A Study of Faricimab (RO6867461) in Participants With Center-Involving Diabetic Macular Edema: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Faricimab (RO6867461) in Participants With Center-Involving Diabetic Macular Edema (NCT02699450), a randomized phase 2 study. The primary outcome is Mean Change From Baseline in BCVA Letter Score at Week 24, in Treatment-Naive Participants.

2. Clinical Question

The primary outcome is a change score: it compares the Week 24 letter score with the participant’s baseline score. The registry explains that higher scores reflect better visual acuity and that a gain from baseline indicates improvement on this test. The study was designed as a randomized, active-comparator, double-masked, parallel-group Phase 2 trial. Treatment-naive participants were assigned among the three arms, while participants previously treated with intravitreal anti-VEGF were assigned between the ranibizumab arm and the 6 mg faricimab arm. The primary question applies only to the treatment-naive population, rather than to every participant enrolled in the trial.

3. Trial Design

This was a randomized, double-masked, active-comparator, parallel-group Phase 2 study in participants with center-involving diabetic macular edema. The registry reports enrollment of 229 participants. The study had a 36-week duration, with a treatment period of 20 weeks followed by an observational period of up to 16 weeks. Only one eye was selected as the study eye; when both eyes met eligibility criteria, the eye with worse baseline best corrected visual acuity was selected.

The three arms were 0.3 mg ranibizumab, 1.5 mg faricimab, and 6 mg faricimab. Each regimen was administered intravitreally every fourth week up to Week 20, for a total of 6 administrations. Participants meeting pre-specified criteria could receive a single 0.3 mg ranibizumab dose and exit the study; the registry record does not give the criteria in full. Treatment-naive participants were randomized in a 1:1:1 ratio across the arms. Participants previously treated with intravitreal anti-VEGF were randomized in a 1:1 ratio to ranibizumab or 6 mg faricimab.

The primary outcome was assessed at Week 24, in the treatment-naive population. The primary analysis population was limited to participants who had received at least one dose and had non-missing assessments at baseline and Week 24. The analysis grouped participants by their randomized assignment. The primary outcome’s arm-level reported analysis counts were 49 for ranibizumab, 49 for 1.5 mg faricimab, and 44 for 6 mg faricimab. These analysis counts are not the same as the overall enrollment figure.

The registry describes masking of participants and study personnel, and specifically notes that the visual-acuity examiner was masked to study-eye and treatment assignment. The examiner also did not perform other direct-care tasks and did not have access to letter scores from previous visits. The registry record does not report the detailed randomization procedure, the method used to maintain masking, or the sample-size justification.

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: A total of 229 patients were randomized, but two participants randomized to Arm C: 6 mg Faricimab were excluded from the analysis populations due to Good Clinical Practice (GCP) non-compliance at a single site.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Efficacy Population: all treatment-naive participants who had received at least one dose of study drug grouped by treatment assigned at randomization (did not differ from actual treatment during the study). This analysis only included participants with non-missing assessments at Baseline and Week 24.

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: Two participants randomized to Arm C: 6 mg Faricimab were excluded from the analysis populations due to Good Clinical Practice (GCP) non-compliance at a single site.

6. Primary Endpoint

The registry lists the primary outcome as Mean Change From Baseline in BCVA Letter Score at Week 24, in Treatment-Naive Participants, measured over Baseline, Week 24.

The primary endpoint is mean change from baseline in best corrected visual acuity letter score at Week 24 among treatment-naive participants. Visual acuity was measured at a starting test distance of 4 meters using one of three distance-acuity charts, with assessment by a trained and certified examiner masked to study drug assignment. The score ranges from 0 to 100, with a higher score representing better measured acuity; a gain from baseline indicates improvement on the letter-score test.

A change score is useful because it expresses the Week 24 assessment relative to each participant’s starting measurement, rather than describing only the final score. The study reports least-squares means from a linear mixed-effects model for repeated measurements. These model-based means account for the analysis approach described in the registry and are not simple averages of the posted arm-level values. The comparison between arms is expressed as a difference in least-squares means, with the sign indicating the direction of the difference according to the stated comparison order.

The endpoint is an assessment of visual-acuity letter score, not a complete description of every aspect of diabetic macular edema or of a participant’s overall experience. The registry also lists secondary visual-acuity outcomes and measures of retinal fluid, but those are distinct outcomes and should not be substituted for the primary endpoint. The primary analysis used only treatment-naive participants with non-missing baseline and Week 24 assessments.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary analysis is described as a linear mixed-effects model for repeated measurements. Its reported quantities are least-squares means for each treatment arm and differences between those means. General explanation: a repeated-measures model uses the pattern of available measurements across visits rather than treating each visit as an unrelated analysis. A mixed-effects framework can account for the correlation among measurements from the same participant and can estimate group means while incorporating the model’s specified terms. The exact model terms and covariance structure are not reported in the registry record.

What does a least-squares mean represent?

A least-squares mean is a model-based estimate of the average outcome for a group. It may differ from the ordinary arithmetic mean because it comes from the fitted statistical model. The registry identifies the primary outcome parameter as a least-squares mean and reports differences in least-squares means for the two comparisons with ranibizumab. It does not report enough detail to reconstruct the model independently, including all covariates or other terms that may have been included.

The reported arm estimates were 10.3 letters for ranibizumab, 11.7 for 1.5 mg faricimab, and 13.9 for 6 mg faricimab. The sponsor’s reported comparison of ranibizumab with 1.5 mg faricimab gave a difference of 1.4 letters, and the comparison with 6 mg faricimab gave a difference of 3.6 letters. The reported confidence intervals are 80% intervals. A confidence interval gives a range around an estimate that reflects sampling uncertainty under the fitted analysis; its width depends on the data and the model. It is not a range of individual participant outcomes.

Why does the analysis population matter?

The stated primary population included treatment-naive participants who received at least one dose of study drug. Participants were grouped according to their randomized assignment, which the registry says did not differ from actual treatment during the study. The analysis further required non-missing assessments at baseline and Week 24. Thus the estimates apply to that analysis definition. They are not automatically estimates for every randomized participant or for the previously treated population.

The primary endpoint is a change from baseline, so both the starting score and Week 24 score contribute to whether the value can be assessed. The registry states that missing values were not imputed. General explanation: imputation would fill in missing observations using a specified procedure; not imputing means the analysis did not replace absent values with modeled substitutes as a separate preprocessing step. The model’s treatment of available repeated observations and the eligibility rule for the endpoint analysis remain relevant. The registry states that missingness was assumed to be at random, but the detailed missing-data diagnostics and sensitivity analyses are not reported in the registry record.

How should the confidence intervals and p-values be read?

The estimated difference of 1.4 letters for 1.5 mg faricimab versus ranibizumab had an 80% confidence interval from -0.6 to 3.4 and p = 0.37. The interval includes zero. The estimated difference of 3.6 letters for 6 mg faricimab versus ranibizumab had an 80% confidence interval from 1.5 to 5.6 and p = 0.03. That interval does not include zero. General explanation: a p-value summarizes how compatible the observed data are with a specified null model, under the assumptions of the analysis. It does not measure the size or clinical importance of an effect. The registry does not state a multiplicity adjustment for these primary comparisons, so the reported p-values should not be presented as adjusted results.

The confidence intervals are 80%, rather than the more commonly encountered 95% level. General explanation: a narrower coverage level generally produces a narrower interval for the same data and method. It is important to keep the stated confidence level attached to the result; the interval should not be silently treated as a 95% interval. The methods description does not report a separate confidence-interval procedure beyond the model-based analysis.

How do the binary secondary comparisons differ?

Some secondary outcomes are percentages of participants with a specified finding, including the presence of subretinal or intraretinal fluid at Week 24. The registry reports Fisher exact analyses for these outcomes. A separate set of simple, unadjusted re-calculations is based on posted summary counts and compares proportions through risk differences. General explanation: a risk difference expresses the difference between two proportions, while Fisher’s exact test evaluates compatibility of a two-group count table with a null association. These methods use the observed counts and are not the repeated-measures model used for the primary visual-acuity endpoint.

Those binary re-calculations are unadjusted, and their intervals and p-values should not be confused with the sponsor’s reported analyses or the model-based primary results. They do not account for the full trial design or for multiplicity across secondary outcomes. Where the registry includes sponsor analyses, those are the trial’s reported analyses; the re-calculations are descriptive checks based on posted summaries, not replacements for the sponsor’s analysis.

What do the planning illustrations mean?

The planning illustrations are based on observed effects and are not the trial’s original sample-size justification. Their stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The effect to detect is expressed as a stated fraction of the observed effect. The illustrations give participants per arm for 80% power and participants per arm for 90% power. General explanation: such calculations show how a hypothetical future study’s size would change under selected effect and design assumptions. They are not observed power, and they do not show the probability that the present study detected an effect. The trial’s own sample-size justification is not reported in the registry record.

The registry reports 46 outcome measures, 36 analyses, and 45 posted secondary outcomes. It does not describe an overall statistical analysis plan for controlling multiplicity across these many analyses, nor does it report whether interim analyses were planned or conducted. Those omissions limit interpretation of the set of p-values as a whole.

10. Analyses Reported by the Sponsor

The primary comparisons are reported using the sponsor’s repeated-measures model, with differences in least-squares means, 80% confidence intervals, and p-values. The reported comparison of 1.5 mg faricimab with ranibizumab is 1.4 letters, with an 80% confidence interval from -0.6 to 3.4 and p = 0.37. The comparison of 6 mg faricimab with ranibizumab is 3.6 letters, with an 80% confidence interval from 1.5 to 5.6 and p = 0.03.

There is no separate simple re-calculation of the primary repeated-measures analysis in the posted results. Reconstructing that model from the arm-level estimates alone would not reproduce its repeated-measures structure or model terms, which are not fully reported. The sponsor’s model-based analysis is therefore the relevant reported analysis for the primary endpoint, rather than a calculation from arm summaries.

For binary secondary outcomes, the registry also reports Fisher exact analyses, while simple unadjusted re-calculations can be made from posted summary counts. Those are different analytic presentations and should not be treated as interchangeable: tests, estimands, and intervals can differ. No discrepancy is flagged in the registry record. The sponsor’s own analysis remains the basis for describing the reported trial result; the simple re-calculations are supplementary checks, not a substitute for the sponsor’s statistical analysis.

11. Multiplicity, Missing Data and Analysis Populations

The registry reports a primary endpoint and a broad set of additional outcomes, with 46 outcome measures, 36 analyses, and 45 posted secondary outcomes. The primary outcome includes two comparisons against the same ranibizumab group. The registry record does not report whether the primary comparisons were adjusted for multiplicity, whether a hierarchy was used, or how error rates were intended to be controlled across the secondary analyses. Consequently, the reported p-values should be read as the values for the individual analyses, not assumed to represent multiplicity-adjusted evidence.

The primary analysis excluded participants without non-missing baseline and Week 24 assessments. The registry says missing values were not imputed and that data were assumed to be missing at random. General explanation: missing-at-random assumptions concern whether missingness can be explained by observed information in the analysis; if missingness depends on unobserved outcomes even after accounting for observed information, estimates may be affected. The registry record does not report the extent or reasons for missing primary assessments, or sensitivity analyses under alternative assumptions. The practical effect of missing data therefore cannot be assessed from the reported information alone.

The defined population also matters. The primary analysis included treatment-naive participants who had received at least one dose and met the assessment requirement. It was not an analysis of all participants enrolled, and it does not answer the same question for previously treated participants. The reported assignment-based grouping supports interpretation by randomized group, but the available outcome analysis still depends on its stated non-missing assessment condition.

The trial’s sample-size justification, detailed statistical analysis plan, and interim-analysis procedures are not reported in the registry record. The planning illustrations based on observed effects are not evidence of the study’s original design rationale or of observed power. Without the unreported planning and multiplicity details, the strength of claims across multiple outcomes should remain cautious.

12. 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 3% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: pneumonia (3.6%), cardiac failure congestive (2.2%), cellulitis (2.2%), visual acuity reduced (1.8%), bronchitis (1.8%).

The registry reports serious adverse events and deaths by treatment arm. For ranibizumab, 9 participants were recorded with serious adverse events among 89 at risk, and 2 deaths were recorded. For 1.5 mg faricimab, the corresponding figures were 7 among 55 at risk and 1 death. For 6 mg faricimab, they were 8 among 80 at risk and 2 deaths. These are descriptive counts from the registry record; they do not establish that study treatment caused an event or death, and they are not by themselves a controlled comparison of safety.

The registry also lists serious-event terms and percentages, including pneumonia at 3.6%, cardiac failure congestive at 2.2%, cellulitis at 2.2%, visual acuity reduced at 1.8%, bronchitis at 1.8%, and gangrene at 1.8%. These figures summarize the listed terms but do not provide a complete clinical account of event timing, severity, relatedness to treatment, or participant-level overlap. The safety follow-up window and the adjudication procedures are not reported in the registry record.

Counts should be interpreted alongside the group-specific numbers at risk and the study’s design and follow-up. The posted figures are useful for describing recorded events, but the registry record does not provide enough information to infer comparative safety or causality.

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

The primary result describes model-estimated changes in a visual-acuity letter score at Week 24 among treatment-naive participants who met the analysis requirements. The reported estimates were positive in all three arms. In the sponsor’s comparison, the 1.5 mg faricimab estimate was higher than the ranibizumab estimate by 1.4 letters, but the 80% confidence interval included zero. The 6 mg faricimab comparison was 3.6 letters, with an 80% confidence interval that did not include zero and a reported p-value of 0.03.

These findings do not, on their own, establish what would occur in other populations or over a longer period. The primary analysis was limited to treatment-naive participants, one selected study eye, and the Week 24 outcome. The trial included previously treated participants, but their primary outcome is not the question addressed by this analysis. The record describes follow-up through Week 36, but a Week 36 primary result is not reported in the registry record.

The result also does not establish that every secondary outcome moved in the same direction, that the difference is clinically important, or that the comparisons were adjusted for multiple testing. The registry record does not provide a threshold for interpreting the clinical importance of a letter-score difference. Statistical uncertainty, the 80% confidence level, the analysis population, and the unreported multiplicity plan all matter when assessing the result.

Finally, the visual-acuity endpoint is one measure of outcome. It does not summarize safety, retinal-fluid findings, or all aspects of participants’ experience. Serious adverse-event counts are descriptive and do not establish causality. The most defensible interpretation is therefore tied to the exact analysis and follow-up reported, without extending it to claims about broader outcomes or populations.

14. Why This Trial Matters Statistically

This trial illustrates why the stated estimand and analysis population matter as much as the treatment labels. The primary endpoint is a model-based mean change in visual-acuity score, and its analysis is restricted to treatment-naive participants with assessments at baseline and Week 24. Reading the result as though it represented every enrolled participant or every study outcome would change the question being answered.

It also demonstrates the importance of distinguishing arm-level estimates from between-arm comparisons. The three least-squares means describe modeled changes in each group, while the difference in least-squares means directly addresses the comparison with ranibizumab. The uncertainty interval and p-value belong to that comparison, not to a simple visual comparison of the arm estimates.

The 80% confidence intervals make the stated uncertainty level especially important. Interpretations can shift when readers assume a different coverage level or overlook the interval’s relation to zero. Multiple dose comparisons and numerous secondary analyses add another statistical consideration: absent a reported multiplicity plan, individual p-values should not be treated as though the overall testing process has been controlled.

Finally, the trial offers examples of different outcome types in the same record. A repeated-measures model for visual-acuity change and Fisher exact analyses for binary fluid outcomes answer different statistical questions. Their estimates and tests should not be conflated. The registry’s missing details on model specification, missing-data sensitivity analysis, and sample-size rationale also show why transparent reporting is necessary for a complete statistical assessment.

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 primary result is restricted to the treatment-naive analysis population and requires non-missing baseline and Week 24 assessments. It therefore does not represent every enrolled participant by definition. The record states that missing values were not imputed and assumes missingness at random, but does not report the amount or reasons for missingness or sensitivity analyses testing other assumptions.

The reported confidence intervals are 80% intervals, and the registry does not report a multiplicity adjustment for the primary comparisons. The study includes multiple treatment comparisons and numerous secondary outcomes, while the overall procedure for controlling false-positive findings is not reported. The registry also does not report the full model specification, such as all included terms or the covariance structure, which limits independent reconstruction of the repeated-measures analysis.

The design was a Phase 2 study with a Week 24 primary endpoint and follow-up through Week 36. A Week 36 primary outcome result is not reported in the registry record. The trial’s sample-size justification and interim-analysis procedures are also not reported. The planning illustrations based on observed effects should not be read as the original design calculation or as evidence of observed power.

Only one eye was selected as the study eye, and the primary outcome concerns visual-acuity letter score rather than every dimension of diabetic macular edema. The registry record does not report detailed participant characteristics, the full criteria for receiving an additional ranibizumab dose and exiting the study, or detailed safety-event timing and relatedness. Those limits constrain assessment of applicability, treatment exposure, and the interpretation of adverse-event counts.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Faricimab (RO6867461) in Participants With Center-Involving Diabetic Macular Edema — NCT02699450.
• Registry record last updated 2020-09-25; status last verified 2020-08.
• Results first posted 2020-09-25.
• 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

The registry record describes a completed, randomized, double-masked, active-comparator, parallel-group Phase 2 study of faricimab and ranibizumab in center-involving diabetic macular edema. Enrollment was 229. The study included treatment-naive and previously treated participants, with different randomization schemes for those populations, and followed participants for 36 weeks.

The primary outcome was mean change from baseline in BCVA letter score at Week 24 among treatment-naive participants. The reported analysis used a linear mixed-effects model for repeated measurements, did not impute missing values, and assumed data were missing at random. The reported difference was 1.4 letters for 1.5 mg faricimab versus ranibizumab and 3.6 letters for 6 mg faricimab versus ranibizumab, each with an 80% confidence interval.

The record also includes secondary outcome analyses and descriptive safety counts. Details of multiplicity control, the original sample-size justification, and several aspects of model and missing-data sensitivity analysis are not reported in the registry record.