This study investigated the prevalence of an HEV infection in patients with GBS who had been treated in our large tertiary care university hospital in Southwestern Germany. Acute HEV infection was very rare in our study population (1.2%), particularly in view of the high seroprevalence of anti-HEV IgG antibodies (41%) indicating generally high HEV infection rates in our region. Based on positive IgM antibodies, both patients in our study had recent HEV infection, whereas none had evidence of current HEV infection as would have been indicated by positive PCR. With regards to considerable geographic differences in HEV infection rates , this is to the best of our knowledge the first study on this issue in a German population, whereas previous studies were performed in the Netherlands, Belgium, Japan and Bangladesh [10,11,12,13].
Apart from the high number of both patients (n = 163) and controls (n = 263) a particular strength of our study was the – compared to previous studies – low threshold to perform HEV PCR in both IgM positive patients and patients with negative serology presenting within 7 days after GBS symptom onset. The latter approach was chosen in order to cover “early cases” of an infection before seroconversion could take place. These cases might be missed if diagnosis would be exclusively based on serology [25, 26], as done in previous studies [10, 11]. In this regard, our approach is in line with current guidelines recommending both serology and PCR testing . For better understanding, the time course of various diagnostic markers of an HEV infection is described in more detail further below.
We regarded comparison of our results with those two studies from Europe, i.e. the Netherlands and Belgium [10, 11], appropriate, since these comply with ours concerning expected relative frequencies of GBS subtypes and HEV genotype (genotype 3 predominant in Europe) that both show relevant geographic differences [9, 23]. In addition, demographic parameters of patients were similar with a median age in the sixth decade and predominance of the male gender. Furthermore, disease severity in the Belgian cohort  was comparable (median GBS disability index of 3), whereas this data was not available for the Dutch study . In contrast to our data, the Dutch and Belgian studies found an acute HEV infection more frequently, namely in 5% (10/201) and 8% (6/73) of GBS patients, respectively. Also in contrast to our data, in the former study by van den Berg et al.  acute HEV infections were more frequent in GBS patients in comparison to controls (1/201 controls, 0.5%) whereas controls were missing in the Belgian study. In addition to regional variations in infection rates, there are other possible explanations for these divergent results: (1) diagnostic criteria for HEV infection, (2) clinical characteristics of the study population, and (3) serological assays used. In view of diagnostic criteria for acute HEV infection, Stevens et al. (Belgium) based the diagnosis exclusively on positive IgM antibodies , whereas van den Berg et al. (Netherlands) used similar criteria as applied in our study, namely positive IgM antibodies in combination with either positive IgG or HEV PCR . Our approach increases the specificity of the result, since HEV IgM antibodies show relevant cross-reactivity with other pathogens . Therefore, as the authors state themselves, the Belgian study might contain false-positive cases, particularly with regards to two patients who showed positive results for other infectious triggers of GBS with known cross-reactive antibodies (CMV, EBV) . Furthermore, clinical characteristics varied between study populations. Van den Berg et al. (Dutch study) included patients with classical GBS only, whereas both Stevens et al. (Belgian study) as well as our study also included patients with MFS and other rarer variants (acute motor axonal neuropathy, AMAN, and acute motor sensory axonal neuropathy, AMSAN), which constitute a minority of GBS patients [11, 12, 23]. Overall, relative frequencies of subtypes in our study were representative, since they were consistent with previous epidemiological studies from Europe [23, 25]. Finally, use of different assays for HEV antibody testing might affect results . The recomWell assay (Mikrogen) used in the Belgian study has higher sensitivity for anti-HEV IgM, but lower sensitivity for anti-HEV IgG compared to the Wantai assay used in the Dutch and our study [29, 30]. To summarize these comparisons of serological testing, our study reveals highest concordance with the Belgian study by Stevens et al. with regards to similar demographic (age, sex) and clinical (GBS subtype and severity) characteristics of the study population. Therefore, different prevalence of an acute HEV infection (1.2% vs. 8%) might be due to regional differences, but might also in part be due to different assays, since the Belgian study used a test with higher sensitivity for IgM antibodies, which potentially increases the risk for false positive results. By contrast, the Dutch study by van den Berg et al. that also revealed a higher IgM seroprevalence (5%) used the same assay as in our study, but included patients diverging with regards to clinical characteristics (patients with classical GBS only). However, it is rather speculative, whether these clinical differences explain differences in the seroprevalence of an acute HEV infection, since none of the studies compared seroprevalence rates of patients with classical GBS and its variants.
The basic question arising from ours as well as previous studies is on how to “prove” an association between HEV infection and GBS assuming that the latter has a parainfectious etiology. For this, HEV infection would have to precede GBS by approximately three to 6 weeks, which is the typical time interval given for other infectious triggers (e.g. gastrointestinal or respiratory tract infections) [1, 14]. However, HEV infections are asymptomatic in 95% of cases . Hence, determining the onset of an HEV infection and therefore temporal relation with GBS, particularly in retrospect, is impossible on clinical grounds, but relies on laboratory tests, i.e. serological and PCR testing. In this regard, incubation time of the virus is 15 to 60 days . Early in the course of the infection, PCR becomes positive, and remains positive for three to 6 weeks (so-called viremia) [27, 31]. In comparison, IgM and IgG antibodies become positive in the second week of viremia, whereas they might be still negative in “very early cases” of an infection [26, 32, 33]. Interestingly, such “very early cases” during the phase of viremia with positive PCR and negative IgM antibodies have been described in HEV-associated neurological injury . These cases were covered by our approach requiring PCR testing in all patients presenting during early stages, i.e. within 7 days from GBS symptom onset. Furthermore, IgM might remain positive for up to 12 months (three to 4 months on average), and IgG even for years [27, 34]. Hence, in retrospect and particularly in asymptomatic cases of an HEV infection, a positive HEV PCR would best determine a close temporal relation and therefore association to GBS, whereas in cases of positive IgM and IgG antibodies, but negative PCR, this would still be possible, but not “proven” [18, 27]. In this regard, HEV PCR indicating current infection was positive in none of our patients, 3/10 IgM positive patients in the Dutch study  and was not performed in any of the six IgM positive patients in the Belgian study . Consequently, current HEV infection in close temporal association to GBS is not excluded in the other IgM positive patients, but has thus far only been proven in three patients from the Dutch cohort.
With regards to potential pathophysiological mechanisms, by which HEV might trigger GBS, HEV PCR was negative in all 75 CSF samples tested. Since also in former systematic studies testing for HEV RNA in GBS patients [10, 12] none had positive PCR in CSF samples indicating direct infection, the virus most likely triggers the disorder indirectly by molecular mimicry as has been described for other pathogens . In contrast, evidence for direct viral infiltration of the nervous system in the clinical setting in GBS patients as would be suggested by its neurotropic properties based on experimental studies [19,20,21] remains thus far anecdotal [16, 17].
The reason for the unexpectedly high IgG seroprevalence in our study remains unclear, even when taking seroprevalence variability dependent on birth year, age  and serological assay  into consideration. It was comparable to the Dutch study (46%) , but considerably higher in comparison to the Belgian study (18%) , and to previous studies investigating infection rates in the general population in the Netherlands (29%) , in Belgium (4–14%) [36, 37], and in Germany (6–24%) [2, 38]. However, high seroprevalence rates indicate high infection rates and thus further emphasize the importance of HEV as a potential trigger of neurological disorders.
The current study has several limitations due to the retrospective, monocentric design. Specifically, serum and CSF samples were not available in some patients, and information about risk factors for HEV infection as consumption of uncooked meat was missing. Additionally, patients were not systematically screened for other infectious triggers of GBS in order to estimate the frequency of potentially cross-reactive antibodies, and the control group was not age- and sex-matched. Concerning the monocentric design, our tertiary care university hospital has a very large catchment area of approximately 2,500,000 inhabitants, and therefore we would regard it as representative for the region of Southwestern Germany.
In summary, in our cohort seroprevalence of an acute HEV infection in patients with GBS was rare and not different from controls. Furthermore, a close temporal association as would be suggested by positive HEV PCR could not be established in any of these patients. Therefore, our data apparently challenge current European guidelines, which – based on data from previous studies – generally recommend HEV testing in GBS patients . Nevertheless, high IgG seroprevalence in our cohort and the control group argues for generally high infection rates with HEV in Southwestern Germany. This underlines the importance to gain more information about the association with GBS as a severe and potentially life-threatening disease that can lead to long-term disability . In this regard, large prospective multicenter studies would be necessary to draw more reliable conclusions.