Ethambutol (EMB), one of the first-line drugs used to treat tuberculosis, works by inhibiting the arabinosyl transferase of mycobacteria (Belanger et al., 1996). However, its use is associated with a well-recognized ocular complication, ethambutol-induced optic neuropathy (EON), which is characterized by blurring of the vision, dyschromatopsia, and central or cecocentral scotoma (Carr and Henkind, 1962). The ocular symptoms usually appear after 7 to 8 months of treatment with EMB, and the development of EON is both time- and dose-dependent (Fraunfelder et al., 2006; Lee et al., 2008). The incidence of EON is about 1% in patients prescribed an EMB dose at 15 mg/kg/d (Santaella and Fraunfelder, 2007; Yang et al., 2016). The risk factors for the occurrence of EON include older age, low body weight, and renal dysfunction (Chen et al., 2012). About 30 to 54% of the patients have varying degrees of vision recovery at 3 to 7 months after EMB discontinuation (Lee et al., 2008; Ezer et al., 2013); however, some patients have severe and permanent visual loss, even without the known risk factors. Therefore, other predisposing factors might exist for EON.

At present, the pathophysiology of EON still remains unclear. Several previous studies have indicated that EMB disrupts energy metabolism and the network structure of mitochondria by inducing severe vacuole formation and by decreasing membrane potential (Chung et al., 2009; Guillet et al., 2010). Dotti et al. (1998) first described the m.G11778A mitochondrial DNA (mtDNA) mutation in an EON patient, pointing to a relationship between mitochondrial mutation and the development of EON. Leber’s hereditary optic neuropathy (LHON) is caused by mtDNA mutations that disrupt complex I activity of the mitochondrial respiratory chain and ATP synthesis (Zhang et al., 2018). Autosomal dominant optic atrophy (ADOA) is another form of mitochondrial optic neuropathy that is caused by mutations of the OPA1 gene (Delettre et al., 2000). The OPA1 gene encodes a dynamin-related GTPase, which is localized at the inner membrane of mitochondria and plays a role in mitochondrial fusion (Kao et al., 2015).

Both LHON and ADOA have a common pathophysiological outcome, which is mitochondrial energy deficiency and retinal ganglion cell apoptosis (Kao et al., 2015; Zhang et al., 2018). The current literature reports 8 patients carrying LHON mtDNA mutations and 2 patients harboring OPA1 mutations who developed optic neuropathy during the use of EMB (Dotti et al., 1998; De Marinis, 2001; Hwang et al., 2003; Chowdhury et al., 2006; Ikeda et al., 2006; Guillet et al., 2010; Pradhan et al., 2010; Seo et al., 2010). Most of these studies are case reports, and the final clinical diagnosis is controversial for those patients. The relationship between the mitochondrial mutations and the clinical features of EON has not been fully studied.

In the present study, we investigated the presence of mitochondrial mutations (OPA1 and LHON-mtDNA) in a cohort of 47 patients with EON. We also studied the effects of these mutations on the clinical features of the patients by performing genetic analysis and identified disease-causing gene mutations in 22 patients.

The current study investigated the mitochondrial mutations in 47 unrelated patients with EON and described the clinical characteristics of the patients. We identified OPA1 mutations in 18 (38.3%) patients and LHON-mtDNA mutations in four (8.5%) patients. EON is a dose-dependent form of toxic neuropathy related to several other risk factors, such as renal dysfunction and older age (Santaella and Fraunfelder, 2007; Chen et al., 2012; Yang et al., 2016). In the current cohort, 91.5% of the patients with EON had taken low-dose EMB (≤ 15mg/kg/day), 95.7% of the patients had a normal renal function, and 89.4% of the patients were younger than 60 years; therefore, the risk of developing ocular toxicity was relatively low in our cohort. Our results suggested that mitochondrial genetic variations, and especially OPA1 mutations, are major predisposing factors for the occurrence of toxic optic neuropathy.

Mutations of the OPA1 gene are responsible for 50–70% of ADOA, (Cohn et al., 2007; Yu-Wai-Man et al., 2010) which is the most common form of inherited optic neuropathy. OPA1-related ADOA is usually a mild and slowly progressive disorder (Cohn et al., 2007). Typically, patients suffer an insidious, symmetrical, and progressive visual defect in their childhood; however, the severity of visual impairment is highly variable (Cohn et al., 2007). About 10–20% of mutation carriers are “asymptomatic,” as they have a normal visual acuity or only a subtle visual disturbance (Cohn et al., 2007; Yu-Wai-Man et al., 2010). In the current cohort, we identified different kinds of mutations of the OPA1 gene, but the type and location of these mutations were similar to those reported in typical patients with ADOA, (Cohn et al., 2007; Ferré et al., 2009; Yu-Wai-Man et al., 2010; Chen et al., 2014) except for the low frequency (13%, 2/15) of missense mutations. This frequency was only the half rate observed previously in patients with ADOA (approximately 27%) (Chen et al., 2014). Another previous study indicated that vision loss was usually more severe in patients with missense mutations than with null mutations (Yu-Wai-Man et al., 2010). Unlike the typical patients with ADOA, the OPA1 mutation carriers in the current study all experienced a subacute visual loss while taking EMB therapy, and none of them noticed a visual abnormality before the treatment; therefore, they might be “asymptomatic” cases. In addition, 71% of the OPA1 mutation carriers presented optic disc hyperemia, which has never been observed in typical patients with ADOA. By contrast, only 20% of those carriers showed temporal disc pallor, which is a prominent optic appearance in patients with ADOA (41–86%) (Cohn et al., 2007; Yu-Wai-Man et al., 2010; Chen et al., 2014). Four patients harboring the most common mutation c.2708_2711delTTAG had different fundus appearance. We speculate that the difference in the fundus performance may be due to the different course of vision loss of patients. In a previous study, Pradhan et al. described a 36-year old man who suffered a bilateral, painless visual loss during his anti-tuberculosis treatment that included EMB (Pradhan et al., 2010). This patient also had optic disc hyperemia and peripapillary hemorrhage. After a series of differential diagnoses, the researchers finally identified a nonsense p.R38X OPA1 mutation in this patient and inferred that the visual loss in that patient had been exacerbated following EMB therapy. We speculate that optic disc hyperemia is an acute response to the toxic effect of EMB; consequently, a concomitant effect of OPA1 mutations and EMB toxicity renders OPA1 mutations carriers prone to optic disc hyperemia. This might be the reason why the OPA1 mutations carriers presented more optic disc hyperemia than was observed in the patients without mutations, even though both groups had a similar disease time course. The mutation rate (38.3%) of OPA1 mutations in this EON cohort was higher than that (9.6 and 7.6%) reported in a group of Chinese patients with suspected hereditary optic neuropathy (Chen et al., 2014) and other Han Chinese population (Zhang et al., 2017). One reason might be the patients harboring OPA1 mutations are more sensitive to the toxicity of EMB and present more obvious visual defects as our mentioned above. Another reason might be due to the small number of patients in the current cohort.

In this cohort, we only identified four male patients carrying a LHON-mtDNA mutation, and this rate (8.5%) was much lower than the rate (38.3%) in patients harboring OPA1 mutations and the rate (33%) in Chinese patients with LHON or suspected with LHON (Ji et al., 2008; Yu et al., 2010). Hwang et al. (2003) screened a cohort of 24 patients with EON for LHON-mtDNA mutations but were unable to identify any LHON-mtDNA mutations. Therefore, the existence of LHON-mtDNA mutations might be uncommon in patients with EON. To date, twelve patients (including our four patients) have been reported to develop optic neuropathy while taking EMB (Table 4; Dotti et al., 1998; De Marinis, 2001; Chowdhury et al., 2006; Ikeda et al., 2006; Seo et al., 2010). Ten of these patients carried mutation m.11778G > A, which is the most common primary mutation of LHON (Chen et al., 2014). In the present study, we described two patients carrying the m.14484T > C mutation (the second commonest primary mutation of LHON) who developed LHON while taking EMB (Table 4). Unlike the eight previously described cases, the four patients in our cohort had a younger onset age (48 years vs. 21.5 years). In addition, these four patients all had a family history, and three of them showed optic disc hyperemia and RNFL pseudoedema, which is a typical fundus appearance in the acute stage of LHON (Chen et al., 2014). Not all individuals who carry LHON-mtDNA mutations will develop visual symptoms, as the occurrence of LHON usually needs other risk factors, like heavy smoking and alcohol consumption (Sadun et al., 2003). EMB might have been a trigger or an epigenetic factor for the manifestation of LHON in these four patients.

The toxicity of EMB is related to its zinc-chelating effect and its metabolites (Chung et al., 2009). At present, the exact mechanism for the toxic neuropathy induced by EMB remains unclear, but increasing evidence indicates a relationship with mitochondrial dysfunction. In an early study, Heng et al. found that EMB was specifically toxic to retinal ganglion cells and that it caused ganglion cell degeneration by a glutamate excitotoxic pathway (Heng et al., 1999). LHON-mtDNA and OPA1 mutations also cause damage to the small-caliber papillomacular bundle axons (Barboni et al., 2010) therefore, the clinical features of EON, LHON, and ADOA partially overlap. All three conditions show color vision abnormality and visual field defects. In the current cohort, most patients presented central or cecocentral scotoma, which could be observed in both EON and LHON or ADOA, whereas only three patients in group 2 showed temporal hemianopia, which is a typical visual field defect of EMB-related optic chiasmopathy (Jayanetti et al., 2017). In this cohort, the majority of patients presented with red/green color vision impairment or total color blindness, while only two patients carrying OPA1 mutations displayed yellow/blue color vision impairment, which is a typical color defect of ADOA (Cohn et al., 2007). Fundus hemorrhage is not common in LHON or ADOA, and three of the four patients with fundus hemorrhage were in group 2, indicating it to be one of the EON clinical features. Our results showed that the mean age was significantly younger in the patients with mitochondrial mutations (27.5 years; group 1) than in the patients without mutations (48 years; group 2), further demonstrating that mitochondrial mutation was an important risk factor for the occurrence of EON, especially in young tuberculosis patients. Patients carrying mitochondrial mutations (OPA1 or mtDNA) may be more vulnerable to the toxicity of EMB. Of the seven patients with a family history, six patients were mutation carriers; therefore, physicians should carefully question patients with tuberculosis or their family members about ophthalmic problems before EMB is prescribed.

Consistent with the observation by Lee et al. (2008) we found that the visual prognosis was related to the initial fundus appearance of the patients with EON. Up to 79.8% of the patients in our cohort showed normal optic disc appearance or optic disc hyperemia, which suggested they were in the early stage of EON. This might be a reason why our patients had a higher visual recovery rate (75.7%) than the rates described in other EON studies (23.1–47%) (Lee et al., 2008; Ezer et al., 2013). Another reason for this high rate might be due to that the 94.6% VA improvement was reported by patients subjectively; this is one limitation of the current study. Nevertheless, we still observed that the visual outcome was better in the patients in group 2 than in group 1, suggesting that the mitochondrial mutations were a critical factor affecting visual prognosis of patients with EON. Of the eight patients carrying mutation m.11778G > A, only one patient who harbored a heteroplasmic mutation showed VA improvement, (De Marinis, 2001) whereas the other patients all experienced VA decreases or no change during their follow-up (Table 4). Another limitation of the current study is that we did not take into account the ocular toxicity of other anti-tuberculosis drugs, such as isoniazid. The third limitation is that since the study was retrospective in nature, the examinations were not standardized.

In conclusion, our results indicated that carriers with OPA1 mutations might be more vulnerable for the toxicity of EMB to develop EON, whereas the exact effect of these OPA1 mutations has not been confirmed in future functional assays. Although mitochondrial mutation screening is not possible in all patients prior to anti-tuberculosis medication, genetic analysis should be strongly recommended for patients with a family history of optic neuropathy.

The data presented in the study are deposited in the NCBI GenBank, accession numbers NM_015560 and AC_000021.2.

The studies involving human participants were reviewed and approved by Medical Ethics Committee of Beijing Tongren Hospital. The patients/participants provided their written informed consent to participate in this study.

X-HZ participated in the data collection, data analysis, and manuscript preparation. YX, KX, and KC participated in the data collection and analysis. Q-GX contributed in the study design and data collection. Z-BJ, YL, and S-HW participated in the study design and the manuscript revision. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.