Leptospira infection and shedding in dogs in Thailand

Background Leptospirosis is a widespread zoonosis and has been recognized as a re-emerging infectious disease in humans and dogs, but prevalence of Leptospira shedding in dogs in Thailand is unknown. The aim of this study was to determine urinary shedding of Leptospira in dogs in Thailand, to evaluate antibody prevalence by microscopic agglutination test (MAT) and enzyme-linked immunosorbent assay (ELISA), and to assess risk factors for Leptospira infection. In Northern, Northeastern, and Central Thailand, 273 stray (n = 119) or client-owned (n = 154) dogs from rural (n = 139) or urban (n = 134) areas were randomly included. Dogs that had received antibiotics within 4 weeks prior to sampling were excluded. No dog had received vaccination against Leptospira. Urine was evaluated by real-time polymerase chain reaction (PCR) specific for lipL32 gene of pathogenic Leptospira. Additionally, urine was cultured for 6 months in Ellinghausen-McCullough-Johnson-Harris (EMJH) medium. Antibodies were measured by ELISA and MAT against 24 serovars belonging to 15 serogroups and 1 undesignated serogroup. Risk factor analysis was performed with backwards stepwise selection based on Wald. Results Twelve of 273 (4.4%; 95% confidence interval (CI): 2.0–6.8%) urine samples were PCR-positive. In 1/273 dogs (0.4%; 95% CI: 0.01–1.1%) Leptospira could be cultured from urine. MAT detected antibodies in 33/273 dogs (12.1%; 95% CI: 8.2–16.0%) against 19 different serovars (Anhoa, Australis, Ballum, Bataviae, Bratislava, Broomi, Canicola, Copenhageni, Coxi, Grippotyphosa, Haemolytica, Icterohaemorrhagiae, Khorat, Paidjan, Patoc, Pyrogenes, Rachmati, Saxkoebing, Sejroe). In 111/252 dogs (44.0%; 95% CI: 37.9–50.2%) immunoglobulin M (IgM) and/or immunoglobulin G (IgG) antibodies were found by ELISA. Female dogs had a significantly higher risk for Leptospira infection (p = 0.023). Conclusions Leptospira shedding occurs in randomly sampled dogs in Thailand, with infection rates comparable to those of Europe and the USA. Therefore, the potential zoonotic risk should not be underestimated and use of Leptospira vaccines are recommended.

There are no comprehensive studies on Leptospira urinary shedding in dogs in Thailand, although several studies demonstrated presence of antibodies against Leptospira in 4.3 to 89.1% of dogs [6][7][8][9][10] (Table 1). Moreover, a recently published small study from Thailand detected Leptospira in the urine of 10.3% (6/58) asymptomatic dogs by rrs nested PCR [32]. Therefore, the aims of the present study were to determine Leptospira urinary shedding prevalence by real-time polymerase chain reaction (PCR), to culture Leptospira from urine, to evaluate Leptospira antibody prevalence by microscopic agglutination test (MAT) and by enzyme-linked immunosorbent assay (ELISA) differentiating immunoglobulin M (IgM) and immunoglobulin G (IgG) antibodies, and to assess risk factors associated with Leptospira infection in dogs in Thailand.
Urine of all 273 dogs was cultured for 6 months. In only 1 urine culture (0.4%; 95% CI: 0.01-1.1%), Leptospira were growing after an incubation period of 3 months. All other 272 cultures remained negative after 6 months. The dog with the positive culture was also positive in urine PCR ( Table 2). In ELISA, this dog had IgM antibodies of 1:320, but no IgG antibodies. No antibodies were found by MAT. Phylogenetic analysis based on secY sequencing showed that this Leptospira strain belonged to the pathogenic genospecies Leptospira interrogans (Fig. 1).

Discussion
This is the first comprehensive study investigating urinary shedding of Leptospira in dogs in Thailand, revealing a shedding prevalence of 4.4% in dogs in Northern, Northeastern, and Central Thailand. The results of this study are of high importance, because Leptospira shedding is a potential infection risk for people in contact with infected dogs. Moreover, shedding dogs contribute to Leptospira spread in the environment [15]. Shedding of Leptospira in dogs starts at day 7-10 after infection and lasts for 4 to 6 weeks [38][39][40], sometimes even over several years [38]. Time period and quantity differ individually and vary between infecting serogroups [15,41]. The shedding prevalence in the present study of 4.4% appears rather low, but the true shedding prevalence might be underestimated because only 12/273 dogs were shedding, while anti-leptospiral antibodies were found in 33/273 (12.1%) dogs in MAT, and in 111/ 252 (44.0%) dogs in ELISA. Thus, almost half of the dogs had been infected at least once in their life, as none of them had ever received a Leptospira vaccine. The shedding prevalence is comparable to other investigations. In Ireland, 7.1% (37/525) of dogs were shedding [29], 8.2% (41/500) in the USA [31], 31.1% in Iran [37], and 19.8% of the dogs in Brazil [35]. In a German study, 1.5% (3/200) of healthy dogs were shedding Leptospira [25], and in Switzerland, the shedding prevalence was 0.2% (1/408) [23]. These low European prevalences are probably due to a broader vaccine-induced immunity in the dog population. Considering the fact that human leptospirosis is endemic in Thailand with its hot humid climate, a much higher shedding prevalence would have been expected in the present survey. Possible explanations are that leptospirosis is more a seasonal disease and no natural disaster, e.g. flooding, which is well documented to enable leptospirosis outbreaks particularly during the rainy season in Thailand, occurred at the time of sampling for the present study [42,43]. A recently published study from Thailand detected a higher shedding prevalence of 10.3% in dogs, as Leptospira were detected in the urine of 6/58 asymptomatic dogs. All these dogs came from Nan province, a rural area in Northern Thailand where leptospirosis is known to be endemic. These urine PCRpositive dogs lived in close contact with livestock and were also used for hunting armadillo and bamboo rats [32]. These facts could explain the higher shedding prevalence found in Nan province compared to the shedding prevalence in the present study.   Culturing Leptospira is not a very sensitive method. The low pH of dog urine kills Leptospira rapidly [15]. Thus, a fast transfer into culture medium is mandatory and was performed in the present study. Nevertheless, only 1/273 (D64) culture samples was positive. Phylogenetic secY analysis revealed that this dog was infected with pathogenic Leptospira interrogans. This dog was also urine PCRpositive with a PCR threshold cycle (Ct) value of 29.0, and had IgM antibodies of 1:320 without IgG antibodies in ELISA or MAT antibodies implying that the infection had been acquired very recently. As culturing Leptospira from urine is not a sensitive method, it is not surprising that no other PCR-positive dog was culture-positive. Another important aspect which could explain the failure to grow Leptospira might be related to a relatively high Ct value of ≥30 in 9/12 PCR-positive dogs, indicating a rather low quantity of excreted Leptospira DNA.
Two different antibody tests were performed. MAT is regarded as the gold standard, but its sensitivity is low. MAT is only at best serogroup-specific and cannot exactly discriminate on serovar level [15]. As no dog in this study had been vaccinated against Leptospira, vaccine-induced interference can be excluded, and the anti-leptospiral antibodies found in 33 dogs (12.1%) were related to exposure. A similar MAT antibody reactivity was found in an older survey in Thailand, revealing an antibody prevalence of 11.0% [9], and a study recently conducted in Northeastern Thailand showed similar results with 10.9% [8]. However, a study on stray dogs in Bangkok in 2009 detected a much higher MAT antibody prevalence of 89.1% (Table 1). All stray dogs of that study were sampled in the center of Bangkok in Buddhist monasteries with close contact to rats which are reservoir hosts of several Leptospira species [6]. There might be a difference in exposure rates of stray dogs and client-owned dogs of which a high number was included in the present study. Client-owned dogs are normally fed by their owners, whereas stray dogs are more likely to hunt rats and mice and thus, to become Leptospira-infected.
In the present study, almost half of the MATpositive dogs (15/33) had antibodies to at least more than one serovar (Table 3) which is presumably due to cross-reactivity which can occur on serovar or even serogroup level [44]. Cross-reactions with other infections, the onset of an acute infection accompanied by a rise in antibodies, or persisting antibodies in a chronic course of infection might be reasons for low antibody titers in MAT [3]. In the present study, the most common reactivity was against serogroup Sejroe, which is present in Rattus rattus, Bandicota indica, and Bandicota savilei in Thailand [45]. This highlights the importance of transmission from rodents. The second frequently reactive serogroup in the present study was Icterohaemorrhagiae which is most commonly involved in human infections worldwide [46], indicating that dogs play an important role in human infection, but rats are also known reservoirs [15]. A high MAT titer could be detected in only 2/ 273 dogs with a titer of 1:640 against Serogroup Sejroe and Bataviae, respectively, that were also positive in urine PCR and had high IgM and IgG titers in ELISA. These results are consistent with an acute but subclinical infection in both dogs, as their physical examination was unremarkable ( Table 2). Only 12.1% dogs had antibodies in MAT in the present study, but it is possible that serovars might have been missed due to an incomplete MAT panel. MAT is also commonly still negative in early infections, while IgM ELISA already reveals positive results [47][48][49][50][51]. This is in line with the finding that only 2/50 dogs that were IgM ELISApositive and IgG ELISA-negative also had detectable MAT antibodies. MAT and ELISA showed a poor agreement indicating a higher sensitivity of ELISA in early infection [47,[49][50][51][52]. Another reason might be a lower specificity of ELISA compared to MAT. In humans living in endemic countries, persistence of IgM antibodies for many months or even years after infection and repeated exposure to nonpathogenic Leptospira was suggested as an explanation for positive IgM results in healthy humans [53][54][55][56].
When comparing antibody findings to PCR results, 9/12 shedders were also ELISA-positive, whereas only 4/12 shedders had antibodies in MAT. The discrepancy between urine PCR and antibody detection is in accordance with a study on 500 dogs in which 41 dogs were shedding Leptospira, while MAT was only positive in 9 dogs [31]. Shedding can occur before MAT antibodies are present [31,[57][58][59]. Another explanation could be immunosuppression or ongoing shedding in chronically infected dogs in which the level of IgM and IgG antibodies had already decreased below detection threshold [15,60].
In the present study, female sex proved to be significantly associated with Leptospira infection in multivariate analysis (Table 4). This finding is in contrast to results of other surveys in which male dogs were at higher risk [61][62][63][64][65] and is also in contrast to a published meta-analysis in which the variable "male dogs" was a significant factor [66]. Interestingly, no further parameters in the present study were significantly associated with Leptospira infection in multivariate analysis. Other studies found a significant predisposition of urban dogs compared to rural living dogs attributed to a higher exposure to wildlife reservoir hosts [66][67][68][69][70]. One could also expect a significantly higher risk of infection for stray dogs. However, client-owned outdoor and stray dogs reside in very similar living environments in both (sub)urban and rural settings in Thailand, and contact to wild-living reservoir hosts occurs in urban and rural areas of Thailand. Sanitation and hygienic standards, including rodent control, might be comparable in both environments. Thus, both stray dogs and client-owned dogs might equally contribute to environmental contamination and potential transmission of Leptospira. Moreover, access to Leptospira-contaminated water sources exists in both environments.

Conclusions
In conclusion, shedding prevalence of Leptospira in dogs taken at random in Thailand was low and not as high as expected for a tropical country. Still, in order to reduce the risk of infection and shedding, vaccines against Leptospira for dogs that are available in Thailand should be recommended as core vaccination, at least for clientowned outdoor dogs. Molecular genetic assays would be of particular importance in order to determine Leptospira strains in dogs in Thailand and globally.

Dogs
In total, 273 randomly selected dogs from rural (n = 139) and urban (n = 134) areas with outdoor access from Northern, Northeastern, and Central Thailand were included. Dogs were presented for either spaying/neutering or for rabies vaccination at public and private castration and vaccination programs. Dogs living indoors only and dogs treated with antimicrobials within the last 4 weeks prior to sampling were excluded. The study population consisted of 266 mixed-breed and 7 pure breed dogs; 154 dogs were client-owned and 119 stray; 185 dogs were female (2 spayed) and 88 male (1 neutered). No dog had received vaccination against Leptospira. After spaying/ neutering and/or rabies vaccination, all client-owned dogs returned to their owners. Of the stray dogs, 77 (64.7%) were brought back and released to the territories where they had been trapped by private and public services for spaying/neutering and rabies vaccination. Forty-two of 119 stray dogs (35.3%) were admitted to governmental dog shelters (no-kill shelters) after spaying/neutering and rabies vaccination. None of the dogs were euthanized.

Sample collection
Blood samples were obtained via puncture of the cephalic or femoral vein. Serum samples were stored at −20°C until further processing. Urine samples were collected by ultrasound-guided cystocentesis (sample volumes of 1.5 ml to 16.0 ml) and stored at 4°C for a maximum of 24 h, and then transferred into 1.5 ml Eppendorf tubes (Eppendorf AG, Hamburg, Germany). Tubes were centrifuged (14,000 x g, room temperature) for 15 min; supernatants were discarded. Pellets were washed with phosphate buffered saline (PBS) and transferred into Eppendorf tubes. A second centrifugation (14,000 x g, room temperature) was performed for 15 min, supernatants were discarded, and pellets were resuspended in 180 μl animal tissue lysis (ATL) buffer (Qiagen, Hilden, Germany) and stored at −20°C until deoxyribonucleic acid (DNA) extraction.

DNA extraction and real-time PCR
For further analysis, urine samples were submitted to IDEXX Laboratories (Ludwigsburg, Germany). Total nucleic acid was extracted from urine as previously described [25]. Leptospira real-time PCR was performed using Light-Cycler 480 (Roche, Mannheim, Germany) with proprietary forward, reverse primers, and hydrolysis probes. The target gene was lipL32/hap1 (accession number AF245281.1), detecting only pathogenic Leptospira. This PCR had been shown to have a reproducible average analytical sensitivity of 10 DNA molecules per reaction.
Urine culture and sequencing of culture-positive sample Of each urine sample, 0.5 ml were cultured in Ellinghausen-McCullough-Johnson-Harris (EMJH) medium as described previously [71][72][73]. Within 2 h after collection, 0.5 ml of each urine sample were added to a tube with 5 ml of liquid EMJH medium plus 0.2 mg/ml 5fluorouracil (5-FU) [74] to a final dilution of 1:10. After mixing and transferring 0.5 ml of the mixture to a second tube with EMJH medium plus 0.2 mg/ml 5-FU, a dilution of 1:100 was prepared. Cultures were stored at 24-28°C and controlled for growth of Leptospira under dark field microscope for a total of 6 months [73] approximately every 7 days. In culture with growth of Leptospira, number of grown Leptospira was estimated by microscope, using a ×20 objective. A 1:10 dilution containing Leptospira was filtered with a 0.2 μm pore size filter (Corning® Sterile Syringe Filter; Corning Incorporated, Wiesbaden, Germany) and then transferred into 9 ml fresh EMJH medium. At a density of >200 Leptospira/field, a passage of the culture into 30 ml EMJH medium was performed. At late exponential phase of leptospiral growth, aliquots of purified Leptospira were frozen with 5% dimethylsulfoxid (DMSO) (Merck KGaA, Darmstadt, Germany) at −80°C in Eppendorf tubes (Eppendorf AG, Hamburg, Germany) until DNA extraction. QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) was used to extract leptospiral DNA, The DNA extract was stored at −20°C until further analysis. For phylogenetic analysis, secY sequencing was performed as described previously [75]. A Neighbor Joining Tree (Fig. 1) was constructed using the software MEGA 7.

Risk factor analysis
Dog owners were requested to answer a standardized questionnaire in order to evaluate risk factors associated with Leptospira infection in dogs (Table 4). Lifestyle and activity parameters of the questionnaire were not recorded in case of stray dogs (n = 119). The approximate age of stray dogs was estimated based on dental examination. Health status of each dog was determined using a standardized physical examination protocol.

Statistical analysis
For sample size calculation, an a priori power analysis using EpiTools epidemiological calculators (Ausvet, Australia) was performed to determine an appropriate sample size to achieve adequate power. Assuming an expected prevalence of urinary shedding of Leptospira (4.0%) with a 95% confidence interval (CI) of the estimation with a 5% precision, at least 236 dogs were necessary to achieve adequate power (>90.0%). Statistical analysis to determine risk factors was performed with SPSS version 24 for Windows (IBM Cooperation, Armonk, USA). For univariate analysis, Fisher's exact test was used to assess risk factors associated with Leptospira infection including all dogs with urinary shedding and/or presence of antibodies in MAT and/or in ELISA, (defined as "Leptospira-infected") ( Table 4). In addition, 2 risk factor analyses were performed for the following 2 subgroups: presence of antibodies against Leptospira determined by MAT (see Additional file 1: Table S1), and urinary shedding of Leptospira detected by PCR (see Additional file 2: Table S2). Multivariate logistic regression analysis was performed using parameters with at least 205 observations as independent variables, as available data of ≥75.0% of the dogs was chosen as mandatory criteria for entering. Backwards stepwise selection based on Wald was performed to detect the most important variables for being Leptospira-infected. Following parameters met the inclusion criteria: age; breed; sex; neutering status; origin; environment. A value of p < 0.05 was considered significant.