Introduction and Anatomy

The papillomacular bundle, more accurately the maculopapillary bundle (MPB), is the name given to the dense collection of retinal ganglion cell (RGC) axons projecting from the retinal macula to the optic nerve. This axonal collection accounts for over 90% of the total optic nerve axons [1]. Notably, the MPB is characterized by the highest axon density in the retinal nerve fiber layer (RNFL), and, like all RGC axons anterior to the lamina cribrosa (LC), the fibers in retinal layer are unmyelinated [2]. The high axonal density in this region limits vascularization, which consequently affects oxygen and nutrient supply to this area [2].

Centrally located within the macula, the anatomical fovea features a sparse ganglion cell layer (GCL) and a sparse RNFL at the periphery. The central 350 μm of the fovea, referred to as the foveola, lacks both a GCL and RNFL [3,4,5]. This arrangement maximizes photoreceptor photon absorption by reducing light scattering from the inner retinal layers. Consequently, RNFL fibers circumnavigate the fovea to form the MPB (Fig. 1). Axons from the nasal macula, both in the superior and inferior foveal periphery, directly and compactly traverse to the optic nerve head (ONH), also referred to as the optic disc. In contrast, RGC fibers temporal to the fovea curve around it to reach the optic nerve head, forming structures known as arcuate bundles (Fig. 1) [1]. Upon reaching the disc, the MPB fibers collect and turn 90 degrees at the superficial nerve fiber layer of the disc to form the prelaminar optic nerve [6].

Fig. 1
figure 1

Configuration of the RNFL. Schematic depiction of the retinal ganglion cell fibers (RGC) in the retinal nerve fiber layer projecting to the optic nerve head (ONH). In the maculopapillary bundle there is a greater density of retinal ganglion cell axons than in the peripheral retina, leading to a greater density of RGC fibers in the temporal ONH compared to the nasal ONH

Full size image

Adjacent to the fovea, the thickness of the GCL and RNFL in the macula increases sharply, exhibiting the highest ganglion cell density in the entire retina [4, 5]. RGC density then rapidly decreases from the foveal perimeter to the retinal periphery by over 1000-fold [4]. Early studies overestimated the ganglion cell concentration outside the macula along the horizontal meridian because of confusion with amacrine cell bodies [4, 5]. The pattern of amacrine cell density mirrors that of RGCs and peaking between 2 and 6 mm from the center of the fovea [7]. Amacrine cell concentration declines at a slower rate than RGCs following increasing retinal eccentricities. At the peak ganglion cell density, the proportion of displaced amacrine cells in the GCL has been measured as low as 3%, rising to approximately 70% at the ora serrata [4, 7].

Primates are the closest animal model for the study of human neuroretinal diseases due to their retina’s close homology to the human organ. Studies in macaques have characterized 18 distinct classes of RGC cells by electrophysiological response properties and gene expression [8]. More recent transcriptomic analysis of human postmortem retinas has identified 12 types of RGCs. The overall transcriptome similarity between macaque and human retinal cells validates the macaque as a suitable model [9]. ON-midget and OFF-midget RGCs (P-cells) make up over 80% of all RGCs in the retina, followed by ON-parasol and OFF-parasol RGCs (M-cells), which comprise approximately 10% [8, 9]. Midget RGCs are 20 times more abundant than parasol RGCs in the GCL of the maculopapillary area [10]. P-cells predominantly establish synaptic contact with amacrine cells and single contact with midget bipolar cells, while M-cells have diffuse contact with several parasol bipolar cells [1].

In accordance with their high prevalence in the macula, P-cells are smaller than M-cells in terms of both cell body size and axonal diameter. Additionally, P-cells comprise most of the MPB [11,12,13]. However, in spite of the general difference in average size between the two cell types, both midget and parasol cells in the peripheral retina tend to have larger receptive fields than those in the fovea, serving to facilitate converging input in retinal areas corresponding to lower acuity vision [12, 13].

Foveal and macular RGCs tend to have smaller diameter axons than those emanating from the peripheral retina [14]. Similarly, axons that terminate on the temporal side of the ONH tend to have lower diameters than those on the nasal side [15•]. Both facts are consistent with the high ratio of P-cell to M-cell fibers present in the MPB [15•]. The grouping of dense small-diameter axonal fibers emanating from the macula creates the characteristic dense parallel fibers of the MPB, especially in the region temporal to the optic disc and nasal to the macula.

Different segments of the optic disc can be distinguished based on the origin of their axonal fibers in the retina. Specifically, fibers originating from the macula, which constitute the MPB, are primarily found in the central and medial temporal portions of the disc [16,17,18]. Subsequent research corroborated these findings by showing that partial vision loss in specific visual field areas aligns with RNFL thinning in regions of the ONH. These findings were later incorporated into what is now known as the Garway-Heath mapping [19,20,21].

Optical coherence tomography (OCT) is a noninvasive imaging technology that employs laser pulses to measure the thickness of the tissue based on the tissue’s optical reflectivity [22]. OCT was first employed to measure in-vivo retinal thickness followed shortly after its development [23]. Segmentation of the various retinal layers imaged by OCT allows for the measurement of individual retinal layer thicknesses and comparison to age-adjusted norms. Over the years, OCT has been employed to identify RNFL changes in multiple optic neuropathies [24,25,26,27]. Later, OCT’s application was expanded as a tool to measure the thickness of the GCL and inner plexiform layer (IPL) in many optic neuropathies [28]. Since then, OCT has developed into a powerful tool for detecting neural degradation in the retina.

OCT has been instrumental in identifying thinning of the MPB in several optic neuropathies, including Leber hereditary optic neuropathy (LHON), dominant optic atrophy (DOA) nutritional optic neuropathies, and toxic optic neuropathies [15•, 29, 30]. Conditions associated with classical degeneration of the MPB are distinguished by cecocentral scotomas, a diminishing of central visual acuity, a reduction in color sensitivity, and are concurrent with atrophy of the ganglion cells within the MPB. Cecocentral scotomas are predominantly associated with a characteristic atrophy in the macular GCL and temporal peripapillary RNFL, and are often present in the early stages of inherited, toxic, and nutritional optic neuropathies [15•, 29, 31,32,33].

The various MPB disorders have different etiologies, many of which implicate mitochondrial mechanisms. For example, LHON and DOA are both associated with the existence of specific pathogenic mutations in genes encoding mitochondrial proteins [34]. The current state of the literature has yet to provide a consensus on why MPB degeneration emerges as a shared consequence across these distinct causative factors. Beyond these conditions, non-selective MPB degradation is also observed in other optic neuropathies, including glaucoma. We review the current literature on the subject, both clinical and experimental, and suggest possible mechanisms that could trigger optic neuropathies involving this highly vulnerable area.

Pathophysiology and Disease

Hereditary Optic Neuropathies

Leber Hereditary Optic Neuropathy

Leber hereditary optic neuropathy (LHON) and dominant optic atrophy (DOA) are optic neuropathies with a genetic basis and which are characterized by cecocentral scotomas leading to bilateral blindness (Fig. 2) [35,36,37]. Despite similar characteristics, each disease has a distinct etiology and disease progression.

Fig. 2
figure 2

Example of structural and functional deficits in LHON.. A) Fundus images from a patient with LHON. B) Optical coherence tomography (OCT) of the ganglion cell complex (ganglion cell and inner plexiform layers) in the maculae. There is significantly more thinning in the right eye than the left. C) OCT scan of the retinal nerve fiber layer (RNFL) of the right optic nerve of the OD demonstrated thinning with a temporal peripapillary predominance. D) A 24–2 visual field analysis of the OD shows profound visual sensitivity loss in the superior paracentral visual field, consistent with dysfunction corresponding to the inferior macular retinal ganglion cells and inferotemporal optic nerve head

Full size image

LHON is a rare condition whose onset requires the existence of one of three primary point mutations in the mitochondrial genome: G3460A, G11778A, and T14484C (Fig. 3) [38]. Recent studies have estimated the prevalence of cases for the G3460A and T14484C mutations to 1:66,000 in India [38], and 1:65,000 in Europe [39]. Males are eight times more likely to be affected than females [38, 39]. However, it is crucial to note that being a carrier of the mutations does not guarantee the development of visual loss, with LHON penetrance being 50% in males and 10% in females [35, 38, 40].

Fig. 3
figure 3

Current summary of implicated proteins in LHON and DOA. LHON mutations all affect complex I (NADH dehydrogenase) in the electron transport chain. These mutations have been shown to increase \({\text{O}}_{2}^{-}\), which is converted to H2O2 by superoxide dismutase 2 (SOD2) in mitochondria. H2O2 is converted to water by catalase (Cat) or glutathione peroxidase (GPX). Reactive oxygen species are direct drivers of apoptosis. Mutations in KIF5A also increase the amount of apoptosis in LHON cybrid cells. DOA has several driving mutations, the most common implicating the pathways for mitochondrial fission and fusion. DRP1 stimulates proteins on the outer mitochondrial membrane to initiate fission mediated through several outer membrane proteins: FIS1, MFF, MiD49, and MiD51. MFN1/2 is responsible for homomeric adhesion between mitochondrial outer membranes to initiate outer membrane fusion. OPA1 has two forms: a membrane-bound long isoform (OPA1-L) and a short (OPA1-S) isoform, cleaved by YMEL1 and OMA1. The long isoform is responsible for triggering inner mitochondrial membrane fusion. AMPK-stimulated autophagy increases in OPA1 mutant animals and cell lines, and its inhibition prevents neurodegeneration, suggesting a mechanism in the pathophysiology of the disease. PNPT1, a protein responsible for mtRNA transport, and SSBI, responsible for mtDNA replication, are also genes that can be mutated in a DOA disease context

Full size image

LHON is characterized by an initial thinning of the peripapillary RNFL starting in the inferotemporal optic disc, which then extended centrally, sparing the superonasal quadrant, an area devoid of MPB (Fig. 2) [15•]. Recent research has shed light on the pattern of macular ganglion cell layer (GCL) degeneration, noting concurrent changes in vascularization throughout disease progression. Castillo et. Al (2022) found that even in LHON mutation carriers, there is a significant reduction in macular GCL thickness unaccompanied by visual loss or MPB RNFL thinning [41•]. These carriers also had decreased capillary vessel density in the superficial capillary plexus of the nasal macula without an effect on the ONH. In contrast, advanced patients had more pronounced vessel density changes in the temporal disc than in the macula [41•].

Castillo et al. (2022) noted minimal differences in GCL between short-term and long-term LHON patients. However, peripapillary RNFL thinning, especially in the temporal quadrant, was progressive and related to the duration of the illness [41•]. These results suggest that changes in macular GCL parameters are linked to, but not necessary for, subsequent ONH and peripapillary RNFL thinning. This affirms that RNFL thinning is a progressive hallmark of the disease.

Carbonelli et al. (2022) tracked the changes in OCT parameters in four patients with LHON mutations over an interval spanning two months before and five months after the onset of visual loss. They observed that preceding visual loss the superior and inferior peripapillary RNFL (pRNFL) increased in thickness, which then persisted up to 40 days following the onset of visual impairment, [42••]. This regional pRNFL thickening was followed by rapid thinning and decline in visual acuity. In contrast, there were no thickness changes in macular GCL prior to visual loss onset. However, within 30 days of the conversion to affective LHON, there was a decline of up to 40% in macular GCL thickness, most prominently in the nasal region [42••].

Similar observations were made by Wang et al. (2021), who demonstrated pRNFL thickening before temporal pRNFL thinning, accompanied by central visual field defects [43••]. Approximately 3–6 months post visual loss, the average pRNFL thickness was observed to be lower than its original baseline reflecting a transition from MPB deterioration to broader disc atrophy. This trend persisted over the 60-month monitoring period, with the nasal quadrant being the last region to experience pRNFL thinning [43••].

The findings of Carbonelli et al. (2022) and Castillo et al. (2022) provide two perspectives on MPB vulnerability. The first demonstrates that RNFL changes precede the subsequent reduction in VA, and the second suggests that changes in macular vascularization could be an inciting factor for progressive optic nerve damage [41•, 42••].

Mitochondrial features other than the principal genetic mutations can affect the penetrance of LHON. Mitochondrial haplogroups (collections of other correlated single nucleotide polymorphisms in the mitochondrial genome) have been linked to LHON penetrance in LHON carriers [44,45,46,47]. Mitochondrial motility is a factor influencing the vulnerability of RGCs. In vitro studies using human induced pluripotent stem cells (hiPSCs) carrying the G11778A mutation have demonstrated an increase in reactive oxygen species (ROS) within induced RGCs, both in affected patients and unaffected mutation carriers (Fig. 3).

Variations in mitochondrial motility were observed between cells that were likely to undergo apoptosis and unaffected cells. This divergence was correlated with reduced levels of the protein kinesin-1 family member KIF5A, a motor protein involved in microtubule-associated ATP-dependent transport and responsible in part for mitochondrial motility, in cells likely to undergo apoptosis (Fig. 3) [48•]. These results indicate mitochondrial dynamics are crucial in determining RGC axonal vulnerability. In addition to this, older studies have hinted that higher levels of mitochondrial biogenesis in carriers is a protective factor against the development of disease [49••].

Hereditary Optic Neuropathies

Dominant Optic Atrophy

Dominant optic atrophy (DOA) exhibits several clinically similar features to LHON, particularly in terms of the pattern of neuroretinal damage and pathophysiological mechanisms. Early characterizations of the disorder noted the simultaneous onset of a gradually expanding binocular cecocentral scotoma that spares peripheral vision [50, 51]. DOA is triggered by a set of nuclear genetic mutations that affect encoding of mitochondrial proteins (Table 1). The onset of DOA can begin within the first two decades of life and progress into adulthood, leading to legal blindness [50,51,52,53]. The speed of visual deterioration is typically much slower than LHON [52, 54].

Table 1 Summary of the MPB-degenerative diseases 
Full size table

The visual impairment in DOA has been linked to a loss of ganglion cell density in the segments of the retina associated with the MPB. OCT scans reveal both macular GCL and peripapillary RNFL thinning in DOA patients. This thinning is in all quadrants of the ONH and macula in the late stages of DOA [55]. Yet, the most rapid and pronounced DOA-associated ganglion cell atrophy is noted in the MPB-associated areas, especially the peripapillary temporal RNFL and nasal sections of the macular RNFL, GCL, and IPL [56, 57]. The specificity of early DOA atrophy to the MPB is underscored by the relative preservation of ganglion cells in the nasal peripapillary region of the optic disc [55, 56].

In advanced stages of DOA, diminished vascularization is evident in the temporal radial peripapillary plexus [58, 59•]. Both the deep and superficial macular capillary plexuses exhibit thinning in patients with advanced DOA, and the degeneration of this vascular network has been shown to correlate strongly with macular GCL thinning [59•]. Studies consistently show that visual acuity loss correlates most strongly with temporal peripapillary RNFL, nasal macular GCL, and the macular deep capillary plexus [56, 57, 59•]. Interestingly, a comparative study comparing acute and chronic LHON patients to DOA patients found no significant differences in the macular GCL thickness between these groups [60••].

In contrast to LHON, where pathological mutations are in the mitochondrial genome, DOA-causing mutations are in nuclear DNA yet code for mitochondrial proteins [61]. Optic atrophy 1 (OPA1) is a nuclear gene located on chromosome 3q28 and is the predominant gene implicated in DOA, accounting for the majority of cases [52, 62]. OPA1 encodes a mitochondrial inner membrane-bound GTPase membrane that is involved in dynamin-associated mitochondrial internal membrane fusion and fission (Fig. 3) [63, 64]. Mutations in OPA1 that cause DOA can be missense, deletion, or inversions [55, 65]. While less prevalent, other DOA-linked mutations encompass genes integral to mitochondrial fusion, fission, and apoptosis, such as OPA3, MFN1, MFN2, and DRP1 [63, 64, 66,67,68,69]. The breadth of recognized DOA mutations has expanded to include those involved in mitochondrial RNA transport (PNTP1), mitochondrial DNA replication (SSB1), and mitochondrial protease activity (AFG3L2, SPG7) (Fig. 3) [70,71,72]. Mitochondrial homeostasis emerges as a common theme among various DOA-related gene functions (Table 1). The process of mitochondrial biogenesis has been commonly implicated in both LHON and DOA pathophysiology [49••, 64]. A notable biochemical difference between the two diseases is the specific mutation effects on mitochondrial oxygen consumption and total mitochondrial respiration rate. In LHON cybrids, the total oxygen consumption rate and ATP production in cells are unaffected by the disease-associated mutations despite a demonstrated reduction of complex I respiration [73, 74••]. In contrast, OPA1 has been shown to decrease oxygen consumption and oxidative respiration in induced pluripotent stem cell lines [75•, 76].

Another proposed cellular mechanism for DOA involves mitochondrial autophagy. A study of OPA1 mutations in purified RGC cultures demonstrated increased calcium-dependent mitophagy through the calcineurin/AMPK (AMP-dependent protein kinase) signaling pathway (Fig. 3) [77]. In mouse models, it was shown that the neuronal effects of OPA1 can be curtailed via the inhibition of AMPK, which results in a decrease in neuron autophagy and an elevation in the mitochondrial content [78].

Despite the differences in possible mechanisms of LHON and DOA, both conditions possibly respond to idebenone treatment. Idebenone, a synthetic coenzyme Q that passes the blood–brain barrier, was initially formulated to address complex I deficiencies in LHON [79, 80]. Subsequent studies showed that idebenone also is a superoxide scavenger [81]. A randomized control trial of idebenone in LHON failed to meet its primary outcome measure [80], but there were suggestive effects in analysis of other outcome measures in a nonrandomized study [79]. A subsequent nonrandomized trial of OPA1-associated DOA visual loss showed better vision compared to historical controls [82]. Although the clinical evidence is not strong for idebenone in these diseases, it does point to a potential role in ROS scavenging as a treatment.

Nutritional and Toxic Optic Neuropathies

Nutritional and toxic optic neuropathies are distinct yet related conditions which manifest with characteristic clinical features, including optic disc pallor, reduced color vision, and the presence of unilateral or bilateral cecocentral scotomas. In the initial stages of these disorders there is a thickening of the temporal RNFL and GCL, which subsequently progresses to thinning [83,84,85,86,87,88,89,90,91,92,93,94].

The primary etiological difference between the two types of metabolic neuropathies lies in their underlying causes: nutritional neuropathies stem from deficiencies in specific nutrients, whereas toxic neuropathies result from exposure to harmful toxins or medications.

Nutritional optic neuropathies primarily stem from deficiencies that facilitate oxidative metabolism (Table 1). The most prevalent nutritional optic neuropathy results from vitamin B12 deficiency, although deficits of folate, thiamine, pyridoxine (vitamin B6), or copper have also been implicated [29, 94]. Vitamin B12 is essential to protein synthesis, fatty acid metabolism, and the conversion of homocysteine to methionine [95]. It also can scavenge ROS, including superoxide [96,97,98]. Folate acts as a coenzyme in nucleotide synthesis for both purines and pyrimidines [99, 100]. Deficiencies in folate have been hypothesized to alter mitochondrial metabolism through low levels of ADP synthesis [101]. Thiamine or vitamin B1 is a key metabolic cofactor implicated in the Kreb cycle, pentose phosphate pathway, glycolysis, nucleic acid synthesis, and homocysteine metabolism [100, 102]. Copper, a ligand of the cytochrome c oxidase (complex IV) in the mitochondrial electron transport chain, is necessary for oxidative metabolism [103].

Similar biomolecular pathways affecting mitochondria are hypothesized to be caused by drugs responsible for the development of toxic optic neuropathies [104]. Most but not all drugs observed to induce optic neuropathies, including ethambutol, linezolid, ciprofloxacin, chloramphenicol, macrolides, aminoglycosides, fluoroquinolones, and isoniazid, have a primary antimicrobial function, yet are believed to affect mitochondrial functions. Clinically, they induce clinical features similar to those of other mitochondrial optic neuropathies, e.g. cecocentral scotomas, and act in a dose-dependent manner [91, 92, 104,105,106,107,108].

The antimicrobial action of these antibiotics, except for tuberculosis-targeting drugs and fluoroquinolones, is mediated by bacterial rRNA binding which impairs protein synthesis (Table 1) [92, 104, 106, 108,109,110,111,112,113]. The homology between bacterial and mitochondrial ribosomes has led to a hypothesis that the drug impairs mitochondrial protein synthesis through the same mechanism [114]. Ethambutol, a metal chelator, has been shown to reduce zinc and copper levels in primate models and has been observed to accelerate the development of optic neuropathies in patients with LHON- or DOA-associated mutations [115, 116]. Isoniazid’s mechanism of action remains unelucidated, although it has been shown to enhance levels of reactive oxygen species and cause vitamin B6 deficiency [117,118,119]. Such a finding bolsters the hypothesis that these drugs exert their pathological effects in the mitochondrial electron transport chain by reducing ligand availability [115, 120].

Methanol poisoning is a non-medicinal toxic optic neuropathy associated with MPB degeneration, characterized by optic nerve pallor and OCT thinning of the RNFL and GCL [121,122,123,124,125,126]. While temporal ONH RNFL and GCL thinning are typically associated with methanol toxicity, there have been a few cases in which this was not the case, and RNFL thinning spared the MPB [127, 128]. The molecular mechanism of methanol toxicity likely involves its conversion to formic acid, which can interfere with oxidative metabolism by inhibiting complex IV of the electron transport chain [129].

However, it is crucial to distinguish between the above toxic neuropathies and those associated with alcohol ingestion (ethanol) and tobacco use. Although it is controversial whether these substances are direct causes of optic neuropathies, they can exacerbate the penetrance of genetically predisposed conditions like LHON, evidenced by the increase copy numbers of mitochondrial DNA in these predisposed patients [130, 131]. Cases associated with alcohol and tobacco use differ from classical nutritional and toxic neuropathies, which lack a recognized genetic predisposition [30].

Glaucomatous Optic Neuropathy

Glaucomatous optic neuropathy, or glaucoma, is the leading cause of irreversible, progressive blindness worldwide. The incidence increases with age [132,133,134,135,136], and is responsible for moderate to severe visual impairment in about 4.1 million individuals age 50 years or older worldwide, and blindness in about 3.6 million [135]. The prevalence of primary open-angle glaucoma (POAG) and primary-angle closure glaucoma (PACG) are estimated to be 2.4% and 0.6%, respectively [132, 133]. While POAG is more common among men, PACG is more common among women [132, 133].

Pathophysiologically, glaucoma is categorized by degeneration of retinal ganglion cells (RGCs) and subsequent visual impairment. Although the etiological trigger of the disorder has yet to be established, depending on the region of the world 10%-50% of patients with glaucoma have increased ocular pressure (IOP). Currently, IOP reduction is the only established intervention for slowing glaucomatous progression [137]. However, while increased IOP is a common feature, forms of glaucoma without elevated IOP exhibit similar structural and functional patterns of damage.

Formerly it was believed that glaucoma predominantly affects RGC axons arising from the peripheral retina, while MPB damage was only seen late in the disease [138]. This understanding evolved starting around 2013, when the widespread use of OCT to measure retinal layers revealed that early glaucoma commonly affects the macula. Hood et al. [139] showed that glaucoma causes atrophy of the GCL and IPL in the inferior macula, projecting to the inferotemporal ONH and sparing the superior MPB. They called this retinal area the macular vulnerability zone (MVZ). Subsequent large cohort studies have corroborated these finding, emphasizing the significance of MVZ damage in glaucoma [140, 141, 142•, 143••].

Recent studies have highlighted deficits in the MPB in early glaucoma. Traditional evaluation methods to assess visual defects in glaucoma, such as Humphrey 24–2 or 30–2 standard automated perimetry (SAP) and current clinical OCT imaging, may under-estimate the extent of macular damage in the early stages of the disease. Enhanced detection in these stages is achievable with 10–2 SAP and with retinal nerve fiber optical texture analysis (ROTA), a novel axonal detection algorithm based on OCT scans [142•, 144]. In a comprehensive cohort study of 204 patients diagnosed with early-stage glaucoma, MPB anomalies were identified by ROTA in 71.6% of the subjects, with a notable subset showing deficits in the inferior MPB. [142•]. Another study using the ROTA technique found MPB abnormalities overlooked by current clinical OCT in about 5% of patients with ocular hypertension [145•].

Decreased visual acuity typically occurs late in progressive glaucoma. There are rare examples of significantly decreased visual acuity at an early stage, although reported cases could include patients where nonglaucomatous optic neuropathy was inadequately excluded. Examples of misattribution of visual loss to glaucoma could include patients with bilateral central scotomas without assessment for LHON or DOA, or unilateral disease assessed with computed tomography but not magnetic resonance imaging focusing on the optic nerve [146]. Such studies failing to completely rule out other optic neuropathies could occur in studies that demonstrated decreased best-corrected visual acuity (BCVA) and color vision deficiency in association with OCT-detected atrophy of the peripapillary RNFL and macular GCL, specifically in regions corresponding to the MPB [32, 147]. That said, there are certainly patients with MPB thinning as a result of the glaucomatous process at the disc, even if the visual acuity is spared [145•, 148•]. Additionally, OCT-A has also shown that vessel density loss correlates with the changes in RNFL, GCL, and IPL thickness in the peripapillary region but not in the macula in patients with structural damage to both areas [149].

Early studies suggested that glaucoma with normal IOP may be more prone to MPB damage than other forms of glaucoma [138]. However, this assertion remains controversial [143••]. Recent studies have linked other retinal characteristics, such as the position of the central vessel trunk (CVT) and the size of the optic disc, with early-stage glaucoma MPB involvement. Positioning of the prelaminar CVT nasally correlates positively with glaucomatous MPB involvement [140, 150]. This distinction in CVT positioning is only notable at the prelaminar segment of the optic nerve as there is no difference between glaucoma-affected eyes and healthy controls at the level of the lamina cribrosa [150]. Furthermore, optic nerve cup morphology is another factor predictive for MPB involvement. Both ONH size and the ratio of the longest to the shortest axis of the ONH positively correlate with paracentral scotomas, macular damage, and nasal shift of the CVT [145•, 148•].

Intriguingly, while the CVT position in glaucoma patients with normal IOP does not significantly differ from that of healthy controls, their lamina cribrosa morphology does. Specifically, the lamina cribrosa exhibits a more pronounced posterior curvature than healthy controls and DOA patients, with the latter two groups showing no significant differences between them [151]. These findings further delineate glaucoma from other optic neuropathies as having unique anatomical characteristics.

Optic Neuritis

Optic neuritis has a disproportionate effect on fibers of the temporal anterior optic nerve. This is particularly manifested by significant dyschromatopsia, loss of contrast sensitivity, and central scotomas [152]. Those features are all consistent with damage to small RGC axons, i.e. the parvocellular system. However, it could also be consistent with a mitochondria-mediated mechanism similar to that seen in LHON and DOA.

Studies exploring reactive oxygen species generation or mitochondrial pathways in animal models of optic neuritis are limited. One study demonstrated that suppression of ROS generation in mitochondria reduced myelin fiber injury in mouse models of optic neuritis and that elevating ROS levels had the opposite effect [153]. Another group demonstrated that ROS generation was a consequence of pro-inflammatory TNFα stimulation. They also showed that treatment with the multiple sclerosis therapy fingolimod in vitro reduced ROS production and neuronal injury [154]. Other animal model studies of optic neuritis demonstrate the importance of SIRT1 in many pathways, among which are antioxidant effects [155, 156]. However, in contrast with MPB-specific diseases, ROS production likely is a consequence of inflammation as opposed to a driving mechanism of pathology.

Although there is insufficient evidence to implicate mitochondria-specific damage in typical inflammatory optic neuritis, the overlap between multiple sclerosis and LHON mutations and other evidence to implicate mitochondria in optic neuritis suggest that this is a worthwhile area for future research [153, 157, 158].

Hypotheses for the Pathogenesis of Degenerative Diseases of the Maculopapillary Bundle

A consistent feature observed in neuropathies selectively affecting the MPB (including hereditary, nutritional, and toxic neuropathies) is the involvement of impaired mitochondrial bioenergetics and metabolism (Table 1). While these etiological factors universally affect mitochondrial functions at the cellular level, the specific pathogenesis underlying MPB-selective RGC degeneration remains unproven. A prevailing hypothesis posits that mitochondrial impairment facilitates the generation of reactive oxygen species (ROS). Elevation of ROS has been noted directly in LHON [48•, 172, 173] and DOA [174, 175]. Indirect evidence from ethambutol toxic neuropathy suggests a similar ROS elevation, based on antioxidant depletion and coenzyme Q supplementation in animal models [176, 177].

The generation of ROS, a byproduct of oxidative metabolism, can trigger apoptotic cell death, especially when the electron transport chain is impaired [178]. Small fiber-selective degeneration in MPB diseases may be attributed to differing bioenergetics across neuron types. In the MPB, the sodium–potassium pump, responsible for restoring the resting potential after neural depolarization, constitutes the primary energy consumer in axons. This ion exchanger is a membrane protein, whereas the mitochondria are cytosolically distributed. Therefore, the ratio of energetic demand to supply can be thought of as the ratio of axonal surface area to volume, which is greatest in small-diameter axons [15•, 179]. Since ROS production is intricately linked to mitochondrial distribution, it follows that greater per-volume oxidative production takes place in smaller-diameter axons.

We have previously suggested that this gap in ROS density creates a vulnerability for the small-diameter P-cell fibers of the MPB to undergo cell death [179]. Yet, the MPB’s typical degeneration doesn’t solely target individual small fibers. Instead, it manifests as a progressively enlarging defect marked by both visual loss and retinal layer degeneration. One plausible explanation lies in the compact arrangement of the small-diameter axonal fibers in the MPB. The demise of even a few of these fibers could release ROS to neighboring axons, potentially triggering a cascade of apoptosis [180].

Using an in-silico model, we illustrated how this axonal vulnerability aligns with clinical and histological findings in LHON [181•]. Given the similarities in mitochondrial metabolism in other MPB disorders and common patterns of neurodegeneration, we propose a similar mechanism in those disorders. One similarity between LHON and DOA is the implications of mitochondrial proliferation dynamics in each disorder. In LHON, asymptomatic mutation carriers have upregulated mitochondrial biogenesis, which may confer a protective effect on the mutations by distributing ROS production more evenly in the optic nerve [49••]. Conversely, many DOA-associated mutations impair mitochondrial fission and are conducive to mitochondrial autophagy [63, 64, 69]. Thus, the underlying mechanisms in these conditions could be deemed functionally synonymous: a decrease in overall mitochondrial count amplifies the oxidative strain on individual mitochondria, potentially triggering the previously mentioned ROS-induced cascade.

Glaucoma is excluded from this classification because, although recent studies have shown an involvement of the MPB even in early disease, neither the clinical findings nor the pathophysiology mirrors the mitochondrial optic neuropathies. Instead, glaucoma affects the peripheral retina concurrently with the MPB in the early stages of the disease. In addition, MPB vulnerability in glaucoma is pronounced in the inferior segment of the MPB. It is distinct from the other MPB disorders discussed above in terms of its etiology [143••] and clinical presentation.

Conclusion

The MPB is affected in a broad spectrum of optic neuropathies. Many of the diseases that are MPB-selective in their early stages, such as LHON, DOA, and metabolic optic neuropathies, implicate mitochondrial metabolism, biogenesis, and motility. Although each has a distinct etiology, all are consistent with mitochondrial mechanisms. The generation of reactive oxygen species (ROS) presents a promising mechanistic nexus linking mitochondrial impairment with the vulnerability and progressive degeneration of small fibers in the MPB. While systemic propagation of reactive oxygen species has been posited in in-silico models, empirical validation though laboratory experimentation has not yet been done.

Recent empirical studies have started to shed light on the MPB involvement in the early stages of glaucoma. However, glaucomatous involvement of the MPB stands in stark contrast to the optic neuropathies described above because of the difference in effects on central vision, particularly visual acuity and color vision.

We advocate for expanded research in several areas relevant to the MPB and disease. First, work is needed to empirically test the ROS propagation hypothesis, through in vitro or in vivo models, by exploring how ROS release affects adjacent axons. Second, it would be beneficial to explore if the death of a subgroup of MPB axons can cause pathological changes in the entire optic nerve other than the distal axons, e.g. mechanisms for secondary degeneration. Third, a deeper exploration into the axon biology of small fibers within the MPB in early glaucoma could help in understanding whether their loss is different from that occurring in non-MPB axons entering the ONH.