Appropriate patient population for future visual system axon regeneration therapies

Abstract

A number of blinding diseases caused by damage to the optic nerve result in progressive vision loss or loss of visual acuity. Secondary glaucoma results from traumatic injuries, pseudoexfoliation or pigmentary dispersion syndrome. Progressive peripheral vision loss is common to all secondary glaucoma irrespective of the initial event. Axon regeneration is a potential therapeutic avenue to restore lost vision in these patients. In contrast to the usual approach of having the worst possible patient population for initial therapies, axon regeneration may require consideration of appropriate patient population even for initial treatment trials. The current state of axon regeneration therapies, their potential future and suitable patient population when ready is discussed in this perspective. The selection of patients are important for adoption of axon regeneration specifically in the areas of central nervous system regenerative medicine.

Graphical Abstract.

The potential adoptable regeneration workflow from preclinical to clinical stage. A three-step workflow adopted from preclinical to a eventual clinical paradigm. They are establishing conditionality in the eye of appropriate patients, promoting long distance axon regeneration (stages 1–3 with potential timeline) followed by visual function assessment with comprehensive electrophysiological measurements. The re-innervation strategies are multi-modal, which is indicated by grey hexagon.

The overarching challenge in all forms of glaucoma including glaucoma with type 2 diabetes and trauma-induced progressive optic nerve (ON) neurodegeneration is rehabilitation, axonal regeneration of the neurons, and re-connectivity(Benowitz et al., 2017; Williams et al., 2020; Yang et al., 2020). Traumatic vision loss in civilian or military personnel can arise from a spectrum of severe head or thoracic trauma with vascular or brain injuries(Ibanez et al., 2021). Finding more than 100,000 human eyes having some level of traumatic optic neuropathy (TON) in a single military zone of conflict is not uncommon(Justin et al., 2020; Singman et al., 2016). Most frequently those eyes progress to demonstrate some form of secondary glaucoma with progressive peripheral vision loss. When combined, the number of military personnel and civilians with TON-induced secondary glaucoma is estimated to be more than 40 million worldwide, rendering it a huge social and economic burden (Kang & Tanna, 2021). Glaucoma broadly is divided in two anatomic categories: angle closure and open angle, depending upon whether the angle between iris and cornea is 15⁰ or larger(Mueller et al., 2023). They are characterized secondary if a prior disease or injury can be clearly ascribed to its onset or primary if no such clear association can be established. A group of patients show statistically normal range of IOP termed normal tension glaucoma (NTG). In patients leading to glaucoma, large diurnal IOP fluctuations occur(Asrani et al., 2000). In different individuals acute elevated IOP may result in glaucoma whereas in general it is cumulative exposure to elevated IOP that results in slow progression of glaucoma. The primary as well as secondary glaucoma may remain sub-diagnostic due to the slow nature of the progression in loss of visual acuity. Thus, primary open angle glaucoma (POAG) is most common form and primary angle closure (PACG) is more widely prevalent among Asians. The occurrence of sub-diagnostic damage simultaneously with discernable damage is a hallmark of all traumatic optic nerve damage. The sub-diagnostic damage results in long-term, slow progressive vision loss characterized as glaucoma. The patient may experience visual field defects, decreased visual acuity, impaired color perception, and anisocoria. In a range of mild to medium trauma (clinically characterized), the long-term consequence is axonopathy(Goldberg et al., 2016). We detail below how current strategies are not designed to address the underlying axon loss and restore lost peripheral vision. Additionally, we describe the promising emerging efforts for the development of initial therapies, which will hopefully become available within the next decade. However, with the emergence of promise, a look into the criterion for selection of the initial patient cohort is a timely consideration, which is what our aim is in this current perspective.

The current treatment modalities and real gap towards intervention strategies.

Trauma-induced optic nerve damage or glaucoma is currently managed by reduction in intraocular pressure (IOP), which halts the disease progression but does not restore lost vision. For TON the current treatment is decompression, IOP reduction and high dose steroid treatment. However, large optic nerve trauma meta-analyses have shown no significant differences between observation versus treatment with steroids in final visual acuity(Chen et al., 2019; Ropposch et al., 2013). The axonopathy remains unaddressed and this gap remains to be filled. Substantial efforts are being made in this area, from potential future stem cell therapy to gene therapies. Current axon regeneration promoting molecules are a few genes and proteins, for example, the knockout of the Phosphatase and Tensin homolog (PTEN) gene but they are expanding(Park et al., 2008). Achieving long-distance axon regeneration without turning or looping remains a challenge (Qian & Zhou, 2020; Williams et al., 2020; Yang et al., 2020). However, the repertoire of molecules to promote axon regeneration is expanding(Curcio & Bradke, 2018). The potential new molecules to promote axon regeneration now includes metabolites and lipids(Chauhan et al., 2020). The latter group of molecules may have an advantage to degrade after exerting their effects. A potential possibility is to consider re-purposing molecules that have been found effective in peripheral nerve regeneration(Caruso et al., 2019; Mirzakhani et al., 2018). The re-innervation of regenerated axons(Benowitz et al., 2017; Qian & Zhou, 2020) remain another challenge. The new axons generated from existing retinal ganglion cells (RGCs) need to be re-innervated in the correct region of the brain in order to achieve functional re-connectivity. Even in this area a number of innovative approaches are emerging. Our large-scale multi-omics analyses(Arcuri et al., 2021; Arcuri et al., 2020; Chauhan et al., 2020; Trzeciecka, Carmy, et al., 2019) combined with computer language processing and machine-learning(Khosla et al., 2021; Myer et al., 2020) has helped identify convergent pathways for axon regeneration and the critical lipids and metabolites in them. The critical knowledge gap is how the first steps of recognition of a target neuron occur by the growing axon and what are the steps in the sequence that results in growth cone collapse leading to formation of an actual functional synapse formation(Skaper et al., 2001; Spencer et al., 1998). Although a number of molecules have been identified that can promote long distance axon regeneration for the visual system but most such molecules are not suitable for delivery. They are rather suitable for gene therapies, specifically gene deletions. Therefore, for a large number of patients easily deliverable molecules preferably in outpatient clinics remain a gap. The second gap as noted above is effective re-innervation strategies and easily deliverable molecule or methods to promote re-innervation.

Emergence of knowledge to promote axon regeneration therapies.

The axon regeneration in the context of ON requires growth cone formation and axon plasmalemma expansion (Pfenninger, 2009; Pfenninger et al., 2003). Growth cone is an actin-rich dynamic structure composed of filopodia and lamellipodia found at the tip of a growing axon. It functions to direct the axon to its target destination by responding to guidance cues. Axon plasmalemma is the outer membrane of the axon. It provides a scaffold for the growth of new axonal processes. Axon regeneration and underlying signaling mechanisms to promote them show tremendous age-dependent variations (Geoffroy et al., 2016; Sutherland & Geoffroy, 2020).

As noted above, axon regeneration signaling mechanisms vary with age. All drug molecules also encounter a fraction of patients as non-responders, which is part of their biochemical individuality and in addition to alteration in signaling mechanisms due to aging. For a wider population of patients in clinics, diverse molecules with overlapping induction capabilities will be needed. Beyond the proof of principle regarding long-distance axon regeneration and reinnervation established in pre-clinical studies, several additional refinements such as delivery approaches and fine-tuning will be necessary to bring therapies to clinics. We project that this may be achievable in 10 to 15 years given the availability of a large and diverse repertoire of a wide variety of molecules for promotion of long-distance axon regeneration. The regeneration of growth cone and plasmalemma expansion in truncated axons can be achieved with modulation of lipids/lipid pathways(Pfenninger, 2009; Pfenninger et al., 1983; Pfenninger et al., 2003). Lipid synthesis and transport are main components for plasmalemma expansion(Pfenninger, 2009; Pfenninger et al., 2003; Pfrieger & Ungerer, 2011). A comprehensive analysis of growth cones(Trzeciecka & Bhattacharya, 2019) and multiple models of regenerating optic nerve(Trzeciecka, Carmy, et al., 2019; Trzeciecka, Stark, et al., 2019) are now available to help identify these critical molecules. Various innovative approaches are emerging which may be adapted for reinnervation after long distance directed axon regeneration. The drugs or surgeries that reduce elevated IOP preserve vision. It is thought that large diurnal fluctuations precede glaucoma. That the lowering of IOP helps axonal transport. However, to restore vision new axon formation and re-innervation is necessary. Both normal tension and primary open angle glaucoma, secondary glaucoma, traumatic optic neuropathies and potentially other optic neuropathies also may benefit from axon regeneration therapies to regain lost vision. The existence of elevated IOP contributed by the anterior eye chamber is expected to hinder axon regeneration in the posterior chamber. Thus, IOP lowering molecules, RGC pro-survival molecules and axon regeneration promoting molecules would be a future arsenal of combinatorial treatments.

In addition to PTEN other biologics also promote regeneration of optic nerve such as oncomodulin(Yin et al., 2003) and ciliary neurotrophic factor (CNTF)(Hauk et al., 2008; Leibinger et al., 2009; Muller et al., 2007). The challenge with such pro-regenerative proteins is that their expression must be carefully regulated to avoid harmful unintended consequences such as uncontrolled cell division. Conversely, ectopic gene expression using engineered vectors often cease to confer uniform long-term expression. The small molecules, and especially hydrophobic lipids, provide deliverable molecules with a possibility of homogeneous distribution. At cellular levels, machinery exists rendering complete degradation of lipids after their effect making them safe and non-toxic at up 10-fold higher levels(Hilgers et al., 2017; Nahab et al., 2011). Published studies of molecules (Jauregui et al., 2022; Zhang et al., 2022) combined with in vitro studies will help identify molecules that could reduce the negative environmental clues and optimize the therapeutic effect of such lipids. Peripheral nerve regeneration studies suggest some molecules capable of reducing negative environmental effects(Delibas et al., 2021) that may be potentially repurposed for the ON after various injuries. Stem cell therapies are also emerging, including neural, mesenchymal, and induced pluripotent stem cells(Garcia-Lopez et al., 2021; Limoli et al., 2021; Zhang et al., 2016). Despite promising results in preclinical studies, further research is needed to fully understand the potential benefits and risks of such therapies for axon regeneration in the human optic nerve.

Considering the significant population of individuals suffering from both glaucoma and type 2 diabetes (T2D), metabolomics should be taken into consideration for axon regeneration in these patients. Diabetes (T2D) leads to alteration in metabolic pathways and accumulation of toxic metabolites, impairing nerve function and axonal regeneration(Sango et al., 2017). This emphasizes the need for further research in this subset of patients, given the increasing prevalence of T2D.

Large axon regeneration associated proteomic and lipidomic datasets can now be normalized and validated through a data transformation pipeline with three stages prior to statistical and bioinformatic analysis(Lee et al., 2012). The cell-type specific enrichment analysis comparing the gene expression profiles of a particular cell type with those of other cell types, can be performed with published bacterial artificial chromosome – translating ribosome affinity purification (bacTRAP) cell types, which refers to a technique used to selectively isolate and identify the transcriptome of a cell type as background(Dougherty et al., 2010; Heintz, 2004; Xu et al., 2014). For identification and visualization of complex protein-protein interactions, comprehensive resource of mammalian protein complexes (CORUM) database and Cytoscapeare program are available.(Giurgiu et al., 2019) The putative lipid binding of proteins can be identified through MetScape and LipidMaps database(Karnovsky et al., 2012; O’Donnell et al., 2019). Kyoto Encyclopedia of Genes and Genomes (KEGG), Reactome, and GO databases all enable identification of pathways, Visualization and Integrated Discovery (DAVID) and ClueGO with semantic clustering and similarity measures can be performed using REViGO and NaviGO.(Ding et al., 2018; Huang et al., 2007; Mlecnik et al., 2019; Supek et al., 2011) Lipid ontologies similarly can be performed using ligand-receptor interaction networks (LION) database for biophysical and chemical properties and the R package Rodin, a computational tool for category, class, and subclass visualization and enrichment. The ON regeneration lipidomic datasets can be normalized individually followed by ComBat batch correction algorithm to mitigate signal drift and technical heterogeneity. To predict regeneration promotion, receiver operating characteristic (ROC) analysis can be performed allowing area under the ROC curve (AUC) calculations.

A plethora of machine learning algorithms and predictive models are rapidly emerging from the vast amount of omics data generated by axon regeneration research. RapidMiner™ software, a workflow for data processing was built to use split-validation model training and testing on various metabolites associated with long distance regeneration. The three algorithms examined were Support Vector Machine (SVM)—used for classification and regression analyses, Edited Nearest Neighbor (ENN)—used to remove noisy and redundant samples from the training dataset, and Deep Learning—used to learn complex patterns from data using artificial neural networks. The programs allow to set up a percentage of data for training for each of the models, and the accuracy of classification as either control or long-distance regeneration for each model. The selection of convergent pathways, metabolites and lipids used in evaluation of long-distance regeneration is thus based on extensive bioinformatics, computational and machine-learning research of multi-omics data and factors fractionated axon and emerging single cell omics data. The omics strategies with machine learning can help find molecules on convergent pathways for concerted treatment.

Need for large animal validation studies.

Rodent studies will serve as the fundamental basis for executing extensive animal testing. It is important to note that the composition of human optic nerve is different, not only at the cellular level but also structurally, to that of mice. In the human optic nerve, RGC axons are arranged in separate bundles with each bundle wrapped by connective tissue called pial septae. In contrast, mouse and rat optic nerves are devoid of pial septae and RGC axons project to the brain as one whole bundle. What effects the pial septae would exert on growing axons, is currently unknown. Whether the human RGC axons grow together within a “tunnel” and topographically project to the correct brain targets remains unknown. Pigs have been used to test axon regeneration for peripheral nerves(Jones & Redpath, 1998; Smith et al., 2022), there had been consideration of adult monkeys to test axon regeneration(Mertsch et al., 2018), cats, rats and monkeys have been shown to express GAP43 in motor neurons consistent with axon regeneration(Linda et al., 1992). Dog models have been proven to be very helpful to develop therapies for inherited retinal diseases, particularly for Leber’s congenital amaurosis(Petersen-Jones & Komaromy, 2015). Despite known progressive and inherited axonopathy, such dog models are yet to be utilized for optic nerve axon regeneration(Griffiths et al., 1985). Testing in large animals such as rabbits, dogs, pigs and monkeys, which have bundled axons projecting within pial septae similar to humans will be necessary to evaluate the potential of therapeutics molecules or their combinations. This includes stem cell therapeutics approaches. Evaluation of large animal models thus far have been scarce for optic nerve axon regeneration. As the strategies and molecules develop, they will be much needed prior to moving into human trials. Several factors including funding limits usage of large animal models. Consensus on 2–3 large animal models will be highly desirable for evaluation of optic nerve axon regeneration.

Warranted candidate patient population for initial stage evaluation of therapies.

Combat and civilian trauma present a spectrum of damage to optic nerve ranging from milder trauma to complete orbit damage (Bodanapally et al., 2014; Shah et al., 2013). The optic nerve traverses the orbit passing through the intraconal space. The nerve is sealed anteriorly by the globe and laterally by the myofascial cone. Trauma or a lesion that acutely increases the volume within the orbit can also increase the intra-orbital pressure. The acute increase in orbital pressure may lead to ischemic optic neuropathy that can cause permanent vision loss in 60–100 minutes(Nguyen et al., 2017). Glaucoma, optic neuritis and indirect trauma result in irreversible vision loss. Our medium-term goal is arriving at a clinically relevant workflow reducing all the axon regeneration treatment steps into three clinic visits (Fig. 1).

Figure 1. Schematic images superimposed on partial radiologic (CT) image of eye and head structures.

The structures in the eye damaged due to impact is shown.

Closed head trauma and exposure to blast shockwaves can lead to progressive and irreversible vision loss through mechanisms analogous to glaucomatous neuropathic disease, albeit in a much shorter time frame. This type of vision loss, known as traumatic optic neuropathy (TON) is an orphan condition closely associated with traumatic brain injury (TBI), and one in which effective therapy remains elusive. The impact of trauma is unpredictable and heterogeneous, requiring multiple models to capture the spectrum of damage. Although trauma has heterogeneous effects, there are common damages that must be addressed. In this perspective for the clinical consideration, we present patient prioritizing considerations. In addition to other damages, all traumatic events cause some degree of impact on the optic nerve, resulting in slow and progressive damage (common damage). In radiologic evaluations, the impact on the delicate optic nerve bundles is often but not always apparent(Ibanez et al., 2021). The optic nerve, optic canal, and optic chiasm are delicate structures that undergo subtle damages due to impact (see Figure 1)(Ibanez et al., 2021). The damage to the largest elongated structure of the optic nerve housing the axon bundles results in glaucomatous visual loss.

The indirect TON or lesions (see Fig. 2A, B) may result in proptosis, edema, retrobulbar stranding and compression but in many eyes the structure becomes completely normal (Fig. 2C). Towards designing the proposed animal experiments, we have considered these aspects and their future relevance for candidate patients. We have consulted extensively with our clinical colleagues in thinking about the problem of axonal regeneration and have kept their input in mind throughout our experimental design. Since 1985, a subset of glaucoma patients has been shown to undergo reversible improvement in visual acuity due to IOP lowering(Caprioli et al., 2016).

Figure 2. Anatomic features of potential axon regeneration therapy patients.

Globe measurements on computerized tomography (CT). Illustration of normal eye globe and optic nerve after visual loss. A. In right globe, distance (vertical dashed line) between cornea and interzygomatic line (horizontal dashed line) of 21 mm or less suggests normal findings. Protopsis is suggested when the distance is over 21 mm or in greater than 2-mm asymmetry between eyes are observed. In the left globe, posterior globe angle measurement (arrows, curved line) over 150º as indicated suggests normal findings. Less than 130º and 120º suggests tenting and critical tenting respectively. B. Orbital compartment compression due to Thyroid disease in a 50-year-old Caucasian female. Axial contrast-enhanced CT shows severe protopsis as well as tenting of the posterior globe (solid straight arrow) and stretching of the optic nerve (arrowhead). The orbital edema and retrobulbar stranding is evident in the picture in the area between arrowhead and arrow respectively C. The CT after medial wall decompression ( dashed arrow) and orbital access drainage. The radiographic images in B, C has been taken from the Bascom Palmer Eye Institute Oculoplastic services clinic.

There is a lack of reliable clinical parameters for accurately assessing the health of the intraconal space. Electrophysiological and afferent pupillary light reflex tests may offer insights into the intraconal space’s health, but these assessments are prone to confounding factors such as vitreous/lens opacification effects. We use non-invasive assessment methods (MRI, OCT) (Enriquez-Algeciras et al., 2011) and endpoint confirmations to identify animals with structurally intact intraconal space (Fig 2A). In humans, some eyes with damage may be categorized as functionally unsalvageable (without the need of enucleation/exenteration) and may be suitable for Phase I trials.

Multiple rodent and other TON models that retain intact intraconal structures can be adapted for future guidance towards clinical translation. We propose that our future priority treatment patients are those with normal intraconal structures(Ibanez et al., 2021). The two axon regeneration components are: 1) intrinsic promotion of axon growth and 2) reduction of negative environmental cues. The current paradigm suggests that intrinsic promotion is a better effector and deliverable through intravitreal space. Traumatic injuries alter the ON microenvironment differently. There is a great need to make functional assessments such as recovery in flash electroretinogram (ERG), pattern electroretinogram (PERG) amplitude and latency improvement in visual evoked potential (VEP) parameters.

Despite this goal, we do not expect all patients to be candidates for axon regenerative therapy. For axon therapy the patient population that we currently deem would be most appropriate should have following characteristics: 1) lack of substantial optic nerve head alteration, that is, no discernable excavation, 2) intact intraconal space, 3) less than 40 years of age, 4) any responsiveness to steroid is a plus, 5) any evidence of spontaneous reversal of visual acuity. These can be among patients of glaucoma (NTG, POAG, PACG, secondary glaucoma), TON patients and patients of genetic or developmental disorders suffering from optic neuropathies. It is also important to think through a paradigm that can accommodate most treatments in outpatient settings.

In devising the workflow, we have kept in mind how we can arrive at a flow that is feasible for adoption in clinic. As shown in Fig. 2, ocular trauma may result in proptosis. The measurements of proptosis in Computed tomography (CT) figures were developed in 1977(Hilal & Trokel, 1977). The displacement between cornea and interzygomatic line in CT images should be 21 mm or less and the angle across two planes in the posterior globe should be no less than 130 degrees (Fig. 2A)(Hilal & Trokel, 1977; Ibanez et al., 2021). The eyes that show the normal structures, are under ages of 45 with good record of static perimetry are candidates for evaluation of axon regeneration integrative therapy with the workflow and reagents (Fig. 3) that we expect to arrive from multiple preclinical studies.

Figure 3. The potential adoptable regeneration workflow from preclinical to clinical stage.

A three-step workflow adopted from preclinical to a eventual clinical paradigm. They are establishing conditionality in the eye of appropriate patients, promoting long distance axon regeneration (stages 1–3 with potential timeline) followed by visual function assessment with comprehensive electrophysiological measurements. The re-innervation strategies are multi-modal, which is indicated by grey hexagon.

In summary, we have come a long way to demonstrate that CNS axon regeneration is feasible from the historic standpoint that they would never be(Arnemann, 1786, 1787). Current best axon regeneration strategies are usually gene deletion and show robust regeneration only in young animals(Geoffroy et al., 2016), however, within the next few years multi-omics and advanced bioinformatics is likely to identify outpatient clinic compatible deliverable molecules and strategies. We can expect that advances within the next decade in this area may surpass cumulative advances made within the last two centuries. It is therefore time for basic discovery researchers, clinicians and policy makers to think through the patient population and characteristics that may help get some discernible outcomes now rather than when all entities and proven strategies are available. For such late onset and progressive diseases perhaps the consideration of patient population should be the ones where there is a good chance to see clear improvement rather than adhere to the tenet that only end stage most difficult and incurable patients are test first. Choosing a group of patients who are likely to improve need not compromise “do not harm” concept. We have presented some criteria across various optic neuropathy patient groups that should be part of this consideration.