Large-scale profiling of antibody reactivity to glycolipids in patients with Guillain-Barré syndrome

Abstract

Guillain-Barré syndrome is an acute polyradiculoneuropathy in which preceding infections often elicit the production of antibodies that target peripheral nerve antigens, principally gangliosides. Anti-ganglioside antibodies are thought to play a key role in the clinical diversity of the disease and can be helpful in clinical practice. Extensive research into clinical associations of individual anti-ganglioside antibody specificities has been performed. Recent research has highlighted glycolipid complexes, glycolipid combinations that may alter antibody binding, as targets. In this study, we investigated antibody reactivity patterns to glycolipids and glycolipid complexes using combinatorial array, in relation to clinical features in Guillain-Barré syndrome. In total, 1413 patients from the observational International Guillain-Barré syndrome Outcome Study (0-91 years, 60.3% male) and 1061 controls (healthy, family, infectious, vaccination, other neurological disease) were included. Acute-phase sera from patients were screened for IgM, IgG, and IgA reactivity against 15 glycolipids and one phospholipid and their heteromeric complexes, similarly to archived control sera. Antibody specificities and reactivity patterns were analysed in relation to clinical features. Of all patients, 1309 (92.6%) were positive for at least one anti-glycolipid (complex) antibody. Anti-GM1 and anti-GQ1b (complex) antibodies best distinguished motor Guillain-Barré syndrome and Miller Fisher syndrome from controls, with antibodies to glycolipid complexes outperforming antibodies to single glycolipids. Three models consisting of anti-glycolipid (complex) antibodies distinguished patients with Guillain-Barré syndrome, the motor variant, and Miller Fisher syndrome from controls with high sensitivity and specificity, performing better than antibodies to single glycolipids used in clinical practice. Seven patient clusters with particular antibody reactivity patterns were identified. These clusters were distinguished by geographical region, clinical variants, preceding Campylobacter jejuni infection, electrophysiological subtypes, the Medical Research Council sum score at study entry, and the ability to walk 10 m unaided at 26 weeks. Two patient clusters with distinct anti-GM1 (complex) reactivity (broad versus restricted) differed in frequency of the axonal subtype. In cumulative incidence analyses, 15 anti-glycolipid (complex) antibodies were associated with the time required to regain the ability to walk 10 m unaided. After adjustment for known prognostic factors, IgG anti-GQ1b:GM4, GQ1b:PS and GQ1b:Sulfatide remained associated with faster recovery. Addition of anti-glycolipid antibodies to clinical prognostic models slightly improved their discriminative capacity, though insufficiently to improve the models. Measurement of anti-glycolipid antibodies by combinatorial array increases the diagnostic yield compared to assaying single glycolipids, identifies clinically relevant antibody reactivity patterns to glycolipids and glycolipid complexes, and may be useful in outcome prediction in Guillain-Barré syndrome.

Keywords: autoantibody; peripheral neuropathy.

Figure 1

Comparison of receiver operating characteristic curves for models distinguishing patients with Guillain-Barré syndrome, motor Guillain-Barré syndrome or Miller Fisher syndrome from controls. Receiver operating characteristic curves are shown with associated values for the area under the receiver operating characteristic curve, for univariable models (A and B) and multivariable models (C–E). Using univariable models, the differentiating performance of IgG anti-GM1 complex antibodies and IgG anti-GQ1b complex antibodies were compared to IgG antibodies to GM1 or GQ1b alone for the distinction of motor Guillain-Barré syndrome (A) or Miller Fisher syndrome (B) from controls. Additionally, newly created multivariable models containing antibodies to both single gangliosides and ganglioside complexes were compared to currently used multivariable models based on antibodies to single gangliosides, for the distinction of Guillain-Barré syndrome (C), motor Guillain-Barré syndrome (D) or Miller Fisher syndrome (E) from controls. AUC = area under the receiver operator characteristic curve; GBS = Guillain-Barré syndrome; GalC = galactocerebroside; GN-GD1a = N-acetylgalactosaminyl GD1a; MFS = Miller Fisher syndrome; PS = phosphatidylserine; SGPG = sulfated glucuronyl paragloboside; Sulf = sulfatide.

Figure 2

Forest plots depicting the top five anti-glycolipid (complex) antibodies associated with several clinical features in patients with Guillain-Barré syndrome. Associations of anti-glycolipid antibodies with motor Guillain-Barré syndrome, Miller Fisher syndrome, bulbar palsy at study entry, the axonal subtype, and preceding Campylobacter jejuni and Mycoplasma pneumoniae infections. Values indicate the odds ratio with their 95% confidence interval per anti-glycolipid antibody. Antibodies were ranked based on the P-value resulting from univariable logistic regression analyses. CJ = Campylobacter jejuni; GalC = galactocerebroside; GalNAc-GD1a = N-acetylgalactosaminyl GD1a; GBS = Guillain-Barré syndrome; MFS = Miller Fisher syndrome; MP = Mycoplasma pneumoniae; PS = phosphatidylserine; Sulf = sulfatide.

Figure 3

Heat map depicting patient clusters with particular IgG anti-glycolipid antibody reactivity patterns derived from unsupervised hierarchical clustering of anti-glycolipid antibodies in patients with Guillain-Barré syndrome. Patients were clustered on the y-axis (A–G) and anti-glycolipid antibodies were clustered on the x-axis. The clusters are separated by white lines. Each patient cluster is characterized by a distinct antibody reactivity pattern: (Cluster A) broad-ranging GalNAc-GD1a reactivity; (Cluster B) restricted GA1 and broad-ranging GM1 reactivity; (Cluster C) restricted GalNAc-GD1a reactivity; (Cluster D) restricted GA1, GD1b and GM1 reactivity; (Cluster E) non-specific; (Cluster F) restricted GQ1b and GT1a reactivity; and (Cluster G) broad-ranging GT1a reactivity. GalC = galactocerebroside; GN-GD1a = N-acetylgalactosaminyl GD1a; PS = phosphatidylserine; SGPG = sulfated glucuronyl paragloboside; Sulf = sulfatide.

Figure 4

Dot plots and stacked bar plots illustrating the associations between anti-glycolipid antibodies (reactivity patterns) and electrophysiological subtypes in Guillain-Barré syndrome. (A) Box plots with individual anti-GM1 (left) and anti-GM1:Sulfatide (right) fluorescence intensities across electrophysiological subtypes. (B) Stacked bar plot depicting the distribution of electrophysiological subtypes across patient clusters based on anti-glycolipid antibody reactivity patterns. (C) Dot plot illustrating the interaction of GM1 with GD1a in patients, by comparing the sum of fluorescence intensities of anti-GM1 and anti-GD1a (anti-GM1 + anti-GD1a) with the fluorescence intensity of the anti-complex antibody anti-GM1:GD1a per individual patient. Each line connects the fluorescence intensity of anti-GM1 + anti-GD1a to the fluorescence intensity of anti-GM1:GD1a of one patient. Groups are based on electrophysiological subtypes. (D) Stacked bar plot showing the distribution of electrophysiological subtypes across three groups based on the interaction of GM1 with GD1a (complex independent, enhanced or attenuated). *P < 0.05, **P < 0.01, ***P < 0.001. FI = fluorescence intensity; Sulf = sulfatide; U = units.

Figure 5

Cumulative incidence curves for the time to regain the ability to walk unaided in relation to anti-glycolipid antibody reactivity. Cumulative incidence curves are shown for IgG anti-GQ1b:GM4 (A), IgG anti-GM1:Sulfatide (B), IgA anti-GM1:Sulfatide (C) and patient clusters based on anti-glycolipid antibody reactivity patterns (D). Neg. = negative; Pos. = positive; Sulf = sulfatide.