Large-scale consortial efforts now provide convincing evidence that the pro-inflammatory cytokine tumor necrosis factor (TNF) promoter polymorphism (TNF G-308A), which is thought to increase TNF-α protein expression resulting in inflammation, is associated with increased risk of non-Hodgkin lymphoma (NHL) and specifically with the NHL subtype, diffuse large B-cell lymphoma (DLBCL) among Caucasians.1 The largest effort to date from the International Lymphoma Epidemiology Consortium (InterLymph) comprising 7999 incidence NHL cases and 8452 controls report that TNF-308A carriers have a 1.25-fold (per allele) increased risk for DLBCL and a 1.35-fold increased risk for marginal zone lymphoma.2
A limitation to the published association studies, however, is the inability to delineate the association between TNF and NHL from human leukocyte antigen (HLA) alleles that are known to be in linkage disequilibrium. TNF is located on chromosome 6p21.3 among the Class III genes of the major histocompatibility complex, 250 kb centromeric to the HLA-B locus and 850 kb telomeric to the Class II HLA-DR locus. It is well documented that Caucasian populations carry the 8.1 ancestral haplotype (AH) that includes the TNF-308A allele (HLA-A1-B8-TNF-308A-DR3-DQ2).3 Notably, the 8.1 AH is implicated in the risk of numerous autoimmune conditions, including those associated with NHL (for example, systemic lupus erythematosus and Sjogren’s syndrome).3–5 It remains unknown, however, whether the association reported for TNF-308A is due to or independent from HLA alleles and/or haplotypes.
Here, we present data from 555 controls and 610 cases from a US population-based case–control study of NHL where HLA Class I and Class II alleles were evaluated in the context of TNF with regard to their role in risk for NHL and NHL subtypes.
As has been previously described,6 the multi-center National Cancer Institute—Surveillance, Epidemiology and End Results (NCI-SEER) NHL case–control study population comprised 1321 newly diagnosed NHL cases identified in four Surveillance, Epidemiology, and End Results registries (Iowa; Detroit, MI; Los Angeles, CA; Seattle, WA) aged 20–74 years between July 1998 and June 2000 without evidence of HIV infection; 1057 population controls were identified by random digit dialing (<65 years) and from Medicare eligibility files (≥65 years). Written informed consent was obtained from each participant before interview. All study participants were also asked to provide a venous blood or mouthwash buccal cell sample. The present analysis was conducted on subset of non-Hispanic Caucasian cases and controls who provided blood (Supplementary Table 1); DNA were extracted using Puregene Autopure DNA extraction kits (Gentra Systems, Minneapolis, MN, USA). Duplicate samples from 100 participants processed in an identical manner were interspersed for all assays and blinded from the laboratory. Agreement for QC duplicates was ≥99%.
NHL pathology and subtype information were derived from abstracted reports by the local diagnosing pathologist. All cases were histologically confirmed and have been coded according to the International Classification of Diseases for Oncology (ICD), 2nd Edition and updated to the WHO/ICD-O-3. In addition to NHL overall, we evaluated the histologic subtypes DLBCL, follicular lymphoma, marginal zone lymphoma, chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL), and T-cell lymphomas (includes mycosis fungoides/sezary syndrome, peripheral T-cell lymphomas, and T/NK-cell lymphomas not otherwise specified).
Four-digit HLA Class I (A, B, C) and Class II genotyping (DR) was conducted at the NCI Laboratory for Genomic Diversity (Frederick, MD, USA) (AMA) according to sequence-specific oligonucleotide probe hybridization and sequence-based typing protocols developed by the 13th International Histocompatibility Workshop.7 HLA alleles were defined as presence or absence of the specific allele (‘+’ if the individual had either one or two copies of the allele, and ‘−’ if the individual had no copies of the allele). TNF genotyping was conducted at the National Cancer Institute Core Genotyping Facility (Gaithersburg, MD, USA) using the Taqman (Foster City, CA, USA) platform. TNF G-308A was defined by genotype (GG (referent), GA and AA).6 Haplotypes for HLA-A-B-DR were determined using ‘FastHap’, which determines haplotypes by expectation maximization (http://home.ncifcrf.gov/ccr/lgd/bioinformatics/index_n.asp). For HLA and TNF, odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to determine the magnitude and statistical significance of associations. All risk estimates are presented for non-Hispanic Caucasians and adjusted for the study design variables: sex, age (<45, 45–64, 65+ years) and study center (Detroit, Iowa, Los Angeles, Seattle). We also conducted analyses among all cases and controls (665 controls and 715 cases) adjusted for race and results were consistent. All logistic regression models were unconditional and conducted using SAS 9.1.3 (SAS Institute, Cary, NC, USA). All tests of statistical significance are two-sided.
We found the HLA-A*01-B*08-DR*03 haplotype associated with elevated risk of DLBCL (Table 1). The association appeared driven by the individual alleles that are part of the ancestral 8.1 haplotype, notably HLA-B *0801 (ORDLBCL = 1.63, 95% CI = 1.08–2.45). HLA-B *0801 was also associated with elevated risks of marginal zone, follicular and T-cell lymphomas though not statistically significant. Similar associations were observed for HLA-DRB1*0301.
Table 1.
Main associations between HLA-A*0101, HLA-B*0801, HLA-DR*0301 and TNF with all NHL and NHL subtypes (DLBCL, follicular, marginal zone, SLL/CLL and T-cell lymphomas) among non-Hispanic Caucasians, adjusted for age, sex and study center.
| Allele or haplotype* | Control n (%) | NHL n (%) | OR (95%CI) | DLBCL n (%) | OR (95%CI) | Follicular n (%) | OR (95%CI) | Marginal Zone n (%) | OR (95%CI) | CLL/SLL n (%) | OR (95%CI) | T cell n (%) | OR (95%CI) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HLA-A*01-B*08-DR*03 | |||||||||||||
| − | 438 (87) | 455 (84) | 1.0 (ref) | 123 (81) | 1.0 (ref) | 128 (86) | 1.0 (ref) | 36 (80) | 1.0 (ref) | 71 (87) | 1.0 (ref) | 28 (82) | 1.0 (ref) |
| + | 67 (13) | 88 (16) | 1.21 (0.89–1.78) | 29 (19) | 1.59 (0.98–2.59) | 21 (14) | 1.05 (0.61–1.80) | 9 (20) | 1.65 (0.75–3.61) | 11 (13) | 0.97 (0.48–1.94) | 6 (18) | 1.28 (0.49–3.28) |
| HLA-A*0101 | |||||||||||||
| − | 373 (70) | 389 (67) | 1.0 (ref) | 107 (65) | 1.0 (ref) | 108 (67) | 1.0 (ref) | 30 (64) | 1.0 (ref) | 58 (71) | 1.0 (ref) | 25 (69) | 1.0 (ref) |
| + | 162 (30) | 188 (33) | 1.12 (0.88–1.45) | 58 (35) | 1.28 (0.88–1.86) | 54 (33) | 1.19 (0.81–1.74) | 17 (36) | 1.35 (0.72–2.54) | 24 (29) | 0.96 (0.57–1.61) | 11 (31) | 1.07 (0.51–2.29) |
| HLA-B*0801 | |||||||||||||
| − | 435 (81) | 432 (70) | 1.0 (ref) | 120 (73) | 1.0 (ref) | 121 (74) | 1.0 (ref) | 34 (72) | 1.0 (ref) | 64 (78) | 1.0 (ref) | 28 (74) | 1.0 (ref) |
| + | 102 (19) | 148 (25) | 1.45 (1.09–1.94) | 45 (27) | 1.63 (1.08–2.45) | 42 (26) | 1.42 (0.93–2.17) | 13 (28) | 1.69 (0.85–3.33) | 18 (22) | 1.18 (0.67–2.09) | 10 (26) | 1.50 (0.69–3.26) |
| HLA-DR*0301 | |||||||||||||
| − | 415 (80) | 405 (74) | 1.0 (ref) | 110 (72) | 1.0 (ref) | 113 (75) | 1.0 (ref) | 28 (62) | 1.0 (ref) | 65 (79) | 1.0 (ref) | 24 (67) | 1.0 (ref) |
| + | 105 (20) | 145 (26) | 1.41 (1.06–1.88) | 43 (28) | 1.55 (1.02–2.36) | 38 (25) | 1.30 (0.83–2.00) | 17 (38) | 2.47 (1.29–4.71) | 17 (21) | 1.04 (0.58–1.86) | 12 (33) | 1.91 (0.90–4.04) |
| TNF G-308A | |||||||||||||
| GG | 383 (73) | 375 (67) | 1.0 (ref) | 98 (61) | 1.0 (ref) | 107 (69) | 1.0 (ref) | 29 (63) | 1.0 (ref) | 64 (80) | 1.0 (ref) | 22 (58) | 1.0 (ref) |
| GA | 130 (25) | 170 (30) | 1.33 (1.02–1.75) | 56 (35) | 1.72 (1.17–2.54) | 44 (28) | 1.19 (0.79–1.79) | 16 (35) | 1.61 (0.84–3.09) | 16 (20) | 0.70 (0.39–1.27) | 14 (37) | 1.77 (0.86–3.66) |
| AA | 12 (2) | 16 (3) | 1.45 (0.67–3.14) | 7 (4) | 2.37 (0.90–6.27) | 4 (3) | 1.17 (0.36–3.79) | 1 (2) | 1.23 (0.15–9.78) | — | — | 2 (5) | 2.95 (0.57–15.3) |
| GA/AA | 142 (27) | 186 (33) | 1.34 (1.03–1.75) | 63 (39) | 1.78 (1.22–2.59) | 48 (31) | 1.19 (0.80–1.77) | 17 (37) | 1.58 (0.83–3.00) | 16 (20) | 0.65 (0.36–1.18) | 16 (42) | 1.87 (0.93–3.75) |
| HLA-A*01-B*08-DR*03-TNF-A | |||||||||||||
| − | 425 (87) | 441 (84) | 1.0 (ref) | 120 (81) | 1.0 (ref) | 121 (85) | 1.0 (ref) | 36 (82) | 1.0 (ref) | 69 (87) | 1.0 (ref) | 28 (82) | 1.0 (ref) |
| + | 62 (13) | 82 (18) | 1.27 (0.89–1.82) | 28 (19) | 1.65 (1.00–2.72) | 21 (15) | 1.17 (0.68–2.01) | 8 (18) | 1.57 (0.69–3.56) | 19 (13) | 0.96 (0.46–1.98) | 6 (18) | 1.33 (0.52–3.45) |
Abbreviations: CI, confidence intervals; DLBCL, diffuse large B-cell lymphoma; HLA, human leukocyte antigen; NHL, non-Hodgkin lymphoma; OR, odds ratios; TNF, tumor necrosis factor; CLL/SLL, chronic lymphocytic leukemia/small lymphocytic lymphoma.
Numbers are not equivalent for each allele because of their differential genotyping completion rates. Results with P<0.05 are indicated in bold font.
The association between TNF-308A and NHL/DLBCL has been previously reported in this population;7 among the case and control subset comprising the present analysis, the association between TNF-308A and DLBCL remained statistically significant (ORAG/AA = 1.78, 95% CI = 1.22–2.59, referent = GG) (Table 1).
Risks for DLBCL among individuals with the HLA-A*01-B*08-DR*03-TNF haplotype (OR = 1.65, 95% CI = 1.00–2.72) were not more pronounced than that observed for TNF (Table 1).
We further evaluated the association between TNF and DLBCL in the context of each HLA allele constituting the A*01-B *08-DR*03 haplotype. We observed statistically significantly elevated DLBCL risk among those with the TNF-308A variant allele with or without the HLA-B*0801 allele (ORHLA-B*0801+ = 1.78, 95% CI = 1.15–2.76; OR HLA-B*0801− = 1.95, 95% CI = 1.11–3.42) or HLA-A*01-B*08-DR*03 haplotype (ORHLA-A *01−B*08-DR*03+ = 1.80, 95% CI = 1.08–3.00; ORHLA-A*01-B*08-DR*03− = 1.63, 95% CI = 1.00–2.68) (Table 2). Similar risk associations for DLBCL were observed for those with the TNF G-308A variant allele, with or without the HLA-A*01 or HLA-DR*03 alleles.
Table 2.
Risk estimates for NHL, DLBCL, marginal zone and T-cell lymphoma among non-Hispanic Caucasians with the TNF G-308A variant allele delineated by the presence or absence of the HLA-A*01-B*08-DR*03 haplotype, adjusted for age, sex and study center
| Allele or haplotypea | TNF G-308A | Control n (%) | NHL n (%) | OR (95%CI) | DLBCL n (%) | OR (95%CI) | Marginal Zone n (%) | OR (95%CI) |
|---|---|---|---|---|---|---|---|---|
| HLA-A*01 | ||||||||
| − | GG | 305 (59) | 303 (55) | 1.0 (ref) | 81 (51) | 1.0 (ref) | 25 (54) | 1.0 (ref) |
| − | GA/AA | 57 (11) | 74 (13) | 1.32 (0.90–1.94) | 23 (14) | 1.57 (0.91–2.72) | 5 (11) | 1.02 (0.37–2.82) |
| + | GG | 75 (14) | 68 (12) | 0.93 (0.64–1.35) | 17 (11) | 0.90 (0.50–1.62) | 4 (9) | 0.66 (0.22–1.98) |
| + | GA/AA | 83 (16) | 111 (20) | 1.35 (0.97–1.88) | 39 (24) | 1.83 (1.15–2.89) | 12 (26) | 1.84 (0.88–3.86) |
| HLA-B*08 | ||||||||
| − | GG | 376 (72) | 358 (64) | 1.0 (ref) | 94 (59) | 1.0 (ref) | 29 (63) | 1.0 (ref) |
| − | GA/AA | 47 (9) | 58 (10) | 1.30 (0.86–1.97) | 22 (14) | 1.95 (1.11–3.42) | 5 (11) | 1.25 (0.45–3.45) |
| + | GG | 5 (1) | 15 (3) | 2.97 (1.06–8.32) | 3 (2) | 2.55 (0.59–11.0) | 0 (0) | — |
| + | GA/AA | 94 (18) | 128 (23) | 1.43 (1.05–1.94) | 41 (26) | 1.78 (1.15–2.76) | 12 (26) | 1.72 (0.84–3.55) |
| HLA-DR*03 | ||||||||
| − | GG | 341 (68) | 325 (61) | 1.0 (ref) | 85 (57) | 1.0 (ref) | 23 (52) | 1.0 (ref) |
| − | GA/AA | 60 (12) | 67 (13) | 1.17 (0.80–1.71) | 23 (15) | 1.57 (0.91–2.70) | 5 (11) | 1.15 (0.42–3.17) |
| + | GG | 24 (5) | 27 (5) | 1.17 (0.66–2.08) | 7 (5) | 1.08 (0.44–2.63) | 4 (9) | 2.52 (0.79–8.02) |
| + | GA/AA | 78 (15) | 111 (21) | 1.50 (1.07–2.08) | 34 (23) | 1.80 (1.12–2.89) | 12 (27) | 2.35 (1.11–4.98) |
| HLA-A*01-B*08-DR*03 | ||||||||
| − | GG | 355 (72) | 345 (66) | 1.0 (ref) | 91 (61) | 1.0 (ref) | 27 (61) | 1.0 (ref) |
| − | GA/AA | 71 (14) | 95 (18) | 1.37 (0.97–1.94) | 29 (20) | 1.63 (1.00–2.68) | 9 (20) | 1.60 (0.71–3.57) |
| + | GG | 1 (<1) | 1 (<1) | 0.87 (0.05–14.5) | 0 (0) | — | 0 (0) | — |
| + | GA/AA | 63 (13) | 83 (16) | 1.35 (0.94–1.94) | 28 (19) | 1.80 (1.08–3.00) | 8 (18) | 1.70 (0.73–3.96) |
Abbreviations: CI, confidence intervals; DLBCL, diffuse large B-cell lymphoma; HLA, human leukocyte antigen; NHL, non-Hodgkin lymphoma; OR, odds ratios; TNF, tumor necrosis factor.
Numbers are not equivalent for each allele because of their differential genotyping completion rates. Results with P<0.05 are indicated in bold font.
Interestingly, for NHL risk overall, we observed a statistically significant elevated risk among those with the HLA-B*08 allele, regardless of TNF G-308A status (ORTNF GG = 2.97, 95% CI = 1.06–8.32; ORTNF GA/AA = 1.43, 95% CI = 1.05–1.94) (Table 2). Though the numbers of individuals who carried HLA-B*08 without TNF G-308A were small (n = 15 for NHL; n = 5 in controls), we note that there was no significant association between TNF-308A and NHL in the absence of HLA-B*08 (n = 58 in NHL; n = 47 in controls). These results suggest a potential role for HLA-B*08 as an independent susceptibility allele for overall NHL risk.
Taken together, our results show an association between the TNF-308A variant allele with DLBCL in both the presence and the absence of the HLA-A* 01-B *08-DR*03 alleles and haplotype. Our results suggest that the associations between TNF-308A and DLBCL, now consistently reported and demonstrated in large-scale consortial efforts,1,2 do not appear to be confounded by the association with the HLA-A*01-B* 08-DR*03 or any of its constituents, which are known to be in LD with TNF-308A (D′ = 0.7 in our population), though these associations could still be enhanced by it. We note that of the individuals carrying the HLA-A*01-B*08-DR*03 haplotype, 99% had a variant TNF-308A allele (GA or AA genotype). Notably, 15% of the population had a variant TNF-308A allele in the absence of the HLA-A*01-B*08-DR *03 haplotype, which allowed us to evaluate whether the reported increased DLBCL risk with TNF-308A would also be observed independent of the HLA-A*01-B*08-DR*03 haplotype and alleles. Whether the TNF-308A is the causal DLBCL SNP remains to be confirmed, but our data suggest that the biologic role of the TNF gene and, specifically, the immune mechanisms of inflammation through NF-kB, which is activated by TNF-α, are worth further investigation in understanding DLBCL etiology.
We note that the association between HLA-B*08 and DLBCL was more pronounced than that for the HLA-A*01-B*08-DR *03 haplotype. This appears consistent with the literature that has demonstrated that the AH 8.1 haplotype is largely carried by most individuals who have HLA-B8, and which is known for its unique cleft motif.3 The AH 8.1 is associated with numerous autoimmune conditions that are also NHL risk factors (for example, Sjogren’s syndrome and systemic lupus erythematosis)8 and thought to result from higher TNF-α expression and supported by studies showing elevated TNF-α expression among healthy individuals with AH 8.1.3 It has been hypothesized that TNF may influence the HLA alleles within AH 8.1 by influencing the inflammatory reactions and response to infections.3 Larger studies will be required to determine whether DLBCL risks observed for individuals with both the AH 8.1 (or HLA-B*08 allele) and TNF-308A allele differ from those with only TNF-308A.
Finally, our results also suggest a possible role that HLA may have in NHL etiology. In particular, the association between HLA-B*08 allele was significantly associated with elevated NHL risk overall regardless of TNF G-308A status.
We believe our results shed important light on the role of TNF and HLA in NHL etiology but recognize that to definitively delineate potential effects of HLA from TNF-308A will require replication of our results in an independent study and further evaluation in non-Caucasian populations where linkage disequilibrium patterns differ. For example, another AH (58.1) that includes HLA-A33, -B58, TNF-308A, and -DR3 is found among Asian populations and we cannot exclude the potential influence of HLA-A*33 and -B *58 because of the rarity of these alleles in our Caucasian population.
Study strengths include the population-based design and four-digit HLA genotyping for HLA Class I A, B, and C alleles and for HLA Class II DR alleles. As we did not include HLA-DQ alleles or other non-HLA alleles in the major histocompatibility complex region, we cannot discount their possible involvement in NHL etiology and affect on the reported TNF-308A association. We also had limited power to detect modest independent associations and potential interactions between these loci. We also cannot rule out the contributions of other potentially important immune genes in the major histocompatibility complex class III loci.
In summary, our results suggest that individuals with a variant TNF G-308A allele have increased DLBCL risk even in the absence of HLA-A*01-B*08-DR *03 alleles and haplotype, providing further evidence of the importance of TNF in DLBCL etiology. Our results also indicate that HLA-B*08 is associated with NHL risk regardless of TNF G-308A status, implicating a potential role for HLA in NHL etiology. Further investigation of these results in relation to lymphoma prognosis may also be warranted.9,10
Supplementary Material
Acknowledgments
We thank Peter Hui of the Information Management Services Inc. for programming support. We also gratefully acknowledge the contributions of the staff and scientists at the SEER centers of Iowa, Los Angeles, Detroit and Seattle for the conduct of the study’s field effort. We especially acknowledge the contributions of study site principle investigators Richard Severson (Detroit) and Leslie Bernstein (Los Angeles), without whom the study would not have been possible. The NCI-SEER study was supported by the Intramural Research Program of the NIH (NCI), and by Public Health Service (PHS) contracts N01-PC-65064, N01-PC-67008, N01-PC-67009, N01-PC-67010 and N02-PC-71105.
Footnotes
Conflict of interest
The authors declare no conflict of interest.
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
References
- 1.Rothman N, Skibola CF, Wang SS, Morgan G, Lan Q, Smith MT, et al. Genetic variation in TNF and IL10 and risk of non-Hodgkin lymphoma: a report from the InterLymph Consortium. Lancet Oncol. 2006;7:27–38. doi: 10.1016/S1470-2045(05)70434-4. [DOI] [PubMed] [Google Scholar]
- 2.Skibola CF, Bracci PM, Nieters A, Brooks-Wilson A, de Sanjose S, Hughes AM, et al. Tumor necrosis factor (TNF) and lymphotoxin-alpha (LTA) polymorphisms and risk of non-Hodgkin lymphoma in the InterLymph Consortium. Am J Epidemiology. 2010;171:267–276. doi: 10.1093/aje/kwp383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Candore G, Lio D, Colonna RG, Caruso C. Pathogenesis of autoimmune diseases associated with 8. 1 ancestral haplotype: effect of multiple gene interactions. Autoimmun Rev. 2002;1:29–35. doi: 10.1016/s1568-9972(01)00004-0. [DOI] [PubMed] [Google Scholar]
- 4.Newton JL, Harney SM, Timms AE, Sims AM, Rockett K, Darke C, et al. Dissection of class III major histocompatibility complex haplotypes associated with rheumatoid arthritis. Arthritis Rheum. 2004;50:2122–2129. doi: 10.1002/art.20358. [DOI] [PubMed] [Google Scholar]
- 5.Schotte H, Willeke P, Tidow N, Domschke W, Assmann G, Gaubitz M, et al. Extended haplotype analysis reveals an association of TNF polymorphisms with susceptibility to systemic lupus erythematosus beyond HLA-DR3. Scand J Rheumatol. 2005;34:114–121. doi: 10.1080/03009740510017724. [DOI] [PubMed] [Google Scholar]
- 6.Wang SS, Cerhan JR, Hartge P, Davis S, Cozen W, Severson RK, et al. Common genetic variants in proinflammatory and other immunoregulatory genes and risk for non-Hodgkin lymphoma. Cancer Res. 2006;66:9771–9780. doi: 10.1158/0008-5472.CAN-06-0324. [DOI] [PubMed] [Google Scholar]
- 7.Tilanus MG, Hansen JA, Hurley CK. IHWG Technical Manual: Genomic Analysis of the Human MHC. IHWG; Seattle, WA: 2002. [Google Scholar]
- 8.Shiina T, Hosomichi K, Inoko H, Kulski JK. The HLA genomic loci map: expression, interaction, diversity and disease. J Hum Genet. 2009;54:15–39. doi: 10.1038/jhg.2008.5. [DOI] [PubMed] [Google Scholar]
- 9.Juszczynski P, Kalinka E, Bienvenu J, Woszczek G, Borowiec M, Robak T, et al. Human leukocyte antigens class II and tumor necrosis factor genetic polymorphisms are independent predictors of non-Hodgkin lymphoma outcome. Blood. 2002;100:3037–3040. doi: 10.1182/blood-2002-02-0654. [DOI] [PubMed] [Google Scholar]
- 10.Nowak J, Kalinka-Warzocha E, Juszczynski P, Bilinski P, Mika-Witkowska R, Zajko M, et al. Association of human leukocyte antigen ancestral haplotype 8. 1 with adverse outcome of non-Hodgkin’s lymphoma. Genes Chromosomes Cancer. 2007;46:500–507. doi: 10.1002/gcc.20436. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.