We found evidence for a spatiotemporal association between the outbreaks of unknown encephalitis in Bac Giang Province and litchi cultivation. The ecologic regression analysis demonstrated that the annual risk for AME in a commune increased with the proportion of litchi-cultivated surface and that the epidemics occurred earlier in the districts that harvested litchis during May–June than in those that harvested litchis during June–July.
Similar to the clinical features of Chandipura encephalitis (3
), i.e., acute encephalitis with rapid onset, those reported here suggest a viral etiology. Unfortunately, all viral investigations have thus far been uninformative. New techniques, such as high-throughput sequencing and resequencing micro-arrays, are currently performed on serum, CSF, culture supernatants, and brain homogenates from suckling mice inoculated intracerebrally with patients’ CSF. Although electron microscopy of brain tissue is a standard method of identifying new viral pathogens in encephalitis syndromes, we were unable to obtain brain tissue specimens through autopsy because of cultural barriers.
The association between litchis and acute encephalitis remains unclear. As with other emerging viruses, we face a multifactorial problem that seems to have litchi fruit production and harvest as its focal point. One possible scenario is that fruit-bearing litchi trees can attract bats, which might be the reservoir for the putative pathogen. Mosquitoes could feed on the infected bats and transmit the virus to humans who have insufficient protection against mosquito bites. Several species of bats were identified in the province, such as the frugivorous bats Rousettus leschenaultii
, which can feed on litchi. These bats’ highly gregarious, cave dwelling, and migratory characteristics facilitate their role in virus carrying and circulation (10
).The bat population density is high during April–September (which includes the time of litchi harvest) before migration during October–March.
In addition, several mosquito species were identified in Bac Giang Province: Anopheles vagus, Armigeres subalbatus, Culex tritaeniorhynchus, Cx. vishnui, Cx. gelidus, Cx. fuscocephalus, Cx. quinquefasciatus, and Cx. bitaeniorhynchus. Although the peak incidence of May–July correlates with the rice paddy breeding and development of Culex spp. mosquitoes, the paddy fields area was negatively associated with the risk for disease in univariate analysis. No data were collected on vector densities in Bac Giang.
Other modes of transmission, e.g., direct contact with litchis contaminated by bat saliva, urine, or guano (11
) or with other vectors, such as insects found in litchi trees or phlebotomine sand flies, as in the case of Chandipura virus (12
), cannot be excluded. Deforestation in Bac Giang to develop the growing of litchi trees because of their high economic value also might have disrupted the ecologic equilibrium of the province, leading to the emergence of a new vector-borne disease (13
). The putative virus also might be exclusively human; use of human feces as fertilizers to enhance litchi growth in these plantations might have contaminated the soil with enteroviruses, which are known to cause fatal encephalitis in deprived children (14
Last, a toxic origin might be possible; in India, a toxic weed, Cassia occidentalis
, caused an acute hepatomyoencephalopathy syndrome, which was first assumed to be viral encephalitis (15
). However, in our study, the presence of fever and meningeal symptoms and the absence of high elevation of liver enzyme favor a viral etiology rather than a toxic origin. Additionally, we demonstrated that children with AME harbor elevated levels of type I interferon in serum and interferon-inducible cytokines, such as IL-6 and IL-8 in CSF, compatible with an infectious process (16
); however, these cytokines are expected to be elevated in any inflammatory process.
The protective effects of poultry density and mean relative humidity during January–April (pre–epidemic period) are more difficult to interpret. With respect to the hypothesis of an arbovirus-mediated pathogenesis, a possibility is that poultry is a preferred host for the putative vector (18
). Regarding negative correlation with humidity, above a certain limit, high relative humidity can be harmful to insects; insects or their eggs can drown or be infected more readily by pathogens (19
). Litchi growing also depends on climate; ideal conditions include a brief dry, cool, and frost-free winter to lead to flowering, followed by warmer temperatures and moderate rainfall and humidity during fruit development and harvest (20
). Still, climatic variations alone could not explain the spatial differences in disease risk between communes of the same province. Moreover, even if the association is not proved to be causal, the persistence of both variables (litchi surface and mean humidity) in the final model suggests that both factors are independently correlated with the spatiotemporal patterns of the disease. Definitive identification of the infectious agent will help clarify these factors associated with AME incidence.
Because our study concerns an ecologic investigation, the relationship between litchis and AME cannot be inferred at an individual level. Another caveat of the study concerns the nonspecific case definition based on clinical features and negative JEV serologic test results. False-negative results have been observed for JEV serologic tests performed soon after onset of symptoms (21
); nevertheless, the widespread use of JE vaccination, introduced in 1997, has considerably reduced JE incidence in the region. In addition, after JE vaccine introduction in the WHO Expanded Program on Immunization, AES surveillance might have been intensified, and an apparent increased incidence of AES might be simply an artifact of more active surveillance.
Artifact is not likely to account for the cases described in this manuscript, however, because they were identified clinically among patients seeking care at the hospital with AES rather than through public health surveillance. Conversely, surveillance might have failed to capture some cases because of the rapid deaths of infected children (e.g., those who died before reaching the hospital).
Apart from these cases, given the severity of the disease, we can assume that all parents, including those who lived in remote villages, sought medical attention. Unfortunately, because of logistical difficulties, systematic data and sample collection and analysis could not be conducted for all case-patients in Bac Giang hospital. Nevertheless, we have no reason to believe that cases for which biologic data were available () differed from other cases.
Our ecologic analysis of outbreaks of acute encephalitis of unknown origin during the litchi harvest period in Bac Giang Province strengthens the hypothesis that litchis might play a role in these outbreaks by showing that litchi cultivation was spatially and temporally associated with AME. This finding can be useful to guide future prospective studies. The suggested role of litchi trees needs to be more thoroughly investigated to explain disease ecology. Further research should include investigating the specific agricultural practices linked to litchi cultivation, distribution of tasks among adults and children, locations where these activities are conducted, and other activities undertaken around litchi fields before and during the epidemic period; conducting entomologic surveys around litchi plantations; and analyzing potential reservoirs and hosts. Last, research efforts should be continued to identify the causative agent.