T. b. rhodesiense and T. b. gambiense

The two sub-species of human-infective African trypanosomes have been categorised by criteria including geographic distribution, morphology, reservoir host range and disease progression. However, the defining characteristics proposed by Hoare (Hoare, Reference Holzmuller, Biron, Courtois, Koffi, Bras-Goncalves, Daulouede, Solano, Cuny, Vincendeau and Jamonneau1972) were that T. b. rhodesiense causes acute infections leading to a rapid onset of late stage and death and is distributed throughout East and Southern Africa while T. b. gambiense causes chronic infections with a slow onset of late stage and death and is distributed throughout West and Central Africa. Over the past 30 years, with most research into African trypanosomiasis taking place in molecular biology labs far removed from endemic countries, this view of the disease has become fairly entrenched amongst parasitologists. Genetic analyses have shown that the relationships of the taxons within T. brucei sl. are complex and do not fit neatly into the three subspecies originally proposed by Hoare. Essentially T. b. rhodesiense may be regarded as a host-range variant of the underlying non-human infective T. b. brucei population (Gibson, Reference Graf2002). Human infectivity in T. b. rhodesiense is a function of the serum resistance-associated (SRA) gene, which gives a phenotype of resistance to the trypanolytic factors found in human plasma high density lipoprotein (Vanhamme et al. Reference Vanhollebeke, Truc, Poelvoorde, Pays, Joshi, Katti, Jannin and Pays2003). Meanwhile, the human-infective trypanosomes circulating in West Africa are divided into at least two groups. Group 1 T. b. gambiense has a clonal population structure, shows low virulence in experimental rodents (Inoue et al. Reference Inverso, De Gee and Mansfield1998). Group 2 T. b. gambiense are genetically heterogeneous, virulent in rodent infections, show instability of human serum resistance and have biological and genetic similarity to T. b. brucei (MacLeod et al. Reference Macleod, Tweedie, Mclellan, Taylor, Hall, Berriman, El-Sayed, Hope, Turner and Tait2001). Confirmation of this close relationship was provided by the successful genetic cross carried out between an East African T. b. brucei isolate (STIB247) and a West African human Group 2 T. b. gambiense isolate (STIB386) (Jenni et al. Reference Kaushik, Uzonna, Zhang, Gordon and Tabel1986). The mechanism of human serum resistance is not known in either clade of T. b. gambiense and to date no analogue of the SRA gene has been described.

T. b. rhodesiense: Variability in clinical disease

A wide spectrum of clinical symptoms, disease severity and duration of illness has been reported in T. b. rhodesiense HAT in East Africa. T. b. rhodesiense was first described in an individual who had been infected 1908 in the Luangwa Valley in Zambia (Ross and Thomson, Reference Smith and Bailey1910), and soon after an epidemic was recorded in southern Nyasaland (now Malawi) (Stannus and Yorke, Reference Sternberg and Tait1911). The disease was described as a chronic disease with incubation times of several months. In subsequent reports of HAT in the same areas, similar disease symptoms were also observed, with no cases presenting with a chancre, and incubation times of up to 10 months (Rickman, Reference Ross and Thomson1974; Buyst, Reference Caljon, Van Den Abbeele, Stijlemans, Coosemans, De Baetselier and Magez1977; Foulkes, Reference Garcia, Jamonneau, Magnus, Laveissiere, Lejon, N'guessan, N'dri, Van Meirvenne and Buscher1981). There have also been reports of asymptomatic cases (parasites detected by microscopy but no symptoms of HAT) in Zambia (Wurapa et al. Reference Wurapa, Dukes, Njelesani and Boatin1984) and Botswana (Apted et al. Reference Balmer, Stearns, Schotzau and Brun1963), although these must be interpreted with some caution with respect to the identity of the trypanosomes concerned. This mild presentation of HAT was completely different from the disease that manifested in the Busoga epidemic in South East Uganda in 1940, with patients exhibiting severe disease pathology and becoming moribund within 4–6 weeks. This spectrum of disease and the transitional presentations that are evident particularly in Tanzania (Komba et al. Reference Lapeyssonnie1997) led to Ormerod's proposal that T. b. rhodesiense HAT spread from South to North with increasing acuteness and virulence (Ormerod, Reference Ormerod1961, 1967). This is now known to not be entirely correct. A re-evaluation of archival data on the Busoga epidemic in Uganda suggest it actually began in 1900, thus predating the 1908 outbreak in the Luangwa Valley (Welburn et al. Reference Wery and Burke2001). Moreover, rather than a spread of a human-infective and monophyletic clade of T. b. rhodesiense, population genetic analysis indicates at least two and possibly multiple allopatric origins of human infectivity out of the underlying local T. brucei brucei populations, possibly involving genetic exchange (MacLeod et al. Reference Macleod, Tweedie, Mclellan, Taylor, Hall, Berriman, El-Sayed, Hope, Turner and Tait2001). Nevertheless, the overall trend of increasing virulence from South to North is confirmed in a recent set of studies into the clinical immunology of HAT in Malawi and Uganda, with fast progression of HAT to late stage in Uganda and slow progression in Malawi with an under-representation of late stage cases (MacLean et al. Reference Maclean, Odiit, Macleod, Morrison, Sweeney, Cooper, Kennedy and Sternberg2004).

T. b. gambiense: Variability in clinical disease and asymptomatic cases

HAT caused by T. b. gambiense infections also results in a range of outcomes, with evidence having been presented for a range of virulence from acute (Truc et al. Reference Turner, Aslam and Dye1997) at one extreme to asymptomatic cases at the other where parasites are detected either directly by microscopy (Wery and Burke, Reference Wurapa, Dukes, Njelesani and Boatin1972), or indirectly by PCR (Jamonneau et al. Reference Jenni, Marti, Schweizer, Betschart, Le Page, Wells, Tait, Paindavoine, Pays and Steinert2004) or serology (Garcia et al. Reference Garcia, Jamonneau, Sane, Fournet, N'guessan, N'dri, Sanon, Kaba and Laveissiere2000) over sustained follow-up. As with all clinical case reports of HAT, some (particularly older) case studies must be interpreted with caution having been carried out in an era when diagnostic, staging and species identification techniques were quite limited. This problem was highlighted in a recent meta-analysis of 77 published accounts that highlighted difficulties in reaching conclusions about the extent and significance of ‘slow’ or asymptomatic cases (Checchi et al. Reference Clayton2008). These included a bias towards single case studies, the fact that many of studies relied on relatively insensitive microscopic detection of trypanosomes, and the possibility that the infections concerned were not T. b. gambiense. However, a subset of these studies presented very persuasive evidence for asymptomatic HAT. For example, in a large-scale survey of 4500 subjects in Kongo Central Province of the DRC, Wery and Burke (Wery and Burke, Reference Wurapa, Dukes, Njelesani and Boatin1972) identified a group of 17 individuals who displayed no symptoms of HAT but gave a positive serological reaction against fixed trypanosomes. These individuals were monitored over a 12-month follow-up period, and in 3 of these individuals trypanosomes were intermittently observed in thick blood films, although parasites were never observed in lymph node aspirates or CSF. A group of 63 individuals who had been diagnosed with HAT on the basis of serology in the Sinfra focus of Cote d'Ivoire but who refused treatment (Garcia et al. Reference Garcia, Jamonneau, Sane, Fournet, N'guessan, N'dri, Sanon, Kaba and Laveissiere2000) provided further evidence of asymptomatic or mild infection (Checchi et al. Reference Clayton2008) in which mild symptoms presented periodically and parasitaemia could not be confirmed directly but was detectable by PCR. In a follow-up period of 7 years, while some of these individuals went on to develop HAT and others left the study, 6 individuals remained asymptomatic throughout. PCR analysis indicated infection with Group 1 T. b. gambiense plus a second hitherto unknown group of T. brucei (Jamonneau et al. Reference Jenni, Marti, Schweizer, Betschart, Le Page, Wells, Tait, Paindavoine, Pays and Steinert2004). Similarly, in a 2-year longitudinal study of 77 aparasitaemic individuals identified in a mass serological screening programme in Sinfra, 50% remained seropositive and aparasitaemic over the entire study period although one developed HAT (Garcia et al. Reference Garcia, Jamonneau, Sane, Fournet, N'guessan, N'dri, Sanon, Kaba and Laveissiere2000). Finally, no review of asymptomatic T. b. gambiense cases is complete without reference to Feo isolate widely used in laboratory studies which was obtained from a Togolese female who had been resident in Europe for 21 years (Lapeyssonnie, Reference Lejon, Lardon, Kenis, Pinoges, Legros, Bisser, N'siesi, Bosmans and Buscher1960).

Experimental Models

Extensive work with experimental rodent models has demonstrated that host genotype has a major impact on the progression of trypanosome infection. Inbred mouse models range from highly susceptible to sub-tolerant to trypanosome infection, although importantly no resistant model has been identified (Antoine-Moussiaux et al. Reference Apted, Smyly, Ormerod and Stronach2008). Interpretation of the models is complicated by the fact that these studies have used a range of trypanosome isolates, many after extensive serial-passage. Furthermore, there do seem to be subtle differences in the behaviour of T. brucei sl. and T. congolense in models, the latter species being highly relevant to the understanding of the phenomenon of trypanotolerance in cattle (Kemp and Teale, Reference Kennedy1998). Sub-tolerance in T. brucei-infected mice is not related to H-2 (Clayton, Reference Collins, Brooks and Chakravarti1978) and segregation analysis demonstrates that it is under polygenic control (Greenblatt et al. Reference Guilliams, Bosschaerts, Herin, Hunig, Loi, Flamand, De Baetselier and Beschin1984; Murray et al. Reference Naessens, Mwangi, Buza and Moloo1984). Similar complex genetic control of sub-tolerance in mice to T. congolense has been demonstrated (Morrison and Murray, Reference Murray, Trail, Davis and Black1979). There has been considerable effort to identify the genes associated with sub-tolerance in T. congolense infection as it is thought that these will be involved in cattle trypanotolerance. Susceptibility and sub-tolerance to T. congolense infection in mice is associated with three quantitative trait loci (QTL), designated Tir1, Tir2, Tir3 (Kemp et al. Reference Kemp and Teale1997). Fine mapping (Iraqi et al. Reference Iraqi, Sekikawa, Rowlands and Teale2000) has improved the resolutions of these QTLs further. For example, the QTL with greatest effect on survival, Tir 1 on chromosome 17 is now mapped with a 95% CI of 0·9 cm, and contains the genes for TNF-α and HSP70.1. Experimental studies in TNF gene deleted mice have demonstrated that TNF plays and essential role in survival (Iraqi et al. Reference Jamonneau, Ravel, Garcia, Koffi, Truc, Laveissiere, Herder, Grebaut, Cuny and Solano2001; Magez et al. Reference Magez, Stijlemans, Baral and De Baetselier2007) and thus the TNF gene is a strong candidate gene for Tir1. Fine mapping also led to the subdivision Tir3. One of these QTLs (Tir3b) maps close to the chromosomal location of both the IL-10 gene and a key regulatory gene for IL-10 synthesis on chromosome 1 (Kaushik et al. Reference Kemp, Iraqi, Darvasi, Soller and Teale2000), although it must be emphasised that the resolution of this QTL is still limited with a 95% CI of 10 cm. A similar QTL analysis has not taken place in T. brucei infection models, and this is important as TNF-α apparently plays a far more complex role here, including both protective (Magez et al. Reference Magez, Radwanska, Beschin, Sekikawa and De Baetselier1993, Reference Magez, Radwanska, Drennan, Fick, Baral, Allie, Jacobs, Nedospasov, Brombacher, Ryffel and De Baetselier1999) and pathology-inducing effects (Magez et al. Reference Masocha, Robertson, Rottenberg, Mhlanga, Sorokin and Kristensson2004). Evidence from immunological studies of T. brucei-susceptible and sub-tolerant strains of inbred mice indicates that the level of ‘tolerance’ to infection is not directly related to the ability to control parasitaemia but rather the regulation of the inflammatory response (Magez et al. Reference Masocha, Robertson, Rottenberg, Mhlanga, Sorokin and Kristensson2004; Bosschaerts et al. Reference Brookes2008; Guilliams et al. Reference Hoare2008), suggesting a possible functional role for IL-10 and IL-10 regulatory genes within Tir3b as anti-inflammatory factors. Interestingly, this is consistent with earlier studies of experimental mouse-host genetics indicating that survival and parasitaemia control are independently segregating traits (De Gee et al. Reference Diffley, Scott, Mama and Tsen1988).

Thus, as host genotype controls severity of trypanosome infection in mouse models of both T. brucei and T. congolense, it is a reasonable conjecture that this is also the case in human hosts. This could be regarded as analogous to the phenomena of trypanotolerance and susceptibility in bovine trypanosomiasis (Kemp and Teale, Reference Kennedy1998).

Human Infections

The idea that variation in disease severity in human T. b. rhodesiense infections was related to host genetics was proposed over 30 years ago (Buyst, Reference Caljon, Van Den Abbeele, Stijlemans, Coosemans, De Baetselier and Magez1977) after studies of Zambian HAT patients showed that host ethnicity and previous exposure to human infective trypanosomes were related to disease pathology (Buyst, Reference Buyst1974). The hypothesis proposed was that the mild disease phenotype of Zambian T. b. rhodesiense HAT (and possible asymptomatic infection) was due to natural selection for greater tolerance of trypanosome infection in populations of Bantu ethnicity, the ancestral populations of whom were likely to have been exposed to human infective trypanosomes for tens of thousands of years in West Africa (Buyst, Reference Buyst1974; Rickman, Reference Ross and Thomson1974). In contrast Nilo-Saharan populations who migrated to East Africa from tsetse-free areas in the north of sub-Saharan Africa underwent acute disease, as described in Ethiopia (Hutchinson, Reference Inoue, Narumi, Mbati, Hirumi, Situakibanza and Hirumi1971). This hypothesis was extended to account for highly acute infections in Caucasian Europeans with no history of previous exposure in both T. b. rhodesiense (Duggan and Hutchinson, Reference Ebert1966) and T. b. gambiense endemic areas (Graf, Reference Greenblatt, Diggs and Rosenstreich1929). The use of ethnicity as a proxy for genetic variation makes assumptions on the degree of genetic isolation between groups that are difficult to substantiate. Ethnographic variation is often paralleled with a variety of anthropological confounding factors that could impact on both exposure to parasites and the general levels of resilience to infectious disease (Garcia et al. Reference Gibson2002), and the substantial number of case-study reports of ethnographic and geographic variability in response to HAT may result from publication-bias (essentially aetiologies not consistent with the norm being more likely to be published). It is also likely that the spatial separation of different communities could lead to their exposure to parasite genetic variants as has been shown in Uganda (MacLean et al. Reference Maclean, Odiit, Macleod, Morrison, Sweeney, Cooper, Kennedy and Sternberg2004). Therefore, while the ‘mild’ infection phenotype in Malawi is differentiated from ‘acute’ infection in Uganda by a reduced TNF-α host-response and increased counter-inflammatory cytokine production (MacLean et al. Reference Maclean, Odiit, Macleod, Morrison, Sweeney, Cooper, Kennedy and Sternberg2004), it is not possible to determine if this is a function of host or parasite variability.

A more robust method to investigate association between host-genotype and HAT severity involves single nucleotide polymorphism (SNP) analysis in the host-population. The density of SNPs in the human genome has been estimated to be once every 250–1000 bp accounting for about 90% of the DNA sequence variants (Collins et al. Reference Courtin, Argiro, Jamonneau, N'dri, N'guessan, Abel, Dessein, Cot, Laveissiere and Garcia1998). This high density and their mutational stability make SNPs a useful marker for human population genetics and for mapping susceptibility genes for complex diseases (Brookes, Reference Buyst1999). In HAT, this analysis has been applied to SNPs associated with the regulation of expression of key mediators of inflammatory responses, based on their known relevance to disease progression severity. In study populations exposed to T. b. gambiense in Cote d'Ivoire and the DRC, SNPs in the IL-10 gene promoter and in exon 5 of IL-6 gene were associated with a lower risk of developing disease, while a weak but significant association was found for a higher risk of developing disease in individuals with an SNP in the TNF-α promoter (Courtin et al. Reference Courtin, Argiro, Jamonneau, N'dri, N'guessan, Abel, Dessein, Cot, Laveissiere and Garcia2006, Reference Courtin, Milet, Jamonneau, Yeminanga, Kumeso, Bilengue, Betard and Garcia2007). It is of interest that the TNF and IL-10 genes are already candidates for the QTLs Tir1 and Tir3b in the control of T. congolense in mice described above. Also, clinical studies of HAT have shown TNF-α production correlates with pathology (Okomo-Assoumou et al. Reference Ormerod1995), and CNS infection is associated with high levels of intrathecal IL-10 and IL-6 synthesis (Lejon et al. Reference Maclean, Chisi, Odiit, Gibson, Ferris, Picozzi and Sternberg2002; MacLean et al. Reference Maclean, Odiit and Sternberg2006) giving a further possible biological context for the SNP associations. However, if these SNPs are really influencing the likelihood of HAT infection, the implication is that there must be a phenotype that is resistant to infection (as opposed to an asymptomatic state) that has never been demonstrated clinically or experimentally. Furthermore, this type of study also requires assumptions to be made on the uniformity of exposure to infected tsetse flies and a range of environmental and socioeconomic factors in the study population and family groups.

In conclusion, while experimental studies strongly suggest that host-genotype will control the severity of HAT, to date the evidence from clinical studies is still lacking. There is, however, one rather atypical example of a clearly defined host-genetic effect on the development of trypanosomiasis identified in an individual with a mutant Apo-L1 component of HDL who developed infection with T. evansi (Vanhollebeke et al. Reference Vassella, Reuner, Yutzy and Boshart2006). It is not known if Apo-L1 polymorphisms could lead to variable innate resistance to brucei trypanosomes in Africa.

Experimental Models

Could the variation in disease progression and severity observed in HAT be related to virulence variation between parasite genotypes? This question has been addressed in experimental mouse infections, but care must be taken in the interpretation of these studies due to the different traits that have been used as measures of virulence. Also it is important to distinguish variation in virulence between field isolates (inter-clonal) from the stable intra-clonal increased virulence as manifest in peak parasitaemia and decreased survival resulting from serial syringe passage of clonal populations trypanosomes in rodents (Turner et al. Reference Turner, Aslam and Angus1995), although there may, of course, be common mechanisms. While serial-syringe passage of trypanosomes is an important model for natural evolution of virulence in parasites (Ebert, Reference Fairbairn and Godfrey1998), the underlying mechanisms remain unclear but are not related to antigenic variation (Inverso et al. Reference Iraqi, Clapcott, Kumari, Haley, Kemp and Teale1988). Serial passage tends to drive trypanosomes populations to monomorphism (i.e. populations in which slender-stumpy differentiation does not take place), and monomorphic trypanosomes appear to be insensitive to density sensing signals causing slender-stumpy differentiation (Vassella et al. 1997). This cannot be the entire explanation since entirely monomorphic populations also show increases in virulence (as measured by growth rate and parasitaemia) during serial passage (Diffley et al. Reference Duggan and Hutchinson1987).

What is of more interest here is inter-clonal virulence variation between distinct field-isolates of T. brucei sl. For example, when Balb/c mice were infected with early passage T. brucei clonal stocks, considerable variation in acute stage growth rates and peak parasitaemias were observed (Turner et al. Reference Turner, Aslam and Angus1995). Similar variation in virulence has been described for field-isolates of T. congolense (Masumu et al. Reference Morrison, Mclellan, Sweeney, Chan, Macleod, Tait and Turner2006). It is becoming clear in mouse model infection studies that inter-clonal differences in virulence are associated with variation in inflammatory regulation in the host, with more virulent strains of parasite causing increased inflammatory-mediated pathology (anaemia and organomegaly); for example, the survival of experimental mice infected with field isolated stocks of T. b. gambiense was not related to peak parasitaemia but to alternative activation of macrophages that down-regulate inflammatory responses (Holzmuller et al. Reference Holzmuller, Grebaut, Peltier, Brizard, Perrone, Gonzatti, Bengaly, Rossignol, Aso, Vincendeau, Cuny, Boulange and Frutos2008a). The potential role of inflammatory regulation in setting virulence was supported in a reverse genetic study with T. brucei in which the phospholipase-C (PLC) gene responsible for the cleavage of the membrane form variant surface glycoprotein (mfVSG) had been deleted. These mutants showed dramatically reduced virulence (Webb et al. Reference Welburn, Fevre, Coleman, Odiit and Maudlin1997). The cleavage products of mfVSG, soluble VSG and dimyristoyl glycerol, are mediators of classical macrophage activation (Magez et al. Reference Magez, Truyens, Merimi, Radwanska, Stijlemans, Brouckaert, Brombacher, Pays and De Baetselier2002) and therefore potential drivers of inflammatory pathology and thus may be play a key role as virulence factors (Namangala et al. Reference Namangala, De Baetselier, Noel, Brys and Beschin2000). This concept is also supported by the observation that treatment of experimental mice with the GPI moiety of mfVSG prior to trypanosome infection leads to an amelioration of immunopathology and extended survival (Stijlemans et al. Reference Truc, Formenty, Diallo, Komoin-Oka and Lauginie2007).

While reverse genetic approaches using trypanosomes may yield clues into the nature and identity of parasite virulence factors, given the vast range of potential candidates, a more targeted approach is required involving forward (classical) genetics (Morrison et al. Reference Morrison and Murray2009). This has recently become available thanks to the availability of high-resolution genetic maps of T. brucei (MacLeod et al. Reference Magez, Lucas, Darji, Songa, Hamers and De Baetselier2005) and the ability to conduct laboratory crosses between different isolates (Sternberg and Tait, Reference Stevens and Godfrey1990). The inheritance of relevant traits (in this case virulence) may then be subject to genetic linkage analysis. Parental strains of T. b. brucei (STIB247 and TREU927) differed in virulence phenotype as measured by traits including anaemia induction and degree of splenomegaly, hepatomegaly and reticulocytosis. A major QTL, TbOrg1, that accounted for the variation in splenomegaly and hepatomegaly was identified on chromosome 3, although at present the resolution of this QTL is too low to identify candidate genes (Morrison et al. Reference Morrison and Murray2009). A second weaker QTL was identified on chromosome 2 contributing to splenomegaly, hepatomegaly and reticulocytosis. The key to progressing from the characterization of QTLs containing virulence genes to the identification of gene products will lie in post-genomic approaches such as proteomics and transcriptomics. Thus host-responses to strains STIB247 and TREU927 have also been compared using microarray analysis (Morrison et al. Reference Morrison, Tait, Mclellan, Sweeney, Turner and Macleod2010) with the aim of identifying the targets of putative virulence factors. The most significant differentially regulated responses between the virulence variants were associated with macrophage activation phenotype, consistent with previous studies with PLC gene deleted attenuated virulence trypanosomes (Namangala et al. Reference Okomo-Assoumou, Daulouede, Lemesre, N'zila-Mouanda and Vincendeau2001) described above. Meanwhile, the ‘secretome’ of virulence-variant parasites may be analysed using 2D gel electrophoresis and DIGE techniques to identify virulence type-specific proteins (Holzmuller et al. Reference Hutchinson2008b). This approach has been applied to the comparison of T. b. gambiense isolates with differential virulence in experimental mice where both strain-unique spots and differentially expressed spots have been identified (Holzmuller et al. Reference Holzmuller, Grebaut, Peltier, Brizard, Perrone, Gonzatti, Bengaly, Rossignol, Aso, Vincendeau, Cuny, Boulange and Frutos2008a).

Human Infections

It is clear that variation in virulence occurs in mouse models and is related to the induction of inflammatory pathology. To determine if this is the case in clinical infections, information on infecting genotype, disease presentation and severity and host immune response is required, and to date only two studies provide this. The first of these used multilocus enzyme electrophoresis genotyping to study 49 clinical isolates of T. b. rhodesiense collected in Uganda in 1989–1993 (Smith and Bailey, Reference Stannus and Yorke1997).

These HAT cases showed a broad clinical spectrum, with both chronic and acute disease profiles reported. The more acute cases presented with high parasitaemia, acute febrile symptoms and an apparent rapid progression to late stage infection. Those with chronic infection showed low parasitaemia and symptoms of prolonged infection such as weight loss. The parasites causing these chronic and acute clinical phenotypes were genetically distinct, in general but not exclusively correlating with Zambezi and Busoga zymodemes (Stevens and Godfrey, Reference Stijlemans, Baral, Guilliams, Brys, Korf, Drennan, Van Den Abbeele, De Baetselier and Magez1992), respectively. In experimental mice, limited data indicated the slow and fast forms showed distinct disease progression, most significantly in terms of tissue and brain invasion. More recently, an association between disease progression and parasite genotype has been demonstrated in spatially separated outbreaks of T. b. rhodesiense HAT in Tororo and Soroti districts of Uganda (MacLean et al. Reference Maclean, Odiit and Sternberg2007). Analysis of clinical histories indicated a difference in disease virulence between the two foci, with Tororo subjects showing increased progression to the CNS infection stage and increased overt inflammatory pathology (splenomegaly and hepatomegaly) while patient interview data indicated no significant difference in disease duration between the two foci. Furthermore, analysis of Glasgow coma score and CSF cytokine levels indicated that the more severe presentation of the disease in Tororo was also associated with increased neurological dysfunction and CNS inflammatory response. In the Tororo patients, plasma IFN-γ concentrations were directly related to an accelerated progression to late (CNS) stage infection, after controlling for age, sex and ethnicity of the subjects. This observation is entirely consistent with data from a mouse model in which IFN-γ plays a critical role in promoting the trans-migration of trypanosomes across the blood-brain barrier (Masocha et al. Reference Masumu, Marcotty, Geerts, Vercruysse and Van Den Bossche2004).