Introduction

According to the World Health Organization (^{WHO, 2017}), leishmaniases are a group of parasitic diseases caused by over 20 different species of Leishmania spp., protozoa that are transmitted mainly through the bite of sandflies. Four forms of the disease are known: visceral leishmaniasis (or kala-azar), post-kala-azar dermal leishmaniasis, cutaneous leishmaniasis and mucocutaneous leishmaniasis.

In humans, the most common clinical manifestation is the cutaneous form, which is considered endemic in 44% of the countries that notified cases of the disease in 2015 (^{WHO, 2017}); while the visceral form is the most severe, nearly always fatal if untreated, and is endemic in 38% of the countries that reported occurrence of the disease in that same year (^{WHO, 2017}). In addition to affecting humans, leishmaniasis also affects several domestic mammals. Infected dogs are the most important reservoirs for the parasite in urban areas, which makes them the main source of infection for people living in endemic areas (^{PACE, 2014}).

However, several studies have sought to identify other vertebrates that can host and participate in the cycle of these protozoa (^{QUARESMA et al., 2011}; ^{GAO et al., 2015}; ^{KENUBIH et al., 2015}; ^{ROHOUSOVA et al., 2015}). Among the investigations that have sought new possible reservoirs, there have been reports on equids showing clinical manifestations of cutaneous leishmaniasis (^{KOEHLER et al., 2002}; ^{SOLANO-GALLEGO et al., 2003}; ^{MÜLLER et al., 2009}; ^{SOARES et al., 2013}). Furthermore, prevalence studies conducted in Europe (^{FERNÁNDEZ-BELLON et al., 2006}; ^{LOPES et al., 2013}; ^{SGORBINI et al., 2014}), Asia (^{GAO et al., 2015}; ^{AHARONSON-RAZ et al., 2015}), Africa (^{MUKHTAR et al., 2000}; ^{KENUBIH et al., 2015}; ^{ROHOUSOVA et al., 2015}) and South America (^{AGUILAR et al., 1989}; ^{FEITOSA et al., 2012}; ^{TRUPPEL et al., 2014}; ^{OLIVEIRA et al., 2017}; ^{BENASSI et al., 2018}) have demonstrated that horses (Equus caballus), donkeys (Equus asinus), mules (Equus asinus caballus) and ponies (E. caballus) are parasitized by different species of Leishmania, such as Leishmania braziliensis, Leishmania infantum and Leishmania siamensis, including cases of mixed infections.

The importance of research on leishmaniasis in equids lies in the fact that these species of domestic animal, just like dogs and cats, are in close contact with humans, which may be through use as a means of transportation or for work or leisure activities. Moreover, the low socioeconomic level of the population living in endemic areas for the disease could generate risks of infection across the zoonotic cycle, such that equids would play the role of either potential reservoirs or sources of food for sandflies in peridomestic areas.

Therefore, the objective of the present study was to carry out a qualitative and quantitative synthesis (with meta-analysis) based on a systematic review of the literature. Greater clarification regarding the main clinical aspects of leishmaniasis in equids and methods for diagnosing it was sought, with a view to enabling support for future studies on this subject.

Materials and Methods

Results

The initial search of databases and study selection are presented in Figure 1. Out of the total number of studies searched (n = 531), 29 met the eligibility criteria and were divided as follows: 11 studies described the clinical characteristics of leishmaniasis in equids, and thus were included in the qualitative synthesis; and another 18 were considered to be cross-sectional studies (prevalence studies), with sufficient data for quantitative synthesis and meta-analysis.

Figure 1 Flow chart of the search, selection and inclusion process for studies in the systematic review. 

Quantitative synthesis and meta-analysis on leishmaniasis prevalence in equids

The studies included in this stage were conducted in Brazil (n = 10), China (n = 1), Spain (n = 1), Ethiopia (n = 2), Greece (n = 1), Israel (n = 1), Italy (n = 1), Portugal (n = 1), Sudan (n = 1) and Venezuela (n = 2). In one of these studies (^{AGUILAR et al., 1989}), data collected in both Brazil and Venezuela were published. In turn, in two other studies (^{TRUPPEL et al., 2014}; ^{BENASSI et al., 2018}), two different diagnostic techniques were used. We therefore considered the use of these different techniques to constitute different studies and, for this reason, the initial total of 18 studies identified was then counted as 21 in the quantitative analysis phase.

Table 2 summarizes the main characteristics of the studies included in the meta-analysis. In evaluating the prevalence results through Cochran’s Q test (in which p = 0) and the I² statistic of Higgins and Thompson (in which I² = 99.5%), high heterogeneity was observed among the studies. Therefore, a random-effects model was used, according to subgroup, to conduct the meta-analysis. This model yielded a combined prevalence of leishmaniasis in equids of 25% (CI: 15-35%), among the studies included (Figure 2). To assess the possible causes of heterogeneity, the studies were divided into subgroups according to the technique that had been used for the diagnosis: parasitological; enzyme-linked immunosorbent assay (ELISA); direct agglutination test (DAT); indirect fluorescence antibody test (IFAT); and polymerase chain reaction (PCR). These results are shown in Figure 2.

Table 2 Quantitative synthesis regarding the main characteristics of the studies included in the meta-analysis. 

References	Country	Sample	Positive	Prevalence (%)	Diagnosis
Aguilar et al. (1984)	Venezuela	28	6	21,43	Parasitological
Aguilar et al. (1989)	Brazil	26	8	30,77	Parasitological
Aguilar et al. (1989)	Venezuela	32	9	28,13	Parasitological
Mukhtar et al. (2000)	Sudan	96	66	68,75	DAT
Fernández-Bellon et al. (2006)	Spain	112	16	14,29	ELISA
Vedovello et al. (2008)	Brazil	55	42	76,36	DAT
Kouam et al. (2010)	Greece	773	2	0,26	ELISA
Feitosa et al. (2012)	Brazil	466	68	14,59	ELISA
Lopes et al. (2013)	Portugal	173	7	4,05	DAT
Sgorbini et al. (2014)	Italy	277	18	6,50	IFAT
Truppel et al. (2014)	Brazil	227	25	11,01	ELISA
Truppel et al. (2014)	Brazil	227	37	16,30	PCR
Acosta et al. (2014)	Brazil	20	0	0,00	IFAT
Aharonson-Raz et al. (2015)	Israel	319	6	1,88	DAT
Gao et al. (2015)	China	37	8	21,62	PCR
Kenubih et al. (2015)	Ethiopia	15	5	33,33	DAT
Rohousova et al. (2015)	Ethiopia	20	2	10,00	PCR
Oliveira et al. (2017)	Brazil	257	62	24,12	IFAT
Evers et al. (2017)	Brazil	398	183	45,98	IFAT
Benassi et al. (2018)	Brazil	40	40	100,00	PCR
Benassi et al. (2018)	Brazil	40	1	2,50	IFAT

Figure 2 Combination of 21 prevalence studies on leishmaniasis in equids, according to the diagnosis method used. 

The visual analysis on the funnel plot (Figure 3) showed that there was asymmetrical distribution among the 21 studies, thus demonstrating the possibility of publication bias, which was confirmed through applying Egger’s test (p = 0.01).

Figure 3 Funnel plot presenting the asymmetrical distribution of studies on the prevalence of leishmaniasis in equids. 

Discussion

The qualitative analysis on the case reports showed that all the equids were affected by the cutaneous form of leishmaniasis (Table 1), presenting lesions that began with papules and nodules and progressed to ulcers, with presence of crusts, alopecia, exudate and pruritus as a result of the evolving inflammatory process that was underway. There is a consensus among researchers that the common infection site of L. braziliensis is the skin, while L. infantum is responsible for visceral infection (^{KOEHLER et al., 2002}). However, the present review shows that this has not been observed among equids, given that all the reports cited in this review showed predominance of the cutaneous form of the disease, regardless of which species of Leishmania was identified.

When an etiological diagnosis of the disease was possible, the most frequent species found was L. infantum (n = 8), followed by L. siamensis (n = 5), L. braziliensis (n = 2) and mixed infection of L. infantum and L. braziliensis (n = 1). In the remaining cases (n = 6), the agent was not identified. Occurrences of species with higher pathogenicity towards humans (L. infantum and L. siamensis) infecting equids is a reason for greater concern, given that vectors use this mammal species as a source of food and could consequently ingest and become contaminated by the protozoa, although this type of transmission has not yet been demonstrated in equids in a general manner.

The lesions that were described in the studies surveyed here were located mainly on the head and limbs, although they were also observed on the neck, genital organs, abdomen and thorax, which are regions where there are few or even no hairs, which facilitates access by the vector mosquito to begin feeding and, consequently, to inoculate the parasite into the skin. Moreover, research on humans has shown that attractive volatile substances are exhaled from specific parts of the body, such as the ears, which attract greater numbers of insects to that region (^{REBOLLAR-TELLEZ et al., 1999}). Although no such reports relating to equids have been conducted, presence of volatile attractants forms a plausible argument for explaining the presence of lesions on certain body parts rather than on others, thus suggesting that this occurs not just because of absence of hair in these places.

Skin diseases are common among equids, and some present characteristics similar to those described for leishmaniasis, such as equine sarcoid, squamous cell carcinoma, pythiosis, habronemiasis (^{KOEHLER et al., 2002}; ^{SOLANO-GALLEGO et al., 2003}) or any other cutaneous disorder that causes either papular or nodular lesions and/or ulcers in equids (^{SOARES et al., 2013}). Thus, leishmaniasis should be considered in making differential diagnoses of dermatopathies in equids, especially when the animals inhabit regions that are endemic for leishmaniasis and lesions occur in areas such as the genital organs, head, neck, ears and inguinal and axillary regions, and when there is no satisfactory response from administration of antimicrobial or antifungal therapies (^{RAMOS-VARA et al., 1996}).

Most lesions that were described in previous studies either regressed spontaneously without the need for treatment or did not recur after surgical removal (^{RAMOS-VARA et al., 1996}; ^{KOEHLER et al., 2002}; ^{SOLANO-GALLEGO et al., 2003}; ^{ROLÃO et al., 2005}; ^{MÜLLER et al., 2009}; ^{GAMA et al., 2014}). This emphasizes the idea that the immune response of equids against the parasite is effective (^{FERNÁNDEZ-BELLON et al., 2006}).

Use of medication to treat the disease was reported in three studies. In the first, ^{Barbosa-Santos et al. (1994)} initially used immunotherapy and observed an increase in antibody levels and worsening of lesions, which then led them to use conventional chemotherapy (pentavalent antimony), through which regression of lesions was achieved after a second application. In the second report, ^{Ramos-Vara et al. (1996)} used sodium stibogluconate and reached complete cure for the lesions. Lastly, ^{Solano-Gallego et al. (2003)} used anti-inflammatory therapy with dexamethasone, but without any satisfactory response.

Regarding the quantitative synthesis, the meta-analysis indicated that the prevalence of leishmaniasis in equine populations was 25% (CI:15-35%), with distribution over four different continents (Africa, South America, Asia and Europe). However, the high heterogeneity of the studies analyzed gives rise to some reflection regarding the degree of reliability of this combined prevalence among the studies.

In seeking to elucidate this high heterogeneity, we were particularly interested in the number of different diagnostic techniques that have been used to investigate leishmaniasis in equids. Thus, one meta-analysis per subgroup was conducted, with separation according to the method used for the diagnosis: parasitological examinations, IFAT, ELISA, PCR and DAT.

The forest plot (Figure 2) brings the 21 studies included in the meta-analysis, where the points inside the boxes in the center of the graph represent the prevalence of each study individually and the horizontal line the confidence interval, and the diamonds demonstrate the combined prevalence among studies. Prevalence results and confidence intervals are shown in the fourth and fifth columns, respectively. The last column represents the weight with which each study participated in the aggregate result.

The heterogeneity was high both in the aggregate outcome of the 21 studies (I² = 100%) and in the DAT (I² = 99%), ELISA (I² = 97%), IFAT (I² = 98%) and PCR (I² = 100%) subgroups. As shown in Figure 2, heterogeneity is clearly present between the studies that used different diagnostic techniques, and it is also present among studies that used the same technique. One exception were the three studies that used parasitological techniques, which did not show heterogeneity between each other (I²=0%). Parasitological examinations are considered to be the gold standard for diagnosis, and thus their results are more reliable.

Serological tests were the diagnostic techniques that presented greatest methodological differences when used to diagnose leishmaniasis in equids. Regarding DAT, for example, five studies used different antigen concentrations, serum dilutions and cutoff points (^{LOPES et al., 2013}; ^{AHARONSON-RAZ et al., 2015}; ^{KENUBIH et al., 2015}), along with positive control serums from either humans or dogs (^{MUKHTAR et al., 2000}; ^{VEDOVELLO et al., 2008}). In ELISA, there were also methodological differences, especially regarding the types of antigens and conjugates used in serum dilutions (^{FERNÁNDEZ-BELLON et al., 2006}; ^{KOUAM et al., 2010}; ^{FEITOSA et al., 2012}; ^{TRUPPEL et al., 2014}). Other serological studies used IFAT and also presented differences regarding the cutoff points and conjugates used (^{SGORBINI et al., 2014}; ^{ACOSTA et al., 2014}; ^{OLIVEIRA et al., 2017}; ^{EVERS et al., 2017}; ^{BENASSI et al., 2018}).

Regarding molecular tests through PCR, the meta-analysis results also demonstrated high heterogeneity (I² = 100%) among the four studies that used this technique (^{TRUPPEL et al., 2014}; ^{GAO et al., 2015}; ^{ROHOUSOVA et al., 2015}; ^{BENASSI et al., 2018}). Although the authors of these studies used different primers to identify parasite DNA, we do not believe that this was the source of the heterogeneity that was observed in the meta-analysis of this subgroup. In this case, we believe it is coherent to attribute this heterogeneity to the variation in real prevalence of the disease in the studies indicated, given the high sensitivity and specificity of the PCR technique as a diagnostic method and the combined prevalence and confidence interval observed among the studies in which it was applied (37%; 0 – 92%).

Publication bias is often responsible for the heterogeneity between studies that is found through meta-analysis. In Figure 3 each point on the plot represents a study, with the prevalence arranged on the X axis and the standard error on the Y axis. In the absence of bias, the points are expected to have a symmetrical distribution under the dashed area of the triangle (inverted funnel), with the most accurate studies occupying the vertex (lowest standard error) and the least accurate distributed at the base of the triangle. It was possible to verify by observing the funnel plot (Figure 3) a marked asymmetry of the points (studies), indicating possible publication bias in the present meta-analysis, a fact also confirmed by the Egger’s test (p = 0.01). In relation to cross-sectional studies, it is especially common to find that researchers are not interested in publishing their studies when they do not find significant prevalence in their investigations. Moreover, the editors of journals even appear reluctant to publish these negative findings. This lack of negative findings generates bias that can interfere in the results from a meta-analysis (^{Pereira & Galvão, 2014}). However, we believe that the difference in the methodologies that were used for diagnosing the disease in equids was the factor that most contributed towards the high heterogeneity observed.

Conclusion

The clinical disease caused by different species of Leishmania spp. is benign to equids, thus not requiring treatment in most cases, with the exception of basic care to avoid secondary contaminations, or cases in which the lesions reach larger proportions. However, leishmaniasis is clinically important, given that it can be mistaken for other common skin diseases in equids. Thus, leishmaniasis should be considered in the differential diagnosis for these diseases. Moreover, more detailed investigations are needed regarding the pathology of the disease in these hosts, particularly if development of visceral lesions occurs, and regarding the role of the immune response in determining whether clinical signs will appear.

Hence, seroepidemiological studies are important for investigating and monitoring suspected exposure of these hosts to the parasite, especially in endemic areas, given that natural equid-vector-human transmission has not yet been either demonstrated or refuted and could, therefore, represent a risk to vulnerable populations. However, the diagnostic techniques used for this purpose have not yet been standardized for equids and this may generate conflicting results and erroneous conclusions regarding prevalence of the disease in this host population, as observed in the present study.