We searched the database at the First Affiliated Hospital of Sun Yat-sen University (Guangdong, China) between 2014 and 2020 and identified 12 patients diagnosed with anti-GQ1b antibody syndrome. This retrospective research complies with the principles of the Declaration of Helsinki and received approval from the local clinical trial committee of the First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China. All the patient information was de-identified so that the identity of the patients may not be ascertained in any way.

In addition to positive anti-GQ1b antibody, the diagnosis of the clinical entities within the anti-GQ1b antibody syndrome were based on the criteria established by Wakerley et al. [6]. The features of GBS spectrum disorders were mainly characterized by a symmetric pattern of limb and/or motor cranial-nerve weakness. Specifically, the criteria for MFS included relatively symmetric ophthalmoplegia, ataxia and areflexia/hyporeflexia in the course of diseases, as well as absence of limb weakness or hypersomnolence. The criteria for BBE included hypersomnolence, ophthalmoplegia and ataxia, with a normal limb strength. In some cases, prominent limb weakness as well as ophthalmoplegia and ataxia were considered as overlapping MFS and GBS (MFS/GBS), or overlapping BBE and GBS (BBE/GBS) when hypersomnolence were accompanied. On entry to the study, patients did not have toxic exposure, malnutrition or vitamin B deficiency, or mitochondrial diseases. Moreover, patients were classified into the optic neuropathy group if the best-corrected visual acuity (BCVA) on the Standard Logarithmic visual acuity chart was 0.6 or worse. The use of BCVA eliminated decreased vision due to refractive error [7].

The profiles of anti-ganglioside IgG and IgM autoantibodies against GM1, GM3, GD1a, GT1a and GQ1b gangliosides were tested semi-quantitatively using a commercial kit (Generic Assays, Germany). Briefly, patients’ sera as well as positive and negative control sera were acquired and added to the wells with buffer solution. The samples were then incubated with autoantibodies to the gangliosides immobilized on the membrane at 4 ${}^{\circ }$C for 120 minutes. Unbound serum samples were removed by several wash steps followed by an anti-human secondary antibody. The substrate 3,3${}^{\prime }$,5,5${}^{\prime }$-Tetramethylbenzidine (TMB) was subsequently pipetted and incubated for 10 minutes at room temperature (18–25 ${}^{\circ }$C). After drying for 30 minutes, the strips were read using a DotDiver2.0 device (Generic Assays, Germany). Positive response was considered if the coloration of the test line could be discriminated from the background [8].

Cases of anti-GQ1b antibody syndrome were divided into two groups according to the existence of optic neuropathy. We reviewed medical records of each patient and recorded age, gender, antecedent infections at presentation, neurological signs during the course of disease, treatment, and follow-up results. We assessed the following signs: visual impairment, internal or external ophthalmoplegia, facial weakness, bulbar palsy, limb weakness, truncal or limb ataxia, superficial sense impairment, deep tendon reflexes (brisk, normal, decreased, or absent) and Babinski’s sign. The paraclinical characteristics included cerebrospinal fluid anomalies, electrophysiological parameters, orbital magnetic resonance imaging (MRI) and visual evoked potentials (VEPs). Albumino-cytological dissociation in the cerebrospinal fluid (CSF) was defined as a normal white cell count (5 $\times$ 10${}^6$/L) with an elevated protein level ($>$450 mg/L, normal values 150–450 mg/L) [9].

Analysis was performed with SPSS statistics version 26.0 (IBM Corp., Armonk, NY, USA). As the total sample size (n = 12) was less than 40 and the distribution was unclear, numerical variables were analyzed using Wilcoxon rank sum test and the p values were obtained by consulting the rank sum test T-bound table (two-tailed). Differences in proportions were tested by Fisher’s exact test (two-tailed). A p value 0.05 was used for significance in all comparisons.

Twelve patients (5 males and 7 females) with positive anti-GQ1b antibodies were identified. Age at onset was 12 to 63, and the median age was 45. Out of 12 patients, 3 (1 male and 2 females) presenting optic neuropathy with noticeable visual deterioration, were assigned to the ocular neuropathy (ON) group. The age at presentation ranged from 42 to 63. The normal visual acuity (NVA) group included 4 males and 5 females with the median age of 48 (ranging from 12 to 60). One patient in the ON group were diagnosed with GBS, 1 with MFS/GBS, and 1with MFS. In the NVA group, 5 were diagnosed with MFS/GBS, 2 with GBS, 1 with BBE/GBS, and 1 with MFS. All cases had acute onset.

The clinical and paraclinical characteristics of all enrolled patients were summarized (Table 1, Supplementary Fig. 1). Antecedent infections occurred in 9 patients (75%), including upper respiratory tract infections and gastroenteritis. All patients in the ON group presented with visual impairments as their initial symptoms, and 4 in the NVA group also complained of blurring vision due to intraocular muscle paralysis. No signs of pain in eye movement, color desaturation or field deficits were observed in the ON group. Extraocular muscle paralysis was presented in 11 (3 in the ON group and 8 in the NVA group). A higher proportion of facial weakness was presented in the ON group than in the NVA group (100% and 22% respectively, p = 0.045, 2-sided). No statistical significance was found in terms of other cranial nerve impairments, including extraocular muscle paralysis (100% and 88.9%, p $>$ 1.000), intraocular muscle paralysis (0 and 44.4%, p = 0.491) as well as bulbar palsy (66.7% for both group, p $>$ 1.000). As for limb symptoms, the differences in the ratio of ataxia, sensory and motor disturbance, and areflexia/hyporeflexia were not statistically significant between the two groups. Cerebrospinal fluid analysis showed albumin-cytological dissociation in 10 patients. Two patients in the NVA group didn’t present with albumin-cytological dissociation, which may be the result that lumbar punctures were undergone too early (within one week) after onset. The main features of the nerve-conduction study (NCS) were also summarized in the table. None of the 12 patients got signs of brainstem or cerebellar abnormalities by cranial MRI or Computed tomography (CT) scanning.

On admission, 11 cases were administered with intravenous immunoglobulin at 0.4 g/kg/d for five days except for 1 in the NVA group. Ten cases except for 2 in the NVA group received glucocorticoids from 1 week to 1 month. All patients received vitamin B treatment. In the ON group, recovery occurred within a mean period of 2.3 months. In the NVA group, one diagnosed with BBE/GBS experienced altered consciousness and made a slow recovery (up to 14 months), which was considered as an outlier and was removed. Thus, the average period of recovery in the NVA group was 2.2 months. The statistical analysis showed that the recovery time of the two groups was not significantly different (Supplementary Fig. 2). As for prognosis, all patients were followed up from 2 months to 6 years after discharge. Eleven patients made full recoveries, and the other one in the NVA group had mild numbness and weakness in the distal limbs. It was noteworthy that the visual impairments of patients in the ON group recovered much later than other symptoms (Table 2).

Gangliosides, a kind of structurally distinct membrane glycolipids which is abundant in the nervous system, can modulate a wide variety of neuronal functions and act as bacterial toxin receptors as well as the target for autoantibodies [10, 11]. Infection with some Gram-negative bacteria such as Campylobacter jejuni and Haemophilus influenzae that bears GQ1b-like lipopolysaccharides may trigger the production of anti-GQ1b antibody due to molecular mimicry. The autoantibody can bind to the brain and nerve sections rich in GQ1b gangliosides, leading to a series of clinical manifestations [12, 13]. It has been reported that GQ1b gangliosides are abundant in the optic nerve only second to cranial nerves innervating extraocular muscles [14, 15, 16]. However, the accumulation of anti-GQ1b epitope was only observed in the paranodal regions of extramedullary portion of the IIIrd, IVth, and VIth cranial nerve, which was also the main attack site of anti-GQ1b antibody [17]. The damage to the regions blocks impulse generation at the node of Ranvier. The optic nerve, whose myelin is composed of oligodendrocytes, might lack the binding site like the paranodal regions, accounting for a low incidence of optic neuropathy in anti-GQ1b antibody syndrome. Moreover, the blood-brain barrier (BBB) might prevent these autoantibodies from entering the intracranial space and binding to the corresponding antigen. To date, the underlying mechanism in the anti-GQ1b antibody positive patients with visual impairment remains unclear. Seito et al. [18] showed that the sera of BBE patients with positive anti-GQ1b antibody can disrupt BBB via inducing the secretion of matrix metalloproteinase. There have also been reported abnormalities in cerebellar MRI of MFS patients [19]. Moreover, it is well-established that different subclasses of anti-GQ1b IgG and IgM may lead to various clinical spectrum [20]. Thus, we speculate that in some specific cases, disruption of the BBB barrier may occur, while the undefined subtypes of GQ1b antibodies simultaneously enter the central nerve system and attack optic nerve. A comprehensive examination of anti-GQ1b antibody in CSF and a serological test for pathogens in antecedent infections should be emphasized.

Odaka et al. [4] reported that a high proportion of patients (94%) had a history of early infection, of which 75% were respiratory tract infections and 9% were gastrointestinal infections. In our study, we also noted antecedent infections in 9 patients (75%), with respiratory infections in 8 cases and gastrointestinal infection in 1 case. The proportion was slightly lower than that reported in the literature. In terms of clinical manifestations of our study, once the optic nerve was involved, visual deterioration tended to present as an initial symptom and lasted longer than other clinical disorders, suggesting that the damage of GQ1b antibody to optic nerve be rapid and last longer than other nerves. Moreover, the incidence of facial weakness in the ON group was higher than that in the NVA group (p = 0.045), suggesting a higher probability of facial nerve damage for patients with visual impairment.

To better understand the clinical characteristics of patients with visual impairment in anti-GQ1b antibody syndrome, we summarized the clinical features of 8 patients reported in literature (Table 3) [21, 22, 23, 24, 25, 26, 27, 28]. Out of all 11 cases, 7 cases showed bilateral visual impairment, indicating that the anti-GQ1b antibody often attack bilateral optic nerves. Consistent to our study, most patients had visual impairment as initial symptoms. In addition, 2 patients presented with hyperreflexia, which may be attributed to damage to the pyramidal tract.

The results of lumbar puncture suggested that 10 patients (3 in the ON group and 7 in the NVA group) showed albumin-cytological dissociation in CSF, which followed the characteristics of CSF in GBS [4]. Moreover, Nobuhiro et al. [29]. reported that 50–66% of GBS patients presented with an elevated protein level but normal cell count within the first week or over 75% within the third week after the symptom onset. Our data indicated that the occurrence of albumin-cytological dissociation in the anti-GQ1b antibody syndrome followed a similar time course as in GBS.

Lee et al. [30] showed that 54.5% of patients with acute ophthalmoplegia and seral anti-GQ1b antibody also presented intraocular muscle paralysis. In our study, 4 patients in the NVA group had different degrees of unilateral/bilateral pupil dilatation. Interestingly, they also complained of blurring vision as their initial symptoms, which may mislead to the diagnosis of optic neuritis. We noticed that these patients mainly presented with deterioration of near vision accompanied by photophobia due to the dilation of pupils. Moreover, the visual acuity of the patients could be improved after correction, similar to the manifestation of ametropia. The characteristics above can serve as evidence to distinguish from optic neuropathy. It was hypothesized that anti-GQ1b antibody binding to the intraocular neuromuscular junction led to a massive release of acetylcholine, resulting in immunopathologic involvement of the ciliary ganglion and super sensitivity of the iris sphincter muscles due to denervation, which eventually caused the destruction of the motor nerve terminal structure and thus the pupil dilatation [31, 32].

Besides, the manifestation of vision loss and optic neuritis in the ON group should be differentiated from neuromyelitis optica, multiple sclerosis or MOG spectrum disorders. All 3 patients presented with symmetrical glove-sock hypoesthesia or symmetrical limb weakness, without observed lesions in the spinal cord or in the brain by MR scanning. Meanwhile, neither AQP4 nor oligoclonal band was found in serum or cerebrospinal fluid. In terms of MOG spectrum disorders, the anti-MOG antibody was negatively detected in 1 patient. There was no detection for this antibody in the other 2 patients due to unavailability in the earlier years. Furthermore, none of the cases in the ON group had signs of encephalitis or CNS demyelination, either any indication of relapses in our follow-up. Therefore, the diagnosis of neuromyelitis optica, multiple sclerosis or MOG spectrum disorders was not supported in the enrolled cases.

Given the autoimmunity of GBS, immunoglobulin therapy is considered effective. The mechanisms mainly attributed to its ability to provide specific antibodies, block Fc receptors on macrophages, inhibit complement activities, and regulate B cell and T cell functions [33]. The published data suggested that oral and intravenous administration of glucocorticoids have no benefit, even a combination of IVIg and glucocorticoids is not more effective than IVIg alone [34, 35]. Eleven cases in our study were treated with intravenous immunoglobulins except for 1 case, who was initially diagnosed with encephalitis and close to complete recovery at the time of GBS diagnosis. Significantly, visual acuity of patient 1 in the ON group had no improvement after receiving immunoglobulin. According to the Optic Neuritis Treatment Trial (ONTT) [36], high-dose corticosteroids administered intravenously was the standard therapy for optic neuritis. Therefore, methylprednisolone was given intravenously to the patients with optic neuropathy. The usage of glucocorticoids in treating GQ1b antibody syndrome with visual impairment has also been reported elsewhere, suggesting an individualized therapy should be considered for this kind of patients [21, 23, 26].

There are some limitations in our study: (1) As anti-GQ1b antibody syndrome with visual impairment is extremely rare, and only a few cases have been reported by far, we have collected 3 eligible patients. The difference between the ON and NVA groups was difficult to draw a statistically significant conclusion, and the sampling error was large due to the small number of sample cases. Therefore, we summarized the clinical features of the other 8 patients reported in literature and made a discussion together with our cases to illustrate a more comprehensive clinical spectrum. (2) No examination of anti-GQ1b antibody in CSF and no serological tests for pathogens in antecedent infections was performed in our patients. Although these examinations were not taken as routine items in the past, our study suggested that they serve as significant evidence for uncovering the mechanisms of vision impairment in anti-GQ1b antibody syndrome.