Genetic markers of the effectiveness of anti-VEGF therapy in age-related macular degeneration

Introduction

Age-related macular degeneration (AMD) remains one of the leading causes of central vision loss in older adults. According to expert estimates, the global prevalence of AMD will continue to rise as the population ages: the projected prevalence of AMD is expected to reach 288 million by 2040 [1]. Clinically, the greatest threat is posed by the neovascular ("wet") form of AMD (nvAMD), characterized by the development of choroidal neovascularization (CNV), serous and hemorrhagic deposits, and rapid decline in visual acuity. The pathogenesis of nvAMD is driven primarily by hyperactivation of the VEGF-A/VEGFR2 pathway, which enhances angiogenesis and vascular permeability. Inflammation, dysregulation of the complement system, and the Angiopoietin-2/Tie2 signaling pathway also contribute to maintaining the pathological phenotype [2].

Intravitreal agents that inhibit VEGF (ranibizumab, aflibercept, brolucizumab) and/or Ang-2 (faricimab—a bispecific inhibitor of VEGF-A/Ang-2) have radically altered the natural history of nvAMD: in randomized clinical trials (RCTs) and real-world clinical practice, they stabilize and improve functional and anatomical outcomes, reducing the risk of irreversible central vision loss. At the same time, differences in molecular targets (ligand "trap" vs. anti-VEGF-A antibody; dual VEGF-A/Ang-2 blockade) determine nuances in pharmacodynamics and potential predictors of response [3, 4]. Nevertheless, despite their efficacy, response variability remains high. According to large multicenter studies (MARINA, ANCHOR, CATT, VIEW), approximately 20–30% of patients do not achieve clinically significant visual improvement, and some patients exhibit primary or secondary resistance to treatment [5, 6, 7]. Some patients demonstrate incomplete or transient response ("tachyphylaxis"), requiring treatment intensification, molecular switching, or transition to dual blockade (faricimab). Contemporary reviews emphasize the multifactorial nature of outcome heterogeneity: baseline visual acuity, timing of treatment initiation, CNV size/type, treatment regimen, comorbidities, and certain optical coherence tomography (OCT) biomarkers explain part of the variance. Molecular genetic features may also play a significant role in determining individual sensitivity to VEGF-dependent angiogenesis inhibition [8, 9, 10, 18].

Over the past 10–15 years, a substantial body of research has accumulated on genetic predictors of anti-VEGF therapy efficacy in nvAMD. Systematic reviews and meta-analyses indicate associations of several single nucleotide polymorphisms (SNPs) with functional and anatomical outcomes (change in central retinal thickness, presence of subretinal/intraretinal fluid [SRF/IRF]): specifically, in the CFH gene (complement system), IL8, ARMS2/HTRA1 (locus 10q26), VEGFA, and KDR/VEGFR2 (VEGF pathway) [2, 11, 12]. However, the direction and magnitude of effects often vary across cohorts, reflecting heterogeneity in study designs, cohort composition, and treatment protocols. Against this background, the need for independent prospective validation and standardization of response phenotypes remains relevant. Concurrently, data are emerging on rare variants with clinically significant effects (e.g., in C10orf88 and UNC93B1) associated with markedly poorer visual response to anti-VEGF in some patients—these findings have been obtained in genome-wide association studies (GWAS) [13, 14], expanding the spectrum of potential predictors but requiring confirmation and evaluation of applicability in routine practice.

Objective

The aim of this review is to systematically synthesize and critically evaluate the current literature on genetic markers associated with the efficacy of anti‑VEGF therapy in nvAMD.

Mechanisms of Action of Anti‑VEGF Drugs and Molecular Targets

Angiogenesis regulated by vascular endothelial growth factor (VEGF) is a key pathogenic mechanism in nvAMD. The VEGF family includes several ligands (VEGF-A, -B, -C, -D, and PlGF) that bind to tyrosine kinase receptors VEGFR-1 (FLT1), VEGFR-2 (KDR), and VEGFR-3 (FLT4). The cascade from VEGF-A to VEGFR-2 (KDR) is considered the most significant for the pathogenesis of choroidal neovascularization, activating intracellular pathways PI3K-AKT, MAPK/ERK, and PLCγ–PKC, which promote endothelial proliferation, vascular permeability, and the formation of new capillaries in the choriocapillary complex [15, 16].

Excessive VEGF-A expression is induced by hypoxia and inflammation in the retinal pigment epithelium (RPE) and choriocapillaris, leading to the growth of pathological vessels beneath the macula. In addition to VEGF-dependent angiogenesis, inflammatory cytokines (IL-8, TNF-α), complement system activation (particularly via CFH), and extracellular matrix disturbances involving HTRA1 and ARMS2 participate in the pathogenesis [13, 17, 18, 19]. The VEGF and CFH genes are included in the list of genes involved in proliferative processes [20]. Anti-VEGF therapy is directed at neutralizing VEGF-A or blocking its interaction with receptors. Current agents achieve this effect through various mechanisms:

• Ranibizumab — a humanized Fab fragment of an antibody to VEGF-A, with high affinity and the ability to penetrate the retina;

• Aflibercept — a recombinant "trap" protein (VEGF-trap) consisting of extracellular domains of VEGFR-1 and VEGFR-2, binding VEGF-A, VEGF-B, and PlGF;

• Brolucizumab — a single-domain antibody (scFv) to VEGF-A with low molecular weight, providing high tissue concentration and prolonged effect;

• Faricimab — a bispecific antibody simultaneously binding VEGF-A and Angiopoietin-2 (Ang-2), stabilizing the Tie2 signaling complex and thereby additionally reducing vascular permeability and inflammation [21, 22].

The Ang-Tie2 pathway complements VEGF signaling and plays an important role in regulating vascular wall stability. Under normal conditions, Ang-1 activates the Tie2 receptor (TEK), maintaining vascular homeostasis; in AMD, Ang-2 expression increases, competitively inhibiting Tie2 and enhancing inflammation and plasma leakage [23]. Faricimab-mediated Ang-2 blockade promotes Tie2 signaling reactivation and potentiates the effect of VEGF inhibition.

These biochemical differences suggest that genetic variants in the VEGF-A/VEGFR2 and Ang-Tie2 pathways may influence individual response. Numerous studies indicate associations of allelic variants of VEGF-A (rs699947, rs3025039) and KDR (rs2071559, rs2305948) with injection requirements and changes in visual acuity following anti-VEGF therapy [11]. Similarly, genetic variability in CFH (complement factor H gene), ARMS2/HTRA1 (locus 10q26), as well as in inflammatory (IL8) and epigenetic (SIRT1) pathways, may determine resistance or sensitivity to anti-VEGF therapy [14, 24].

Clinical Variability in Response to Anti‑VEGF Therapy

The introduction of VEGF inhibitors has been a key advancement in the treatment of nvAMD. Large randomized studies—MARINA, ANCHOR, VIEW, HAWK, and HARRIER—have convincingly demonstrated that regular intravitreal administration of anti-VEGF agents leads to stabilization or improvement of visual acuity in the majority of patients [6, 7, 21, 25]. Nevertheless, despite significant progress, marked interindividual variability in therapeutic response is observed.

"Response" to anti-VEGF therapy is traditionally understood as a combination of functional and anatomical changes, assessed by dynamics of visual acuity and morphological (OCT) parameters of the retina. Functional response is most often defined as the change in the number of letters on the ETDRS (Early Treatment Diabetic Retinopathy Study) chart, where a gain of ≥10–15 letters is considered clinically significant [5]. In real-world clinical practice, the change in visual acuity in patients with neovascular AMD, for example, when using Aflibercept according to the label, averages +7–8 letters by the end of the first year of therapy. In the VIEW study, this figure was +8.4 letters (with 7.5 injections over one year) [25]. Anatomical response is assessed using OCT by changes in central retinal thickness (CRT), volume of subretinal and intraretinal fluid, reduction in hyperreflective foci, and dynamics of choroidal neovascularization [27]. Based on the combination of these parameters, phenotypes of complete, partial, and absent response are distinguished. Although these parameters partially explain differences in efficacy, they do not provide a complete picture. Even with similar clinical characteristics, patients may demonstrate opposite dynamics—marked response or resistance to therapy.

Current treatment protocols for nvAMD, including:

• Treat and extend (T&E) — treat and extend the interval;

• Pro Re Nata (PRN) — as needed;

• Fixed dosing — fixed dosing regimen;

• Observe-and-Plan (O&P) — observe and plan,

are aimed at optimizing injection frequency and minimizing patient burden; however, the selection of the optimal regimen remains empirical. The absence of biomarkers capable of predicting the efficacy of a specific agent limits the possibilities of personalized therapy. Observations indicate that when switching from aflibercept or ranibizumab to faricimab, some patients with chronic edema demonstrate improvement in morphological parameters, which may reflect individual differences in sensitivity to components of the VEGF-A and Ang-2 signaling pathways [21, 22].

Main Classes of Genetic Markers of Anti‑VEGF Therapy Efficacy

Over the past 15 years, numerous studies have been published on genetic factors influencing the efficacy of anti-VEGF therapy in nvAMD. These include both candidate genes involved in disease pathogenesis (CFH, ARMS2/HTRA1, VEGFA, KDR) and newly associated loci identified through GWAS. Despite heterogeneity of results, certain signals are replicated across several cohorts, allowing these variants to be considered as promising genetic predictors of treatment response.

Complement System Genes: CFH and Related Loci

The complement system plays a key role in maintaining innate immunity and the inflammatory response in the retina. Dysregulation of its regulation is considered one of the central mechanisms of AMD pathogenesis, particularly in dysfunction of the alternative complement pathway. The protein encoded by the Complement Factor H gene (CFH) performs an inhibitory function, preventing excessive activation of C3 convertase and damage to retinal pigment epithelium (RPE) cells [28, 29]. Thus, nucleotide A in the CFH gene, corresponding to the T (non-mutant) allele, exerts a protective role in age-related macular degeneration.

The mutant C allele (nucleotides C), which encodes a tyrosine-to-histidine substitution at position 402 (allelic variant rs1061170 (Y402H)) in exon 7 of CFH, was first identified as one of the strongest genetic risk factors for AMD [30, 31]. Functionally, the Y402H variant alters CFH binding to heparin, C-reactive protein, and the RPE cell surface, reducing inhibition of the alternative complement pathway [32]. This leads to chronic subretinal inflammatory activation, enhanced VEGF-A expression, and reduced efficacy of its blockade by anti-VEGF agents. Experimental models have shown that CFH deficiency increases vascular permeability and enhances response to inflammatory stimuli [33].

Subsequent studies demonstrated that this variant may also modify the efficacy of anti-VEGF therapy. In a retrospective analysis by Brantley et al. (2007), patients with the CC genotype (His/His) had a statistically significantly smaller gain in visual acuity following bevacizumab therapy compared to TT (Tyr/Tyr) carriers (mean difference 3.3 vs. 6.9 ETDRS letters; p = 0.02) [34]. Similar results were obtained by Kloeckener-Gruissem et al. (2011): carriage of the C allele was associated with poorer functional response to ranibizumab in a Swiss cohort (n = 156) [35].

A meta-analysis conducted in 2018, including 76 studies, confirmed the association between this allelic variant and the risk of developing various forms of AMD. It was also noted that in the European ethnic group, this association is more pronounced than in the Asian group: some studies did not demonstrate a link between the allelic variant and disease risk. This is partly explained by the lower frequency of this allele in Asian populations. An association was identified in this ethnic group between the rs1061170 polymorphism and the risk of developing progressive AMD (OR: 2.09; 95% CI 1.67–2.60) and wet AMD (OR: 2.24; 95% CI 1.81–2.77), while in Europeans—early AMD (OR: 1.44; 95% CI 1.27–1.63), dry AMD (OR: 2.90; 95% CI 1.89–4.47), and wet AMD (OR: 2.46; 95% CI 2.15–2.83) [36].

A meta-analysis by Wang et al. (2022), including 15 studies, confirmed a statistically significant, albeit modest, association between rs1061170 and reduced response to anti-VEGF therapy (pooled OR = 1.34, 95% CI 1.10–1.63, p = 0.004). The effect was more pronounced in European populations, whereas in Asian cohorts the associations were borderline, suggesting possible ethnic differences in allele distribution and interactions with other genes in the complement pathway [11].

In addition to rs1061170, intronic and promoter variants of CFH, such as rs1410996 and rs1329428, have attracted research attention. These SNPs are tightly linked to rs1061170 but may have independent functional significance, affecting CFH expression in the retina. For rs1410996, it has been shown that the presence of two GG alleles increased the risk of wet AMD more than twofold compared to two AA alleles [37]. In a study involving Caucasian subjects, rs1410996 was found to be as significant a predictor of AMD development as the more studied rs1061170 [37]. The link between this allelic variant and AMD risk has not yet been explained. It is hypothesized that the mutation does not alter the function of the final protein, as in the case of rs1061170, but negatively affects gene expression levels, leading to reduced protein synthesis [38]. Data on the influence of rs1410996 on anti-VEGF therapy efficacy are conflicting. A meta-analysis found that rs1410996 is associated with poorer therapy response in Asian populations [11]. A later study involving Lithuanian residents did not show an association between the allelic variant and therapy response [37]. A negative correlation of rs1410996 with the efficacy of ranibizumab anti-VEGF therapy was also demonstrated in another European population—among Spaniards [40].

For the intronic variant rs1329428, the association with nvAMD risk remains controversial [41], and studies are sparse. In a Japanese patient cohort, C allele carriers were shown to require additional aflibercept injections more frequently [42]. However, in another Korean patient cohort, no association of this allelic variant with the efficacy of ranibizumab therapy was found [43]. According to Russian researchers Kozhevnikova OS et al. (2022), the aggressive nvAMD phenotype correlates with the minor allele of rs2285714 and is visualized on OCT as persistent subretinal fluid and giant pigment epithelial detachments (PED) [10]. Currently, these SNPs should be regarded as research candidates with respect to treatment response.

Locus 10q26: ARMS2 and HTRA1

Locus 10q26 remains one of the most replicated regions of genetic predisposition to nvAMD; it contains the ARMS2 and HTRA1 genes, which are in strong linkage disequilibrium. Biologically plausible mechanisms include the involvement of HTRA1 (a serine protease) in extracellular matrix remodeling and inflammatory signaling, as well as functional effects of the ARMS2 A69S variant (rs10490924), partially interpreted through regulation of expression in the 10q26 region. Against this background, it is natural that these variants are frequently tested as candidate pharmacogenetic modifiers of response to anti-VEGF therapy.

Several clinical studies have shown associations of ARMS2/HTRA1 genotypes with response parameters, but the nature and strength of associations depend on the chosen outcome. In a Japanese prospective multicenter study of nvAMD patients, the ARMS2 rs10490924 allelic variant was significantly associated with the need for additional injections after the initial three ranibizumab loading doses: the association persisted in the pooled analysis (p = 0.0013), whereas no association was found with visual acuity dynamics or achievement of a "dry" macula. This underscores that the marker may predict treatment burden but not necessarily functional outcome [44]. In another study, in a cohort of patients with polypoidal choroidal vasculopathy, the ARMS2 rs10490924 variant was also associated with anti-VEGF response at 12-month follow-up, extending the applicability of the signal beyond classic nvAMD [45]. Several studies have shown a positive association of the HTRA1 promoter variant rs11200638 with functional/anatomical outcomes, but results are heterogeneous across ethnicities and study designs. A classic example of an early signal is the work by Abedi et al., where risk-allele homozygotes demonstrated worse outcomes following anti-VEGF therapy [46].

Alongside positive signals, a number of studies have failed to confirm an association between ARMS2/HTRA1 and anti-VEGF response. For example, in the study by Cruz-Gonzalez F. et al. in a Spanish cohort (with variable ranibizumab regimen), no associations of response with ARMS2 rs10490923/rs10490924 or HTRA1 rs11200638 were identified [39]. A meta-analysis on HTRA1 rs11200638 (2017) did not find a statistically significant association of this variant with anti-VEGF response in patients with exudative AMD [47], highlighting the role of publication bias and differences in phenotyping. In a later systematic review/meta-analysis by Wang Z. et al. (2022), it was noted that some SNPs in HTRA1/ARMS2 fall into the set associated with response, but the overall level of evidence remains moderate against a background of high inter-study heterogeneity (different drugs, treatment regimens, ethnic composition, response criteria) [11].

Comparison of results indicates that the choice of endpoint for evaluating treatment efficacy remains a fundamental issue. For ARMS2 rs10490924, the most reproducible signal concerns injection frequency/need for additional treatment after the loading phase (treatment burden), whereas for functional outcome (ΔETDRS) and morphology (CRT, "dry" macula), associations are less stable and often fail to replicate in independent cohorts. For HTRA1 rs11200638, the aggregated data are more contradictory: there are both positive signals and significant negative results from meta-analyses, especially for European samples. The strong linkage disequilibrium between ARMS2 and HTRA1 variants complicates attribution of effect to each variant individually. The magnitude of effect and even the direction of associations may differ between Asian and European cohorts; some "HTRA1 signals" may reflect tagging of ARMS2 (and vice versa). This requires careful study design (joint models, conditional analysis) and validation in ethnically diverse samples. Review and methodological articles emphasize precisely this issue as a key source of heterogeneity [24].

Allelic Variants of VEGFA and KDR (VEGFR2) and Their Association with Anti‑VEGF Therapy Efficacy

Vascular endothelial growth factor A (VEGF-A) is a key mediator of pathological angiogenesis in nvAMD. Increased VEGF-A expression in retinal pigment epithelium (RPE) cells and choriocapillaris is induced by hypoxia and oxidative stress, activating the VEGFR-2 (KDR) receptor on endothelial cells and initiating PI3K/AKT, MAPK, and PLCγ/PKC signaling pathways responsible for proliferation, migration, and increased vascular permeability [15, 16]. Since anti-VEGF agents are directed at neutralizing VEGF-A or blocking its interaction with KDR, allelic variants of these genes represent logical targets for pharmacogenetic studies.

Numerous studies have examined the promoter and 3'-untranslated regions of VEGFA, which regulate expression levels of the factor. The most frequently studied are rs699947 (−2578 C>A), rs833061 (−1498 C>T), rs1570360 (−1154 G>A), and rs3025039 (+936 C>T).

In a prospective study by Abedi et al. (2013) in a cohort of 223 patients receiving ranibizumab, carriage of the A allele at rs699947 was associated with greater improvement in visual acuity at 12 months, whereas CC genotype carriers demonstrated a smaller functional response (p = 0.01) [46]. Similar results were obtained by Cruz-Gonzalez et al. (2014) in a Spanish cohort: VEGF-A rs699947 and KDR rs2071559 variants were associated with changes in VA and the frequency of additional injections after 12 months of treatment [40, 48].

In Asian populations, associations were predominantly observed for rs3025039 (+936 C>T). Park et al. (2014) in a Korean cohort (n = 172) showed that T-allele carriers had a less pronounced reduction in CRT and a more frequent need for repeat injections following ranibizumab (p = 0.02), although differences in visual acuity did not reach significance [49].

Several independent studies have confirmed that the influence of VEGF-A variants is generally modest and more often reflected in early morphological outcomes (retinal thickness, presence of fluid) rather than in long-term visual dynamics [50]. However, a large analysis of CATT data (JAMA Ophthalmology, 2014) did not confirm statistically significant associations for the studied VEGF-A or VEGFR2 variants, highlighting possible effects of differences in ethnic composition and small effect sizes [51].

Allelic Variants of KDR (VEGFR2) and Receptor Sensitivity

KDR encodes the primary VEGF-A receptor, whose phosphorylated form initiates angiogenic signaling pathways. The most frequently studied SNPs are rs2071559 (−604 T>C) in the promoter region and rs2305948 (Q472H) in exon 7. The promoter variant may affect binding of transcription factors SP1/ELK1, while rs2305948 may affect the structure of the tyrosine kinase domain.

In the study by Cruz-Gonzalez et al. (2014), the C allele at rs2071559 was associated with a weaker functional response (mean VA improvement of 4.5 letters vs. 7.8 for TT genotype, p < 0.05). In Spanish and Portuguese cohorts, this variant also correlated with an increased need for injections, suggesting possible regulatory influence on VEGFR-2 expression [48].

In the review by Wu et al. (2017), it was noted that KDR variants rs2071559 and rs2305948 demonstrated associations with response in some cohorts, but in large analyses (including CATT), these associations were not replicated, indicating heterogeneity of results and modest predictive value [52].

Synthesis analyses (Balikova I, 2019; Wang Z, 2022) [11, 24] emphasize that associations of VEGFA/KDR with anti‑VEGF therapy efficacy are of a "moderate" effect nature, often driven by additional factors—baseline VA, CNV type, treatment regimen, and ethnic composition of the sample. Since individual SNPs provide small contributions, the most promising approach is the use of polygenic models (Polygenic Response Score) and multifactorial adjustment for clinical covariates. Consistent results are observed mainly for rs699947 and rs2071559, whereas data for rs3025039 and rs2305948 are contradictory.

The overall assessment of the level of evidence is moderate: associations are replicated in several cohorts, but replication in large RCTs and mechanistic in vivo confirmation are lacking. Nevertheless, these variants are of interest for research panels and may complement classical markers (CFH, ARMS2/HTRA1, IL8) in multigene models for predicting response.

Inflammatory and Oxidative Pathways in nvAMD

Inflammation is a key component of the pathogenesis of neovascular age-related macular degeneration (nvAMD). Activation of innate immunity, cytokines, and chemokines contributes to damage of retinal pigment epithelium (RPE) cells, microglial activation, and maintenance of the neovascular process [53, 54]. Among inflammatory mediators, interleukin-8 (IL-8) has attracted particular attention due to its potent proangiogenic properties, acting as a chemoattractant for neutrophils and an activator of endothelial cells.

IL-8 expression is upregulated in RPE under oxidative stress and exposure to AGE products, as well as following photochemical damage [55]. In the pathological retina in nvAMD, IL-8 participates in recruiting inflammatory cells and activating endothelium, enhancing VEGF-A production and potentiating angiogenesis [56].

Given these mechanisms, the IL8 gene (locus 4q13–q21) is considered a candidate marker for sensitivity to anti‑VEGF therapy.

IL8 rs4073 (−251A/T) Allelic Variants and Anti‑VEGF Efficacy

The rs4073 (−251A/T) variant, located in the promoter region of IL8, affects transcription and secretion levels of IL-8: A-allele carriers are characterized by increased expression and higher inflammatory potential [57].

A number of early studies established an association between rs4073 and the risk of AMD and age of disease onset. In the study by Hautamäki et al. (2015), the A allele was associated with earlier onset of AMD (p = 0.008), but not with disease type (exudative vs. atrophic) [58].

However, data on the association of rs4073 with anti‑VEGF therapy efficacy remained contradictory. In the study by Thomsen et al. (2024) (Acta Ophthalmologica), a prospective evaluation of 12‑month anti‑VEGF response was conducted in 346 nvAMD patients. The authors did not find a statistically significant association of rs4073 with changes in visual acuity, retinal thickness, or number of injections. No differences were detected between homozygotes for the A and T alleles either, although a possible association with disease predisposition was demonstrated [59].

Overall conclusion: IL8 rs4073 may influence AMD risk and the severity of the inflammatory component and may be considered a candidate marker for sensitivity to anti‑VEGF therapy.

Epigenetic Regulation of Angiogenesis: The SIRT1 Gene

SIRT1 is an NAD⁺-dependent class III deacetylase (sirtuin) that regulates a wide range of processes: from metabolism and aging to inflammation and angiogenesis. At the retinal level, SIRT1 controls the activity of HIF-1α, NF-κB, and p53, thereby influencing the expression of VEGF-A, VEGFR-2, MMP-14, and proinflammatory cytokines.

In an experimental model, Lin et al. (2018) showed that deletion of SIRT1 in retinal endothelial cells impairs vascular endothelial migration and both physiological and pathological angiogenesis, mediated by altered deacetylation of HIF-1α and subsequent regulation of VEGF‑A/VEGFR‑2 and MMP-14 expression [60]. Other studies have shown that SIRT1 can both stimulate and suppress angiogenesis depending on context, regulating VEGF/VEGFR‑2 and ICAM-1 expression through NF-κB and HIF-1α [61, 62].

At the clinical level, SIRT1 has been studied primarily as an AMD risk gene and systemic factor. In the work of Liutkeviciene et al. (2019), the SIRT1 variant rs12778366 was shown to be associated with an increased risk of AMD (C allele and TC genotype increased risk approximately 2‑to 2.5‑fold), especially in women and patients over 65 years of age [63]. Kaikaryte et al. (2022) extended these data by studying three SNPs (rs3818292, rs3758391, rs7895833) and serum SIRT1 levels: AMD patients showed changes in both genotypes and SIRT1 concentrations, which the authors interpret as reflecting an imbalance between pro- and anti-angiogenic influences of SIRT1 [64].

A review by Velmurugan et al. (2024) emphasizes that SIRT1 integrates epigenetic, metabolic, and inflammatory mechanisms in the retina and may potentially serve as a target for modifying AMD progression and sensitivity to therapy, but there are as yet no direct clinical data linking specific SIRT1 polymorphisms or expression levels with response to anti‑VEGF agents [65]. SIRT1 in this sense is a candidate for future pharmacogenetic and epigenetic studies. Moshetova L.K. et al. assert that SIRT1 represents an attractive candidate for the development of therapeutic strategies to prevent premature aging of ocular tissues, in particular age-related macular degeneration [66].

Comparative Analysis of Data from Meta‑Analyses and Reviews

Early reviews on the pharmacogenetics of anti‑angiogenic therapy for AMD (Agosta, 2012) summarized the initial studies on CFH, ARMS2/HTRA1, VEGF‑A, and several other genes, encompassing both anti‑VEGF and photodynamic therapy, and emphasized "promising but preliminary" results [67]. In the work of Dedania et al. (2015), data specifically on anti‑VEGF response in nvAMD were systematically reviewed, with the authors concluding that most associations remain controversial and the level of evidence is insufficient for clinical application [68]. In the work of Fauser S. et al. (2015), 39 publications on genetic predictors of anti‑VEGF response were analyzed, with the authors also concluding that none of the tested SNPs could at that time be considered a clinically useful marker, despite recurring signals for CFH and ARMS2/HTRA1 [69].

Wu et al. (2017) performed the first targeted meta‑analysis for VEGFA and VEGFR2 and did not identify robust associations of the studied allelic variants with therapy response (including rs699947, rs833061, rs3025039, and several KDR SNPs), noting high heterogeneity of outcomes and small effect sizes [52]. In another meta‑analysis by Wang Z. et al. (2022), including 33 studies, the authors identified 9 SNPs in four genes (CFH, ARMS2, HTRA1, OR52B4) that were associated with anti‑VEGF therapy efficacy in the pooled analysis. However, the authors separately noted that the data require confirmation in large and ethnically diverse cohorts [11]. Strunz et al. (2022) performed their own genome‑wide association study (GWAS) followed by an analytical review: none of the tested SNPs reached the GWAS significance threshold, and the results for individual candidates (CFH, ARMS2/HTRA1, VEGFA/KDR) were not confirmed at a level sufficient for clinical interpretation [70].

A recent 2024 scoping review on molecular biomarkers in nvAMD concluded that despite dozens of studies on the contribution of genetic biomarkers, no response marker has progressed to being validated; the main reasons are small and heterogeneous samples, differences in design, and the absence of standardized outcomes [2]. Similarly, a genetic sub‑analytical GWAS study of the VIEW 1 and 2 RCTs showed that in a standardized RCT cohort, neither CFH, nor ARMS2/HTRA1, nor VEGFA/KDR demonstrated significant association with key clinical outcomes [71].

Limitations of the Existing Evidence Base

All major reviews and meta‑analyses emphasize common methodological problems:

1. Heterogeneity of response phenotypes. In different studies, response was defined by:

• gain of ≥5, ≥10, or ≥15 ETDRS letters;

• change in CRT on OCT;

• presence/absence of subretinal/intraretinal fluid;

• number of injections over 3–12 months.

This substantially complicates direct data pooling.

2. Different drugs and treatment regimens. Meta‑analyses included patients receiving bevacizumab, ranibizumab, aflibercept, with various regimens (fixed, PRN, treat‑and‑extend). Even within the same drug, the regimen strongly influences VA and CRT dynamics, and failure to account for these factors in models leads to confounding of effects.

3. Ethnic heterogeneity and sample size. Many studies were conducted in relatively small and ethnically specific cohorts (Japanese, Korean, European, Latin American samples). Meta‑analyses emphasize that the effects of CFH and ARMS2/HTRA1 are more pronounced in Europeans, whereas in Asian populations results are more contradictory.

4. Publication bias. Reviews by Dedania (2015), Fauser (2015), Bobadilla (2022), and the recent scoping review by Dervenis N (2024) directly point to selective publication of "positive" results with no reporting of neutral/negative associations, inflating effect estimates in early meta‑analyses.

5. Lack of standardized pharmacogenetic study protocols. Most studies are single‑center, with varying inclusion criteria, design, and statistical models.

Thus, comparative analysis of meta‑analyses and reviews allows several fundamental conclusions to be formulated:

No single SNP (including CFH rs1061170, ARMS2 rs10490924, HTRA1 rs11200638, and VEGF‑A/KDR variants) possesses sufficient sensitivity and specificity to serve as an independent clinical test for predicting response to anti‑VEGF therapy.

The most promising remain polygenic models using combined information from multiple genes (CFH, ARMS2/HTRA1, and others), in conjunction with clinical and OCT parameters.

According to Budzinskaya M.V. et al. (2013), negative prognostic signs for anti‑angiogenic therapy include the presence of 402H, (−625)A, and (−251)A in both copies of the CFH, HTRA1, and IL‑8 genes [72].

Existing work underscores the deficit of well‑phenotyped prospective cohorts, especially for individual drugs (brolucizumab, faricimab) and for combining genetics with other classes of biomarkers (e.g., cytokines, etc.) (Table 1).

Table 1. Main SNPs for which data are available on specific rs‑numbers, association with treatment efficacy, and direction of effect

Conclusion

Thus, current evidence confirms that individual variability in the efficacy of anti‑VEGF therapy in neovascular AMD has a complex nature, encompassing both clinical and molecular‑genetic factors. The most reproducible associations have been identified for the CFH, ARMS2/HTRA1, VEGFA, IL8, and KDR genes; however, their predictive value is limited and does not permit the use of individual allelic variants in clinical practice. New GWAS studies indicate the existence of subgroups of patients with genetically determined resistance to therapy, opening perspectives for a personalized approach. Further progress is possible with a transition from fragmented candidate gene studies to large multicenter projects with unified response phenotype criteria and the use of multigene, multi‑omic, and clinical‑genetic models. Integration of such data into clinical practice will improve the accuracy of predicting anti‑VEGF treatment efficacy and bring the implementation of personalized ophthalmogenetics closer. Nevertheless, one should not forget about non‑parametric methods for comparing observed and expected frequencies in statistical analysis of small samples, which are very widely used in biomedical research. Therefore, individual studies also retain great scientific interest in the development of personalized anti‑VEGF therapy approaches.