Guillain-Barré syndrome and dengue-like disease in 2015: temporal relationship in Piauí state and implications on Zika virus surveillance.

Teresina, February 28, 2017

Dear editor

Neurological symptoms of approximately two thirds of patients with Guillain-Barré syndrome (GBS) begin a few weeks after an apparently benign febrile infection with respiratory or gastrointestinal manifestations[1]. Using serological methods, several studies identified as major infectious agents involved in GBS: Campylobacter jejuni, Mycoplasma pneumoniae, Cytomegalovirus, Epstein-Barr virus and Haemophilus influenzae [2-5], but some flaviviruses with a documented circulation in Brazil were identified as triggers of GBS in the last two decades, especially dengue virus (DENV), West Nile virus (WNV) and Saint Louis encephalitis virus (SLEV)[4,6-9].

The introduction of the Zika virus (ZIKV) in Brazil was laboratory confirmed in May 2015 in the midst of an epidemic of a mild and sub-febrile exanthematic disease that affected especially inhabitants of the Northeast region of the country[10]. The disease was considered by some, at that time, as a mild form of infection by DENV. Thus, many symptomatic human infections by ZIKV may have been reported as dengue cases to health authorities. A few months later, the benign character attributed to ZIKV infection was questioned in face of its association with an increased number of GBS cases, microcephaly and other congenital malformations[11]. The first cases of Zika virus infection in Piauí State were laboratory confirmed in July 2015, but medical records point to its significant occurrence in that State up to five months before[12]. Phylogenetic analysis and molecular clock showed that the introduction of ZIKV in Brazil occurred between May and December 2013[13].

The local State Division of Epidemiology was already carrying out a surveillance program of neurological cases since the laboratory confirmation of the first West Nile human case in the country in August 2014, in Aroeiras do Itaim municipality – Piauí State – Brazil[9]. Since then, cases of aseptic meningitis, viral encephalitis, transverse myelitis and GBS have been submitted to epidemiological and laboratory investigation, which allowed monitoring the occurrence of these disorders[12].

Considering the State records on the occurrence of neurological diseases and data from the National System of Notifiable Diseases - Piauí section (SINAN-PI, in Portuguese), the existence of a temporal correlation between GBS time series and notifications of suspected cases of dengue in 2015 were verified. The cross-correlation analysis of the time series was chosen instead of a simple linear correction because: (i) a gap was expected between the previous infectious syndrome and the onset of neurologic manifestations, and (ii) the record of suspected dengue is more dynamic than the occurrence of GBS due to factors related to the natural history of each disease and the patient’s access to a medical notifier service. Correlation calculations were performed using the software Bioestat 5.0® and Free Statistic Software v1.1.23-r7[14].

In 2015, in Piauí State, 7,659 suspected dengue cases were reported in parallel with 42 cases of GBS. Serological tests (ELISA-IgM) were positive for dengue in 23.8% of GBS cases. There were no positive results for WNV. Testing for ZIKV had not yet been included in the GBS case investigation panel in 2015. Distributions of monthly notifications of suspected dengue and GBS cases in 2015 obeyed a monophasic up and down pattern, with the maximum occurrence reported for April and June, on an annualized rate of 5.5 GBS cases per 1,000 dengue notifications (Figure 1). The cross-correlation analysis (Figure 2) showed the maximum parallelism upon assigning a delay from one (rho = +0.78, CI 95% = 0.36 to 0.93; p = 0.003) to two (rho = +0.86, CI 95% = 0.59 to 0.96; p = 0.0003) months between time series. The maximum correlation between the time series of GBS cases and suspected dengue notifications recorded in Piauí State (1-2 months) is consistent with the description of the syndrome as a post-infectious phenomenon, characterized by an autoimmune attack on the myelin sheath triggered by an antigenic stimulation a few weeks before[1,4].

Figure 1

Reported cases of dengue fever, Guillain-Barré syndrome and monthly rainfall in the State of Piauí in 2015. (Piauí State Secretary for Health, Brazilian National Institute of Meteorology)

Figure 2

Cross-correlation between the time series of dengue-suspected cases reported monthly and hospitalizations due to Guillain-Barré syndrome in the state of Piauí, Brazil, 2015. The dashed lines indicate a 95% confidence interval.

It is possible that a significant proportion of notifications of suspected dengue cases have corresponded to the infection caused by ZIKV because there is great similarity between the clinical manifestations of these viral diseases. In addition, the availability of laboratory tests in Piauí municipalities is limited. Until the first laboratory confirmation of the infection caused by ZIKV in the country by the Ministry of Health in May 2015, the disease was little known by most health-care medical services and epidemiological surveillance teams in Piauí State. The first five months of 2015 concentrated 82% of dengue-suspected notifications in Piauí. In the same year, 22% of the 7,659 reported cases were serologically negative for dengue, and only 19% of the cases were confirmed by laboratory testing. The remaining cases were classified by clinical and epidemiological criteria. All the confirmatory laboratory tests were carried out by ELISA-IgM on serum samples[12]. A temporal relationship has been observed between the occurrence of peaks in rash Zika-simile and GBS outbreaks in several other regions[15-18].

It is difficult to quantify accurately the occurrence of Zika virus disease in Brazil in 2015. The universal notification of disease suspicion has only been established by health authorities in March 2016. So far, only laboratory-confirmed cases were reported, except for sentinel units deployed in some cities in the second half of 2015, when the epidemic had already decreased in Northeastern Brazil. The Ministry of Health estimates that up to 1.5 million Brazilians were affected by the Zika disease in 2015[19]. There is a likelihood of cross-reactivity in serological tests used to confirm dengue cases before non-recognized ZIKV infections - especially in patients with a previous history for that disease[20]. Therefore, even dengue cases classified as confirmed are subject to questioning due to the lack of a molecular biology diagnostic method and a simultaneous serological testing for ZIKV.

The inferences to be drawn from this study are limited. The proportion of dengue cases reported as suspicious in 2015, which corresponded to infections by Zika, remains in fact speculative. The similarities between the clinical presentations of the two diseases and the possibility of cross-reactions to serological examinations make the confirmation of cases by clinical, epidemiological and laboratory criteria are unpredictable when a diagnostic by molecular biology is dismissed. Until recently, there were no commercial kits for diagnosing ZIKV infection in the country, and detection of the viral genome by large-scale molecular biology is still impracticable. The magnitude of the occurrence of GBS is much lower than that of dengue - this did not allow the comparison of data in the temporal unit of an epidemiological week. This fact, combined with differences between the speed of recognition, the reporting of dengue and GBS and the ease of access to adequate diagnostic services to perform them, makes the delay calculated at the maximum cross-correlation subject to rough approximations. The time of higher occurrence of dengue in Piauí State coincides with the rainy season, which is also associated with more respiratory and diarrheal diseases, which are potentially involved in the pathogenesis of GBS.

There was a correspondence between GBS time series and notifications of suspected dengue in Piauí State in 2015. The delay of up to eight weeks between the correlated time series is compatible with the natural history of the syndrome and its post-infectious nature. In the face of evidence that ZIKV may act as an important GBS set-off and the unavailability (at that time) of notification and confirmation instruments of prior infection with Zika virus, the results of this study reinforce the need for clinical, laboratory and epidemiological study to differentiate DENV and ZIKV infections, encouraging systematic studies on the etiology of neurological infectious diseases in different regions of the country.