Evaluating the impact of targeting livestock for the prevention of human and animal trypanosomiasis, at village level, in districts newly affected with T. b. rhodesiense in Uganda

Background Uganda has suffered from a series of epidemics of Human African Trypanosomiasis (HAT), a tsetse transmitted disease, also known as sleeping sickness. The area affected by acute Trypanosoma brucei rhodesiense HAT (rHAT) has been expanding, driven by importation of infected cattle into regions previously free of the disease. These regions are also affected by African Animal Trypanosomiasis (AAT) demanding a strategy for integrated disease control. Methods In 2008, the Public Private Partnership, Stamp Out Sleeping Sickness (SOS) administered a single dose of ﻿try﻿p﻿anocide﻿ to ﻿31 486﻿ head of cattle in 29 parishes in Dokolo and Kaberamaido districts. This study examines the impact of this intervention on the prevalence of rHAT and AAT trypanosomes in cattle from villages that had (HAT+ve) or had not (HAT-ve) experienced a recent case of rHAT. Cattle herds from 20 villages were sampled and screened by PCR, pre-intervention and 6-months post-intervention, for the presence or absence of: Trypanosoma brucei s.l.; human infective T. b. rhodesiense; Trypanosoma vivax; and Trypanosoma congolense savannah. Results Post-intervention, there was a significant decrease in the prevalence of T. brucei s.l. and the human infective sub-species T. b. rhodesiense in village cattle across all 20 villages. The prevalence of T. b. rhodesiense was reduced from 2.4% to 0.74% (P < 0.0001), with the intervention showing greater impact in HAT-ve villages. The number of villages containing cattle harbouring human infective parasites decreased from 15/20 to 8/20, with T. b. rhodesiense infection mainly persisting within cattle in HAT+ve villages (six/eight). The proportion of T. brucei s.l. infections identified as human infective T. b. rhodesiense decreased after the intervention from 8.3% (95% CI = 11.1–5.9%) to 4.1% (95% CI = 6.8–2.3%). Villages that had experienced a recent human case (HAT+ve villages) showed a significantly higher prevalence for AAT both pre- and post-intervention. For AAT the prevalence of T. vivax was significantly reduced from 5.9% to 0.05% post-intervention while the prevalence of T. congolense increased from 8.0% to 12.2%. Conclusions The intervention resulted in a significant decrease in the prevalence of T. brucei s.l., human infective T. b. rhodesiense and T. vivax infection in village cattle herds. The proportion of T. brucei s.l. that were human infective, decreased from 1:12 T. brucei s.l. infections before the intervention to 1:33 post-intervention. It is clearly more difficult to eliminate T. b. rhodesiense from cattle in villages that have experienced a human case. Evidence of elevated levels of AAT in livestock within village herds is a useful indicator of risk for rHAT in Uganda. Integrated veterinary and medical surveillance is key to successful control of zoonotic rHAT. Electronic supplementary material The online version of this article (doi:10.1186/s40249-016-0224-8) contains supplementary material, which is available to authorized users.


Background
Human African Trypanosomiasis (HAT) or sleeping sickness, comprises two distinct tsetse transmitted parasitic diseases, both of which are fatal if they are not treated: chronic, Trypanosoma brucei gambiense HAT (gHAT) and acute Trypanosoma brucei rhodesiense HAT (rHAT) [1]. Uganda has active foci for both rHAT and gHAT which have been geographically separated for more than a century [1,2]. gHAT lies in the northwest in districts adjacent to South Sudan and is relatively stable [3] while the area affected by rHAT, originally restricted to the shores of Lake Victoria [4] began expanding from the mid-1980s. An rHAT epidemic began in Tororo district in the mid-1980s spreading to Butaleja and Busia districts [5][6][7] spreading to Soroti and Serere districts in 2000 [7,8] and subsequently to Kaberamaido and Dokolo districts by 2004 [9,10]. By 2005, the areas affected by gHAT and rHAT were only 150 km apart [11].
The causal agent of rHAT, T. b. rhodesiense, co-exists in a range of non-human hosts [12] with a suite of other trypanosomes that cause African Animal Trypanosomiasis (AAT). A series of studies showed that cattle in Uganda carried human infective T. b. rhodesiense [6,7,9] and that movements of infected cattle from districts within the rHAT focus to rHAT free districts [8,13] were responsible for the spread of rHAT in Uganda. The epidemic in Soroti was directly linked to importation of cattle through Brooks Corner cattle market [8].
Both HAT and AAT are major priorities for Uganda [14,15] with approximately one-third of the national herd at risk from AAT [15], with serious economic losses [16]. Infection with T. b. rhodesiense in indigenous cattle is largely asymptomatic; livestock in rural areas of Uganda are not routinely treated unless they show co-infection with T. congolense or T. vivax. Cattle are long-term investments and untreated infected animals comprise a persistent zoonotic reservoir of rHAT parasites posing a threat to human health.
The expansion of the rHAT focus poses a significant health risk (Fig. 1) and convergence of the gHAT and rHAT foci would compromise existing approaches for diagnosis and treatment [17]. T. b. rhodesiense HAT is significantly under-reported [18] and identification of human cases remains a challenge to the health system [19]. Controlling rHAT requires veterinary intervention [20]. A single trypanocidal intervention, targeted to cattle in Kamuli and Soroti districts, at high risk for rHAT in 2002 reduced the prevalence of T. b. brucei s.l. and T. b. rhodesiense and resulted in a significant decrease in human rHAT cases [21] suggesting that if sufficient head of cattle were treated, rHAT transmission could be stopped [22]. In 2006, a Public Private Partnership, Stamp Out Sleeping Sickness (SOS) [23], aimed to remove the reservoir of T. b. rhodesiense in Soroti, Dokolo, Lira, Apac and Kaberamaido districts by mass treating cattle [24].
Here we examine the impact of mass cattle trypanocidal intervention on the parasite burden (T. brucei s.l., T. vivax and T. congolense and human infective T. b. rhodesiense parasites) in cattle within the SOS intervention zone, at village level, examining villages that had and had not previously experienced a human case of rHAT.

Description of the study area
The study area comprised Kaberamaido and Dokolo districts (in the Eastern and Northern regions respectively) in Uganda, which have been affected by rHAT since 2004. The districts border the northern shore of Lake Kyoga with a combined area of approximately 2 467 km 2 and a human population of approximately 261 000 [25]. The main economic activities within the study area are agriculture and fishing, with the majority of the population engaged in subsistence farming [26]. The first cases of rHAT were reported in Kaberamaido and Dokolo districts in 2004 and the continued presentation of new cases indicates active transmission in both districts. A mass chemotherapuetic intervention was undertaken in cattle, in these districts in 2006, in an attempt to prevent the northwards spread of rHAT [24]. According to the livestock census undertaken in 2008, Dokolo had 58 902 cattle and Kaberamaido 76 109 cattle (Table 1).

Description of the intervention
Between April and May 2008, due to the persistence of rHAT cases in parishes lying on the border between Dokolo and Kabaramiado [10,27], a further intervention, comprising a single dose of trypanocide, was given to 31 486 cattle across 29 rHAT affected parishes ( Table 2). All villages in the districts were georeferenced using hand held global positioning systems prior to the intervention (GPS: Garmin, E-trex venture) [26]. Intervention sites were selected at parish level (between 3-10 villages); an average village had less than 100 households with between 30 and 50% of households owning cattle, typically small herds of 3-10 cattle although some larger herds of 50-100 cattle were present [28]. All available village cattle were injected with a single dose of Veriben (diminazene aceturate -Ceva Santé Animale) at 7.5 mg/kg; at the same time a single application of the insecticide Vectocid (Ceva Santé Animale) was applied to the legs and belly [29][30][31]. Treatment coverage data was obtained from estimates based on the total number of cattle treated in the parishes under the SOS intervention and from farmer reporting of intervention status of cattle sampled 6 months after intervention.

Study design
To examine the parasite dynamics before and after intervention, two groups of villages were selected: villages that had experienced a case of human rHAT in the previous 6 months (rHAT +ve villages) and those that had never reported a case (rHAT -ve villages). Records detailing all rHAT patients between January and June 2007 were obtained from the two health centres established for treatment of rHAT in Dokolo and Kaberamaido: Lwala health centre in Kaberamaido district and Serere health centres in Soroti district [18]. Cases were cross-referenced to the georeferenced village data to classify all villages into two groups: case villages and no-case villages. A case village (rHAT +ve village) was defined as having one or more rHAT cases diagnosed in the 6 month period January -June 2007. Twenty villages were randomly selected to monitor the impact of the intervention: 10 rHAT +ve villages that had reported a rHAT case between January -June 2007 and 10 that had never reported a case (HAT -ve villages). Figure 2 shows the location of all study villages and the intervention sites. The village HAT case status is shown in Fig. 3 which also indicates the location of the HAT treatment centre.

Sample size calculation
The required sample size was based on the requirement for comparing two proportions (rHAT case vs. non rHAT case villages) [32], with adjustment for cluster sampling [33] at the village level. An estimated T. brucei s.l. prevalence of 20% was used, alongside an intercluster variance (V c ) of 0.013 [34] to allow for a differences in trypanosome prevalence to be detected at 95% confidence and 80% power [32]. This called for a  (1 340). Cattle were systematically sampled by order of presentation at each village sampling site, up to a maximum of 100 animals per village, to maintain sufficient sample size of cattle from rHAT case and non-rHAT case villages overall, and to compensate for villages with fewer than 67 cattle.

Sample collection
Samples of blood (100 μl) were taken from the ear vein of cattle in the rHAT+ve (case) villages and rHAT+ve (no case) villages, before (April -May 2008) and 6 months after intervention (October 2008). In total 3 549 cattle were sampled: 1 658 before intervention and 1 891 after 6 months intervention. Before taking blood samples, the livestock owner was asked whether the animal had been treated during the mass intervention. One hundred microliters of blood was drawn from the ear capillary using a lancet, collected into two 50 μl capillary tubes and spotted on a FTA™ card (Whatman, Maidstone, Kent, UK). Cards were air dried at room temperature for 24 h and stored in multi-barrier pouches with desiccant (Whatman, UK).

Laboratory analysis
Samples were analysed by PCR to identify all trypanosome species and subspecies present. Five 3 mm discs were punched from each blood sample and processed as previously described [35][36][37]. Discs were washed four times, twice with FTA purification reagent and twice with TE buffer, before being air dried and then heated at 90°C for 30 min suspended in 5% (w/v) Chelex solution. Five microliters of eluted sample solution, was used to seed all PCR reactions. Previously published species specific primers were used to detect T. brucei s.l., T. vivax, and T. congolense savannah, in three separate PCR reactions [35][36][37]. Samples positive for T. brucei s.l. were further tested for T. b. rhodesiense using a multiplex PCR to discriminate T. b. brucei and T. b. rhodesiense [38]. Standard PCR amplifications for all reactions were carried out in 25 μl mixtures; reaction conditions, primer sequences and cycling conditions were as previously published [35][36][37]. One positive control [genomic deoxyribonucleic acid (DNA)] and one negative control (extract from blank FTA™ disc) were run with each reaction. PCR products were resolved via electrophoresis on 1.5% agarose gels using GelRed DNA stain and run at 100v for a minimum of 45 min in the presence of a molecular marker, until band size could be easily determined. PCR products were visualised and documented with a BioRad GelDock™ imaging system.

Statistical analysis
The prevalence of T. brucei s.l., T. b. rhodesiense, T. vivax and T. congolense savannah trypanosome infection detected by PCR were expressed as a percentage, and exact binomial 95% confidence intervals were computed (R, version 2.0.1). Fisher's Exact test was used to compare trypanosome prevalence in rHAT +ve and rHAT -ve villages and between the pre-and 6 months post-intervention sampling.

Results
One thousand six hundred and fifty-eight cattle blood samples were obtained before intervention from the 20 village sites; 854 from animals in HAT +ve villages and 804 from cattle in HAT -ve villages. Of these, 642 samples were positive for one or more species of trypanosome; 362 from HAT +ve villages (prevalence 42.4%, 95% CI = 45.8-39.1%) and 280 were from HAT -ve villages (prevalence 34.8%, 95% CI = 38.2-31.5%). The prevalence of cattle trypanosomiasis (AAT) was significantly higher in HAT +ve villages (P < 0.001).
T. brucei s. l T. brucei s.l. was the most common species detected both before and after intervention (Fig. 4). T. brucei s.l. prevalence varied from 15.6 to 53.0% between villages. The prevalence of T. brucei s.l. was significantly higher in HAT +ve villages than in HAT -ve villages pre-intervention (P = 0.002) and post-intervention (P = 0.008).
Across all 20 villages, the prevalence of T. brucei s.l. was significantly lower 6 months after the intervention, irrespective of village HAT status. For HAT +ve villages, the prevalence of T.
The intervention also impacted in the proportion of human infective T. b. rhodesiense within the population of circulating T. brucei s.l. in cattle (Fig. 5). Critically, the proportion of T. b. brucei that was T. b. rhodesiense (human infective) decreased after intervention. Prior to the intervention in rHAT +ve and rHAT -ve villages combined 8.3% (39/472, 95% CI = 11.1-5.9%) of circulating T. brucei s.l. were human infective; post-intervention this reduced to 4.1%, (14/342, 95% CI = 6.8-2.3%). Prior to the intervention, cattle in 15 of the 20 study villages had T. b. rhodesiense prevalences ranging from 1 to 5%. T. b. rhodesiense was not detected in any cattle in three HAT +ve villages and two HAT -ve villages. Six months post-intervention, only eight of the 20 study villages contained T. b. rhodesiense positive cattle; 75% (6/8) of these were from HAT +ve villages. The locations of villages harbouring cattle infected with T. b. rhodesiense before and after intervention are shown in Fig. 6.
Intervention coverage impacted on identification of Trypanosoma infection by PCR (Table 4). Of the 342 animals found positive for T. brucei s.l., 47% at the 6 month post-intervention sampling were from cattle that had not participated in the intervention (162/342). Within the HAT +ve villages 52% (101/195) of the animals found positive for T. brucei, had not participated in the intervention. Of the 12 T. b. rhodesiense infections identified in HAT +ve villages, 66% (8/12) were observed in cattle reported not to have particpated in the intervention . Of the 231 T. congolense infections identified 37% were from animals reported not to have participated in the intervention and 63% from those reported to have particpated.

Discussion
Microscopy consistently underestimates the prevalence trypanosomes in livestock [38,39], so the use of sensitive and specific 'field friendly' diagnostic tools are key to our understanding of the epidemiology of bovine and zoonotic trypanosomiasis [39]. Cattle have been shown to play a major role in the epidemiology of rHAT in Uganda [6-9, 21, 33]. This is one of the first studies to use molecular tools to evaluate large-scale field interventions targeted at T. b. rhodesiense, the causal agent of rHAT, in cattle -the principal animal reservoir. Previous studies have shown the importance of eliminating the reservoir of human infective T. b. rhodesiense parasites from domestic cattle to prevent human infection [21,32].
Across the 20 randomly selected villages within the intervention zone we observed a significant reduction of T. b. rhodesiense in cattle 6 months after the intervention (from 2.40 to 0.74%), a 10-fold decrease in prevalence of T. b. rhodesiense in HAT -ve villages compared to a 50% reduction of T. b. rhodesiense in HAT +ve villages. The number of villages with T. b. rhodesiense Table 4 Trypanosoma infection and intervention participation status of cattle sampled at 6 months post-intervention as reported by livestock-keepers Village HAT status Intervention participation Cattle sampled (n) TBSL + (n) % TBSL + SRA + (n) % SRA + TV + (n) % TV + TC + (n) % TC +  [33,40] suggesting that once T. b. rhodesiense is established in a village, spilling over from cattle into humans, eliminating the cattle reservoir of T. b. rhodesiense in the village becomes more difficult. HAT transmission is influenced by many factors, including tsetse population density, tsetse susceptibility [41] and proximity to around a central point such as a market, watering point or swampland [8,10,40]. In this study villages that had experienced a case of rHAT also showed a significantly higher prevalence of T. brucei s.l. and T. congolense in the village cattle than those that had not had a case of rHAT, further evidence of a link between AAT and T. b. rhodesiense HAT. Conditions for transmission of all species of trypanosome appeared more favourable in HAT +ve villages.
T. vivax all but disappeared across the study villages post-intervention, but the prevalence of T. congolense increased (although a large proportion of infected animals had not participated in the intervention). Drug resistance in cattle has been reported for T. congolense in parts of Ethiopia [42] and Burkina Faso [43] but this has not been reported in Uganda where T. congolense infection is rarely reported or treated. Eliminating T. vivax and reducing T. brucei s.l. in cattle may simply offer opportunities for T. congolense to establish in these animals [44].
To interrupt T. b. rhodesiense transmission, where cattle are the principal reservoir of T. b. rhodesiense, it is calculated that 86% of the cattle population must be treated [22]. The discrepancy between the coverage estimates calculated from the livestock census data (91.3%) and the lower estimates obtained provided from the livestock-keepers (55.2%) show the difficulties of calculating coverage in the absence of robust and verifiable village and parish cattle numbers; true coverage probably lay somewhere between the two.
In Dokolo, cattle were treated in 13 of 33 parishes in the district, or 39% of the district; coverage was estimated at 20% of the district cattle. Likewise, for Kaberamaido, the mass intervention covered 50% of the district (16 of 32 parishes) but using, 2008 livestock census data, only 27% of the district cattle population are estimated to have been treated. The parishes treated during the mass intervention had relatively low cattle populations compared to other areas of the district.
Accurate information about the number of cattle owned by individuals and households in this region is difficult to obtain. Extensive cattle rustling in the late 1980s decimated livestock in the region and the Lango and Teso economy [45] and cattle restocking projects [46] were ongoing during 2008 when this intervention took place [13]. Villagers may answer questions about donor-led interventions in ways that take into account how their answers will influence future benefits [47] and it is likely that a proportion of livestock-keepers misrepresented the treatment status of their cattle in the hope that their herd would be re-treated.

Conclusions
Treatment of cattle produced a significant decrease in the prevalence of T. brucei s.l., human infective T. b. rhodesiense and T. vivax infection in village cattle herds and impacted on the proportion of T. brucei s.l. that were human infective: from 1:12 human infective T. brucei s.l. before the intervention to 1:33 post-intervention.
The observation that is more difficult to remove T. b. rhodesiense from villages that have experienced a rHAT case, makes a strong case for improved information sharing between human health providers, animal health providers and the community. The prevalence of AAT in villages can serve as an additional indicator of human health risk in these communities, alerting medical services to survey human populations. Incorporating participatory research in conjunction with traditional disease modelling approaches could lead to improved disease control [48].
To prevent the continued migration of rHAT, point of sale trypanocide treatments should be applied for all inter-district cattle movements and interventions that target rHAT and AAT should be scaled up in all rHAT affected districts. Interventions that additionally target tick-borne diseases are welcomed by local communities [45], where trypanosomiasis and tick-borne diseases reduce draft cattle output by 21% and household income from the use of oxen (estimated at $245 USD annually) by 32% [49]. Appropriate scaled application of trypanocidal drugs, followed by routine application of veterinary pyrethroids to prevent re-infection and manage tick infestation, offer a sustainable solution for zoonotic HAT, AAT and tick-borne disease control [29,31,49,50] offering a win-win for human and animal health. Availability of data and materials All data and materials are available on request from the corresponding author.