Animal Models and New Approach Methodologies in Retinal Disease Research: A Comprehensive Review

Abstract

Studies employing animal models play a pivotal role in advancing our understanding of the pathophysiology of retinal diseases. These models enable the investigation of molecular and cellular mechanisms underlying retinal structural damage, as well as the assessment of genetic and environmental factors contributing to disease development. The application of appropriate experimental models provides essential insights into the progression of degenerative processes and tissue responses to therapeutic interventions. The advancement of modern molecular biology and genetic engineering techniques has facilitated the development of increasingly precise animal models, which have proven crucial for identifying pathological alterations occurring in the course of retinal diseases. In recent years, research has demonstrated that, depending on the model employed, the observed changes may involve inflammatory processes, oxidative stress, photoreceptor dysfunction, extracellular matrix remodeling, and aberrant glial cell responses. It has also been shown that the nature and dynamics of these alterations vary according to the specific disease entity and the animal species used. The aim of this review is to compile and systematize current knowledge regarding the most commonly used animal models in retinal disease research and to discuss their utility in analyzing potential pathogenetic mechanisms and therapeutic targets. The review also highlights emerging complementary research strategies associated with New Approach Methodologies (NAMs), including retinal progenitor and iPSC-derived cell-based approaches, advanced retinal imaging techniques, and alternative experimental platforms such as the chorioallantoic membrane (CAM) assay, which may support translational retinal research and reduce reliance on traditional animal models. The authors hope that this work will contribute to the refinement of preclinical research methodologies and, through an improved understanding of the processes underlying the development of retinal diseases, facilitate the advancement of more effective diagnostic and therapeutic strategies in the future.

1. Introduction

Retinal diseases represent one of the leading causes of visual impairment and blindness in developed countries. These conditions include both inherited (monogenic) disorders and pathological changes associated with aging or environmental factors. In research on emerging therapies—including cell-based and gene-based approaches the use of appropriate animal models constitutes a critical stage [1,2], as these models allow for the analysis of pathway-specific disease mechanisms and the assessment of the safety and efficacy of therapeutic interventions. Animal models of retinal diseases are essential for investigating the mechanisms underlying progressive vision loss and for evaluating emerging therapeutic strategies [3,4]. However, no single model fully reproduces the complexity of human retinal pathology, and each system captures only selected molecular, structural, or functional aspects of disease.

Animal models remain important tools for understanding complex ocular structures, the blood–retinal barrier, the neurovascular unit, and systemic pharmacokinetic and toxicological responses; however, their use should be more focused and prudent, and they should be integrated with NAMs in a complementary manner. In March 2026, the FDA additionally published a draft guidance for the validation of NAMs, emphasizing four key principles: context of use, human biological relevance, technical characterization, and fit-for-purpose validation. Small animal models, particularly murine systems, provide powerful platforms for studying gene-specific mechanisms, inflammatory signaling, oxidative stress, metabolic dysfunction, and early-stage neurodegeneration due to their genetic manipulability and rapid breeding cycles [5]. Nevertheless, important anatomical differences, including the absence of a macula and distinct photoreceptor composition, limit their translational fidelity for modeling human central retinal degeneration [3].

Many inherited retinal degeneration models additionally demonstrate convergence of distinct genetic defects toward shared downstream pathogenic pathways. For example, mutations affecting phototransduction proteins or ciliary function, including RPE65, PDE6A/B, and ciliopathy-associated genes, are associated with progressive accumulation of cellular stress, impaired protein turnover, mitochondrial dysfunction, and activation of inflammatory signaling cascades within the retina [3,6]. Several murine and canine models exhibit increased oxidative stress accompanied by reactive oxygen species accumulation and chronic microglial activation, which contribute to secondary photoreceptor degeneration beyond the initial genetic defect [3]. Similarly, metabolic dysregulation caused by altered photoreceptor energy demand and impaired retinal pigment epithelium function has been implicated in disease progression across multiple inherited retinal disorders [6].

Proteostasis imbalance also represents a critical shared mechanism in retinal degeneration models. Defective protein folding, impaired trafficking of phototransduction components, and insufficient clearance of misfolded proteins may induce endoplasmic reticulum stress and trigger apoptotic pathways in photoreceptors [6]. These convergent mechanisms are particularly important for translational research because they suggest that therapeutics targeting inflammation, oxidative damage, or cellular stress responses may complement gene-specific interventions and potentially provide broader neuroprotective effects across genetically heterogeneous retinal diseases.

In contrast, large animal models such as dogs and pigs more closely resemble human ocular anatomy, retinal organization, and surgical accessibility, thereby offering greater predictive value for translational and preclinical studies, especially in the context of gene therapy, retinal imaging, and device-based interventions [3,4]. Canine models have been particularly valuable for evaluating inherited retinal degeneration therapies, including RPE65-targeted approaches, while porcine models better replicate vitreoretinal architecture and retinal size relevant to human surgical procedures [4]. However, these models are associated with higher maintenance costs, longer generation times, reduced scalability, and limited genetic tractability compared with rodents.

Accordingly, the broad spectrum of available animal models enables complementary investigation across multiple levels of biological complexity, from mechanistic studies of disease pathways to translational assessment of therapeutic safety and efficacy. The selection of an appropriate model therefore depends on the specific experimental objective, including the targeted retinal structure, disease mechanism, and intended translational application [3,4,6].

An important area of investigation is also the assessment of the integrity of the inner blood–retinal barrier. Animal models used for this purpose allow for the analysis of vascular permeability, inflammatory processes, and imaging modalities for detecting vascular leakage, which is of critical importance for understanding the pathogenesis of diabetic retinopathy and other diseases associated with microvascular dysfunction [7].

2. Materials and Methods

A comprehensive literature search was conducted in English using the PubMed, PMC, and Google Scholar databases, covering publications from 1985 to 2025. The search employed the following keywords: “retinal disease,” “retinal degeneration,” “animal models,” “rodent models,” “canine models,” “NHP models,” and “retinal therapy.” The initial search yielded a substantial number of results, which were screened for relevance based on titles and abstracts. Exclusion criteria included non-English publications and studies not directly related to retinal diseases or experimental animal models. After applying the inclusion and exclusion criteria, a total of 255 articles were selected for in-depth review. Reference management software (Zotero v.7.0.16) was used to organize the retrieved articles and remove duplicates.

3. Types of Retinal Diseases

One of the most prevalent categories comprises disorders characterized by damage to photoreceptors (cones and rods) and retinal pigment epithelium (RPE) cells, ultimately leading to cell death and loss of visual function [8].

RPE cells play a critical role in maintaining photoreceptor homeostasis, including through phagocytosis of shed outer segment fragments of rods and cones, ion transport, and the production of trophic factors. Damage to the RPE frequently results in secondary photoreceptor degeneration [9].

The group of diseases involving primary photoreceptor degeneration includes, among others, retinitis pigmentosa (RP), Leber congenital amaurosis (LCA), and other cone–rod dystrophies [1,10].

In genetic models, photoreceptor degeneration arises from mutations in specific genes, such as RHO, PDE6B, or RPE65 [5,11]. The rd1 and rd10 mouse strains are classical models of RP, characterized by progressive rod loss followed by secondary cone degeneration. These models have enabled detailed investigations into mechanisms of apoptosis, oxidative stress, and visual cycle dysfunction [5,11,12,13].

Dogs carrying mutations in RPE65 represent an excellent model of LCA, in which gene therapy was first successfully tested. This work ultimately led to the development of the first clinically approved gene therapy, Luxturna [14,15].

As noted by Thomas et al. (2025), models of photoreceptor degeneration are particularly valuable for evaluating cell-based therapies aimed at replacing lost photosensitive cells or promoting their survival through transplantation of retinal pigment epithelium (RPE) cells [1]. In experimental studies, retinal progenitor cells (RPCs) or cells differentiated from induced pluripotent stem cells (iPSCs) are delivered into the subretinal space, and their effects are assessed using morphological methods (histology, immunofluorescence) as well as functional techniques (electroretinography [ERG], optokinetic testing) [1].

In the context of vascular retinal diseases, such as diabetic retinopathy or retinal ischemia, animal models constitute essential tools for analyzing mechanisms of pathological angiogenesis and for evaluating the efficacy of therapies targeting vascular endothelial growth factor (VEGF) [16,17]. The most commonly employed inducible models reproduce conditions of hyperglycemia or hypoxia, thereby enabling investigation of the molecular mechanisms underlying disease development and assessment of potential therapeutic strategies, including pharmacological treatments and surgical interventions [17].

The chick chorioallantoic membrane (CAM) assay is widely used to investigate pathological angiogenesis in retinal diseases, including proliferative diabetic retinopathy. In 7–8-day-old chick embryos, which lack a fully developed immune system, angiogenesis induced by pathological stimulants, such as tumor-derived factors or hypoxia, can be examined, thereby modeling neovascularization observed in retinopathy [18,19]. Angiogenic effects are quantified by counting blood vessels within a defined area using a stereomicroscope; in the shell-less variant, embryos are cultured in Petri dishes, allowing assessment of a larger CAM surface [19]. Originally described in the 1970s, this model continues to demonstrate its utility in the preclinical screening of anti-angiogenic agents prior to studies in mammals [20].

4. Animal Models of Retinal Diseases: Genetic and Phenotypic Determinants

To facilitate orientation within the diversity of available animal models of retinal diseases,

Table 1. Selected Animal Models of Retinal Diseases—Genetic and Phenotypic Characteristics and Research Applications.

Species	Model/Genes	Human Disease	Key Phenotypic Features	Primary Research Application	References
Dog	PDE6A	Retinitis pigmentosa	Accumulation of cGMP, rod photoreceptor degeneration, thinning and loss of the outer nuclear layer (ONL)	Phototransduction, gene therapy	Tuntivanich et al. [21] Mowat et al. [22]
Dog	PDE6B	Retinitis pigmentosa	Elevated cGMP, outer segment abnormalities	AAV therapy	Pichard et al. [23]
Dog	RPE65	Leber congenital amaurosis, Retinitis pigmentosa	Visual cycle defects, ERG improvement following AAV treatment	Reference model for gene therapy	Acland et al. [24] Le Meur et al. [25]
Dog	CNGA3/CNGB3	Achromatopsia	Severe cone dysfunction	CNG channel function	Wissinger et al. [26] Tanaka et al. [27] Sidjanin et al. [28]
Dog	CNGB1	RP45	Slow rod degeneration with cone preservation	Long-term therapies	Winkler et al. [29] Petersen-Jones et al. [30]
Dog	BEST1 (CMR)	Bestrophinopathy	Multifocal retinal detachments	Retinal pigment epithelium disorders, optical coherence tomography	Guziewicz et al. [31] Zangerl et al. [32]
Cat	CEP290 (rdAc)	Leber congenital amaurosis	Slow degeneration with relative preservation of the area centralis	“Mini-gene” therapies	Menotti-Raymond et al. [33] Coppieters et al. [34]
Cat	CRX (CrxRdy)	Leber congenital amaurosis, retinitis pigmentosa	Early-onset, dominant photoreceptor dystrophy	Transcriptional regulation	Menotti-Raymond et al. [35] Occelli et al. [36]
Cat	AIPL1	LCA4	Very early photoreceptor loss	PDE6 stabilization	Yadav et al. [37] Kolandaivelu et al. [38]
Cat	RDH5	Fundus albipunctatus	Delayed visual recovery	Visual cycle	Gonzalez-Fernandez et al. [39] Nakamura et al. [40] Yamamoto et al. [41] Kuehlewein et al. [42]
Cat	Diabetic retinopathy (post-pancreatectomy)	diabetic retinopathy	Mild microangiopathy, absence of proliferative diabetic retinopathy	Long-term diabetic retinopathy	Mansour et al. [43] Hatchell et al. [44]
Mouse	STZ/alloxan	Diabetic retinopathy	Neurodegeneration, retinal ganglion cell apoptosis	Early stages of diabetic retinopathy	Kumar et al. [45] Yang et al. [46] Martin et al. [47] Feit-Leichman et al. [48]
Mouse	Ins2Akita	Diabetic retinopathy (type 1)	Pericyte loss, increased vascular permeability	Vascular models	Gastinger et al. [49]
Mouse	Ccl2−/−Cx3cr1−/−	Age-related macular degeneration	Drusen-like deposits, retinal pigment epithelium alterations	Inflammation, aging	Ross et al. [50] Chan et al. [51]
Mouse	CFH Y402H	Age-related macular degeneration	Complement activation	Immunopathogenesis of age-related macular degeneration	Toomey et al. [52]
NHP	PDE6C	Cone dysfunction	Central vision impairment	Gene therapy	Moshiri et al. [2]
NHP	BBS7	Bardet–Biedl syndrome	Central dystrophy with systemic manifestations	Syndromic models	Peterson et al. [53]
NHP	CLN7	Batten disease	Retinal and central nervous system neurodegeneration	Neurodegenerative diseases	McBride et al. [54]
NHP	Diabetic retinopathy	Diabetic retinopathy	Restricted, variable phenotype	Pathophysiology of diabetic retinopathy	Kim et al. [55] Tso et al. [56]
Rat	STZ-induced diabetes	Diabetic retinopathy	Retinal vascular leakage, Müller cell activation, VEGF upregulation	Blood–retinal barrier dysfunction, anti-VEGF therapy	Oliveira et al. [57] Wu et al. [58]
Rat	Oxygen-induced retinopathy (OIR)	Retinopathy of prematurity	Retinal neovascularisation, vaso-obliteration	Angiogenesis, anti-VEGF treatment	Budd et al. [59] McCloskey et al. [60]
Rat	Microbead/episcleral vein cauterisation	Glaucoma	Elevated intraocular pressure, retinal ganglion cell loss	Neuroprotection, glaucoma pathophysiology	Ruzafa et al. [61] Araujo et al. [62] Du et al. [63]
Rat	Optic nerve crush (ONC)	Optic neuropathy	Axonal degeneration, retinal ganglion cell apoptosis	Axonal regeneration, stem cell therapies	Liu et al. [64] Xie et al. [65] Nguyen et al. [66]
Rat	NMDA excitotoxicity model	Retinal neurodegeneration	Retinal ganglion cell degeneration, oxidative stress	Excitotoxicity, neuroprotective therapies	Lam et al. [67] Suo et al. [68] Zhang et al. [69]

presents a concise overview of the most commonly used models, together with their genetic background, phenotypic characteristics, and principal research applications. The table serves as a reference point for the detailed descriptions of individual models discussed in the subsequent sections of this chapter.

4.1. Canine Models

Dogs represent one of the best-characterized large-animal models of progressive retinal atrophy (PRA) and have played a central role in translational retinal research. Over the past decades, at least 31 recessive and dominant mutations associated with retinal degeneration have been identified across different dog breeds, enabling the investigation of diverse pathogenic mechanisms underlying inherited retinal disease [70]. Importantly, canine PRA models do not fully replicate all aspects of human retinal degeneration; however, they provide high translational value due to their retinal size, cone-rich visual streak, long-term disease progression, and suitability for surgical and gene therapy approaches [70,71]. These features were instrumental in the preclinical development of the first FDA-approved RPE65-targeted gene therapy for Leber congenital amaurosis (LCA), highlighting the predictive validity of canine models for translational ophthalmic research [71].

The spectrum of canine PRA mutations encompasses defects in phototransduction, visual cycle function, ion channel activity, ciliogenesis, synaptogenesis, and retinal development, thereby allowing investigation of both gene-specific and convergent degenerative pathways, including oxidative stress, inflammatory signaling, and metabolic dysfunction within photoreceptors [70]. Nevertheless, important limitations remain, including breed-specific variability, restricted genetic scalability compared with murine systems, and incomplete reproduction of the full heterogeneity observed in human retinitis pigmentosa [70,71].

4.1.1. Phototransduction Defect Models: Rhodopsin and Phosphodiesterase Mutations

Mutations affecting rhodopsin and rod phosphodiesterase subunits (PDE6A and PDE6B), as well as the SAG gene, constitute some of the most extensively studied canine PRA models. In PDE6A-mutant dogs, a frameshift mutation introduces a premature stop codon leading to complete loss of functional phosphodiesterase activity [71,72]. This defect causes pathological accumulation of cGMP in photoreceptor outer segments, prolonged Ca²⁺ influx, mitochondrial dysfunction, and activation of non-apoptotic photoreceptor cell death pathways associated with oxidative stress and proteostasis imbalance [73,74]. The model closely reproduces several key molecular and structural features of human autosomal recessive RP, including progressive rod degeneration followed by secondary cone loss, thereby providing strong predictive validity for evaluating gene supplementation strategies [72,75].

Importantly, disease progression in these dogs can be longitudinally monitored using clinically relevant imaging modalities such as indirect ophthalmoscopy and OCT, further enhancing their translational utility [37,75]. AAV-mediated gene therapy has demonstrated substantial preservation of retinal morphology and restoration of rod function in PDE6A-mutant dogs, supporting the relevance of this model for human therapeutic development [37,75]. However, despite these advantages, canine models remain limited by relatively small cohort availability, inter-animal variability, and higher maintenance costs compared with rodent systems [71].

A similar mechanism is observed in Irish Setter dogs carrying PDE6B mutations, in which nonsense variants disrupt outer segment morphogenesis and elevate intracellular cGMP levels [21]. These models successfully reproduce early photoreceptor dysfunction characteristic of human RP and have shown responsiveness to AAV-mediated gene replacement therapy [21]. Nevertheless, as with other monogenic retinal degeneration models, they primarily capture pathway-specific aspects of disease and do not fully encompass the complex inflammatory and metabolic interactions contributing to retinal degeneration in human patients.

4.1.2. Dogs as Models of RPE65-Associated Diseases

Mutations in canine RPE65 provide one of the best-characterized large-animal models of human Leber congenital amaurosis. RPE65 encodes an isomerase essential for the visual cycle, catalyzing the conversion of all-trans-retinyl esters to 11-cis-retinal. Its deficiency leads to accumulation of all-trans-retinal and formation of toxic bisretinoids such as A2E, a major component of lipofuscin, contributing to photoreceptor dysfunction and degeneration. Gene augmentation therapy in affected dogs improves visual behavior and restores ERG responses, while significantly slowing photoreceptor loss [38,76,77,78]. However, in humans, the therapeutic effect is generally weaker and less durable, likely reflecting interspecies differences in visual cycle efficiency, chronic accumulation of retinoid by-products, and retinal remodeling processes [79].

At the mechanistic level, disrupted retinoid cycling induces oxidative and metabolic stress, including increased reactive oxygen species and mitochondrial dysfunction in photoreceptors and RPE. Accumulation of toxic retinoid derivatives such as A2E further impairs lysosomal function and promotes lipofuscin deposition [80,81]. These stressors are associated with activation of inflammatory pathways, including upregulation of cytokines such as TNF-α, IL-1β, IL-6, and chemokine CCL2, as well as complement activation (e.g., C3 deposition and C5b-9 formation). Reactive gliosis, reflected by increased GFAP expression in Müller cells and microglial activation (Iba1+ cells), further contributes to a chronic inflammatory microenvironment that may accelerate secondary photoreceptor loss [39,82].

4.1.3. Canine Ciliopathy Models and Associated PRA

A substantial portion of canine PRA is linked to ciliary defects, as the photoreceptor outer segment is a modified primary cilium. Data from Gene Ontology, SYSCILIA Gold Standard, and CiliaCarta indicate that nearly a thousand proteins participate in ciliary function [40,41,83]. Mutations in genes such as BBS4, C2orf71, CCDC66, FAM161A, NPHP4, NPHP5, RPGR, RGRIP1, and TTC8 have been reported in dogs, leading to defective cilium formation or maintenance and consequent photoreceptor degeneration [42]. Mutations in STK38L, LRIT3, and MERTK affect retinal development, congenital stationary night blindness (CSNB), and RPE phagocytosis. Other genes linked to PRA include ADAM9, PRCD, RD3, NECAP1, PPT1, and SLC2A3 [40,41,83].

4.1.4. Models of Developmental Retinal Degeneration

Norwegian Elkhounds with a Stk38L mutation represent a canine model of developmental retinal degeneration [84]. The mutation involves a SINE insertion in exon 4, resulting in exon skipping in mature mRNA [34]. Retinal development in young dogs appears normal initially, but from week 8, mislocalized rhodopsin, outer segment abnormalities, and increased TUNEL-positive cells are observed, alongside a transient proliferation of rod-like photoreceptors. This process ceases around week 14, followed by gradual outer nuclear layer (ONL) loss by approximately week 48 [43].

4.1.5. Canine Achromatopsia: CNGA3 and CNGB3 Mutations

Dogs are also valuable models of achromatopsia. In humans, roughly 25% of cases are linked to CNGA3 mutations [22], and canine mutations include R424W and V644del. The R424W mutation destabilizes the salt bridge between the S6 and S4–S5 helices, impairing proper protein folding and preventing CNG current generation [85]. The V644del mutation disrupts the CLZ domain, essential for channel assembly, resulting in partial loss of function [85]. CNGB3 mutations, including a complete gene deletion (Cngb3^-/-) and a missense D262N mutation, account for over half of human achromatopsia cases [23,44,86].

4.1.6. Rod CNG Channels and CNGB1 Mutations

Mutations in the CNGB1 gene in Papillon dogs include insertions and deletions that lead to frameshift variants, skipping of exon 26, and production of a truncated CNGB1 protein that fails to correctly localize to the rod outer segments [24,87]. This canine phenotype closely resembles human RP45, a rare form of autosomal recessive retinitis pigmentosa accounting for less than 4% of arRP cases [45]. Clinically, rod degeneration progresses relatively slowly, while cone photoreceptors remain preserved for extended periods, allowing maintenance of functional daylight vision for many years [87]. Gene augmentation therapy administered in young dogs (3.5–6.5 months of age) significantly slows disease progression, highlighting the unusually broad therapeutic window of this model and its translational relevance.

Beyond the structural consequences of the mutation, the CNGB1 canine model also provides important insight into shared pathogenic mechanisms underlying retinal degeneration. CNGB1 encodes the β-subunit of the cyclic nucleotide-gated (CNG) channel in rod photoreceptors, a key component of phototransduction [46]. Defective assembly or trafficking of the CNG channel disrupts ion homeostasis and photoreceptor signaling, leading to chronic cellular stress. Misfolded or truncated CNGB1 proteins are likely retained within the endoplasmic reticulum (ER), where they can activate proteostasis-related stress pathways, including the unfolded protein response (UPR). Persistent ER stress has been implicated in photoreceptor apoptosis across multiple inherited retinal degenerations and may contribute to the slow but progressive rod loss observed in this model [47,48].

In addition, secondary inflammatory responses increasingly appear to play an important role in disease progression in CNGB1-associated degeneration. Rod photoreceptor stress and degeneration can activate Müller glia and retinal microglia, leading to release of pro-inflammatory cytokines, complement activation, and remodeling of the retinal environment. Although inflammation is initially protective, chronic neuroinflammatory signaling may accelerate degeneration of surviving rods and contribute to secondary cone dysfunction. The relatively prolonged survival of cones in CNGB1 mutant dogs makes this model especially valuable for studying the transition from primary rod pathology to secondary cone degeneration, a hallmark shared by many forms of retinitis pigmentosa [24,88].

4.1.7. Canine Bestrophinopathies: Canine Multifocal Retinopathy (CMR)

Dogs also model BEST1-related diseases, equivalent to human bestrophinopathy (BVMD) [49]. Canine multifocal retinopathy (CMR) is an autosomal recessive disorder with a predictable course. Identified mutations include p.Arg25Ter (cmr1) [25], p.Gly161Asp (cmr2), and two exon 10 variants—p.Pro463fs and p.Gly489Val (cmr3) [89]. All result in a similar phenotype characterized by multifocal retinal detachments with subretinal fluid accumulation, especially in cone-rich regions. Initial OCT changes appear around week 11, involving central elevations that progressively develop into macro- and micro-detachments [25].

4.2. Feline Models

Cats exhibit diverse inheritance patterns of congenital retinal degenerations, including autosomal recessive, autosomal dominant, and X-linked forms. The best-characterized is the autosomal recessive form of progressive retinal atrophy (PRA), designated rdAc, which occurs in certain high-risk breeds and results from a mutation in the CEP290 gene [38,90,91]. The degeneration is late-onset and slow, distinguishing it from many human CEP290-associated phenotypes, including severe forms of Leber congenital amaurosis (LCA), which account for approximately 20% of cases [92,93]. Because CEP290 is a ciliary protein, mutations in this gene can also lead to multisystem syndromes such as Meckel [94,95], Joubert [50,51,96], and Senior–Løken [97,98]. In rdAc cats, gradual photoreceptor loss and ellipsoid zone atrophy occur, with initial preservation of the area centralis, reflecting features of early human LCA [52].

4.2.1. CRX Model (CrxRdy): Spontaneous Homeobox Mutation

Another significant model is the CrxRdy cat, harboring a spontaneous single-base deletion in the CRX gene, resulting in a frameshift and truncated protein lacking the C-terminal 114 amino acids [91]. Analogous human mutations cause LCA7 and other forms of severe early-onset blindness [99]. This feline model recapitulates dominantly inherited, rapidly progressive photoreceptor degeneration consistent with CRX-related dystrophies in humans, including retinitis pigmentosa (RP), cone dystrophies, and macular degeneration [53,54,100,101].

CrxRdy kittens exhibit non-recordable ERG from birth, with arrested rod responses that are completely extinguished by 20 weeks. Photoreceptor outer segments develop abnormally, and degeneration begins centrally, following a cone–rod dystrophy pattern [36]. The mutation exerts a dominant-negative effect: the mutant protein retains the DNA-binding domain but lacks the transactivation domain, disrupting transcriptional regulation of photoreceptor genes, including RHO and ARR3 [36,99]. This model provides a unique large-animal platform to study LCA7 pathomechanisms and CRX dysfunction.

4.2.2. AIPL1 Model in Persian Cats (LCA4)

Persian cats carry a spontaneous mutation in AIPL1, corresponding to a recessive, severe retinal dystrophy phenotypically similar to human LCA4 [34]. AIPL1 is essential for proper maturation and stabilization of PDE6 in photoreceptors and accounts for less than 10% of human LCA cases [102]. A C>T mutation at c.577 introduces an early stop codon, producing a short, nonfunctional protein [103]. Affected cats display very early and rapid photoreceptor loss, resulting in blindness before adulthood [104]. This model is instrumental for studying LCA4 pathogenesis and PDE6-related therapies [55,56,105,106,107].

4.2.3. RDH5 Model: Feline Analog of Fundus Albipunctatus and Cone Dystrophy

The most recently described feline model involves a mutation in RDH5, encoding the enzyme that converts 11-cis-retinol to 11-cis-retinal in the retinal pigment epithelium [108]. In humans, RDH5 mutations typically cause fundus albipunctatus [109,110], but may also lead to macular dystrophy [26,111,112] or cone dystrophy [27].

The feline RDH5 model recapitulates key human disease features, including severely delayed photoreceptor recovery following light exposure and central retinal changes. This is particularly valuable, as Rdh5 knockout mice do not exhibit a comparable phenotype [113].

4.2.4. Detailed Characterization of CEP290 Mutation in rdAc Cats

The rdAc model is one of the best-characterized large-animal retinal degeneration models. An intronic c.6966+9T>G mutation creates an alternative, stronger acceptor splice site, resulting in the inclusion of four additional nucleotides in exon 50, causing a frameshift and premature stop codon [90]. Some transcripts still use the wild-type acceptor site, allowing production of small amounts of normal mRNA. The presence of partially functional truncated protein explains the milder disease course in cats compared to the severe, often lethal phenotypes in humans [114]. This model has high translational relevance, particularly for testing “minigene”-based therapies, as the full CEP290 sequence exceeds the packaging capacity of AAV vectors.

4.2.5. Feline Models in Diabetic Retinopathy Research

Unlike genetic LCA models, feline diabetic retinopathy (DR) studies mainly rely on diabetes induction via pancreatectomy, sometimes combined with alloxan administration [28,115,116]. Animals become hyperglycemic within 1–2 weeks. The earliest microangiopathy changes—capillary basement membrane thickening—appear around 3 months, with no significant endothelial or pericyte loss even up to 10 months [108].

This relative resistance of retinal capillaries to early cell dropout may reflect species-specific differences in pericyte survival pathways and a slower progression of hyperglycemia-induced oxidative injury compared with rodent models.

Importantly, vascular pathology in feline DR is not driven solely by hyperglycemia but involves a complex interplay of metabolic dysregulation, oxidative stress, low-grade chronic inflammation, and impaired neurovascular coupling. Hyperglycemia-induced overproduction of ROS, particularly within retinal mitochondria, contributes to endothelial dysfunction, pericyte stress, and altered intracellular signaling, even in the absence of overt early cell loss [117,118,119,120]. In parallel, activation of the polyol pathway and accumulation of advanced glycation end-products (AGEs) promote ECM stiffening, basement membrane thickening, and further amplification of oxidative injury through receptor for AGEs (RAGE)-mediated signaling [117,121].

In addition, hypoxia-inducible factor-1α stabilization under relative retinal hypoperfusion promotes upregulation of pro-angiogenic mediators, including VEGF, which contributes to subtle blood–retina barrier impairment and early microvascular leakage. However, compared with human disease, the downstream angiogenic cascade appears attenuated, limiting progression to overt proliferative stages [122,123]. Dysfunction of Müller cells and retinal neurons further disrupts metabolic support and potassium/glutamate homeostasis, weakening neurovascular coupling and impairing fine regulation of capillary blood flow [124].

Inflammatory mechanisms also play a contributory role, with increased expression of cytokines such as TNF-α and adhesion molecules like ICAM-1 promoting endothelial activation, although leukostasis is generally less pronounced than in rodent diabetic models. Together with pericyte stress and subtle endothelial dysfunction, these processes lead to gradual breakdown of vascular autoregulation and progressive capillary nonperfusion over time [123,124].

Extracellular matrix remodeling, including increased collagen deposition and basement membrane thickening, further stabilizes early microangiopathic changes but may paradoxically limit vessel regression and hemorrhagic transformation [122]. This may partly explain the relatively slow and incomplete progression toward advanced proliferative diabetic retinopathy in this model [125].

After 5 years of diabetes, the first microaneurysms appear; by 6.5 years, small intraretinal hemorrhages are observed in the area centralis [115]. At 7.5 years, capillary nonperfusion and IRMA develop, and by ~8.5 years, early neovascularization foci are detectable. Advanced DR features, such as extensive cotton-wool spots, venous beading, or widespread retinal neovascularization, do not occur. Another study reported a very limited and variable phenotype—only one of two cats developed microaneurysms after 7 years, and none exhibited hemorrhages or ischemia [116]. Notably, only mild cataracts develop, permitting long-term fundus imaging and ERG assessment.

4.3. Rodent Models

4.3.1. Pharmacological Models

Type 1 diabetes can be induced in mice using β-cell–toxic compounds such as streptozotocin (STZ) or alloxan, or via a galactose-enriched diet. STZ remains the most commonly used method for DR induction, producing hyperglycemia within days depending on dose and administration protocol. In STZ models, early astrocyte activation occurs by 4–5 weeks [126], with retinal ganglion cell (RGC) loss observable from week 6 [29]. Inner nuclear layer (INL) and outer nuclear layer thickness decrease after 10 weeks, and vascular and RGC apoptosis is evident between 6 weeks and 6 months of hyperglycemia [30,127]. A galactose diet allows long-term studies (up to 26 months), enabling observation of complications due exclusively to elevated sugar levels [125].

Streptozotocin-induced diabetic rat models are also extensively used in retinal research and provide several advantages over murine systems, particularly in studies requiring repeated intraocular procedures or longitudinal imaging [128,129,130]. Compared with mice, diabetic rats develop more pronounced retinal vascular leakage, oxidative stress, Müller cell activation, as well as inflammatory changes, which facilitates the investigation of blood–retinal barrier dysfunction and neurovascular impairment [128]. Increased VEGF expression, retinal oedema, and enhanced glial fibrillary acidic protein (GFAP) immunoreactivity are commonly observed in these models and reflect pathological alterations seen in the early stages of diabetic retinopathy. Due to their larger eye size and improved surgical accessibility, rats are frequently used in studies evaluating intravitreal therapies, anti-VEGF treatment, and neuroprotective strategies targeting retinal inflammation and vascular dysfunction [17,131].

4.3.2. Endogenous Mutation Models

Spontaneous hyperglycemia occurs in mice carrying mutations in Ins2^Akita, NOD, db/db, and KKAy genes. Ins2^Akita mice model type 1 diabetes, with an Ins2 mutation causing β-cell death. Heterozygous Ins2^Akita retinas show early microglial changes at 8 weeks and peripheral RGC loss by 22 weeks [132]. NOD mice experience autoimmune β-cell destruction, with retinal vascular and cellular pathology appearing after 4–12 weeks of hyperglycemia [133]. Type 2 diabetes models, db/db and KKAy, allow study of obesity, hyperglycemia, and subsequent retinopathy; some exhibit limited utility due to low fertility or spontaneous phenotype regression [134].

4.3.3. Proliferative Retinopathy Models

Among type 1 diabetes genetic models used for DR research, the Akita mouse (Ins2^Akita) is prominent. The spontaneous Ins2 mutation causes chronic hyperglycemia and gradual development of retinal microangiopathy, including pericyte loss and increased vascular permeability. This model is highly reproducible and does not require chemical induction, minimizing toxic confounders [135]. Another approach is the Kimba (trVEGF029) model, which overexpresses VEGF in photoreceptors and is used to study advanced vascular changes and pathological neovascularization [136]. Crossing Kimba with Akita mice recreates features of proliferative DR [137]. These models permit investigation of vascular, neuronal, and glial alterations, though spontaneous regression of neovascularization in oxygen-induced retinopathy (OIR) limits their utility for drug testing.

In addition to murine systems, oxygen-induced retinopathy (OIR) rat models are commonly used in studies of retinopathy of prematurity and ischaemia-driven retinal neovascularisation. In these models, neonatal rats are exposed to fluctuating oxygen concentrations, leading initially to vaso-obliteration and subsequently to pathological neovascularisation during the hypoxic phase [138,139,140]. Compared with mouse OIR systems, rat models better reflect the oxygen fluctuations experienced by premature infants and develop retinal vascular abnormalities similar to those observed in human retinopathy of prematurity [141,142]. Increased VEGF expression, activation of hypoxia-inducible pathways, and abnormal intravitreal neovascularisation are commonly observed in these models, making them useful for studies of angiogenesis and anti-VEGF therapies [142,143,144].

4.3.4. Mouse Models in AMD Research

Genetic mouse models are widely used to study age-related macular degeneration (AMD). One example is the Ccl2^−/−Cx3cr1^−/− double knockout line, in which retinal changes accumulate with age. Histologically, these models can display AMD-like features, such as drusen accumulation, photoreceptor degeneration, and RPE alterations. The extent and progression of retinal changes vary by age and strain, which must be considered in data interpretation [31,145].

A major limitation is the potential presence of the Crb1^rd8 mutation, which alone can cause retinal degeneration. Lines lacking rd8 show significantly milder phenotypes, preventing full recapitulation of the human disease [146,147].

Complement-related models provide an alternative approach. For instance, transgenic mice expressing the human CFH Y402H variant develop sub-RPE deposits and RPE cellular stress, exacerbated by age and high-fat diet, highlighting the complement system’s role in AMD pathogenesis [32].

4.3.5. Rat Models of Retinal Neurodegeneration and Optic Neuropathy

Rat models are extensively employed in studies of retinal neurodegeneration and optic nerve injury due to their relatively large ocular size, well-characterised retinal structure, and suitability for surgical manipulation and intravitreal procedures [148,149]. These models are particularly useful for investigating retinal ganglion cell degeneration, neuroinflammatory responses, oxidative stress, and potential neuroprotective strategies [150].

Experimental glaucoma in rats is commonly induced by episcleral vein cauterisation, laser photocoagulation, or microbead injection into the anterior chamber, resulting in chronic elevation of intraocular pressure and progressive retinal ganglion cell loss [151,152,153]. Histopathological changes observed in these models include optic nerve degeneration, retinal thinning, glial activation, and impaired axonal transport, reflecting several pathological features associated with human glaucomatous neuropathy [154,155]. Due to their reproducibility and relatively gradual disease progression, rat glaucoma models are frequently used in studies evaluating neuroprotective agents and mechanisms involved in retinal ganglion cell death [156,157,158].

Optic nerve crush (ONC) models are widely applied in retinal research to investigate axonal injury and secondary neuronal degeneration. Mechanical compression of the optic nerve leads to rapid r, activation of inflammatory pathways, and progressive axonal damage within the optic nerve [159]. These models have contributed substantially to the understanding of neuronal apoptosis, regenerative signalling pathways, and stem cell-based therapeutic approaches aimed at restoring retinal function following optic nerve injury [160,161,162].

Glutamate excitotoxicity models, most commonly induced by intravitreal N-methyl-D-aspartate (NMDA) administration, are used to examine retinal neuronal damage associated with excessive glutamatergic stimulation. NMDA-induced injury results primarily in selective degeneration of retinal ganglion cells and inner retinal neurons, accompanied by mitochondrial dysfunction, oxidative stress, and inflammatory activation [163,164]. Such models are widely utilised in studies of glaucoma, retinal ischaemia, and other neu pies targeting excitotoxic damage [165,166,167,168].

Light-induced retinal degeneration models in albino rats are commonly employed to investigate photoreceptor injury associated with oxidative stress and age-related retinal degeneration. Exposure to prolonged or high-intensity light results in photoreceptor apoptosis, disruption of retinal pigment epithelium homeostasis, and progressive thinning of the outer nuclear layer [57,169,170]. Increased production of reactive oxygen species and inflammatory mediators has also been reported in these models, making them useful for studies evaluating antioxidant therapies and mechanisms underlying retinal degeneration [58,171].

4.4. Non-Human Primate Models

Non-human primates (NHPs) are used less frequently than rodents but offer high translational relevance due to the presence of a macula and retinal organization closely resembling humans, allowing studies of central vision disorders. High costs, long study durations, and ethical constraints limit NHP use to late preclinical stages [172,173].

4.4.1. Bardet–Biedl Syndrome (BBS7 Mutation)

An NHP model of Bardet–Biedl syndrome arises from a spontaneous BBS7 gene deletion [33]. This model exhibits progressive retinal degeneration, primarily affecting the central retina, along with systemic features typical of the syndrome. The concordance between ocular and systemic phenotypes with human disease makes it valuable for translational studies.

4.4.2. Batten Disease (CLN7 Mutation)

Batten disease associated with CLN7 mutation has been described in Japanese macaques [35]. Affected animals show progressive neurodegeneration involving both the retina and central nervous system, closely mimicking the severe human phenotype, making the model suitable for studying therapies targeting neuronal ceroid lipofuscinoses.

4.4.3. Retinitis Pigmentosa in Cynomolgus Macaques

Non-syndromic retinal dystrophies are reported sporadically in NHPs. In cynomolgus macaques, a familial RP-like phenotype with severe photoreceptor dysfunction has been identified [174]. Genetic studies did not reveal a clear molecular cause, highlighting the complexity of such models.

4.4.4. Bilateral Macular Dystrophy in Rhesus Macaques

A single case of bilateral macular dystrophy in a rhesus macaque was evaluated using clinical imaging and electrophysiology [175]. The symmetric nature suggests a possible genetic basis, though it remains unconfirmed. This model illustrates NHP utility for central vision disorders.

4.4.5. Idiopathic Bilateral Optic Atrophy

Rhesus macaques with idiopathic bilateral optic atrophy serve as a model of selective optic nerve injury [176]. Changes are largely confined to fibers mediating central vision, while photoreceptor function remains relatively intact. This model is used to study optic neuropathies.

4.4.6. Cone Dysfunction Due to PDE6C Mutation

Spontaneous NHP models with homozygous PDE6C mutations exhibit severe cone dysfunction. Phenotypes include macular changes and significant central vision loss, closely resembling human cone disorders, making them particularly relevant for preclinical gene therapy studies [2].

4.4.7. Diabetic Retinopathy Models in NHPs

Non-human primates (NHPs) have also been used in diabetic retinopathy (DR) research. Both type 1 and type 2 diabetes models in NHPs exhibit relatively limited and variable retinal pathology, more closely resembling early non-proliferative stages of human DR than advanced proliferative disease [59,177,178]. A key feature of these models is that hyperglycemia alone is often insufficient to drive robust retinopathy, and systemic comorbidities—particularly arterial hypertension—act as critical disease modifiers. Hypertension exacerbates retinal microvascular injury through increased hydrostatic pressure, endothelial shear stress, and impaired autoregulatory responses of arterioles and capillaries, thereby accelerating breakdown of vascular homeostasis and promoting focal ischemic changes [60].

At the molecular level, early vascular dysfunction in diabetic NHP retina is associated with endothelial nitric oxide dysregulation, increased endothelin-1 signaling, and reduced vasodilatory reserve, contributing to impaired neurovascular coupling [179,180]. In parallel, hyperglycemia-driven mitochondrial ROS production and activation of protein kinase C (PKC) isoforms promote endothelial activation, basement membrane thickening, and subtle blood–retina barrier compromise. However, compared with rodent models, pericyte dropout and capillary acellularity are generally less pronounced, which may explain the slower progression of microangiopathy [181].

Separately, intravitreal VEGF administration has been used as an experimental approach to induce vascular alterations in NHPs [182]. VEGF acts primarily through VEGFR-2 signaling on endothelial cells, leading to phosphorylation of tight junction proteins (e.g., occludin and claudin-5), cytoskeletal rearrangement, and increased vascular permeability. This results in transient breakdown of the blood–retina barrier, characterized by reversible plasma extravasation, interendothelial gap formation, and increased leukocyte-endothelial interactions. Importantly, these changes mimic the permeability-dominant phase of diabetic macular edema but do not reproduce the chronic structural remodeling of diabetic retinopathy, such as sustained pericyte loss, acellular capillaries, or progressive neurodegeneration [183]. VEGF-induced models are therefore best interpreted as acute permeability and leakage models rather than full-spectrum DR models [183].

Complementary NAM-based systems, including retinal endothelial co-cultures, retina-on-chip platforms, and retinal organoids, may help address some translational limitations of current DR models [61,62,63,184]. These systems are particularly valuable for mechanistic studies and drug screening, whereas animal models remain essential for investigating systemic metabolic interactions and long-term disease progression [1,185].

4.5. Comparative Analysis of Translational Relevance and Limitations of Retinal Disease Models

Table 2. Comparative analysis of selected animal models of retinal diseases with emphasis on translational relevance, predictive validity, major limitations, and clinical applicability.

Model/Species	Disease Context	Key Phenotype	Main Translational Strengths	Major Limitations	Clinical and Translation Relevance	References
rd1/rd10 mouse	Retinitis pigmentosa (RP)	Progressive rod degeneration followed by secondary cone loss	Useful for studies of apoptosis, oxidative stress, and visual cycle dysfunction	Accelerated degeneration and limited reproduction of human disease complexity	Widely used in preclinical retinal degeneration studies	[11,12,13]
STZ/alloxan mouse	Diabetic retinopathy (DR)	Hyperglycemia, retinal ganglion cell loss, vascular apoptosis	Reproducible induction of diabetic retinal changes	Limited progression toward advanced proliferative disease	Used in studies of neurovascular dysfunction and retinal injury	[29,30,125,126,127]
Ins2Akita mouse	Diabetic retinopathy	Chronic hyperglycemia with retinal vascular and neuronal changes	Genetic model without chemical induction	Mild retinal phenotype compared with advanced human DR	Frequently used in mechanistic DR studies	[132,135]
Akimba mouse	Proliferative diabetic retinopathy	Retinal neovascularization, hemorrhages, vascular leakage	Combines hyperglycemia and VEGF-driven pathology	VEGF overexpression represents an artificial experimental condition	Model for proliferative DR and anti-VEGF studies	[136,137,186,187]
OIR rat	Retinopathy of prematurity/ischemic retinopathy	Hypoxia-induced retinal neovascularization	Reproduces oxygen fluctuation-related vascular pathology	Spontaneous regression may limit long-term drug studies	Commonly used for angiogenesis and anti-VEGF research	[138,139,140,141,142,143,144]
STZ rat	Diabetic retinopathy	Vascular leakage, oxidative stress, Müller cell activation	Suitable for intravitreal procedures and longitudinal imaging	Faster and more pronounced retinal pathology than in some human cases	Used in anti-inflammatory and anti-VEGF studies	[128,129,130,131]
RPE65 dog	Leber congenital amaurosis (LCA)	Visual cycle dysfunction and photoreceptor degeneration	Large-eye anatomy, clinically relevant imaging, suitability for gene therapy	High maintenance costs and limited cohort availability	Contributed to development of Luxturna gene therapy	[14,15,38,76,77,78,79]
PDE6A/PDE6B dog	Retinitis pigmentosa	Progressive rod degeneration with cGMP accumulation	Longitudinal monitoring and AAV-mediated gene therapy studies	Limited scalability and inter-animal variability	Used for preclinical gene supplementation studies	[21,37,71,72,75]
CNGB1 dog	RP45	Slow rod degeneration with prolonged cone preservation	Broad therapeutic window for gene therapy studies	Limited availability of affected animals	Used in long-term gene augmentation studies	[24,45,87,88]
CEP290/CRX cat	LCA/cone–rod dystrophy	Progressive photoreceptor degeneration and central retinal changes	Mimics several features of human retinal degeneration	Limited molecular and genetic tools compared with mice	Useful for evaluation of gene-based therapeutic strategies	[36,50,51,52,53,54,90,91,92,93,94,95,96,97,98,99,100,101]
Ccl2−/−Cx3cr1−/− mouse	Age-related macular degeneration (AMD)	Drusen-like deposits, RPE atrophy, photoreceptor loss	Useful for studies of inflammation and chemokine signaling	Variable phenotype severity depending on strain background	Used in mechanistic AMD studies	[31,145,146,147]
CFH Y402H mouse	AMD	Sub-RPE deposits and complement activation	Models interaction between complement dysregulation and environmental stress	Does not fully reproduce human AMD phenotype	Used in complement-targeted AMD research	[32,64,65,66,67,68]
Non-human primates (NHPs)	Macular and retinal diseases	Human-like retinal organization with macula	High anatomical and functional similarity to humans	High costs, ethical limitations, long study duration	Used in late-stage preclinical validation	[33,35,59,172,173,174,175,176,177,178,182]

provides a comparative overview of the most commonly used retinal disease animal models that were discussed in this review with emphasis on their major phenotypic characteristics, translational strengths, key limitations, and clinical and translational relevance. The comparison highlights substantial differences between rodent, large-animal, and non-human primate models regarding their predictive value, applicability to human retinal diseases, and suitability for translational research and therapeutic development.

It is important to note that no single animal model fully reproduces all aspects of human retinal disease, emphasizing the need for complementary experimental approaches depending on the specific translational and therapeutic objective.

5. Hybrid Genetic–Environmental Models in Retinal Degenerative Diseases

Retinal degenerative diseases often have a multifactorial origin, in which strong genetic risk factors interact with environmental influences and lifestyle. For example, in age-related macular degeneration, smoking and a high-fat diet increase disease risk in individuals carrying CFH variants [64]. In many single-gene mouse models, pathological changes are mild or appear only late in life. However, introducing an additional stressor, such as dietary imbalance, oxidative stress, excessive light exposure, or metabolic disturbance, can reveal underlying pathology and accelerate disease progression [69,188]. Representative examples of these combined approaches are summarized in Table 3.

These gene–environment combinations produce more robust and clinically relevant phenotypes that better resemble human disease. In age-related macular degeneration, combining variants in complement-related genes such as CFH and C3, or lipid metabolism genes such as APOE and LDLR, with a high-fat diet or increased oxidative stress leads to drusen-like deposits, thickening of Bruch’s membrane, and damage to the retinal pigment epithelium [64,65,69]. In diabetic retinopathy, the combination of genetically driven hyperglycemia with VEGF overexpression results in pronounced neovascularization and vascular leakage [189]. In glaucoma, a genetic predisposition to elevated intraocular pressure, together with aging or additional stressors, leads to earlier loss of retinal ganglion cells [190,191].

Table 3. Representative Hybrid Genetic–Environmental Animal Models of Retinal Degenerative Diseases and Their Major Retinal Phenotypes.

Genetic Background (Species/Strain)	Environmental/Experimental Challenge	Main Retinal Phenotypes	Key References
CFHY402H KI (mouse human CFH allele)	High-fat/cholesterol diet (HFC)	Sub-RPE lipid deposits, complement activation, RPE dysfunction	[65,66]
ApoE/– (C57BL/6)	High-fat diet (HFD)	Retinal hypercholesterolemia, ischemia, inflammation, VEGF upregulation, neovascularization	[192]
Ccl2–/–Cx3cr1–/– (double KO, mouse)	Aging (spontaneous)	Early-onset drusen-like deposits, RPE atrophy, microglial accumulation (AMD-like)	[145]
Ins2Akita (mouse, type 1 diabetes) × trVEGF029 (Kimba mouse)	– (genetic cross)	Severe proliferative DR: retinal NV, hemorrhages, vascular leakage	[187]
DBA/2J (mouse; GpnmbR150X, Tyrp1b mutations)	Aging (2–15 mo)	Iris atrophy, pigment dispersion, chronic IOP elevation, sectoral RGC loss11	[193]

Table 3. Representative Hybrid Genetic–Environmental Animal Models of Retinal Degenerative Diseases and Their Major Retinal Phenotypes.

Genetic Background (Species/Strain)	Environmental/Experimental Challenge	Main Retinal Phenotypes	Key References
CFHY402H KI (mouse human CFH allele)	High-fat/cholesterol diet (HFC)	Sub-RPE lipid deposits, complement activation, RPE dysfunction	[65,66]
ApoE/– (C57BL/6)	High-fat diet (HFD)	Retinal hypercholesterolemia, ischemia, inflammation, VEGF upregulation, neovascularization	[192]
Ccl2–/–Cx3cr1–/– (double KO, mouse)	Aging (spontaneous)	Early-onset drusen-like deposits, RPE atrophy, microglial accumulation (AMD-like)	[145]
Ins2Akita (mouse, type 1 diabetes) × trVEGF029 (Kimba mouse)	– (genetic cross)	Severe proliferative DR: retinal NV, hemorrhages, vascular leakage	[187]
DBA/2J (mouse; GpnmbR150X, Tyrp1b mutations)	Aging (2–15 mo)	Iris atrophy, pigment dispersion, chronic IOP elevation, sectoral RGC loss11	[193]

5.1. Age-Related Macular Degeneration

Several hybrid models have been developed to reproduce the complex pathology of AMD, and complement dysregulation is a central mechanism. Humanized mice carrying the CFH Y402H risk variant develop AMD-like retinal lesions only when exposed to stress [66]. When placed on a cholesterol-rich Western diet, these mice accumulate subretinal lipid-rich deposits, show complement activation in Bruch’s membrane, and develop RPE dysplasia, whereas chow-fed controls do not [66]. Similarly, heterozygous CFH-deficient mice (Cfh^+/–) on a high-fat, high-cholesterol diet develop prominent sub-RPE deposits and recruit mononuclear phagocytes, while paradoxically, mice lacking CFH entirely (Cfh^–/–) show little pathology. These findings implicate partial CFH loss together with dietary lipids in drusen formation [65]. In addition, knocking out other complement components, such as C3 or factor B, alters choroidal neovascular responses after laser injury, further supporting a key role for complement in AMD pathogenesis [67,68].

Disruption of cholesterol-handling genes further sensitizes the retina to dietary challenge. For example, ApoE–/– mice fed a high-fat diet show marked cholesterol accumulation in Bruch’s membrane and the neural retina, accompanied by inflammation and retinal vascular alterations that exceed the effects of either the diet or the genotype alone [194]. In that study, high-fat feeding or ApoE deletion by itself produced only modest functional changes, whereas the combination induced retinal hypoxia, activated NF-κB and TNFα signaling and upregulated VEGFR2, and ultimately drove retinal ganglion cell apoptosis and peripheral neovascularization [192]. Mice carrying the human ApoE4 allele also develop age-dependent AMD-like features, including drusen and a thickened Bruch’s membrane. Likewise, transgenic mice expressing human ApoB100 on an LDLR-deficient background exhibit ERG deficits and photoreceptor thinning, reflecting retinal lipid dysregulation [195]. Together, these models show that genetic dyslipidemia combined with a fat-rich diet produces more robust, AMD-like phenotypes.

Complementary NAM-based platforms, including retinal pigment epithelium cultures, retinal organoids, and retina-on-chip systems, may improve modeling of oxidative stress, complement activation, and retinal aging processes under conditions more closely resembling human biology [62,63,184,196,197]. Nevertheless, animal models remain indispensable for studying systemic inflammatory and metabolic interactions occurring in vivo [1,71].

Dual chemokine deficiency also produces AMD-like features. Mice lacking both Ccl2 and Cx3cr1 spontaneously develop retinal pigment epithelium atrophy, photoreceptor loss, and drusen-like subretinal deposits within a few months—much earlier and more severe than seen in single-knockout animals—suggesting that impaired microglial and macrophage trafficking promotes debris accumulation [31]. Although these double-knockout mice do not require an external stressor beyond normal aging, they effectively model the interaction between defective chemokine signaling and age-related processes in driving retinal degeneration.

5.2. Diabetic Retinopathy

The Ins2Akita mouse (heterozygous Ins2 mutation) develops persistent hyperglycemia from weaning and models type 1 diabetes, but on its own shows only modest diabetic retinopathy (DR) features—mild vascular leakage and limited neuronal loss by 6–9 months of age [198]. The Kimba (trVEGF029) mouse, which overexpresses human VEGF-A specifically in photoreceptors, is non-diabetic yet exhibits profound retinal neovascularization and edema driven by VEGF excess. When these strains are combined to produce the Akimba mouse (Ins2^Akita × trVEGF029), the result is a severe, DR-like phenotype: by young adulthood Akimba mice show widespread retinal neovascular tufts, hemorrhages and exudates on fundus examination and fluorescein angiography, together with robust vascular leakage and worsened neuronal loss compared with either parental line—an effect that models the interaction of hyperglycemia and VEGF excess in multifactorial DR [137,186]. Wada et al. (2021) reveal in their study that retinal hemorrhages and extensive neovascularization in Akimba mice are absent from wild-type and Akita-only controls, illustrating the gene–gene (hyperglycemia × VEGF) interaction that underlies the hybrid phenotype [194]. The Akimba model therefore provides a useful in vivo system for studying mechanisms and therapies for advanced, proliferative DR and VEGF-driven edema in a hyperglycemic background.

5.3. Glaucoma and Other Retinal Degenerations

Glaucoma entails progressive retinal ganglion cell (RGC) and optic nerve damage, often linked to elevated intraocular pressure (IOP). The DBA/2J mouse carries mutations in the Gpnmb and Tyrp1 genes that cause iris pigment dispersion and atrophy, leading to synechiae formation and progressive ocular hypertension beginning around 8–10 months of age; consequently this strain is widely used as a model of inherited pigmentary glaucoma [199,200]. Longitudinal studies that followed DBA/2J mice from 2 to 15 months report that intraocular pressure rises first and is followed by retinal ganglion cell (RGC) dysfunction and degeneration beginning at approximately 10 months, temporally correlated with the IOP increase [193]. Combining the DBA/2J background with additional stressors—such as genetic reduction in antioxidant defenses or an experimental optic nerve insult—accelerates RGC loss and shortens the time to neurodegeneration, demonstrating that secondary stressors can interact with the strain’s intrinsic susceptibility to speed glaucoma progression [201].

Less common retinal dystrophies can also be modeled using hybrid gene–environment approaches. For example, rodent models of retinitis pigmentosa (such as rhodopsinP23H or rd10 mice) show markedly accelerated photoreceptor degeneration when exposed to intense light or other oxidative challenges [202,203]. Likewise, experimental models of Leber hereditary optic neuropathy (LHON)—which carry pathogenic mitochondrial mutations—demonstrate worsened retinal ganglion cell (RGC) loss after hypoxic episodes or optic nerve injury, indicating that environmental stressors unmask or amplify mitochondrial vulnerability [204]. Although these disorders are primarily monogenic rather than complex diseases like AMD or DR, they illustrate the general principle that external insults can reveal subclinical pathology and accelerate neurodegeneration in genetically susceptible retinas.

6. Animal Models and NAMs in Translational Retinal Research

In April 2026, the FDA published the report Reducing Animal Testing in Nonclinical Studies, emphasizing the need to reduce animal testing in preclinical research. This report forms part of the FDA “Roadmap”, which aims to make animal studies the exception rather than the standard by promoting the use of integrated human biology-based methods, referred to as New Approach Methodologies (NAMs). In practice, the FDA encourages the implementation of advanced cellular systems and organ-on-chip technologies, sophisticated in silico models, and three-dimensional cultures such as organoids, which enable the generation of data more directly relevant to human biology, improve prediction of toxic and pharmacokinetic effects, and substantially reduce the need for laboratory animal use.

In retinal research, these developments have contributed to the rapid expansion of alternative in vitro platforms. Recent reviews have highlighted the growing use of both two-dimensional and three-dimensional retinal models capable of reproducing structural and functional features of the human retina while reducing dependence on animal experimentation [205]. In particular, retinal organoids derived from human stem cells can recapitulate retinal lamination and cellular differentiation patterns and respond to pathological stimuli in a manner more consistent with human tissue, making them promising tools for preclinical investigations [62]. Similarly, retina-on-chip systems and advanced 3D retinal cultures provide scalable, reproducible, and cost-effective platforms for drug screening and mechanistic studies, enabling the assessment of larger numbers of compounds and experimental conditions with reduced experimental burden compared with traditional in vivo studies [63,184].

Nevertheless, animal models remain an important component of translational retinal research. Commonly used retinal models include zebrafish, mice, rats, pigs, dogs, and selected non-human primates, all of which have contributed substantially to understanding retinal disorders such as diabetic retinopathy, inherited retinal degeneration, and age-related macular degeneration [71,88,172]. However, the translational limitations of many traditional animal models are increasingly recognized. Both the FDA report and contemporary preclinical literature emphasize that rodent models do not fully reproduce the anatomical and physiological complexity of the human retina, which was discussed in the previous part of this article. Likewise, several large-animal models are used primarily because of practical considerations such as availability, lower costs, or ease of breeding, rather than because they fully replicate human retinal pathology.

Accordingly, in line with the current FDA recommendations and the broader transition toward human-relevant methodologies, the present review focuses primarily on retinal models with translational and practical relevance for contemporary preclinical ophthalmic research. Emphasis was placed on models that have demonstrated utility in mechanistic studies, therapeutic validation, and regulatory acceptance, particularly in the context of gene therapy, retinal degeneration, and vascular retinal diseases. For this reason, several historically utilized but presently less relevant systems, including selected rabbit and hamster models, were not discussed in detail.

Although models such as bilateral macular dystrophy in rhesus macaques, idiopathic bilateral optic atrophy, sporadic RP-like phenotypes in cynomolgus macaques, the feline RDH5 model, and the AIPL1 Persian cat model are less widely used than established rodent or canine systems, they were included because they provide unique insights into specific aspects of retinal pathology that are not fully replicated in more commonly utilized models. In particular, several of these models exhibit phenotypic features involving the central retina, cone-dominant regions, or slowly progressive degeneration patterns that may partially resemble selected features of human retinal diseases. Furthermore, rare spontaneous models can contribute to the investigation of disease heterogeneity, genotype–phenotype relationships, and mechanisms of retinal degeneration that are difficult to reproduce experimentally. Although their translational applicability is more limited due to restricted availability, reduced standardization, and lower scalability, these models remain valuable for mechanistic studies and may support the development of targeted therapeutic strategies in selected retinal disorders.

7. Development of Retinal NAMs Platforms

New Approach Methodologies (NAMs) in retinal research include a broad spectrum of human-relevant experimental platforms, such as retinal organoids, iPSC-derived retinal cultures, retina-on-chip systems, advanced co-cultures, and computational approaches. Among these, retinal organoids currently represent one of the most advanced in vitro models of the human retina. They provide a three-dimensional system derived from human embryonic stem cells or induced pluripotent stem cells and enable the investigation of retinal development, disease mechanisms, therapeutic responses, and toxicity under conditions more closely resembling human retinal biology than conventional two-dimensional cultures.

Despite significant advances in molecular biology, gene therapies, and regenerative medicine, the development of effective treatments for retinal diseases remains a major challenge in modern ophthalmology. One of the main limitations in the development of new therapies is the imperfection of preclinical models used to investigate disease pathogenesis and evaluate the safety and efficacy of potential therapeutic strategies [61,185,206]. Traditional two-dimensional (2D) cell cultures enable the analysis of selected molecular mechanisms; however, they fail to reproduce the spatial organization of the retina and the complex intercellular interactions occurring within neural retinal tissue [61]. Animal models, in turn, have constituted the foundation of ophthalmic research for decades due to their ability to assess processes occurring within the entire organism, including immune responses, metabolic interactions, and the influence of vascularization on retinal function [1]. Despite their high translational value, significant interspecies differences between the human retina and laboratory animals limit the direct translation of preclinical findings into clinical settings [1,61]. These differences particularly concern macular organization, photoreceptor distribution, and the progression of degenerative processes characteristic of human retinal diseases [185,206].

In response to the limitations of conventional experimental models, retinal organoid technology has emerged and is currently considered the most advanced in vitro model of the human retina [61,207,208]. Retinal organoids are three-dimensional structures derived from human embryonic stem cells (hESCs) or induced pluripotent stem cells (hiPSCs), capable of self-organization and recapitulating sequential stages of retinogenesis [62,197,209]. Unlike conventional 2D cultures, retinal organoids enable the formation of a layered architecture resembling the developing human retina while preserving more physiological interactions among neuronal cells, photoreceptors, and glial cells [197,210]. Consequently, organoids provide a more accurate platform for modeling developmental and degenerative processes characteristic of human retinal biology [206,211].

A particular advantage of retinal organoids over animal models is the possibility of using cells derived directly from patients carrying specific genetic mutations [212,213]. This approach enables the generation of models corresponding to the individual genotype and phenotype of the disease, which is especially relevant for inherited retinal disorders and the development of personalized medicine [61,214]. Organoids derived from patients with mutations associated with retinal degenerative diseases exhibit impaired photoreceptor maturation, altered gene expression, and increased susceptibility to degeneration, enabling the investigation of disease mechanisms under conditions more closely resembling human biology than animal models [146,148]. Furthermore, the application of single-cell RNA sequencing technologies enables the identification of heterogeneous cellular populations and the tracking of retinogenesis at single-cell resolution [210,215]. Such analyses facilitate the detection of subtle molecular alterations and differences in cellular maturation between healthy and diseased organoids [215].

Retinal organoids are currently widely applied in studies on gene therapies, drug toxicity assessment, and regenerative medicine [208,211,212]. In contrast to animal models, they enable the evaluation of therapeutic responses in human cells without the influence of species-specific differences, potentially increasing the predictive value of preclinical studies [61]. Nevertheless, animal models remain indispensable in translational research because they enable the assessment of long-term treatment safety, drug biodistribution, and complex interactions occurring between the retina and other organ systems [1,185]. Therefore, retinal organoids should not be regarded as a complete replacement for animal models but rather as complementary tools enabling more precise modeling of processes specific to human biology [1,61]. The combined application of both approaches may substantially improve translational research and enhance the transferability of preclinical findings into clinical practice [207].

Despite their numerous advantages, retinal organoids still exhibit significant biological and technological limitations. One of the most important challenges is the lack of vascularization, which restricts oxygen and nutrient diffusion into deeper layers of the organoid [196,206]. Under physiological conditions, retinal development and function are highly dependent on interactions between neuronal cells, vascular endothelial cells, and pericytes responsible for maintaining the blood–retina barrier [194]. Conventional retinal organoids fail to reproduce the full vascular architecture, which contributes to their limited functional maturation and hampers long-term culture maintenance [211]. In response to these limitations, partially vascularized retinal organoids and organoid-on-a-chip systems are currently being developed to achieve a more physiologically relevant tissue microenvironment [196,216,217]. These technologies may improve cell survival, enhance photoreceptor maturation, and increase the translational value of organoid-based models in the future [218].

Another major limitation is the absence of a complete immune microenvironment [219,220]. Most standard retinal organoids do not contain microglia or other immune cells involved in neuroinflammatory processes [219]. This limits their applicability in modeling diseases in which immune responses and chronic inflammation play a central pathogenic role [185,220]. Recent efforts have focused on the development of more complex co-culture systems integrating microglia into retinal organoids, which may enable a more accurate reconstruction of the human retinal immune environment [219].

Retinal organoid immaturity also remains a major challenge [62,185,197]. Most currently available models correspond to embryonic or early postnatal developmental stages, limiting their applicability in studies of age-related retinal diseases such as age-related macular degeneration [221]. Although organoids contain photoreceptors and exhibit partial functional activity, they do not achieve the full maturation characteristic of the adult human retina [222]. Additional challenges include high variability among pluripotent stem cell lines and limited reproducibility between culture batches and research laboratories [208,211,215]. The lack of fully standardized differentiation protocols complicates cross-study comparisons and affects the translational reliability of organoid-based models [211]. Furthermore, organoids do not recapitulate systemic interactions present in the human body, including the influence of circulation, endocrine signaling, multi-organ interactions, and long-range neuronal connectivity [223]. Consequently, animal models remain essential for evaluating systemic treatment safety and whole-organism responses [1].

Despite these limitations, retinal organoids represent a highly promising research platform in ophthalmology and regenerative medicine [207,224]. Further integration of organoid technologies with advanced bioengineering approaches, imaging techniques, and computational analysis is expected to play a crucial role in future developments [208,218]. One of the key future directions involves improving the physiological relevance of retinal organoids through the development of partially vascularized models, organoid-on-a-chip systems, and organoids integrated with immune cells [196,216,217,219,220]. These approaches may enhance cell survival, improve photoreceptor maturation, and enable more reliable modeling of neuroinflammatory and neurovascular interactions [196,218].

Artificial intelligence (AI)-based tools are also expected to play an increasingly important role in the automated analysis of organoid imaging data and in the standardization of organoid maturation and functionality assessment [225,226]. AI-driven algorithms may facilitate the identification of subtle morphological and degenerative changes undetectable by conventional microscopy, while also supporting high-throughput drug screening and prediction of therapeutic responses [225]. Such technologies may partially reduce variability between culture batches and improve reproducibility across different research centers [208,211].

Advanced systems for monitoring neuronal activity and functional photoreceptor responses in real time are also under development [222,227,228]. The integration of retinal organoids with optoelectronic platforms may enable more precise evaluation of functional maturation and light signal transmission under in vitro conditions [222,227]. In addition, multispectral imaging approaches may allow non-invasive assessment of cellular metabolism, viability, and spatial organization during long-term organoid culture [229,230]. These technologies may further increase the translational value of retinal organoids and improve their applicability in preclinical studies and personalized medicine [208,218].

Increasingly, it is emphasized that the future of retinal disease research will rely on the integration of different experimental models rather than the replacement of one model by another [1,61,185]. Retinal organoids enable detailed investigation of processes specific to human biology and facilitate personalized studies, whereas animal models remain indispensable for evaluating complex systemic interactions that cannot yet be fully reproduced in vitro [1,185]. Therefore, retinal organoid models and animal models should currently be regarded as complementary research platforms whose combined use may significantly improve drug discovery processes and enhance the clinical translation of preclinical findings [61,207].

8. Discussion

Appropriate animal models provide essential platforms for understanding the etiology and pathogenesis of retinal diseases and for evaluating emerging therapeutic strategies [231,232]. Nevertheless, no currently available model fully reproduces the complexity and heterogeneity of human retinal disorders, and increasing ethical and societal pressure continues to encourage reduction and refinement of animal experimentation [232]. The translational value of retinal models therefore depends not only on their ability to reproduce structural degeneration, but also on how accurately they mimic disease-associated molecular pathways, including inflammatory signaling, oxidative stress, metabolic dysfunction, and proteostasis imbalance observed in human retinal pathology [212,231].

Animal models have been particularly valuable for identifying diagnostic biomarkers, monitoring disease progression, and evaluating therapeutic response under controlled experimental conditions [233]. Unlike clinical studies, they enable precise determination of disease onset, longitudinal assessment of retinal degeneration, and analysis of early pathogenic events preceding overt vision loss [233]. Importantly, many retinal degeneration models demonstrate convergence of distinct genetic or environmental insults toward shared downstream mechanisms. Experimental studies in rodents have shown that chronic oxidative stress and mitochondrial dysfunction contribute substantially to progressive photoreceptor degeneration, secondary cone death, and activation of retinal microglia [212,231]. Similarly, inflammatory signaling pathways involving Müller glia activation, microglial recruitment, cytokine release, and complement cascade dysregulation are increasingly recognized as central contributors to retinal neurodegeneration across both inherited and acquired retinal diseases [231]. In several retinal degeneration models, activated Müller glia and retinal microglia exhibit increased expression of pro-inflammatory mediators, including tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), interleukin-6 (IL-6), and monocyte chemoattractant protein-1 (MCP-1/CCL2), which promote photoreceptor dysfunction and neuronal loss [234]. In rd1 and rd10 murine models of retinitis pigmentosa, microglial activation occurs early during degeneration and is associated with migration of microglia into the outer nuclear layer, where they contribute to photoreceptor phagocytosis and oxidative damage [82].

Complement dysregulation has also been implicated in retinal degeneration, particularly through increased activation of C3 and complement factor B pathways, which amplify chronic neuroinflammation and synaptic remodeling within the retina [235].

In diabetic retinopathy models, chronic hyperglycemia further enhances inflammatory signaling via NF-κB activation, vascular endothelial growth factor (VEGF) upregulation, leukostasis, and blood–retinal barrier disruption, linking vascular dysfunction with progressive neurodegeneration [236]. These findings increase the translational relevance of animal models by demonstrating that retinal degeneration is driven not only by primary genetic defects, but also by secondary inflammatory and immune-mediated mechanisms that may represent therapeutic targets independent of the underlying mutation.

Proteostasis imbalance also represents a major pathological mechanism reproduced in several inherited retinal degeneration models. Multiple retinal diseases are associated with the accumulation of misfolded or improperly processed proteins, particularly mutant rhodopsin (RHO), phosphodiesterase subunits (PDE6A/PDE6B), interphotoreceptor retinoid-binding protein (IRBP), fibulin-3 (EFEMP1), and ELOVL4-associated proteins involved in Stargardt-like macular degeneration [47,237,238]. In rhodopsin-associated retinitis pigmentosa, especially in RHO P23H and T17M models, mutant rhodopsin undergoes misfolding and retention within the endoplasmic reticulum ER, resulting in activation of unfolded protein response (UPR) pathways mediated by PERK, IRE1, and ATF6 signaling [47,238]. Persistent accumulation of misfolded proteins overwhelms ER-associated degradation (ERAD) systems, induces oxidative stress and calcium dysregulation, and ultimately activates pro-apoptotic pathways contributing to progressive photoreceptor degeneration [239]. Similarly, PDE6B-mutant rd1 mice exhibit ER stress and intracellular Ca²⁺ imbalance preceding overt retinal degeneration, supporting the predictive validity of these models for studying early proteostasis-related mechanisms in human retinitis pigmentosa. In addition, metabolic stress associated with the exceptionally high energetic demands of photoreceptor cells contributes to retinal vulnerability in both genetic and vascular retinal diseases, particularly under conditions of impaired oxygen supply or mitochondrial dysfunction [231,232]. These convergent pathways increase the translational relevance of animal models by enabling evaluation of mutation-independent neuroprotective therapies targeting oxidative damage, inflammation, or cellular stress responses.

Small animal models, particularly mice, remain the most widely used systems for in vivo retinal research due to their low maintenance costs, short reproductive cycles, and extensive genetic manipulability [212]. Murine models are especially valuable for investigating monogenic inherited retinal diseases, mechanistic disease pathways, and early-stage therapeutic interventions. However, their predictive validity for human retinal disease remains partial rather than comprehensive. Rodents are nocturnal animals lacking a macula and possessing a substantially lower cone-to-rod ratio than humans, limiting their ability to accurately reproduce cone-rich central retinal degeneration and complex visual function deficits characteristic of human retinal disorders [3,212]. In different vertebrate species, the proportions and spatial arrangement of rods and cones vary significantly, further complicating the translation of experimental results to human retinal diseases. Although rods and cones share common phototransduction mechanisms, the expression levels and activity of individual cascade proteins differ significantly between photoreceptor types [240]. Interpretation of experimental results is further complicated by functional interactions between rods and cones, including electrical coupling and complex neuronal signaling in the retina [241]. Translational constraints also apply to retinal metabolism and stress responses, as cones exhibit higher energy requirements and greater susceptibility to mitochondrial dysfunction and oxidative stress than rods [240]. Consequently, rod-dominant animal models may underestimate the metabolic and oxidative mechanisms contributing to cone degeneration in human retinal diseases.

Furthermore, the human inner limiting membrane (ILM) differs in thickness and morphology across retinal regions. In the foveal and optic nerve head regions, the retina is exceptionally thin and smooth, contrasting sharply with the uniform sensory retina of rodents. Unlike the human retina, the rodent sensory retina does not display age-related changes seen in humans [242]. Rodents also have a short lifespan of approximately two years, which may limit the faithful representation of cumulative retinal cell damage and disease progression influenced by lifestyle and environmental factors in patients [221]. Furthermore, although murine models effectively recapitulate selected molecular pathways, including oxidative stress responses and inflammatory activation, they often fail to reproduce the slower progression, phenotypic heterogeneity, and systemic metabolic interactions observed in human disease [212,231]. Consequently, findings derived from rodent studies frequently require validation in larger animal models with retinal anatomy and physiology more closely resembling those of humans.

There are significant limitations in mouse-based research programs. Although there are many genetic similarities between mice and humans, the differences are substantial enough that several well-known models fail to fully replicate human disease phenotypes, including models of Usher syndrome [243] and Stargardt disease [244]. Currently available models of degenerative retinal diseases are useful for testing cell-based therapies and understanding mechanisms underlying vision restoration. However, many mouse and rat models of RP have limited applicability due to the small size of their eyes, which restricts access to the subretinal space and limits the size of cells that can be injected subretinally or intravitreally [1].

Large animal models offer significant advantages over commonly used laboratory rodents. Eye size and globe dimensions are more comparable to humans, and the retina exhibits a high density of cones and more tightly packed photoreceptors, providing high visual acuity [3]. NHPs can bridge the gap between small mammals and humans, although studies are costly, ethically complex [245], and comprehensive genetic tools are currently limited [221].

Animal models are invaluable for studying visual physiology and the pathogenesis of inherited retinal diseases. Among these, NHP models are particularly useful due to their anatomical and physiological similarity to the human retina. The presence of a macula in NHP, along with extensive cortical representation of central vision, enables high visual acuity, a feature shared with humans [172]. No other animal model replicates this effect, highlighting its importance in research.

A key consideration in developing transgenic animal models is the choice between germline and somatic genome editing. Germline editing allows genetic modifications to be transmitted to subsequent generations, whereas somatic editing does not. The broader use of CRISPR/Cas9-based approaches in ophthalmic research has increased interest in animal models compatible with long-term in vivo genome editing. Mohan et al. reported that Rosa26-Cas9 knock-in mice retained normal retinal morphology and visual function during prolonged observation, indicating that these models may be suitable for studies involving sustained retinal gene editing [246]. The use of such systems may contribute to more reproducible preclinical studies and further development of experimental therapies for inherited retinal disorders. However, ethical and technical challenges are particularly significant for germline modifications in NHP. The disadvantages of large species generally include higher research costs, long gestation periods, and typically single offspring per pregnancy [3]. Additionally, these models require specialized infrastructure, long-term maintenance of breeding colonies, and the limited number of animals also makes it difficult to obtain adequate statistical power for the research. Research involving NHP is subject to strict regulatory oversight, potentially slowing progress and increasing costs [247]. There are also concerns regarding iatrogenic damage from procedures used to create models, as well as unintended phenotypic traits that may negatively impact animal welfare [245].

Many naturally occurring IRDs in NHP with pathogenic mechanisms similar to humans likely remain undiscovered. Their identification is challenging because observable behavioral deficits usually appear only after significant retinal degeneration [172]. Conducting comprehensive ophthalmic examinations or genetic testing on all NHP in a facility is costly and often impractical. Additionally, despite high genetic similarity, some naturally occurring NHP mutations may not be pathogenic in humans, limiting the clinical relevance of such models for human IRD research.

Pigs have long been utilized as large animal models in biomedical research, with increasing use due to the ability to induce mutations that phenocopy several human hereditary diseases. Advantages include large litter sizes, the possibility of artificial insemination, and applicability of all current genetic modification techniques [4]. The similarity of pig eyes to human eyes makes them ideal for developing surgical or transplant procedures, as well as subretinal or intravitreal drug delivery [248,249,250,251]. Pigs also have a holangiotic retinal vasculature similar to humans, lack a tapetum (a non-cellular reflective layer present in other large animal eyes, such as dogs and cats), and are more cost-effective than NHP, with research expenses roughly ten times lower [252].

A major limitation of pig models is the limited availability of genetically modified lines with outer retinal degeneration [252]. Additionally, pigs exhibit substantial inter-individual and strain-dependent variability, which complicates standardization of study outcomes [253].

Over the years, dogs have proven to be invaluable large animal models for studying hereditary retinal diseases, as spontaneous IRDs are common in the canine population [3]. Numerous naturally occurring mutations closely mirror those in human patients [254], providing advantages over rodent models by allowing characterization of disease natural history, mutation-specific protein interactions, and evaluation of therapeutic interventions. One reason clinical trials in humans may yield less impressive results compared to canine studies is that therapeutic interventions in animals were often preceded by exten application of advanced technologies, such as adaptive optics and optoretinography, may provide valuable insights for the development of novel interventions [254].

Cats also frequently display hereditary retinal degeneration analogous to human IRDs caused by mutations in homologous genes. Both dogs and cats possess a region of high retinal acuity, the area centralis, which corresponds to the human macula. Combined with eye size similar to humans, these models provide insights unattainable in rodents [12]. Comparable eye size allows similar injection procedures, implants, and surgical interventions as those used in humans. However, unlike humans, these species have a tapetum, a reflective layer within the inner choroid [71].

The limitations of canine and feline models include the reliance on naturally occurring mutations, lack of engineered models for specific genes, high costs and time associated with colony maintenance, slow disease progression, and restricted access to specialized research tools and facilities. Ethical considerations further constrain research using these species, and such research may be difficult or impossible to conduct in some countries [71].

In summary, animal models have led to the identification of disease genes and elucidation of numerous molecular and cellular mechanisms underlying retinal pathology. They also provide a foundation for testing therapeutic approaches, including pharmacological, genetic, and surgical interventions [255].

Although indispensable for studying pathomechanisms and therapies, animal models have significant limitations. Many species lack full human ocular anatomy and physiology, including the macula, and certain disease processes progress slowly. Different experimental models address different research questions. Rodents are most useful for analyzing molecular mechanisms and genetic studies, while large animals better represent the surgical and anatomical conditions relevant to translational therapies. Large animals, such as cats, dogs, and NHP, are more similar to humans anatomically, but their maintenance is costly, access to colonies and research tools is limited, and ethical regulations add complexity. No single model is ideal; selection requires a balance between translational relevance, cost, and availability. The limitations of animal models may stem from interspecies differences and incomplete understanding of disease pathogenesis. Therefore, a logical approach in retinal disease research is to deconstruct the problem into smaller analytical units, such as cell and tissue cultures [231]. Understanding the molecular foundations of the human retina increases the likelihood of developing precisely targeted therapies that could restore or preserve vision for many individuals [185].

9. Conclusions

Animal models have played a crucial role in elucidating the molecular and cellular mechanisms of retinal diseases and in the development of the first effective therapies, including gene therapies. Each group of models, from rodents to dogs, cats, and non-human primates, provides unique insights; however, none fully replicates the human anatomy and pathophysiology of retinal disorders. Key limitations include the absence of a macula in most species, differences in photoreceptor distribution, and distinct disease progression dynamics. Large animal models offer higher translational value due to their anatomical similarity to the human eye and the feasibility of surgical and therapeutic procedures analogous to clinical interventions. Nonetheless, their use is associated with high costs, limited availability, prolonged study duration, and significant ethical constraints. Retinal organoids represent a promising alternative and complement to animal models, enabling the study of human retinogenesis and disease pathogenesis in a three-dimensional system while preserving cell–cell interactions and the patient-specific genetic background. Despite limitations such as lack of vascularization and incomplete maturity, organoids currently constitute the most advanced in vitro model of the human retina.

The greatest research and clinical potential lies in a complementary approach that integrates animal models with modern in vitro systems, including organoids. Such multi-level strategies enhance the likelihood of a deeper understanding of retinal disease pathogenesis and the development of effective, personalized therapies.