¹Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2XY, UK

²Department of Molecular Medicine, Karolinska Hospital, CMM L8:03, S171 76 Stockholm, Sweden

³School of Biology, University of Leeds, Leeds LS2 9JT, UK

⁴Human Genetics, Duthie Building (808), Southampton General Hospital, Tremona Road, Southampton SO16 6YD, UK

⁵MRC HGMP Resource Centre, Hinxton, Cambridge CB10 1SA, UK

⁶Research & Development Genetics Department, Zeneca Pharmaceuticals, Alderly Park, Macclesfield, Cheshire SK10 4TG, UK

⁷Instituto Evandro Chagas, Caixa Postal 3, 66.001 Belem, Brazil

⁸Departamento de Patologica, Universidade de Federal do Maranho, São Luis, Maranhão, Brazil

⁹Departamento de Parasitologia, Universidade de Piaui, Terasina, Piaui, Brazil

¹⁰London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK

Correspondence to: Professor J M Blackwell, CIMR, Wellcome Trust/MRC Building, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2XY, UK. E-mail: jennie.blackwell@cimr.cam.ac.uk

This work was supported by grants from The Wellcome Trust, Heather Cordell (CIMR) provided valuable advice on power calculations for linkage. We would also like to thank the people of northeastern Brazil for their hospitality and for their contribution to this study.

Familial aggregation, high relative risk to siblings, and segregation analysis, suggest genetic control of visceral leishmaniasis in Brazil. Class II gene effects in mice, and high circulating tumour necrosis factor in humans, provide reasons to target HLA. Fifteen polymorphic markers across 1.03 Mb (DQB1 to TNFa) were genotyped (87 multicase families; 638 individuals). Model-based parametric analyses using single-point combined segregation and linkage in COMDS, or multi-point linkage in ALLEGRO, failed to detect linkage. Model-free nonparametric affected sibling pair (SPLINK) or NPL_all score (ALLEGRO) analyses also failed to detect linkage. Information content mapping confirmed sufficient marker information to detect linkage. Analysis of simulated data sets demonstrated that these families had 100% power to detect NPL_all scores of 5 to 6 (>LOD4; P < 0.00001) over the range (7% to 61%) of age-related penetrances for a disease susceptibility gene. The extended transmission disequilibrium test (TDT) showed no consistent allelic associations between disease and the 15 loci. TDT also failed to detect significant associations between extended haplotypes and disease, consistent with failure to detect significant linkage disequilibrium across the region. Linkage disequilibrium between adjacent groups of markers (HLADQ/DR; 82-1/82-3/-238bpTNFA; LTA/62/TNFa) was not accompanied by significant global haplotype TDT associations with disease. The data suggest that class II/III regions of HLA do not contain major disease gene(s) for visceral leishmaniasis in Brazil.

Genes and Immunity (2002) 3, 350-358. doi:10.1038/sj.gene.6363852

visceral leishmaniasis; HLA; multicase families; genetic susceptibility

Introduction

Leishmaniasis occurs in most areas of the tropics and subtropics, including large areas of Africa, India and South America. Visceral leishmaniasis, caused by Leishmania donovani, L. chagasi and L. infantum, accounts for 500 000 of the 2 million new cases of leishmaniasis that occur annually (http://www.who.int/emc/diseases/leish/leisburtre.html). Although fatal in susceptible individuals, there is evidence to suggest that most people are resistant to clinical disease. Evidence from specific skin-test reactivity and lymphocyte proliferation assays^1,2,3 indicates that only a small subset of people infected with leishmanial parasites develop clinical disease. Familial aggregation is a feature of visceral leishmaniasis caused by L. chagasi in northeastern Brazil,⁴ providing a high relative risk (_2S = 34) of disease in further siblings of affected sibling pairs.⁵ Segregation analysis supports single dominant or additive gene control, with a frequency of ~0.002 for the disease allele and penetrances ranging from 7% (<1 year olds) to 61% (9-10 year olds) according to age liability classes. Although this does not discount minor contributions from multiple loci, a hypothesis worth testing is that susceptibility to visceral leishmaniasis in this region of Brazil is controlled by one (or a small number of) major gene(s).

In mice, genes encoding the IA (DQ in man) and IE (DR in man) class II molecules in the major histocompatibility complex (MHC: H-2 in mice, HLA in man) exert a profound influence on susceptibility to L. donovani infection.⁶ Genetic mapping data^6,7 are supported by functional data showing that treatment in vivo with anti-IE monoclonal antibodies abrogates the non-cure response of IE-bearing H-2^d mice.⁸ Treatment of H-2^b mice, which do not express IE, with monoclonal antibodies to IA reduces their ability to clear liver and spleen parasite burdens and self-cure. Transgenic introduction of the gene encoding IE chain to restore IE expression in non-IE expressing mice also abrogates the self-curing response.⁹ Hence, there is strong evidence that polymorphism at class II molecules is itself functionally responsible for MHC-regulated control of infection.

In man, high levels of circulating tumour necrosis factor (TNF) associated with clinical visceral leishmaniasis¹⁰ suggest that regulatory polymorphisms for genes encoded in other regions of the HLA complex might also be important in determining genetic susceptibility. This would be consistent with a range of other infectious diseases,^{11,12,13,14,15,16,17} including mucocutaneous leishmaniasis,¹⁸ where allelic associations between disease and polymorphisms in the gene (TNFA) encoding TNF or the closely linked gene (TNFB) encoding lymphotoxin A (LTA or TNF) have been observed. This may be functionally related to regulation of TNF transcription by single nucleotide polymorphisms (SNPs) in the promoter region of TNFA.^19,20,21,22 Hence, there are multiple reasons to target the HLA complex as a candidate disease susceptibility locus for visceral leishmaniasis.

To date, only a few studies have examined polymorphisms across the HLA complex in relation to susceptibility to visceral leishmaniasis. These have included population-based case-control studies,^23,24,25 as well as family-based genetic analysis.²⁶ Linkage or allelic associations have been negative^23,26 or weak,^24,25 but sample sizes have been small. Here we examine 15 polymorphic loci spanning 1.03 Mb from HLADQB1 to the TNFa microsatellite in 87 multicase families of visceral leishmaniasis from northeastern Brazil.^5,27 We use parametric and nonparametric linkage analyses, information content mapping, power calculations, single marker and haplotype transmission disequilibrium testing, and analysis of linkage disequilibrium, to evaluate the role of HLA genes in determining disease susceptibility.

Results

Linkage analyses

Allele frequencies for markers, and cumulative physical distances between markers, are provided in Table 1. A breakdown of the family structures is provided in Table 2. Combined segregation and linkage analysis performed within COMDS under an additive model provided no evidence for positive linkage between a putative disease susceptibility locus (DSL) controlling visceral leishmaniasis and HLA markers (Table 3). Multipoint parametric linkage analysis using this model and performed within ALLEGRO (Table 3) met criteria (LOD scores -2) for excluding loci in both class II and class III regions of HLA in controlling susceptibility to visceral leishmaniasis. However, since HLA does not appear to be the major gene regulating visceral leishmaniasis in this population, the model of inheritance predicted by segregation analysis may be inappropriate. Hence, nonparametric model-free linkage analyses were performed. Singlepoint (data not shown) and multipoint nonparametric analysis in ALLEGRO (Figure 1a) again failed to provide evidence (all NPL_all scores <1; lowest P value 0.23) for HLA genes in controlling visceral leishmaniasis, despite the fact that the information content across the region (Figure 1b) was high (0.59 to 0.85). Affected sib-pair analyses performed within SPLINK also failed to demonstrate significant linkage (Table 4). The highest LOD score observed in SPLINK was 0.701 (P = 0.055) for HLADQA1 with 0:1:2 identity-by-descent (IBD) allele sharing of 20:41:39. This equates to a locus-specific risk for siblings of patients (_s; calculated from the ratio of the expected proportion of affected sibling pairs sharing zero alleles identical by descent (0.25) and the observed proportion²⁸) of 1.25.

Power to detect linkage

In order to be sure that our families had sufficient power to detect linkage, simulations were performed within ALLEGRO. One hundred replicates were generated using parameters (single gene additive model; allele frequency for disease gene q = 0.0016; extremes of age-related penetrances of 7% and 61%) defined initially by segregation analysis performed for these families,⁴ and actual HLA marker information for all individuals in the pedigrees. The simulated data sets were then used to carry out nonparametric linkage analysis within ALLEGRO. NPL_all scores for over the range of critical values (1.65, 2.33, 3.09, 3.27, 4.27) equivalent to P values of 0.05, 0.01, 0.001, 0.0001 and 0.00001 are shown in Table 5. These critical values are equivalent to allele-sharing LOD scores of 0.59, 1.17, 2.07, 3.00 and 3.95. This analysis demonstrates that these families have 100% power to detect NPL_all scores of 5 to 6 (=LOD>4; P < 0.00001) over the range of age-related penetrances for a putative DSL in this population.

Family-based allelic association testing

Although linkage of visceral leishmaniasis susceptibility to HLA markers was not observed, it was possible that family-based allelic association testing performed would be more powerful in detecting weaker associations. Using the extended transmission disequilibrium test (ETDT) and transmission to all affected offspring (Table 6), significant genotype-wise association was observed for microsatellite marker 82-1 (²₂₉, P = 0.024), and for the adjacent (30 kb distal) marker 82-3 (²₄₂ = 61.4, P = 0.027). The allele-wise ETDT was not significant for these markers. The significant ² for goodness of fit comparing allele-wise and genotype-wise tests (Table 6) indicates poor fit to the data that most likely results from multiple rare alleles at these markers. A single test statistic showing significant biases in transmission of individual alleles (82-1₉₄ +ve P = 0.03) did not survive correction for multiple testing. No other markers showed significant allele-wise or genotype-wise ETDT statistics. Again, tests for significant biases (+ve indicates higher transmission of this allele to affected offspring than expected) in transmission of individual alleles (DQB1_*0603 +ve P = 0.01; DQA1_*0103; DR₁₅ +ve P = 0.04; LH1₉₉ +ve P = 0.03; 62₁₆₈ +ve P = 0.04) did not survive correction for multiple testing. Another reason for a significant goodness of fit test may be the use of ETDT to test multiple transmissions within a family. The more conservative use of haplotype relative risk (HRR) that examines transmission to a single affected individual in each family failed to provide evidence for allelic associations for any of the markers across HLA (data not shown). There was no evidence for significant linkage disequilibrium that extended across the class II and class III regions in this population (data not shown). No significant global ETDT statistics were obtained using TDT Phase for any haplotype combinations across or within the class II and class III regions. Individual haplotype associations were observed for DQA1-DRB1 (3-13; passed 17, not passed 7; ²₁ = 4.2, P = 0.041) and DQB1-DQA1-DRB1 (11-3-13; passed 14, not passed 4; ²₁ = 5.6, P = 0.018) when transmissions to all affected sibs in a family were analysed. This is consistent with data showing strong linkage disequilibrium between DQB1 and DQA1 (D' = 0.849; P = 2.6e-132), DQB1 and DR (D' = 0.818; P = 3.2e-138) and DQA1 and DR (D' = 0.801; P = 9.4e-94). However, these individual haplotype associations were not significant using the more conservative association test examining transmissions to a single affected sib per family, and did not survive correction for multiple testing. No significant individual haplotype associations were observed involving haplotypes with 82-1/82-3/-238bpTNFA or LTA/62/TNFa (data not shown), even though some evidence (D' > 0.7; P < 1.6e-6) for linkage disequilibrium was observed for markers within these two clusters.

Discussion

In a previous study of visceral leishmaniasis in northeastern Brazil, Amendoeira and colleagues²⁶ typed polymorphisms at HLA A, B, C, DR, DQ, Bf, C2, C4a or C4b in 10 families with multiple cases of visceral leishmaniasis caused by L. chagasi and found no evidence for linkage. This result might be put down to small sample size. However, we have also failed to find evidence for linkage following extensive analysis using 15 polymorphic markers spanning a 1.03 Mb region of HLA from DQB1 to the TNFa microsatellite in 638 individuals from 87 two and three-generation pedigrees with multiple cases of visceral leishmaniasis. We approached this first as a test of the hypothesis that the HLA region might contain the major single gene predicted by segregation analysis carried out using extended information on disease status from these families.⁵ Hence, we applied parametric linkage analyses, performed as combined segregation and linkage analyses within COMDS, and as multipoint analysis in ALLEGRO to maximise marker information across the region. All of these tests failed to demonstrate linkage. This is in striking contrast to similar analysis¹⁷ of 73 multicase families (376 individuals) of leprosy from the same region of Brazil. In this case, combined segregation and linkage analysis performed in COMDS provided highly significant linkage to DQB1 (LOD = 4.978; P = 0.00000171), DQA1 (4.870; 0.00000221), DRB1 (5.783; 0.00000025), TNF-308 (4.000; 0.00001791), and LTA (1.935; 0.00285171). This was supported by nonparametric multipoint analysis in GENEHUNTER with P values ranging from P = 0.0009 to P = 0.003, and locus-specific _s values of 1.66 for DQA1 and 1.79 for DQB1 (M-A Shaw and JM Blackwell, unpublished analysis) compared to the 1.25 obtained for DQA1 in the present study. The information content mapping carried out within ALLEGRO shows that our result with visceral leishmaniasis is unlikely to be due to insufficient marker information to detect linkage. Our conclusion is that HLA does not contain the major gene controlling visceral leishmaniasis as predicted by segregation analysis.

To determine whether HLA contained a disease susceptibility locus with lesser effect, or not detectable under the model of inheritance predicted by COMDS, we extended our analyses to nonparametric linkage analyses and transmission disequilibrium testing. Because of the small effective sample sizes (Table 4), our set of families may not have been ideal for the affected sibling pair analysis performed using SPLINK. Others have shown²⁹ that in excess of 4000 affected sibs may be required to demonstrate linkage if the disease allele is at low (q = 0.01) frequency. In an attempt to maximise information over our larger two or three-generation pedigrees, we performed multipoint analysis of all affected relative pair combinations in ALLEGRO. This should have improved our power to detect linkage. Again, we failed to find evidence for linkage of a putative DSL to HLA, even though analysis of simulated data sets within ALLEGRO demonstrated that these families had 100% power to detect NPL_all scores of 5 to 6 (=LOD4; P < 0.00001) over the range (7% to 61%) of age-related penetrances for disease. Nor did transmission disequilibrium testing provide any powerful evidence of strong allelic associations between alleles at markers across the HLA complex and susceptibility to visceral leishmaniasis. In this case, our maximum number of transmissions tested (244, Table 6) may also have fallen short of that required²⁹ to detect associations at low frequencies (q = 0.01) of the putative disease gene. Hence, an extended sample from northeastern Brazil may find evidence for allelic associations between visceral leishmaniasis and HLA. All of our primary data are available on request for meta-analysis by others. It is also possible that HLA associations will be observed in other geographical regions. For example, Faghiri and coworkers²⁴ found that the relative risk of disease was 13.27- higher in individuals carrying HLA A26 when 52 cases (allele frequency 15.38%) of antibody diagnosed visceral leishmaniasis caused by L. infantum in Iran were compared with 226 matched controls (allele frequency 1.35%) (Fisher's exact test P = 0.0001; P = 0.004 after correction according to the number of antigens studied). No other significant HLA A or Cw allele P values survived correction for multiple testing. No significant associations were observed for HLA B alleles. Overall, the evidence for HLA associations in Iran was weak. In India, no significant bias in transmission of HLA A, B, or DR alleles from heterozygous parents to 51 unrelated parasitologically proven L. donovani kala-azar patients was observed.²³ More recently Meddeb-Garnaoui and coworkers²⁵ examined HLA-DRB1, -DQB1, TNFalpha, TNFbeta, HSP70-2 and HSP70-hom genetic polymorphisms in 156 unrelated patients with mediterranean visceral leishmaniasis caused by L. infantum compared with 154 unrelated leishmanin skin test positive healthy controls. They found reduced allele (DR2) and genotype (DR2/DR13; DQB1*0201/-) frequencies in patients compared to controls but none of these remained significant after application of a Bonferroni correction factor. A higher frequency of homozygotes for the HSP70-2/PstI negative allele in cases compared to controls also failed to achieve significance after Bonferroni correction. No associations were found for the -308 base pair TNFA gene polymorphism or the NcoI polymorphism in the first intron of the TNFB/LTA gene. Like us, these authors²⁵ conclude that their results do not support association between susceptibility to visceral leishmaniasis and the MHC class II and class III loci.

In conclusion, evidence available to date does not support a major role for polymorphism in genes of the HLA complex in determining susceptibility to human clinical visceral leishamaniasis. This is of interest considering the previous genetic and functional studies in mice.^6,8,9 Also in view of the major role which class I and II genes of the MHC have in presenting antigen to protective or disease exacerbatory CD4^30,31 or CD8³² T cells, and the high levels of circulating TNF associated with clinical disease in man.¹⁰ There are, of course, many other genes/ mechanisms that regulate these important immunological pathways. The negative result obtained here simply fuels the search for the non-MHC genes and mechanisms influencing the severe clinical manifestations of human visceral leishmaniasis.

Methods

Ascertainment of families

The study was based in three sites (Santerem, Marajo and Igaripe Miri) in the State of Para, one site (São Luis Island) in the State of Maranhão, and one site (Terasina) in the State of Piaui in northeastern Brazil. Ethical approval for the study was obtained from the Ethics Committee of the Instituto Evandro Chagas, Belem, Para, Brazil. Epidemiological and demographic details relating to the study sites are described in detail elsewhere.⁵ Multicase families were ascertained from the medical records of the Fundacão National de Saude in the States of Para, Maranhão and Piaui. Families were collected on the basis of data from the 1983-85 and the 1993-94 epidemics. Families were pursued when there was indication from their medical records that additional family members had been, or were currently, affected with visceral leishmaniasis. All of the specific sites studied were areas of low incidence of cutaneous leishmaniasis, although the northeastern region of Brazil as a whole contains areas of high incidence of cutaneous disease.³³ The separation of cutaneous and visceral disease areas relates to microgeographical/ecological differences.^34,35 The populations studied in northeastern Brazil represent admixtures of Caucasian, Negroid and Native Indian ethnic backgrounds that have interbred extensively for over 150 years. Family-based analysis was therefore considered preferable to a population-based genetic study. The total data for segregation analysis⁵ consisted of 87 two or three-generation pedigrees with 824 individuals yielding 138 nuclear families (117 with at least one affected individual). Families were collected from the five sites. For genetic studies, 15 families (99 individuals) were from Santerem, Marajo and Igaripe Miri in the State of Para; 23 families (175 individuals) from Terasina in the State of Piaui; and 49 (364 individuals) from the island of São Luis in the State of Maranhão. All families were of equivalent socio-economic status. Blood was collected by venepuncture from all available members of the families and peripheral blood mononuclear cells cryopreserved for later preparation of Epstein-Barr virus (EBV)-transformed B cells. Informed consent for venepuncture was obtained from adults, and from the parents of children <18 years old. EBV cells were expanded and DNA prepared for genetic analysis. A total of 87 multicase families (117 nuclear families; 638 individuals) were available for analysis after checking for genetic integrity within all families.

Diagnosis

Diagnosis was made on the basis of clinical, parasitological and serological criteria as described.⁵ All of the individuals classified as affected in these families were diagnosed with clinical visceral leishmaniasis requiring treatment with pentavalent antimony. Data on sub-clinical disease or asymptomatic infections were not available. At initial presentation, symptoms suggestive of visceral leishmaniasis included fever, often prolonged and not cyclical (differential diagnosis for malaria), pale countenance of skin due to anaemia, weight loss and hepatosplenomegaly (predominantly splenomegaly in this region of Brazil). A hard and palpable spleen was a significant clinical indicator. These examinations were carried out by experienced local clinicians. Bone marrow or splenic aspirates were taken from all suspected cases. Diagnosis of 98-100% of patients was supported by immunofluorescent antibody tests to detect leishmania-specific antibodies using antigen prepared from L. chagasi promastigotes, and 84-100% by direct observation of Giemsa stained parasites in bone marrow or splenic smears, depending on location.⁵ Cultured parasites from a subset (~10% over the three sites) of visceral leishmaniasis patients were confirmed as L. chagasi based on monoclonal antibody or isoenzyme identification performed at the Instituto Evandro Chagas in Belem. L. amazonenesis and L. mexicana were never observed.

Genetic typing

Fifteen polymorphic loci were typed across a 1.03 Mb region of HLA from DQB1 to the TNFa microsatellite distal to the genes encoding TNF (TNFA) and LTA. The HLA class II DQA1, DQB1 and DRB1 loci were typed by PCR amplification and sequence-specific oligonucleotide (PCR-SSO) hybridization as described.^36,37 HLA-DRB1 probe sequences were taken from the 1991 HLA Workshop.³⁸ HLA-DQB1 and DQA1 sequences were from the same workshop and from other published sources.^39,40 SSO probes were labelled with P₃₂ and hybridization observed by explosure to X-ray film (Kodak). Typings were read in duplicate for each individual DNA. Nine novel microsatellite markers (LH1, D3A, 9N1, 82-2, T2, 82-1, 82-3, and 62; Hsieh et al⁴¹) and the TNFa microsatellite⁴² were typed by PCR amplification using marker-specific FAM-, TAMRA-, TET- or HEX-labelled forward and unlabelled reverse primers. PCR products were coloaded with ROX -350 fluorescent labelled size standard (Applied Biosystems, Foster City, CA, USA) onto 6% polyacrylamide gels and run on an ABI 373 or ABI377 sequencer analyser (Applied Biosystems). Gels were analysed using the computer software programmes genescan and genotyper (Perkin-Elmer, Foster City, CA, USA). The -238 bp⁴³ SNP for TNFA was typed by PCR-restriction fragment length polymorphism PCR-RFLP) by designing a primer that would incorporate a BsawI restriction site into one of the variants. The -308 bp⁴⁴ SNP for TNFA, and the LTA NcoI biallelic polymorphism⁴⁵ were typed as described. The gene order across the HLA class II and class III regions is DQB1, DQA1, DRB1, LH1, D3A, 9N1, 82-2, T2, 82-1, 82-3, -238 bp TNFA, -308 bp TNFA, LTANcoI, 62 and TNFa.⁴¹

Parametric linkage analysis

Combined segregation and linkage analysis was performed within the computer program COMDS.⁴⁶ This is achieved by estimating segregation model parameters with the recombination fraction (TH) fixed at 0.5, and against the model with segregation parameters and recombination fraction estimated. The model employed was generated using COMDS to carry out segregation analysis of the complete pedigree sets for visceral leishmaniasis.⁵ The difference between -2 log likelihood (-2lnL) in the two models is a ² with 1 degree of freedom. This can be expressed as a lod (Z) through division by 2 log(10). COMDS is only capable of using nine allele frequencies so for microsatellite markers with >9 alleles, the alleles of lowest frequency were pooled and used as a single 'allele'. Multipoint parametric linkage analysis was performed within the computer program ALLEGRO,⁴⁷ using the COMDS⁵ best-fit model (additive single gene; gene frequency q = 0.0016 for the putative disease susceptibility allele; penetrances 7% and 61%, the extremes predicted for different age liability classes).

Nonparametric linkage analyses

Single- and multi-point nonparametric linkage analyses were performed within ALLEGRO.⁴⁷ This analysis compares IBD allele sharing for all affected relatives in a pedigree to generate an NPL_all score and associated P value. Since complex pedigrees were used, this score has more power to detect linkage than scores generated only by comparison of IBD allele sharing between affected sibling pairs.⁴⁸ For comparison, affected sib-pair linkage analysis was performed within SPLINK version 1.08 that uses the maximum likelihood IBD method,^49,50 modified to allow for uncertainty of IBD assignment due to missing or incomplete parental data.^51,52 In nuclear families with more than one affected sib-pair, all comparisons are weighted at 2/A where A equals the number of affected siblings. This is considered to be a conservative strategy.⁵² Gene frequencies for marker loci were derived from unrelated individuals using SPLINK (Table 1).

Linkage simulations

To determine whether the 87 multicase families had sufficient power to detect linkage, simulations were performed within ALLEGRO.⁴⁷ One hundred replicates were generated using parameters (single gene additive model; allele frequency for disease gene q = 0.0016; extremes of age-related penetrances of 7% and 61%) defined initially by segregation analysis performed for these families,⁴ and actual HLA marker information for all individuals in the pedigrees. The simulated data sets were then used to carry out nonparametric linkage analysis within ALLEGRO to determine power to detect linkage at critical values of NPL_all scores.

Allelic association using transmission disequilibrium testing

The extended transmission disequilibrium test^53,54 was used to determine whether there was any bias in transmission of particular marker alleles from heterozygous parents to affected offspring. Analysis was performed using the program ETDT.⁵⁴ Nuclear families with marker information for only one parent were discarded by ETDT from the analysis. ETDT considers a genotype-wise (saturated model) where transmission probabilities for each allele of each genotype are individually examined to determine deviation from 50% transmission to affected offspring. An alternative allele-wise (parsimonious model) considers transmission probabilities of each allele across genotypes, to test the extent to which each allele is associated with disease. The test also provides information on transmission of individual alleles of the locus to affected offspring. In this case, a Bonferroni correction for multiple testing was applied by multiplying the P value by the number of alleles informative for transmission testing. When testing families where there are multiple transmissions, significant P values observed may be due to linkage rather than association.⁵⁴ Hence the haplotype relative risk (HRR; Falk and Rubinstein⁵⁵) was also used to restrict the analysis to one informative transmission per family, usually to the eldest affected child in each family. ETDT for haplotypes across multiple loci was performed within the computer program TDT Phase⁵⁶ that provides a global ETDT statistic for haplotype associations, as well as individual haplotype associations.

Linkage disequilibrium testing

Parents of all pedigrees were used as a sample of genetically independent population sample to measure linkage disequilibrium between markers. Conventional pairwise disequilibrium statistics (D'),⁵⁷ and associated P values, were determined for all pairwise combinations of markers across the class II and class III regions.

1 Ho M, Siongok TK, Lyerly WH, Smith DH. Prevalence and disease spectrum in a new focus of visceral leishmaniasis in Kenya. Trans R Soc Trop Med Hyg 1982; 76: 741-746. MEDLINE

2 Sacks DL, Lal SL, Shrivastava SN, Blackwell JM, Neva FA. An analysis of T cell responsiveness in Indian Kala-azar. J Immunol 1987; 138: 908-913. MEDLINE

3 Davies CR, Llanos-Cuentas EA, Pyke SD, Dye C. Cutaneous leishmaniasis in the Peruvian Andes: an epidemiological study of infection and immunity. Epidemiol Infect 1995; 114: 297-318. MEDLINE

4 Cabello PH, Lima AM, Azevedo ES, Kriger H. Familial aggregation of Leishmania chagasi infection in northeastern Brazil. Am J Trop Med Hyg 1995; 52: 364-365. MEDLINE

5 Peacock CS, Collins A, Shaw MA et al. Genetic epidemiology of visceral leishmaniasis in northeastern Brazil. Genet Epidemiol 2001; 20: 383-396. MEDLINE

6 Blackwell J, Freeman J, Bradley D. Influence of H-2 complex on acquired resistance to Leishmania donovani infection in mice. Nature 1980; 283: 72-74. MEDLINE

7 Blackwell JM. Leishmania donovani infection in heterozygous and recombinant _H-2_ haplotype mice. Immunogenetics 1983; 18: 101-109. MEDLINE

8 Blackwell JM, Roberts MB. Immunomodulation of murine visceral leishmaniasis by administration of monoclonal anti-Ia antibodies: differential effects of anti-I-A vs anti-I-E antibodies. Eur J Immunol 1987; 17: 1669-1672. MEDLINE

9 Kaye PM, Cooke A, Lund T, Wattie M, Blackwell JM. Altered course of visceral leishmaniasis in mice expressing transgenic I-E molecules. Eur J Immunol 1992; 22: 357-364. MEDLINE

10 Barral-Netto M, Badaro R, Barral A et al. Tumor necrosis factor (cachectin) in human visceral leishmaniasis. J Infect Dis 1991; 163: 853-857. MEDLINE

11 McGuire W, Hill ASV, Allsop CEM, Greenwood BM, Kwiatkowski D. Variation in the TNF- promoter region associated with susceptibility to cerebral malaria. Nature 1994; 371: 508-511. MEDLINE

12 Khoo SH, Pepper L, Snowden N et al. Tumour necrosis factor c2 microsatellite allele is associated with the rate of HIV disease progression. AIDS 1997; 11: 423-428. MEDLINE

13 Roy S, McGuire W, Mascie-Taylor CGN et al. TNF promoter polymorphism and susceptibility to lepromatous leprosy. J Infect Dis 1997; 176: 530-532. MEDLINE

14 Conway DJ, Holland MJ, Bailey RL et al. Scarring trachoma is associated with polymorphism in the tumor necrosis factor alpha (TNF-alpha) gene promoter and with elevated TNF-alpha levels in tear fluid. Infect Immun 1997; 65: 1003-1006. MEDLINE

15 McGuire W, Knight JC, Hill AV, Allsopp CE, Greenwood BM, Kwiatkowski D. Severe malarial anemia and cerebral malaria are associated with different tumor necrosis factor promoter alleles. J Infect Dis 1999; 179: 287-290. Article MEDLINE

16 Nadel S, Newport MJ, Booy R, Levin M. Variation in the tumor necrosis factor-alpha gene promoter region may be associated with death from meningococcal disease. J Infect Dis 1996; 174: 878-880. MEDLINE

17 Shaw MA, Donaldson IJ, Collins A et al. Association and linkage of leprosy phenotypes with HLA class II and tumour necrosis factor genes. Genes Immun 2001; 2: 196-204. Article MEDLINE

18 Cabrera M, Shaw M-A, Sharples C et al. Polymorphism in TNF genes associated with mucocutaneous leishmaniasis. J Exp Med 1995; 182: 1259-1264. MEDLINE

19 Wilson AG, Symons JA, McDowell TL, McDevitt HO, Duff GW. Effects of a polymorphism in the human tumour necrosis factor alpha promoter on transcriptional activation. Proc NY Acad Sci USA 1997; 94: 3195-3199.

20 Louis E, Franchimont D, Piron A et al. Tumour necrosis factor (TNF) gene polymorphism influences TNF-alpha production in lipopolysaccharide (LPS)-stimulated whole blood cell culture in healthy humans. Clin Exp Immunol 1998; 113: 401-406. MEDLINE

21 Knight JC, Udalova I, Hill AV et al. A polymorphism that affects OCT-1 binding to the TNF promoter region is associated with severe malaria. Nat Genet 1999; 22: 145-150. Article MEDLINE

22 Skoog T, van't Hooft FM, Kallin B et al. A common functional polymorphism (C--&gt;A substitution at position -863) in the promoter region of the tumour necrosis factor-alpha (TNF-alpha) gene associated with reduced circulating levels of TNF-alpha. Hum Mol Genet 1999; 8: 1443-1449. MEDLINE

23 Singh N, Sundar S, Williams F et al. Molecular typing of HLA class I and class II antigens in Indian kala-azar patients. Trop Med Int Health 1997; 2: 468-471. MEDLINE

24 Faghiri Z, Tabei SZ, Taheri F. Study of the association of HLA class I antigens with kala-azar. Hum Hered 1995; 45: 258-261. MEDLINE

25 Meddeb-Garnaoui A, Gritli S, Garbouj S et al. Association analysis of HLA-class II and class III gene polymorphisms in thesusceptibility to mediterranean visceral leishmaniasis. Hum Immunol 2001; 62: 509-517. MEDLINE

26 Amendoiera R, Guilherme L, Martin M et al. HLA and visceral leishmaniasis in families of endemic area in northeast Brazil. Mem I Oswaldo Cruz 1988; 83: 119.

27 Blackwell JM, Black GF, Peacock CS et al. Immunogenetics of leishmanial and mycobacterial infections: The Belem Family Study. Phil Trans Roy Soc B 1997; 352: 1331-1345.

28 Risch N. Assessing the role of HLA-linked and unlinked determinants of disease. Am J Hum Genet 1987; 40: 1-14. MEDLINE

29 Risch N, Merikangas K. The future of genetic studies of complex human diseases. Science 1996; 273: 1516-1517. MEDLINE

30 Kemp M, Kurtzhals JAL, Bendtzen K et al. Leishmania donovani-reactive Th1- and Th2-like T-cell clones from individuals who have recovered from visceral leishmaniasis. Infect Immun 1993; 61: 1069-1073. MEDLINE

31 Kurtzhals JA, Hey AS, Jardim A et al. Dichotomy of the human T cell response to Leishmania antigens. II. Absent or Th2-like response to gp63 and Th1-like response to lipophosphoglycan-associated protein in cells from cured visceral leishmaniasis patients. Clin Exp Immunol 1994; 96: 416-421. MEDLINE

32 Stern JJ, Oca MJ, Rubin BY, Anderson SL, Murray HW. Role of L3T4+ and LyT-2+ cells in experimental visceral leishmaniasis. J Immunol 1988; 140: 3971-3977. MEDLINE

33 Lainson R, Shaw JJ. Evolution, classification and geographical distribution. In: Peters W, Killick-Kendrick R (eds). The Leishmaniasis in Biology and Medicine Academic Press: London, 1987, 1-120.

34 Nascimento MD, Costa JM, Fiori BI et al. The epidemiological determinant aspects in the maintenance of visceral leishmaniasis in the state of Naranhao, Brazil. Rev Soc Bras Med Trop 1996; 39: 233-240.

35 Grimaldi G, Tesh RB, McMahon-Pratt D. A review of the geographic distribution and epidemiology of leishmaniasis in the New World. Am J Trop Med Hyg 1989; 41: 687-725. MEDLINE

36 Saiki RK, Bugawan TL, Horn GT, Mullis KB, Erlich HA. Analysis of enzymatically amplified beta-globin and HLA-DQalpha DNA with allele-specific oligonucleotide probes. Nature 1986; 324: 163-166. MEDLINE

37 Bidwell J. Advances in DNA-based HLA-typing methods. Immunol Today 1994; 15: 303-307. MEDLINE

38 Kimura A, Sasazuki T. Eleventh International Histocompatibility Workshop reference protocol for the HLA DNA typing technique. In: Tsuji K (ed) HLA: OUP, 1999, pp 397-419.

39 Lundin KE, Ronningnen KS, Aono S et al. HLA-DQ antigens and DQ beta amino acid 57 of Japanese patients with insulin-dependent diabetes mellitus: detection of a DRw8DQw8 haplotype. Tissue Antigens 1989; 34: 233-241. MEDLINE

40 Ronningen KS, Iwe T, Halstensen TS, Spurkland A, Thorsby E. The amino acid at position 57 of the HLA-DQ beta chain and susceptibility to develop insulin-dependent diabetes mellitus. Hum Immunol 1989; 26: 215-225. MEDLINE

41 Hsieh SL, March RE, Khanna A, Cross SJ, Campbell RD. Mapping of 10 novel microsatellites in the MHC class III region: application to the study of autoimmune disease. J Rheumatol 1997; 24: 220-222. MEDLINE

42 Nedospasov SA, Udalova IA, Kuprash DV, Turetskaya RL. DNA sequence polymorphism at the human tumor necrosis factor (TNF) locus. Numerous TNF/lymphotoxin alleles tagged by two closely linked microsatellites in the upstream region of the lymphotoxin (TNF-) gene. J Immunol 1991; 147: 1053-1059. MEDLINE

43 D'Alfonso S, Richiardi PM. A polymorphic variation in a putative regulation box of the TNFA promoter region. Immunogenetics 1994; 39: 150-154. MEDLINE

44 Wilson AG, di Giovine FS, Blakemore AIF, Duff GW. Single base polymorphism in the human Tumour Necrosis Factor alpha (TNF) gene detectable by NcoI restriction of PCR product. Hum Mol Genet 1993; 1: 353.

45 Messer G, Spengler U, Jung MC et al. Polymorphic structure of the tumor necrosis factor (TNF) locus: an NcoI polymorphism in the first intron of the human TNF- gene correlates with a variant amino acid in position 26 and a reduced level of TNF- production. J Exp Med 1991; 173: 209-219. MEDLINE

46 Morton NE, Shields DC, Collins A. Genetic epidemiology of complex phenotypes. Ann Hum Genet 1991; 55: 301-314. MEDLINE

47 Gudbjartsson DF, Jonasson K, Frigge ML, Kong A. Allegro, a new computer program for multipoint linkage analysis. Nat Genet 2000; 25: 12-13. Article MEDLINE

48 Kruglyak L, Daly MJ, Reeve-Daly MP, Lander ES. Parametric and nonparametric linkage analysis: a unified multipoint approach. Am J Hum Genet 1996; 58: 1347-1363. MEDLINE

49 Risch N. Linkage strategies for genetically complex traits. II. The power of affected relative pairs. Am J Hum Genet 1990; 46: 229-241. MEDLINE

50 Risch N. Linkage strategies for genetically complex traits. III. The effect of marker polymorphism on analysis of affected relative pairs. Am J Hum Genet 1990; 46: 242-253. MEDLINE

51 Holmans P. Asymptotic properties of affected-sib-pair linkage analysis. Am J Hum Genet 1993; 52: 362-374. MEDLINE

52 Holmans P, Clayton D. Efficiency of typing unaffected relatives in an affected sib-pair linkage study with single locus and multiple tightly-linked markers. Am J Hum Genet 1995; 37: 1221-1232.

53 Spielman RS, McGinnis RE, Ewens WJ. Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet 1993; 52: 506-516. MEDLINE

54 Sham PC, Curtis D. An extended transmission/disequilibrium test (TDT) for multi-allele marker loci. Ann Hum Genet 1995; 59: 323-336. MEDLINE

55 Falk CT, Rubinstein P. Haplotype relative risks: an easy reliable way to construct a proper control sample for risk calculations. Ann Hum Genet 1987; 51: 227-233. MEDLINE

56 Dudbridge F, Koeleman BP, Todd JA, Clayton DG. Unbiased application of the transmission disequilibrium test to multilocus haplotypes. Am J Hum Genet 2000; 66: 2009-2012. MEDLINE

57 Klitz W, Stephens JC, Grote M, Carrington M. Discordant patterns of linkage disequilibrium of the peptide-transporter loci within the HLA class II region. Am J Hum Genet 1995; 57: 1436-1444. MEDLINE

Figure 1 Multipoint nonparametric analysis for linkage between a putative disease susceptiblity allele for visceral leishmaniasis and 15 markers tested across the HLA class II/III region. (a) NPL_all scores, none achieved a NPL_all score of 1.5 required for statistical significance at the P = 0.05 level. (b) Information content mapping across the region. CentiMorgan (cM) distances on the X axes based on physical distances (see Table 1) where 1 cM = 1 Mb.

Table 1 Candidate gene marker information for visceral leishmaniasis families, northeastern Brazil

Table 2 Breakdown of number of affected sibs per nuclear family for 117 nuclear families contained within the 87 multicase families. Nuclear families with only one affected offspring were always part of more complex pedigree with multiple affected relative pairs that contribute to the NPL_all linkage analysis

Table 3 Combined segregation and linkage analysis using COMDS, and multipoint parametric linkage analysis using ALLEGRO. For COMDS, the best-fit additive single locus model was employed⁵ with parameters D = Dominance = 0.50, displacement on liability scale between homozygotes = T = 6.04, frequency of disease allele = Q = 0.0016. For each locus examined for linkage, the model was tested iterating the recombination fraction theta (), and with a fixed value of = (0.5). Results are expressed as ²₁ one-sided tests and Lod (Z). For ALLEGRO (multipoint LOD), the best-fit additive single locus model was employed⁵ with parameters D = 0.50, Q = 0.0016, and the extremes of age-related penetrances (7% and 61%) as computed in COMDS

Table 4 Affected sib-pair (SPLINK) analyses for HLA for visceral leishmaniasis families from northeastern Brazil

Table 5 Percent power to detect linkage at critical values of the NPL_all score determined by ALLEGRO linkage simulations at extremes (7% and 61%) of age-related penetrance for a putative disease susceptibility gene in 87 multicase families from northeastern Brazil

Table 6 ETDT analyses for HLA markers for visceral leishmaniasis in northeastern Brazil