The Major Histocompatibility Complex Conserved Extended Haplotype 8.1 in AIDS-related Non-Hodgkin’s Lymphoma

doi:10.1097/QAI.0b013e3181b017d5

J Acquir Immune Defic Syndr. Author manuscript; available in PMC 2011 January 5.

Published in final edited form as:

J Acquir Immune Defic Syndr. 2009 October 1; 52(2): 170–179.

doi: 10.1097/QAI.0b013e3181b017d5

PMCID: PMC3015185

NIHMSID: NIHMS250460

The Major Histocompatibility Complex Conserved Extended Haplotype 8.1 in AIDS-related Non-Hodgkin’s Lymphoma

Brahim Aissani, PhD,¹ Kisani M Ogwaro,² Sadeep Shrestha, PhD,¹ Jianming Tang, PhD,² Elizabeth C Breen, PhD,³ Hui-lee Wong, PhD,⁴ Lisa P Jacobson, ScD,⁵ Charles S Rabkin, MD,⁴ Richard F Ambinder, MD, PhD,⁶ Otoniel Martinez-Maza, PhD,^7,⁸ and Richard A Kaslow, MD^1,²

¹Department of Epidemiology, University of Alabama at Birmingham, Birmingham, AL

²Department of Medicine, University of Alabama at Birmingham, Birmingham, AL

³Department of Psychiatry and Biobehavioral Sciences, UCLA, Los Angeles, CA

⁴Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD

⁵Department of Epidemiology, Johns Hopkins University, Baltimore, MD

⁶Department of Oncology, Johns Hopkins University, Baltimore, MD

⁷Department of Obstetrics and Gynecology, UCLA, Los Angeles, CA

⁸Department of Microbiology, Immunology & Molecular Genetics, UCLA, Los Angeles, CA

Correspondence to: Brahim Aissani, PhD, Program in Epidemiology of Infection and Immunity, Department of Epidemiology, University of Alabama at Birmingham, 1665 University Blvd, RPHB Room 220A, Birmingham, Alabama 35294-0022. (Email: baissani/at/uab.edu; Email: rkaslow/at/uab.edu)

The publisher's final edited version of this article is available at J Acquir Immune Defic Syndr

See other articles in PMC that cite the published article.

Publisher's Disclaimer

Abstract

Two single nucleotide polymorphisms (SNPs) in adjacent genes, lymphotoxin alpha (LTA +252G, rs909253 A>G) and tumor necrosis factor (TNF −308A, rs1800629 G>A), form the G-A haplotype repeatedly associated with increased risk of non-Hodgkin’s lymphoma (NHL) in individuals uninfected with HIV-1. This association has been observed alone or in combination with HLA-B* 08 or HLA-DRB1*03 in the major histocompatibility complex (MHC). Which gene variant on this highly conserved extended haplotype (CEH 8.1) in Caucasians most likely represents a true etiologic factor remains uncertain. We aimed to determine whether the reported association of the G-A haplotype of LTA-TNF with non-AIDS NHL also occurs with AIDS-related NHL. SNPs in LTA and TNF and in six other genes nearby were typed in 140 non-Hispanic European American pairs of AIDS-NHL cases and matched controls selected from HIV-infected men in the Multicenter AIDS Cohort Study. The G-A haplotype and a 4-SNP haplotype in the neighboring gene cluster (rs537160 (A) rs1270942 (G), rs2072633 (A) and rs6467 (C)) were associated with AIDS-NHL (OR=2.7, 95% CI: 1.5–4.8, p=0.0009 and OR=3.2, 95% CI: 1.6–6.6 p=0.0008; respectively). These two haplotypes occur in strong linkage disequilibrium with each other on CEH 8.1. The CEH 8.1-specific haplotype association of MHC class III variants with AIDS-NHL closely resembles that observed for non-AIDS NHL. Corroboration of an MHC determinant of AIDS and non-AIDS NHL alike would imply an important pathogenetic mechanism common to both.

Keywords: Human Leukocyte Antigen, HIV, CD4, Multicenter AIDS Cohort NHL Study

INTRODUCTION

In individuals not infected with human immunodeficiency virus (HIV), single nucleotide polymorphisms (SNP) in the gene encoding lymphotoxin alpha (LTA) and those in the gene for tumor necrosis factor (TNF) have been associated with non-Hodgkin’s lymphoma (NHL)¹^–⁸. Specifically, the ‘G’ allele at nucleotide position +252 (rs909253 A>G) of LTA alone or in combination with the ‘A’ allele at position −308 (rs1800629 G>A) of TNF have been repeatedly associated with NHL in some ²^–⁵^,⁷ but not other ¹^,⁶^,⁸ studies. Associations with alleles at specific human leukocyte antigen (HLA) loci in the vicinity of TNF and LTA have also occasionally been reported ⁹^–¹¹. However, for several reasons it remains uncertain whether variants in LTA (+252G) and TNF (−308A) or markers in adjacent loci represent true etiologic factors. First, NHL often occurs in the context of autoimmunity, immunosuppression, infection ¹² and atopy ¹³; but neither high levels of the immunostimulatory TNF and LTA molecules nor other indications of their direct involvement in NHL have been well documented. Second, while the −308A variant has been associated with high levels of TNF production, the numerous attempts to relate sequence variation in those genes to functional variation as measured by production, levels, or biologic activity of the molecules have been controversial ¹⁴^–¹⁹. Finally, the SNPs in those two genes repeatedly associated with NHL most often occur on one of the most conserved extended haplotypes yet found in the human genome—the Caucasian HLA-B*08-containing ancestral haplotype CEH 8.1 ²⁰^,²¹ spanning more than 2 Mb in the MHC region.

No study to date has examined the MHC-related genetic determinants of NHL in the context of HIV/AIDS; and no class III genetic markers other than those in the LTA-TNF have been examined, leaving unexplored a large (~ 1 Mb) chromosomal region. We have sought to confirm the relationships seen in non-AIDS NHL through a case-control study of AIDS-NHL within a large cohort study by analyzing haplotypes formed by SNPs in the central MHC class III region. For this reason, we designed the present study to incorporate two sets of MHC class III SNPs in weak or no LD with each other (based on publicly available population data) to evaluate the effects associated with extended haplotypes in central MHC. In addition to the target TNF-LTA gene cluster, we selected SNPs from four genes in the region that lies about 330 Kb centromeric to TNF. We chose this region because: 1) it is one of the most polymorphic regions of the human genome; 2) it hosts a complement gene cluster including CFB (complement factor B) involved in the proliferation of B-lymphocytes, a hallmark of NHL pathogenesis; and 3) it lies between TNF and HLA-DR, two loci previously associated with NHL.

Taken together, our observed association of AIDS-NHL with the same LTA (+252G)-TNF (−308A) haplotype repeatedly associated with non-AIDS NHL, along with a comparable association of a nearby segment of CEH 8.1 bearing the C2 and CFB genes, corroborates the overall haplotypic effect of CEH 8.1 on NHL pathogenesis within and outside of the context of HIV/AIDS.

MATERIAL AND METHODS

Cohort characteristics and study design

We studied participants in the Multicenter AIDS Cohort Study (MACS), a prospective investigation of the natural history of HIV infection and AIDS ²² among 4954 homosexual men enrolled in 1984–1985 plus 668 men enrolled in 1987–1991 at four study centers in the US (Baltimore, Chicago, Los Angeles and Pittsburgh). Clinical information and blood samples were obtained at six month intervals. The MACS cohort study was approved by the Institutional Review Board at each center, where individual participants gave informed consent; testing for genetic variants potentially related to HIV/AIDS outcomes was included. Within the cohort, a case-control study was designed to compare serologic features in NHL cases and HIV-infected controls ²³. Cases were participants who were diagnosed with AIDS-NHL as of April 2002, had at least one serum sample from a time point preceding the diagnosis, and could be a matched with HIV-infected control as described below (n=180). Longitudinal serum samples were available at all or at any of the three designed time points prior to NHL diagnosis: > 3 years pre-NHL (closest to 4 years), 1–3 years pre-NHL (closest to 2 years) and/or 0–1 year pre-NHL (closest to 0.5 year).

Controls were HIV+ participants who had not been diagnosed with NHL as of April 2002. Matching criteria were: 1) duration of HIV infection based on the known date of seroconversion (n=21) or date of entry to the study as HIV seroprevalent (n=159), and 2) expected sample availability at equivalent time points (± 1 year). One-to-one matching of a case with a control minimized the case-control differences in the duration of HIV infection (6.7 and 6.9 years, respectively).

To facilitate control for confounding by HIV duration, up to 39 serial measurements of CD4+ count for each subject permitted estimation of a slope of the CD4+ count over time and backward imputation of the count in the control at the time of diagnosis in the index case (see Statistical analyses). Eleven matched pairs had no CD4+ count measurement in the cases (n=7) or in the controls (n=2) or in both cases and controls (n=2) within two years preceding the NHL diagnosis. Exclusion of these subjects led to a total of 169 matched case-control pairs.

Subphenotypic classification based on 112 (67%) of the NHL cases identified diffuse large B cells (28.0%); large B cell diffuse, immunoblastic (22%); Burkitt lymphoma (BL) and BL-like (17%); lymphoma/NHL, not otherwise specified (22%); and other (11%). An overlapping subset of NHL cases (n=95) were further classified as either systemic (56%) or central nervous system (CNS) (44%). Finally, 46 controls (29%) and 73 cases (44%) were also diagnosed with Kaposi’s sarcoma (KS) during the course of their HIV infection.

The analysis reported here was restricted to subjects of non-Hispanic European ancestry (n=140 pairs) because of the strong differentiation of HLA genes in ethnically diverse populations and because of insufficient numbers of subjects in other ethnic groups. Only crude screening analysis of NHL subphenotypes could be conducted in a study of this size and degree of histologic heterogeneity.

Laboratory methods

DNA polymorphisms at candidate MHC class III genes were genotyped using a commercially available genotyping platform (BeadArray®, Illumina Inc., San Diego, CA). SNP selection and genotyping.

A total of 63 SNPs were selected in candidate genes clustering in two regions of central MHC (Fig.1). A first telomeric set of 27 SNPs was selected from across LTA and TNF and extending up to NCR3 (natural cytotoxicity triggering receptor 3 gene) and a second centromeric set of 36 SNPs in the region containing genes from C2 (complement component 2) to CYP21A2 (cytochrome P450, family 21, subfamily A, polypeptide 2 gene). Within each candidate gene, a systematic search for informative SNPs in populations of Western European ancestry was conducted in public databases (HapMap I and SNP500 cancer database) as well as in private databases (Illumina Technologies, San Diego, CA and Applied Biosystems Inc., Foster City, CA). Primary criteria for SNP selection included a) aggressive (r² > 0.80) haplotype tagging potential (htSNP) across gene loci ²⁴, b) minor allele frequency (MAF) > 5%, c) predicted functionality (identified in PupaSuite) ²⁵ or documented association with NHL or location in genes relevant to B-cell activation and lymphomagenesis, and d) two-hit SNPs (or Illumina-validated) with a “designability” score of 1 (anticipated success rate > 80%). Quantification of DNA was performed with Molecular Probes (Invitrogen, Carlsbad, CA) prior to genotyping on the Illumina platform.

Figure 1

Portions of the subjects had previously been typed at HLA class I and class II loci for several different studies of HIV-related outcomes. HLA typing methods and the relationship between the effects at the HLA loci and the effects of the MHC class III loci are detailed in Supplementary Material.

Quality Control

Reliability of SNP typing was assessed through comparisons of duplicate data available for an average of 31 SNPs typed in a parallel study including a subset of NHL cases (n=65) and controls (n=44). Whenever departure from Hardy-Weinberg equilibrium (HWE) was observed for a SNP, the genotypic call score was checked prior to including a given SNP for analyses.

STATISTICAL ANAYSES

Statistical tests were performed in SAS 9.1 (SAS Institute Inc., Cary, NC), unless otherwise indicated. The effects of the genetic variants were evaluated in a case-control design after adjustment for other covariates as shown below.

Covariates

The distributions of ages between cases and controls were comparable in a previous study of these subjects ²³. To control for potential confounding by differential decline in CD4⁺ count during HIV disease progression, a well known effect of variants in HLA class I genes ²⁶, we assumed that the raw CD4⁺ count declined according to a linear function and used regression methods to impute the CD4⁺ count in each control at the time of NHL diagnosis in his matched case. In a minority of subjects (n=13) where data were insufficient to estimate the slopes reliably, the actual CD4⁺ count at the study visit closest to but preceding (within 2-years) NHL diagnosis was used. For the seroprevalent subjects we included all available CD4⁺ counts in the estimation of the slopes; for the seroconverters we included counts during the interval between one year post-seroconversion and the NHL diagnosis. Following log-transformation, standardization of the raw CD4⁺ counts and removal of the outliers (> 3 standard errors from zero), backward imputations of the CD4⁺ counts in the matched controls at the date of the index diagnoses were performed. The resulting covariate was included in the multivariate models for estimating the risk associated with selected DNA polymorphisms.

Population structure

Because cases and controls were selected from the same large cohorts of heterogeneous non-Hispanic European Americans recruited into the study before any NHL occurred, inadvertent bias due to systematic population stratification should have been minimal. Although we did not include standard genomic control markers ²⁷ in this analysis, the contribution of population structure among those of non-Hispanic European descent was nevertheless evaluated by testing the allelic diversity at the highly polymorphic and population-specific HLA class I and II loci among cases and controls (See Supplementary Data). We estimated the degree of gene differentiation among cases and controls using Weir and Cockerham’s variance-based method ²⁸ to estimate θs, an approximation of the Wright’s F-statistic ²⁹ and to assess the correlation of pairs of alleles between cases and controls. Confidence intervals (95%) around the estimates of θs were assessed with 10,000 bootstrap replicates and differences in overall allelic distributions among cases and controls were assessed using a G-like exact test as implemented in Genepop 4.0 ³⁰.

Single locus analysis

SNP markers were examined separately in case and control groups for adherence to HWE using Pearson’s chi-squared test. Pairwise LD between SNPs were measured using the coefficient of determination r². Graphical representations of LD patterns across the central MHC were obtained separately for cases and controls using Haploview ³¹. Estimates of the risk associated with each marker genotype were determined by the prevalence odds ratio (OR) and corresponding 95% confidence interval (95% CI) using logistic regression with or without adjustment for the imputed CD4⁺ counts. In the OR calculations the most frequent homozygous genotype in the controls served as the referent genotype. If a high level of heterozygosity or a low level of homozygosity was observed, dominant and additive (p for trend) genetic models were considered.

Multiple locus analysis

Haplotypes were assigned from the unphased genotypes at SNPs showing significant association (p < 0.05) in the single-locus tests. Haplotype frequencies were estimated using the expectation-maximization (EM) algorithm ³². To account for the long-range LD in the MHC, maximum likelihood estimates of haplotype frequencies were obtained iteratively starting with initial frequencies of 1/h, where h denotes the number of possible haplotypes in the sample. The 95% confidence intervals for haplotype frequencies were calculated using a binomial method ³³. Frequency estimates were obtained for the common haplotypes (>3%) ³⁴. MHC haplotypes were constructed by expanding the associated TNF-LTA haplotypes to include haplotypes formed by the associated SNPs in the complement region.

Overall differences in the distribution of haplotypes between cases and controls were assessed using the haplotype trend regression (HTR) approach ³⁵. This approach, assuming an additive model, estimates posterior probabilities for each subject for all EM-inferred haplotypes. These posterior probabilities were treated as independent variables in the HTR model with the weights in the design matrix reflecting various alternative inferences about haplotypes. A logistic regression model containing weighted haplotypes was applied to accommodate our case-control design ³⁶ and to allow control for confounding by differential CD4 effects. In the multivariate logistic model, the adjusted odds ratios represent the risk increase per haplotype copy. Haplotypes with a frequency < 3% were aggregated as a single term in the model. Haplotype associations were tested using the most prevalent haplotype in the controls as reference.

Multiple testing

No formal correction for comparison of multiple class III SNPs was applied because the primary objective of the study was to test the prior hypothesis that the two linked SNPs previously associated with non-AIDS NHL (LTA +252 and TNF −308), along with others in CEH 8.1, were also associated with AIDS NHL.

RESULTS

Marker statistics

Of the 63 SNPs assayed in the central MHC; 47 could be amplified and were polymorphic. For those SNPs, we compared 137 cases and 140 controls. The estimated average frequency of typing error (2–4%) was unlikely to affect the results materially.

Allele frequency and HWE

The panel of 47 SNPs analyzed is shown in Table 1. The SNPs (m30–m76) are shown ordered in telomere-to-centromere orientation from the LTA/TNF region (top) toward the complement gene region located 333 kilobases (Kb) away (bottom). Minor allele frequency (MAF) and deviation from HWE are shown separately for the cases and controls. Several markers (~ 38%) that had MAFs < 5% in cases and controls were excluded from further analyses. We considered the SNPs whose proportions departed from HWE only in cases or controls but not in both as potentially indicative of marker-disease associations (cases) or signs for selective constraints (controls) and analyzed them carefully. No significant (p < 0.05) departure from HWE was found in both cases and controls at any SNP loci, with exception to TNF-863 SNP (p=0.078 and p=0.004 in the cases and controls, respectively).

Table 1

Allele frequency and HWE at markers in the major histocompatibility complex ^a

Single locus analysis

Globally, 9 SNPs were found to be associated with increased risk for AIDS-NHL (Table 2). Of these, four located in the LTA/TNF gene cluster (m33, m34, m37, m39) included the commonly studied LTA +252G (m34) and TNF−308A (m39) polymorphisms, and five other polymorphisms (m51, m67, m70, m71 and m75) occurred in genes in and around the complement gene cluster. Three LTA markers (m33, m34 and m37) were in complete LD, and LTA +252G (m34) was subsequently used to tag all three in analyses. In univariate logistic models with matched data and adjustment for the CD4⁺ counts, 4 of these SNPs (m39, m67, m70 and m71) were found to be associated with increased risk for NHL (ORs=2.1 to 3.2). With the exception of SNP m70, the elevations of the ORs were of modest significance (0.01 < p <0.06) and occurred with the heterozygous genotype at every marker. The low frequencies of the variant homozygotes and the excess of heterozygotes observed at several SNPs supported analyses under dominance and additive genetic models. Similar increases of ORs were observed under dominant and additive genetic models (p for trend: 0.009–0.08). We excluded possible distortion of estimates due to a combination of matching by duration of HIV infection and controlling for the rate CD4⁺ change by comparing risk estimates from a model tested without adjustment for the CD4⁺ count with those obtained from a logistic model using unmatched data and adjustment for the effect of CD4⁺ counts. The two models yielded comparable risk estimates and more consistent trends in the genotypic risks (Table 2). We have concentrated on the unconditional model with adjustment for the CD4⁺ counts because it was based on a larger sample size (n=277 vs. n=238). Markers that significantly (p for trend < 0.05) modified the risk for NHL (LTA m34, TNF m39, CFB m67 and m70, RDBP m71 and CYP21A2 m75) were selected for multilocus-based tests of association.

Table 2

Association of selected SNPs in the MHC with AIDS-NHL in a case-control study of HIV-1-infected European American men

Linkage disequilibrium

To examine the relationships among the tested SNPs, long-range LD patterns were examined separately for cases and controls (supplemental Figs 1 & 2). The LD patterns and haploblock ³⁷ structures indicated that LTA/TNF markers are not in strong LD with markers in the complement gene cluster, with the exception of the TNF m39 and CFB m70 markers (r²=0.54 and r²=0.58 in cases and controls, respectively).

Haplotype analysis

With no overall correlation found between markers of the LTA/TNF and the complement gene clusters, we evaluated the risk associated with the MHC haplotypes separately across the two gene clusters. We first inferred haplotypes across the well-studied markers LTA +252G (m34) and TNF−308A (m39) and tested whether specific haplotypes containing these two SNPs were associated with AIDS-NHL. Sixty-one cases (44.5%) and 32 controls (22.8%) carried the LTA+252(G)-TNF−308(A) haplotype (henceforth haplotype G-A or hap1) (Table 3A). Consistent with previous studies of non-AIDS NHL ¹, the haplotype G-A conferred a 2.7-fold increase of risk (p=0.0009).

Table 3

Risk of AIDS-NHL associated with MHC haplotypes in HIV-1 infected European American men

For haplotypes across the complement gene region we found a unique haplotype [CFB m67(A)-CFB m70(G)-RDBP m71(A)-CYP21A2 m75(C)] (hap2) associated with a 3-fold increase in risk (OR=3.2; 95% CI:1.6–6.6; p=0.0008). To determine whether the risk haplotypes hap1 and hap2 are independent of each other, we evaluated the association with the combined 6-locus haplotypes (m34-m39-m67-m70-m71-m75). A haplotype formed by the juxtaposition of hap1 and hap2 [(m34(G)-m39(A)-m67(A)-m70(G)-m71(A)-m75(C)] (hap3) was the only haplotype that modified the risk for AIDS-NHL (OR=4.2; 95% CI: 2.0–8.9; p=0.0002).

The apparent discrepancy between the estimated ORs for Hap3 and Hap2 in all of the three sets A, B and C is due to the varying frequencies of the referent haplotype; adjustment had only a minor effect. Referent haplotype G-A-G-A used to estimate the OR for hap2 in the set A occurred in 28.7% of cases and 38.7% of controls. Referent haplotype A-G-G-A-G-A occurred in 19.3% of cases and 32.5% of controls.

Examination of the haplo-specific alleles at the 6 SNP sites on the CEH 8.1 contig NT_113891 (c6_COX cell line-derived genomic contig ³⁸) revealed that this haplotype is part of the conserved CEH 8.1. Interestingly, approximately one-third of the cases with G-A carried this haplotype in combination with haplotypes other than hap2 suggesting that the G-A haplotype may occur on other common haplotypes across the class I and III regions of the MHC.

With the partial existing HLA data we examined the connection between the early reported associations of NHL with HLA-DR3 and HLA-B8 and the more recently reported MHC class III associations. To this end, we expanded hap3 to include the closest telomeric (HLA-B) and centromeric (HLA-DRB1) HLA genes and found that hap4 (a CEH 8.1-specific haplotype formed by the combination of HLA-B*0801, hap3 and HLA-DRB1*0301) is the only 8-locus haplotype that was significantly associated (OR=7.8; 95% CI: 2.5–24.0, p=0.0004) with the NHL (supplementary Table S1). Further support for a CEH 8.1 effect comes from analyses that included partial data from other HLA class I (namely HLA-Cw) and class II (HLA-DQA1 and HLA-DQB1) genes (supplementary Table S1).

We evaluated the potential confounding or additional risk associated with comorbidity due to Kaposi’s sarcoma (KS) in a subset of cases (n=62) and controls (n=44) who also developed that condition during the course of their HIV infection. In the KS-free subset, the associations with haplotypes hap1–3 remained significant (Table 3B). The subset of subjects who developed KS was too small to permit meaningful estimates of risk (data not shown).

The availability of subphenotypic data for the anatomic location of AIDS-NHL [systemic vs. central nervous system (CNS)] permitted limited analysis of stratified data. For the subset of systemic cases, all three haplotypes (hap1–3) were positively associated with AIDS-NHL (Table 3C), whereas for the much smaller subset of CNS cases (n=45), the ORs could not be calculated with confidence (Supplementary Table S2).

DISCUSSION

In our case-control study of homosexual men in the MACS, carriage of LTA (+252G) and TNF (−308A), which closely tag the conserved extended MHC haplotype CEH 8.1, were associated with an approximately two-fold higher risk of AIDS-NHL. These LTA and TNF variants have repeatedly been reported in association with NHL unrelated to HIV infection ²^–⁵^,⁷. We detected associations of similar magnitude with SNPs tagging a nearby segment of the CEH 8.1 that contains a complement gene cluster. Alleles of HLA genes present on CEH 8.1 and less frequently studied in association with non-AIDS NHL ⁹^–¹¹ also showed comparable relationships among the subset of MACS subjects with available HLA typing. This first study of MHC effects in AIDS-NHL thus succeeded in its purpose of replicating the established association with non-AIDS NHL.

The multiple previous positive studies of the associated SNPs may have focused attention on TNF because its encoded protein is involved in a range of neoplastic processes ³⁹^,⁴⁰ and because its promoter variant −308A has been implicated, albeit not invariably, in relatively high TNF production ¹⁵^,¹⁶. However, that TNF marker and its companion in LTA are well documented elements of the most extensively conserved haplotype in the genome yet observed. CEH 8.1 stretches for at least 2 Mb between HLA class I and class II loci ²⁰^,²¹. Although this inordinately strong conservation of the haplotypic relationships of those two SNPs to CEH 8.1 is not in doubt, we nevertheless verified that our observed associations did not mainly reflect unusual recombinant events in that central MHC region. We demonstrated that more cases than controls who displayed the LTA-TNF G-A combination carried other alleles in the neighboring loci from LTA to CYP21A2 that are also recognized CEH 8.1 variants ³⁸. While these associations increase the likelihood that one or more causal loci in this extended haplotype will be implicated in NHL, they also diminish the probability that any population association with a given marker in that haplotype will actually signify a causal relationship. In order to emphasize the CEH 8.1-wide effect, we have reported associations with haplotypes rather than with SNPs.

Our study tested a specific hypothesis, and it is most unlikely that our detection of the identical associations seen multiple times before in non-AIDS NHL represents a chance finding. On the other hand, the small size of our study sample may have led to unstable and potentially somewhat inflated risk estimates; replication in another HIV cohort would not only confirm but also better quantify the risk.

We analyzed different groups of cases and controls and applied alternative statistical models to ensure that our results were not method-dependent. We have emphasized the unmatched case-control design with adjustment for the CD4⁺ count because it provided greater statistical power and probably yielded a less conservative statistical test. Further, the CEH 8.1 was earlier thought to be associated with a rapid decline of CD4⁺ cells in HIV-1 infection ²⁶. It was therefore important that an analysis of cases and controls matched by duration of HIV infection effectively eliminated the possibility of confounding by differential decline of CD4⁺ cells among pairs with CEH 8.1-positive NHL cases.

The extraordinary conservation in CEH 8.1 will make identification of the precise causal determinant on this haplotype by typical genetic approaches more difficult. By the same token, we documented that one-third of the G and A alleles were carried on non-HLA-B*08 lineages. Smaller numbers of subjects appear to carry both G and A alleles either on the same or the opposite chromosome but in the absence of the full CEH 8.1. If these cases represent recombinant chromosomes carrying CEH 8.1 with reduced conservation, they could be highly informative for fine mapping of the candidate region of the central MHC. However, since the EM-inferred haplotypes may not be entirely accurate for LTA +252 and TNF −308 double heterozygotes, direct experimental assessment of the phase in such subjects is warranted for future studies.

NHL is a neoplasm for which the strongest risk factors identified to date reflect dysregulation of the immune system. However, regardless of whether immune deficiency is congenital, iatrogenic in the setting of post-transplant immunosuppression, or acquired as a consequence of HIV infection, its presence could complicate the evaluation of true causal factor(s) whose function is altered in NHL. In the context of HIV/AIDS, factors other than the decline of CD4⁺ count with disease progression are suggested by 1) a disproportionally higher risk of NHL (10–100-fold higher than the population risk) in HIV-infected individuals even in the setting of moderate immune deficiency ⁴¹^,⁴² and 2) a less dramatic decline of systemic NHL incidence as compared with that of KS and primary CNS lymphoma in the era of highly active antiretroviral therapy⁴³^–⁴⁵. Prolonged immune deficiency and low CD4⁺ count one year prior to the time of NHL diagnosis have also been reported as independent predictors of NHL outcome⁴⁶. As noted above, we controlled for possible confounding of the observed genetic relationships by the underlying decline in immunity.

Inclusion of a substantial proportion of subjects who also developed KS did not distort the association with CEH 8.1 because the HLA alleles of that haplotype have not been associated with HIV-related KS in the MACS ⁴⁷. Clinical information on the histological type was available for only a subset of our cases; that limitation precluded statistically meaningful tests by tumor type. Likewise, the study showed a clear association with systemic AIDS-NHL but that with central nervous system AIDS-NHL could not be accurately assessed because of the small sample size; however the trend was similar to that observed with the unstratified sample (Supplementary Table S2).

Recent reports ¹^,⁶^,⁸ have described a protective association with TNF −857T, a SNP that was not included in our genotyping scheme. Although no earlier study detected significant associations with both the TNF −308A - LTA +252G haplotype and the TNF −857 SNP, we cannot exclude a protective role for the latter in our AIDS-NHL. The documented occurrence of the two TNF SNPs on distinct MHC lineages ⁸ suggests that they are differentially distributed in different Caucasian populations. That may explain the consistency in the findings for the haplotype containing −308A in our study and the others including relatively heterogeneous European populations, and the contrasting –857T association in subjects of predominantly British ancestry. Moreover, the –857T association has appeared stronger with follicular lymphoma ¹^,⁸, while the −308A association has appeared stronger with diffuse large B cell disease ⁴^,⁶, the form more frequently represented in our population as well.

The pathogenetic mechanisms underlying the increased susceptibility of CEH 8.1 carriers to NHL and to a number of other conditions of autoimmunity and immune dysfunction ²¹, are poorly understood. Adaptive and demographic considerations have been invoked to account for the apparent contrast between increased disease susceptibility and high population frequency (selective advantages) of CEH 8.1 ⁴⁸. Our data show that, except for TNF −308A and CFB m70, the remaining four SNPs associated with AIDS-NHL have elevated minor allele frequencies in both the controls (MAF ≥ 0.27) and the cases (MAF≥ 0.38), emphasizing the necessity to focus future mapping efforts on those MHC markers in strong LD with the presumably less extended TNF −308(A)-CFB m70(G) subhaplotype.

To summarize, the present study extends previously reported association of LTA +252G and TNF −308A from non-AIDS to AIDS-NHL and, as with the recent observations in non-AIDS NHL ¹⁰^,¹¹, shows that the positive association with these two polymorphisms extends across CEH 8.1. In light of the extraordinary allelic invariance across several megabases of central and extended MHC of CEH 8.1-bearing chromosomes ²⁰^,³⁸^,⁴⁹^–⁵¹, our results as well as those reported for non-AIDS NHL strongly support an association with one or more genetic variants somewhere in the G-A-bearing haplotypes rather than suggesting an LTA-TNF-specific causal relationship. Fortunately, few large HIV-infected populations now go untreated for long enough to permit NHL to develop in numbers sufficient for an investigation similar to ours. It could therefore be difficult to use another population study to confirm our observation of an MHC haplotype determinant common to the pathogenesis of AIDS- and non-AIDS NHL alike.

Supplementary Material

Click here to view.^{(441K, pdf)}

Click here to view.^{(971K, pdf)}

ACKNOWLEDGMENT

Data in this manuscript were collected by the Multicenter AIDS Cohort Study (MACS) with centers (Principal Investigators) at The Johns Hopkins University Bloomberg School of Public Health (Joseph B. Margolick, Lisa Jacobson), Howard Brown Health Center and Northwestern University Medical School (John Phair), University of California, Los Angeles (Roger Detels), and University of Pittsburgh (Charles Rinaldo). Website located at http://www.statepi.jhsph.edu/macs/macs.html.

Sources of support

Partial funding was provided by R01-CA106168 (R.A.K.), R01-CA73475 (O.M.M.), P30-AI045008 Penn Center for AIDS Research subcontract (B.A.) and P50-CA96888 (R.A.). The Multicenter AIDS Cohort Study (MACS) is funded by the National Institute of Allergy and Infectious Diseases, with additional supplemental funding from the National Cancer Institute and the National Heart, Lung and Blood Institute. UO1-AI-35042, 5-MO1-RR-00722 (GCRC), UO1-AI-35043, UO1-AI-37984, UO1-AI-35039, UO1-AI-35040, UO1-AI-37613, UO1-AI-35041.

Footnotes

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Part of the data were presented at the 58^th annual meeting of the American Society of Human Genetics (ASHG), Philadelphia, Nov 11–15, 2008.

REFERENCES

1. Spink CF, Keen LJ, Mensah FK, Law GR, Bidwell JL, Morgan GJ. Association between non-Hodgkin lymphoma and haplotypes in the TNF region. Br J Haematol. 2006 May;133(3):293–300. [PubMed]

2. Seidemann K, Zimmermann M, Book M, et al. Tumor necrosis factor and lymphotoxin alfa genetic polymorphisms and outcome in pediatric patients with non-Hodgkin's lymphoma: results from Berlin-Frankfurt-Munster Trial NHL-BFM 95. J Clin Oncol. 2005 Nov 20;23(33):8414–8421. [PubMed]

3. Wang SS, Cozen W, Cerhan JR, et al. Immune mechanisms in non-Hodgkin lymphoma: joint effects of the TNF G308A and IL10 T3575A polymorphisms with non-Hodgkin lymphoma risk factors. Cancer Res. 2007 May 15;67(10):5042–5054. [PubMed]

4. Wang SS, Cerhan JR, Hartge P, et al. Common genetic variants in proinflammatory and other immunoregulatory genes and risk for non-Hodgkin lymphoma. Cancer Res. 2006 Oct 1;66(19):9771–9780. [PubMed]

5. Rothman N, Skibola CF, Wang SS, et al. Genetic variation in TNF and IL10 and risk of non-Hodgkin lymphoma: a report from the InterLymph Consortium. Lancet Oncol. 2006 Jan;7(1):27–38. [PubMed]

6. Lan Q, Zheng T, Rothman N, et al. Cytokine polymorphisms in the Th1/Th2 pathway and susceptibility to non-Hodgkin lymphoma. Blood. 2006 May 15;107(10):4101–4108. [PubMed]

7. Warzocha K, Ribeiro P, Bienvenu J, et al. Genetic polymorphisms in the tumor necrosis factor locus influence non-Hodgkin's lymphoma outcome. Blood. 1998 May 15;91(10):3574–3581. [PubMed]

8. Purdue MP, Lan Q, Kricker A, et al. Polymorphisms in immune function genes and risk of non-Hodgkin lymphoma: findings from the New South Wales non-Hodgkin Lymphoma Study. Carcinogenesis. 2007 Mar;28(3):704–712. [PubMed]

9. Juszczynski P, Kalinka E, Bienvenu J, et al. Human leukocyte antigens class II and tumor necrosis factor genetic polymorphisms are independent predictors of non-Hodgkin lymphoma outcome. Blood. 2002 Oct 15;100(8):3037–3040. [PubMed]

10. Nowak J, Kalinka-Warzocha E, Juszczynski P, et al. Haplotype-specific pattern of association of human major histocompatibility complex with non-Hodgkin's lymphoma outcome. Tissue Antigens. 2008 Jan;71(1):16–26. [PubMed]

11. Nowak J, Kalinka-Warzocha E, Juszczynski P, et al. Association of human leukocyte antigen ancestral haplotype 8.1 with adverse outcome of non-Hodgkin's lymphoma. Genes Chromosomes Cancer. 2007 May;46(5):500–507. [PubMed]

12. Engels EA. Infectious agents as causes of non-Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev. 2007 Mar;16(3):401–404. [PubMed]

13. Grulich AE, Vajdic CM, Cozen W. Altered immunity as a risk factor for non-Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev. 2007 Mar;16(3):405–408. [PubMed]

14. Wilson AG, Symons JA, McDowell TL, McDevitt HO, Duff GW. Effects of a polymorphism in the human tumor necrosis factor alpha promoter on transcriptional activation. Proc Natl Acad Sci U S A. 1997 Apr 1;94(7):3195–3199. [PubMed]

15. Bouma G, Xia B, Crusius JB, et al. Distribution of four polymorphisms in the tumour necrosis factor (TNF) genes in patients with inflammatory bowel disease (IBD) Clin Exp Immunol. 1996 Mar;103(3):391–396. [PubMed]

16. 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 Sep;113(3):401–406. [PubMed]

17. He B, Navikas V, Lundahl J, Soderstrom M, Hillert J. Tumor necrosis factor alpha-308 alleles in multiple sclerosis and optic neuritis. J Neuroimmunol. 1995 Dec 31;63(2):143–147. [PubMed]

18. Huizinga TW, Westendorp RG, Bollen EL, et al. TNF-alpha promoter polymorphisms, production and susceptibility to multiple sclerosis in different groups of patients. J Neuroimmunol. 1997 Feb;72(2):149–153. [PubMed]

19. Turner DM, Grant SC, Lamb WR, et al. A genetic marker of high TNF-alpha production in heart transplant recipients. Transplantation. 1995 Nov 27;60(10):1113–1117. [PubMed]

20. Degli-Esposti MA, Leaver AL, Christiansen FT, Witt CS, Abraham LJ, Dawkins RL. Ancestral haplotypes: conserved population MHC haplotypes. Hum Immunol. 1992 Aug;34(4):242–252. [PubMed]

21. Price P, Witt C, Allcock R, et al. The genetic basis for the association of the 8.1 ancestral haplotype (A1, B8, DR3) with multiple immunopathological diseases. Immunol Rev. 1999 Feb;167:257–274. [PubMed]

22. Kaslow RA, Ostrow DG, Detels R, Phair JP, Polk BF, Rinaldo CR., Jr The Multicenter AIDS Cohort Study: rationale, organization, and selected characteristics of the participants. Am J Epidemiol. 1987 Aug;126(2):310–318. [PubMed]

23. Breen EC, Magpantay L, Epeldegui M, Hung YP, Detels R, Kaslow R, Rabkin C, Ambinder R, Jacobson LP, Martínez-Maza O. Treatment and Prevention. Australia: Sydney; 2007. Serum and cellular markers of B cell activation are increased several years preceding the diagnosis of AIDS-associated non-Hodgkin’s B cell lymphoma. Paper presented at: IAS Conference on HIV Pathogenesis.

24. de Bakker PI, Yelensky R, Pe'er I, Gabriel SB, Daly MJ, Altshuler D. Efficiency and power in genetic association studies. Nat Genet. 2005 Nov;37(11):1217–1223. [PubMed]

25. Conde L, Vaquerizas JM, Dopazo H, et al. PupaSuite: finding functional single nucleotide polymorphisms for large-scale genotyping purposes. Nucleic Acids Res. 2006 Jul 1;34:W621–W625. (Web Server issue) [PMC free article] [PubMed]

26. Kaslow RA, Duquesnoy R, VanRaden M, et al. A1, Cw7, B8, DR3 HLA antigen combination associated with rapid decline of T-helper lymphocytes in HIV-1 infection. A report from the Multicenter AIDS Cohort Study. Lancet. 1990 Apr 21;335(8695):927–930. [PubMed]

27. Devlin B, Roeder K. Genomic control for association studies. Biometrics. 1999 Dec;55(4):997–1004. [PubMed]

28. Weir BSC, C.C. Estimating F-statistics for the analysis of population structure. Evolution. 1984;38:1358–1370.

29. Wright S. Isolation by Distance. Genetics. 1943 Mar;28(2):114–138. [PubMed]

30. Raymond M, Rousset F. GENEPOP (version 1.2): Population genetics software for excat tests and ecumenicism. J. Heredity. 1995;86:248–249.

31. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005 Jan 15;21(2):263–265. [PubMed]

32. Excoffier L, Slatkin M. Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol. 1995 Sep;12(5):921–927. [PubMed]

33. Hawley ME, Kidd KK. HAPLO: a program using the EM algorithm to estimate the frequencies of multi-site haplotypes. J Hered. 1995 Sep–Oct;86(5):409–411. [PubMed]

34. Clayton D. SNPHAP: A Program for Estimating Frequencies of Large Haplotypes of SNPs. " http://www-gene.cimr.cam.ac.uk/clayton/software/snphap.txt”.

35. Zaykin DV, Westfall PH, Young SS, Karnoub MA, Wagner MJ, Ehm MG. Testing association of statistically inferred haplotypes with discrete and continuous traits in samples of unrelated individuals. Hum Hered. 2002;53(2):79–91. [PubMed]

36. Shrestha S, Strathdee SA, Galai N, et al. Behavioral risk exposure and host genetics of susceptibility to HIV-1 infection. J Infect Dis. 2006 Jan 1;193(1):16–26. [PubMed]

37. Gabriel SB, Schaffner SF, Nguyen H, et al. The structure of haplotype blocks in the human genome. Science. 2002 Jun 21;296(5576):2225–2229. [PubMed]

38. Stewart CA, Horton R, Allcock RJ, et al. Complete MHC haplotype sequencing for common disease gene mapping. Genome Res. 2004 Jun;14(6):1176–1187. [PubMed]

39. Anderson GM, Nakada MT, DeWitte M. Tumor necrosis factor-alpha in the pathogenesis and treatment of cancer. Curr Opin Pharmacol. 2004 Aug;4(4):314–320. [PubMed]

40. Szlosarek P, Charles KA, Balkwill FR. Tumour necrosis factor-alpha as a tumour promoter. Eur J Cancer. 2006 Apr;42(6):745–750. [PubMed]

41. Clifford GM, Polesel J, Rickenbach M, et al. Cancer risk in the Swiss HIV Cohort Study: associations with immunodeficiency, smoking, and highly active antiretroviral therapy. J Natl Cancer Inst. 2005 Mar 16;97(6):425–432. [PubMed]

42. Beral V, Peterman T, Berkelman R, Jaffe H. AIDS-associated non-Hodgkin lymphoma. Lancet. 1991 Apr 6;337(8745):805–809. [PubMed]

43. Rabkin CS. AIDS and cancer in the era of highly active antiretroviral therapy (HAART) Eur J Cancer. 2001 Jul;37(10):1316–1319. [PubMed]

44. Matthews GV, Bower M, Mandalia S, Powles T, Nelson MR, Gazzard BG. Changes in acquired immunodeficiency syndrome-related lymphoma since the introduction of highly active antiretroviral therapy. Blood. 2000 Oct 15;96(8):2730–2734. [PubMed]

45. Gates AE, Kaplan LD. AIDS malignancies in the era of highly active antiretroviral therapy. Oncology (Williston Park) 2002 May;16(5):657–665. discussion 665, 668–670. [PubMed]

46. Grulich AE, Wan X, Law MG, et al. B-cell stimulation and prolonged immune deficiency are risk factors for non-Hodgkin's lymphoma in people with AIDS. Aids. 2000 Jan 28;14(2):133–140. [PubMed]

47. Dorak MT, Yee LJ, Tang J, et al. HLA-B, -DRB1/3/4/5, and -DQB1 gene polymorphisms in human immunodeficiency virus-related Kaposi's sarcoma. J Med Virol. 2005 Jul;76(3):302–310. [PubMed]

48. Traherne JA. Human MHC architecture and evolution: implications for disease association studies. Int J Immunogenet. 2008 Jun;35(3):179–192. [PMC free article] [PubMed]

49. Aly TA, Eller E, Ide A, et al. Multi-SNP analysis of MHC region: remarkable conservation of HLA-A1-B8-DR3 haplotype. Diabetes. 2006 May;55(5):1265–1269. [PubMed]

50. Horton R, Gibson R, Coggill P, et al. Variation analysis and gene annotation of eight MHC haplotypes: the MHC Haplotype Project. Immunogenetics. 2008 Jan;60(1):1–18. [PMC free article] [PubMed]

51. Smith WP, Vu Q, Li SS, Hansen JA, Zhao LP, Geraghty DE. Toward understanding MHC disease associations: partial resequencing of 46 distinct HLA haplotypes. Genomics. 2006 May;87(5):561–571. [PubMed]