Efficacy, durability, and safety of faricimab with extended dosing up to every 16 weeks in diabetic macular edema: 2-year results from the Japan subgroup of the phase 3 YOSEMITE trial

Purpose To evaluate the 2-year efficacy, durability, and safety of faricimab in patients with diabetic macular edema (DME) in the YOSEMITE Japan subgroup. Study design YOSEMITE/RHINE (NCT03622580/NCT03622593) subgroup analysis: global, multicenter, randomized, double-masked, active-comparator–controlled, phase 3 faricimab trials. Methods Patients were randomized 1:1:1 to intravitreal faricimab 6.0 mg every 8 weeks (Q8W) and per treat-and-extend (T&E) dosing, or aflibercept 2.0 mg Q8W. Outcomes were assessed through year 2 for the YOSEMITE Japan subgroup (N = 60) and the pooled YOSEMITE/RHINE global cohort (N = 1891). Results In the YOSEMITE Japan subgroup, 21, 19, and 20 patients were randomized to faricimab Q8W, faricimab T&E, and aflibercept Q8W, respectively (632, 632, and 627 patients in the pooled YOSEMITE/RHINE cohort). Vision gains and anatomic improvements with faricimab at year 1 were maintained over 2 years and were generally consistent between groups. Mean best-corrected visual acuity changes from baseline at year 2 (weeks 92–100 average) for the YOSEMITE Japan subgroup were +12.5, +9.0, and +5.0 letters in the faricimab Q8W, faricimab T&E and aflibercept Q8W arms, respectively (+10.8, +10.4, and +10.3 letters in the pooled YOSEMITE/RHINE cohort). At week 96, 61.1% of the YOSEMITE Japan subgroup and 78.1% of the pooled YOSEMITE/RHINE cohort were on ≥ Q12W dosing. Faricimab was well-tolerated with a safety profile comparable with aflibercept. Conclusion Faricimab up to Q16W offered durable vision gains and anatomic improvements up to 2 years in patients with DME in the YOSEMITE Japan subgroup. Outcomes were generally consistent with the pooled YOSEMITE/RHINE cohort. Supplementary Information The online version contains supplementary material available at 10.1007/s10384-024-01078-y.


Introduction
Diabetic macular edema (DME), a consequence of diabetic retinopathy, is a leading cause of visual impairment in Japanese adults [1][2][3].In Japan, as is the case globally, intravitreal anti-vascular endothelial growth factor (VEGF) therapy has become the standard of care for patients with center-involving DME with visual impairment [2,3].Survey data show that the proportion of ophthalmologists choosing anti-VEGF injections as the first-choice therapy for DME in Japan increased from 73% in 2015 to 81% in 2016-2017 [4,5].
Real-world studies suggest that the outcomes achieved with anti-VEGF therapies fall short of those anticipated from clinical trials [6,7].This is likely in part due to undertreatment associated with the burden of frequent monitoring visits and injections.Data from a large claims database study in Japan indicate that the number of anti-VEGF injections in patients with DME within 1 year after the first injection was 3.5 in 2020 [8] compared with more than 8 in clinical trials [9].This highlights an unmet need for novel therapies that can alleviate the treatment burden.There is also considerable heterogeneity in response to anti-VEGF treatment in real-world patient populations [10].
Based on the premise that multitargeted treatment strategies for DME may promote vascular stability and durable efficacy beyond those offered by VEGF inhibition alone, faricimab was developed as the first bispecific antibody for intraocular use [11].Faricimab independently binds and neutralizes both VEGF-A and angiopoietin (Ang)-2, which are involved in regulation of retinal blood vessel growth and integrity, and are upregulated in DME [11,12].Faricimab has been approved in multiple countries for the treatment of patients with DME, based on positive 1-year data from the pivotal YOSEMITE and RHINE trials, which are collectively the largest registrational DME program conducted to date [13].These randomized active-controlled trials of faricimab in patients with DME demonstrate that faricimab 6.0 mg dosed every 8 weeks (Q8W), or according to a personalized treat-and-extend (T&E) regimen, offered noninferior vision gains and anatomic improvements at 1 year compared with aflibercept 2.0 mg Q8W; the primary efficacy endpoint of YOSEMITE and RHINE was change in best-corrected visual acuity (BCVA) from baseline at 1 year, averaged over weeks 48, 52, and 56 [13].The strong durability of faricimab was evident in the faricimab T&E arm, with > 50% of patients receiving every 16 week (Q16W) dosing at week 52 and > 70% of patients receiving every 12 week (Q12W) dosing or longer [13].The safety and tolerability profile of faricimab was comparable with aflibercept [13].Robust vision gains and anatomic improvements achieved with faricimab were maintained up to 2 years with up to Q16W dosing in the faricimab T&E arms [9].
In an analysis of the YOSEMITE Japan subgroup, the efficacy, durability, and safety of faricimab at year 1 were generally consistent with the global results from the pooled YOSEMITE/RHINE cohort [14].Now that YOSEMITE and RHINE have completed their 96-week treatment periods, we present data from the 2-year analysis of the YOSEMITE Japan subgroup along with an assessment of consistency with the pooled YOSEMITE/RHINE cohort through study end.

Study design
The study design and rationale for YOSEMITE (NCT03622580) and RHINE (NCT03622593) have been previously described [13,15].Briefly, YOSEMITE and RHINE were identically designed, global, multicenter, randomized, double-masked, active-comparator-controlled, phase 3 trials of faricimab in patients with DME.Both trials were conducted in accordance with the International Council for Harmonization E6 Guidelines for Good Clinical Practice, the tenets of the Declaration of Helsinki, applicable US Food and Drug Administration regulations, the European Union Clinical Trials Directive (2001/20/EC), and relevant local, state, and federal laws.Study protocols were approved by institutional review boards and ethics committees as applicable, and all patients provided written informed consent.
Patients eligible for inclusion were ≥ 18 years of age and had center-involving DME, defined as BCVA 25-73 Early Treatment Diabetic Retinopathy Study (ETDRS) letters (approximate Snellen equivalent, 20/320-20/40), and central subfield thickness (CST) of ≥ 325 µm for Spectralis spectral domain-optical coherence tomography (SD-OCT) or ≥ 315 µm for Cirrus SD-OCT or Topcon SD-OCT (measured as the average thickness between the internal limiting membrane and Bruch's membrane in the central 1-mm diameter of the ETDRS grid).
The T&E regimen was a personalized treat-andextend-based dosing regimen that allowed adjustable dosing up to Q16W [13,15].Patients randomized to the faricimab T&E arm initially received faricimab at Q4W intervals.Following first achievement of CST < 325 µm at or after week 12, patients switched to adjustable dosing whereby dosing intervals were extended by 4 weeks (up to Q16W), maintained, or reduced by 4 or 8 weeks (down to Q4W) based on prespecified CST and BCVA criteria at active dosing visits.
To maintain masking, all patients attended Q4W study visits where they received active treatment or sham up to week 96, with a final study visit at week 100.

Outcome measures
Outcomes were evaluated for Japanese patients enrolled from Japanese study sites in YOSEMITE (no Japanese patients or study sites were included in RHINE) and the pooled YOSEMITE/RHINE trial population.Two-year trial outcomes reported herein were consistent with prespecified endpoints in the primary analysis [13], and included changes in BCVA from baseline at 2 years (defined as the average of weeks 92, 96, and 100) and over time; the proportion of patients in the faricimab T&E dosing arms on Q4W, Q8W, Q12W, or Q16W dosing intervals at week 96 and over time; change in CST from baseline at 2 years and over time; the proportion of patients with absence of DME (CST < 325 μm based on protocol-defined DME) over time; the proportion of patients with absence of intraretinal fluid (IRF) and subretinal fluid (SRF) over time (measured in the central 1 mm ETDRS circle); the proportion of patients with ≥ 2-step improvement on the ETDRS Diabetic Retinopathy Severity Scale (DRSS) at week 96; and the incidence and severity of ocular and nonocular adverse events (AEs) through study end.

Statistical analysis
Japan subgroup analyses were prespecified in the YOSEM-ITE trial protocol and efficacy and safety analyses performed were consistent with the primary analysis [13].Efficacy analyses were based on the intent-to-treat population, grouped by treatment arm at randomization.Adjusted means for continuous endpoints were assessed using a mixed model for repeated measures, adjusted for treatment group, visit, visit-by-treatment group interaction, baseline BCVA or CST (continuous) as applicable, and randomization factors of baseline BCVA and previous intravitreal anti-VEGF therapy.For binary secondary endpoints, weighted proportions were estimated using the Cochran-Mantel-Haenszel method stratified by baseline BCVA score and prior intravitreal anti-VEGF therapy.
Efficacy outcomes for the YOSEMITE Japan subgroup are reported with 95.04% confidence intervals (CIs) to adjust for interim safety assessments conducted through to the completion of the primary analysis [13].Analyses were conducted by an independent data monitoring committee.Efficacy analyses for the pooled YOSEMITE/RHINE cohort were additionally adjusted or stratified by geographic region and study (YOSEMITE versus RHINE) and 95% CIs are reported.No formal statistical comparisons were made between the YOSEMITE Japan subgroup and pooled YOSEMITE/RHINE cohort; results should be interpreted as exploratory and descriptive.
For all efficacy analyses, intercurrent events due to the COVID-19 pandemic were handled using a hypothetical strategy where all values were censored after the intercurrent event, and intercurrent events not due to COVID-19 were handled using a treatment policy strategy where all observed values were used regardless of the intercurrent event.The robustness of these assumptions and the primary results were demonstrated using sensitivity and supplemental analyses as previously described [13].Statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc.).
Safety and tolerability were assessed through descriptive summaries of ocular and systemic AEs (coded using Medical Dictionary for Regulatory Activities thesaurus terms), deaths, and ocular assessments through study end.
The YOSEMITE Japan subgroup included 60 patients enrolled from 27 study sites in Japan (Fig. S1, Online Resource 1) [14].Of these patients, 21, 19, and 20 were randomized to faricimab Q8W, faricimab T&E up to Q16W, and aflibercept Q8W, respectively [14].The number of patients who discontinued study treatment and the reasons for discontinuation were generally similar across treatment arms.

Baseline characteristics
Baseline characteristics for the YOSEMITE Japan subgroup have been previously reported, and were generally well balanced across treatment arms and consistent with the pooled YOSEMITE/RHINE cohort [14].

Durability outcomes
In the YOSEMITE Japan subgroup and the pooled YOSEM-ITE/RHINE cohort, durable vision gains with faricimab were achieved with T&E dosing up to Q16W (Fig. 2a,b

Anatomic outcomes
In the YOSEMITE Japan subgroup, CST reductions achieved with faricimab at year 1 were maintained over 2 years and the mean CST change numerically favored faricimab Q8W over faricimab T&E and aflibercept Q8W at 2 years (Fig. 3).The mean CST changes from baseline across weeks 92-100 were -255.In the YOSEMITE Japan subgroup and the pooled YOSEMITE/RHINE cohort, the proportion of faricimabtreated patients achieving absence of DME was maintained or improved from year 1 through year 2 (Fig. 4).Numerically more patients in the YOSEMITE Japan subgroup and the pooled YOSEMITE/RHINE cohort achieved absence of DME (defined as CST < 325 μm) in the faricimab arms compared with the aflibercept arm at year 2. The proportion of patients in the YOSEMITE Japan subgroup achieving absence of DME across weeks 92-100 was 100%, 71.0-77.4%, and 63.0-76.0% in the faricimab Q8W, faricimab T&E, and aflibercept Q8W arms, respectively.In the pooled YOSEMITE/RHINE cohort, the proportion of patients achieving absence of DME across weeks 92-100 was 87.5-92.2%,81.4-85.9%,and 78.6-82.6% in the faricimab Q8W, faricimab T&E, and aflibercept Q8W arms, respectively [9].
A similar pattern was observed for absence of IRF in the YOSEMITE Japan subgroup and the pooled YOSEM-ITE/RHINE cohort, except that the proportion of patients was numerically higher in the faricimab arms than the aflibercept Q8W arm in the YOSEMITE Japan subgroup (Fig. S2a, b, Online Resource 1).In the YOSEMITE Japan subgroup, the proportion of patients with absence of IRF was numerically reduced in the faricimab T&E arm at year 2 versus year 1.The proportions of patients achieving absence of IRF across weeks 92-100 were 55. 7 [9].Absence of SRF through week 100 was high and generally consistent across treatment arms and across the YOSEMITE Japan subgroup and the pooled YOSEMITE/ RHINE cohort; 58-76% of patients in the YOSEMITE Japan subgroup and 62-66% of patients in the pooled YOSEM-ITE/RHINE cohort had absence of SRF at baseline, which increased to near 100% for all groups by week 16 and was maintained through week 100 (Fig. S2c, d, Online Resource 1).

DRSS improvement
Rates of at least 2-step DRSS improvement were largely consistent across treatment arms and between the YOSEM-ITE Japan subgroup and the pooled YOSEMITE/RHINE cohort at weeks 52 and 96, although there was a numerical trend for higher proportions in the faricimab arms versus the aflibercept arm in the YOSEMITE Japan subgroup (Fig. 5).The proportion of patients achieving at least 2-step DRSS improvement at week 96 was 50%, 48%, and 36% in the YOSEMITE Japan subgroup in the faricimab Q8W, faricimab T&E, and aflibercept Q8W arms, respectively, and 52%, 44%, and 43% in the pooled YOSEMITE/RHINE cohort [9].

Safety outcomes
In the YOSEMITE Japan subgroup and the pooled YOSEMITE/RHINE cohort, faricimab was well tolerated through study end, with a safety profile comparable with aflibercept (Table 1).In the pooled YOSEMITE/RHINE cohort and the YOSEMITE Japan subgroup, the incidence of ocular AEs through study end was comparable between patients receiving faricimab Q8W (49.7% and 42.9%, respectively), faricimab T&E (49.2% and 42.1%), and aflibercept Q8W (45.4% and 50.0%), with most events being mild or moderate in severity.Serious ocular AEs in the pooled cohort and the YOSEMITE Japan subgroup were low and comparable between the faricimab Q8W (4.1% and 4.8%, respectively), faricimab T&E (5.4% and 5.3%), and aflibercept Q8W (3.2% and 10.0%) treatment arms.The incidence of nonocular AEs was also comparable across treatment arms and similar between the YOSEMITE Japan subgroup and the pooled YOSEMITE/ RHINE cohort.No Antiplatelet Trialists' Collaboration events were reported in the YOSEMITE Japan subgroup.

Discussion
Following the completion of the YOSEMITE and RHINE trials and publication of year 2 data [9], the current analysis was performed to extend the previously reported YOSEM-ITE Japan subgroup year 1 analysis [14] through to study end along with an assessment of consistency with the pooled YOSEMITE/RHINE cohort.Data from the YOSEMITE Japan subgroup demonstrated that vision gains, anatomic improvements, and extended durability with faricimab up to Q16W were maintained through year 2, in line with what was observed in the pooled YOSEMITE/RHINE cohort and the individual YOSEMITE and RHINE trial findings [9].
BCVA gains, anatomic outcomes, and safety of faricimab in the YOSEMITE Japan subgroup were generally consistent with the global YOSEMITE/RHINE results [9].BCVA gains with faricimab at year 1 were maintained through year 2 and remained comparable with those seen with aflibercept.Improved anatomic outcomes with faricimab versus aflibercept in the YOSEMITE Japan subgroup were maintained over 2 years: numerical advantages in favor of faricimab were observed in the change from baseline in CST and in the proportion of patients achieving absence of DME and IRF.
Durable vision gains and anatomic outcomes with faricimab were achieved in the YOSEMITE Japan subgroup, with more than 60% of patients on Q12W or Q16W dosing at week 96.In a real-world study from 2019, a maximum injection interval of 14 weeks and a mean injection interval of 8.5 ± 2.9 weeks was reported among 75 treatment-naïve eyes with DME treated with anti-VEGF (ranibizumab or aflibercept) according to a T&E schedule [16].This is lower than the maximum injection interval of 16 weeks achieved in this study.Furthermore, in another retrospective consecutive case study, the mean recurrence interval (time between previous injection and criteria for an additional injection being met) of previous anti-VEGF injection was 5.8 ± 2.5 weeks, which was significantly extended to 10.8 ± 4.9 weeks (p=0.0005)following switching to faricimab [17].These findings reflect an opportunity for reduced treatment burden with faricimab relative to current options.Faricimab continued to be well tolerated in the YOSEM-ITE Japan subgroup through year 2, with a safety profile comparable with that of aflibercept.Additional data regarding the longer-term safety and efficacy of faricimab in DME will come from the RHONE-X extension trial (NCT04432831) for patients completing YOSEMITE or RHINE and the VOYAGER study (NCT05476926), a realworld study of patients with DME or neovascular age-related macular degeneration who are treated with faricimab or the Port Delivery System with ranibizumab in routine clinical practice.
Although general trends were similar between the YOSEMITE Japan subgroup and the pooled YOSEMITE/ RHINE cohort [9], some differences were noted, including increased variability between treatment arms in the YOSEMITE Japan subgroup versus the pooled cohort.This was apparent in the adjusted mean BCVA changes from baseline between weeks 92 and 100 and in the proportion of patients with ≥ 2-step DRSS improvement at study end.Additionally, the proportion of patients in the faricimab T&E arm who achieved Q12W or Q16W dosing was lower for the YOSEMITE Japan subgroup compared with the pooled YOSMITE/RHINE cohort.These differences may be a reflection of the smaller number of patients in the YOSEMITE Japan subgroup.There was a trend toward greater improvement in DRSS with faricimab in the pooled YOSEMITE/RHINE cohort [9].Although a similar trend was observed in the YOSEMITE Japan subgroup, the small number of patients preclude any definite conclusions from being drawn.However, these findings are nevertheless consistent with faricimab improving diabetic retinopathy by inhibiting Ang-2 and stabilizing blood vessels [18].This is further highlighted in a recently published descriptive review from Japan, which shows a reduction in microaneurysm count in a patient treated with faricimab [19].Another difference is that the proportions of patients achieving absence of IRF decreased slightly in the faricimab T&E arm during the second year in the YOSEMITE Japan subgroup, but not in the pooled YOSEMITE/RHINE cohort [9].Of note, there was a high degree of variability and a small overall sample size in the YOSEMITE Japan subgroup, which may have contributed to these differences.
This study was limited by the small sample size of the YOSEMITE Japan subgroup.Findings from the prospective, open-label, multicenter SWAN study (jRCTs031230213) will provide further information on the efficacy and safety of faricimab in a larger cohort of Japanese patients with DME.Additionally, no formal statistical comparisons were made between treatment arms or with the pooled cohort; thus, these findings should be interpreted as exploratory and descriptive.
In conclusion, year 2 data from the YOSEMITE Japan subgroup are generally consistent with the pooled YOSEMITE/RHINE cohort [9], and show that dual Ang-2/ VEGF-A inhibition with faricimab offered comparable vision stability and numerically greater fluid control versus aflibercept, while also demonstrating strong durability, with the potential for extended dosing up to every 16 weeks.These results support dual Ang-2/VEGF-A inhibition with faricimab as a treatment for DME that provides durable efficacy, extended treatment intervals, and, consequently, may reduce the treatment burden for patients.

Fig. 3
Fig. 3 Adjusted mean CST change from baseline in the YOSEMITE Japan subgroup (a) and the pooled YOSEMITE/RHINE cohort (b).Pooled global cohort includes patients in Japan who were enrolled during the global phase of YOSEMITE/RHINE.Data originally published in Wong et al [9].a Adjusted mean CST change from baseline

Fig. 5
Fig. 5 Proportion of patients with ≥ 2-Step DRSS improvement in the YOSEMITE Japan subgroup (a) and the pooled YOSEMITE/RHINE cohort (b).Pooled global cohort includes patients in Japan who were enrolled during the global phase of YOSEMITE/RHINE.Data originally published in Wong et al [9].Analyses included patients with evaluable color fundus photograph images at baseline and week 52 and/or week 96.Weighted proportions were estimated using the CMH method, stratified by baseline BCVA (< 64 letters versus ≥ 64 letters), prior intravitreal anti-VEGF therapy (yes versus no) in the YOSEMITE Japan subgroup and the pooled YOSEMITE/RHINE cohort, and by region (United States and Canada versus rest of the world) and study (YOSEMITE versus RHINE) in the pooled YOSEMITE/RHINE cohort.Weighted proportions for the aflibercept Q8W arm presented for the faricimab Q8W versus aflibercept Q8W comparison.97.52% CI error bars are shown at week 52; 95.04% CI error bars are shown at week 96.BCVA best-corrected visual acuity, CI confidence interval, CMH Cochran-Mantel-Haenszel, DRSS Diabetic Retinopathy Severity Scale, ETDRS Early Treatment Diabetic Retinopathy Study, Q8W every 8 weeks; T&E treat-and-extend, VEGF vascular endothelial growth factor RHINE cohort (Fig.2c, d).In the YOSEMITE Japan subgroup and the pooled YOSEMITE/RHINE cohort, 46.2% and 79.3% of patients who achieved ≥ Q12W dosing at week 52 maintained ≥ Q12W dosing through week 96.A similarly high proportion of patients in the YOSEM-ITE Japan subgroup (71.4%) and the pooled YOSEMITE/ [9].2Proportion of patients in the faricimab T&E arm who achieved Q4W, Q8W, Q12W, or Q16W dosing at week 96, and dosing intervals in the faricimab T&E arm through week 96 in the YOSEMITE Japan subgroup (a, c) and the pooled YOSEMITE/RHINE cohort (b, d).Pooled global cohort includes patients in Japan who were enrolled during the global phase of YOSEMITE/RHINE.Data originally pub-lished in Wong et al[9].Proportion of patients in the faricimab T&E arm on Q4W, Q8W, Q12W, or Q16W dosing at week 96, among those who had not discontinued the study at the week 96 visit.Treatment interval at week 96 was defined as the treatment interval decision made at that visit.Q4W every 4 weeks, Q8W every 8 weeks, Q12W every 12 weeks, Q16W every 16 weeks, T&E treat-and-extend YOSEMITE/ VEGF vascular endothelial growth factor of patients with absence of IRF was generally numerically higher in the faricimab Q8W and T&E arms than the aflibercept Q8W arms.The proportions of patients achieving absence of IRF across weeks 92-100 were 57.2-62.7%,44.1-48.4%, and 36.2-41.3% in the faricimab Q8W, faricimab T&E, and aflibercept Q8W arms, respectively a Absence of DME was defined as CST < 325 μm, measured as the distance from the internal limiting membrane to Bruch's membrane.Weighted proportions were estimated using the CMH method, stratified by baseline BCVA (< 64 letters versus ≥ 64 letters), prior intravitreal anti-VEGF therapy (yes versus no) in the YOSEMITE Japan subgroup and the pooled YOSEMITE/RHINE cohort, and by region (United States and Can-cept Q8W comparison.95% CI error bars are shown; estimates < 0% or > 100% are imputed as 0% or 100%, respectively.BCVA bestcorrected visual acuity, CI confidence interval, CMH Cochran-Mantel-Haenszel, CST central subfield thickness, DME diabetic macular edema, Q8W every 8 weeks, T&E treat-and-extend,

Table 1
Summary of key adverse events through week 100 for the YOSEMITE Japan subgroup and the pooled YOSEMITE/RHINE cohort Results are presented for safety-evaluable population.Percentages are based on n values in the column headings; multiple occurrences of the same AE in an individual are counted only once AE adverse event, APTC Antiplatelet Trialists' Collaboration, BCVA best-corrected visual acuity, Q8W every 8 weeks, T&E treat-and-extend a Ocular AEs in the study eye only are presented; b Ocular AEs of special interest were defined as events associated with severe intraocular inflammation, events requiring surgical or medical intervention to prevent permanent loss of sight or events associated with BCVA loss of ≥ 30 letters for > 1 hour; c Excluding endophthalmitis; d APTC events were adjudicated by an external independent committee; all other events were investigator reported.Includes AEs with onset from the first dose of study drug through study end