustrative and should not be read as evidence of observed power or as the original rationale for enrollment.

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 5% reporting threshold.

Most frequent non-serious adverse event terms

The adverse-event summary records zero serious events among 32 participants at risk in the ABP 938 arm and zero among 16 participants at risk in the aflibercept arm. It also records zero deaths in each arm. The summary lists two other adverse events in each arm. These are descriptive counts in the registry record, not a formal statistical comparison.

The study lists separate secondary outcomes for ocular treatment-emergent adverse events, serious ocular treatment-emergent adverse events, and non-ocular serious treatment-emergent adverse events. However, the posted summary does not give separate arm-level results for each listed outcome, or provide event terms for the serious events. The overall event counts should not be assigned to a specific category unless the registry reports that breakdown.

Zero recorded serious events does not establish that serious events are impossible or that the arms have the same safety profile. Safety estimates based on a limited set of observations can be imprecise, and the record does not report a formal comparison or detailed event follow-up. These figures describe what was recorded in this study, rather than proving the presence or absence of a risk in other settings.

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

The primary result addresses a focused procedural question: whether the retinal specialist judged that the assigned prefilled syringe allowed safe and effective administration of the prescribed dose on the injection day. The posted percentage is 100.0% in both arms, with arm-specific 95% confidence intervals. This means all injections included in the posted endpoint results met the registry’s stated success definition.

The result does not establish that ABP 938 and aflibercept are equivalent. A formal equivalence conclusion would require a prespecified margin and an appropriate comparative analysis; neither is reported in the registry record. Nor does the absence of a difference in the displayed percentages prove that there is no difference in the underlying populations. The record provides no formal test or comparative estimate for the primary outcome.

The endpoint is not a measure of disease control, vision, or longer-term clinical benefit. It records a specialist’s assessment of administration using a syringe. The result therefore should not be extended to outcomes that were not measured by this primary endpoint. Similarly, the adverse-event counts are limited descriptive information, not a comprehensive comparison of safety.

Precision also matters. The confidence intervals extend from 89.1% to 100.0% for ABP 938 and from 79.4% to 100.0% for aflibercept. These ranges show that the observed percentages are estimates, not exact measures of the underlying success rates. The relatively broader interval for aflibercept indicates more uncertainty around that arm’s percentage in the posted results.

Finally, the endpoint population includes only participants with available data, and the registry record does not explain unavailable data. Together with the short endpoint timeframe, absence of a reported comparative analysis, and lack of detailed secondary safety results, this limits the conclusions that can be drawn from the posted results. The findings are best read as descriptive evidence about recorded injection administration in this study.

13. Why This Trial Matters Statistically

This trial is statistically distinctive because its primary question concerns administration with a prefilled syringe, not a conventional clinical endpoint such as a change in disease status. The binary definition—whether a retina specialist answered yes to a specific question—makes the measured outcome direct and interpretable. It also makes clear that interpretation depends on the consistency and meaning of that assessment.

The result illustrates why an observed percentage alone does not convey precision. Both arms have a posted estimate of 100.0%, yet their confidence intervals differ in width. The intervals remind readers that an observed proportion is subject to sampling uncertainty, and that the amount of uncertainty can differ between arms. In general, this is particularly important when an outcome is near the boundary of its possible scale.

The study also demonstrates the distinction between parallel arm summaries and a comparative statistical conclusion. Randomization is part of the design, but a formal claim of equivalence or noninferiority still requires an appropriate analysis and decision framework. The registry record does not report those elements for the primary endpoint. The matching percentages alone cannot supply them.

Finally, the record’s available-data population and limited safety breakdown show why endpoint definitions, analysis populations, and reporting completeness matter. A procedural success measure can answer a narrow question about administration while leaving longer-term clinical outcomes and detailed safety comparisons unresolved. Keeping those boundaries explicit helps readers understand both the value and the limits of the observed result.

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 primary outcome is based on the administering specialist’s answer to a subjective yes-or-no question. The registry record does not report how assessments were standardized across specialists, whether responses were independently reviewed, or how disagreement or uncertain responses were handled. Because the study was open label, assessors were not described as masked to the assigned product; the record does not establish whether awareness influenced responses.

The endpoint is assessed on the day of injection and is limited to syringe-assisted administration. It does not measure longer-term clinical outcomes. The registry record also does not report a formal comparative analysis, an equivalence or noninferiority margin, or a statistical plan for interpreting the arm-level results. As a result, equal observed percentages cannot be treated as evidence of formal equivalence.

Only participants with available endpoint data are included in the stated analysis population. The record does not report the number or reasons for missing endpoint observations or how missingness was handled. This leaves uncertainty about the extent to which the posted results represent all randomized participants.

For safety, the record lists secondary outcomes but does not post separate arm-level estimates for each one or event-term detail. The brief summary of serious events and other adverse events does not replace a detailed safety analysis. Additional information about eligibility, study procedures, and the study’s sample-size rationale is not reported in the registry record.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: A Study to Evaluate ABP 938 and Aflibercept (Eylea®) in Participants With Chorioretinal Vascular Disease (CVD) — NCT05704725.
• Registry record last updated 2025-02-14; status last verified 2025-02.
• Results first posted 2025-02-14.
• 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

NCT05704725 is a completed, randomized, open-label, parallel Phase 3 study sponsored by Amgen. It enrolled participants with chorioretinal vascular disease and compared a single intravitreal injection of ABP 938 in a prefilled syringe with a single injection of aflibercept in a prefilled syringe, using a stated 2:1 allocation ratio.

The primary endpoint was the percentage of injections judged successfully administered on the day of injection. The posted percentage was 100.0% in each arm, with a Clopper-Pearson 95% confidence interval of 89.1% to 100.0% for ABP 938 and 79.4% to 100.0% for aflibercept. The registry record does not report a formal between-arm analysis for this endpoint.

The record lists three secondary safety outcomes and reports zero serious events and zero deaths in each arm, with two other adverse events in each arm. It does not provide separate results for each listed safety outcome. Only participants with available endpoint data were included, and reasons for unavailable data are not reported.