Introduction

Diabetic retinopathy (DR) is a frequent microvascular complication of diabetes mellitus (DM) that can result in macular thickening due to retinal vascular hyperpermeability, a condition known as diabetic macular oedema (DMO). DMO is the most common cause of moderate visual impairment in patients with DR.¹ The global prevalence of DMO is estimated to be around 6.8%, equating to about 27 million adults worldwide being affected by DMO.² The number of adults with DMO worldwide is anticipated to increase by 51.9% in the next 20 years.^3,4

Intravitreal injection of anti-vascular endothelial growth factor (anti-VEGF) is currently the mainstay treatment of DMO.^5,6 The landmark trials demonstrated improved anatomical and functional outcomes when followed up for at least two years.^7–10

However, a subset of patients does not respond satisfactorily to anti-VEGF injections. A post hoc analysis of Diabetic Retinopathy Clinical Research (DRCR) Retina network Protocol T showed that 44% of patients receiving aflibercept 2mg for DMO had persistent macular oedema after two years of treatment.^11,12 Additionally, real-world data often demonstrate nuanced variations between clinical trial outcomes and routine healthcare practice¹³ and some patients exhibit an inadequate response even with frequent and regular anti-VEGF therapy.^14,15

This resistance phenomenon is believed to stem from a complex interplay of genetics, tachyphylaxis, and alternative angiogenic pathways beyond VEGF. Thus, various pathogenic mechanisms are implicated in DMO progression.¹⁶ Persistent DMO results in the propagation of oedematous changes driven by multiple inflammatory mediators earlier in the cascade, including angiopoietin-2 (Ang-2), which contributes to vascular destabilisation, vascular permeability, and neovascularization.^17–19 The accumulation of reactive oxygen species (ROS) further exacerbates this process.

Faricimab (VabysmoTM, Roche/Genentech, Basel, Switzerland) is the first and currently the only bispecific molecule that targets both Ang-2 and VEGF. Blocking the Ang-2 pathway improves vascular stability, culminating in DMO reduction.²⁰ We aim to assess the real-world outcomes of intravitreal faricimab in DMO with sub-optimal response to previous treatment with aflibercept 2mg.

Materials and Methods

Study Design and Population

This retrospective study was conducted in a single tertiary centre: Bristol Eye Hospital, University Hospitals Bristol and Weston, Bristol, United Kingdom (UK). The study included 62 eyes of 50 patients with sub-optimally responsive DMO that were switched to faricimab. A sub-optimal response was defined after at least six 4-weekly aflibercept 2mg doses, as:

• A central subfield thickness (CST) greater than 325 µm after the loading dose.
• An increase of more than 20% from their best CST was recorded despite treatment intervals of 8-weekly or less.

Exclusion criteria for the study included: 1) less than 6 doses of aflibercept at 4-weekly intervals, 2) less than four doses of 4-weekly faricimab injection, 3) intravitreal dexamethasone implant in the previous six months, 4) fluocinolone acetonide intravitreal implant in the previous three years, 5) macular laser within the previous 3 months to commencing faricimab. If an eye received an intravitreal steroid implant or macular laser, they were excluded from the study at that point. Data collected from our Electronic Medical Records (EMR) included demographics, diabetes type, diabetic retinopathy (DR) grading, HbA1c, and treatment intervals.

Outcome measures were best recorded visual acuity (BRVA) in early treatment diabetic retinopathy study (ETDRS) letters, central subfield thickness (CST) measured using optical coherence tomography (OCT) (Topcon DRI OCT Triton plus; Tokyo, Japan) and treatment intervals in weeks. These were recorded at baseline (day of first faricimab injection) after four loading doses of faricimab and at the latest clinic appointment.

Treatment Protocol

Four doses of faricimab at 4-weekly intervals were given, with a review after the fourth loading dose. From this point, they moved on to a treat and extend (T&E) regime, with intervals being maintained, extended, or reduced guided by anatomical outcomes compared to a reference CST value. The reference CST was the lowest CST recorded after loading (aflibercept 2mg or faricimab). Treatment intervals were reduced by 4 weeks if CST increased by >20% of the reference CST, maintained if CST increased by 10–20% of the reference CST, or extended by 4 weeks if CST increased by <10% of this reference value. This local treatment protocol was adapted and modified from the Personalised Treatment Interval (PTI) arm of the landmark trials in YOSEMITE and RHINE.²¹

Statistical Analysis

The study’s data were analysed using JASP statistical software. Descriptive statistics were presented as mean averages with standard deviation. A paired t-test was utilized for normally distributed data, while the Wilcoxon signed-rank test was applied to non-normally distributed datasets. Statistical significance was established at a p-value less than 0.05. Effect sizes were calculated using Cohen’s d. The effect size interpretation followed established parameters: negligible (d < 0.2), small (0.2 ≤ d < 0.50), moderate (0.5 ≤ d < 0.80), and large (d ≥ 0.8).

Results

The mean age of participants was 63.9 (±11.4) years, with the majority being male (56%) and predominantly of white ethnicity (80%). Table 1 summarises patient’s demographics. The mean HbA1c within 6 months of baseline was 64.5 (±18.2) mmol/mol. At the latest follow-up, the mean HbA1C remained stable at 65.5 (±18.7) mmol/mol, CI[3.9, −1.8](p=0.283).

Table 1 Demographic Data from the Electronic Medical Records (EMR), 50 Patients, 62 Eyes

Baseline (n=62)

The mean number of injections received before the switch to faricimab was 17.3 (±10.7). All patients had received aflibercept 2mg before the switch to faricimab and six patients (9.7%) had initially been treated with ranibizumab followed by aflibercept 2mg before being switched to faricimab. Treatment received prior to faricimab initiation can be found in Table 2. At baseline, the BRVA was 67.6 (±11.8) letters, and CST was 406.4 (±105.9) µm (Table 3).

Table 2 A Summary of Treatment Received for All Eyes Prior to Switching to Faricimab. All Previous Treatments Were Given in Line with the Study’s Inclusion Criteria

Table 3 Summary of the Functional and Anatomical Outcomes for Faricimab T&E at Baseline, Post-Loading, and Latest Follow-up

Visit After Loading

After the fourth injection in the loading phase, the mean review time was 6.5 (±1.9) weeks. Fifty-four eyes continued in the T&E pathway as per the local protocol; 5 eyes were switched to an intravitreal dexamethasone implant due to poor response, and three were switched to a pro re nata (PRN) treatment regime and therefore excluded from further analysis (Table 4).

Table 4 Summary of Treatment Intervals at Baseline, Post-Loading, and Follow-up

Table 5 Summary Table of Current Literature on the Use of Faricimab in DMO

BRVA was 70.4 (±12.7) letters, with a gain of 2.8 letters, 95% CI [0.8, 4.7] letters from baseline (p=0.008, d=0.23). 11 eyes (17.7%) gained ≥10 letters. CST improved to 372.8 (±132.0) µm (95% CI [9.7, 81.0], p=0.002, d=0.28). At this point, 31 eyes (50%) had CST <325 µm, 4 eyes (6.5%) had no IRF at this visit compared to 0% at baseline. The treatment intervals prescribed from this visit were as follows: 13 eyes (21%) 4-weekly, 6 eyes (9.7%) 6-weekly, 26 (41.9%) 8-weekly, 1 eye (1.6%) 10-weekly, and 8 eyes (12.9%) 12-weekly. This equates to 56.5% being on treatment intervals of 8 weeks or more. The mean treatment interval was now 7.4 (±2.6) weeks, a gain from baseline of 0.8 weeks (p=0.030).

Latest Follow-up Outcomes

The mean follow-up was 57.1 (±19.7) weeks, receiving a mean of 7.92 (±2.53) faricimab injections from baseline. Forty-one eyes remained on the T&E pathway with faricimab; 14 switched to intravitreal dexamethasone implants; 2 patients (3 eyes) died, and one discontinued treatment (Table 4).

Outcome metrics were available for 42 eyes; at their latest follow-up, the BRVA was stable at 68.7 (±14.6) letters (p=0.572, d=0.08). Eight (12.9%) gained ≥10 letters. CST was 343.1 (±117.5) µm with a reduction from baseline of 56.5µm (95% CI [4.6, 108.3], p = 0.020, d=0.57). Two eyes (3.2%) had resolution of diabetic macular oedema. Thirty-nine eyes (62.9%) had a reduction in CST compared to baseline.

The treatment intervals recommended at the latest follow-up were based on CST reference values. These are as follows: 2 eyes (3.2%) 4-weekly, 6 eyes (9.7%) 6-weekly, 19 eyes (30.6%) 8-weekly, 2 eyes (3.2%) 10-weekly, 7 eyes (11.3%) 12-weekly, 1 eye (1.6%) 14-weekly, 4 eyes (6.5%) 16-weekly. The mean injection interval was 9.2 (±3.2) weeks, a gain of 2.4 weeks (95% CI[1.3,3.6], p < 0.001, d = 0.96). 53.2% of those on the T&E regime at the latest clinic visit were on a treatment interval of 8 weeks or more.

A total of 14 eyes (22.6%) had poor response to treatment, demonstrated by either no improvement or a significant increase in CST, and these were switched to intravitreal dexamethasone implant. Two eyes (3.2%) received adjunctive treatment with macular laser and were subsequently excluded from the analysis at this point.

Safety

A total of 491 faricimab injections were analysed. Two patients (3 eyes) died from pre-existing health conditions unrelated to their intravitreal treatment. Two eyes, accounting for 0.42% of our total injections, had a flare-up of pre-existing anterior uveitis and responded well to topical steroid treatment only. This is in line with landmark trials, and no other safety concerns were reported.²¹

Discussion

This study evaluated faricimab’s effectiveness in treating DMO in patients who showed sub-optimal response to prior aflibercept 2mg therapy. Switching this cohort to faricimab allowed us to extend treatment intervals, enhance macular anatomy, and maintain stable vision outcomes.

The cohort’s demographic profile and significant retinal disease aligns with DMO epidemiology. These challenging cases had a high treatment burden, averaging 17.3 aflibercept 2mg injections prior to the switch. A small subset remained poorly responsive and required intraocular steroid implants post-switch.

Real-world studies corroborate faricimab’s benefits in similar challenging cohorts, summarised in Table 5.^22–29 Like ours, these studies demonstrate stable vision, anatomical improvements, and no new safety concerns following the switch. Visual acuity in our cohort was maintained throughout, with modest, non-significant improvements after the loading phase. Anatomical outcomes showed a statistically significant CST reduction, with 67.7% achieving CST reduction after loading, increasing to 84.8% at the latest visit.

Compared to YOSEMITE and RHINE trials, where 78–86% achieved <325 µm CST by 52 weeks in the PTI group, 50% of our cohort achieved this after the loading phase, and 45.2% maintained it at the most recent visit. Similarly, our findings align with real-world data from Rush et al, where 37.5% achieved a CST of <300 µm after four months. Given the complexity of our treatment-experienced cohort, these outcomes are highly encouraging.

A key finding in our study was the statistically significant extension of the injection interval to an average of 9.2 weeks at 12 months. This is particularly noteworthy and has significant real-world benefits given the significant psychological and economic burden that frequent injections impose on patients, their caregivers, and healthcare systems. These burdens, well-documented in the literature, can adversely impact adherence and overall quality of life.^30,31 In our study, the average injection interval increased by 2.4 weeks, with the proportion of patients needing injections every 8 weeks dropping from 56.5% to 30.6%. Notably, 53.2% of our cohort achieved intervals of 8 weeks or longer by 12 months, surpassing the 39.2% reported by Rush et al.²³ Comparable results were seen for 12-week intervals, and fewer patients required 4-week injections.

While Durrani et al demonstrated interval extension, their study lacked long-term follow-up and a defined loading protocol.²⁸ Reducing the injection burden significantly benefits patients and caregivers by easing logistical and financial pressures. For example, this approach could free up 2040 appointments and save £230,520 annually per 1000 patients in appointment costs alone, not accounting for additional savings from transportation or lost work hours.

The psychological relief of extended intervals and fewer treatments is invaluable. As these data reflect only the first year with mandatory loading doses, further improvements in the second year could enhance long-term outcomes and healthcare efficiency.

No new safety signals were identified across the 491 injections administered. Two unrelated deaths and two cases of anterior uveitis flare in a patient with a known history of the condition were reported. Importantly, no endophthalmitis, retinal vasculitis, or new intraocular inflammation were reported.

The variability in BRVA and CST outcomes highlights the need for predictive biomarkers to better determine which patients are likely to respond favourably to faricimab. Given the real-world nature of our study, there were certain limitations, including our cohort’s predominantly white demographic, which limits the generalisability of these findings to other ethnic groups. CST recordings were not completed at a fixed, pre-determined time, which, with the diurnal variation on fluid accumulation in DMO, may contribute to variation in the recorded CST.³² Additionally, the retrospective nature of this study imposes inherent limitations.

Our cohort size was relatively small due to our deliberate inclusion of only patients with complete datasets, ensuring data integrity despite reducing overall numbers. Notably, this is the first study to clearly delineate and implement a defined T&E pathway for faricimab administration. Despite these constraints, we maintained methodological rigor by implementing strict control measures for confounding factors through comprehensive inclusion and exclusion criteria. We further enhanced study validity by excluding patients who deviated from protocol requirements. HbA1c levels remained consistently elevated at both baseline and final visits, which may contribute to the challenges in managing diabetic macular oedema in this cohort. Future prospective, longitudinal real-world studies will provide more comprehensive insights into effective management strategies for this challenging patient population.

Conclusion

This study demonstrated that in DMO cases with sub-optimal response to aflibercept 2mg, switching to faricimab can maintain visual acuity, whilst improving anatomical outcomes, and extend injection intervals. These benefits, combined with absence of new safety concerns in our cohort, not only enhance patient outcomes but also reduce the treatment burden. This approach allows for better alignment with patient needs while improving efficiency and lowering healthcare costs.