Between a bug and a hard place: Trypanosoma cruzi genetic diversity and the clinical outcomes of Chagas disease

Over the last 30 years, concomitant with successful transnational disease control programs across Latin America, Chagas disease has expanded from a neglected, endemic parasitic infection of the rural poor to an urbanized chronic disease, and now a potentially emergent global health problem. Trypanosoma cruzi infection has a highly variable clinical course, ranging from complete absence of symptoms to severe and often fatal cardiovascular and/or gastrointestinal manifestations. To date, few correlates of clinical disease progression have been identified. Elucidating a putative role for T. cruzi strain diversity in Chagas disease pathogenesis is complicated by the scarcity of parasites in clinical specimens and the limitations of our contemporary genotyping techniques. This article systematically reviews the historical literature, given our current understanding of parasite genetic diversity, to evaluate the evidence for any association between T. cruzi genotype and chronic clinical outcome, risk of congenital transmission or reactivation and orally transmitted outbreaks.

Over the last 30 years, concomitant with successful transnational disease control programs across Latin America, Chagas disease has expanded from a neglected, endemic parasitic infection of the rural poor to an urbanized chronic disease, and now a potentially emergent global health problem. Trypanosoma cruzi infection has a highly variable clinical course, ranging from complete absence of symptoms to severe and often fatal cardiovascular and/or gastrointestinal manifestations. To date, few correlates of clinical disease progression have been identified. Elucidating a putative role for T. cruzi strain diversity in Chagas disease pathogenesis is complicated by the scarcity of parasites in clinical specimens and the limitations of our contemporary genotyping techniques. This article systematically reviews the historical literature, given our current understanding of parasite genetic diversity, to evaluate the evidence for any association between T. cruzi genotype and chronic clinical outcome, risk of congenital transmission or reactivation and orally transmitted outbreaks.
KEYWORDS: cardiomyopathy . Chagas disease . congenital transmission . diagnostics . genetic diversity . oral outbreaks . reactivation . treatment Background Chagas disease is the most important parasitic infection in Latin America, affecting an estimated 5-6 million individuals, with a further 70 million at risk [1]. The geographical range of the etiological agent, Trypanosoma cruzi (Kinetoplastida: Trypanosomatidae), extends from the southern USA to Argentinean Patagonia, where it is transmitted by more than 100 species of hematophagous triatomine bugs (Hemiptera: Reduviidae: Triatominae) [2,3] to at least eight orders of domestic, synanthropic and sylvatic mammalian hosts [4]. Human disease occurs when infected triatomine feces enter through intact mucosa or abraded skin [5]. Oral transmission is an important secondary infection route, responsible for regional microepidemics of acute Chagas disease in areas often devoid of domestic triatomine species, for example, the Amazon Basin [6]. In recent years, a significant proportion of the infected population has emigrated from rural areas, leading to the urbanization of Chagas disease in endemic countries as well as internationally [7]. Chagas disease is now considered an emergent global public health problem associated with congenital transmission [8], blood transfusions [9] and organ transplantations [10].
Following T. cruzi exposure, human infection begins with an acute phase, lasting up to 3 months, during which circulating trypomastigotes can be visualized in peripheral blood films or buffy coat smears. Most individuals are asymptomatic or present with a non-specific, self-limiting febrile illness [8]. Mortality during the acute phase is rare (<1% of cases) and may result from severe myocarditis, pericardial effusion and/or meningoencephalitis [11]. Acute mortality occurs more frequently in infants and immunocompromised patients than other infected persons. Even in the absence of treatment, the acute phase spontaneously resolves in most individuals [8,11].
Chronic infection is initially asymptomatic and the majority of patients will remain clinically indeterminate for life. However, over a period of 10-30 years, approximately 20-30% of infected individuals will develop irreversible, potentially fatal cardiac syndromes (chronic chagasic cardiomyopathy [CCM]) and/or dilatation of the GI tract (megacolon or megaesophagus) [11]. Early CCM is typically characterized by conduction system abnormalities, particularly right bundle branch block and/or left anterior fascicular block, and premature ventricular contractions [12]. More advanced manifestations include ventricular tachycardia, high-degree atrioventricular block and progressive dilated cardiomyopathy with congestive heart failure [13]. Sudden death accounts for 30-65% of CCM-related mortality and can affect patients with end-stage heart disease as well as those who were previously asymptomatic [14]. Gastrointestinal (GI) megasyndromes are rarer than cardiac sequelae, resulting from denervation, decreased motility, sphincter dysfunction and eventual luminal dilatation of the esophagus and/or colon [15]. However, the prevalence of different clinical forms, especially digestive disease, varies considerably between geographical regions [16,17].
Clinical staging of CCM Multiple expert committees have published guidelines for the standard evaluation of patients with chronic T. cruzi infection [18][19][20]. All recommend thorough history and physical examination, and at a minimum, a 12-lead ECG. Some committees also recommend echocardiograms and/or barium studies of the esophagus and colon. If all of these are normal, the patient is considered to have the indeterminate form of Chagas disease (IND). The expert committee convened by the US CDC advised that barium studies only be performed if the patient reported GI symptoms, based on the low prevalence of the digestive form of Chagas disease in Mexico and Central America, the source of most infected individuals in the USA [18]. This committee also recommended against performing echocardiograms on patients with no cardiac symptoms or ECG abnormalities, because significant abnormal findings on echocardiogram are rare in the absence of other indications of Chagas cardiomyopathy.
Several systems have been used to characterize the severity of CCM for clinical staging and epidemiological studies. The most commonly cited schemes are the modified Kuschnir classification, the Los Andes classification and the more recent system that incorporates the American College of Cardiology/ American Heart Association criteria for congestive heart failure staging [21][22][23][24]. All use similar general criteria, including specific ECG abnormalities plus chest radiography or echocardiogram to provide measures of left ventricular size and/or ejection fraction (see comparative table in [18]). However, many investigators add other modifications based on their own clinical experience, to provide finer scale classification or increase the specificity of staging [25,26]. Many ECG findings that are listed in these schemes to define early CCM, such as low voltage and right bundle branch block, occur in other cardiac diseases and are fairly common in older age groups, independent of T. cruzi infection status [27]. Others, such as moderate bradycardia and incomplete right bundle branch block, are normal variants in healthy young people [28]. While the most severe stage in any of the common classification schemes (equivalent to advanced congestive heart failure) is highly predictive of mortality [29], there is no obvious method to verify the clinical or epidemiological validity of criteria for the earlier stages. The use of heterogeneous classification schemes directly impedes comparisons of epidemiological studies of CCM severity and prevalence, and meta-analyses of the relationship between Chagas disease pathogenesis and T. cruzi strain diversity.
A brief history of T. cruzi taxonomy T. cruzi is an ancient parasite, estimated to have diverged from its most recent common ancestor 3-4 million years ago [30], and as such, displays considerable genetic diversity. Current international consensus recognizes a minimum of six stable genetic lineages: TcI-TcVI [31]. A potential seventh batrestricted genotype (TcBat), with genetic affiliations to TcI, has recently been reported in Central and northern South America [32][33][34]. Historically, the taxonomy of T. cruzi has been hindered by a lack of standardized molecular typing methods and the use of various alternative nomenclatures ( The earliest attempts to characterize T. cruzi strain variation, based on multilocus enzyme electrophoresis (MLEE), classified isolates into three major groups or 'zymodemes I, II and III' [36,37], which were later subdivided into 43 'clonets' [38]. Subsequent genotyping of additional strains using MLEE [39], random amplification of polymorphic DNA [40] and nuclear loci [41][42][43] grouped isolates into two major lineages, designated T. cruzi I and T. cruzi II [44]. More recently, supported by MLEE and multilocus sequence typing, TcII was separated into TcIIa-e [45,46], which were latterly renamed as TcII-TcVI to remove any presumptive sublineage designations [31]. Each of the six former T. cruzi genetic lineages is now considered a discrete typing unit (DTU), defined as 'a collection of strains that are genetically more closely related to each other than to any other strain and that share one or several specific characters' [47]. However, the criteria for division, number of subgroups, and their precise biological and evolutionary relevance are still a popular subject of debate [48][49][50][51][52][53][54][55][56][57].
Molecular epidemiology of T. cruzi Molecular epidemiology studies have made substantial progress defining the phylogeographical and ecological niche of each T. cruzi lineage (TABLE 2) [63]. Sylvatic DTU distribution data are still largely aggregated due to differences in ease of capture between reservoir species, paucity of supporting ecological information and our inability to genotype subpatent zoonotic infections [64].
In general, TcI, TcII, TcV and TcVI are most frequently isolated from domestic cycles and responsible for the majority of human infections. TcI has the widest distribution; it is the principal cause of Chagas  Levels of intra-patient parasite multiclonality might be expected to increase proportionally with vector exposure. However, this assumes a constant force of infection, incomplete cross-genotypic immunity, and lack of genotype interaction (e.g., genotype displacement, reciprocal inhibition, potentiation or recombination) [110][111][112][113][114], transmission population bottlenecks (as observed in related trypanosomes [115]) or any additional mechanisms that might alter the establishment of secondary infections. The complexity of natural multiclonal parasite populations is largely unknown and our ability to detect them is restricted by genetic marker resolution [107,116]. The study of this phenomenon conventionally necessitates deriving biological clones from live parasite populations (by micromanipulation [117], limiting dilution [118], plating on semi-solid media [106] or FACS [116]), prior to genetic typing, which introduces a range of potential adaptation biases, discussed below.
Genotyping of T. cruzi can be performed either directly from clinical samples (blood or tissue biopsies) or following parasite isolation by hemoculture or xenodiagnosis. Due to the scarcity of parasites in peripheral blood, especially in chronically infected patients, the former method has limited sensitivity. The primary drawback associated with parasite isolation is selection bias for particular subpopulations, initially by preferential outgrowth due to faster dividing rates and/or culture media [55,119,120] and subsequently by loss of clonal diversity from serial maintenance in axenic culture or animals [121][122][123][124][125][126]. Hemoculturing is laborious; recovery rates are usually less than 30% among chronic patients [127] and almost entirely determined by parasite load and distribution within the starting sample. Xenodiagnosis, which can facilitate greater parasite recovery, has also been shown to vary depending on vector permissibility to local strains [128][129][130]. Furthermore, due to differential strain tropisms, circulating clones isolated by hemoculture or xenodiagnosis are often genetically distinct from those sequestered in tissues [102][103][104] and can vary even between sequential blood samples [131]. Together, these observations strongly suggest that intra-host parasite diversity is routinely underestimated.    [55,140,141]). One major confounder associated with the use of any multi-copy gene is the level of intra-clone copy number and position homology to ensure comparability between strains. Genome size [142,143], karyotype [144][145][146][147][148] and chromosomal arrangements of tandem repeat regions [149,150] are known to differ widely between natural T. cruzi strains and even biological clones derived from the same population. Similar caveats affect minicircle-based genotyping, which vary in copy number and complement between major DTUs [136,151], are susceptible to contamination [152] and whose profiles are highly sensitive to minor changes in reaction conditions, raising issues of reproducibility [121,153]. With many of these methods, strain DTU assignment is dependent on absence of PCR products/restriction fragment bands, which can also result from novel variation in as yet untested strains; a typing methodology is only as 'good' as the panel of reference strains used to validate it.
Additional genotyping options are available for axenic parasite cultures, including karyotyping [154,155], DNA fingerprinting [156] and microsatellite analyses [53,70,78,89,157]. To date, no single, widely validated genetic marker affords complete, unequivocal DTU resolution [158], and reliance on only one target is inadvisable given the potential confounding influence of genetic exchange [57]. The availability of reference whole genome sequences [159][160][161][162] has reinvigorated interest in exploring the relevance of T. cruzi genetic diversity to clinical outcomes of Chagas disease. However, comparative genomics of representative T. cruzi field isolates is not yet a reality, as is the case with other more experimentally tractable trypanosomatid species [163][164][165].

Methods for the systematic literature review
To date, few correlates of chronic disease progression and clinical manifestations have been identified, although both host and parasite genetics are presumed to be involved [140,166,167]. Herein, the authors systematically review the literature, given our current understanding of T. cruzi genetic diversity, to re-evaluate the evidence for any association between parasite genotype and clinical outcome, risk of congenital transmission or reactivation, and orally transmitted outbreaks.
Independent queries of the literature were performed using the electronic databases MEDLINE/PubMed, Web of Science v5.15, EMBASE and Scopus, with no restrictions to language or calendar date. To retrieve chronic patient studies, the following search terms were used: 'chronic' or 'patients' AND 'Chagas disease' or 'cruzi' AND 'DTU' or 'DTUs' or 'lineage' or 'lineages' or 'genotype' or 'genotypes' or 'genotyping'. To retrieve congenital studies, the following search terms were used: 'congenital' or 'maternal' or 'neonate' AND 'Chagas disease' or 'cruzi' AND 'DTU' or 'DTUs' or 'lineage' or 'lineages' or 'genotype' or 'genotypes' or 'genotyping'. To retrieve reactivation studies, the following search terms were used: 'HIV' or 'transplant' or 'reactivation' AND 'Chagas disease' or 'cruzi' AND 'DTU' or 'DTUs' or 'lineage' or 'lineages' or 'genotype' or 'genotypes' or 'genotyping'. To retrieve oral transmission studies, the following search terms were used: 'oral' or 'acute' AND 'Chagas disease' or 'cruzi' AND 'DTU' or 'DTUs' or 'lineage' or 'lineages' or 'genotype' or 'genotypes' or 'genotyping'. In addition, reference lists from retrieved articles were manually checked to identify further relevant studies.
For cohorts of chronic chagasic patients, original research studies that met all of the following criteria were included: at a minimum, patients were classified as 'acute', 'indeterminate' (i.e., asymptomatic) or 'chronic' (i.e., symptomatic) following clinical examination; T. cruzi genotyping was performed on a subset of infected patients, at least to DTU level; and results were reported with reference to a recognizable T. cruzi nomenclature scheme, as detailed in TABLE 1.
For reactivation patients, original research studies that met all of the following criteria were included: patients were clinically classified as immunocompromised by confirmation of HIV co-infection or following organ transplantation; T. cruzi genotyping was performed on a subset of infected patients, at least to DTU level; and results were reported with reference to a recognizable T. cruzi nomenclature scheme (TABLE 1).
For congenital patients, original research studies that met all of the following criteria were included: neonatal T. cruzi infection was established at birth, or shortly thereafter, but prior to early childhood to be considered 'congenitally infected'; T. cruzi genotyping was performed on a subset of infected neonates and/or mothers, at least to DTU level; and results were reported with reference to a recognizable T. cruzi nomenclature scheme (TABLE 1). An additional criterion (T. cruzi genotyping performed on matched mother-infant samples) was dropped when the initial literature search indicated that it would have excluded all but three articles with a total of 23 mother-infant pairs [135,168,169].
For oral outbreaks, original research studies that met all of the following criteria were included: oral transmission was confirmed on the basis of epidemiological indicators (e.g., familial clustering), incrimination of the contaminated food source, clinical presentation (severe acute morbidity and/or mortality) or exclusion of local vector-borne transmission; T. cruzi genotyping was performed on a subset of infected patients, at least to DTU level; and results were reported with reference to a recognizable T. cruzi nomenclature scheme (TABLE 1).
Exclusion criteria included articles in languages other than English or Spanish, unpublished reports (including dissertations or conference abstracts), papers presenting solely animal data, book chapters, prospective study protocols and review articles. Patient case studies with sample size <5 were excluded from the analysis of chronic chagasic patients. Because the literature for congenital, reactivation and oral transmission is sparse by comparison, we included all publications in these categories that provided genotyping data from human-derived specimens within a recognizable T. cruzi nomenclature scheme.

Chronic Chagas disease
The earliest evidence that chronic Chagas disease manifestations may differ according to parasite strain came from reports of geographical variation, in which the rarity of megasyndromes in Venezuela compared to central and eastern Brazil was circumstantially linked to radical genetic differences between T. cruzi zymodemes ( To date, these remain some of the largest cross-sectional studies in which infecting T. cruzi genotype was directly examined in conjunction with patient clinical data (TABLE 4). Investigators noted that digestive Chagas disease was frequent in the Southern Cone, coinciding with the absence of TcI and preponderance of TcII/V, and drew a contrast with the rarity of digestive disease and predominance of TcI further north [16]. However, in these studies, very little genotypic data were obtained for digestive patients (n = 16; 5 TcI, 8 TcII and 3 TcV) compared to those with CCM (n = 159; 38 TcI, 43 TcII and 78 TcV) or without symptomatic disease (n = 217; 25 TcI, 30 TcII, 1 TcIV and 161 TcV) (TABLE 4). Furthermore, some of the aforementioned experimental design and biological limitations must be acknowledged alongside these observations. In all of these early reports, MLEE was performed using lysate prepared from parasites isolated through a combination of hemoculture, xendiagnosis and/or inoculation into animals [129,173] and in each study, positive hemocultures were obtained for less than one-third of patients ( With improved molecular techniques and the advent of direct genotyping from clinical specimens, current evidence suggests that parasite strains detected in peripheral blood from patients with or without morbidity reflect the principal lineage circulating in the local domestic cycle (TABLE 5). However, it should also be noted that due to differential strain tropisms, bloodstream parasites are not necessarily the same genotype responsible for pathology [ [199] Santander, Colombia Between a bug & a hard place Review informahealthcare.com peripheral blood. Additional smaller studies also corroborate these observations with domestic genotypes detected in both patients' blood [190,191] and cardiac, esophagus and colon tissue specimens [192,193]. One recent method, developed to circumvent some of these technical limitations, is to adopt an indirect approach, exploiting serological detection of antibodies produced in response to DTU-specific T. cruzi antigens, with the advantage of potentially revealing both historical and contemporary infecting parasite lineages. Consistent with genotypic data from the same area, serosurveys in southern endemic regions using recombinant peptides directed at trypomastigote small surface antigen indicate pervasive infection with TcII/V/VI [194,195] and even a putative association of ECG abnormalities with seroreactivity to this particular protein [196].
Lastly, on a more cautionary note, three publications from Chile report conflicting genotypic data from patients in longitudinal follow-up after a clinical trial of allopurinol and itraconazole . In the first publication, using kDNA Southern blots, more than 75% of the results in both groups were described as 'TcI + TcII + hybrid' [197], perhaps reflecting cross-reactivity between minicircle hybridization probes [200]. Subsequent papers presenting nuclear microsatellite genotyping from patients from the same clinical trial, with unspecified degrees of overlap, demonstrated a predominance of TcI (79% of 33 IND and 57% of 28 CCM in [198]; 86% of 28 IND and 71% of 24 CCM in [199]).

Chagas disease reactivation
Reactivation of Chagas disease may occur in infected individuals who become immunocompromised through immunosuppressive treatment (e.g., transplant recipients) or co-infection with HIV [11,201]. Reactivation is characterized by increased parasite multiplication, a return to microscopically detectable parasitemia levels and symptoms more typical of acute T. cruzi infection, including meningoencephalitis, acute myocarditis and skin chagomas [201]. High parasite loads lead to more frequent strain identification in immunocompromised than immunocompetent patients [188]. In addition, the loss of immunological control may allow parasites previously sequestered in deep tissues to replicate and return to the circulation. Thus, patients with reactivation arguably provide the best in vivo indicator of the complexity of natural infections in humans [102][103][104].
Compared to results from immunocompetent individuals, the detection of patent mixed infections is striking in some immunocompromised patients, with multiple genotypes in a single specimen or different genotypes in tissue compared to blood (TABLE 6). TcI, hypothesized to have cardiac muscle tropism, demonstrated contrasting distributions in tissue versus blood in immunocompetent patients (38% in cardiac tissue vs 0% in blood) [188], but was detected at equal frequencies in tissue and blood from cardiac transplant patients with reactivation (31% in tissue vs 33% in blood) (TABLE 6) [188]. This result may reflect unmasking of mixed infections that were below the level of detection prior to reactivation. Current data are insufficient to evaluate any association between strain and clinical symptoms or risk of reactivation. Analyses need to be designed to account for known risk factors for reactivation (severity of immunocompromise and specific immunosuppressive regimens) [201]. The most powerful design is a longitudinal approach, for example, in end-stage CCM patients evaluated for heart transplantation, combined with parasitological monitoring following surgery. However, pre-and post-immunosuppression comparisons are difficult, given that genotyping from blood of immunocompetent patients is limited by low peripheral parasitemia [104].

Congenital Chagas disease transmission
With improved vector control, congenital transmission has become proportionately more important among chronically infected populations, estimated to account for 22% of new T. cruzi cases in 2015 [1]. Even if vector-borne transmission were interrupted today, infected girls and women will continue to transmit the infection to their children, sustaining the cycle across generations in the absence of the vector [202].
Congenital T. cruzi infection is most often clinically silent, but can cause a spectrum of presentations, including low birth weight, prematurity and low Apgar scores to meningoencephalitis, hepatosplenomegaly, anemia, thrombocytopenia and respiratory distress syndrome [203][204][205]. Higher morbidity and mortality rates were described in the 1980s and 1990s compared to more recent cohort data [204,206]. Clinically severe congenital infection is reported to be associated with higher levels of neonatal parasitemia than less-severe or asymptomatic disease [207]. Congenital Chagas disease is also assumed to carry the same risk of chronic cardiac and/or GI manifestations as vector-borne infections.
Since the earliest descriptions of congenital T. cruzi infection [208][209][210], researchers have struggled to explain why vertical transmission is restricted to only a small proportion of infected mothers. Congenital transmission rates are highly variable both within and between endemic areas, ranging from 4.4 to 11.3% in Argentina [211][212][213][214], from 3.4 to 17.1% in Bolivia [204,205,215,216], from 0.2 to 5.2% in Brazil [217,218], from 2.5 to 11.1% in Chile [219][220][221] and from 5.6 to 10% in Paraguay [222,223]. Factors now known to be associated with higher risk of congenital transmission include younger maternal age (presumed to reflect more recent infection) [204], maternal and neonatal immunological responses [224,225], higher maternal parasitemia [168,205,214], and HIV and other immunodeficiencies [226,227]; the evidence for any influence of T. cruzi genetic diversity is more equivocal (TABLE 7). The majority of congenital genotyping studies have been performed in southern endemic areas, particularly Argentina, Bolivia and Chile, and in general, mirror the distribution of TcII/V/VI lineages observed among local chronic adult populations; additional studies are needed from regions of domestic TcI transmission in northern South and Central America [228,229].
A number of limitations prevent the accurate assessment of the interaction between T. cruzi genotype and risk of congenital transmission. Studies which examine congenital cohorts frequently perform genotyping only on maternal [219] or neonatal specimens [230], and those that incorporate both often present results from unmatched mothers and infants [200]. Even fewer studies compare parasite genotypes between mothers who transmitted to their infants and those who did not [135]. Due to the small volume of neonatal blood, most congenital genotyping is reliant on DTU-specific minicircle probes (TABLE 7) and some studies do not test for all lineages, but only those predicted to be found circulating locally (conventionally TcI, TcII, TcV and TcVI) [200,169,219]. Cross-reactivity between minicircle probes for closely related DTUs (TcII and TcVI) has been reported [168,200], which casts some doubt on studies detecting co-infections with these lineages [231]. Furthermore, most congenital genotyping studies are constrained by small sample sizes and suboptimal sensitivity of current  conventional diagnostic methods. Microscopic examination in a single specimen fails to identify over half of infected neonates [205], and subsequent loss to follow-up is high [212,232,233], thereby routinely underestimating the rates of congenital transmission and providing only a fraction of potential parasite strains for genotyping. Differential diagnostic sensitivities and rates of follow-up render it difficult to draw conclusions about geographical In Argentina, women with one congenitally infected child were significantly more likely to transmit to that child's siblings than mothers who had not previously transmitted [234,235]. Considering the extent of T. cruzi genetic diversity at the intra-DTU level, it is unlikely that all representatives of a lineage would be equally permissible to vertical transmission, but it is conceivable that particular parasite clones may be better adapted for transplacental infection [235]. Animal data also support the possibility that some T. cruzi strains are more predisposed to vertical transmission than others [236].
To date, only one study has directly examined infected human placental tissue, detecting additional minicircle signatures not observed in matched maternal blood samples and tentatively supporting the existence of T. cruzi subpopulations with placental tropism [234]. Others have described parallel discordant minicircle profiles between paired maternal-neonate blood specimens, implying either the generation of novel mutations by rapid parasite multiplication during acute neonatal infection or selective transmission of parasite subpopulations [168]. In the only study to address neonatal morbidity, no significant association was found between DTU and clinical severity of congenital infection, but nearly all genotyped specimens in this study were classified as TcV [200].
In one of the aforementioned Argentinean studies, women from areas with high triatomine infestation had the lowest risk of congenital transmission compared to those with vector control (intermediate risk) or from urban areas that had never been infested (highest risk) [235]. These findings have been confirmed in the Bolivian Chaco region, where pregnant women who had resided longer in an infested house had significantly lower parasitemia and were less likely to transmit to their child, compared to those living in areas without vector infestation [237]. Sustained vector exposure and/or repeated re-or super-infection by T. cruzi may act as an immune booster, allowing women to maintain effective control of parasitemia, thereby decreasing their risk of congenital transmission.
Oral T. cruzi transmission and outbreaks of acute Chagas disease T. cruzi trypomastigotes in triatomine feces are infectious when ingested by experimental animals [238], and consumption of infected vectors or contaminated material is considered the predominant transmission modality in non-human mammals [6]. Acute T. cruzi infections in humans attributed to oral transmission have been reported in increasing numbers in recent decades, especially in the Brazilian Amazon [6]. Three basic scenarios are described: sporadic cases in areas with sylvatic but not domestic vectors in which the attribution is one of exclusion [6]; small rural family-or village-based clusters of acute cases traced to shared food or drink [239][240][241][242]; and rare large outbreaks, sometimes in urban areas considered to be free of vectorial transmission, with an identified common source such as contaminated fruit or sugarcane juice [243][244][245][246]. Most outbreaks are small, often affecting family groups in the Amazon region, where the palm fruits açaí and baçaba are dietary staples easily contaminated by infected triatomine vectors that live in the trees themselves [241,247].
The largest reported outbreak, attributed to locally prepared guava juice, comprised 103 infections among students and staff at a school in Caracas [244]. Orally transmitted T. cruzi infection appears to be associated with more severe acute morbidity and higher mortality than vector-borne infection [239,244,248]. The most frequent symptoms are fever, dyspnea, myalgias, and generalized and facial edema; ECG changes are also common. In the Caracas outbreak, 75% of 103 infected individuals were symptomatic, 66% had ECG abnormalities, 20% were hospitalized and there was one death from acute myocarditis [244,248]. Among 13 patients infected in two outbreaks associated with contaminated sugarcane juice in northeastern Brazil, 92% had ECG abnormalities, 27% had left ventricular ejection fractions below 55% and two individuals (age 9 and 16 years) died of rapidly progressive congestive heart failure [249]. Among survivors, nearly all cardiac abnormalities resolved after treatment with benznidazole. The higher proportion of symptoms and severe morbidity and clustering of infections may facilitate easier detection of acute, oral infection compared to vectorial cases, and may contribute to the predominance of oral infections in acute Chagas disease surveillance data [6].
Because of logistical constraints, few outbreaks [241,244] receive thorough epidemiological investigations, and direct incrimination of the contaminated item is infrequent. At the DTU level, T. cruzi genotypes identified in oral outbreaks principally reflect the predominant lineages circulating in that geographical area (TABLE 8). In the Caracas school outbreak, molecular typing demonstrated identical TcI strains in 3 of the 103 infected individuals and similar strains in a single P. geniculatus captured at the site where the implicated guava juice was prepared [244,250]. Human specimens from the outbreak in Santa Catarina in 2005 attributed to sugarcane juice were typed as TcII [40]. Small family outbreaks in the Amazon region generally yield TcIV or TcI [95,93,97,99], while rural outbreaks outside of the Brazilian Amazon (Colombia, Venezuela, French Guiana) detect TcI (sometimes identified specifically as sylvatic TcI) [240,[251][252][253].
Most investigations occur months after the outbreaks and consist of vector and animal reservoir studies in the area of the outbreak [243,95,98,245,254], based on the assumption that outbreak vehicles were contaminated by infected triatomine feces or anal gland secretions of infected opossums, which contain infective trypomastigotes [255]. The investigation in the area of the Santa Catarina outbreak (typed in humans as TcII) demonstrated TcI in opossums and both TcI and TcII in triatomine vectors, implicating local triatomines as the most likely source of infection [243].
Recent laboratory data suggest that parasite contact with host gastric acid may render trypomastigotes more invasive through changes in parasite surface glycoproteins, and that this interaction may underlie the increased clinical severity observed in orally acquired Chagas disease [256][257][258]. The T. cruzi surface glycoprotein gp82 is highly resistant to proteolysis and has been shown to mediate migration through the stomach mucin layer and invasion of gastric mucosal cells. Different parasite strains express distinct isoforms of gp90, a second surface glycoprotein, which present differential susceptibility to digestion by pepsin. Strains with pepsin-digested gp90 have highly efficient gastric mucosal invasion and establish patent T. cruzi infections, in contrast to those with pepsin-resistant gp90 which invade poorly. Experimental infection with a Colombian TcI isolate demonstrated efficient oral and intraperitoneal transmission, in contrast to a Peruvian strain (TcV) which was much less efficient by the oral than the intraperitoneal route [259]. Similarly, a strain from the 2005 Santa Catarina outbreak (TcII) was highly invasive, producing a robust infection and high mortality in a mouse model, unlike CL Brener (TcVI) and G (TcI) strains, with these differences attributed to the pepsin susceptibility of the gp90 isoforms of the strains [256].

T. cruzi lineage in Chagas disease diagnostics and chemotherapy
Diagnosis of Chagas disease in the chronic phase is based on serological detection of anti-T. cruzi IgG antibodies. However, no single assay has sufficient sensitivity and specificity to be used alone; confirmed diagnosis relies on concordant results from at least two tests using different antigens and/or formats (usually ELISA, indirect immunofluorescence and/or indirect hemagglutination) [260]. Differential sensitivities to serodiagnostic tests have been reported between Bolivia and Peru, where two different commercial rapid tests based on recombinant antigens demonstrated sensitivities of 87.5 and 90% versus 30 and 54%, respectively [261]. Rapid test sensitivities were closely correlated with absorbance values on whole parasite lysate-based ELISAs. Similarly, low sensitivities of recombinant antigen ELISA and rapid tests have also been reported from Panama [262,263] and Mexico [264]. No clear correlation between serodiagnostic test reactivity and local T. cruzi DTU has been observed; instead, these discrepancies may reflect weaker adaptive immune responses to parasite antigens between endemic populations [265]. Current treatment options for Chagas disease are limited to benznidazole and nifurtimox. While both drugs have high cure rates in the acute phase, efficacy during the chronic phase has been much harder to document, largely due to the lack of a timely, sensitive test of cure [18,266]. Two recent trials have successfully employed quantitative real-time PCR to rapidly detect treatment failure of the new drug candidates posaconazole and E1224, a related drug [268][269][270][271]. In contrast, fewer than 10% of those who completed the 60-day benznidazole course had positive results by PCR during the follow-up period. The posaconazole trial required positive pretreatment results by PCR as a prerequisite for enrollment, while the E1224 trial was preceded by an optimization exercise that achieved 92% sensitivity for PCR using multiple specimens and optimized techniques. These two trials provide strong support for the use of PCR as a primary outcome measure in clinical trials in the chronic phase. No data on T. cruzi strain are currently available from these trials, but parasite diversity was unlikely to be high because the E1224 trial was conducted in Bolivia, and nearly all patients treated in the posaconazole trial in Spain were also of Bolivian origins. Small human studies, clinical impression and findings in animal models tentatively suggest that parasite susceptibility varies with geographical location and parasite lineage [272][273][274][275] as might be expected given the crucial role played by some genetic loci (e.g., TcNTR) in resistance to both benznidazole [276] and nifurtimox [277]. However, direct human data are sparse and in vitro epimastigote assays have poor correlation with in vivo mouse models of drug response [278]. Natural drug resistance to either drug has been reported in both patient-derived isolates as well as sylvatic unexposed strains [273,279]. The most comprehensive in vivo analysis of parasite susceptibility was conducted in the 1980s using 47 parasite strains isolated from patients, sylvatic reservoirs and vectors, and subsequently inoculated into mice [273]. In mouse models, cure rates to both benznidazole and nifurtimox were close to 100% for parasite strains from Argentina and the southernmost region of Brazil, where the predominant DTUs are presumed to be TcV and TcVI. In central and Atlantic Brazil, where TcII is likely to be more common, cure rates ranged from 50 to 65%, while strains from sylvatic reservoir hosts and vectors showed highly variable responses [273].
The inability to dissect the relative contribution of parasite genetics to treatment failure, disease pathology and progression represents a major hurdle to the assessment of novel drug candidates. The advent and optimization of a non-invasive, in vivo, bioluminescent imaging system which can facilitate real-time monitoring of chronic parasite burden has the potential to address these key questions using current and prospective chemotherapies, prior to human evaluations [280,281].

Expert commentary & five-year view
Establishing or excluding any relationship between T. cruzi genetic diversity and clinical outcome will require significant improvements in study design and reporting, patient sampling and parasite genotyping (BOX 1). As encountered throughout this article, meta-analyses of historical studies are impeded by the paucity (and in many cases, complete absence) of descriptive clinical data among genotyping publications. Use of a standardized CCM stratification method (or supplemental data to allow reclassification) would facilitate comparison between studies. Because of the long asymptomatic period before the onset of clinical signs and symptoms, crosssectional evaluations will inevitably misclassify some individuals as IND who will later develop CCM; if parasite genetics contribute to the later progression of disease, this misclassification will introduce a bias toward the null hypothesis ('parasite genetics are not associated with CCM'). This bias will be exacerbated if CCM and IND groups are not matched by age distribution; without deliberate age-matching (individual or group frequency-matching), the IND group will always be younger than the symptomatic group due to the natural history of the disease [25]. Only one genotyping study ensured that the age distributions of IND and CCM groups were comparable [189]; no other publications in this review took this potential confounder into account. Cohort studies of Chagas  valuable opportunities to examine whether T. cruzi strain affects the risk of development and progression, while avoiding the biases of cross-sectional studies. The strict clinical criteria used in the two most prominent current cohorts (the control arm of the BENEFIT trial [282] and the REDS-II study [27]) should provide a firm basis for interpretation, although thus far, neither has published T. cruzi genotypic data.
The pathogenesis of Chagas cardiomyopathy is widely believed to be multifactorial, with contributions from both parasite and host genetics; evaluating parasite genotype in isolation is therefore likely inadequate. The addition of immunology, human genetics and more detailed epidemiological evaluations could provide a richer assessment of the possible interactions of factors modifying clinical outcome. Similarly, studies of congenital Chagas disease could be made more rigorous by ensuring that both mother and infant specimens are tested, and that the source population, criteria for and age at congenital diagnosis are more clearly detailed in publications. The maternal immune response clearly plays an essential role in modulating congenital transmission risk, and optimally would be evaluated in tandem with parasite strain diversity [224,225]. Finally, the impact of vector control initiatives, including changes in force of infection and circulating strain compositions, needs to be considered, particularly when comparing genotypic data from the same endemic area over time.
Epidemiological as well as genotype data are extremely sparse for GI Chagas disease. The widely accepted observation that gastrointestinal Chagas disease is much rarer in Central and northern South America than in the Southern Cone still rests on clinical reports rather than quantitative population-based epidemiological studies. In part, this stems from the impracticality of performing barium studies in field conditions and on large numbers of people. A study in Bolivia used clinical swallowing time (by placing a stethoscope on the neck, asking the subject to swallow and timing the duration) to assess prevalence of early esophageal changes; 22% of seropositive participants had swallowing times longer than the clinically established cutoff of 10 s, compared to none of the seronegative participants [283]. If field-friendly methods such as swallowing times can be validated by comparison with barium studies, this could provide a useful tool for epidemiological assessments in conjunction with parasite genetic analysis.
Many articles are difficult to interpret because the relationship between data by specimen and by patient is not explicit. Synthesis of the literature as a whole is challenging because multiple manuscripts by the same authors often present genotyping data from samples from the same source population, without specifying the degree of overlap [197,198,284]. Recharacterization of the same specimens in consecutive papers may inflate the prevalence estimates of the resulting DTUs. For all of these limitations, more transparent data presentation would be invaluable. The current 'open-access' movement [193] may assist in this area, and could be enhanced by the use of common identifiers between datasets from the same source population to facilitate such meta-analyses.
Modifications to study design must be complemented by parallel improvements in clinical T. cruzi genotyping. As previously discussed, parasite isolation by hemoculture, xenodiagnoses or animal inoculation is not ideal and introduces unquantifiable biases; novel methodologies to enrich T. cruzi DNA in clinical specimens and circumvent loss of clonal diversity during parasite culturing stages warrant further investigation. Crucially, the current repertoire of clinical genotyping techniques is restricted to DTU-level classification and insufficient to explore the potential interaction between parasite multiplicity of infection and clinical outcome. Illumina amplicon sequencing, recently developed to explore intra-host pathogen genetic diversity, involves the generation of millions of 'short' sequencing reads from individual samples, potentially allowing correlation of read depth with genotype abundance [285][286][287][288]. This strategy has been used to examine intra-patient multiclonality among chronic Chagas disease patients, across the clinical spectrum (asymptomatic to severe cardiomyopathy, megaesophagus or megacolon) in Goiás, Brazil and matched motherinfant pairs from Cochabamba, Bolivia. While no relationship between parasite multiclonality and patient sex, age or clinical symptoms has been observed thus far, putative evidence of diversifying selection affecting antigenic genes was detected, suggesting a link between genetic diversity in this gene family and survival in the mammalian host [289].
As yet unidentified, diverse genetic characteristics of T. cruzi may influence clinical outcome. However our systematic review demonstrates no unequivocal evidence for an association between T. cruzi genotype and chronic morbidity, risk of reactivation, or congenital or oral transmission. In recent publications, the most consistent finding is that specimens from all groups reflect the predominant genotypes circulating in the local area. Results from patients with reactivation indicate that mixed infections may be the rule, rather than the exception. Further elucidation of the role of T. cruzi genotype in disease pathogenesis will require improvements in both study design and genotyping techniques.

Acknowledgment
Much of the research covered in this review was funded by the Wellcome Trust and the European Commission Framework Programme Project "Comparative epidemiology of genetic lineages of Trypanosoma cruzi" ChagasEpiNet (contract #223034). CB received partial salary support from National Institutes of Health R01 AI107028-01A1. LAM was supported by a BBSRC Doctoral Training Grant.

Financial & competing interests disclosure
The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Key issues
. Trypanosoma cruzi displays remarkable genetic diversity, which is believed to contribute to the biological, epidemiological and clinical variation observed among Chagas disease foci.
. T. cruzi strains can be classified into six stable genetic lineages known as discrete typing units (TcI-TcVI), with distributions loosely defined by geography, ecology and transmission cycle.
. Establishing an association between T. cruzi genotype and clinical outcome is primarily complicated by low peripheral parasitemia in chronic infections.
. Current evidence suggests that parasite strains detected in patients with or without morbidity reflect the principal lineage circulating among domestic transmission cycles in that geographical area.
. Likewise, local parasite genotypes are frequently implicated in acute T. cruzi oral outbreaks, which have been increasingly reported in recent years.
. Loss of immunological control through immunosuppression or HIV co-infection allows sequestered parasites to return to the circulation, thus providing the best indicator of the complexity of natural infections.
. Women who transmit T. cruzi to one child are more likely to transmit to other offspring, suggesting that parasite genotype may be one factor influencing the risk of vertical infection. However, to date, the discrete typing units isolated from mothers and infants reflect the principal lineages circulating in that area.
. Improvements in both study design and genotyping techniques are required to advance our understanding of the role of T. cruzi genotype in disease pathogenesis.