Zika virus (ZIKV), a mosquito-borne RNA flavivirus, has caused a major outbreak in the Americas that began in 2014.1 ZIKV infection manifests as a self-limited febrile syndrome associated with rash, conjunctivitis, and arthralgias.2-4 In 2013 and 2014, an increase in the number of cases of the Guillain–Barré syndrome was observed during an outbreak of ZIKV infection in French Polynesia.5,6 Recently, clusters of the Guillain–Barré syndrome and microcephaly have been spatially and temporally related to the current outbreak of ZIKV infection in the Americas.7 In Colombia, the government reported the first autochthonous case of ZIKV infection in October 2015.8 In December 2015, the Colombian Instituto Nacional de Salud (INS) documented an unusual number of cases of the Guillain–Barré syndrome in the Caribbean and the northeastern regions of Colombia. By January 2016, the outbreak of ZIKV infection had spread to most regions of Colombia. Concomitantly, an increase in the number of neuroinflammatory disorders was reported.7 Here, we describe an observational clinical and virologic study of the Guillain–Barré syndrome cases that were evaluated in the context of the ZIKV outbreak in Colombia, which further supports the association between ZIKV infection and the Guillain–Barré syndrome — in particular, the acute inflammatory demyelinating polyneuropathy (AIDP) form of the syndrome.
During the outbreak of ZIKV infection in Colombia, all patients in whom the Guillain–Barré syndrome was diagnosed at six university-based centers from January through March of 2016 were evaluated prospectively as part of the Neuroviruses Emerging in the Americas Study (NEAS) (see the Supplementary Appendix, available with the full text of this article at NEJM.org). Patients underwent clinical and neurologic evaluation by internal medicine and neurology specialists. Nerve-conduction studies and electromyography were performed as part of the standard of care, and the results were classified in accordance with previously established criteria.9,10 Samples of blood and cerebrospinal fluid (CSF) were obtained as part of the standard of care and, when available, aliquots of these samples along with urine were used for virologic testing for ZIKV infection. The clinical and laboratory information was documented with the use of standardized questionnaires (NEAS forms), as well as the Spanish version of the evaluation form of the International GBS Outcome Study (IGOS) (see the Supplementary Appendix). The diagnosis of the Guillain–Barré syndrome was based on the Brighton Collaboration GBS Working Group criteria.11 Brighton criteria levels indicate the certainty of a diagnosis of the Guillain–Barré syndrome. Level 1, in which the diagnosis is supported by nerve-conduction studies and the presence of albuminocytologic dissociation in CSF, indicates the highest degree of certainty. A level 2 diagnosis is supported by either a CSF white-cell count of less than 50 cells per cubic millimeter (with or without an elevated protein level) or nerve-conduction studies consistent with the Guillain–Barré syndrome (if the CSF white-cell count is unavailable). A level 3 diagnosis is based on clinical features without support from nerve-conduction or CSF studies.
Because all the patients with the Guillain–Barré syndrome we studied were residing in areas that were endemic for mosquito-borne virus transmission, their illnesses were suspected to be associated with ZIKV disease as defined by the Pan American Health Organization (PAHO).12 In patients with a diagnosis of the Guillain–Barré syndrome fitting level 1, 2, or 3 of the Brighton criteria, the diagnosis of ZIKV infection was defined as definite, probable, or suspected. Definite cases of ZIKV infection were those that were confirmed by a positive real-time reverse-transcriptase–polymerase-chain-reaction (RT-PCR) assay for ZIKV RNA in blood, CSF, or urine. Probable cases were those that were characterized by positive results of enzyme-linked immunosorbent assays (ELISAs) for antiflavivirus antibodies in the CSF, serum, or both but negative results of RT-PCR for ZIKV and for the four dengue virus (DENV) serotypes. Suspected cases were characterized by a clinical syndrome compatible with ZIKV infection with two or more features of the PAHO case definition12 (rash, fever, nonpurulent conjunctivitis, arthralgia, myalgia, and periarticular edema) without laboratory confirmation. To characterize the temporal profile of the disorder, the onset of suspected ZIKV infection was defined as the day of onset of systemic symptoms outlined in the case definition. The onset of neurologic symptoms was defined as the first day of onset of limb weakness, sensory symptoms, facial paralysis, or other neurologic symptoms.
Virologic testing was performed at the Virology Laboratory, Universidad del Valle, Cali, Colombia. The TaqMan RT-PCR assay used for the diagnosis of ZIKV infection was based on a protocol from Lanciotti and colleagues.13 Serum, CSF, and urine were considered to be positive for ZIKV if the two distinct genomic regions targeted by the RT-PCR were amplified. Serum and CSF samples were also tested for the four DENV serotypes by means of nested RT-PCR.14,15 DENV IgM-capture and IgG-capture ELISAs (Panbio Diagnostics) were performed to detect the presence of flavivirus cross-reactive antibodies. A patient was defined as having had a recent flavivirus infection if an ELISA for IgM or IgG was positive in any of the examined fluids.16 To determine the presence of ZIKV infectious particles, viral isolates were obtained from ZIKV RT-PCR–positive serum and urine samples from four patients and cultured in C6/36 Aedes albopictus cell and Vero cell lines. Inoculated cells were cultured for at least 14 days, with imaging performed once daily by microscopy to assess cytopathic changes,17 and the culture supernatants were tested for ZIKV by RT-PCR (details are provided in the Supplementary Appendix).
The study protocol was approved by the institutional review board at the Johns Hopkins University School of Medicine and by the ethics committee at each participating center. The ethics committee of each participating center provided research guidelines, and either oral or written consent was obtained from all patients.
From October 2015 through March 2016, there were 2603 laboratory-confirmed ZIKV infections in Colombia and more than 58,790 suspected cases. In addition, there were 401 patients with a neurologic syndrome who had a history of ZIKV infection; 270 of the cases (67%) corresponded to the Guillain–Barré syndrome18 (Figure 1Figure 1Cases of ZIKV Infection and the Guillain–Barré Syndrome in Colombia.). On the basis of data from the registry of individual records of health care services, the INS estimated that approximately 250 cases of the Guillain–Barré syndrome per year occurred in the whole country between 2009 and 2015, for a mean of approximately 20 cases per month (unpublished data). The frequency was increased relative to that baseline rate during the ZIKV outbreak, during which more than 270 cases of the Guillain–Barré syndrome were registered up to epidemiologic week 12 of 2016, for a mean of approximately 90 cases per month.19 (In a given year, epidemiologic week 1 ends on the first Saturday in January, as long as it falls at least 4 days into the month.) According to surveillance data from the INS, DENV had circulated in Colombia during the last decade and caused periodic outbreaks. A chikungunya virus outbreak occurred in the region during most of 2015 (see the Supplementary Appendix). However, it was not until the end of 2015 and the beginning of 2016 that ZIKV was first introduced to the region, and this period coincided with the first documented increase in the incidence of the Guillain–Barré syndrome (Figure 1).
A total of 68 patients who fulfilled the Brighton criteria for the Guillain–Barré syndrome and related variants and presented to the participating centers were included: 56 patients (82%) fulfilled level 1 or 2 criteria on the basis of evidence from CSF analysis, neurophysiological studies, or both. Four patients (6%) had the Miller Fisher syndrome, and 2 patients (3%) had other Guillain–Barré syndrome variants (bilateral facial palsy with areflexia and a pure sensory syndrome). The median age of the patients was 47 years (interquartile range, 35 to 57), 38 patients (56%) were male, and 61 patients (90%) were of mixed race. A total of 66 patients (97%) had symptoms of ZIKV infection in the 4 weeks preceding the onset of neurologic symptoms (Table 1Table 1Clinical and Demographic Characteristics of the 68 Patients with the Guillain–Barré Syndrome., and Table S1 in the Supplementary Appendix). Two patients did not report having had any systemic symptoms before the onset of the Guillain–Barré syndrome but were residents of a region affected by the ZIKV infection outbreak. The median duration of symptoms of ZIKV infection was 4 days; the condition manifested mainly with fever (in 69% of the patients), rash (59%), headaches (34%), myalgias (34%), nonpurulent conjunctivitis (25%), and arthralgias (22%). The median time between the onset of the ZIKV infection symptoms and the onset of the Guillain–Barré syndrome was 7 days (interquartile range, 3 to 10).
The clinical and laboratory features of the patients with the Guillain–Barré syndrome are summarized in Table 2Table 2Clinical and Laboratory Findings in the 68 Patients with the Guillain–Barré Syndrome., and in Table S2 in the Supplementary Appendix. The symptoms at presentation included limb weakness (97%), paresthesias (76%), and facial palsy (32%). A total of 56 patients (82%) reported an ascending pattern of weakness. On neurologic examination, the median Medical Research Council (MRC) sum score (which indicates muscle strength in 12 different muscle groups and ranges from 0 to 60, with higher scores indicating more preserved muscle strength) was 40 (interquartile range, 26 to 47).20 Cranial neuropathies were present in 43 patients, with bilateral facial palsy being the most common (in 50% of the 68 patients). Autonomic dysfunction was present in 21 patients (31%). A total of 40 patients (59%) were admitted to intensive care units, and 31% of all patients required mechanical ventilation. Treatment was administered to 46 patients (68%); intravenous immune globulin was the most commonly used treatment (62% of the 68 patients). Three patients (4%) died after respiratory failure and sepsis. The median modified Rankin score (which indicates the severity of disability and ranges from 0 to 6, with 0 indicating no symptoms and 6 indicating death) at nadir was 4 (interquartile range, 3 to 5).
Nerve-conduction studies and electromyography were performed with the use of standard techniques in 46 patients (68%). In accordance with published criteria,9,10 36 patients (78% of the 46 patients) were determined to have the AIDP subtype of the Guillain–Barré syndrome, 1 patient (2%) had the acute motor axonal neuropathy (AMAN) subtype, and 4 patients (9%) had equivocal studies that did not allow a subtype classification (Table 2). No abnormalities were noted in hematologic testing performed at admission. CSF analysis was performed in 55 patients (81%); the median white-cell count was 0 cells per cubic millimeter (interquartile range, 0 to 2.5), and the median protein concentration was 116 mg per deciliter (interquartile range, 67 to 171). A total of 45 patients (82%) had albuminocytologic dissociation in CSF, indicated by increased protein levels (>52 mg per deciliter) in the absence of pleocytosis (<10 cells per cubic millimeter).
Of the 68 patients, 42 (62%) underwent testing for ZIKV by RT-PCR in at least one of three biologic samples: urine (24 patients), serum (31 patients), and CSF (30 patients) (Figure 2Figure 2Laboratory Testing and Temporal Profiles of Illness in 42 Patients with the Guillain–Barré Syndrome during the ZIKV Infection Outbreak in Colombia. and Table 3Table 3Laboratory Studies for the Investigation of Flavivirus Infection in 42 Patients with the Guillain–Barré Syndrome., and Fig. S1 in the Supplementary Appendix). A total of 17 patients (40%) tested positive for ZIKV by RT-PCR; most of the positive results were in urine samples (16 patients). Three patients had positive ZIKV RT-PCR results in CSF (Fig. S2 in the Supplementary Appendix); only 1 patient had a positive result in serum, and this patient’s serum remained positive at 31 days after the onset of ZIKV infection (Patient 29 in Figure 2B). The median time from the onset of the symptoms of viral illness to the collection of the first ZIKV-positive urine sample was 16.5 days (interquartile range, 11.5 to 19.7), with 1 patient remaining positive at 48 days after onset (Patient 29 in Figure 2B). The results of RT-PCR for all four DENV serotypes were negative in the 39 patients tested.
ZIKV was cultured from the serum and urine of Patient 29 and from the urine of Patients 32 and 36 (Figure 2B) in C6/36 and Vero cell lines. The presence of ZIKV in the culture supernatants was confirmed by RT-PCR. Light microscopic imaging showed cytopathic changes consistent with flavivirus infection (Fig. S3 in the Supplementary Appendix). The profile of antiflavivirus antibodies is shown in Table 3 and Figure 2A, and in Table S3 in the Supplementary Appendix. A total of 32 of the 37 patients with the Guillain–Barré syndrome who were tested (86%) had evidence of a recent flavivirus infection, as indicated by the presence of cross-reactive IgM or IgG antiflavivirus antibodies. The pattern of expression of antiflavivirus antibodies stratified according to the results of the ZIKV RT-PCR is shown in Table S3 in the Supplementary Appendix. On the basis of clinical profiles and laboratory testing, the diagnosis of ZIKV infection was classified as definite in 17 patients, probable in 18 patients, and suspected in 33 patients (Table 1 and Table 3).
Of the 68 patients with the Guillain–Barré syndrome, 42 underwent laboratory testing for the identification of ZIKV infection. Figure 2 shows the laboratory and clinical temporal profiles of the infection in these 42 patients. The period from the onset of symptoms of ZIKV infection to the onset of neurologic symptoms and the time to nadir is outlined for each case. Two patients did not have any symptoms of ZIKV infection preceding the neurologic symptoms, and 2 patients had simultaneous onset of ZIKV infection and neurologic symptoms. A total of 20 patients (48%) in this group had a rapid onset of neurologic symptoms without an asymptomatic period after ZIKV infection symptoms (parainfectious onset), whereas the other patients had a variable asymptomatic period between ZIKV infection and the onset of neurologic symptoms (postinfectious onset).
The identification of the ZIKV genome by RT-PCR in biologic samples from 17 patients with the Guillain–Barré syndrome, together with the presence of immune responses (IgG, IgM, or both) to flaviviruses in the CSF in most of the patients tested, supports the involvement of ZIKV in these cases of the Guillain–Barré syndrome during the outbreak of ZIKV infection in Colombia. In addition, the clinical features of a preceding viral illness consistent with ZIKV infection and the evidence indicating that DENV infection was not present (i.e., negative RT-PCR results for the four DENV serotypes and the absence of laboratory features typical of DENV infections) are also supportive of such a relationship. However, the fact that there are cross-reactive antiflavivirus antibodies between DENV and ZIKV complicates the serologic assessment. The increase in cases of the Guillain–Barré syndrome during the time of the ZIKV outbreak in Colombia and the absence of such an increase while DENV and chikungunya virus were circulating within the region in previous years21-23 provides epidemiologic evidence of the link between ZIKV infection and the Guillain–Barré syndrome. Before our study, the most compelling evidence of an association between the Guillain–Barré syndrome and ZIKV infection came from a case–control study conducted during the 2013–2014 ZIKV outbreak in French Polynesia. In that study, 42 patients with the Guillain–Barré syndrome had serologic evidence of recent flavivirus infection.5
The clinical features of the Guillain–Barré syndrome that were observed during the Colombian outbreak of ZIKV infection, including a preceding viral illness of short duration (median, 4 days) in 97% of the patients, are similar to those described in French Polynesia.5 Similar to the symptoms seen in patients who had the Guillain–Barré syndrome during the ZIKV infection outbreak in French Polynesia, the neurologic symptoms at presentation in the patients in our series consisted of ascending limb weakness (82%), paresthesias (76%), and facial palsy (32%). In our study, 46 patients (68%) underwent electrophysiological studies, and the results of these studies were consistent with the AIDP form of the Guillain–Barré syndrome in 78% of these patients. This observation is consistent with the more classical presentation of the Guillain–Barré syndrome and contrasts with the AMAN form described from French Polynesia,5 a finding that may reflect a variable clinical phenotype of ZIKV-associated Guillain–Barré syndrome, evolutionary changes of the virus, or host-dependent factors in the two countries.
In our study, an analysis of the manifestation of neurologic symptoms among the patients with a diagnosis of definite or probable ZIKV infection suggests that the temporal profile of neurologic symptoms does not follow the classical postinfectious profile of the Guillain–Barré syndrome that is associated with other conditions, such as Campylobacter jejuni infection.10,24-26 Although the overall median time from the onset of the viral syndrome to the Guillain–Barré syndrome in our study was similar to that among cases in French Polynesia5 (7 and 6 days, respectively), an analysis of the temporal profile of the illnesses in the 42 patients who underwent laboratory testing showed that 20 patients (48%) had neurologic symptoms during or immediately after the viral syndrome associated with ZIKV infection. These observations suggest that in cases of the Guillain–Barré syndrome associated with ZIKV infection, the Guillain–Barré syndrome may follow the pattern of a parainfectious disorder rather than the classic postinfectious profile.24,26 The reason for this is uncertain, but possible explanations include the following: that ZIKV starts a process of immune molecular mimicry against nervous system antigens before the clinical symptoms of viral infection are manifested, that ZIKV produces immune dysregulation that leads to the Guillain–Barré syndrome through a mechanism or mechanisms not related to molecular mimicry, that ZIKV produces a hyperacute immune response, or that there are direct viral neuropathogenic mechanisms that are as yet unknown for the Guillain–Barré syndrome. Although the presence of ZIKV in the CSF and the replicating capability of the virus in three cases may suggest a ZIKV neuroinvasive process in the Guillain–Barré syndrome, more studies are needed to assess such a mechanism.
Another important observation in our study is the finding that in patients with the Guillain–Barré syndrome and definite ZIKV infection, there is a prolonged period of viruria, which persists for days after the viral syndrome is over. Although the frequency of detection of ZIKV genome in CSF and serum was low, the higher frequency of detection of ZIKV in urine makes this biologic sample one that can be considered potentially useful for the diagnosis of ZIKV infection. In our study, the median time between the onset of ZIKV infection symptoms to collection of the first urine sample that tested positive was 16.5 days, and in one of our patients, ZIKV viruria was observed up to 48 days after the onset of the viral syndrome. These observations are consistent with reported cases of prolonged ZIKV viruria in patients with the Guillain–Barré syndrome.27
We also found a potential relationship between the Guillain–Barré syndrome in association with ZIKV infection and previous exposure to DENV infection. A total of 32 of the 37 patients (86%) with the Guillain–Barré syndrome who were tested for antiflavivirus antibodies had evidence of a recent flavivirus infection, as indicated by positivity for antiflavivirus cross-reactive IgM antibodies, IgG antibodies, or both. The antibody titers detected by the IgG-capture ELISA are consistent with an anamnestic response to DENV (see the Supplementary Appendix). These data, along with negative DENV RT-PCR results, suggest that these patients had previously been exposed to DENV and that the ZIKV infection may have been a secondary flavivirus infection. There were 5 patients with the Guillain–Barré syndrome who had no detectable flavivirus antibodies; these patients may have had a primary flavivirus infection with ZIKV,13 as is suggested by the negative antibody profile of one patient who tested positive for ZIKV by RT-PCR in urine.
Our study provides virologic evidence of ZIKV infection in patients with the Guillain–Barré syndrome in Colombia. The onset of the Guillain–Barré syndrome can parallel the onset of systemic manifestations of ZIKV infection, indicating a so-called parainfectious onset, which suggests that factors different from the known postinfectious mechanisms may be present in ZIKV-related Guillain–Barré syndrome. Most of the patients had the AIDP form of the Guillain–Barré syndrome. Our results indicate that RT-PCR testing of urine is a valuable diagnostic tool for the identification of ZIKV infection in patients with the Guillain–Barré syndrome.