Figure

Figure. Wildlife trade sites and POFI sites (black circles) where wildlife samples were collected for study of zoonotic pathogens in wildlife traded in markets for human consumption, Laos. Provinces are labeled...

During December 2014–September 2017, we collected samples from 9 wildlife trade hotspots (8) and 2 roadside stalls (hereafter all referred to as trade sites) in Laos (Figure; Appendix Table 1). In addition, 3 Provincial Offices of Forest Inspection (POFI) collected samples from wildlife confiscated in markets by law enforcement. After identifying wildlife at trade sites (9), we asked vendors for permission to sample their animals. Depending on whether the animal was alive, dead, or butchered, we collected urogenital swabs, urine and blood samples, and kidney, liver, and spleen tissue samples (Appendix Table 2).

We extracted nucleic acid using QIAamp Viral RNA Mini Kits (QIAGEN, https://www.qiagen.com) with modifications (Appendix). We conducted PCRs targeting Leptospira spp., Rickettsia spp., Orientia tsutsugamushi, Anaplasmataceae, Ehrlichia chaffeensis, Anaplasma phagocytophilum, Coxiella burnetti, flaviviruses, hantavirus, dengue virus, Zika virus, and universal bacterial 16S rRNA (Appendix Table 3). Where necessary, PCR products were sequenced (Macrogen Inc., https://www.macrogen.com) and compared against GenBank through blastn (https://blast.ncbi.nlm.nih.gov). We performed descriptive, univariate, and multivariate analyses by using R version 3.6.2 (https://www.r-project.org). We assessed the effect of the wild meat processing status (alive, fresh, or frozen) on the risk for Leptospira detection by using a mixed effects logistic regression with species as random effect. Statistical significance was set at α = 0.05 (Appendix).

We collected 717 samples from 359 animals (trade sites: 461 samples from 324 animals; POFI: 256 samples from 35 animals); animals sampled were from >37 identifiable vertebrate species from 12 families (Appendix Table 4). Most were Sciuridae squirrels (73.0%, 262/359) and represented 16 species, most frequently Pallas’s squirrel (Callosciurus erythraeus) (20.3%, 73/359). From trade sites, 69 animals (21.3%, 95% CI 17.0%–26.2%) had >1 samples positive for >1 pathogens in 10 of 11 sites (90.9%, 95% CI 57.1%–99.5%) (Appendix Table 5). Of 324 animals tested, 65 (20.1%, 95% CI 15.9%–24.9%) were positive for Leptospira spp.; 4/41 were positive for Rickettsia spp. (9.8%, 95% CI 3.2%–24.1%), 0 for O. tsutsugamushi (0%, 95% CI 0%–10.7%), and 2 for Anaplasmataceae (4.9%, 95% CI 0.8%–17.8%) (Table 1). Positivity was higher among animals collected by POFI; 25/35 (71.4%) animals tested positive for >1 pathogens. Of those, 9 were positive for Leptospira spp. (25.7%, 95% CI 13.1%–43.6%), 20 for Rickettsia spp. (57.1%, 95% CI 39.5%–73.2%), 2 for O. tsutsugamushi (5.7%, 95% CI 1.0%–20.5%), and 6 for Anaplasmataceae (17.1%, 95% CI 7.2%–34.3%) (Table 2). Sequencing identified R. typhi, R. felis, R. conorii, an Anaplasma species (either A. centrale, A. capra, or A. marginale), A. platys, A. bovis, A. phagocytophilum, Ehrlichia chaffeensis, Lactococcus garvieae, and Kurthia populi (Tables 1, 2). No samples were positive for C. burnetii (0/76), flaviviruses (0/359), dengue virus (0/359), or Zika virus (0/358).

Among species for which >10 individual animals were sampled in trade sites, 2 had particularly high proportions of Leptospira spp.–positive specimens: the variable squirrel (Callosciurus finlaysonii) (13/28; 46.4% 95% CI 28.0%–65.8%) and the common palm civet (Paradoxurus hermaphroditus) (10/22; 45.5%, 95% CI 25.2%–67.3%). Leptospira spp.–positivity was higher in dry (50/195; 25.6%, 95% CI 19.8%–32.5%) than wet season (15/129; 11.6%, 95% CI 6.9%–18.8%) (χ² = 8.7; p = 0.003). Data disaggregation by species and province suggested that observed seasonality was driven by results in common palm civets and variable squirrels in Champasak Province. No association was detected between the probability of an animal testing positive for Leptospira and the animal being alive (3/22; 14%, 95% CI 3.6%–36%), freshly dead (58/293; 20%, 95% CI 16%–25%; p = 0.6), or frozen (4/9; 44%, 95% CI 15%–77%; p = 0.1). In a subset of Leptospira spp.–positive animals with multiple samples, 75% (18/24; 95% CI 53%–89%) of urogenital swab samples and 50% (9/18; 95% CI 29%–71%) of blood samples were positive (p = 0.11 by Fisher exact test). Rickettsia spp. were detected exclusively in solid organs (liver, kidney, and spleen).

Zoonotic pathogens were nearly ubiquitous across sites; 10/11 sites yielded >1 pathogens. Squirrels are frequently traded in Lao markets (8) and had the greatest pathogen diversity in this study. Leptospira spp. was identified most frequently, found in 20.1% of animals (>45% in variable squirrels and common palm civets). Variable squirrels are commonly traded, often in batches of 2 to 3 squirrels (8); hence, on average, someone purchasing 3 variable squirrels would have an 83% likelihood of buying >1 infected squirrel (p = 1 – (1 – prevalence)³ = 1 – 0.55³ = 0.83). The higher risk for Leptospira detection in the dry season is at odds with the typically described correlation of transmission with precipitation and flooding (10), suggesting that much remains to be understood of Leptospira ecology. Other studies have shown higher prevalence in rats (11), and although we are confident of the results from trade sites, storage of animals from POFI sites might have resulted in cross-contamination, which warrants cautious interpretation of results in this subset. Among Leptospira spp.-positive animals, detection was more likely in urogenital swab samples, highlighting the risk for transmission through infected urine (10). Although reservoir rodents are characterized by chronic renal infections, septicemia occurs during initial infection (10), and the high proportion of positive blood samples indicates a public health risk in relation to the consumption of uncooked or undercooked meat, organs, and blood. The PCR used to detect leptospires is specific for pathogenic and intermediate species (Appendix Table 3), but we could not confirm their human pathogenicity. The high volume of squirrel trade combined with high infection frequency suggests a high risk for exposure among wildlife consumers. Because leptospirosis is a key cause of fever in rural Laos (12), further work is needed to learn more about the relevance of contact with wildlife through trade and consumption.

The Rickettsiales species identified here are known to cause human infections in Laos (13). R. typhi causes murine typhus, a major underrecognized cause of fever (13). O. tsutsugamushi is responsible for up to 23% of fever (14), and although commonly associated with ground-dwelling rodents, the vectors (Leptotrombidium mites) parasitize squirrels (15), and O. tsutsugamushi has been isolated from Callosciurus notatus squirrels in Malaysia (16). Other bacteria identified are reviewed elsewhere (Appendix Table 6).

Although many of the human pathogens identified are transmitted by arthropod vectors, we found few arthropods in the wildlife sampled, probably because vectors leave animals quickly after animal death (17). Therefore, because most market vendors sell dead animals obtained from hunters or intermediaries (8), vendors are less likely to be exposed to disease vectors, and hunters are possibly at greater risk than market vendors or consumers. O. tsutsugamushi and R. typhi can cause infections through aerosol exposure, bites from infected animals, and needlestick injuries (18), but whether such routes of infection occur at trade sites is unclear. The frequent occurrence of Leptospira, which can be transmitted by direct contact with abraded skin and mucous membranes, may pose health risks to hunters, vendors, and consumers.