Abstract
Objective: To test the hypothesis that different preceding infections influence the neurophysiologic classification and clinical features of Guillain–Barré syndrome (GBS).
Methods: We tested pretreatment sera, 7 ± 3 (mean ± SD) days from onset, from 229 patients with GBS in a multicenter trial of plasma exchange and immunoglobulin, for serological markers of infection, adhesion molecules, and cytokine receptors, and compared these with neurophysiologic and clinical features.
Results: Recent infection by Campylobacter jejuni was found in 53 patients (23%), cytomegalovirus in 19 (8%), and Epstein–Barr virus in four (2%). Patients with C. jejuni infection were more likely than others to have neurophysiologic criteria of axonal neuropathy or inexcitable nerves, antiganglioside GM1 antibodies, pure motor GBS, lower CSF protein, and worse outcome. Patients with cytomegalovirus infection were younger and more likely than others to have raised serum concentrations of molecules important in T lymphocyte activation and migration, soluble intercellular adhesion molecule-1 (sICAM-1), soluble vascular cell adhesion molecule-1 (sVCAM-1), soluble leukocyte selectin, and soluble interleukin-2 receptor (sIL-2R). Concentrations of sICAM-1 and soluble tumor necrosis factor receptor were higher in patients with inexcitable nerves than those with demyelinating neurophysiology. Logistic regression analysis showed death or inability to walk unaided at 48 weeks were associated with diarrhea, inexcitable nerves, severe arm weakness, age over 50, raised sIL-2R concentration and absence of immunoglobulin (Ig) M antiganglioside GM1 antibodies.
Conclusions: Subtypes of GBS defined by preceding infections were only approximately associated with different patterns of clinical, neurophysiologic, and immunologic features. A single infectious agent caused more than one type of pathology in GBS, implying interaction with additional host factors. Most patients had no identified infection.
Guillain–Barré syndrome (GBS) is recognized to have different pathologic subtypes and different antecedent infections, but the relationships between these are still the subject of debate. In North America and Europe the commonest subtype has acute inflammatory demyelinating polyradiculoneuropathy as its pathologic substrate and fewer than 10% of cases are due to acute motor axonal neuropathy and acute motor and sensory axonal neuropathy.1-3⇓⇓ These axonal forms are more frequent in China.4
The most common antecedent infection recognized before the onset of GBS is Campylobacter jejuni enteritis, which has been associated with more severe axonal degeneration in Western patients3 and with acute motor axonal neuropathy in Chinese patients.5 The next most common infection preceding GBS is cytomegalovirus (CMV),6,7⇓ which has been associated with prominent sensory involvement and severe disease.8
Antibodies to gangliosidelike molecules on C. jejuni or virus-infected cells may cross-react with axonal or Schwann cell membranes and provide a possible explanation for the different GBS subtypes. Prominent T lymphocyte (T cell) activation occurs in GBS and is important in the pathogenesis of its rat model, experimental autoimmune neuritis (EAN).9 Similarly, up-regulation of integrins on T cells and adhesion molecules on the endoneurial endothelium has been demonstrated in GBS and EAN.9 The differential role of these T cell and endothelial immune modulating factors in the pathogenesis of GBS subtypes has not been explored in detail.
We conducted a large, randomized treatment trial1 and classified the patients into electrophysiologic subtypes.2 We tested pretreatment sera and reported that patients with electrophysiologic criteria of axonal neuropathy and inexcitable nerves were more likely to have antibodies to ganglioside GM1 than those with demyelinating neuropathy.2 This study was designed to test the hypothesis that different preceding infections contribute to the development of different GBS subtypes. This is the first study to test the association of cytokine receptors, adhesion molecules, CMV, and Epstein–Barr virus (EBV) infection with the other variables in a large group of patients early in the disease.
Methods.
Three hundred eighty-three patients with GBS were enrolled in a randomized, controlled trial in 11 countries in Europe, North America, and Australia between 1993 and 1995 that compared treatment with plasma exchange (PE), IV immunoglobulin (IVIg), or both.1 Entry criteria included age over 16 years, inability to walk unaided, and onset less than 14 days before random assignment. Blood was taken from each patient on the day of randomization (mean ± SD of 7 ± 3 days after onset of neuropathic symptoms and before starting treatment), and again 14 days later (after finishing treatment). All sera were coded and analyzed at a single group of laboratories in Würzburg; researchers were unaware of the clinical details. The code was not broken until all serum analysis had been done. We excluded 154 of the 383 patients: 14 because of incomplete neurophysiologic data or trial protocol violations,2 128 because a satisfactory serum specimen was not received,2 four because of absence of test results for C. jejuni, six because of absence of results for cytomegalovirus, and two because of absence of a CSF result. Therefore 229 patients were included in the present analysis. Of these, 30 did not have a record of clinical and electrophysiologic sensory examination, so 199 were included in the subanalysis of pure motor GBS.
Clinical1 and electrophysiologic2 methods for this study have been described previously. A respiratory infection was defined as a history of a recent upper or lower respiratory infection or an influenzalike illness. Pure motor GBS was defined as the presence of both a normal clinical sensory examination at randomization and normal sensory action potentials when first tested. CSF was sampled before treatment (a median of 4 days after GBS onset; interquartile range [IQR], 2 to 7 days) and analyzed in the patient’s local laboratory. A relapse was defined clinically as a worsening of at least one disability grade1 due to GBS, lasting at least 7 days and following at least 7 days of stabilization or improvement. Poor outcome was defined as death or inability to walk 5 meters without the assistance of a person, frame, or stick at 48 weeks after randomization.
Campylobacter jejuni infection.
Pretreatment serum was tested for antibodies to C. jejuni by ELISA to define a recent infection, and by Western immunoblot to seek associations with specific C. jejuni proteins. We defined a recent C. jejuni infection by the presence of either positive ELISA results for both immunoglobulin (Ig) A and IgG classes, or both a history of recent diarrhea and positive ELISA results for IgG or IgA, where a positive assay could be for either or both strains.
Campylobacter jejuni strains ATCC 43446 (Penner serotype O:19) and LIO11 (Penner serotype O:15, Lior type 11) were obtained from the National Reference Center for Intestinal Pathogens, Hamburg, Germany. Strains were cultured on Campylobacter agar (Oxoid Ltd., Basingstoke, UK) with 8% lysed horse blood (Oxoid) and Campylobacter-selective supplement (Skirrow formula, Oxoid) at 42 °C for 48 hours in an oxygen-depleted atmosphere containing 10% CO2 (Anaerocult C, Merck, Darmstadt, Germany).
Outer membrane protein antigens were prepared from each C. jejuni strain using the N-lauroyl-sarcosine extraction method described by Achtman,10 with small modifications. Briefly, bacteria from 20 plates were harvested using a Trigalsky spatula, suspended in 0.9% NaCl and centrifuged at 4,000 g for 15 minutes at 4 °C. The bacterial pellet was washed in saline, resuspended in 10 mM Tris-HCl (pH 8.0), disintegrated by sonication (four times 30 seconds) using a Branson Sonifier Cell Disruptor B15 (Danburg, CT), and centrifuged at 1,500 g for 20 minutes at 4 °C to remove intact bacterial cells. The supernatant containing cell membranes was centrifuged at 48,200 g for 1 hour at 4 °C. The subsequent pellet was dissolved in 150 μL distilled water, then 1.2 mL of 1.67% N-lauroyl-sarcosine solution in 11.1 mM Tris-HCl (pH 7.6) was added, mixed well and extraction was performed for 20 minutes at room temperature. The sample was then ultracentrifuged at 46,000 g for 90 minutes at 20 °C and the pellet resuspended in 500 μL distilled water.
ELISA.
Microtiter plates (U-bottomed; Greiner, Frickhausen, Germany) were coated with antigens from C. jejuni strains ATCC 43446 or LIO11. A 0.1-μg outer membrane protein preparation in 100 μL disodium carbonate coating buffer (pH 9.4) was added to each well and incubated overnight at 4 °C in a moist chamber. After washing five times with washing buffer (0.05 M Tris-HCl, pH 7.4; 0.35 M NaCl; 0.1% Tween 20), 100 μL of serum samples diluted 1:100 in dilution buffer (0.05 M Tris-HCl, pH 7.4; 0.35M NaCl; 0.1% Tween 20; 0.0005% phenol red) were added in duplicate onto the coated plates and incubated at 37 °C for 1 hour. After additional washing, 100 μL of alkaline phosphatase–labeled antihuman IgG diluted 1:4000 (Dianova, Hamburg, Germany) or antihuman IgA diluted 1:3000 in dilution buffer were added to each well and incubated for 1 hour at 37 °C. The plates were developed with 100 μL of p-nitrophenyl-phosphate substrate solution by incubation for 15 minutes at room temperature in the dark. The reaction was stopped with 100 μL 2M NaOH and optical densities at 405 nm were read using the MR 700 Microplate ELISA reader (Dynex, Denkendorf, Germany). A positive result was defined as a mean optical density more than two standard deviations above the mean of 226 normal control German blood donors.
Western blotting.
Identifying the molecular specificity of antibodies to C. jejuni might help to clarify the hypothesis of molecular mimicry in the pathogenesis of GBS. Western blotting and immunostaining were performed as described previously11 for ATCC 43446 and LIO11 strains. IgA antibodies were measured against C. jejuni outer membrane proteins of molecular weight 28, 29, 30, and 31 kD (PEB 1, 2, 3, and 4 molecules, probably involved in adhesion or transport), 42 kD (porin, the major outer membrane protein), 50 kD, 55 kD (flagellin), 70 kD (a heat-shock protein), and 92 kD.
Cytomegalovirus and Epstein–Barr virus infection.
Pretreatment serum samples were tested for CMV IgM and EBV IgM with commercial enzyme immunoassays (Enzygnost anti-CMV/IgM and anti-EBV/IgM; Dade Behring, Marburg, Germany). The IgG avidity was measured because IgM alone is not specific for acute infection.12 EBV IgG avidity was analyzed according to a semiautomated protocol using Enzygnost anti-EBV/IgG test kits (Dade Behring) and a 4.7 M urea solution as described previously.13 Similarly, CMV IgG avidity was measured using Enzygnost anti-CMV/IgG test kits (Dade Behring) and a 6 M urea solution. Patients were considered to have an acute CMV or EBV infection when IgM was positive or borderline and IgG avidity was low (<45%). Acute infections were excluded when IgM was negative or IgG avidity was high (>50%). The serologic status of all other patients was considered indeterminate.
Antibodies to ganglioside GM1.
Antibodies to ganglioside GM1 were measured by ELISA and verified by dot blot. This method has been described,2 as modified from reference 11.
Cytokine receptors and adhesion molecules.
Concentrations of the following molecules in serum were measured by quantitative sandwich ELISA and considered to be raised if they exceeded the following upper limits of normal: soluble leukocyte selectin (sL-selectin, 1,350 ng/mL), soluble vascular cell adhesion molecule-1 (sVCAM-1, 714 ng/mL), soluble intercellular adhesion molecule-1 (sICAM-1, 400 ng/mL), 60-kD soluble tumor necrosis factor receptor (sTNF-R, 5.0 ng/mL), soluble interleukin-2 receptor (sIL-2R, 4.0 ng/mL), soluble interleukin-4 receptor α (sIL-4Rα, 50 pg/mL), and interleukin-10 (IL-10, 3.0 pg/mL). All assays were performed with commercially available kits according to the manufacturers’ instructions (sVCAM-1 and sIL-4Rα from R & D Systems, Bad Nauheim, Germany; all others from Bender/Boehringer, Vienna, Austria).
Statistical analysis.
Multiple statistical tests were necessarily done to explore the complex relationships. We did not attempt formally to correct for this or to define a stricter threshold for significance, but instead took these results as generating hypotheses for future testing, and used p values simply to assess the strength of association. Results with p values close to 0.05 should be interpreted with caution.
Dichotomous data were analyzed using the χ2 test with Yates’ continuity correction, Fisher’s test if frequencies were small, or McNemar’s test for paired data. Continuous variables were analyzed using Student’s t-test, or one-way analysis of variance (ANOVA) if more than two groups were compared. The log rank test was used to analyze number of days until walking. All tests were two tailed. All analyses were done using SPSS for Windows software version 6.1 (SPSS, Chicago, IL).
Logistic regression was used in analyses investigating the simultaneous effects of several factors on a dichotomous outcome. Nonsignificant variables were removed in turn to leave a final model containing only significant variables. The analyses included the following dichotomous explanatory variables at the time of randomization: sex; age 50 or more; C. jejuni, respiratory, or CMV infection; diarrhea; inexcitable motor nerves; assisted ventilation at randomization; severe distal and proximal upper limb weakness (arm grade 3 or worse); pure motor GBS; treatment assigned; short delay (4 days or fewer) from onset until randomization; distally-evoked compound muscle action potential amplitude <20% of lower limit of normal; cardiac arrhythmia requiring treatment; abnormal serum liver enzymes; IgG, IgA, and IgM antibodies to ganglioside GM1; and raised concentrations of sL-selectin, sVCAM-1, sICAM-1, sTNF-R, sIL-2R, sIL-4Rα, or IL-10.
Results.
There were no significant differences between the 229 included and the 154 excluded patients in age, disability grade at randomization or at 48 weeks, or the proportions with diarrhea or in each neurophysiologic or treatment group.
Infections and immune factors.
Prevalence of preceding infections.
Evidence of recent infection by C. jejuni was present in 53 of 229 patients (23%), by CMV in 19 (8%), and by EBV in four (2%). One patient had infection by both C. jejuni and CMV, and one had infection by both CMV and EBV. One patient had indeterminate CMV status. Patients with CMV infection were younger (mean age ± SD, 35 ± 12 years) than those with no infection (52 ± 16 years, p < 0.001) or C. jejuni infection (55 ± 15 years). The proportion of patients with GBS who had C. jejuni infection varied among different countries (p = 0.0005), principally due to the high proportion in Portugal (seven of eight; 88%). There was no significant difference between the overall proportions in Europe and North America.
Of 369 patients in whom data on a clinical history of infection were available, 64 (17%) had recently had diarrhea only, 145 (39%) respiratory infection only, 11 (3%) both, and 149 (41%) neither. C. jejuni infection was present in 31 of 47 patients (66%) with diarrhea compared with 22 of 182 patients (12%) without diarrhea (p < 0.0001). Diarrhea occurred a median of 9 days (IQR, 6 to 13 days) before onset of GBS. The time delay between diarrhea and GBS onset was not significantly related to presence or absence of C. jejuni infection, nor to neurophysiologic subtype.
Infections and neurophysiologic type.
Campylobacter jejuni infection was more common in the groups with neurophysiologic evidence of axonal neuropathy or inexcitable nerves. Neurophysiologic criteria classified 148 patients as having demyelinating neuropathy, six as having axonal neuropathy, seven as inexcitable, eight as normal, and 60 as equivocal. Axonal neurophysiology was present in five of 52 patients (10%) with C. jejuni infection alone, compared with one of 155 (1%) with no infection (p = 0.004). Patients with axonal neurophysiology were more likely to have IgA antibodies on Western blot to the 28-, 29-, 30-, and 55-kD C. jejuni outer membrane proteins. Nerves were inexcitable in four patients (8%) with C. jejuni infection alone compared with two (1%) with no infection (p = 0.04). Demyelinating neurophysiology was present in 29 of 52 patients (56%) with C. jejuni infection alone, 13 of 17 (77%), with CMV infection alone, and 103 of 155 (66%) with no infection. By comparison, using the Dutch criteria,14 demyelinating neurophysiology was present in 21 patients (40%) with C. jejuni infection, 10 (59%) with CMV infection, and 80 (52%) with no infection. Patients with C. jejuni infection and axonal neurophysiology were younger (44 ± 8 years, mean ± SE) than those with C. jejuni infection and demyelinating neurophysiology (58 ± 3 years, p = 0.05). The proportions with IgA Western blot antibodies did not differ among patients with C. jejuni infection with different neurophysiologic types. There was no obvious relationship between CMV or EBV infection and neurophysiologic type, although numbers were small. No patient with CMV or EBV infection had axonal neurophysiology.
Infections and pure motor GBS.
Pure motor GBS occurred in 19 of 131 patients (15%) with no infection, compared with 17 of 46 (37%) with C. jejuni infection only (p = 0.002) and one of 16 (6%) with CMV infection only (p = 0.3). Pure motor GBS was associated with IgA antibodies on Western blot to the 28-, 30-, 31-, 42-, and 55-kD C. jejuni outer membrane proteins. Five of the eight Portuguese patients (63%) had pure motor GBS.
Infections and antibodies to ganglioside GM1.
Patients with C. jejuni infection were more likely than those with no infection to have IgA, IgG and IgM antibodies to ganglioside GM1 ( table 1). IgA antibodies on Western blot to the 28-, 29-, 30-, 31-, 42-, and 70-kD C. jejuni outer membrane proteins were more frequent in patients with IgA antibodies to ganglioside GM1. Considering patients with C. jejuni infection in different neurophysiologic groups, IgM (but not IgG or IgA) antibodies to ganglioside GM1 were more frequent in the axonal (4/5, 80%) and inexcitable (3/4, 75%) groups than the demyelinating group (6/30, 20%; p = 0.02).
Preceding infections, antibodies to ganglioside GM1, and outcome
Infections, adhesion molecules, and cytokine receptors.
Patients with CMV infection had higher mean serum concentrations of the adhesion molecules sICAM-1, sVCAM-1, and sL-selectin than patients with C. jejuni or no infection (figure 1). Similarly, a greater proportion of patients with CMV infection had raised concentrations of these adhesion molecules and sIL-2R than those with no infection. Pretreatment serum concentrations of sICAM-1 were raised (above the upper limit of normal) in 10 of 17 patients (59%) with CMV infection alone compared with 16 of 155 (10%) with no infection (p < 0.001); concentrations of sVCAM-1 were raised in 14 patients (82%) with CMV and 75 (49%) with no infection (p = 0.02); concentrations of sL-selectin were raised in 16 patients (94%) with CMV and 96 (62%) with no infection (p = 0.02); and concentrations of sIL-2R were raised in 12 patients (71%) with CMV and 65 (42%) with no infection (p = 0.05). Patients with EBV infection had raised concentrations of the same molecules, although this was not significant (but the numbers were small). C. jejuni infection was not associated with raised concentrations of cytokine receptors or adhesion molecules compared with patients without infection.
Figure 1. Patients with cytomegalovirus (CMV) and Epstein–Barr virus (EBV) infection had higher mean concentrations of adhesion molecules and soluble interleukin-2 receptor (sIL-2R) than patients with Campylobacter jejuni or no infection. Lines show upper limit of normal concentration. One-way analysis of variance: soluble leukocyte (sL)-selectin, p < 0.001; soluble vascular cell adhesion molecule-1 (sVCAM-1), p < 0.001; soluble intercellular adhesion molecule-1 (sICAM-1), p < 0.001; sIL2-R, p = 0.002. CMV versus no infection (Student’s t-test): sL-selectin, p < 0.001; sVCAM-1, p = 0.009; sICAM-1, p < 0.001; sIL2-R, p = 0.1. EBV versus no infection (Student’s t -test): not significant. Extreme values for EBV: (a) 12,700 ± 6400 ng/mL, (b) SE 3400 ng/mL.
Infections and CSF.
The CSF protein concentration was lower in patients with C. jejuni infection (median, 450 mg/L; IQR, 332 to 600 mg/L) than those with no infection (745 mg/L; IQR, 360 to 1,327 mg/L; p = 0.0002) or with CMV infection (1,200 mg/L; IQR, 575 to 1,465 mg/L). CSF leukocyte count was higher in patients with EBV infection (median, 3/μL; IQR, 2 to 16/μL) than in those with no infection (1/μL; IQR, 0 to 2/μL; p = 0.03). There was no significant difference in the CSF protein concentration or leukocyte count among the neurophysiologic groups.
Cytokine receptors, adhesion molecules, and neurophysiologic groups.
Patients with inexcitable nerves had higher mean concentrations of sICAM-1 and sTNF-R than those in all other groups ( figure 2). There was no difference between the axonal and demyelinating groups in the mean concentrations of any cytokine or adhesion molecule measured.
Figure 2. Mean serum concentrations of proinflammatory cytokine receptors and soluble interleukin-2 receptor (sIL-2R) were greater in patients with inexcitable nerves than other neurophysiologic groups. Lines show upper limit of normal concentration. Soluble intercellular adhesion molecule-1 (sICAM-1), p = 0.005 (analysis of variance); soluble tumor necrosis factor receptor (sTNF-R), p = 0.007 (analysis of variance); sIL-2R, p = 0.02 (Student’s t-test, inexcitable versus demyelinating).
Outcome.
Thirty-five of the 229 patients (15%) in the main analysis had poor outcome (defined as death or inability to walk unaided at 48 weeks), compared with 58 of 369 (16%) patients in the whole trial.
Associations of single variables with outcome.
Poor outcome was associated with C. jejuni infection (see table 1), diarrhea, and IgA antibodies on Western blot to the 28-, 29-, 30-, 31-, and 55-kD outer membrane proteins of C. jejuni. Of 369 patients with data available on clinical history of infection, 26 of 75 (35%) with diarrhea had a poor outcome compared with 32 of 294 (11%) without diarrhea (p < 0.0001). Patients with pretreatment concentrations of sTNF-R or sIL2-R above the upper limit of normal had a worse outcome than those with normal concentrations ( figure 3). Raised pretreatment concentrations of sL-selectin were associated with a better outcome than normal concentrations. Patients with IgA or IgG antibodies to ganglioside GM1 before treatment had a worse outcome ( table 2).
Figure 3. Raised concentrations of soluble tumor necrosis factor receptor (sTNF-R) and soluble interleukin-2 receptor (sIL-2R) and normal concentrations of soluble leukocyte- (sL)-selectin before treatment predict poor outcome at 48 weeks. Proportions are derived from patients with poor outcome/total patients with normal or raised concentration. sL-selectin: 19/81 normal, 16/148 raised, p = 0.02; sTNF-R: 25/193 normal, 10/36 raised, p = 0.04; sIL-2R: 11/119 normal, 24/110 raised, p = 0.01.
Pretreatment immunoglobulin (Ig)A and IgG antibodies to ganglioside GM1 predict worse outcome
Logistic regression analysis of outcome, relapse, and death.
Logistic regression analysis showed that poor outcome was significantly associated with diarrhea, older age, severe arm weakness, inexcitable motor nerves, raised sIL-2R concentrations, and absence of IgM antibodies to ganglioside GM1 ( table 3).
Factors associated with poor outcome (death or inability to walk unaided at 48 weeks) by logistic regression analysis
A relapse occurred in 19 of 369 patients (5%) (including eight of 229 patients in the main analysis, 3%). Logistic regression analysis of all variables found that risk of relapse was not significantly associated with treatment type or any other factor tested.
Seventeen of 369 patients (5%) (including 10 of the 229 patients in the main analysis, 4%) died within 48 weeks. Logistic regression analysis of all variables showed that the factors at randomization significantly associated with death were raised sTNF-R concentration, cardiac arrhythmia requiring treatment, C. jejuni infection, and artificial ventilation.
Logistic regression analysis was repeated using only simple clinical variables available to most neurologists at the time of diagnosis. This excluded all serologic results (i.e., antibodies, serologically defined infections, and cytokine receptors and adhesion molecules) and therefore included all 369 patients, and used all other variables listed in the Methods section. Poor outcome was significantly associated with cardiac arrhythmia requiring treatment, diarrhea, inexcitable nerves, age 50 years or over, and severe arm weakness.
Treatment differences in subgroups.
Patients with pure motor GBS walked sooner and had a better outcome at 48 weeks after combined PE and IVIg treatment than after PE alone (p = 0.03), although IVIg alone was no better than PE (p = 0.2). Of the 53 patients with pure motor GBS, five of 10 (50%) treated with PE had a poor outcome, compared with four of 21 (19%) treated with IVIg and two of 22 (9%) treated with combined PE and IVIg (p = 0.03 across three treatments). Motor GBS has no universally accepted definition, and there was no significant difference among the three treatments if motor GBS was defined either as the presence of normal sensory action potentials alone or as the presence of normal clinical sensory examination alone.2 There was no significant difference among outcomes in the three treatment groups in patient subgroups with preceding diarrhea, age over 50 years, preceding respiratory infection, C. jejuni infection, CMV infection, those who were ventilated at random assignment, or those with antibodies to ganglioside GM1.
Discussion.
Infections and immune factors.
This is the largest study of associations between preceding infections, neurophysiologic subtype, markers of immune activation, and outcome in GBS. Patients with C. jejuni infection were more likely to have neurophysiologic criteria of axonal neuropathy, antibodies to ganglioside GM1, pure motor GBS, and a less-elevated CSF protein concentration. Patients with CMV infection were more likely to have raised concentrations of molecules associated with T cell activation and migration. This supports the emerging idea that subtypes of GBS defined by infection are associated with different patterns of disease. However, these associations were only approximate and most patients had no identified infection.
We found C. jejuni infection in 23% of GBS patients, similar to the 13% to 39% reported in Europe and North America,3,11,15-17⇓⇓⇓⇓ but fewer than the 24% to 52% in Japan18 and 47% to 72% in China.4,19,20⇓⇓ This study defined C. jejuni infection by the presence of antibodies of two classes, or one class and recent diarrhea, and so may have been more specific than other studies that required antibodies of only one class. We tested for the two serotypes of C. jejuni most commonly implicated in GBS, and about two thirds of the sera positive to one strain also cross-reacted with the other, but we may not have detected infection in a few patients infected with other strains. We confirm the association of C. jejuni and antibodies to ganglioside GM1.15,18,20,21⇓⇓⇓ However, C. jejuni isolated from patients with GBS and with enteritis alone had similar gangliosidelike moieties, suggesting that host susceptibility is important too.22,23⇓ Western immunoblot did not show that an antibody to any single protein on C. jejuni was particularly associated with the pathogenesis of GBS. It would be interesting to investigate antibodies to C. jejuni oligosaccharides, which would not have been detected by our Western blot assay.
We found acute CMV infection in 8% of patients with GBS, compared with 5% to 22% in previous reports, and acute EBV infection in 2%, compared with 1% to 13% in previous reports.6-8,18,24,25⇓⇓⇓⇓⇓ Patients with EBV-associated GBS tended to have features similar to those with CMV infection. Two patients (1%) had more than one infection, compared with 8% in the Dutch study.7 Apparent double infection may represent a single infection with reactivation of antibodies to another previous infection.12 Our method, using a combination of IgG avidity and IgM, is very specific for acute CMV or EBV infection and minimizes false-positive IgM results. Patients with CMV and EBV infection often had IgM but never IgG or IgA antibodies to ganglioside GM1. Previous studies of patients with CMV-associated GBS found that 5% to 29% had IgM antibodies to GM1, and antibodies to ganglioside GM2 were also usually IgM.26-28⇓⇓
Adhesion molecules are essential for T cell activation and transendothelial migration in GBS and EAN.29,30⇓ We measured the soluble extracellular portions of cytokine receptors because these are more stable in stored serum than cytokines themselves. CMV infection was associated with raised concentrations of adhesion molecules and sIL-2R, suggesting greater activation and migration of T lymphocytes than C. jejuni or no identified infection. Raised concentrations of sTNF-R and sIL-2R were associated with a worse outcome, suggesting that a Th1 response is harmful.9 Concentrations of sIL-4R (which inhibits the Th1 and promotes the Th2 phenotype) and IL-10 (which suppresses macrophages and Th1 activity) did not differ between neurophysiologic groups or following different infections.
Although we confirmed that patients with C. jejuni infection were more likely to develop axonal neurophysiology or inexcitable nerves than patients with CMV or no identified infection,3-5,19⇓⇓⇓ most patients with C. jejuni infection had demyelinating neurophysiology. What other factors might influence this? Patients with C. jejuni infection and axonal neurophysiology were younger and more likely to have IgM antibodies to ganglioside GM1 than patients with C. jejuni infection and demyelinating neurophysiology. However, we could not predict an individual’s neurophysiologic type with certainty from the clinical and immunologic features. The pathogenesis is likely to depend not only on the immunogenic components of the infecting organism, but also on the host’s immune response. Subtypes of GBS have overlapping features and some authorities prefer to consider them as a continuous spectrum. Concentrations of sTNF-R and sICAM-1 were raised in the group with inexcitable nerves, implying greater T cell activation with more severe neurophysiologic abnormality.31 Pure motor GBS had similar associations to axonal GBS,32 perhaps due to autoimmune attack on antigen(s) more abundant on motor axons, for example gangliosides GM133 or GD1a.4 We found no specific associations with the most common demyelinating subtype of GBS. Demyelinating GBS may be caused by any of the infections, and patients with any of the individual features associated with axonal neurophysiology were actually more likely to have demyelinating neurophysiology.
This study was primarily designed as a treatment trial, but the intent to do many of these subgroup analyses was expressed prospectively. Patients were selected only for GBS and willingness to participate in the treatment trial, and although only 60% could be included in the serologic analyses because sera were missing, we believe that this was a representative sample of all GBS patients.
Outcome.
We confirmed that a poor outcome is associated with serologic evidence of C. jejuni infection, diarrhea, older age,34 severe weakness, and small or inexcitable motor action potentials,3,35-38⇓⇓⇓⇓ and showed that poor outcome is also predicted by raised sIL-2R concentrations and absence of IgM antibodies to ganglioside GM1. Outcome may be affected by psychosocial factors39 and the development of chronic inflammatory demyelinating polyradiculoneuropathy as well as axonal degeneration. Death has been associated with older age and the presence of underlying lung disease.40 In a prospective study of 172 patients, 9% relapsed and relapse was associated with absence of antibodies to ganglioside GM1, smaller sensory action potentials, and proximal weakness.41 In this study, 5% of patients relapsed and none of the investigated factors predicted relapse.
Ganglioside GM1-like moieties occur on axons and Schwann cells.33 We confirmed that IgA and IgG antibodies to ganglioside GM1 were no longer an independent predictor of poor outcome when C. jejuni infection was also included in the multiple variable regression model.21 Japanese patients with these antibodies fell into two groups, recovering either faster or slower than antibody-negative patients.42 Antibodies to ganglioside GM1b had similar associations to GM1, except antibodies to GM1b but not GM1 were associated with slow recovery.43 It is difficult to explain why IgM antibodies to ganglioside GM1 should be a good prognostic factor, but it may be partly explained by confounding with other good prognostic factors that were not independently significant, particularly raised sL-selectin levels and CMV infection.
Retrospective subgroup analysis in the Dutch study suggested that patients with motor GBS, diarrhea, antibodies to ganglioside GM1 or GM1b, rapid onset, and CMV or C. jejuni infection might have a better outcome if treated with IVIg than with PE.8,15,32,38,43⇓⇓⇓⇓ In our study, patients with pure motor GBS had a better outcome if treated with both PE and IVIg compared with PE alone, but IVIg alone was not significantly better than PE. Little weight should be put on these analyses until they are confirmed on another group of patients or, preferably, tested in a trial with prospectively defined subgroups.
Acknowledgments
Supported by Novartis, the Guillain–Barré Syndrome Foundation International, and the BIOMED Inflammatory Neuropathy Cause and Treatment Program (R.D.M.H.), and the Gemeinnützige Hertie Foundation and Würzburg University Research Funds for all Würzburg-based coauthors and for laboratory analyses.
Acknowledgment
The authors thank H. Klüpfel, M. Piepenschneider, and K. Zehe for expert technical assistance in the determination of antiganglioside antibodies; T. Fritz, C. Köhler, M. Schorn, and B. Steinmetz for expert technical assistance in the assays for CMV and EBV; and K.W. Pflughaupt, PhD, for organizational help.
Footnotes
- Received May 2, 2000.
- Accepted November 21, 2000.
References
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan