Real-World 12 and 18-Month Outcomes of Faricimab in Patients Previously Treated for Neovascular Age-Related Macular Degeneration

1. Introduction

The Age-related macular degeneration (nAMD) is a significant cause of visual impairment globally.¹ Intravitreal anti-vascular endothelial growth factor (anti-VEGF) is effective in reducing the risk of visual loss.² Faricimab (Vabysmo, Roche/Genetech, Basel, Switzerland) is a bispecific antibody targeting both angiopoietin-2 (ANG-2) and VEGF-A.3 ANG-2 is a key growth factor in regulating angiogenesis and vascular permeability.4 Simultaneous inhibition of ANG-2 and VEGF-A is postulated to enhance treatment durability. TENAYA and LUCERNE demonstrated comparable vision gains between faricimab and aflibercept in nAMD, with most patients on faricimab achieving extended dosing intervals.5 This retrospective study evaluates the efficacy and safety of faricimab in a cohort of nAMD patients with suboptimal response or high treatment burden on other anti-VEGF therapies.

2. Materials and Methods

Study design and patient selection 

This observational study was conducted at Torbay Hospital, Devon, UK, in accordance with the Declaration of Helsinki. Given its retrospective nature, no ethical approval was required.

Consecutive patients aged ≥50 years switched to faricimab for nAMD between September and December 2022 were included. Patients were switched if they had a poor morphological response to their anti-VEGF therapy, and/or a high treatment burden. Poor morphological response was defined as disease activity including the persistence of subretinal fluid (SRF) and/or intraretinal fluid (IRF), and/or new macular haemorrhage despite optimal treatment. Minimal, unchanged, or worsening disease activity immediately following a loading phase of anti-VEGF therapy is also considered poor response. High treatment burden is defined as treatment intervals ≤8 weeks to maintain dry macular status or stable disease activity.

Exclusion criteria were as follows: (1) treatment-naïve status; (2) second eye of bilateral treatment unless both met the inclusion criteria; (3) receipt of other treatment affecting vision such as cataract surgery during follow-up; (4) less than 12 months’ follow-up. All patients who received at least one dose of faricimab were included in the safety analysis to minimise bias. Electronic healthcare data were collected retrospectively.

Treatment protocol and outcome measures 

Patients were treated with the following clinician-directed regimen: (1) full loading phase consisting of 4 monthly injections; (2) partial loading phase consisting of two to three monthly injections; or (3) a treat-and-extend (T&E) regimen starting with an interval of 4-6 weeks. At each visit, the treatment interval was adjusted based on disease activity such as change in macular fluid (SRF and/or IRF) and any accompanying visual acuity change. Treatment intervals were reduced by up to 4 weeks to a minimum 4-week interval if disease activity increased, or extended by up to 4 weeks to a maximum of 16 weeks if stable.

Baseline data collected included sex, age and eye laterality, number of the last intravitreal anti-VEGF injections, anti-VEGF drug before switching, pre-switch treatment interval, visual acuity, and macular anatomical status including central subfield thickness (CST). CST was defined as the mean thickness measured from internal limiting membrane to retinal pigment epithelium within the 1-mm diameter circle centred around fovea on optical coherence tomography (OCT) (Spectralis, Heidelberg Engineering Inc, Franklin, Massachusetts, USA). Endpoint data including visual acuity, OCT anatomical status and treatment interval were collected within prespecified time windows; at 12 months (2 weeks before to 14 weeks after) and 18 months (2 weeks before to 14 weeks after). Time windows were specified to account for asynchronous follow-up in a real-world cohort. The number of faricimab injections received at each endpoint was recorded within the respective 12-month and 18-month periods.

A complete case analysis (CCA) was performed, including only eyes with complete data for each outcome variable at each prespecified follow-up visit.

Statistics 

Descriptive statistics summarised patient characteristics. Continuous data were reported as mean ± standard deviation (SD) or median (interquartile range [IQR]) depending on data normality. Data were entered into Excel (Microsoft, Redmond, Washington, USA) and analysed using R (R Foundation for Statistical Computing version 4.2.2, Vienna, Austria). Data normality was assessed with skewness, kurtosis, histograms and Q-Q plots. Wilcoxon signed-rank and McNemar tests were used to analyse non-parametric continuous and dichotomous data, respectively. A p value of <0.05 was deemed statistically significant.

3. Results

Baseline characteristics and treatment history 

Data from 71 eyes (62 patients) and 60 eyes (54 patients) were analysed at 12 and 18 months, respectively. Mean age at baseline was 82.29 ± 7.04 years (range 65-98 years). 80.1% (50/62) of patients were female and all were Caucasian.

Prior to switching, 97.2% (69/71) of eyes received aflibercept 2 mg (Eylea, Regeneron Pharmaceuticals, NY, USA) and 2.8% (2/71) Ranibizumab 0.5 mg (Lucentis, Genentech, CA, USA). Patients had received a median of 20 injections (IQR: 9-33.5) per eye of their last anti-VEGF therapy over a mean of 2.99 ± 2.36 years (range 0.07-8.67). The mean treatment interval of patients’ prior anti-VEGF treatment was 5.18 ± 1.57 weeks. Immediately before switching to faricimab, 97.2% (69/71) of eyes were on treatment intervals ≤8 weeks.

At baseline, 94.4% (67/71) of eyes showed suboptimal response to prior anti-VEGF treatment. After faricimab switch, 7.0% (5/71) of eyes received a full loading course of four monthly injections, 8.5% (6/71) a partial loading course of three monthly injections, 1.4% (1/71) two monthly injections, and 83.1% (59/71) were switched directly to a T&E regimen with an initial interval of 4-8 weeks.

A mean of 8.97 ± 1.97 faricimab injections were administered over a mean period of 18.3 ± 2.1 months (range: 11.6-21.2).

Patients’ demographics and baseline characteristics are summarised in Table 1.

Visual acuity outcomes 

At baseline, median best corrected visual acuity (BCVA) was 68 ETDRS letters (IQR 57-74). Median BCVA was 66 letters (IQR 55-75, p = 0.119) and 67 letters (IQR 56-73, p = 0.009) at 12 and 18 months, respectively.

At 12 months, the median BCVA change was a loss of two ETDRS letters (IQR: 7 letter loss to 3.5 letter gain). 5.6% (4/71) of eyes gained 5-9 ETDRS letters, 9.9% (7/71) gained 10-14 letters, 5.6% (4/71) gained ≥15 letters. Among those experiencing a loss, 25.4% (18/71) of eyes lost 5-9 letters, 9.9% (7/71) lost 10-14 letters, 5.6% (4/71) lost ≥15 letters.

At 18 months, the median BCVA change was a loss of three ETDRS letters (IQR: 8 letter loss to 2 letter gain). 8.3% (5/60) of eyes gained 5-9 letters, 5.0% (3/60) gained 10-14 letters, 3.3% (2/60) gained ≥15 letters. Among those with a loss, 16.7% (10/60) of eyes lost 5-9 letters, 11.7% (7/60) lost 10-14 letters, 10.0% (6/60) lost ≥15 letters. Table 2 shows BCVA change over time.

BCVA best-corrected visual acuity.

For the 83.1% of eyes (59/71) switched directly to a T&E regimen, median baseline BCVA was 69 letters (IQR 60-74), changing to 66 letters (IQR 59-75, p = 0.0316) and 67 letters (IQR 56-73, p = 0.0065) at 12 and 18 months, respectively.

Anatomical outcomes 

There was a statistically significant CST change from a baseline median value of 296 µm (IQR 270-339) to 273 µm (IQR 245-313, p = 0.0001) and 263 µm (IQR 240-300, p < 0.000) at 12 and 18 months, respectively. Table 3 shows CST results.

Table 4. CST change over time. 

	Number of eyes	CST, median (IQR), µm	p value
Baseline	69	296
12 months	69	273	0.0001
18 months	60	263	<0.0000

CST central subfield thickness, IQR interquartile range.

Table 4. CST change over time. 

	Number of eyes	CST, median (IQR), µm	p value
Baseline	69	296
12 months	69	273	0.0001
18 months	60	263	<0.0000

CST central subfield thickness, IQR interquartile range.

At baseline, 5.6% (4/71) of eyes had a dry macula. After the first faricimab injection, 33.9% (21/62) of eyes had a dry macula (p = 0.0002). Since some eyes received partial or full loading regimens and were reviewed only afterwards, macular fluid status after the first injection could not be assessed in all cases. Compared with baseline, the proportion of eyes with a dry macula was higher at 26.8% (19/71) (p = 0.0023) and 31.7% (19/60) (p = 0.0008) at 12 and 18 months, respectively.

Of the four eyes with a dry macula at baseline, one developed a macular haemorrhage at the 18-month visit whilst on a 16-week treatment interval. Another eye had trace IRF on OCT at 18 months whilst receiving 8-weekly faricimab injections. The remaining two eyes maintained a dry macula at 18 months.

Table 5. CST change in eyes switched directly onto T&E regimen. 

	Number of eyes	CST, median (IQR), µm	p value
Baseline	57	289
12 months	57	271	0.0025
18 months	49	261	<0.0000

CST central subfield thickness, IQR interquartile range.

Table 5. CST change in eyes switched directly onto T&E regimen. 

	Number of eyes	CST, median (IQR), µm	p value
Baseline	57	289
12 months	57	271	0.0025
18 months	49	261	<0.0000

CST central subfield thickness, IQR interquartile range.

In patients switched directly to a T&E regimen, median CST was 289 µm (IQR 270-326) at baseline, 271 µm (IQR 245-309, p = 0.0025) at 12 months, and 261 µm (IQR 240-299, p < 0.0000) at 18 months.

Treatment interval 

At the switching point, the mean treatment interval was 5.2 ± 1.6 weeks. This increased to 7.4 ± 2.8 weeks (p < 0.0000), and 7.5 ± 3.1 weeks (p < 0.0001) at 12 and 18 months, respectively.

9.9% (7/71) and 13.3 (8/60) of eyes achieved an extended treatment interval of ≥12 weeks (≥Q12W) at 12 and 18 months, respectively. A treatment interval extension of at least 4 weeks was achieved in 28.2% (20/71) and 36.7% (22/60) of eyes at 12 and 18 months, respectively. 46.5% (33/71) and 46.7% (28/60) of eyes achieved a treatment interval ≥Q8W at 12 and 18 months, respectively. Figure 1 shows the distribution of dosing intervals at 12 and 18 months.

Q4-6W (4-6 weekly interval), Q8-10W (8-10 weekly interval), ≥Q12W (12 weekly interval).

Treatment intervals were reduced to 4 weeks in 2.8% (2/71) and 3.3% (2/60) of eyes due to suboptimal treatment response at 12 and 18 months, respectively. At 12 months, 5.6% (4/71) of eyes were switched from faricimab to aflibercept 8 mg due to inadequate response based on clinician judgement.

In eyes switched directly to a T&E faricimab regimen, mean treatment interval was 5.2 ± 1.4 weeks at baseline, 7.2 ± 2.8 weeks (p < 0.000) at 12 months and 7.3 ± 2.9 weeks (p < 0.000) at 18 months. In this subgroup, 8.5% (5/59) and 11.8% (6/51) of eyes achieved ≥Q12W treatment interval at 12 and 18 months, respectively. A treatment interval extension of at least 4 weeks was achieved in 30.5% (18/59) and 33.3% (17/51) of eyes at 12 and 18 months, respectively. 42.4% (25/59) and 43.1% (22/51) of these eyes achieved ≥Q8W treatment interval at 12 and 18 months, respectively. Figure 2 illustrates the distribution of dosing intervals at 12 and 18 months.

Q4-6W (4-6 weekly interval), Q8-10W (8-10 weekly interval), ≥Q12W (12 weekly interval).

Safety outcomes 

In this study, one eye developed a flare of hypertensive acute anterior uveitis (AAU) two days after faricimab injection. It was subsequently treated with oral antiviral therapy, topical steroids, and topical and oral carbonic anhydrase inhibitors. Another eye developed hypotony after faricimab injection without permanent sequelae. No cases of post-injection endophthalmitis were recorded.

4. Discussion

We present our real-world experience of one of the early nAMD cohorts switched to faricimab after a suboptimal response to prior anti-VEGF treatment. This retrospective study has one of the longest follow-up periods to date. Our results demonstrated stable visual acuity, significant improvement in anatomical outcomes, and extension of treatment intervals in a cohort in which the majority required dosing intervals ≤8 weeks on prior anti-VEGF treatment. Approximately one third of eyes (36.7%) had treatment intervals extended by at least 4 weeks at 18 months.

Real-world visual acuity outcomes are often inferior to those reported in randomised controlled trials such as TENAYA and LUCERNE.5 In real-world faricimab switching studies, visual acuity has predominantly been reported as stable rather than demonstrating gains in previously treated cohorts.6-17 Studies including TRUCKEE, FARWIDE-nAMD and FARETINA-AMD reported smaller improvements or stable visual acuity in previously treated subgroups compared with treatment-naïve cohorts.6,18,19 Other switching studies reported no significant BCVA change at 12 months.14-16 Another retrospective study reported a statistically significant decline in median BCVA at 12 months in a cohort of 215 eyes characterised by higher treatment burden and resistance.17

At 12 months, there was no statistically significant change in median BCVA in our cohort (p = 0.140). There was a statistically, but not clinically, meaningful change in median BCVA at 18 months. 39.4% (28/71) and 45% (27/60) of eyes had a visual acuity change of <5 letters from baseline at 12 and 18 months, respectively. Although more eyes experienced a loss than a gain in acuity at both follow-up time points, most lost <15 letters from baseline. Overall, 94.4% (67/71) and 90.0% (54/60) of eyes maintained stable visual acuity, avoiding a loss of ≥15 ETDRS letters (≥3 LogMAR lines) at 12 and 18 months, respectively. A loss of ≥3 LogMAR lines corresponds to a doubling of the visual angle which is considered moderate visual loss, an outcome traditionally reported in clinical trials.20,21

In this study, the vast majority (94.4%) of patients had a suboptimal response to prior anti-VEGF treatment. Five of the six patients who lost >15 letters at 18 months were classified as poor responders at the point of switching. Previously treated patients may introduce selection bias towards more treatment-resistant disease and are more likely to have structural damage than treatment-naïve patients, indicating worse prognosis.23,24 None of the six patients who experienced at least moderate visual loss at 18 months received a faricimab loading regimen, as per the UK label of four monthly injections, at switching. The necessity of a fixed loading phase of faricimab, as used in TENAYA and LUCERNE, remains debated. Some suggest a fixed loading phase may be required to maximise Ang-2 inhibition for better long-term visual prognosis.11,25

Given a significant proportion of our cohort (83.1%) was switched directly to a T&E regimen, a subgroup analysis was performed. In this subgroup, there was a statistically, but not clinically, significant change in visual acuity at 12 and 18 months. The interquartile range of visual acuity change highlights heterogeneity in anti-VEGF response in previously treated eyes and underscores the need for a personalised therapeutic approach. Genetic variation has been proposed as a potential mechanism for this phenomenon.22 In this cohort, prior anti-VEGF treatment duration varied widely, from less than one year to over 8 years, representing a source of heterogeneity. Interestingly, a meta-analysis examining real-world 10-year outcomes of anti-VEGF therapy for nAMD found a slow, progressive decline in visual acuity following the first year of treatment, with progression towards macular atrophy proposed as a potential explanation.26 No further subgroup analyses were performed due to limited sample size.

Similar to other real-world studies,6-13 this cohort saw a reduction in median CST from 296 µm (IQR 270-339) at baseline to 273 µm (IQR 245-313, p = 0.000) and 263 µm (IQR 239.75-299.75, p < 0.000) at 12 and 18 months, respectively. Similar findings were observed in the subgroup switched directly to a T&E regimen. The proportion of eyes with a dry macula at 12 months in our study was 26.8% (19/71), lower than the 33.7% and 40.2% reported by Sim et al. and Janmohamed et al., respectively.15,17 Patients in these studies received an initial loading phase of four monthly faricimab injections. Goodchild et al. demonstrated that the maximum macular fluid drying effect of faricimab was achieved following the third loading injection in a cohort of 98 switched eyes.10 As the majority of eyes in our cohort did not receive a loading phase of faricimab, this may partly explain the comparatively lower proportion of dry maculae at 12 months. At 18 months, 31.7% (19/60) of eyes had a dry macula, which was more comparable to that of the aforementioned studies.15,17

Frequent anti-VEGF treatment burdens patients and places substantial strain on public healthcare resources. Our analysis demonstrated encouraging treatment interval extension. Approximately one third of eyes achieved an extension of at least four weeks at 18 months. Specifically, 9.9% (7/71) and 13.3% (8/60) of eyes achieved a ≥Q12W treatment interval at 12 and 18 months, respectively. Similar outcomes were observed in the subgroup switched directly to a T&E regimen. Nevertheless, consistent with findings of Ambati et al. at one year, approximately 50% of eyes in this cohort did not reach a ≥Q8W treatment interval at either 12 or 18 months.16 This contrasts with the results of TENAYA and LUCERNE, and may reflect selection bias towards more chronic cases, in which higher VEGF levels are associated with ongoing inflammation, and under-treatment due to real-world constraints.5,26 Pharmacological factors such as tachyphylaxis have also been proposed as a plausible mechanism contributing to reduced anti-VEGF efficacy over time.27,28

For safety, there were no cases of endophthalmitis, retinal vasculitis or systemic side effects associated with faricimab after a total of 787 intravitreal injections administered in 136 eyes that received at least one dose during the study period. One case of hypertensive acute anterior uveitis (AAU) and one case of post-injection hypotony were reported, both managed non-invasively with no long-term sequelae. Notably, the patient who developed hypertensive AAU had a previous history of herpetic uveitis. Nevertheless, faricimab-related intraocular inflammation could not be excluded in this patient. The hypotony case was likely related to scleral thinning secondary to repeated intravitreal injections in the same quadrant.

The main strength of this study is its longer follow-up compared with existing real-world studies. Limitations include its retrospective design, and the absence of predefined criteria for switching to faricimab, which may have introduced selection bias. The non-standardised treatment regimen at our centre also represents an important source of heterogeneity. Furthermore, we did not account for potential confounders such as the presence of polypoidal choroidal vasculopathy, which may reduce the efficacy of anti-VEGF monotherapy.29

5. Conclusions

In conclusion, switching nAMD patients to faricimab was safe and led to overall stable visual acuity, anatomical improvement and extended treatment intervals. These benefits were maintained after 18 months. Nevertheless, individual responses varied, and personalised treatment approaches are warranted. Future research in larger cohorts with prospective study design and longer follow-up is required to validate the durability of faricimab in nAMD treatment.