As stated above, AMD is known to be a multifactorial disease, the development and progression of which is governed by the complex interaction of various environmental and genetic elements; aging is the primary factor, and drives the overexpression of VEGF-A in the macular microenvironment among elderly patients. Advancements in technologies, such as single-cell sequencing and genome-wide association studies (GWASs), have revealed mutations and factors that contribute to the progression of AMD. Through GWASs, specific genes, including CFH on chromosome 1 and ARMS2 and HTRA1, both residing on chromosome 10, have emerged as significant loci closely linked to advanced AMD [26,95]. The CHF variant is primarily connected to the presence of drusen, whereas the ARMS2-HTRA1 variant is correlated with the occurrence of subretinal or sub-RPE hemorrhages [96]. Although these genes are involved in the development of nAMD and may be useful predictors of treatment response, they have yet to be shown to have a significant role in its treatment. Other genes including MMP9, CETP, and TIMP3 have been linked to nAMD due to their roles in regulating the extracellular matrix remodeling [97], and the FGD6, HTRA1, and CFH genes play pivotal roles in governing oxidative stress and inflammation, which in turn regulate the advancement of angiogenesis, thereby contributing to the progression of nAMD [98].

However, the RPE hypoxia previously described promotes an over-expression of the hypoxia-inducible factor alpha (HIF-α) and VEGF-A by RPE cells, with the consequent degeneration of the RPE cells themselves and of Bruch’s membrane [99]. Anti-VEGF treatments have really shown that VEGFA/HIF-α-related genes (VEGF, VEGFR, PDGF, PEGF) can be useful treatments; this makes VEGF, VEGFR, PDGF, and PEGF the primary targets for the current gene therapy [100].

Gene therapy for nAMD faces challenges due to the complexity of the genes associated with the condition. Unlike monogenic disorders with a small gene that can fit into an AAV for the standard gene augmentation therapy, nAMD involves multiple genetic factors; the already mentioned genes contribute to disease susceptibility, making it challenging to devise a one-size-fits-all gene therapy. The diverse genetic landscape of nAMD adds a layer of complexity, requiring a nuanced approach in developing gene therapies tailored to the specific genetic factors.

Exploring gene silencing through small interfering RNA (siRNA) or microRNA (miRNA) targeting VEGF is considered as a potential approach for AMD treatment [101]. Numerous clinical trials are currently underway, focusing on the utilization of precise gene silencing methods [102,103,104]. After being introduced into cells, siRNA binds and activates the RNA-induced silencing complex, which in turn targets and degrades any cells complementary to the siRNA sequence, thereby preventing protein synthesis.

Bevasiranib, a modified naked RNA, results in the downregulation of VEGF-A by means of its intracellular transcriptional inhibitor action and possibly its TLR3-mediated activity, and may be the treatment of nAMD. A phase III human trial, which involved the intravitreal administration of siRNA bevasiranib (NCT00499590), was halted, as it was deemed unlikely to achieve its primary objective [105]. As bevasiranib may only inhibit new VEGF synthesis, without impacting existing VEGF levels, a phase III trial (NCT00499590) was also performed to assess the efficacy of the combined bevasiranib and ranibizumab therapy for nAMD treatment, but this too was unlikely to meet its primary endpoint and was terminated.

AGN211745 (formerly Sirna-027) is a chemically modified naked siRNA that has VEGFR-1 as the target gene, inducing gene silencing by binding the complementary target RNA with the lytic cytoplasmic protein complexes known as RNA-induced silencing complexes, thereby reducing the level of VEGFR-mRNA and significantly inhibiting MNV development, with the potential to treat nAMD. However, despite positive findings in the phase I/II study, a phase II trial administering Sirna-027 (NCT00395057) did not meet crucial efficacy endpoints. (NCT00363714) [106].

Despite many efforts in multiple trials exploring gene silencing, studies have never advanced beyond phase III, as gene silencing methods encounter several obstacles, including RNA instability, limited bioavailability, and the potential for non-specific targeting. These challenges, common to most drug delivery systems, significantly hamper the successful application of siRNA therapeutics in the treatment of nAMD. Additionally, while siRNA-based therapies have demonstrated theoretical advances for patients with nAMD, this approach has not shown any superiority compared to conventional anti-VEGF treatments. This is primarily because even with siRNA therapies, the requirement for repeated injections persists, as their effect is temporary (3–7 days) due to their degradation by tissue nucleases. Nonetheless, the possibility of extending these effects exists through chemical alterations or the use of viral vectors, which could help maintain the efficacy of therapies based on RNA interference.

An alternative to siRNAs involves the use of microRNAs (miRNAs) which are small (18–22 nucleotide), single-stranded, noncoding RNAs that down-regulate the gene expression post-transcriptionally [107]. Various research studies have shown that the dysregulation of miRNAs is relevant both in experimental AMD models and in AMD subjects, and may therefore potentially be associated with an increased risk of developing AMD [108,109,110]. MicroRNA mimics or anti-miRNA have the potential to be biomarkers, diagnostic tools, or targets for the control and treatment of this disease, by modulating retinal cellular function [111]. Unfortunately, the miRNAs evaluated in animal models of AMD behave differently compared to AMD patients; thus, their role in the disease remains unclear [112].

To achieve successful results in gene therapy, it is essential to use a vector that ensures prolonged gene expression levels while minimizing the risks of toxicity and immune reactions. Different types of vectors have been used.

Viral vectors are modified viruses commonly used in gene therapy approaches to deliver therapeutic genes or RNA-based molecules to target disease cells. They have been used as delivery vehicles to precisely transport therapeutic genetic material into the target cells within the eye and achieve a sustained therapeutic effect. In gene therapy for nAMD, vector selection is of paramount importance. For retinal gene supplementation, the optimal selection is the recombinant adeno-associated viral vector (AAV) [113,114]. Its small, single-stranded DNA genome of approximately 4.6 kilobases (kb) with organized capsid structure makes it conducive to genetic modifications [115]. AAVs are currently the most commonly used vector for retinal gene transfer in both preclinical studies and clinical trials [116]. They provide advantages like extended transgene expression, minimal risk of insertional mutagenesis, only slight inflammatory responses induced, and a low chance of germline transmission [117,118]. The most extensive AAV serotypes studied in ocular gene therapies are AAV2, AAV5 and AAV8 [119,120,121]. Gene therapy products utilizing AAV vector systems, including Glybera (alipogene tiparvovec to treat hereditary lipoprotein lipase deficiency) [122], Luxturna (voretigene neparvovec-rzyl), Zolgensma (onasemnogene abeparvovec to treat spinal muscular atrophy type 1) [123] and Hemgenix (Etranacogene dezaparvovec for the treatment of hemophilia B) [124], have received notable approvals. Among these, Luxturna, the first approved gene therapy for a genetic disease, is a recombinant AAV 2 vector containing human RPE65 complementary DNA that enables RPE cells to produce the retinoid isomerohydrolase RPE65. After its efficacy and safety were ultimately confirmed in an open-label, randomized and controlled phase 3 trial conducted at two centers in the United States, Luxturna was authorized for gene augmentation therapy in RPE65-associated retinal dystrophy [125] and stands out as a retinal gene therapy designed to treat Leber congenital amaurosis (LCA) [126]. However, a subset of patients undergoing subretinal Luxturna injection developed progressive perifoveal chorioretinal atrophy following surgery. Despite that, most patients did well on visual function measures. Although the mechanism for chorioretinal atrophy is not known at this time, there are several potential factors that must be considered, alone or in combination, namely: direct toxicity of the AAV2 vector to the photoreceptors and RPE, inflammation or immune response to the vector, surgical delivery and ocular factors [127]. Further studies are necessary to determine what potential factors predispose patients to this complication and to clarify what the implications are for gene therapy in nAMD, especially in terms of a immune response.

Moreover, retroviruses and lentiviruses have been employed in various gene therapy products, such as RetinoStat^® (Oxford BioMedica, Oxford, UK, OXB-201) targeted for nAMD (NCT01301443) and stem cell therapy. Notably, subretinal administration of RetinoStat, a lentiviral vector expressing endostatin and angiostatin, demonstrated safety and good tolerance. Patients with severe nAMD exhibited signs of clinical improvement, including visual acuity stabilization and reduction in vascular leakage [128]. Nonetheless, retroviruses and lentiviruses carry risks such as the potential for insertional mutagenesis and germline transmission. Additionally, they might trigger more pronounced inflammatory responses compared to AAVs. An important aspect of gene therapy is the possible immune reaction towards the AAV capsid: in humans, administering AAV vectors, unlike in many animal models, triggers antigen-specific T-cell activation, posing an increased risk during the initial postoperative phase. A brief period of immunosuppression around the surgery can help regulate immune responses until the capsid antigens are eliminated from the infected cells [129]. The route of vector delivery significantly influences immunogenicity. Subretinal delivery is a favorable option for disorders primarily affecting the RPE and/or photoreceptors. Given that the majority of inherited retinal disorders (IRD) involve either or both of these cell types, the subretinal delivery emerges as the prevailing administration route in gene therapy trials targeting monogenic conditions. This method involves the creation of a retinotomy near the temporal vascular arcades, allowing the bleb to slowly spread toward the foveal region, creating a shallow elevation [68]. Despite this type of delivery method involving a temporary detachment of the retina, the existing trial data indicates that it is generally safe and has the potential to offer effective therapeutic outcomes [69,70,71].

Among the non-viral delivery techniques, the most straightforward approach is physical delivery, which involves injecting naked plasmid DNA, siRNA, mRNA or miRNA. However, this method has a limited efficacy due to the rapid degradation and minimal uptake [130]. Non-viral gene delivery through chemical techniques is attractive due to its lower potential to trigger immune responses, straightforward scalability and cost savings in production [131].

DNA nanoparticles consisting of a single molecule, compacted using polyethylene glycol (PEG)-substituted lysine peptides (CK30-PEG), have been utilized for transporting payloads of up to 20 kb in size [132,133]. These nanoparticles have demonstrated safety in diverse mouse models of retinal degeneration [134,135].

Lipid-based transfection systems have proven to be effective in delivering target genes to retinal cells in various studies. Numerous lipid-based drugs designed for eye diseases are accessible for transporting CRISPR or ribonucleoproteins for base editing [136,137]. Niosomes, consisting of cholesterol and uncharged single-chain surfactant [138,139], exhibit potential as non-viral carriers for gene delivery [140,141,142]. In ocular gene therapy, polymer-based platforms like chitosan, hyaluronic acid, polyethyleneimine (PEI), poly(amidoamine) (PAMAM), PEG, poly (lactic-glycolic acid) (PLGA) and poly(L-lysine) (PLL) have been under investigation [101,130].

Gene editing technology involves the manipulation of the target gene at the DNA or genomic level. The most common gene editing system to date uses clustered, regularly interspaced, short palindromic repeat (CRISPR) endonucleases such as Cas9, which can cut the DNA at a precise, targeted location, to either ablate or repair a destructive mutation [143]. The CRISPR/Cas9 system comprises a guide RNA targeting the gene of interest and an endonuclease that creates a site-specific double-stranded DNA “cut”, enabling precise genetic modification [144]. This allows for the lasting and accurate modification or removal of a mutation associated with a specific disease [145]. However, when addressing mutations in a single gene, CRISPR may not be effective for patients without a recognized genetic diagnosis. The CRISPR-Cas9 system has several potential advantages over other editing tools such as its simplicity of target design, ease of generating large-scale libraries and relatively low cost [146,147]. Moreover, the genome editing with CRISPR-Cas9 enables multiple editing through the engagement of multiple guide RNAs (gRNAs) [148,149]. Treatment for AMD patients can involve the use of the adeno-associated viral vector (AAV)-CRISPR tool, utilizing CjCas9 (Campylobacter jejuni) [150,151], and type-V CRISPR-Cas systems with LbCpf1 nucleases. AAV-delivered CjCas9 can accurately target and modify specific sites in the human or mouse genome, inducing mutations in RPE cells. In this context, CjCas9 can target the VEGFA or Hif1a gene in RPE cells, potentially reducing the size of laser-induced neovascularization. This approach may evolve into an in vivo genome editing therapy for nAMD [152]. Progress in CRISPR/Cas9 technology, including base and prime editing, holds the promise of improving the efficiency and cost-effectiveness of using CRISPR/Cas9 to treat retinal diseases such as nAMD.

However, a key challenge in the application of CRISPR/Cas9 technology remains the manufacturing and production for in vivo editing [153], and all CRISPR applications in retinal diseases including nAMD have been largely experimental; clinical trials of CRISPR for nAMD are lacking, as the field is still exploring safety and efficacy concerns.

The genomic impacts of transduction using AAV vectors encoding CRISPR-Cas nucleases are still under investigation; high levels of AAV integration (up to 47%) into Cas9-induced double-strand breaks (DSBs) are in therapeutically relevant genes in cultured murine neurons, mouse brain, muscle, and cochlea, and this should be recognized as a common outcome for applications that utilize AAV for genome editing [113]. Moreover, efficient gene delivery and editing can be achieved through the ocular delivery of mRNA packaged in lipid nanoparticles (LNPs). Subretinal injections of LNPa containing Cre mRNA in the mouse show a tdTomato signal in the RPE, enabling genome editing in the retina; in the future, this can be used to correct genetic mutations that lead to blindness [114].

Clinical trials investigating gene therapy for nAMD currently adopt two strategies: the intraocular administration of modified viral vectors expressing antiangiogenic proteins, and RNA interference molecules to contrast the VEGF overexpression.

To this purpose, PEDF, endostatin, angiostatin, secreted extracellular domain of VEGFR1 and sFLT-1 have been targeted by gene therapy [154].

A phase I clinical trial (ClinicalTrials.gov: NCT00109499) explored the safety of AdGVPEDF.11D in patients affected by advanced nAMD. The investigators delivered the PEDF gene via an adenoviral vector with deficient replication (by deletion of E1, E3, and E4). PEDF is an important endogenous antiangiogenic factor, and its levels are low in the presence of nAMD. Adenovirus, a double-strand DNA virus, can carry up to 37 kb for transgene delivery [155,156,157,158]. The participants received an intravitreal injection of AdGVPEDF.11D with dosages ranging from 1E6 and 1E9 particle units (PU). In 25% of cases, there were reports of mild and temporary intraocular inflammation, with no severe adverse events. Although the study was not designed to assess the therapeutic efficacy, neovascularization was observed to be stable or reduced in patients receiving 1E8 or 1E9 PU, compared to those receiving lower doses.

Intravitreal and subretinal injection of FLT-1 (also known as VEGFR-1) or FLT-1 derivates have been tested on nAMD patients after encouraging results on animal models [159]. FLT-1 expression is normally upregulated by hypoxia, neutralizing VEGF-A, and thereby preventing its dimerization with membrane receptor VEGFR-2 and the consequent pro-angiogenic pathway. The intravitreal injection of AAV2-sFLT01, encoding for a fusion protein composed by sFLT-1 domain 2 and the Fc domain of IgG₁, was tested in a phase I trial (ClinicalTrials.gov: NCT01024998, Sanofi Genzyme, Paris, France), whereas the subretinal administration of recombinant AAV (rAAV).sFLT-1, encoding the natural soluble FLT-1, was experimented on in a phase I/IIa trial (ClinicalTrials.gov: NCT01494805, Avalanche Biotechnologies).

In the first trial, the viral vector was demonstrated to be safe, not detectable systemically and not eliciting immunogenic activity. Moreover, the encoded protein was detectable within 52 weeks in 5 of the 10 patients treated with the highest dosage (2E10 vector genomes). In general, the expression was dose-related, but variable among the subjects, with 80% of non-expressers showing, at baseline, anti-AAV2 antibody titers of 1:400 or greater, indicating a considerable impact of individual characteristics in determining the response to treatment. Although the treatment was well tolerated at all dosages, it did not produce any significant anatomical (retinal thickness) and functional (BCVA) improvement [160].

The phase I/IIa trial NCT01494805 confirmed the safety and effectiveness of the subretinal injection of the rAAV.sFlt-1 vector, resulting in an increase in retinal sFLT-1 levels. Forty patients suffering from nAMD were assigned to low-dose, high-dose or control arms. A regular intravitreal injection of ranibizumab was administered when patients showed a BCVA reduction or intraretinal/subretinal fluid increase on OCT or augmented leakage on fluorescein angiography during the 36-month follow-up. The number of intravitreal treatments and changes in BCVA and retinal thickness were recorded during the 36-month follow-up. This gene therapy demonstrated safety and good tolerance; however, no notable changes were observed in the examined endpoints [161,162,163]. The induced endogenous expression of anti-VEGF has also been explored in humans, after encouraging results on animal models.

Positive results were also obtained with the delivery of a gene encoding a soluble monoclonal portion of an anti-VEGF antibody structurally similar to ranibizumab. The safety and tolerability of this gene treatment, called RGX-314 and administered via subretinal injection, was tested in a phase I/IIa trial (ClinicalTrials.gov: NCT03066258, REGENXBIO). The 42 enrolled nAMD patients had previously been treated with anti-VEGF intravitreal injections. They were divided into 5 cohorts receiving the adeno-associated viral vector (NAV AAV8) at different doses (3E9, 1E10, 6E10, 1.6E11 and 2.5E11 genome copies [GC] per eye). The rescue treatment consisted of intravitreal anti-VEGF in the case of vision loss of 5 or more ETDRS letters; persistent, increased or new intra/subretinal fluid on OCT; or the appearance of new macular hemorrhage. The aqueous levels of the encoded protein were observed to increase in a dose-dependent manner in the five subgroups, with the RGX-314 protein reaching 260.5 ng/mL in 1 year in the 6E10 cohort (six patients). In the same sub-group at 24 months, BCVA was improved by 14 ETDRS letters, and the central retinal thickness remained stable at the baseline. BCVA remained stable at 2 years, with changes within one ETDRS line in cohort 2, 4, and 5. The rescue treatment after 2 years was necessary in all cohorts, with a lower mean number in the higher dose cohorts (2.8, 4.4 and 2 ranibizumab injections in cohorts 3, 4 and 5, respectively), whereas the first two cohorts received a higher number of rescue injections (10.3 and 9.3 in cohort 1 and 2, respectively). RGX-314 showed a good tolerability; overall, the intervention demonstrated no severe adverse events at the lower doses; 2 participants in the highest dose group developed retinal pigmentary changes that resulted in vision loss. As a consequence, the protocol was amended [166]. These results encouraged several more studies on RGX-314 safety and efficacy on nAMD patients: a phase II trial (ClinicalTrials.gov: NCT04832724) comparing the effects of two different doses in two subretinal formulations, the clinical and the eventual commercial formulations, a phase II trial (ClinicalTrials.gov: NCT04514653) comparing 3 different doses of RGX-314 with ranibizumab, a randomized combination of a fixed dose of RGX-314 with either topical or local steroid formulations post-treatment, and a 5-year follow-up trial with a sub-study on the affected fellow-eye (ClinicalTrials.gov: NCT03999801). Unfortunately, no conclusive data from these trials are currently available [167].

The subretinal injection of viral vectors encoding endostatin and angiostatin, which are endogenous inhibitors of angiogenesis, showed good preclinical results on mice with laser-induced neovascularization [168,169]. These results prompted a phase I clinical study (ClinicalTrials.gov: NCT01301443, Oxford Biomedica) on subretinal treatment with the non-replicating bicistronic EIAV vector encoding both endostatin and angiostatin (RetinoStat) on humans with advanced nAMD. The trial enrolled 21 patients that were divided into three cohorts receiving a different treatment dose (4E4, 2.4E5 and 8E5 transduction units [TU]). The gene therapy was safe, well tolerated and generated a sustained expression of angiostatin and endostatin, which was detected in aqueous humor samples of eight patients for up to 2.5 years and in 2 patients for more than 4 years. Unfortunately, despite a documented reduction of fluorescein leakage, the treatment produced no functional improvement [170].

The complement cascade activation with membrane attack complex (MAC) accumulation has been observed to be upregulated in AMD patients, with consequent RPE cell damage. This process is thought to play an important pathogenetic role in both atrophic and nAMD [171]. CD59 is a membrane that prevents MAC formation on the cell membrane in the final phases of the complement cascade leading to cell lysis. Therefore, a soluble form of this molecule has been studied for gene therapy applications in dry AMD and nAMD. For the latter, a phase I trial (NCT03585556) adopting the intravitreal injection of AAVCAGsCD59, a viral vector encoding for soluble CD59 was initiated, but the results have yet to be made available.

Another gene-based therapeutic strategy to reduce the expression of VEGF and its receptors is gene silencing with siRNAs. As previously mentioned, these artificial RNA strands are capable of forming complexes with complementary mRNA, selectively silencing their expression after transcription. Preclinical successes on nAMD models led to clinical trials adopting bevasiranib, a 21-nt RNA silencing the VEGF encoding mRNA, and ANG 211745, a 21-nt RNA that silences the mRNA encoding for FLT-1, also known as VEGFR1.

Bevasiranib was the first siRNA approved for IVT use in clinical trials on nAMD patients. This treatment was proved safe in a phase I trial (ClinicalTrials.gov: NCT00722384, OPKO Health) adopting five dosing regimens (0.1, 0.33, 1, 1.5 and 3 mg). Since bevasiranib inhibits VEGF synthesis, but does not affect the preexisting VEGF levels, and its efficacy as a monotherapy was demonstrated insufficient in a phase II trial (ClinicalTrials.gov: NCT00259753), resulting in BCVA loss and neovascular lesion enlargement, this gene therapy was subsequently associated with intravitreal ranibizumab in patients affected by nAMD in a phase III trial (ClinicalTrials.gov: NCT00499590). Despite its promising rationale, the trial did not meet its primary endpoint and was terminated.

Another phase III trial testing the efficacy and safety of the combined bevasirinab-ranibizumab therapy was aborted even before the enrollment started due to concerns regarding its Toll-like receptor (TLR) action, which was detected in murine models, and observed to induce RPE cell apoptosis [105].

The first phase I trial (ClinicalTrials.gov: NCT00363714, Allergan, Dublin, Ireland) assessing another intravitreal si RNA (ANG 211745) safety in nAMD patients showed good results, but the following phase II trial failed to reach its therapeutic targets [106]. Further concerns emerged on the TLR3 pathway activation; thus, more specific gene treatments were developed to overcome this limitation. In a phase I trial, the intravitreal injection of PF-045236 (ClinicalTrials.gov: NCT00725685), a 19-nt siRNA silencing the hypoxia-induced gene RTP801, was tested on patients with MNV or DME and was demonstrated to be safe and well tolerated [172]. In the subsequent phase II MONET trial (ClinicalTrials.gov: NCT00713518), this gene therapy showed no superiority in improving BCVA when compared with ranibizumab, but the two treatments combined showed synergetic efficacy [173].

The need to frequently combine treatment regimes in order to obtain the best outcome for the patient highlights the complexity of nAMD pathogenesis and, consequently, the need for a multifactorial therapeutic approach. To this purpose, a single gene therapy that regulates the expression of different proangiogenic molecules simultaneously would represent the ideal solution; preclinical studies by Askou et al. evaluated multigenic lentiviral vectors in human cells and in mouse retina that encode for both PEDF and anti-VEGF miRNA [174].