Document G5BQR66ndy3N6XRG0X7b8m44Y

Articles Genetic variation in TNF and IL10 and risk of non-Hodgkin lymphoma: a report from the InterLymph Consortium Nathaniel Rothman, Christine F Skibola, Sophia S Wang, Gareth Morgan, Qing Lan, Martyn T Smith, John J Spinelli, Eleanor Willett, Silvia De Sanjose, Pierluigi Cocco, Sonja I Berndt, Paul Brennan, Angela Brooks-Wilson, Sholom Wacholder, Nikolaus Becker, Patricia Hartge, Tongzhang Zheng, Eve Roman, Elizabeth A Holly, Paolo Boffetta, Bruce Armstrong, Wendy Cozen, Martha Linet, F Xavier Bosch, Maria Grazia Ennas, Theodore R Holford, Richard P Gallagher, Sara Rollinson, Paige M Bracci, James R Cerhan, Denise Whitby, Patrick S Moore, Brian Leaderer, Agnes Lai, Charlotte Spink, Scott Davis, Ramon Bosch, Aldo Scarpa, Yawei Zhang, Richard K Severson, Meredith Yeager, Stephen Chanock, Alexandra Nieters Summary Background Common genetic variants in immune and inflammatory response genes can affect the risk of developing non-Hodgkin lymphoma. We aimed to test this hypothesis using previously unpublished data from eight European, Canadian, and US case-control studies of the International Lymphoma Epidemiology Consortium (InterLymph). Methods We selected 12 single-nucleotide polymorphisms for analysis, on the basis of previous functional or association data, in nine genes that have important roles in lymphoid development, Th1/Th2 balance, and proinflammatory or anti-inflammatory pathways (IL1A, IL1RN, IL1B, IL2, IL6, IL10, TNF, LTA, and CARD15). Genotype data for one or more single-nucleotide polymorphisms were available for 3586 cases of non-Hodgkin lymphoma and for 4018 controls, and were assessed in a pooled analysis by use of a random-effects logistic regression model. Findings The tumour necrosis factor (TNF) 308GA polymorphism was associated with increased risk of nonHodgkin lymphoma (p for trend=0005), particularly for diffuse large B-cell lymphoma, the main histological subtype (odds ratio 129 [95% CI 110151] for GA and 165 [116234] for AA, p for trend 00001), but not for follicular lymphoma. The interleukin 10 (IL10) 3575TA polymorphism was also associated with increased risk of non-Hodgkin lymphoma (p for trend=002), again particularly for diffuse large B-cell lymphoma (p for trend=0006). For individuals homozygous for the TNF 308A allele and carrying at least one IL10 3575A allele, risk of diffuse large B-cell lymphoma doubled (213 [137332], p=000083). Interpretation Common polymorphisms in TNF and IL10, key cytokines for the inflammatory response and Th1/Th2 balance, could be susceptibility loci for non-Hodgkin lymphoma. Moreover, our results underscore the importance of consortia for investigating the genetic basis of chronic diseases like cancer. Introduction The mechanisms underlying differences in immune response between individuals are complex and include inherited genetic variation and cumulative antigenic exposure to infectious and other environmental challenges that give rise to immunological memory. Common variations in genes of the immune system have evolved through selective pressure to ensure hostpathogen coexistence. However, variants selected to protect against infection could inadvertently lead to a greater risk of other diseases that are less susceptible to selection. Such variants might be expected to predispose to chronic inflammatory disease and malignant diseases of the lymphoid system. Lymphoid development and differentiation and T-helper (Th)1/Th2 balance (ie, cellular vs humoral immunity) are regulated in part by key cytokines including interleukin (IL)1, IL2, IL6, IL10, tumour necrosis factor (TNF), and lymphotoxin (LT).17 Furthermore, deregulated concentrations of several cytokines (eg, IL6, IL10, and TNF) have been detected in patients with lymphoma and were associated with an adverse prognosis.810 Evidence that genetic susceptibility plays a part in lymphomagenesis is provided by strong and consistent findings from registry and populationbased epidemiological studies that show an increased risk of non-Hodgkin lymphoma in individuals with a family history of this or other haemopoietic malignant diseases.11,12 Here, we tested the hypothesis that single-nucleotide polymorphisms in nine candidate genes that have important roles in lymphoid development, proinflammatory or anti-inflammatory pathways, and Th1/Th2 balance7 are associated with risk of non-Hodgkin lymphoma. Methods Study characteristics The International Lymphoma Epidemiology Consortium (InterLymph, http://epi.grants.cancer.gov/InterLymph) is a voluntary consortium established in 2000 to facilitate collaboration between epidemiological studies of Lancet Oncol 2005; 7: 2738 Published online November 29, 2005 DOI:10.1016/S1470-2045(05) 70434-4 See Reflection and Reaction page 3 Division of Cancer Epidemiology and Genetics (N Rothman MD, S S Wang PhD, Q Lan MD, S I Berndt PharmD, S Wacholder PhD, P Hartge ScD, M Linet MD, M Yeager PhD, S Chanock MD), Core Genotyping Facility, Advanced Technology Center (M Yeager, S Chanock), and Pediatric Oncology Branch, Center for Cancer Research (S Chanock), National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, USA; Division of Environmental Health Sciences, School of Public Health, University of California, Berkeley, CA, USA (C F Skibola PhD, Prof M T Smith PhD); Institute of Cancer Research, The Royal Marsden, London, UK (Prof G Morgan PhD); Cancer Control Research Program (J J Spinelli PhD, R P Gallagher MA, A Lai MSc) and Genome Sciences Centre (A Brooks-Wilson PhD), British Columbia Cancer Agency, Vancouver, British Columbia, Canada; Epidemiology and Genetics Unit, Department of Health Sciences, University of York, York, UK (E Willett PhD, Prof E Roman PhD); Epidemiology and Cancer Registry Unit, Catalan Institute of Oncology, Barcelona, Spain (S De Sanjose MD, F X Bosch MD); Department of Public Health, Occupational Health Section (P L Cocco MD) and Department of Cytomorphology (M G Ennas PhD), University of Cagliari, Cagliari, Italy; Unit of Environmental Cancer Epidemiology, International http://oncology.thelancet.com Vol 7 January 2006 27 Articles Agency for Research on Cancer, Lyon, France (P Brennan PhD, P Boffetta MD); Division of Clinical Epidemiology, German Cancer Research Centre, Heidelberg, Germany (Prof N Becker PhD, A Nieters PhD); Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, CT, USA (Prof T Zheng ScD, Prof T R Holford PhD, Prof B Leaderer PhD, Y Zhang PhD); Department of Epidemiology and Biostatistics, School of Medicine, University of California San Francisco, San Francisco, CA, USA (Prof E A Holly PhD, P M Bracci MS); School of Public Health, University of Sydney, Sydney, Australia (Prof B Armstrong PhD); Department of Preventive Medicine, University of Southern California Keck School of Medicine, Los Angeles, CA, USA (W Cozen DO); School of Medicine, University of Leeds, Leeds, UK (S Rollinson PhD); Department of Health Sciences Research, Mayo Clinic College of Medicine, Rochester, MN, USA (Prof J R Cerhan PhD); Viral Epidemiology Section, AIDS Vaccine Program (D Whitby PhD) and Intramural Research Support Program (M Yeager) Science Applications International Corporation-Frederick, National Cancer Institute-Frederick, Department of Health and Human Services, Frederick, MD, USA; Department of Pathology, University of Verona, Verona, Italy (P S Moore PhD, Prof A Scarpa MD); Department of Pathology and Microbiology, School of Medical Sciences, University of Bristol, Bristol, UK (C Spink PhD); Fred Hutchinson Cancer Research Center and Department of Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA, USA (Prof S Davis PhD); Department of Pathology, Hospital Verge de la Cinta, Tortosa, Spain (R Bosch MD); and Department of Family Medicine and Karmanos Cancer Institute, Wayne State University, Detroit, MI, USA (Prof R K Severson PhD) Correspondence to: Dr Nathaniel Rothman, Division of Cancer Epidemiology and Genetics, EPS8116, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD 20892, USA rothmann@mail.nih.gov EPILYMPH--Italy* EPILYMPH--Spain15 University of California San Francisco16 EPILYMPH--Germany17 Connecticut18 UK19|| NCI-SEER12** British Columbia Number with genotype data Cases Controls 144 113 354 569 309 685 Specific study characteristics Both sexes, all subtypes Both sexes, all subtypes Both sexes, all subtypes 482 481 497 561 461 461 963 747 376 401 Both sexes, all subtypes Women only, all subtypes Both sexes, DLBCL and follicular lymphoma only Both sexes, all subtypes Both sexes, all subtypes DLBCL=diffuse large B-cell lymphoma. *Cagliari, Nuoro, and Oristano, Italy. Barcelona, Tortosa, Reus, and Madrid, Spain. Santa Clara, San Mateo, San Francisco, Marin, Contra Costa, and Alameda counties, San Francisco Bay Area, CA, USA. Ludwigshafen/Upper Palatinate, Heidelberg/Rhine-Neckar County, Wrzburg/Lower Frankonia, Hamburg, Bielefeld, and Munich, Germany. Connecticut, USA. ||Counties of North, East, and West Yorkshire; Lancashire, district of South Lakeland; Caradon district of Cornwall, South Devon, Dorset, and South Hampshire, UK. **Detroit, Iowa, Los Angeles, Seattle, USA. Greater Victoria and Vancouver, Canada. Table 1: Description of studies participating in InterLymph genotyping project lymphoma worldwide.13,14 It was formed to coordinate selected analyses across similarly designed studies of lymphoma; to increase statistical power to detect associations, especially for histological subtypes of nonHodgkin lymphoma that might have differing causes; and to provide protection against false-negative and false-positive findings. In this study, we analysed data from the eight studies in InterLymph who were willing and able to participate in the genotyping project. Detailed information on participant recruitment and pathology review has been published for six of the eight studies.12,1519 The EPILYMPH--Italy study enrolled controls from a random sample of the general population by use of population lists, and the British Columbia study enrolled controls from a random sample of the population by use of Provincial Health Insurance records. Both studies used WHO classification for non-Hodgkin lymphoma.20 All studies were population-based, with the exception of the EPILYMPH--Spain study, which was hospital-based. We excluded patients with non-Hodgkin lymphoma who were HIV-positive, and included only white participants, almost all of whom were of European descent, to keep population homogeneity to a maximum. All studies provided details of age and sex and indicated whether a case was diagnosed with diffuse large B-cell lymphoma, follicular lymphoma, or other histology. The InterLymph genotype working group decided a priori not to investigate genotype associations for other, less common histological subtypes, because the statistical power would have been very limited, even in a pooled analysis of this size. In addition, the pathological diagnosis of diffuse large B-cell lymphoma and follicular lymphoma has been stable and comparable in Canada and the USA and in Europe, allowing data to be pooled for these subgroups across studies, even though pathology samples were not reviewed centrally. Although some studies did not divide diffuse large-cell lymphomas into B and T subtypes, we use the term diffuse large B-cell lymphoma throughout because almost all diffuse largecell tumours derive from B cells. All studies were approved by their local ethics review committee, and written informed consent was obtained from all participants. All eight studies provided genotype data for all cases of non-Hodgkin lymphoma enrolled in their study, with the exception of the UK study, for which only cases diagnosed with diffuse large B-cell lymphoma and follicular lymphoma and their individually matched controls had been genotyped; as a consequence, the UK data were used only in histology-specific analyses. Laboratory analysis We chose 12 single-nucleotide polymorphisms (minor allele frequency range 002044), each of which could be functionally important, in nine genes for coordinated genotyping and analysis: in location 2q14, IL1A 889CT (rs1800587), IL1B 511CT (rs16944), and IL1B 31CT (rs1143627); in 2q14.2, IL1RN 9589AT (rs454078); in 4q2627, IL2 384TG (rs2069762); in 7p21, IL6 174GC (rs1800795) and IL6 597GA (rs1800797); in 1q3132, IL10 1082AG (rs1800896) and IL10 3575TA (rs1800890); in 6p21.3, TNF 308GA (rs1800629) and LTA 252AG (rs909253); and in 16q21, CARD15 Ex1135C (rs2066847). DNA samples were analysed at one of six laboratories. Five laboratories used the TaqmanTM platform (Applied Biosystems, Foster City, CA, USA) exclusively or mainly for genotyping, and one laboratory, which analysed samples for the EPILYMPH--Germany study, used PyrosequencingTM or allele-specific PCR. Sequence data and assay conditions for TaqmanTM assays are available on the NCI SNP500 website http://snp500cancer. nci.nih.gov. To ensure that genotyping results were consistent across studies, every laboratory analysed the same set of DNA samples from 102 ethnically diverse individuals that had previously been sequenced and genotyped on one or more platforms as part of the SNP500Cancer project.21 All laboratories completed genotype analysis before a comparison with the publicly available genotypes on the NCI SNP500 website. We assessed concordance across laboratories and rechecked quality control data for assays not in Hardy-Weinberg equilibrium at p005 to confirm accuracy. Statistical analysis To investigate the association between the singlenucleotide polymorphisms and risk of non-Hodgkin lymphoma, risk estimates were estimated with a random-effects logistic regression model that adjusted for age (50, 5059, 6069, 70 years), sex, and study centre. An exact test was used to calculate risk estimates 28 http://oncology.thelancet.com Vol 7 January 2006 Articles Controls Cases Odds ratio (95% CI) TNF 308GA GG 2312 (74%) 1927 (71%) 1 (ref) GA 719 (23%) 705 (26%) 118 (104133) AA 87 (3%) 86 (3%) 125 (091170) GA or AA 806 (26%) 791 (29%) 119 (105133) Trend 3118 (100%) 2718 (100%) 116 (104128) LTA 252AG AA 1699 (48%) 1465 (47%) 1 (ref) AG 1484 (42%) 1281 (42%) 100 (086116) GG 326 (9%) 339 (11%) 118 (096144) AG or GG 1810 (52%) 1620 (53%) 101 (088116) Trend 3509 (100%) 3085 (100%) 105 (095115) IL10 3575TA TT 1419 (41%) 1172 (39%) 1 (ref) TA 1604 (46%) 1423 (47%) 110 (098122) AA 439 (13%) 435 (14%) 119 (100141) TA or AA 2043 (59%) 1858 (61%) 111 (101123) Trend 3462 (100%) 3030 (100%) 109 (101117) IL10 1082AG AA 972 (31%) 804 (30%) 1 (ref) AG 1513 (49%) 1326 (49%) 108 (095122) GG 623 (20%) 580 (21%) 113 (096132) AG or GG 2136 (69%) 1906 (70%) 109 (097122) Trend 3108 (100%) 2710 (100%) 106 (099114) IL1A 889CT CC 1740 (50%) 1494 (49%) 1 (ref) CT 1436 (42%) 1306 (43%) 106 (096118) TT 281 (8%) 253 (8%) 107 (089129) CT or TT 1717 (50%) 1559 (51%) 106 (096117) Trend 3457 (100%) 3053 (100%) 105 (097113) IL1B 511 CT CC 1559 (45%) 1371 (45%) 1 (ref) CT 1566 (45%) 1338 (44%) 099 (089110) TT 365 (10%) 358 (12%) 111 (092135) CT or TT 1931 (55%) 1696 (55%) 101 (092112) Trend 3490 (100%) 3067 (100%) 103 (096111) IL1B 31CT TT 1520 (44%) 1362 (45%) 1 (ref) CT 1569 (45%) 1304 (43%) 095 (085105) CC 379 (11%) 364 (12%) 107 (090128) CT or CC 1948 (56%) 1668 (55%) 097 (088107) Trend 3468 (100%) 3030 (100%) 101 (093108) IL1RN 9589AT AA 1870 (54%) 1558 (52%) 1 (ref) AT 1345 (39%) 1230 (41%) 110 (099122) TT 254 (7%) 232 (8%) 113 (093137) AT or TT 1599 (46%) 1462 (48%) 110 (100122) Trend 3469 (100%) 3020 (100%) 108 (100117) IL2 384TG TT 1755 (51%) 1491 (49%) 1 (ref) TG 1389 (40%) 1281 (42%) 109 (097123) GG 310 (9%) 267 (9%) 099 (079125) TG or GG 1699 (49%) 1548 (51%) 107 (097118) Trend 3454 (100%) 3039 (100%) 104 (096112) IL6 174GC GG 1277 (36%) 1097 (36%) 1 (ref) GC 1658 (47%) 1470 (48%) 102 (092114) CC 564 (16%) 499 (16%) 101 (087118) GC or CC 2222 (64%) 1969 (64%) 102 (092113) Trend 3499 (100%) 3066 (100%) 101 (094108) IL6 597GA GG 1151 (38%) 998 (38%) 1 (ref) GA 1423 (46%) 1243 (47%) 101 (090113) AA 494 (16%) 417 (16%) 095 (081112) GA or AA 1917 (62%) 1660 (62%) 099 (089111) Trend 3068 (100%) 2658 (100%) 098 (091106) p p for heterogeneity NA 0009 016 0005 0005 NA NA 047 036 048 NA NA 100 NA 011 0014 089 0011 026 0030 NA 0098 0044 0037 002 NA NA 065 082 043 NA NA 023 NA 013 075 013 086 011 055 NA NA 026 NA 047 067 023 061 025 042 NA NA 086 NA 028 014 079 0073 039 0035 NA NA 032 NA 043 0096 058 0044 086 0032 NA 0086 021 0053 0062 NA NA 04 022 055 NA NA 014 NA 096 0073 016 034 031 040 NA NA 066 NA 086 054 068 047 078 033 NA NA 092 NA 056 041 089 023 065 017 (continues) (continued) Controls Cases Odds ratio (95% CI) p CARD15 Ex1135C 3347 (96%) 2926 (95%) 1 (ref) 149 (4%) 141 (5%) 108 (085137) 1 (1%) 2 (1%) 229 (012135) or 150 (4%) 143 (5%) 109 (086139) Trend 3497 (100%) 3069 (100%) 110 (087140) NA 054 060 047 041 p for heterogeneity NA NA 031 027 025 Data are number of individuals (%) unless otherwise stated. NA=not applicable. *Includes up to seven studies that enrolled all histological types of non-Hodgkin lymphoma with genotype data for specific single-nucleotide polymorphisms. Test for heterogeneity for codominant model. Test for heterogeneity for dominant model. Test for heterogeneity for additive model (ie, trend). Table 2: Pooled genotype frequencies and risks for all histologies by single-nucleotide polymorphism* and p values for homozygous carriers of the CARD15 Ex1135C variant, and for homozygous carriers of TNF 308GA for diffuse large B-cell lymphoma in the National Cancer Institute-Surveillance Epidemiology and End Results (NCI-SEER) Detroit centre, because there was only one homozygous case or control in these analyses. Heterogeneity across studies was assessed by comparison of the logistic-regression model with and without the cross-product terms of the genotypes and study centre by use of a likelihood-ratio test. Heterogeneity between subtypes of non-Hodgkin lymphoma was assessed by comparing them directly in a logisticregression model and testing for differences in the genotype association. The test for trend was assessed with an additive model--that is, with a single variable for genotype coded as the number of variant alleles, in the logistic-regression model. All genotype analyses were done with STATA version 8.2. We assessed the robustness of the findings by calculating the false-discovery rate,22 defined as the expected ratio of erroneous rejections of the null hypothesis to the total number of rejected hypotheses, which yields a p value corrected for multiple comparisons, and by application of the false-positive report probability method.23 Before analysis, investigators were asked to provide a range of prior probabilities of association with non-Hodgkin lymphoma for every single-nucleotide polymorphism, based on their interpretation of all sources of information; prior probability values for TNF 308GA and IL10 3575TA varied from 0001 (ie, that a given single-nucleotide polymorphism has a one in one thousand chance of being truly associated with non-Hodgkin lymphoma) to 01. We divided these values by two and applied them to the two histology-specific results presented here, using the observed risk estimates from the additive model. A false-positive report probability value rejection criterion of 02 was used to designate findings as noteworthy.23 http://oncology.thelancet.com Vol 7 January 2006 29 Articles Controls Cases Diffuse large B-cell lymphoma n (%) Odds ratio (95% CI) TNF 308GA GG 2597 (73%) 716 (66%) GA 854 (24%) 312 (29%) AA 113 (3%) 53 (5%) GA or AA 967 (27%) 365 (34%) Trend 3564 (100%) 1081 (100%) LTA 252AG AA 1876 (47%) 519 (44%) AG 1701 (43%) 491 (42%) GG 380 (10%) 159 (14%) AG or GG 2081 (53%) 650 (56%) Trend 3957 (100%) 1169 (100%) IL10 3575TA TT 1593 (41%) 422 (36%) TA 1816 (46%) 567 (48%) AA 512 (13%) 180 (15%) TA or AA 2328 (59%) 747 (64%) Trend 3921 (100%) 1169 (100%) IL10 1082AG AA 1089 (31%) 294 (27%) AG 1734 (49%) 537 (50%) GG 742 (21%) 253 (23%) AG or GG 2476 (69%) 790 (73%) Trend 3565 (100%) 1084 (100%) TNF 308GA and IL10 3575TA GG/TT 1077 (30%) 258 (24%) GG/TA 1180 (33%) 342 (32%) GA/TT 339 (10%) 115 (11%) GA/TA 390 (11%) 143 (13%) GG/AA 322 (9%) 109 (10%) AA/TT 46 (1%) 16 (2%) GA/AA 120 (3%) 51 (5%) AA/TA or AA 66 (2%) 36 (3%) IL1A 889CT CC 1962 (50%) 601 (51%) CT 1631 (42%) 485 (41%) TT 319 (8%) 84 (7%) CT or TT 1950 (50%) 569 (49%) Trend 3912 (100%) 1170 (100%) IL1B 511CT CC 1744 (44%) 517 (45%) CT 1773 (45%) 513 (44%) TT 426 (11%) 131 (11%) CT or TT 2199 (56%) 644 (55%) Trend 3943 (100%) 1161 (100%) IL1B 31CT TT 1707 (44%) 517 (45%) CT 1778 (45%) 508 (44%) CC 437 (11%) 135 (12%) CT or CC 2215 (56%) 643 (55%) Trend 3922 (100%) 1160 (100%) IL1RN 9589AT AA 1870 (54%) 474 (53%) AT 1345 (39%) 354 (40%) TT 254 (7%) 63 (7%) AT or TT 1599 (46%) 417 (47%) Trend 3469 (100%) 891 (100%) IL2 384TG TT 1977 (51%) 605 (52%) TG 1585 (40%) 459 (39%) GG 349 (9%) 102 (9%) TG or GG 1934 (49%) 561 (48%) Trend 3911 (100%) 1166 (100%) 1 (ref) 129 (110151) 165 (116234) 133 (114155) 129 (114146) 1 (ref) 106 (088129) 147 (118184) 113 (096133) 116 (104129) 1 (ref) 120 (104 139) 128 (104157) 122 (106140) 115 (104126) 1 (ref) 114 (097136) 123 (100152) 117 (100137) 111 (100122) 1 (ref) 126 (104152) 144 (111186) 151 (118192) 139 (107181) 157 (086286) 169 (117243) 213 (137332) 1 (ref) 097 (084111) 085 (065111) 095 (083108) 094 (085105) 1 (ref) 099 (086115) 101 (081126) 100 (087114) 100 (091111) 1 (ref) 096 (084111) 101 (081126) 097 (085111) 099 (090110) 1 (ref) 107 (091126) 100 (072139) 106 (091123) 103 (092116) 1 (ref) 093 (079109) 092 (068123) 092 (081106) 095 (086105) Follicular lymphoma p n (%) Odds ratio (95% CI) NA 576 (71%) 0002 209 (26%) 0006 25 (3%) 000021 234 (29%) 00001 810 (100%) NA 052 0001 014 0007 424 (48%) 371 (42%) 97 (11%) 468 (52%) 892 (100%) NA 0015 002 0006 0006 323 (37%) 418 (47%) 142 (16%) 560 (63%) 883 (100%) NA 012 0053 0048 0043 227 (28%) 388 (48%) 194 (24%) 582 (72%) 809 (100%) NA 223 (28%) 0016 258 (32%) 0006 64 (8%) 000094 115 (14%) 0015 91 (11%) 014 6 (1%) 00047 29 (4%) 000083 19 (2%) NA 435 (49%) 062 372 (42%) 023 80 (9%) 042 452 (51%) 027 887 (100%) NA 396 (44%) 094 400 (45%) 093 98 (11%) 098 498 (56%) 097 894 (100%) NA 396 (45%) 062 385 (44%) 092 99 (11%) 070 484 (55%) 088 880 (100%) NA 350 (51%) 043 285 (41%) 100 52 (8%) 048 337 (49%) 059 687 (100%) NA 431 (48%) 035 383 (43%) 055 75 (8%) 024 458 (52%) 034 889 (100%) 1(ref) 103 (086123) 092 (055154) 101 (085121) 100 (086116) 1 (ref) 091 (078107) 104 (080133) 093 (080108) 098 (087110) 1 (ref) 110 (093129) 124 (099155) 113 (097132) 111 (100124) 1 (ref) 102 (085123) 112 (090140) 105 (088125) 106 (095118) 1 (ref) 103 (085127) 087 (064119) 127 (098165) 124 (094165) 060 (025145) 101 (065156) 118 (069204) 1 (ref) 099 (084116) 100 (068148) 100 (086116) 102 (090114) 1 (ref) 101 (086118) 102 (080132) 101 (087118) 101 (090113) 1 (ref) 095 (081112) 099 (077127) 096 (082112) 098 (088110) 1 (ref) 114 (096136) 119 (085167) 115 (097136) 111 (098127) 1 (ref) 106 (091124) 100 (076131) 105 (091122) 102 (091115) p for difference between histological types p NA .. 078 .. 074 .. 088 .. 098 00037 NA .. 025 .. 078 .. 037 .. 071 0015 NA 028 0066 013 0059 .. .. .. .. 062 NA .. 083 .. 030 .. 057 .. 031 049 NA .. 074 .. 039 .. 007 .. 013 .. 025 .. 098 .. 055 .. NA .. 088 .. 100 .. 100 .. 079 018 NA .. 089 .. 086 .. 086 .. 084 100 NA .. 055 .. 093 .. 059 .. 073 083 NA .. 014 .. 032 .. 010 .. 011 043 NA .. 044 .. 098 .. 051 .. 068 037 (continues) 30 http://oncology.thelancet.com Vol 7 January 2006 Articles (continued) Controls IL6 174GC GG 1427 (36%) GC 1858 (47%) CC 664 (17%) GC or CC 2522 (64%) Trend 3949 (100%) IL6 597GA GG 1151 (38%) GA 1423 (46%) AA 494 (16%) GA or AA 1917 (62%) Trend 3068 (100%) CARD15 Ex1135C 3770 (96%) 161 (4%) 1 (1%) or 162 (4%) Trend 3932 (100%) Cases Diffuse large B-cell lymphoma n (%) Odds ratio (95% CI) 419 (36%) 527 (45%) 217 (19%) 744 (64%) 1163 (100%) 1 (ref) 097 (083112) 108 (089131) 100 (087115) 103 (093113) 300 (37%) 386 (47%) 127 (16%) 513 (63%) 813 (100%) 1 (ref) 104 (088124) 099 (078125) 103 (087121) 100 (090112) 1096 (95%) 49 (4%) 3 (1%) 52 (5%) 1148 (100%) 1 (ref) 101 (073142) 1032 (083542) 108 (078149) 113 (083155) p NA 066 045 096 059 NA 064 092 074 095 NA 093 0038 066 043 Follicular lymphoma n (%) Odds ratio (95% CI) 313 (35%) 417 (47%) 163 (18%) 580 (65%) 893 (100%) 232 (39%) 270 (45%) 96 (16%) 366 (61%) 598 (100%) 834 (95%) 43 (5%) 2 (1%) 45 (5%) 879 (100%) 1 (ref) 097 (082115) 101 (081125) 098 (084115) 100 (090111) 1 (ref) 089 (073109) 088 (068116) 089 (074107) 093 (082106) 1 (ref) 119 (070202) 904 (047533) 126 (083194) 129 (082202) NA=not applicable. *Includes data for up to eight studies with data for specific single-nucleotide polymorphisms. Table 3: Pooled genotype frequencies and risks for diffuse large B-cell lymphoma and follicular lymphoma* p for difference between p histological types NA .. 073 .. 095 .. 081 .. 097 087 NA .. 025 .. 037 .. 022 .. 027 036 NA 053 0087 028 027 .. .. .. .. 078 Haplotypes were estimated from single-nucleotide polymorphisms within the same chromosomal region using the expectation-maximisation algorithm.24 Measures of linkage disequilibrium, D and r2, were assessed with Haploview.25 Overall differences in the haplotype distribution between cases and controls were assessed with a global score test,26 which was adjusted for age, sex, and study centre. The effects of individual haplotypes were estimated from the additive model by fitting of a logistic-regression model and use of the estimated probabilities of the haplotypes as weights to update the regression coefficients in an iterative manner.26 All haplotype analyses were done with the statistical package Haplo Stats version 1.1.0. Fixed-effects pooled analyses for haplotype associations are reported, although similar results were obtained when haplotypes were estimated for each individual study and then combined in a meta-analysis with random-effects models (not shown). Role of the funding source The study sponsors had no role in study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication. Results Table 1 shows brief details of the eight case-control studies of the InterLymph consortium that participated in this project. The median response rate for participants who were both interviewed and provided a source of genomic DNA (blood, or in some studies, buccal cells) was 712% (range 454846) for cases and 496% (276639) for controls. Sensitivity analyses showed that these results were unchanged after exclusion of the study with the lowest response rate for either cases or controls (not shown). The study populations included only adults; the mean age of cases was 587 years (SD 129) and of controls was 581 years (140). All studies provided data for all single-nucleotide polymorphisms at the time analysis began, except the UK study, which did not have data for IL1RN 9589AT or IL6 597GA, and the British Columbia study, which did not have data for IL6 597GA, IL10 1082AG, or TNF 308GA. IL2 384TG and TNF 308GA in the EPILYMPH--Spain study and IL1B 511CT and TNF 308GA in the University of California San Francisco study were not consistent with HardyWeinberg equilibrium at 001p0001. IL1B 31CT in the University of California San Francisco study; IL1RN 9589AT, IL10 1082AG, and LTA 252AG in the EPILYMPH--Germany study; IL1RN 9589AT in the Connecticut study; and IL6 174GC in the UK study were not consistent with HardyWeinberg equilibrium at 005p001. Exclusion of studies with a single-nucleotide polymorphism out of Hardy-Weinberg equilibrium had a minimum effect on risk estimates for the polymorphisms or for haplotypes containing the polymorphism (not shown). Table 2 shows results from the analyses of singlenucleotide polymorphisms for all cases of nonHodgkin lymphoma, and table 3 for diffuse large B-cell lymphoma and follicular lymphoma separately. TNF 308GA was associated with increased risk of Linkage disequilibrium The non-random association of two or more genetic markers on the same chromosome, usually in close proximity, that tend to be inherited together either more or less frequently in any given population than would be expected from the distance between them. Haplo Stats http://mayoresearch.mayo.edu/ mayo/research/biostat/schaid.cfm http://oncology.thelancet.com Vol 7 January 2006 31 Articles TNF 308GA Controls EPILYMPH--Italy GG GA AA GA or AA Trend 100 (88%) 13 (12%) 0 13 (12%) 113 (100%) EPILYMPH--Spain* GG GA AA GA or AA Trend 434 (79%) 103 (19%) 15 (3%) 118 (21%) 552 (100%) University of California San Francisco GG 487 (72%) GA 160 (24%) AA 26 (4%) GA or AA 186 (28%) Trend 673 (100%) EPILYMPH--Germany GG GA AA GA or AA Trend 338 (71%) 130 (27%) 11 (2%) 141 (29%) 479 (100%) Connecticut GG GA AA GA or AA Trend 402 (72%) 139 (25%) 18 (3%) 157 (28%) 559 (100%) UK GG GA AA GA or AA Trend 285 (64%) 135 (30%) 26 (6%) 161 (36%) 446 (100%) NCI--SEER GG GA AA GA or AA Trend 551 (74%) 174 (23%) 17 (2%) 191 (26%) 742 (100%) Detroit GG GA AA GA or AA Trend 83 (73%) 29 (25%) 2 (2%) 31 (27%) 114 (100%) Iowa GG GA AA GA or AA Trend 189 (73%) 65 (25%) 5 (2%) 70 (27%) 259 (100%) Los Angeles GG GA AA GA or AA Trend 97 (80%) 25 (20%) 0 25 (20%) 122 (100%) Seattle GG GA AA GA or AA Trend 182 (74%) 55 (22%) 10 (4%) 65 (26%) 247 (100%) Cases 54 (90%) 5 (8%) 1 (2%) 6 (10%) 60 (100%) 55 (72%) 16 (21%) 5 (7%) 21 (28%) 76 (100%) 61 (62%) 32 (33%) 5 (5%) 37 (38%) 98 (100%) 87 (67%) 37 (29%) 5 (4%) 42 (33%) 129 (100%) 103 (66%) 49 (32%) 3 (2%) 52 (34%) 155 (100%) 154 (61%) 82 (32%) 17 (7%) 99 (39%) 253 (100%) 202 (65%) 91 (29%) 17 (5%) 108 (35%) 310 (100%) 45 (75%) 14 (23%) 1 (2%) 15 (25%) 60 (100%) 68 (64%) 32 (30%) 6 (6%) 38 (36%) 106 (100%) 39 (67%) 18 (31%) 1 (2%) 19 (33%) 58 (100%) 50 (58%) 27 (31%) 9 (10%) 36 (42%) 86 (100%) Odds ratio (95% CI) 10 (ref) 075 (025227) NA 092 (032261) 110 (043285) 10 (ref) 115 (063211) 265 (092762) 134 (077231) 139 (090214) 10 (ref) 155 (097248) 157 (057430) 155 (099243) 140 (097200) 10 (ref) 110 (071171) 177 (059528) 116 (076176) 118 (082170) 10 (ref) 136 (091202) 065 (019227) 128 (087187) 115 (082160) 10 (ref) 112 (080157) 122 (064233) 114 (083157) 111 (087143) 10 (ref) 148 (109201) 283 (141570) 160 (120214) 156 (122199) 10 (ref) 086 (038192) 092 (0021818) 084 (038182) 084 (042168) 10 (ref) 140 (084234) 342 (1001172) 155 (095252) 156 (103237) 10 (ref) 189 (092391) NA 199 (097409) 206 (103413) 10 (ref) 191 (108337) 357 (135943) 216 (128366) 190 (127284) p NA 061 NA 088 084 NA 064 007 030 014 NA 0069 038 0055 007 NA 066 031 050 037 NA 013 050 022 042 NA 050 054 042 040 NA 0012 00035 00015 00003 NA 071 100 065 062 NA 019 005 0079 003 NA 0085 NA 006 004 NA 0026 001 00042 00019 IL10 3575TA Genotype Controls Cases TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend TT TA AA TA or AA Trend 73 (65%) 37 (33%) 2 (2%) 39 (35%) 112 (100%) 34 (58%) 22 (37%) 3 (5%) 25 (42%) 59 (100%) 271 (49%) 233 (42%) 50 (9%) 283 (51%) 554 (100%) 34 (44%) 39 (51%) 4 (5%) 43 (56%) 77 (100%) 238 (35%) 343 (51%) 96 (14%) 439 (65%) 677 (100%) 35 (38%) 41 (44%) 17 (18%) 58 (62%) 93 (100%) 192 (40%) 215 (45%) 71 (15%) 286 (60%) 478 (100%) 47 (36%) 64 (50%) 18 (14%) 82 (64%) 129 (100%) 240 (43%) 268 (48%) 53 (9%) 321 (57%) 561 (100%) 55 (35%) 74 (48%) 26 (17%) 100 (65%) 155 (100%) 174 (38%) 212 (46%) 73 (16%) 285 (62%) 459 (100%) 82 (31%) 128 (49%) 52 (20%) 180 (69%) 262 (100%) 286 (39%) 329 (45%) 116 (16%) 445 (61%) 731 (100%) 107 (35%) 151 (49%) 51 (16%) 202 (65%) 309 (100%) 46 (41%) 49 (44%) 16 (14%) 65 (59%) 111 (100%) 21 (36%) 28 (48%) 9 (16%) 37 (64%) 58 (100%) 82 (32%) 128 (50%) 46 (18%) 174 (68%) 256 (100%) 36 (34%) 50 (47%) 20 (19%) 70 (66%) 106 (100%) 56 (46%) 48 (40%) 17 (14%) 65 (54%) 121 (100%) 22 (38%) 29 (50%) 7 (12%) 36 (62%) 58 (100%) 102 (42%) 104 (43%) 37 (15%) 141 (58%) 243 (100%) 28 (32%) 44 (51%) 15 (17%) 59 (68%) 87 (100%) Odds ratio (95% CI) 10 (ref) 126 (064248) 303 (0471955) 135 (070261) 140 (079250) 10 (ref) 140 (085231) 068 (023200) 128 (079207) 106 (073154) 10 (ref) 077 (047125) 116 (062220) 085 (054134) 101 (073140) 10 (ref) 125 (081192) 112 (060207) 121 (081183) 110 (083146) 10 (ref) 120 (081178) 212 (122370) 136 (094197) 139 (106182) 10 (ref) 129 (092182) 151 (097235) 135 (097186) 124 (100154) 10 (ref) 128 (095173) 120 (081180) 126 (095167) 113 (093137) 10 (ref) 120 (057254) 097 (034278) 114 (056231) 103 (063169) 10 (ref) 099 (058167) 106 (055207) 101 (062165) 102 (074142) 10 (ref) 175 (087351) 110 (039307) 157 (081302) 120 (076189) 10 (ref) 152 (088264) 144 (069300) 150 (089252) 125 (088177) p NA 051 024 037 025 NA 018 048 032 076 NA 028 064 049 095 NA 032 073 035 052 NA 035 00081 011 0016 NA 014 0069 0071 0054 NA 010 036 010 021 NA 063 095 072 091 NA 096 086 097 088 NA 011 086 018 044 NA 014 033 013 021 (continues) 32 http://oncology.thelancet.com Vol 7 January 2006 Articles (continued) British Columbia GG GA AA GA or AA Trend TNF 308GA Controls .. .. .. .. .. Cases .. .. .. .. .. Odds ratio (95% CI) p .. .. .. .. .. .. .. .. .. .. IL10 3575TA Genotype Controls TT TA AA TA or AA Trend 119 (34%) 179 (51%) 51 (15%) 230 (66%) 349 Cases 28 (33%) 48 (56%) 9 (11%) 57 (67%) 85 Odds ratio (95% CI) 10 (ref) 119 (070201) 074 (033170) 1085 (065180) 094 (066135) p NA 052 048 075 075 Data are number of individuals (%) unless otherwise indicated. NA=not applicable. *p=0005 for test of Hardy-Weinberg equilibrium. Expected distribution of genotypes in controls is GG: 42701 (7736%), GA: 11698 (2119%), and AA: 801 (145%). Risk estimate for carrying AA genotype would be higher if control population was in Hardy-Weinberg equilibrium. p=00069 for test of Hardy-Weinberg equilibrium. Expected distribution of genotypes in controls is GG: 47770 (7098%), GA: 17861 (2654%), and AA: 1670 (248%). Risk estimate for carrying AA genotype would be higher if control population was in Hardy-Weinberg equilibrium. Study sites for NCISEER. Table 4: Study-specific genotype frequencies and risks for TNF 308GA and IL10 3575TA and diffuse large B-cell lymphoma non-Hodgkin lymphoma for both the GA and AA genotypes (table 2). When restricted to diffuse large B-cell lymphoma, the main histological subtype, which comprised about 27% of cases of non-Hodgkin lymphoma, the risk estimates were stronger (table 3). Risk estimates for diffuse large B-cell lymphoma remained significant when any study was removed from the analysis (not shown). Although the tests for departure from Hardy-Weinberg equilibrium for TNF 308GA in controls were significant for the EPILYMPH--Spain and University of California San Francisco studies, the observed frequencies differed little from that expected under Hardy-Weinberg equilibrium (eg, about 1% for homozygotes in both studies; table 4). No genotyping errors were seen when quality control samples were rechecked in either study. Furthermore, genotyping done for non-InterLymph related projects in these studies found that 95% or more of the single-nucleotide polymorphisms assessed were in Hardy-Weinberg equilibrium, as expected, which lessens the possibility that these two control populations were not in Hardy-Weinberg equilibrium. Finally, a sensitivity analysis showed that risk estimates were much the same when we excluded these two studies from the analysis (p for trend=00013): the odds ratio was 129 (95% CI 110151) before exclusion and 126 (106150) after exclusion for the TNF 308GA genotype, and 165 (116234) before exclusion and 158 (101247) after exclusion for the AA genotype. By contrast, TNF 308GA was not associated with follicular lymphoma (table 3), the second most common subtype, comprising about 19% of cases, and the association was significantly different from that seen for diffuse large B-cell lymphoma (table 3). Furthermore, this polymorphism was not associated with risk of the other histological subtypes of nonHodgkin lymphoma combined (ie, for all non-Hodgkin lymphoma cases minus diffuse large B-cell lymphoma and follicular lymphoma): risk estimates were 115 (099135) for heterozygotes and 100 (065154) for homozygotes (p for trend=016), and the effect differed significantly from the association with diffuse large B-cell lymphoma (p=001). However, we cannot exclude the possibility of an effect in a small histological subgroup that we did not assess in this study. Figure 1 shows study-specific associations between TNF 308GA and diffuse large B-cell lymphoma and follicular lymphoma, under an additive model. The TNF 308GA association was consistent for diffuse large B-cell lymphoma, with all studies showing risk estimates of higher than 10 (figure 1, table 4). The pooled estimate from the test for trend (ie, additive model) remained highly significant after adjustment by the false-discovery rate method (ie, original p value of 0000055 adjusted for 12 comparisons with each of diffuse large B-cell lymphoma and follicular lymphoma became p=00013), and the finding was deemed noteworthy as determined by the false-positive report probability approach for even the lowest prior probability estimate of 00005. TNF 308GA was in linkage disequilibrium with LTA 252AG (pooled D=097, r2=038), which is consistent with previous reports,27 and LTA 252AG was also associated with increased risk of diffuse large B-cell lymphoma (table 3). Because studies done on LTA 252AG in stimulated mononuclear cells have shown raised concentrations of LT, a potent proinflammatory cytokine,28 we attempted to distinguish its effect from that of the TNF variant by estimating haplotypes. Assuming an additive model, we found that the haplotype with both TNF 308GA and LTA 252AG (ie, AG haplotype) was associated with increased risk of diffuse large B-cell lymphoma, with risk estimates of more than 10 for all studies (figure 2). By contrast, the GG haplotype was not associated with risk of diffuse large B-cell lymphoma (figure 2). The two haplotypes AG and GG differed significantly in risk of diffuse large B-cell lymphoma (p=0003). The other key finding of our study was that IL10 3575TA was associated with an increased risk of non-Hodgkin lymphoma (table 2), particularly with risk of diffuse large B-cell lymphoma, but not follicular lymphoma (table 3, figure 3) or with other histologies combined (p for trend=022). This result remained http://oncology.thelancet.com Vol 7 January 2006 33 Articles A EPILYMPH--Italy EPILYMPH--Spain15 University of California, San Francisco16 EPILYMPH--Germany17 Connecticut18 UK19 NCI-SEER12 Pooled Odds ratio (95% CI) 110 (043285) 139 (090214) 140 (097200) 118 (082170) 115 (082160) 111 (087143) 156 (122199) 129 (114146) 01 Decreased risk of DLBCL 10 Increased risk of DLBCL Odds ratio (95% CI) 100 B Odds ratio (95% CI) 180 (043750) 104 (053206) 094 (064139) 100 (064157) 100 (069146) 082 (061110) 123 (092162) 100 (086116) 01 10 100 Decreased risk of Increased risk of follicular lymphoma follicular lymphoma Odds ratio (95% CI) Figure 1: Forest plots for study-specific and pooled risk estimates from additive model of TNF 308GA for diffuse large B-cell lymphoma (DLBCL, A) and follicular lymphoma (B) DLBCL pooled estimate p00001; follicular lymphoma pooled estimate p=098. significant for diffuse large B-cell lymphoma when any one study was excluded (data not shown) and was generally consistent, with most studies having risks from the additive model of more than 10 (figure 3, table 4). The p value from the additive model was 0056 after adjustment by the false-discovery rate method and the finding was noteworthy as determined by the falsepositive report probability method for only the highest prior probability estimate of 005. IL10 3575TA was in strong linkage disequilibrium with IL10 1082AG (pooled D=097 and r2=063), which also was associated with increased risk of diffuse large B-cell lymphoma (table 3). We attempted to separate the effects of each polymorphism, since both could be functional.29,30 Assuming an additive model, the haplotype with IL10 3575TA and IL10 1082AG (ie, AG haplotype) was associated with increased risk of diffuse large B-cell lymphoma, and this risk was consistent across most studies (figure 4). By contrast, the TG haplotype was not associated with risk of diffuse large B-cell lymphoma (figure 4). The two haplotypes AG and TG differed significantly in risk of diffuse large B-cell lymphoma (p=0007). Associations did not differ by age or sex, and there was no multiplicative genegene interaction between TNF 308GA and IL10 3575TA (not shown). For individuals homozygous for TNF 308GA and homozygous or heterozygous for IL10 3575TA, risk of diffuse large B-cell lymphoma was doubled, with no association with follicular lymphoma (table 3). None of the other single-nucleotide polymorphisms investigated in our study was associated with risk of all non-Hodgkin lymphoma (table 2) or of diffuse large B-cell lymphoma or follicular lymphoma (table 3). We note that the homozygous IL1B 511CT genotype was significantly associated with increased risk of nonHodgkin lymphoma in one study and with decreased risk in another study, whereas the pooled analysis of all studies showed no overall association (figure 5). A EPILYMPH--Italy EPILYMPH--Spain15 University of California, San Francisco16 EPILYMPH--Germany17 Connecticut18 UK19 NCI-SEER12 Pooled Odds ratio (95% CI) 119 (046308) 159 (100252) 132 (090193) 112 (077164) 113 (081159) 108 (083141) 165 (128212) 129 (113147) B Odds ratio (95% CI) 185 (095359) 096 (058158) 079 (050124) 01 Decreased risk 10 of DLBCL Increased risk 100 of DLBCL Odds ratio (95% CI) 079 (052119) 080 (055117) 101 (074138) 125 (097162) 099 (087114) 01 Decreased risk of DLBCL 10 Increased risk 100 of DLBCL Odds ratio (95% CI) Figure 2: Forest plots for study-specific and pooled risk estimates from additive model of haplotypes with TNF 308GA and LTA 252AG for diffuse large B-cell lymphoma (DLBCL) (A) AG versus GA. Pooled estimate p=000014. (B) GG versus GA. Pooled estimate p=095. Frequency of haplotypes: AG=016, GG=016, GA=068, and AA=0003. Global omnibus test p=000125. 34 http://oncology.thelancet.com Vol 7 January 2006 Articles A Odds ratio (95% CI) B Odds ratio (95% CI) EPILYMPH--Italy EPILYMPH--Spain15 University of California, San Francisco16 EPILYMPH--Germany17 Connecticut18 UK19 NCI-SEER12 British Columbia Pooled 140 (079250) 106 (073154) 101 (073140) 110 (083146) 139 (106182) 124 (100154) 113 (093137) 094 (066135) 115 (104126) 196 (073525) 087 (051151) 099 (073135) 114 (083156) 144 (106195) 109 (086138) 109 (088134) 093 (064136) 111 (100124) 01 Decreased risk of DLBCL 10 100 Increased risk of DLBCL Odds ratio (95% CI) 01 10 100 Decreased risk of Increased risk of follicular lymphoma follicular lymphoma Odds ratio (95% CI) Figure 3: Study-specific and pooled risk estimates from additive model of IL10 3575TA for diffuse large B-cell lymphoma (DLBCL, A) and follicular lymphoma (B) (A) Pooled estimate p=0006. (B) Pooled estimate p=0059. Discussion In a large pooled analysis of data from eight studies of non-Hodgkin lymphoma participating in the InterLymph consortium, we noted that TNF 308GA and IL10 3575TA were associated with an increased risk of non-Hodgkin lymphoma, particularly diffuse large B-cell lymphoma. TNF and IL10 are good candidate genes for the study of lymphomagenesis because they code for important immunoregulatory cytokines that are crucial mediators of inflammation, apoptosis, and Th1/Th2 balance, and function as autocrine growth factors in lymphoid tumours.5,31,32 Moreover, studies3335 of TNF and IL10 knock-out mice have shown that each cytokine affects B-cell lymphomagenesis either indirectly or directly. Furthermore, clinical studies9,10,30 suggest that serum concentrations of TNF and IL10 affect prognosis of non-Hodgkin lymphoma, particularly diffuse large B-cell lymphoma. Lastly, research36 on monozygotic twins suggests that production of TNF and IL10 have a strong heritable basis. Other studies10,30,37,38 that have assessed TNF 308GA and several IL10 polymorphisms and risk of nonHodgkin lymphoma have been small and not populationbased, and could not provide conclusive results about the role of these variants in the development of this disease. The results we report here are derived from a pooled analysis that included at least ten times more study participants than these previous reports. Our findings, especially for TNF 308GA, were highly significant, and effects were consistent across studies. TNF 308GA has been associated with increased susceptibility for cerebral malaria, and other infections and inflammatory conditions such as rheumatoid arthritis and Sjgren's syndrome.39,40 Several studies4143 have shown that this allele results in higher constitutive and inducible expression of TNF. However, conclusions based on other model systems have varied,40 probably because of differences in cell type and stimuli used, and, as such, further work is needed to clarify the function of this polymorphism. Th1/Th2 balance T-helper (Th) 1 and Th 2 cells are characterised by secretion of specific cytokine patterns that direct distinct immune response pathways. Whereas Th1 cells drive cellular immunity to fight intracellular pathogens including viruses and eliminate cancerous cells, Th2 cells drive humoral immunity via upregulation of antibody production to protect against extracellular pathogens. An imbalance of the Th1/Th2 system could be responsible for both the occurrence and the progression of infectious, autoimmune, and neoplastic diseases. A EPILYMPH--Italy EPILYMPH--Spain15 University of California, San Francisco16 EPILYMPH--Germany17 Connecticut18 UK19 NCI-SEER12 Pooled Odds ratio (95% CI) 145 (081259) 103 (069153) 102 (073142) 115 (086155) 145 (110192) 119 (094149) 111 (090135) 117 (105130) B Odds ratio (95% CI) 122 (050294) 090 (053155) 091 (049170) 130 (083202) 108 (069167) 070 (048103) 081 (056118) 091 (076108) 01 10 100 Decreased risk Increased risk of DLBCL of DLBCL Odds ratio (95% CI) 01 10 100 Decreased risk Increased risk of DLBCL of DLBCL Odds ratio (95% CI) Figure 4: Study-specific and pooled risk estimates from additive model of haplotypes with IL10 3575TA and IL10 1082AG for diffuse large B-cell lymphoma (DLBCL) (A) AG versus TA. Pooled estimate p=0004. (B) TG versus TA. Pooled estimate p=029. Frequency of haplotypes: AG=036; TG=010, TA=053, and AA=0006, respectively. Global omnibus test p=00075. http://oncology.thelancet.com Vol 7 January 2006 35 Articles EPILYMPH--Italy EPILYMPH--Spain15 University of California, San Francisco16 EPILYMPH--Germany17 Connecticut18 NCI-SEER12 British Columbia Pooled Odds ratio (95% CI) 075 (032175) 060 (039095) 121 (074196) 164 (107252) 117 (078177) 105 (076145) 142 (085236) 111 (092135) 01 10 100 Decreased risk Increased risk of non-Hodgkin lymphoma of non-Hodgkin lymphoma Odds ratio (95% CI) Figure 5: Study-specific and pooled risk estimates for IL1B 511CT TT genotype compared with CC genotype for all cases of non-Hodgkin lymphoma. Pooled estimate p=028. Test for heterogeneity p=014. We found that the haplotype with TNF 308GA and LTA 252AG (ie, the AG haplotype), but not the GG haplotype, was associated with increased risk of diffuse large B-cell lymphoma, suggesting that TNF 308GA could act alone or in conjunction with LTA 252AG. However, we cannot exclude the possibility that its association with diffuse large B-cell lymphoma is from another variant in linkage disequilibrium within TNF or LTA27 or in other neighbouring immunomodulatory genes such as NFKBIL1 or BAT1. Furthermore, because TNF is located within the HLA class III region (250 kb centromeric to the HLA-B locus and 850 kb telomeric to the class II HLA-DR locus), and HLAs have a crucial function in regulation of the immune response to infection and malignant transformation, linkage dis- equilibrium of the TNF promoter polymorphism with other alleles within this region, including the extended HLA haplotype A1-B8-DR3,44 could be responsible for increasing risk of diffuse large B-cell lymphoma. The IL10 3575A allele, which results in lower production of IL10 compared with the 3575T allele,29 was associated with increased risk of non-Hodgkin lymphoma, particularly for diffuse large B-cell lymphoma. The higher risk for diffuse large B-cell lymphoma was restricted to the haplotype containing IL10 3575TA and IL10 1082AG (ie, AG haplotype) rather than the TG haplotype, suggesting that IL10 3575TA is more important than IL10 1082AG in determining risk of non-Hodgkin lymphoma. A possible mechanism of lymphomagenesis consistent with our findings is that higher expression of TNF and LT upregulates antiapoptotic regulators and proinflammatory effectors mediated via the nuclear transcrip- tion factor (NF)-B pathway, which provides key signals to support B-cell survival and differentiation in the germinal centre.45 NF-B target genes are highly expressed in activated B-cell-like diffuse large B-cell lymphoma, a major subgroup of diffuse large B-cell lymphoma.46 A tightly regulated balance between proapoptotic and antiapoptotic processes is of utmost importance and thus a slight imbalance towards increased cell survival could favour lymphomagenesis. At the same time, IL10 is a potent downregulator of the production of macrophage proinflammatory cytokines, notably TNF.47 Consequently, decreased expression of IL10 would less efficiently suppress proinflammatory cytokine production and, therefore, could increase risk of non-Hodgkin lymphoma. The observed association of the TNF promoter single-nucleotide polymorphism with diffuse large B-cell lymphoma underscores the role of this key cytokine in regulation of the immune response and perhaps surveillance. In this regard, the TNF pathway could be a suitable target for intervention. Concerns have been raised that population stratification--ie, confounding by unrecognised ethnic admixture--can lead to a test of association with a misleadingly low p value, especially in large studies.48 We note that a small bias from any source could result in significant associations as the sample size increases. Marchini and colleagues48 discussed the potential effects of population stratification on the test for trend for scenarios of modest differences in allele frequency and disease risk that might be present in studies of mixed European populations. Their simulations suggest that population stratification is unlikely to have biased our highly significant p value for the TNF 308GA association by more than about one order of magnitude. Furthermore, the consistency of our findings across studies and across sites of the multicentred NCI-SEER study (table 4) provides additional evidence against population stratification, because this potential bias is unlikely to change risk estimates in the same direction and extent in studies done in diverse settings and in different study populations.49 In conclusion, our results of common cytokine single-nucleotide polymorphisms and non-Hodgkin lymphoma identified two genetic variants of probable importance in risk of this disease and showed the effectiveness of a large consortium in identification of genetic associations. The large scale of our pooled analysis helps to mitigate against false-negative and false-positive results--issues that hamper smaller studies.23,50 Our findings provide an important clue to lymphomagenesis; nevertheless, they need to be replicated and to that end, several thousand additional cases and controls will become available for analysis in the near future from ongoing studies of non-Hodgkin lymphoma. Finally, these results suggest that exploration of additional variants in TNF, LTA, and IL10, in their receptors and other related genes, and in genes in linkage disequilibrium including the class III region of the MHC locus, should provide further insight into the pathogenesis and ultimately the prevention and treatment of non-Hodgkin lymphoma, the incidence of which has risen steadily worldwide over the last half of the twentieth century and for which the factors governing development remain elusive.51,52 36 http://oncology.thelancet.com Vol 7 January 2006 Articles Contributors The project was conceived and led by members of the genotyping working group within InterLymph: N Rothman, C F Skibola, S S Wang, G Morgan, Q Lan, M T Smith, J J Spinelli, A Brooks-Wilson, P Hartge, S Chanock, and A Nieters. The cochairs of the InterLymph consortium at the inception of this project were P Boffetta, M Linet, and B Armstrong. Investigators who obtained and provided data from the studies are: NCI-SEER: P Hartge, S S Wang, J R Cerhan, W Cozen, S Davis, R K Severson, M Linet, and N Rothman; Connecticut: T Zheng, Q Lan, T R Holford, B Leaderer, and Y Zhang. EPILYMPH--Germany: N Becker and A Nieters; UK: E Roman, G Morgan, E Willett, S Rollinson, and C Spink; British Columbia: J J Spinelli, R P Gallagher, A Brooks-Wilson, and A Lai; EPILYMPH--Spain: S de Sanjose, X Bosch, D Whitby, and R Bosch; University of California San Francisco: E A Holly, C F Skibola, P M Bracci, and M T Smith; and EPILYMPH--Italy: P Cocco, P S Moore, A Scarpa, and M G Ennas. EPILYMPH is coordinated by P Brennan and P Boffetta. Bioinformatics support was provided by C F Skibola, A Nieters, M Yeager, and S Chanock, and quality control samples were provided by S Chanock. Genotyping was done by M Yeager and S Chanock for the NCI-SEER, Yale, and EPILYMPH--Spain studies; C F Skibola and M T Smith for the University of California San Francisco study; A Brooks-Wilson for the British Columbia study; S Rollinson and C Spink for the UK study; A Scarpa and P S Moore for EPILYMPH--Italy study; and A Nieters for EPILYMPH--Germany study. Statistical analysis was done by S I Berndt and SS Wang, with input from S Wacholder, E Willett, J J Spinelli, Q Lan, P Hartge, and N Rothman. The manuscript was drafted and revised by N Rothman, C F Skibola, S S Wang, G Morgan, Q Lan, M T Smith, E Willett, S de SanJose, P Cocco, S I Berndt, P Brennan, S Wacholder, P Hartge, E Roman, P Boffetta, S Chanock, and A Nieters. All authors reviewed and approved the manuscript. Conflict of interest We declare no conflicts of interest. Acknowledgments We thank Randy D Gascoyne, Joseph M Connors, and Stephen Leach from the British Columbia study; Graham Law and Alexandra Smith from the UK study; Giorgio Broccia, Emanuele Angelucci, Attilio Gabbas, and Giovannino Massarelli from the EPILYMPH--Italy study; Rebeca Font, Yolanda Benavente, Elisabeth Guino, Alberto Fernandez de Sevilla, Tomas Alvaro, Mercedes Garcia, and Vicens Romagosa from the EPILYMPH--Spain study; Ina Koegel and Evelin Deeg from the EPILYMPH--Germany study; Brian Chiu from the InterLymph genotyping working group; and Nilanjan Chatterjee, Geoffrey Tobias, Lindsay Morton, Robert N Hoover, and Joseph F Fraumeni Jr, from the US National Cancer Institute (NCI). This project was supported by a European Commission grant to EPILYMPH (Grant number QLK4-CT-2000-00422); US National Institutes of Health (NIH) grants CA104862 (M T Smith, principal investigator) and CA45614, CA89745 and CA87014 (E A Holly, principal investigator), University of California San Francisco; Federal funds from the NCI, NIH, under contract NO1-CO-12400; Fondazione Cariverona (2004, A Scarpa; 2005, PS Moore), and Compagnia di S Paolo--Programma Oncologia (P Cocco), EPILYMPH--Italy; German Jos Carreras Leukemia Foundation DJCLS-R04/08 (A Nieters, principal investigator) and Federal Office for Radiation Protection (StSch4261 and StSch4420) (N Becker, principal investigator), EPILYMPH--Germany; FISS grant PI040091 (S De Sanjose, principal investigator), RCESP 03/09, and FISS grant PI041467 (R Bosch, principal investigator.), EPILYMPH--Spain; National Cancer Institute of Canada, the Chan Sisters Foundation, and the Canadian Institutes for Health Research, British Columbia; Leukaemia Research, UK; and the Intramural Research Program of the US NIH, NCI, NCI-SEER. References 1 Hofmann SR, Ettinger R, Zhou YJ, et al. Cytokines and their role in lymphoid development, differentiation and homeostasis. Curr Opin Allergy Clin Immunol 2002; 2: 495506. 2 Pfeffer K. Biological functions of tumor necrosis factor cytokines and their receptors. Cytokine Growth Factor Rev 2003; 14: 18591. 3 Dinarello CA. Biologic basis for interleukin-1 in disease. Blood 1996; 87: 2095147. 4 Hilbert DM, Kopf M, Mock BA, et al. Interleukin 6 is essential for in vivo development of B lineage neoplasms. J Exp Med 1995; 182: 24348. 5 Khatri,VP, Caligiuri MA. A review of the association between interleukin-10 and human B-cell malignancies. Cancer Immunol Immunother 1998; 46: 23944. 6 Moore KW, de Waal MR, Coffman RL, O'Garra A. Interleukin-10 and the Interleukin-10 receptor. Annu Rev Immunol 2001; 19: 683765. 7 Lucey DR, Clerici M, Shearer GM. Type 1 and type 2 cytokine dysregulation in human infectious, neoplastic, and inflammatory diseases. Clin Microbiol Rev 1996; 9: 53262. 8 Kurzrock R. Cytokine deregulation in hematological malignancies: clinical and biological implications. Clin Cancer Res 1997; 3: 258184. 9 Blay JY, Burdin N, Rousset F, et al. Serum interleukin-10 in nonHodgkin's lymphoma: a prognostic factor. Blood 1993; 82: 216974. 10 Warzocha K, Ribeiro P, Bienvenu J, et al. Genetic polymorphisms in the tumor necrosis factor locus influence non-Hodgkin's lymphoma outcome. Blood 1998; 91: 357481. 11 Altieri A, Bermejo JL, Hemminki K. Familial risk for non-Hodgkin lymphoma and other lymphoproliferative malignancies by histopathologic subtype: the Swedish Family-Cancer Database. Blood 2005; 106: 66872. 12 Chatterjee N, Hartge P, Cerhan JR, et al. Risk of non-Hodgkin's lymphoma and family history of lymphatic, hematologic, and other cancers. Cancer Epidemiol Biomarkers Prev 2004; 13: 141521. 13 Morton LM, Zheng T, Holford TR, et al, for the InterLymph Consortium. Alcohol consumption and risk of non-Hodgkin lymphoma: a pooled analysis. Lancet Oncol 2005; 6: 46976. 14 Morton LM, Hartge P, Holford TR, et al. Cigarette smoking and risk of non-Hodgkin lymphoma: a pooled analysis from the International Lymphoma Epidemiology Consortium (InterLymph). Cancer Epidemiol Biomarkers Prev 2005; 14: 92533. 15 de Sanjose S, Shah KV, Domingo-Domenech E, et al. Lack of serological evidence for an association between simian virus 40 and lymphoma. Int J Cancer 2003; 104: 52224. 16 Holly EA, Bracci PM. Population-based study of non-Hodgkin lymphoma, histology, and medical history among human immunodeficiency virus-negative participants in San Francisco. Am J Epidemiol 2003; 158: 31627. 17 Becker N, Deeg E, Nieters A. Population-based study of lymphoma in Germany: rationale, study design and first results. Leuk Res 2004; 28: 71324. 18 Zheng T, Holford TR, Leaderer B, et al. Diet and nutrient intakes and risk of non-Hodgkin's lymphoma in Connecticut women. Am J Epidemiol 2004; 159: 45466. 19 Willett EV, Smith AG, Dovey GJ, et al. Tobacco and alcohol consumption and the risk of non-Hodgkin lymphoma. Cancer Causes Control 2004; 15: 77180. 20 Jaffe ES, Harris NL, Stein H, Vardiman JW, eds. World Health Organization classification of tumours: pathology and genetics of tumours of haematopoietic and lymphoid tissues. IARC Press: Lyon, 2001. 21 Packer BR, Yeager M, Staats B, et al. SNP500Cancer: a public resource for sequence validation and assay development for genetic variation in candidate genes. Nucleic Acids Res 2004; 32: D52832. 22 Benjamini Y, Drai D, Elmer G, et al. Controlling the false discovery rate in behavior genetics research. Behav Brain Res 2001; 125: 27984. 23 Wacholder S, Chanock S, Garcia-Closas M, et al. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst 2004; 96: 43442. 24 Excoffier L, Slatkin M. Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol 1995; 12: 92127. 25 Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005; 21: 26365. 26 Schaid DJ, Rowland CM, Tines DE, et al. Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am J Hum Genet 2002;70: 42534. http://oncology.thelancet.com Vol 7 January 2006 37 Articles 27 Knight JC, Keating BJ, Rockett KA, Kwiatkowski DP. In vivo characterization of regulatory polymorphisms by allele-specific quantification of RNA polymerase loading. Nat Genet 2003; 33: 46975. 28 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-beta gene correlates with a variant amino acid in position 26 and a reduced level of TNF-beta production. J Exp Med 1991; 173: 20919. 29 Gibson AW, Edberg JC, Wu J, et al. Novel single nucleotide polymorphisms in the distal IL-10 promoter affect IL-10 production and enhance the risk of systemic lupus erythematosus. J Immunol 2001; 166: 391522. 30 Lech-Maranda E, Baseggio L, Bienvenu J, et al. Interleukin-10 gene promoter polymorphisms influence the clinical outcome of diffuse large B-cell lymphoma. Blood 2004; 103: 352934. 31 Aggarwal BB. Signalling pathways of the TNF superfamily: a double-edged sword. Nat Rev Immunol 2003; 3: 74556. 32 Balkwill F. Tumor necrosis factor or tumor promoting factor? Cytokine Growth Factor Rev 2002; 13: 13541. 33 Korner H, Cretney E, Wilhelm P, et al. Tumor necrosis factor sustains the generalized lymphoproliferative disorder (gld) phenotype. J Exp Med 2000; 191: 8996. 34 Batten M, Fletcher C, Ng LG, et al. TNF deficiency fails to protect BAFF transgenic mice against autoimmunity and reveals a predisposition to B cell lymphoma. J Immunol 2004; 172: 81222. 35 Czarneski J, Lin YC, Chong S, et al. Studies in NZB IL-10 knockout mice of the requirement of IL-10 for progression of B-cell lymphoma. Leukemia 2004; 18: 597606. 36 Westendorp RG, Langermans JA, Huizinga TW, et al. Genetic influence on cytokine production in meningococcal disease. Lancet 1997; 349: 191213. 37 Chouchane L, Ahmed SB, Baccouche S, Remadi S. Polymorphism in the tumor necrosis factor-alpha promotor region and in the heat shock protein 70 genes associated with malignant tumors. Cancer 1997; 80: 148996. 38 Fitzgibbon J, Grenzelias D, Matthews J, et al. Tumour necrosis factor polymorphisms and susceptibility to follicular lymphoma. Br J Haematol 1999; 107: 38891. 39 Hajeer AH, Hutchinson IV. TNF-alpha gene polymorphism: clinical and biological implications. Microsc Res Tech 2000; 50: 21628. 40 Bayley JP, Ottenhoff TH, Verweij CL. Is there a future for TNF promoter polymorphisms? Genes Immunol 2004; 5: 31529. 41 Wilson AG, Symons JA, McDowell TL, et al. Effects of a polymorphism in the human tumor necrosis factor alpha promoter on transcriptional activation. Proc Natl Acad Sci USA 1997; 94: 319599. 42 Heesen M, Kunz D, Bachmann-Mennenga B, et al. Linkage disequilibrium between tumor necrosis factor (TNF)-alpha-308 G/A promoter and TNF-beta NcoI polymorphisms: association with TNF-alpha response of granulocytes to endotoxin stimulation. Crit Care Med 2003; 31: 21114. 43 Bouma G, Crusius JB, Oudkerk Pool M, et al. Secretion of tumour necrosis factor alpha and lymphotoxin alpha in relation to polymorphisms in the TNF genes and HLA-DR alleles. Relevance for inflammatory bowel disease. Scand J Immunol 1996; 43: 45663. 44 Wilson AG, de Vries N, Pociot F, et al. An allelic polymorphism within the human tumor necrosis factor alpha promoter region is strongly associated with HLA A1, B8, and DR3 alleles. J Exp Med 1993; 177: 55760. 45 Husson H, Lugli SM, Ghia P, et al. Functional effects of TNF and lymphotoxin alpha1beta2 on FDC-like cells. Cell Immunol 2000; 203: 13443. 46 Shaffer AL, Rosenwald A, Staudt LM. Lymphoid malignancies: the dark side of B-cell differentiation. Nat Rev Immunol 2002; 2: 92032. 47 Wanidworanun C, Strober W. Predominant role of tumor necrosis factor-alpha in human monocyte IL-10 synthesis. J Immunol 1993; 151: 685361. 48 Marchini J, Cardon LR, Phillips MS, Donnelly P. The effects of human population structure on large genetic association studies. Nat Genet 2004; 36: 51217. 49 Wacholder S, Rothman N, Caporaso N. Population stratification in epidemiologic studies of common genetic variants and cancer: quantification of bias. J Natl Cancer Inst 2000; 92: 115158. 50 Colhoun HM, McKeigue PM, Davey SG. Problems of reporting genetic associations with complex outcomes. Lancet 2003; 361: 86572. 51 Hartge P, Devesa SS. Quantification of the impact of known risk factors on time trends in non-Hodgkin's lymphoma incidence. Cancer Res 1992; 52 (suppl): S556669. 52 Cartwright R, Brincker H, Carli PM, et al. The rise in incidence of lymphomas in Europe 198592. Eur J Cancer 1999; 35: 62733. 38 http://oncology.thelancet.com Vol 7 January 2006