GLP-1 receptor drugs and their current and possible future roles in eye diseases

The structural diversity of these agents underpins their distinct pharmacological properties. Both exenatide and lixisenatide, derivatives of exendin-4, feature a modification at the second N-terminal amino acid to confer resistance to DPP-4-mediated degradation [20]. Liraglutide, an acylated human GLP-1RA produced via recombinant DNA technology, exhibits 97% sequence homology with endogenous human GLP-1(7–37). Following subcutaneous administration, liraglutide reaches peak plasma concentrations within 8–12 hours and has an elimination half-life of approximately 13 hours [23]. Semaglutide shares 94% sequence homology with human GLP-1, and its prolonged action is attributed to strong albumin binding and resistance to DPP-4 degradation [24]. Dulaglutide is a fusion protein consisting of two identical disulfide-linked chains, each composed of a GLP-1 analog (90% homology with native GLP-1) covalently linked to a modified human IgG4 Fc domain [25]. Tirzepatide represents a new class of single-molecule dual agonists for both GIP and GLP-1 Rs, developed from pioneering research on multi-receptor peptides [26]. It is a 39-amino acid linear peptide structurally modified for an extended plasma half-life [27]. While tirzepatide has a comparable binding affinity for the GIP receptor, its affinity for the GLP-1 R is approximately fivefold weaker than that of native GLP-1. Notably, it acts as a biased agonist at the GLP-1 R, preferentially stimulating cyclic adenosine monophosphate (cAMP) production [28].

While the metabolic efficacy of GLP-1RAs is extensively validated, their potential in neuro-ophthalmology is contingent upon their ability to penetrate the blood-brain barrier (BBB) and the BRB. The neuroprotective potential of GLP-1 has been established in models of neurodegenerative disorders, including Parkinson's disease and Alzheimer's disease (AD), with evidence suggesting it can modify disease progression and ameliorate pathology [29 –32]. This therapeutic effect is predicated on the ability of GLP-1RAs to cross the BBB and exert direct physiological effects within the brain [21, 22]. Although much of this evidence derives from animal models, making direct extrapolation to humans complex, significant functional commonalities exist between the murine and human BBB, including glucose transport, the presence of circumventricular organs (CVOs), and saturable transport systems [33]. Whereas central nervous system (CNS) effects are determined by BBB penetration, ocular efficacy is governed by the traversal of the BRB. Analogous to the BBB, the BRB has an inner component (iBRB) of retinal endothelial cells and an outer component (oBRB) formed by tight junctions between retinal pigment epithelial (RPE) cells [34]. These barriers can exhibit differential permeability to pharmaceutical compounds. Therefore, a pivotal question is whether the demonstrated BBB permeability of certain GLP-1RAs translates to effective penetration of the BRB to achieve therapeutic concentrations in retinal tissues.

While the pharmacokinetic properties of GLP-1RAs determine their systemic availability, their therapeutic potential in ocular diseases is fundamentally contingent upon the presence and localization of GLP-1 R in target tissues. The GLP-1 R is a member of the class B family of G protein-coupled receptors (GPCRs) [35], and investigations have confirmed its expression within the human retina [36]. The localization of this receptor is highly specific; current evidence indicates that GLP-1 R expression is found predominantly within the ganglion cell layer (GCL), exclusively detected in a discrete subpopulation of neurons [37]. Morphologically, these GLP-1 R-positive cells display characteristic neuronal phenotypes, such as spherical nuclei and horizontal projections.

This specific localization within the GCL is of profound clinical relevance, as retinal ganglion cells are the primary neurons that degenerate in glaucoma and are also critically affected during the early neurodegenerative stages of diabetic retinopathy [38, 39]. However, it is equally notable that GLP-1 R expression is reportedly undetectable in retinal and choroidal vasculature, the RPE, or other ocular structures [37]. This finding suggests that the primary mechanism of GLP-1RAs in the retina is likely direct neuroprotection. It also implies that any beneficial effects on vascular or RPE pathology may be mediated indirectly, either through neuronal signaling pathways within the retina or via systemic anti-inflammatory and metabolic improvements.

The established anti-inflammatory and neuroprotective effects of GLP-1RAs within the central nervous system suggest their therapeutic potential across a spectrum of neurodegenerative disorders [40, 41]. In preclinical models of glaucoma and DR, GLP-1RAs have demonstrated robust neuroprotective effects in ocular tissues, primarily by modulating oxidative stress, inflammation, mitochondrial function, and synaptic plasticity [42, 43].

In preclinical models of glaucoma, neuroprotection appears to be strongly linked to the anti-inflammatory properties of GLP-1RAs. For instance, NLY01, a long-acting PEGylated GLP-1RA with an extended half-life [41], effectively mitigates retinal ganglion cell (RGC) apoptosis induced by elevated intraocular pressure, a process mediated by a well-defined inflammatory cascade [44]. Ocular hypertension triggers the production of pro-inflammatory cytokines by immune cells, which in turn drives the transformation of neuroprotective astrocytes into a neurotoxic A1-reactive phenotype, ultimately leading to RGC degeneration [45]. NLY01 exerts its neuroprotective effect by potently suppressing this pathogenic inflammatory axis, thereby attenuating the formation of cytotoxic A1 astrocytes and preserving RGC viability [42]. Corroborating these findings, other GLP-1RAs have similarly shown neuroprotective properties in murine models of ocular hypertensive glaucoma, manifesting as enhanced RGC survival, reduced microglial/macrophage activation, and amelioration of optic nerve axon degeneration [46].

In the context of diabetic retinopathy, Müller cells, which constitute approximately 90% of retinal glia and are critical for neuronal support [47], emerge as a key therapeutic target. Under diabetic conditions, these cells undergo reactive gliosis, contributing to the inflammatory environment [48]. Exendin-4 (E4), a long-acting GLP-1RA, has demonstrated neuroprotective effects in early experimental DR [49, 50]. In the Goto-Kakizaki (GK) rat, a spontaneous model of T2DM [51], intravitreal E4 prevented cell loss in the GCL and outer nuclear layer (ONL) without affecting systemic glucose levels, suggesting a local retinal mechanism. At the cellular level, E4 directly targets Müller cells to attenuate reactive gliosis, preserving the critical Bcl-2/Bax and Bcl-xL/Bax pro-survival ratios and consequently mitigating mitochondrial-mediated apoptosis via reduced caspase-3 activation [52].

Further studies in DR models have elucidated additional protective pathways. Lixisenatide inhibits neuroinflammation by promoting the proteasomal degradation of thioredoxin-interacting protein (TXNIP) in ocular tissues via the PI3K/Akt signaling pathway, an effect more potent than that of insulin. This intervention significantly attenuates retinal cell apoptosis and ameliorates pathological thickening of the inner nuclear layer (INL) and nerve fiber layer (NFL) [53]. Similarly, in a diabetic rat model induced by a high-fat diet and streptozotocin (STZ) injection [54, 55], liraglutide was shown to mitigate the hyperglycemia-induced decline in RGC-5 cell viability and improve the structural integrity and organization of the GCL [56].

Mechanistically, these neuroprotective effects are underpinned by the activation of canonical GLP-1 R signaling pathways within the retina. In the retinas of diabetic mice treated with GLP-1RAs, a significant elevation in both cAMP and pAKT levels was observed, indicating successful target engagement and triggering of downstream pro-survival signals [36]. Furthermore, GLP-1RAs can counteract the diabetes-induced downregulation of the glutamate-aspartate transporter (GLAST), a key glial transporter responsible for clearing extracellular glutamate [57]. By restoring GLAST expression, GLP-1RAs may protect retinal neurons from excitotoxicity and subsequent apoptosis [36]. Collectively, these preclinical studies illustrate that GLP-1RAs protect retinal neurons through a multi-pronged approach involving anti-inflammatory, anti-apoptotic, and anti-excitotoxic mechanisms.

Emerging evidence suggests that GLP-1RAs hold considerable therapeutic promise in ophthalmic pathologies, particularly through their vasoprotective and anti-inflammatory mechanisms. This pharmacological effect is highly relevant for addressing the fundamental pathogenic processes of DR. While long understood as a microvascular disease, the central role of neuroinflammation and immune dysregulation is now increasingly clear. Recent multi-omics analyses by Cappellani et al., for instance, have provided strong genetic and proteomic evidence linking DR pathogenesis directly to the activation of specific immune pathways, including the complement system and adaptive T-cell responses [53, 58, 59]. This refined understanding of DR as an immunoinflammatory disease provides a strong rationale for exploring immunomodulatory agents like GLP-1RAs. GLP-1RAs appear to exert dual therapeutic benefits by concurrently ameliorating neuronal and vascular pathologies through stabilization of the BRB and suppression of inflammation-driven angiogenesis.

The vasomodulatory effects of GLP-1RAs also extend to glaucoma models, although the observed outcomes differ. Systemic administration of semaglutide conferred discernible vasoprotective and anti-neuroinflammatory benefits in a rodent model of glaucoma, primarily manifesting as attenuated intraocular pressure elevation and suppression of pathological astrocyte activation and vascular remodeling. Intriguingly, however, semaglutide treatment did not directly improve retinal ganglion cell survival in this context. Instead, the pharmacological intervention significantly enhanced the structural complexity (fractal dimension and lacunarity) of astrocytic networks and promoted a more organized retinal vasculature. These morphological improvements suggest a potential mechanism for mitigating vascular dropout and capillary degeneration in glaucomatous conditions, even if direct, short-term neuroprotection was not observed [60].

The vasoprotective effects are most evident in models of diabetic retinopathy, where a hallmark of the disease is the breakdown of the BRB and subsequent retinal vascular leakage [61]. In GK rats, E4 treatment significantly suppressed Evans blue leakage from retinal vessels. This was achieved by activating retinal GLP-1 R and inhibiting the overexpression of placental growth factor (PLGF) and intercellular adhesion molecule-1 (ICAM-1), thereby stabilizing tight junction proteins and preserving BRB integrity. Mechanistically, E4's primary effect was to alleviate inflammatory responses and vascular abnormalities, with relatively modest inhibitory effects on VEGF [62]. Consistent with these findings, locally administered GLP-1RAs in db/db mice also prevented diabetes-induced albumin leakage and suppressed the overexpression of VEGF and IL-1β, thus preserving BRB integrity [36]. These studizes collectively suggest that GLP-1RA treatment not only mitigates neurodegeneration but also ameliorates early-stage microvascular damage.

While the vasomodulatory and anti-inflammatory effects of GLP-1RAs target downstream pathological events, their fundamental therapeutic potential may lie in the direct modulation of metabolic dysregulation. Ocular tissues exhibit exceptionally high energy demands: photoreceptors require substantial ATP for phototransduction; the RPE engages in active transport to maintain the BRB; and high-frequency synaptic transmission imposes further metabolic burdens [63]. This high metabolic demand renders ocular tissues particularly vulnerable to metabolic dysregulation, which can precipitate an energy crisis leading to retinopathy, AMD, and other pathologies. This metabolic vulnerability, however, also presents a prime therapeutic target for metabolic modulators such as GLP-1RAs.

In AMD, this metabolic link is evident. Drusen, the pathological hallmark of AMD, are lipid-rich extracellular deposits from the RPE that impair cellular function and serve as a nidus for chronic inflammation [64]. Elevated free fatty acids (FFAs) can drive the intracellular accumulation of lipid droplets (LDs) within RPE cells, inducing oxidative stress and epithelial-mesenchymal transition (EMT)—processes strongly associated with AMD progression [65, 66]. Liraglutide has been shown to ameliorate this lipid droplet accumulation in ARPE-19 cells by activating the AMP-activated protein kinase (AMPK) pathway, which in turn attenuates EMT, enhances lipophagy, and mitigates oxidative stress [67].

Beyond lipid metabolism, GLP-1RAs exert profound effects on mitochondrial homeostasis in retinal neurons. Research has confirmed that RGC-5 cells express GLP-1 R and are susceptible to high-glucose-induced mitochondrial damage [43]. Under hyperglycemic conditions, RGCs exhibit significant mitochondrial impairment, including cristae dissolution, swelling, vacuolar degeneration, elevated reactive oxygen species (ROS) production, and diminished mitochondrial membrane potential (MMP) [68, 69]. GLP-1RAs appear to counteract this damage at multiple levels. Exenatide treatment ameliorates these morphological alterations and inhibits the mitochondrial apoptotic pathway by suppressing cytochrome c release and favorably modulating the expression of pro-apoptotic (Bax, caspase-3) and anti-apoptotic (Bcl-2) proteins, thereby protecting RGC-5 cells from hyperglycemic injury [43].

In addition to regulating apoptosis, GLP-1RAs also fine-tune mitochondrial quality control via mitophagy. Mitochondria are highly dynamic organelles pivotal for ATP production, metabolism, and ROS signaling [70]. Liraglutide administration significantly mitigates excessive high glucose-induced mitophagy, as evidenced by reduced autophagosome formation and a normalization of LC3A/B and p62 expression. In energy-demanding RGCs, while basal mitophagy is physiologically essential, excessive activation can lead to energy depletion and cellular demise. By restoring this crucial balance, liraglutide helps preserve mitochondrial integrity and neuronal viability [56].

Beyond their established metabolic functions, GLP-1RAs exert direct immunomodulatory effects within the central nervous system. A key mechanism involves their interaction with glial cells, as recent studies show that activated microglia and astrocytes can induce the expression of GLP-1 R, making them direct targets for GLP-1 mimetics [41]. For instance, GLP-1 treatment attenuates the endotoxin-induced release of the potent proinflammatory cytokine IL-1β [71], which is known to inhibit neuronal transmission and promote apoptosis [72]. In preclinical models of AD, this translates to significant therapeutic effects. Liraglutide treatment reduced the population of activated microglia by 50% in APP/PS1 mice [73]. Mechanistically, GLP-1RAs suppress neuroinflammation by shifting microglia from a pro-inflammatory to a phagocytic, anti-inflammatory phenotype. They reduce the release of cytokines like IL-1β, IL-6, and TNF, upregulate the anti-inflammatory cytokine TGF-β, and downregulate the expression of disease-associated microglial (DAM) genes, ultimately enhancing Aβ clearance [74]. This demonstrates a profound capacity to quell neuroinflammation.

This potent anti-inflammatory capacity also translates to pathologies of the ocular surface, such as diabetic dry eye disease. Immunohistochemical analysis has revealed GLP-1 R expression in the acinar and ductal epithelium of the lacrimal gland, although this expression is significantly downregulated in diabetic models. Topical administration of a liraglutide-containing ophthalmic solution reversed this pathology, markedly increasing tear fluid secretion and attenuating diabetes-induced inflammation in the lacrimal gland [75]. Similarly, SM102, a novel long-acting GLP-1RA structurally related to semaglutide, demonstrated robust efficacy in the db/db mouse model. SM102 treatment enhanced tear secretion, prolonged tear film breakup time, reversed corneal tissue damage, and increased conjunctival goblet cell density. These benefits were associated with a rebalancing of the cytokine milieu—reducing pro-inflammatory IL-6, IL-1β, and TNF-α while elevating anti-inflammatory TGF-β1—mediated by the suppression of the MAPK signaling pathway [76]. These findings highlight the therapeutic versatility of GLP-1RAs, addressing inflammatory conditions from the CNS to the anterior segment of the eye.

The significance of these immunomodulatory effects is further underscored by recent advances in understanding DR pathogenesis. As highlighted by Cappellani et al., the molecular landscape of DR is deeply intertwined with the dysregulation of the adaptive immune system and the complement cascade [59]. The established ability of GLP-1RAs to suppress glial cell activation and modulate cytokine release—key drivers of these very pathways—positions them as a uniquely targeted therapeutic strategy. This offers a compelling mechanistic rationale for their use in DR, suggesting they do not merely offer a generic anti-inflammatory benefit, but may directly intervene in the specific immunological events now known to perpetuate retinal damage.

As a leading cause of vision loss in working-age adults, DR presents a major therapeutic challenge [61, 77]. Given the established neuroinflammatory and vasomodulatory mechanisms of GLP-1RAs, their potential to modify DR progression has become a topic of intense clinical investigation and debate. Indeed, early retrospective cohort studies and post-hoc analyses suggested that GLP-1RAs might delay DR progression or improve outcomes, with one study reporting that 80% of patients on continuous therapy achieved either improvement or stabilization of their DR status [78, 79]. However, the clinical picture is far from straightforward. A recent retrospective cohort analysis revealed that, compared to the combination of SGLT-2 inhibitors with insulin, the concomitant use of GLP-1RAs and insulin was associated with a significantly elevated risk of DR and diabetic macular edema [80].

This complexity is further echoed in analyses that differentiate between disease stages. While GLP-1RAs have been linked to a statistically significant elevation in the risk of early-stage DR events compared to placebo, they may conversely demonstrate a protective effect against the progression to advanced-stage diabetic retinopathy when compared to insulin therapy [81]. The concern regarding an early increase in risk was most prominently highlighted by the SUSTAIN-6 trial, which found that patients undergoing semaglutide treatment exhibited an elevated risk of diabetic retinopathy complications [82]. This apparent paradox—an early risk coupled with a potential long-term benefit—demanded a mechanistic explanation.

The leading hypothesis to reconcile these findings is the "early worsening" phenomenon, where rapid and substantial glycemic improvement temporarily exacerbates pre-existing DR. This has been a long-recognized event in patients with type 1 diabetes [83, 84]. To probe whether semaglutide exerts a direct adverse vascular effect, a separate randomized controlled trial specifically investigated its impact on retinal physiology. The trial failed to demonstrate any statistically significant impact of semaglutide on retinal oxygenation or vascular diameter, arguing against a direct vasotoxic mechanism. However, semaglutide administration was associated with a marginal increase in central retinal thickness; importantly, this effect was attenuated after adjusting for glycated hemoglobin (HbA1c) levels, suggesting a strong correlation with the degree of glycemic improvement [85]. This concept is powerfully substantiated by a comprehensive meta-analysis encompassing seven major CVOTs, which concluded that the elevated risk of DR progression associated with semaglutide was driven by the magnitude and rapidity of HbA1c reduction, rather than a direct drug effect [86].

The underlying pathophysiology of this early worsening is believed to involve hemodynamic shifts in a retinal microvasculature that has lost its capacity for autoregulation. The retinal capillary damage central to DR leads to tissue hypoxia, altered venous oxygen saturation, and a compromised ability to regulate blood flow [87 –89]. These changes are not uniform, with the macular area typically exhibiting dilation while the periphery suffers from ischemic hypoxia [90, 91]. A rapid metabolic correction can overwhelm this fragile, dysregulated system, potentially worsening vascular leakage and macular edema.

Despite the coherence of the early worsening hypothesis, the narrative is further complicated by a large body of real-world evidence that does not reflect the risk signals seen in certain trials. Multiple large-scale analyses of observational data, including those comparing GLP-1RAs to SGLT-2 inhibitors and other oral agents, have found no statistically significant association between GLP-1RA use and the exacerbation of DR or the development of sight-threatening retinopathy [92 –95]. This discrepancy between controlled trials and real-world evidence highlights the complexities of translating trial findings to diverse patient populations.

In conclusion, the role of GLP-1RAs in the management of DR remains a nuanced and evolving topic. The potent glycemic-lowering capacity of some agents can induce a temporary worsening of DR, a phenomenon driven by rapid metabolic correction rather than direct toxicity. Yet, this risk must be balanced against a largely neutral safety profile in broad observational studies and a potential for long-term benefit. Definitive answers are eagerly awaited from ongoing trials like the FOCUS study (NCT03811561), which is specifically designed to clarify the long-term impact of semaglutide on diabetic eye disease [96].

Glaucoma is a leading cause of irreversible blindness characterized by progressive RGC degeneration [97]. Mounting evidence implicates chronic neuroinflammation and glial cell dysregulation as central drivers of this neurodegenerative process [98 –100].

Given this neuroinflammatory underpinning, the established immunomodulatory and neuroprotective properties of GLP-1RAs make them a compelling therapeutic candidate for glaucoma. Preclinical evidence provides a potential mechanism, with animal models showing that GLP-1RAs can reduce intraocular pressure by inhibiting sodium-potassium ATPase and enhancing nitric oxide synthesis [101]. This hypothesis is strongly supported by a consistent body of evidence from large-scale retrospective clinical studies. Across diverse patient populations, GLP-1RA use has been associated with a reduced risk of glaucoma. In obese patients without diabetes, their use was linked to a significantly lower incidence and progression of primary open-angle glaucoma (POAG) and ocular hypertension compared to other weight-loss drugs [102]. Furthermore, studies have shown GLP-1RAs are associated with a modest decrease in intraocular pressure, particularly in individuals with elevated baseline intraocular pressure (IOP) [103], and a reduced risk of developing glaucoma in patients with T2DM [104]. This protective association has been repeatedly observed, linking GLP-1RA use to lower risks of POAG, ocular hypertension, and the need for first-line glaucoma treatment in large cohorts of T2DM patients [105 –107]. A systematic meta-analysis has corroborated these findings, confirming a significant association between GLP-1RA utilization and reduced glaucoma incidence in retrospective studies [108]. While this body of evidence is highly encouraging, it is important to note that these findings are derived from studies with considerable methodological heterogeneity, a limitation that will be addressed in the Discussion. Consequently, prospective randomized controlled trials (RCTs) are imperative to establish definitive causality. Future directions should prioritize RCTs targeting early-stage glaucoma, with primary endpoints encompassing both structural (e.g., OCT-based RGC metrics) and functional (e.g., visual field preservation) outcomes, to validate the potential of GLP-1RAs as a true disease-modifying strategy in glaucoma management.

As a leading cause of central vision loss in the elderly, AMD is a multifactorial disease driven by chronic oxidative stress, inflammation, and metabolic dysregulation [109, 110].

Intriguingly, the current clinical evidence regarding GLP-1RAs in AMD is profoundly dichotomous, with their effect appearing to depend entirely on the disease subtype. On one hand, a retrospective cohort study demonstrated a significant protective association, linking GLP-1RA use to a reduced risk of nonexudative AMD, with an efficacy reported to be superior to that of metformin, insulin, and statins [111]. This benefit is putatively driven by the systemic amelioration of metabolic risk factors like high body mass index (BMI) and hyperlipidemia, though direct preclinical evidence remains lacking. However, this potential protective mechanism is overshadowed by alarming findings in nAMD. Another large-scale cohort study demonstrated that long-term GLP-1RA use was associated with a more than twofold, exposure-dependent increase in the risk of developing nAMD. The proposed mechanism for this adverse outcome involves a paradoxical effect where GLP-1RA-induced metabolic shifts cause transient retinal hypoxia, triggering the upregulation of pro-angiogenic chemokines like CXCL12 [112].