Global distribution, host range and prevalence of Trypanosoma vivax: a systematic review and meta-analysis

Trypanosomosis caused by Trypanosoma vivax is one of the diseases threatening the health and productivity of livestock in Africa and Latin America. Trypanosoma vivax is mainly transmitted by tsetse flies; however, the parasite has also acquired the ability to be transmitted mechanically by hematophagous dipterans. Understanding its distribution, host range and prevalence is a key step in local and global efforts to control the disease. The study was conducted according to the methodological recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. A systematic literature search was conducted on three search engines, namely PubMed, Scopus and CAB Direct, to identify all publications reporting natural infection of T. vivax across the world. All the three search engines were screened using the search term Trypanosoma vivax without time and language restrictions. Publications on T. vivax that met our inclusion criteria were considered for systematic review and meta-analysis. The study provides a global database of T. vivax, consisting of 899 records from 245 peer-reviewed articles in 41 countries. A total of 232, 6277 tests were performed on 97 different mammalian hosts, including a wide range of wild animals. Natural infections of T. vivax were recorded in 39 different African and Latin American countries and 47 mammalian host species. All the 245 articles were included into the qualitative analysis, while information from 186 cross-sectional studies was used in the quantitative analysis mainly to estimate the pooled prevalence. Pooled prevalence estimates of T. vivax in domestic buffalo, cattle, dog, dromedary camel, equine, pig, small ruminant and wild animals were 30.6%, 6.4%, 2.6%, 8.4%, 3.7%, 5.5%, 3.8% and 12.9%, respectively. Stratified according to the diagnostic method, the highest pooled prevalences were found with serological techniques in domesticated buffalo (57.6%) followed by equine (50.0%) and wild animals (49.3%). The study provides a comprehensive dataset on the geographical distribution and host range of T. vivax and demonstrates the potential of this parasite to invade other countries out of Africa and Latin America.


Introduction
Trypanosomes are protozoan parasites belonging to the family of Trypanosomatidae and the genus Trypanosoma (T.). The genus Trypanosoma comprises many species such as T. brucei, T. congolense, T. equiperdum, T. evansi, T. simiae, T. suis and T. vivax, which cause diseases called trypanosomoses in different mammalian hosts including humans [1]. Trypanosomoses are widely distributed in Africa, Latin America and Asia [2,3].
Trypanosoma vivax is one of the most important Trypanosoma species known to infect both domestic and wild animals [4,5]. Trypanosoma vivax is reported from cattle, dromedary camel, [6], goat, sheep, pig, dog [7], horse, donkey [8], both domesticated and wild buffalo, warthog, hippopotamus, reedbuck, waterbuck [9], antelope [10], giraffe [11], rhinoceros [12], rodents, pangolins, primates, reptiles and different wild ungulates and carnivores [13]. In Sub-Saharan Africa, T. vivax is mainly transmitted by tsetse flies (Diptera: Glossinidae) in which the parasite can multiply and remain infective throughout the insect's life [14]. The parasite has the ability to be transmitted mechanically by hematophagous flies such as Tabanus spp., Stomoxys calcitrans and Haematobia irritans, which are responsible for the spread of T. vivax in tsetse-free areas of Africa and in Latin America [4,[15][16][17][18] Trypanosoma vivax infection can be suspected by clinical and/or serological evidence and can be confirmed by parasitological or molecular methods [19]. Trypanosoma vivax prevalence shows considerable variation with geography, abundance of tsetse or blood-sucking flies, and host species. In tsetse-infested areas of tropical Africa, the T. vivax prevalence is typically reported between 5-15% and often accounts for up to half of the total trypanosome prevalence. Outside of the tsetse belt, T. vivax prevalence is lower, between 2-10%, and it is related to local and seasonal variation in biting fly abundance [20].
Trypanosomosis caused by T. vivax is an important cause of economic losses related to morbidity, mortality, reproductive issues and decreased production [4]. For example, economic losses associated with bovine trypanosomosis have been estimated to be around US$5 billion a year in Africa, and the continent spends at least $30 million every year to control bovine trypanosomosis in terms of curative and prophylactic treatments [21]. Estimates outside Africa indicate that > 11 million head of cattle with a value of > US$ 3 billion are at risk from T. vivax infection in the Brazilian Pantanal and Bolivian lowlands, with potential losses in excess of US$ 160 million [16].
Many studies have been conducted on T. vivax over the past 100 years. Studies before the 1950s focused more on the morphology and taxonomy [22,23], pathogenicity [24] and treatment [25,26]. However, since the 1950s, a considerable number of epidemiological studies have been conducted. Notwithstanding the excellent review on livestock trypanosomoses and their vectors in Latin America [18] and a recent general review on T. vivax [20], a systematic literature review on the global distribution, prevalence and host range of T. vivax is lacking. Moreover, no information on the global distribution of T. vivax is available at the World Animal Health Information System of the World Health Organization (https :// www.oie.int/wahis _2/publi c/wahid .php/Disea seinf ormat ion/Disea sedis tribu tionm ap).
Thus, this study was conducted to provide the global distribution of T. vivax and to estimate the pooled prevalence of trypanosomosis caused by T. vivax in naturally infected domestic and wild animals.

Methods
The systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist [27]. Screening and data extraction were performed by two authors (SL and EF) independently. All disagreements were discussed and resolved by consensus. A third author (PB) was also involved in the search for full-text papers to ensure that all relevant publications were included.

Literature search
On 30 August 2019, a systematic literature search was conducted on three databases to identify all publications reporting natural infection of T. vivax across the world. PubMed, Scopus and CAB Direct were screened using the search term Trypanosoma vivax without time and language restrictions. All references found were imported into Mendeley Desktop reference manager software.

Inclusion and exclusion criteria
To be considered, articles were required to meet the following inclusion criteria: (i) should be observational studies such as cross sectional, longitudinal, case report or outbreak investigation, published in indexed journals, reporting any natural infection of T. vivax using any diagnostic test or tests available; (ii) the study design, sample size, sample type, diagnostic methods and number of T. vivax-infected animals or prevalence, including 0%; (iii) species of animals with T. vivax infections must be provided. Experimental studies; publications which fail to describe diagnostic tools, study design and/or sample sources; and reports solely based on clinical signs were removed despite reporting the prevalence of the disease. In addition, studies reporting T. vivax from multiple species without stratifying the report at species level were removed.

Data extraction
All relevant information such as author names, year of publication, study period, country, region, province, district, latitude, longitude (if provided or if they can be retrieved), host species, number of samples analyzed, type of samples collected, diagnostic method used, number of positives and prevalence or percentage were extracted to a pre-prepared Microsoft Excel spreadsheet (Microsoft Corp., Redmond, WA, USA). When publications only reported the number of animals tested and the prevalence, the numbers of positives were calculated. When publications only reported the number of animals tested and the number of positives, prevalence values were calculated. Publications in other languages than English were translated using Google Translate.

Data analysis
Owing to heterogeneity within and between studies, random-effects meta-analysis was used to estimate the pooled prevalence and its 95% confidence interval (CI) in different hosts [28]. The estimation was carried out after categorization of the results according to the diagnostic tests used and the host species tested. Accordingly, diagnostic tests were categorized into three categories: (i) parasitological methods, including wet blood smear, stained blood smear and microhematocrit concentration; (ii) serological methods, including enzyme-linked immunosorbent assay (ELISA) both antigen and antibody based, indirect fluorescence antibody test (IFAT) and antigen detection LATEX agglutination; (iii) molecular methods, including reverse line blot hybridization assay, realtime and conventional polymerase chain reaction (PCR). Species-wise, sheep and goat were categorized into "small ruminants, " horse, donkey and mule into "equine" and all studied wild animals including Cape buffalo into "wild animals. " For cattle, domestic buffalo, dromedary camel, pig and dog, pooled prevalence was estimated without categorization.
Heterogeneity between studies was evaluated through the Cochran's Q test (reported as p value), and the inverse variance index (I 2 ). I 2 describes the percentage of observed total variation between studies due to heterogeneity rather than to random error (intra-study variation). I 2 values < 25% correspond with low heterogeneity, up to 50% with moderate and up to 75% with high heterogeneity [29]. Sub-group analysis using the variable test method was performed to determine the potential sources of heterogeneity among studies. The across-study bias was examined by a funnel plot and Egger's regression asymmetry test. A funnel plot was used to visually examine the presence of publication bias, and Egger's regression asymmetry test was used to test whether the bias was statistically significant or not [30]. The unbiased estimates were calculated using the Duval and Tweedie non-parametric 'fill and trim' linear random method [31].
The meta-analysis was done using 'meta' package of R statistical software version 3.6.2 (R Foundation for Statistical Computing). The map representing the global distribution of T. vivax was created, using Quantum GIS software version 3.4.5 (Open Source Geospatial Foundation, Boston, MA, USA).

Literature search selection and data extraction
A total of 1691 publications were retrieved, 348 from PubMed, 1269 from Scopus and 74 from CAB Direct (Fig. 1). After removal of 390 duplicates, the remaining 1301 articles were screened based on their titles and abstracts. Reviews and articles reporting on laboratory and field experiments (n = 993) were excluded of further analysis. Articles without an abstract or without sufficient information to make a decision were left for full text review. Of the remaining 308 articles, 11 of the full text files remained inaccessible [32][33][34][35][36][37][38][39][40][41][42]. Finally, one additional article, missed by the systematic literature search, was included manually. Full-text papers of 298 articles were retrieved online or via the library of the Institute of Tropical Medicine Antwerp and eligibility assessed according to the pre-established inclusion/exclusion criteria. Further 53 articles were excluded leaving 245 articles fulfilling all inclusion criteria for the qualitative analysis [4-10, 12-14, 43-277]. Among these 245 articles published between 1958 and 2019, 10 are case reports, 186 report on a cross-sectional study, 35 on a longitudinal study and 14 on an outbreak investigation. All relevant data from these articles were recorded, according to diagnostic method and host species, in a Microsoft Excel file, thus containing 899 records used in the meta-analysis (Additional file 1).
Of these 245 articles, 187 are conducted in 27 African countries, with Ethiopia taking the lead with 43 articles, followed by Nigeria with 29, Uganda with 21 and Kenya with 15 articles. In Latin America, 57 studies were conducted of which 32 were from Brazil, 9 from Venezuela and 6 from Colombia.

Geographic distribution
All the studies conducted in the 27 African countries reported the presence of T. vivax in at least one host species; natural T. vivax infections were found in 12 of the 13 studied Latin American countries ( Fig. 2 and Table 1). In Martinique, Alonso and co-workers did not find clinical or serological evidence of T. vivax in cattle on this island [50]. One article mentions a cross-sectional study on 300 equines in Pakistan, but all animals were negative in molecular tests for T. vivax [231]. We could not find any other reports on the presence of T. vivax in Asia, Antarctica, Australia, Europe and North America.

Host range
A total of 232,627 tests were performed, and 24,420 of them were positive for natural infection of T. vivax. Trypanosoma vivax was reported from nine domestic animal species: cattle, domestic buffalo, dog, donkey, dromedary camel, goat, horse, pig and sheep. Among them, cattle were the most studied species with 198,593 tests performed on cattle in 36 countries and two territories (192 publications) and 20,964 were positive for T. vivax. Next to cattle, goat, sheep, pig and donkey were the most frequently studied species. The protozoal parasite was also reported from wild animals including diverse species of antelopes, Cape buffalo, hippopotamus, black rhinoceros, pangolin and warthog. Trypanosoma vivax was reported from 39 wild fauna species, including many antelope species and Cape buffalo (Tables 2, 3).

Pooled prevalence estimates according to host species and type of diagnostic test
Pooled prevalence estimates by test methods for different hosts are presented in Table 4, and forest plots of the meta-analysis and the subgroup analyses can be found in Additional files 2 and 3. Substantial heterogeneity was observed in the pooled estimate except for dog, which remained significant (P < 0.05) even after sub-group analysis.
A total of 145 cross-sectional studies from 32 countries were included in estimation of natural infection of T. vivax in cattle. The random effect model indicates the pooled prevalence to be 6.4% (5.7-7.2, 95% CI). For small ruminants, pooled prevalence of T. vivax was estimated from 33 studies in 16 countries and found to be 3.8% (2.5-5.6, 95% CI). A total of 15 studies from 10 different countries were used to estimate the pooled prevalence of T. vivax in equines. The random effect model estimates the pooled prevalence to be 3.7% (2.0-6.8, 95% CI). Pooled prevalence of T. vivax in camels was estimated from four studies in three different countries. The model estimates a pooled prevalence of 8.4% (3.4-19.3, 95% CI). A total of 12 studies from 8 different countries were included in the estimation of pooled prevalence in pigs, which was found to be 5.5% (3.0-10.1, 95% CI). Five studies from five countries were used in the estimation pooled prevalence of T. vivax in dogs. The pooled prevalence was estimated to be 2.6% (1.0-6.3% 95% CI). Three studies reported natural infection of T. vivax in domestic buffaloes from Venezuela, and the random effect model estimates a pooled prevalence of 30.6% (14.2-54.1, 95% CI). For wild animals, a pooled prevalence of 12.9% (9.9-16.6, 95% CI) was estimated from six studies in five countries. Subgroup pooled prevalences estimated according to the type of diagnostic test, as represented in Table 4, were lowest with parasitological techniques (from 1.1% in pigs to 13.2% in wild animals) and highest with serological techniques (from 13.8% in small ruminants to 57.6% in domestic buffalo).

Discussion
This study presents the first systematic review of published literature since the 1950s describing global distribution, host range and prevalence of trypanosomosis caused by T. vivax. Not surprisingly, most publications report on T. vivax infections in domestic mammalian species, in particular in cattle and small ruminants, while few publications describe natural infections in wildlife.
Looking at the T. vivax distribution map (Fig. 2), there is an evident data gap for some sub-Saharan African countries where tsetse flies are present and therefore T. vivax may be endemic. Although our formal search strategy could not retrieve any publication on these "missing" countries, conventional Google search confirms the presence of T. vivax in South Sudan and Zimbabwe [284,285], and Genevieve et al. [286] reported on the presence  [13] of potential vectors in the Central African Republic. Since Angola, the Central African Republic and the Republic of Congo are endemic for human African trypanosomosis, the presence of T. vivax in these countries is likely [287]. Due to its adaptation to mechanical transmission, T. vivax is also present outside the tsetse belt in Africa, e.g. in Ethiopia and Sudan [114,288]. As a consequence, the trypanosomosis control efforts with focus on tsetse eradication might have little effect on T. vivax. Also, economic impact assessments that are solely based on tsetse distribution alone could seriously underestimate the problem of trypanosomosis because of T. vivax. Out of Africa, T. vivax is present in Latin America but not in North America, Australia, Asia and the Pacific regions. Trypanosoma vivax is believed to be introduced into Latin America in cattle and horses imported from Africa, possibly in the sixteenth century, and spread to different Latin American countries including Brazil, Colombia, French Guiana, Guadeloupe, Guyana, Martinique, Panama, Suriname and Venezuela [18]. Stephen [289] reviewed the presence of the parasite in Costa Rica, Ecuador, El Salvador, Paraguay and Peru, and according to Gardiner et al. [15], T. vivax was present in the Caribbean thus posing a threat to the livestock industries.
From our literature search, we can only confirm T. vivax to be endemic in 12 Latin American countries of which 7 (Argentina, Bolivia, Brazil, Colombia, Guyana, Peru, Venezuela) are also endemic for T. evansi [290]. Although, our literature search provides information on the potential spread of T. vivax in Latin America, it is important to note that the distribution could be much wider, for example, T. vivax was only detected in Argentina in 2018; this is this due to the lack of previous studies. Apparently, T. vivax has never spread into Asia, unlike T. evansi, although similar to the latter; it can be mechanically transmitted by bloodsucking flies. Unless there is a particular biological or environmental factor preventing T. vivax from invading the Middle East and Asia, as well Northern Africa, North America and Europe, we must remain alert about the risk of importing T. vivax into non-endemic countries as happened in Latin America.
This review suggests that T. vivax has a very diverse host range, including 9 domestic mammals and almost 40 wild fauna species. Regarding the latter, however, data should be interpreted with caution. Diagnostic tests, whether parasitological, serological or even molecular, have their limitations. For examples, by sequencing of PCR amplicons, Auty and co-workers [11] clearly demonstrate that wildlife may harbor a diversity of trypanosomes, including taxonomically undefined species. Therefore, it is likely that many reports on T. vivax infection in wildlife and tsetse in fact deal with other trypanosome species that are not necessarily pathogenic for domestic animals.
The pooled prevalence of trypanosomosis in different hosts varies significantly depending on the detection methods; significantly higher estimates were reported in publications using serological techniques. Higher estimates using a serological technique could be due to the persistence of the antibody over several months after curative treatment and the possibility of low undetectable parasitemia in parasitological techniques [20,93,291,292]. Moti et al. [187] compared the percentage positivity obtained with different diagnostic techniques and showed that relative to the microhematocrit centrifugation technique the percent positivity increased by 50 and 250% when using PCR-RFLP. Also, Garcia et al. [118] reported that for the detection of trypanosomes, PCRbased assays are twice as sensitive as parasitological techniques such as the microhaematocrit centrifugation.
The study has the following limitations. The literature search was almost exclusively based on electronic databases whereby some older literature must have been missed. The data showed a large degree of heterogeneity among studies, which remain significant after subgroup analysis. There is a significant publication bias, which could be due to incomplete or inaccurate information provided in the publications. In addition, studies were conducted between 1956 and 2017, and the result may not accurately reflect the current epidemiological situation and therefore could limit interpretation of the result to some degree. Furthermore, we suspect numerous data gaps mainly because of two reasons. First, due to the lack of a country-level monitoring and reporting system for trypanosomosis, most of the data included in this analysis are from research activities. Second, trypanosomosis diagnosis in most endemic countries relied to a great extent on low-sensitivity parasitological methods, while more sensitive molecular tools are rarely used. In addition, the majority of studies analyzing trypanosome's presence in the field may not have a sampling strategy that allows a robust estimation of prevalence. This is for multiple and understandable reasons-samples can be difficult and expensive to collect, and many studies rely on purposive sampling, or sampling of, for example, animals presented to veterinary clinics. While these kinds of studies provide a rough idea of pathogen presence/ absence, they may not provide an accurate estimate of prevalence. Thus, caution should be taken when interpreting the results presented here.

Conclusion
With this study, we intended to provide comprehensive information on the geographical distribution, host range and prevalence of trypanosomosis caused by T. vivax worldwide. The results confirm the wide geographical distribution and a diverse host range of T. vivax. The parasite parasitizes almost all domestic mammals and many wild animal species, thus suggesting the potential to get established in other countries with favorable environmental conditions, e.g. in the Middle East, Asia and Australia. The meta-analysis showed a high degree of variability in estimated prevalence values. The variability can be attributed to diagnostic tests used and the species of the animal infected.
Additional file 1. Global Trypanosoma vivax occurrence records.

Additional file 2. Forest plots showing an overview of studies reporting
Trypanosoma vivax in different host species.

Additional file 3. Forest plots showing an overview of studies reporting
Trypanosoma vivax grouped by test methods in different host species.