Glutamate has a neurotoxic effect when it accumulates at the synapses (Gazulla and Cavero-Nagore, 2006). In patients with ALS (both spinal cord and motor cortex involvement), and in the superoxide dismutase 1 (SOD1) transgenic mouse model, a decrease in glutamate receptors (GluRs) was found in astrocytes. This decrease induces extracellular glutamate accumulation, which causes overstimulation of GluRs and neuronal death via excitotoxicity (Lin et al., 1998; Trotti et al., 1999; Barbeito et al., 2004; Pratt et al., 2012). In ALS patients, a decrease in glutamate transporters is mainly due to an alteration in messenger RNA (Lin et al., 1998; Honig et al., 2000).

Mitochondrial function disturbances, such as fragmentation and aggregation, are frequently found in ALS patients (Vielhaber et al., 2000; Chung and Suh, 2002; Krasnianski et al., 2005; Echaniz-Laguna et al., 2006; Crugnola et al., 2010; Cozzolino and Carrì, 2012). Increased crests, swelling, and fragmentation have been observed in the mitochondria of the spinal motor neurons and proximal axons of skeletal muscle in ALS-related tissues (Chung and Suh, 2002; Echaniz-Laguna et al., 2002; Boillée et al., 2006a). The increase in the misfolded SOD1 enzyme in the mitochondria of the spinal cord of mice is considered to be the main cause of mitochondrial dysfunction. In addition, aggregates of the enzyme SOD1 may also interact with the apoptosis regulator protein Bcl-2, inducing an apoptotic cascade and contributing to the deterioration of neurons and neuro-muscular degeneration (Boillée et al., 2006a).

Axonal transport (retrograde and anterograde) is impaired in ALS patients and in mutant SOD1 mice, as evidenced by the accumulation of altered structures, such as mitochondria, neurofilaments, and autophagosomes, in the spinal motor neuron axons (Ikenaka et al., 2012; Zarei et al., 2015). Mutations in the dynein genes have been seen in models of ALS mice. Dyneins are responsible for the transport of mitochondria and autophagosomes, causing both to accumulate in the axon (Ikenaka et al., 2012). Autophagosomes are necessary for the elimination of altered mitochondria and dilated endoplasmic reticules, which accumulate in the axons of motor neurons and cause them to malfunction (Zarei et al., 2015). All of the above suggest that an alteration in axonal transport could be fundamental for the development of ALS.

Increased free radical and oxidative damage has been found in biopsies from ALS patients, as well as in cerebrospinal fluid, serum, and urine samples (Simpson et al., 2004; Shin Hee and Lee Keun, 2013). This oxidative damage also affects RNA, which has been shown in both human central nervous system (CNS) biopsies and in mouse ALS models of SOD1 (Chang et al., 2008). The enzyme SOD1 is an important anti-oxidant. The alterations in the redox reactions are one of the first theories of how mutations in SOD1 can cause cytotoxicity (Carrì et al., 2015). In addition, it has been observed in the motor cortex, using positron emission tomography (PET), that increased oxidative stress is related to the severity of the disease in ALS patients (Tsujikawa et al., 2015).

It has been shown that in ALS, abnormal protein accumulations are produced that aid in the pathogenesis of the disease (Blokhuis et al., 2013). The ubiquitin–proteasome (UP) system, which repairs and removes proteins, plays an important role in ALS, with ubiquitin-reactive inclusions being characteristic of this pathology (Han-Xiang et al., 2011; Pratt et al., 2012). Among these, the inclusions of proteins TDP-43 and p62 are indicative of this pathology (Al-Sarraj et al., 2011; Williams et al., 2016). Some ALS patients present numerous inclusions positive for p62 but negative for TDP-43 in the hippocampus and cerebellum (Al-Sarraj et al., 2011). These p62-positive and TDP-43-negative inclusions have also been observed in other CNS areas including the retina (Brettchneider et al., 2013; Fawzi et al., 2014).

Neuroinflammation occurs in many neurodegenerative diseases, such as Parkinson’s, Alzheimer’s, and ALS, resulting in the activation of astroglial and microglial cells (Ramírez et al., 2017). Specifically, in ALS, it has been observed that reactive glia (microglia and astrocytes) can influence the damage and subsequent death of motor neurons (Vargas and Johnson, 2010; Philips and Robberecht, 2011; Liao et al., 2012). In the ALS, the presence of mutant proteins (SOD1 and TDP-43), oxidative stress, mitochondrial damage, etc., produces continuous damage, which can trigger a chronic activation of glial cells and, therefore, a sustained inflammatory process that could exacerbate neuronal damage (Philips and Robberecht, 2011). It has been observed that the mutant SOD1 protein can have a toxic function on motor neurons only when the activated microglia is present through a mechanism involving a toll receptor (CD14-TLR) that causes an increase in free radicals (Zhao et al., 2010). Astrocyte activation (Vargas and Johnson, 2010), microglial activation (Alexianu et al., 2001), and lymphocyte appearance (Engelhardt et al., 1993) have been found in animal models of ALS (mutants of SOD1) (Alexianu et al., 2001; Boillée et al., 2006b) and in ALS patients (Kushner et al., 1991; Kawamata et al., 1992; Nagy et al., 1994; Schiffer et al., 1996; Turner et al., 2004). In both cases, the mutant SOD1 causes the microglia to increase the expression of pro-inflammatory cytokines, such as IL-1β and TNF-α (Weydt et al., 2004), and inflammatory mediators, such as cyclooxygenase 2 (COX-2) (Almer et al., 2001) and nitric oxide (NO). The NO released by the microglia could induce apoptosis in motor neurons through the activation of Fas via (Raoul et al., 2002).

Mutant protein TDP-43 can produce microglial activation through the surface receptor CD14, which stimulates the NF-κB and AP-1 pathways and the inflammasome. This can cause a neurotoxic cascade leading to motor neuron death (Zhao et al., 2015). In the ALS SOD1 model, the microglia isolated at the onset of the disease has an M2 or anti-inflammatory phenotype; however, the microglia isolated at the end of the disease has a neurotoxic M1 phenotype. This demonstrates a dual role of microglial cells during the disease process in this ALS model (Liao et al., 2012).

Astrogliosis has been observed in ALS models with mutations in C9orf72, FUS, SOD1, and TARDBP genes (Wong et al., 1995; Liu et al., 2016; Sharma et al., 2016). In the last model, the death of the motor neurons could be due, in part, to a loss of the function of the TDP-43 protein in the astrocytes (Yang et al., 2014). In ALS SOD1 models, astrocytes can release ATP, causing the activation of microglial cells via purinergic receptors (P2X7) (Gandelman et al., 2010). They can also release transforming growth factor-β1 (TGF-β1), which can induce microglial inactivation, thereby eliminating the possible beneficial effects of the microglia and, thus, accelerating the progression of the pathology (Endo et al., 2015). Moreover, the microglia in ALS can affect astrocytes by favoring the appearance of a neurotoxic subtype (Liddelow et al., 2017).

Considering all of the above, both the microglia and reactive astrocytes can affect neural function in ALS by playing an important role in the progression of the disease.

This gene is located at the locus 9p21 of chromosome 9. It has been proposed that the C9orf72 mutation can decrease the C9orf72 protein, thus causing less endocytosis, which is necessary for autophagy (Mathis et al., 2019). Moreover, the massive accumulation of the expanded hexanucleotide GGGGCC could be neurotoxic and sequester proteins that bind to RNA, thereby causing the disruption of the machinery that processes RNA (Mathis et al., 2019). This mutation could also lead to increased vulnerability to excitotoxicity due to increased calcium permeability mediated by AMPA type receptors (Selvaraj et al., 2018). The form of inheritance is autosomal dominant, although some carriers do not develop the disease, so it has an incomplete penetrance (Renton et al., 2011). This variant represents the most frequent cause of SALS (7%) and FALS (30–40%) (Majounie et al., 2012). The manifestations at the onset of the disease are typical of ALS and have a bulbar onset. This variant is often associated with an earlier age of onset, a faster clinical course, and shorter survival (Schymick and Traynor, 2010).

The FUS gene, located on the short arm of chromosome 16 (16p11.2), encodes a protein called sarcoma fusion protein (FUS) (Hübers et al., 2015; Scekic-Zahirovic et al., 2017). In ALS, mutations in the FUS gene have been observed, most of which can produce changes in amino acids in the region of the protein related to DNA binding and mRNA processing. These mutations may interfere with the importation of FUS into the cell nucleus, which will cause an accumulation of FUS in the cytoplasm. This has been observed particularly in the nerve cells that control muscle movement (Scekic-Zahirovic et al., 2017). This mutation occurs in 3–5% of FALS and in 1% of SALS cases (Deng et al., 2010; Mackenzie et al., 2010). In addition, patients with FUS mutations tend to develop the disease earlier and have a shorter life expectancy than those observed with mutations in other genes. Patients with ALS and FUS mutations may also develop frontotemporal dementia (FTD) (Kwiatkowski et al., 2009; Vance et al., 2009).

The optineurin gene (OPTN) is located on chromosome 10 and encodes a protein called optineurin, which is a multifunctional ubiquitin-binding phosphoprotein found in the cytoplasm. Changes of this protein can cause alterations in intracellular traffic and lead to inclusions in ALS. In addition, this protein is also involved in the signaling of the tumor necrosis factor α/NF-κB pathway (Zhu et al., 2007) and the mGluR (Ying et al., 2010). In 2010, Maruyama et al. (2010) and Schymick and Traynor (2010) found mutations in the OPTN gene in ALS patients. These mutations are present in 1.2% of patients with FALS (Del Bo et al., 2011). ALS patients with OPTN mutations presented typical spinal-onset disease (Weishaupt et al., 2013). Mutations in this gene were previously shown to be involved in primary open-angle glaucoma (Maruyama et al., 2010). This gene has also been linked to normotensional glaucoma in ALS patients (Weishaupt et al., 2013). In cells from patients with SALS and FALS, optineurin can be placed in inclusion bodies with TDP-43 (Maruyama et al., 2010), FUS (Ito et al., 2011), and SOD1 (Maruyama et al., 2010). Both retinal ganglion cells (RGCs) and motor neurons share common susceptibility factors (Weishaupt et al., 2013).

The SOD1 gene is located on the long arm of chromosome 21 (21q22.11). This gene encodes a cytosolic enzyme called copper zinc (Cu/Zn) SOD1 (Gatchel and Zoghbi, 2005; Valentine et al., 2005) that plays a very important role in the elimination of superoxide radicals, thereby protecting against free radicals (Azadmanesh and Borgstahl, 2018). Around 2.5–23% of patients with FALS and 0.44–7% of those with SALS have mutations in SOD1 (Andersen, 2006; Van Es et al., 2010). Most of the inherited forms of SOD1 gene mutations are dominant (Hayward et al., 1998; Marucci et al., 2007). Mutations in the SOD1 enzyme can induce configurational changes in the SOD1 protein leading to motor neuron toxicity (Mathis et al., 2019). ALS patients who present mutations in SOD1 show an earlier time of onset and a longer duration of disease. In addition, they usually do not present cognitive disorders, and their motor symptoms usually begin in the lower extremities (Mathis et al., 2019).

The Tar DNA binding protein (TARDBP) gene is located on the short arm of chromosome 1 (1p36.22) and encodes the TDP-43 protein. More than 40 mutations have been identified in the TARDBP gene and result in 6.5% of FALS cases. The form of inheritance is dominant (Kühnlein et al., 2008; Van Deerlin et al., 2008) and represents 0–5% of SALS cases (Kühnlein et al., 2008; Rutherford et al., 2008; Sreedharan et al., 2008). Most of the mutations cause changes in the amino acids in the TDP-43 protein and affect the region of the protein involved in the processing of RNA (Neumann et al., 2006; Da Cruz and Cleveland, 2011). Cytoplasmic inclusions of ubiquitin-reactive hyperphosphorylated TDP-43 proteins have been found in tissues of patients with FTD (Arai et al., 2006; Neumann et al., 2006). These inclusions have also been seen in the glial tissue and neurons of patients with SALS (Maekawa et al., 2009). These inclusions can often be placed with p62 and ubiquitin but are not found in FALS with FUS mutation (Vance et al., 2009) or SOD1 mutation (Mackenzie et al., 2007). Some patients with FALS and TARDBP mutation may also develop FTD (Arai et al., 2006; Neumann et al., 2006).

Few studies have analyzed the VEPs in ALS patients (Table 2 and Figure 2). It was not until 1986 that alterations in VEPs were found in patients with ALS, with abnormal relative difference in latencies between each eye. In this study of 32 patients analyzed, only four had VEP abnormalities (increased latency time); however, these alterations were mild (Matheson et al., 1986). In other studies, while wave latency and amplitudes were within normal limits in all ALS patients (Ghezzi et al., 1989; Palma et al., 1993), somatosensory evoked potentials were abnormally delayed (N9–N13 and N13–N19 latencies), but no correlation was found among these abnormalities and the duration and severity of the disease (Ghezzi et al., 1989).

In addition, in ALS patients the P1 component was also found to be absent, and the P300 component was delayed and attenuated (Münte et al., 1998). In a study of six ALS patients, only one patient with a 9-month disease evolution had abnormal P100 bilateral extension VEPs and a significant interocular difference of P100. In addition, none of the patients presented alterations in their electroretinogram (González Díaz et al., 2004). Subsequently, electrophysiological studies of ALS patients showed new evidence of cortical participation involving visual areas, with alterations in the early sensory components of VEPs (Münte et al., 1998).

While one study revealed impaired contrast sensitivity during an eye examination in two ALS patients with C9orf72 mutation (Fawzi et al., 2014), another study concluded that this function is not affected in ALS patients (Volpe et al., 2015; Table 2 and Figure 2).

Visual acuity in ALS patients is controversial. While a study found a lower visual acuity in ALS patients in both high-contrast and low-contrast (2.5% and 1.5%) visual acuity with Sloan charts (Moss et al., 2012), in others, the visual acuity exam revealed no differences in monocular high-contrast visual acuity (Volpe et al., 2015; Moss et al., 2016; Rojas et al., 2019) or low-contrast visual acuity (Moss et al., 2016).

Only two studies describe VF in early ALS spinal onset patients. With a decrease of mean sensitivity and an increase of the square of loss variance (Liu et al., 2018), ALS patients presented a worse reliability index (fixation losses, false positives, and negatives) due to motor difficulties. For this reason, the authors suggested that VF is not a suitable test to assess ALS patients (Rojas et al., 2019; Table 2 and Figure 2).

Few studies have focused on the histopathology of retinal tissue in both ALS patients and animal models of mammals with this condition. Histopathological studies in ALS patients demonstrated intraretinal protein inclusions. The first histopathological analysis of the retinas of patients with ALS was performed in 2014 on a patient with the C9orf72 mutation. In this study, the authors found p62-positive and pTDP43-negative intracytoplasmic perinuclear inclusions in the inner nuclear layer (INL). These deposits were similar to those observed in the dentate gyrus of ALS patients with the C9orf72 mutation. The p62-positive staining was colocalized for both the poly-(GA)n dipeptide repeat and ubiquitin in the retina, which is similar to the perinuclear inclusions located in the brains of patients with this mutation. The p62-positive inclusions found were likely located in specific cone bipolar cells and within the amacrine and horizontal cells, as they were also marked with GLT-1 and recoverin. The authors suggested that these deposits may be related to the affectation of contrast sensitivity (Fawzi et al., 2014). In addition, in other ALS patients with C9orf72 mutations, specific p62 inclusions were observed in the retinal ganglion cell layer (GCL) in a far smaller proportion than in INL (94.9% in INL vs. 5.1% in GCL). Numerous positive ubiquitin2 + aggregates were also observed in a mutant UBQLN2 transgenic mice experimental model, mainly in the INL, with fewer in the outer plexiform layer (OPL) and some in the GCL (Volpe et al., 2015). The accumulation of ubiquitin2 aggregates in the layers of the retina with more synapses is related to the accumulation of these aggregates in the dendritic spines of the hippocampus. The location of the aggregates at the synapses, and their spines, may be associated with the dementia observed in this experimental model of ALS. In addition, few ubiquitin2 positive aggregates were detected in the subretinal space of the retinal pigment epithelium, which sometimes elevated these levels in the same way as druses (Volpe et al., 2015). Furthermore, lipofuscin deposits sometimes related to subretinal drusen-like aggregates were found in progranulin-deficient FTD patients (Ward et al., 2017). Retinal thinning in these patients was detected by OCT before symptoms, suggesting that the eye is affected in progranulin-deficient FTD disease (Ward et al., 2014).

As noted previously, microglial and astroglial cell activation occurs in ALS. However, to our knowledge, there are only two works that analyzed the glial cells of the retina in relation to ALS. hSOD1 + vacuoles located in the dendrites of excitatory retinal neurons were observed in a mouse model of ALS SOD1 (SOD1^G93A), mainly in the inner plexiform layer (IPL) and rarely in the GCL and INL. However, there were no signs of activation of either the astroglia or the microglia of the retina compared with those of wild-type mice (Ringer et al., 2017). In contrast, microglial activation was demonstrated in a mouse model of ALS [devoid of ran-binding protein2 (Ranbp2)]. Ranbp2 is a protein involved in nucleo-cytoplasmic transport whose regulation is impaired in both SALS and FALS (Ferreira, 2019). In ALS mice without this protein (with respect to wild-type controls), there was an increase in the number of CD45 +, CD11b +, and F4/80 + microglial cells surrounding the RGCs, in addition to a notable increase in amoeboid forms. An increase of metalloproteinase 28, which is an immune regulator, was also observed in the RGCs, suggesting that Ranbp2 may be involved in the signaling between the microglia and the RGCs in the immune response of ALS disease (Cho et al., 2019; Figure 2).

Although the first study performed in 75 ALS patients with Cirrus OCT showed no changes in the peripapillary and macular areas (Roth et al., 2013), further studies have shown that there are changes in the retina of ALS patients (Ringelstein et al., 2014; Volpe et al., 2015; Hübers et al., 2016; Simonett et al., 2016; Mukherjee et al., 2017; Abdelhak et al., 2018; Liu et al., 2018; Rohani et al., 2018; Rojas et al., 2019; Table 3).

In the retina of ALS patients, high-resolution SD-OCT revealed reduction in the mean total macular thickness, in the peripapillary retinal nerve fiber layer (pRNFL), the INL (Ringelstein et al., 2014; Hübers et al., 2016), and the outer nuclear layer (ONL) (Abdelhak et al., 2018), suggesting neurodegeneration of the retina in ALS patients.

Volpe et al. studied whether clinical and histopathological findings were present in the eyes of ALS patients and explored their correlation with an animal model (ALS/dementia transgenic mice with dysfunctional ubiquilin2, UBQLN2P497H) (Volpe et al., 2015). Using SD-OCT, the authors observed that (i) in ALS patients compared with the controls, there was a decrease in total macular volume; (ii) 37.5% of ALS patients showed an average pRNFL below the first percentile, and temporal and papillomacular bundle were more affected; and (iii) in ALS patients, the total macular thickness and pRNFL thickness correlated inversely with the time of evolution of ALS (Volpe et al., 2015).

Recently, in a study performed in early ALS patients with spinal onset and without ocular diseases, a significant macular thickness increase was found in the temporal and inferior areas of the inner macular ring in comparison with that in a healthy control, suggesting that this retinal thickening in early ALS patients could be due a microglial activation in the neuroinflammatory process (Ringelstein et al., 2014; Rojas et al., 2019). In contrast, in another study in ALS patients, only the macular retinal nerve fiber layer (mRNFL) showed a significant thickness decrease by OCT, which correlated positively with pulmonary function tests (Simonett et al., 2016). However, neither total macular thickness nor macular thickness showed changes (Simonett et al., 2016).

Only one study exists that detected a thinning in the ONL (Abdelhak et al., 2018), suggesting a possible impact on the photoreceptors and linking this finding with this subclinical visual acuity impairment (Moss et al., 2012; Abdelhak et al., 2018). Using SD-OCT, this work analyzed the retinal vessels, finding that in ALS patients compared with the control, the outer wall thickness of the retinal vessels was higher than in the control group (Abdelhak et al., 2018). Similar microvascular alterations in the brain and spinal cord of ALS model mice were found to precede the degeneration of motor neurons (Zhong et al., 2008).

In ALS patients, an analysis of the relationship of clinical features and retinal changes using SD-OCT and diffusion tensor imaging (DTI) found no significant correlation between clinical features and retinal thickness; however, there was a direct correlation between retinal thickness and fractional anisotropy of the corticospinal tract (Hübers et al., 2016). On the basis of these observations, it was suggested that retinal changes could be related to damage of white matter in the corticospinal tract and may be a possible biomarker in ALS (Hübers et al., 2016).

Although several papers have demonstrated the pRNFL decrease in ALS patients without ocular pathology (Mukherjee et al., 2017; Rohani et al., 2018; Rojas et al., 2019), there is controversy about its correlation with ALSFRS-R values. While one study did not demonstrate a correlation between RNFL thickness and the ALSFRS-R score and their progression rates (Mukherjee et al., 2017), other authors found a correlation with some OCT parameters (Abdelhak et al., 2018; Liu et al., 2018; Rohani et al., 2018; Rojas et al., 2019; Salobrar-García et al., 2019). In a study of the pRNFL thickness in four quadrants, the average pRNFL thickness showed a significant positive correlation with the ALSFRS-R score (Rohani et al., 2018), and the pRNFL thickness in the inferior sector was negative (Rojas et al., 2019). When the analysis was more detailed, dividing the papilla into 12 hourly sectors, sectors H5 and H6 had a positive direct significant correlation, and H8 had an inverse significant correlation of pRNFL with the ALSFRS-R values (Rojas et al., 2019). In addition, the entire retinal thickness correlated negatively with the ALSFRS-R score (Abdelhak et al., 2018).

Follow-up OCT studies of ALS patients are very rare. In the only one performed of early ALS patients with spinal onset (basal patients) who were examined 6 months after the basal scan (follow-up patients), the SD-OCT follow-up analysis of ALS patients showed a significant macular thickness decrease in the inferior areas of the inner and outer macular ring and a significant pRNFL thickness decrease in the superior and inferior quadrants, compared with the baseline (Rojas et al., 2019).

Asymmetry is a typical hallmark of ALS, as mentioned above. In the left eye (LE) of ALS patients, after adjustment for multiplicity, there was a significant decrease in pRNFL thickness in the nasal quadrant compared with the corresponding quadrant thickness in the right eye (RE) (Rohani et al., 2018). In the follow-up study mentioned above (Rojas et al., 2019), there were significant interocular asymmetries in some areas between the LE, which were always thinner, and the RE were observed. In the baseline ALS group, in the LE, the inferior-nasal quadrant of the macular ganglion cell complex (GCC), and the H7 and H9 hourly-sectors of the pRNFL were significantly thinner than those in the RE, while in the follow-up ALS group, both the supero-nasal quadrant of macular GCC and the temporal quadrant, and the H8 and H9 hourly sectors of the pRNFL, were significantly thinner in the LE than those in the RE. Therefore, the asymmetric participation of the CNS in this disease is not exclusive to the motor system (Rohani et al., 2018).

Thinning of the inner retina was also observed by SD-OCT in other neurodegenerative diseases, including Alzheimer’s disease and Parkinson’s disease, in which both RNFL and GCL thinning were detected (Moreno-Ramos et al., 2013; Coppola et al., 2015; Salobrar-García et al., 2015b; Garcia-Martin et al., 2016). However, in FTD, OCT identified ONL thinning, specifically in the ellipsoid zone, and this thinning of the outer retina correlated with cognitive changes (Kim et al., 2017). Moreover, the relationship between ONL thinning and FTD was especially apparent in the subgroup of patients considered to have “likely tauopathy” based on their symptoms or their genetics (Kim et al., 2017).

The differences observed in retinal thickness measurements using OCT may be due to the following: (i) the different disease stages of ALS patients included in the studies; (ii) the heterogeneous nature of this pathology (Hübers et al., 2016); and (iii) the small number of participants in these studies, since most ALS patients do not manifest visual problems.

Many authors observed changes in retinal thickness, as shown in this section. Therefore, retinal changes constitute a biomarker of neurodegeneration and progression of ALS disease (Hübers et al., 2016; Rohani et al., 2018), and OCT analysis is a useful tool for the study of this pathology (Rojas et al., 2019).