Bacteria of the genus Campylobacter are important human pathogens and are common causes of gastrointestinal illness in both developed and developing countries [9]. Extraintestinal complications of Campylobacter jejuni infection such as reactive arthritis, pancreatitis, and carditis are well described [10], and in the last decade, clinical and epidemiologic evidence has suggested that infection with C. jejuni may be a precipitating factor for development of the Guillain-Barre syndrome. In small serologic studies of patients with the Guillain-Barre syndrome, as many as one third to one half of patients had increased levels of C. jejuni antibodies at the time of onset of neurologic symptoms [11-13]. Routine stool cultures of eight patients with the Guillain-Barre syndrome in Japan yielded C. jejuni in seven (88%) patients [14]. In an important U.S. investigation of 106 patients with the Guillain-Barre syndrome, C. jejuni was isolated from 4 (44%) of 9 patients who had antecedent clear-cut diarrheal illness; however, cultures had not been obtained from 97 other patients [15].

To assess the extent to which the Guillain-Barre syndrome is associated with recent C. jejuni infection, we did serologic screening with defined assays to determine the frequency of C. jejuni antibodies in a large group of patients with the syndrome and in two control groups.

Methods

Top Methods Results Discussion Author & Article Info References

Study Population

Case Patients

Case patients included 126 persons with the Guillain-Barre syndrome admitted to the University of Maryland Medical Center, the University of Medicine and Dentistry of New Jersey, or to one of several hospitals in St. Louis between 1983 and 1990. A consecutive sample was used. All patients met the National Institute of Neurologic and Communicative Disorders and Stroke [16] criteria for diagnosis of the Guillain-Barre syndrome; none had any exclusion criteria. Most of the patients had demyelinating polyneuropathy, although some had axonal damage. All had a monophasic illness that progressed during 1 to 4 weeks, plateaued, and almost all patients recovered to some extent; one patient died. Blood samples were collected within 3 weeks after the onset of neurologic symptoms from 118 of the patients; 8 patients had blood samples collected from 24 days to more than 3 months after onset of neurologic symptoms (median, 10 weeks). Because antibody response to C. jejuni infection is usually transient [17], specimens from these eight patients are not included in the analysis. Histories of symptoms of gastrointestinal or respiratory infections before onset of the Guillain-Barre syndrome were not available.

Controls

Two groups of controls were studied. One control group, a convenience sample, consisted of 56 patients with neurologic diseases other than the Guillain-Barre syndrome who were admitted to the neurology services at the University of Maryland, the University of Medicine and Dentistry of New Jersey, or one of the St. Louis hospitals. Diagnoses of these patients included optic neuritis, multiple sclerosis, amyotrophic lateral sclerosis, polyneuropathy associated with IgM paraproteinemia, chronic inflammatory demyelinating polyneuropathy, Charcot-Marie-Tooth disease, unexplained peripheral neuropathy, ischemic neuropathy, neuropathy associated with diabetes, and drug-induced neuropathy. The second control group, a voluntary sample, included 18 healthy employees of the neurology departments of the three universities and 29 healthy family members of case patients and of disease controls. Medical histories of healthy controls were not obtained.

Sera and Data Collection

Sera from case patients and controls was stored frozen at the collection sites until the serologic assays were done. Data collected included disease severity of case patients, sex, age, and date of sera collection. Information about race was available only from the patients at the University of Medicine and Dentistry of New Jersey.

Serologic Methods

Presence of serum IgA, IgG, and IgM antibodies specific for C. jejuni was determined by enzyme-linked immunosorbent assay (ELISA) as previously described [17]. Antigens from three C. jejuni strains of Penner (O-) types 1, 2, and 3, which are commonly isolated from humans, were prepared as described by McCoy and colleagues [18]; they contained a mixture of acid-extracted proteins common to C. jejuni strains [17, 19, 20]. The serum dilutions used were as follows: 1:50 for IgA, 1:200 for IgG, and 1:400 for IgM. The class-specific second antibody used (and the use dilution) was peroxidase-conjugated goat anti-human IgA (1:2000), IgG (1:2500), or IgM (1:1000) (Tago Inc; Burlingame, California). The plates were read at 414 nm on a Dynatech ELISA reader (Alexandria, Virginia), and results were expressed as an optical density value for each well. All sera had been coded and were assayed blindly three times on different days in duplicate wells for each run.

Calculation of Optical Density Ratios

The assays were standardized using sera from a patient with the Guillain-Barre syndrome and culture-proven C. jejuni infection (positive control) and sera from 38 healthy children in New York who had had no recent history of diarrheal illness (negative controls). Optical density values were determined serially for plates containing these sera, and the mean optical density of the positive control was plotted versus the mean optical density + 1 SD of the negative controls. For each immunoglobulin class, a regression line was calculated and then used each day the assay was done (using the same positive control serum) to generate a threshold value for determination of the optical density ratio. (Regression analysis for IgA assay, r = 0.973, P < 0.001; regression analysis for the IgG assay, r = 0.998, P < 0.001; regression analysis for the IgM assay, r = 0.981, P < 0.001.) To control for day-to-day assay variation, the optical density ratio was defined as the ratio of the optical density of each unknown serum sample to the observed threshold on that day. A similar method has been used to standardize Helicobacter ELISA determination [21].

Case Definitions and Statistical Methods

A positive serologic response was defined as an optical density ratio greater than or equal to a specified threshold in one or more immunoglobulin classes. Differences in the proportion of positive responses among case patients and controls were ascertained using an optical density ratio 1, 2, or 3 as the threshold. Statistical analyses were done in a univariate manner using the Mantel Haenszel Chi-squared analysis or the Fisher exact test when indicated and by the calculation of odds ratios.

The sensitivity and specificity of the assays were determined as follows. Optical density ratios were determined as described above for 17 patients in the convalescent phase of culture-proven C. jejuni infection. These were U.S. soldiers who had become infected while on military exercise in Thailand in 1988; convalescent sera were obtained 4 weeks after onset of diarrhea. The comparison group comprised 19 healthy adults who were employees of or visitors to the Infectious Diseases Division of Vanderbilt University and included 20 healthy children in Nashville who had no recent history of diarrheal illness.

Results

Top Methods Results Discussion Author & Article Info References

Patients

Demographic characteristics of the 118 case patients and 103 controls are shown in Table 1. Age was known for 112 (95%) case patients and 93 (90%) controls; gender was known for 116 (98%) case patients and for 99 (96%) controls. Information about the month blood was drawn was available for 116 (98%) case patients and 80 (78%) controls. The groups were similar except that controls were somewhat younger (median age, 33 years compared with 41 years for case patients) and were more likely to have had their blood drawn during the summer months. Race of patients and controls (from the University of Medicine and Dentistry of New Jersey only) also was similar in the two groups. In both groups there were more male patients.

Among 17 persons with culture-proven C. jejuni infections, in the IgA assay, 15 (88%) had an optical density ratio 1. Increasing the threshold for defining a positive serologic response to an optical density ratio 2 and 3 decreased the sensitivity of the test (Table 2). The specificity of the test was evaluated by the serologic responses in the control group. Defining a positive serologic response as an optical density ratio 1, the IgA assay was 85% specific (specificity determined by subtracting the percentage of controls exceeding threshold from 100%) (see Table 2). Increasing the threshold to an optical density ratio 2 or 3 improved the specificity of the test. The sensitivity and specificity of the IgG and IgM assays are also shown in Table 2. As expected, raising the stringency by increasing the optical density ratios or numbers of positive classes required to define a positive serologic response decreased the sensitivity but improved the specificity of the assays.

Serologic Assays in Patients with the Guillain-Barre Syndrome and in Controls

Fifty-eight percent of the case patients (69 of 118) had an optical density ratio 1 in the IgA assay compared with 35% (36 of 103) of the controls (odds ratio, 2.6; P < 0.001) (Table 3). In the IgG assay, 56% of case patients and 41% of controls had an optical density ratio 1 (odds ratio, 1.9; P = 0.02). In the IgM assay, 61% of case patients and 42% of controls had optical density ratios 1 (odds ratio, 2.2; P = 0.004). When the threshold for defining a positive response was increased to an optical density ratio 2 or 3, the strength of the association between the Guillain-Barre syndrome and serologic evidence of previous C. jejuni infection also increased. Seventy (59%) of case patients and 39 (38%) of controls had an optical density ratio 1 in two or more immunoglobulin classes (odds ratio, 2.3; P = 0.001). Increasing the required optical density ratio to 2 or 3 and increasing the number of immunoglobulin classes that must exceed the threshold for a serum sample to be classified as having a positive serologic response increased the strength of the association between the Guillain-Barre syndrome and serologic evidence of previous C. jejuni infection (see Table 3). Distribution of the optical density ratios was analyzed using the most specific definition of a serologic response that still provided more than 50% sensitivity, which was an optical density ratio 2 in two or more classes (specificity = 97%; see Table 2). Use of this criterion for seropositivity showed that patients with the Guillain-Barre syndrome were 5.3 times more likely to have serologic evidence for C. jejuni infection than were the controls; we used this definition in all subsequent assays.

Epidemiologic Characteristics of Patients with the Guillain-Barre Syndrome Who Had Campylobacter jejuni Infection

Male patients with the Guillain-Barre syndrome were more likely to have had serologic evidence of previous C. jejuni infection. Of 67 male patients with the Guillain-Barre syndrome, 31 (46%) had positive C. jejuni serologic results compared with 11 (22%) of 49 female patients (odds ratio, 3.0; P = 0.009). In contrast, no differences were observed in the rate of seropositivity between male and female controls. The association between the Guillain-Barre syndrome and seropositive responses was observed in all age groups but was strongest in persons older than 60 years (Table 4). The rate of seropositive responses among patients with the Guillain-Barre syndrome increased from 15% in those younger than 20 years to 47% in those older than 60 years; among controls, the rate of seropositive responses did not increase with age (see Table 4). In a separate analysis, the mean age of patients with the Guillain-Barre syndrome and a positive seropositive response (42 years) was greater than that of patients with the syndrome who had a seronegative response (36 years) (P = 0.02).

An association between the Guillain-Barre syndrome and previous C. jejuni infection was observed in all seasons but reached statistical significance only in summer and autumn (Table 5). Because the peak incidence of C. jejuni infections in the United States occurs from August through October [22] and because in previous anecdotal case reports the Guillain-Barre syndrome had occurred from 1 to 3 weeks after C. jejuni infection [23-25], we compared the proportion of serologically positive patients who had the syndrome occurring from September to November with those occurring during the rest of the year. Of 39 patients with the Guillain-Barre syndrome who had onset of symptoms from September to November, 20 (51%) had serologic evidence of previous C. jejuni infection, compared with 22 (29%) of 77 patients with onset of neurologic symptoms during other months (odds ratio, 2.6; 95% CI, 1.10 to 6.25; P = 0.02).

Discussion

Top Methods Results Discussion Author & Article Info References

Our study showed a strong association between the Guillain-Barre syndrome and serologic evidence of recent C. jejuni infection. Although the assay used was not highly sensitive or specific, increasing the specificity of the test by increasing the required optical density ratio for classification as a positive result increased the strength of the association. Furthermore, in our study the greatest proportion of patients with the Guillain-Barre syndrome who had C. jejuni antibodies occurred during the autumn months, which immediately follow the period of peak incidence of culture-proven C. jejuni infections [22]. These observations suggest that the assays were detecting specific C. jejuni antibodies in truly infected persons.

Assay Validation

A number of features lend further credence to our results of this study. First, the assays were done using standardized techniques in a blinded manner, substantially reducing the possibility of investigator bias. Each serum sample was assayed several times in duplicate to assure reproducibility of results. Regression analyses provided more precise thresholds for determination of positive results than were used in previous investigations [11-13]. The number of sera tested was large, and samples were obtained from three university medical centers making it unlikely that the results represented an institutional or regional phenomenon.

Description of Patients and Controls

The patient and control groups were similar for gender and race but were somewhat different for age (the control group was younger) and for the time of year blood was obtained (the control group was more likely to have blood drawn during the summer). However, because the incidence of C. jejuni infections is greater among younger adults and during the summer months [22], any bias introduced by these differences would have resulted only in an underestimation of the association between the Guillain-Barre syndrome and serologic evidence of C. jejuni infection. Furthermore, because C. jejuni infection may be present among household contacts, use of family members of patients with the Guillain-Barre syndrome as controls also may have resulted in an underestimation of the association between the syndrome and C. jejuni infection. Indeed, the optical density ratios among the healthy control group were greater than those observed in the disease controls and in the healthy control group used initially to determine the accuracy of the assay.

The distribution of seropositive responses among patients with the Guillain-Barre syndrome in our study suggests that subgroups of patients may exist who are more likely to have had previous C. jejuni infection. The proportion of patients with the Guillain-Barre syndrome and C. jejuni antibodies was greatest in male patients, in persons infected in summer months. The proportion also increased with age. Although C. jejuni infections are known to peak during late summer [22], male gender and older age may represent additional risk factors for the development of the Guillain-Barre syndrome after infection with C. jejuni.

Previous Studies

Because the duration of convalescent excretion of C. jejuni is brief (mean, 16 days after onset of diarrhea) [22], and the Guillain-Barre syndrome associated with C. jejuni typically occurs 1 to 3 weeks after onset of diarrhea [24-27], reliance on isolation of C. jejuni from fecal cultures is an insensitive measure of previous infection. Other investigators have used serologic tests to determine the frequency of C. jejuni infection as an antecedent to the Guillain-Barre syndrome. However, the studies have been limited by use of nonstandardized methods, absence of appropriate controls, or failure to "blind" the investigators to the source of the sera studied [11-13]. Using immunodot assays [28] to determine the frequency of C. jejuni antibodies, Gruenwald and colleagues [11] found that 3 (18%) of 17 patients with the Guillain-Barre syndrome had increased titers in two or more immunoglobulin classes. Similarly, using a complement fixation technique, Winer and colleagues [5] found that 14% of 99 patients with the Guillain-Barre syndrome had positive C. jejuni serologic test results. However, the absence of control groups in these investigations makes it difficult to interpret these results. Increased C. jejuni antibodies were found in 41% of 32 children diagnosed with the Chinese paralytic syndrome, but the control groups studied were from Thailand [29], and the relevance of this disease to the Guillain-Barre syndrome has not been established [30]. Kaldor and Speed [12] used nonstandardized methods to study serum from 56 patients with the Guillain-Barre syndrome and from 57 controls; in their unblinded analysis, they found 38% of the patients and none of the controls met their criteria for positive serologic responses. Despite their limitations, each of these previous studies yielded results consistent with our findings.

Incidence and Pathologic Mechanism

The incidence of the Guillain-Barre syndrome is slightly higher in men [4], and in our study, male patients outnumbered female patients, 3 to 2. The incidence of Campylobacter infection is also slightly higher in men [22]. Male patients with the Guillain-Barre syndrome were three times as likely as female patients to have serologic evidence of previous C. jejuni infection. Host factors, including immunologic response to C. jejuni infection that is genetically linked [31], may predispose patients to the development of the Guillain-Barre syndrome after C. jejuni infection. Of 6 patients with the Guillain-Barre syndrome in Japan who had antecedent C. jejuni infection, all had the HLA-B35 antigen, compared with only 14% of 3090 healthy controls [32]. Previous studies of patients with the Guillain-Barre syndrome had not shown an association with a particular HLA type [33], but the data had not been stratified for presence of C. jejuni infection.

The hallmark of the Guillain-Barre syndrome is segmental demyelination of peripheral nerves accompanied by mononuclear infiltrates and edema. However, the pathogenic mechanism involved in the Guillain-Barre syndrome after C. jejuni infection is not known. Some investigators have suggested that relatively few C. jejuni strains are capable of triggering an immunologic response that is associated with myelin destruction. The findings of two Japanese groups indicated that a particular serotype of C. jejuni, Penner type O19, was associated with the initiation of the Guillain-Barre syndrome [34, 35]. Whether the O19 strains represent a particularly virulent clone, or whether the O19 polysaccharide is itself a virulence determinant or a marker for other virulence factors is not known. The neurologic target for immunologic injury also is unknown. Fujimoto and Amako [36] showed that serum from a patient who acquired the Guillain-Barre syndrome shortly after a C. jejuni infection reacted strongly with P0, a peripheral nerve myelin-specific protein. They suggested that C. jejuni has antigenic characteristics that stimulate production of antibodies that react with peripheral nerve myelin and cause the Guillain-Barre syndrome. Similarly, other investigators have proposed that gangliosides found in myelin may be the ultimate target of antibodies formed in response to C. jejuni infection. Thus, anecdotal reports exist of the Guillain-Barre syndrome occurring after administration of parenteral gangliosides [37, 38].

Conclusion

It is difficult to estimate the exact number of patients with the Guillain-Barre syndrome who had previous C. jejuni infection. Previous epidemiologic studies have found that 10% to 30% of patients with the Guillain-Barre syndrome have antecedent acute gastrointestinal illnesses. Our results suggest that the specificity of the test is close to 100% (the number of controls exceeding threshold is close to zero) when the definition of a positive response is an optical density ratio 3 in two or more immunoglobulin classes or optical density ratio 2 in all three classes. Using this most stringent definition of "positive," about 15% to 20% of patients with the Guillain-Barre syndrome are classified as positive; this provides a conservative estimate of the number of patients with the syndrome who also had C. jejuni infection. Although the biological significance of this association cannot be established by this study, our results provide further evidence that C. jejuni infection is a common initiator of the Guillain-Barre syndrome.

Request For Reprints: Ban Mishu, MD, Infectious Diseases Division, A3310 Medical Center North, Vanderbilt University School of Medicine, Nashville, TN 37232.