In 2003, Antonellis et al. confirmed that GARS gene was the pathogenic gene of CMT2D/HMN5A for the first time . Up to now, only 20 variants of GARS gene have been reported to be associated with CMT2D/HMN5A [5, 12, 13]. Classic phenotype of CMT2D/HMN5A is weakness and atrophy of upper extremities, especially the thenar eminence and the first dorsal interosseous muscle groups. The main distinguishing characteristic of the two disorders is sensory deficits in a gloving-stock pattern in patients with CMT2D. Sensory loss may vary between family members and may be normal in some. Patient 2 and Patient 3 in our study presented with classic phenotype of HMN5A, while the Patient 1 who presented with prominent superficial sensory loss in distal extremities was diagnosed as CMT2D. Upper limbs were the most frequent onset sites, and lower limbs onset (in Patient 1) could also be seen . Compared with HMN5A, weakness of lower limbs in patients with CMT2D tended to be more prominent . Most patients were of adolescent onset, while infant onset was also reported and tended to be more severe [14-16]. Severity of phenotype could be varying even in one family [14, 16].
Motor peripheral neuropathy in the first three patients were confirmed with electrophysiological studies. NCS of the Patient 1 and Patient 2 showed distinctively lower CMAP in median nerve than in ulnar nerve, and this imbalance involvement argued against a primary length-dependent distal axonopathy and was more in favor of a motor neuronopathy . Sensory loss was the characteristic of CMT2D, and in other studies, sensory nerve action potential amplitude was decreased or diminished in CMT2D patients [6, 12]. In our study, normal sensory nerve conduction was revealed in the Patient 1, however, sural nerve biopsy showed uncompaction of the myelin sheath, axonal degeneration and necrosis of peripheral nerve myelin sheath and axon, which indicated the diagnosis of CMT2D. Not only large myelinated fibers, but also small- to middle-size fibers may occur in CMT2D [9, 16].
A Korean patient with c.794C>T (p.S265F) was diagnosed with HMN5A , however our patient with c.794C>T (p.S265F) was diagnosed with CMT2D, because sensory nerve damage was revealed in sural nerve biopsy . Previous studies showed that the mutation c.374A>C (p.E125G) was associated with both CMT2D and HMN5A [9, 11]. Coexistence of CMT2D and HMN5A phenotypes caused by same variant remained unknown etiology. Recently, the GARS c.794C>A (p.S265Y) variant was reported in a Malian family with CMT2D , and c.373G>A (p.E125K) was associated to an infant patient with failure to thrive and severe muscle weakness . These two amino acid sites might be critical to GARS. Previously, c.1000A>T (p.I334F) variant was reported in a patient with HMN5A , and in our study, we found Patient 3, a 21-year-old female also carried c.1000A>T variant. Unlike c.794C>T (p.S265F) and c.374A>G (p.E125G) which occupied in the core catalytic domain of GARS, c.1000A>T (p.I334F) was located in the anticodon binding domain.
An unreported GARS gene variant c.781T>G (p.Y261D) was found in Patient 4. This variant was predicted as pathogenic by using different prediction tools and located in the core catalytic domain of GARS, the same as c.794C>T (p.S265F) and c.374A>G (p.E125G). However, no genetic cosegregation was found in this family and no pure sensory neuropathy associated with GARS gene variant was reported yet. According to the guidelines of ACMG, the pathogenicity of this mutation was considered indeterminate. With subacute onset of sensory neuropathy, long-term positive anti-Hu antibody, and no any other possible reason of sensory neuropathy, the patient was considered anti-Hu antibody neuropathy [19-22]. However, malignancy was not confirmed in seven-year of follow-up. Asymptomatic carriers with pathogenic variant were found in different studies [12, 23], so we cannot exclude the possibility that descendants of Patient 4 may show symptoms later. Continuous follow-up should be conducted on Patient 4 and her family members.
An American study that screened 100 patients diagnosed with dSMA, HMN, or motor axonal CMT for variants in GARS found 3 variants , while a Taiwan study reported two heterozygous variants found from 340 CMT patients (0.6%) . Another American study showed very low frequency of GARS gene variant, only 0.4% of 3216 CMT patients . In Asia, one Japanese study reported that one disease-causing mutation in one patient was found from 89 patients with axonal CMT (1.1%).  In this study, we found three pathogenic variants of GARS (1.46%, 3/206), including c.794C>T (p.S265F), c.374A>G (p.E125G) and c.1000A>T (p.I334F) in 206 unrelated Chinese Han patients with a clinical diagnosis of IPN (Table 3). So far, as GARS variants are rare, few studies focus on the mutation frequency of GARS. The issue of the above studies on the mutation frequency of GARS is that the inclusion criteria of patients were inconsistent, which makes horizontal comparison difficult among people from different countries and regions. Studies with larger sample size, especially multicenter studies with unified inclusion criteria of patients are required to assess the mutation frequency of GARS among different groups of people.
In general, GARS variant is a rare cause of CMT and the phenotype tends to be CMT2D or HMN5A. Clinical characteristics, NCS, and even sural nerve biopsy and skin biopsy are essential to distinguish them. As the advance of next-generation sequencing technologies including disease-specific gene panels, whole-exome sequencing and whole-genome sequencing etc, novel likely pathogenic genes and variants would be found increasingly . In this study, the results suggest the importance of comprehensive assessment by using clinical phenotype, ancillary tests and genetic evidence to evaluate the pathogenicity of genetic variants in patients suspected as IPN.