& prevention germline variation in apoptosis pathway genes...

Rese

GerNon

JennifAhmeThom

Abst

Intro

Nocommlifetimand 1dencefor en

AuthorRochesof Mediof HeaDepartmLaboraDepartMedicin

Note: Smiology

CorresSciencSouthw266-24

doi: 10

©2010

www.a

Do

Published OnlineFirst September 20, 2010; DOI: 10.1158/1055-9965.EPI-10-0581

Cancer
Epidemiology,
iomarkersrevention

arch Article

mline Variation in Apoptosis Pathway Genes and Risk of

B& P

–Hodgkin's Lymphoma

er L. Kelly1, Anne J. Novak2, Zachary S. Fredericksen3, Mark Liebow4, Stephen M. Ansell2,
t Dogan5, Alice H. Wang3, Thomas E. Witzig2, Timothy G. Call2, Neil E. Kay2, as M. Habermann2, Susan L. Slager3, and James R. Cerhan6
ractBac

in nonditionlinked

MebasedcompoMACH

Res(P = 0confidBCLAsevensuccesand tw

Con

of NHLvironme

s' Affiliatioter, Rochestcine, 3Divisilth Scienceent of Medtory Medicment of Hee, Rocheste

upplement, Biomarke

ponding Aes Researchest, Roche78. E-mail:

.1158/1055-

American A

acrjourna

wnloaded

kground: The t(14;18)(q32;q21) translocation is the most commonly observed chromosomal translocation–Hodgkin's lymphoma (NHL), resulting in constitutive Bcl-2 expression and apoptosis inhibition. In ad-, germline variation in both BCL2L11 (BIM) and CASP9, known regulators of apoptosis, has recently beentoNHL risk.We conducted a comprehensive evaluation of 36 apoptosis pathway geneswith risk ofNHL.thods: We genotyped 226 single-nucleotide polymorphisms (SNP) from 36 candidate genes in a clinic-study of 441 newly diagnosed NHL cases and 475 frequency-matched controls. We used principalnents analysis to assess gene-level associations, and logistic regression to assess SNP-level associations.was used for imputation of SNPs in BCL2L11 and CASP9.

ults: In gene-level analyses, BCL2L11 (P = 0.0019), BCLAF1 (P = 0.0097), BAG5 (P = 0.026), and CASP9.0022) were associated with NHL risk after accounting for multiple testing (tail strength, 0.38; 95%ence interval, 0.05-0.70). Two of the five BCL2L11 tagSNPs (rs6746608 and rs12613243), both genotypedF1 tagSNPs (rs797558 and rs703193), the single genotyped BAG5 tagSNP (rs7693), and three of thegenotyped CASP9 tagSNPs (rs6685648, rs2020902, and rs2042370) were significant at P < 0.05. Wesfully imputed BCL2L11 and CASP9 SNPs previously linked to NHL, and replicated all four BCL2L11o of three CASP9 SNPs.clusion: We replicated the association of BCL2L11 and CASP9 with NHL risk at the gene and SNPand identified novel associations with BCLAF1 and BAG5.
level,
Impact: Closer evaluation of germline variation of genes in the apoptosis pathway with risk of NHL and itssubtypes is warranted. Cancer Epidemiol Biomarkers Prev; 19(11); 2847–58. ©2010 AACR.

Howea relaThe

locatimostlocati

duction

n–Hodgkin's lymphoma (NHL) is the fifth moston cancer overall in the United States, and thee odds of developing NHL is 1 in 45 for menin 53 for women (1). The remarkable rise in inci-

over the last 50 years suggests a major rolental factors in the etiology of this cancer.

ulin hBcl-2trol olymphsis toinfectinappand ogenesphombalandeaththermtion ilympphoc(CLL/Bcl-2

ns: 1School of Medicine and Dentistry, University ofer, New York and 2Division of Hematology, Departmenton of Biomedical Statistics and Informatics, Departments Research, 4Division of General Internal Medicine,icine, 5Division of Anatomic Pathology, Department ofine and Pathology, and 6Division of Epidemiology,alth Sciences Research, Mayo Clinic College ofr, Minnesota

ary data for this article are available at Cancer Epide-rs & Prevention Online (http://cebp.aacrjournals.org/).

uthor: James R. Cerhan, Department of Health, Mayo Clinic College of Medicine, 200 First Streetster, MN 55905. Phone: 507-538-0499; Fax: [email protected]

9965.EPI-10-0581

ssociation for Cancer Research.

ls.org

on August 15, 2019. © 20cebp.aacrjournals.org from

ver, established risk factors to date account for onlytively small fraction of the cases (2).t(14;18)(q32;q21) translocation is a hallmark trans-on in follicular lymphoma (FL; ref. 3), one of thecommon lymphoma subtypes (2). With this trans-on event, BCL2 becomes fused to the immunoglob-eavy chain (IgH) locus, leading to constitutiveexpression and apoptosis inhibition under the con-f the IgH enhancer (4, 5). Under normal conditions,ocytes must strictly regulate growth and apopto-provide adequate immunologic defenses againstions while not overwhelming the organism withropriate cell numbers. With the t(14;18)(q32;q21)ther less commonly observed translocations ofin the apoptosis pathway observed among lym-a cases (6, 7), it is clear that dysregulation of thece between cell proliferation and programmed cellis a central feature in lymphomagenesis (8). Fur-ore, evidence that the t(14;18)(q32;q21) transloca-s also present in ∼30% of diffuse large B-cellhomas (DLBCL) and in 1% to 2% of chronic lym-ytic leukemias/small lymphocytic lymphomas
SLL; ref. 6), and the deletion/downregulation ofinhibiting microRNA species (mir-15 and mir-16)
2847

10 American Association for Cancer Research.

http://cebp.aacrjournals.org/

in CLin lymBcl-

antiapand inenzym

The hhealthwouldprotei

Table ul Ca ol (2

Gene (alias)* Function HapMapSNPs (n)†

SNPsgenotyped (n)

Gene coverage‡ SN Prin lmp nts

BCL2BAD Pro 7 7 0 1BAG Ant 10 8 0 9BAG Ant 42 5 0 6BAG Ant 10 10 0 8BAG Ant 10 5 1 26BAK Pro 19 5 0 0BAX Pro 10 4 0 4BCL Ant 179 5 7 1 8BCL Ant 13 7 0 7BCL ti/P 41 8 0 4BCL Ant 9 5 0 5BCL Pro 31 4 2 019BCL Pro 4 7 0 9BCL Pro 83 1 9 1 5BCL Pro 82 2 5 4 6BCL Ant 6 6 0 3BCL Ant 8 8 2 097BID Pro 44 1 4 1 6BIK Pro 31 7 0 1BNIBNIHRK Pro 12 5 0 4

CaspaAIF1 Ant 12 7 0 7APA Pro 50 1 8 6 65BIR Ant 9 8 0 5CAS Pro 30 7 0 6CAS Pro 11 10 1 2CAS Pro 10 9 0 9CAS Pro 24 2 0 0CAS Pro 22 7 0 5CAS Pro 22 3 0 5CAS Pro 15 6 0 9CAS Pro 59 1 8 2 1CCD

Abb*As†To‡Ge s not

Kelly et al.

Cance2848

Do


L (9), suggests a broad role for Bcl-2 and apoptosisphoma.2 is a member of a large family of proapoptotic andoptotic proteins, which coordinate both extrinsic

ne coverage is defined as (total HapMap SNPs − number of SNP

trinsic cell signals to activate caspases, the effectores necessary for apoptosis execution (4, 5, 10).

cientaccum

r Epidemiol Biomarkers Prev; 19(11) November 2010

on August 15, 2019. © 20cebp.aacrjournals.org wnloaded from

igh prevalence of the t(14;18) translocation amongy individuals, estimated as high at 66% at age 50,indicate that perhaps overexpression of the Bcl-2

n as a result of this transformation may not suffi-

tagged)/total HapMap SNPs.

1. Gene-level res
ts, Mayo se-Contr Study of NHL
for mulatin

Can

10 Am

002-2005)

alignant transformg evidence that oth

cer Epidemiology, B

erican Association

ation (11-13).er Bcl-2 family

iomarkers & P

for Cancer Res

Psco

cipaone

Thereprotei

reventi

earch.

analysis

P ≤ 0.05 (n) DF P

family members
(BCL2L8) 2 1.4% 2 0.7 1 i 2 0.0% 2 0.1 3 i 8 7.1% 4 0.3 4 i 2 0.0% 1 0.7 5 i 1 0.0% 1 0.0 1 (BCL2L7) 2 2.6% 2 0.2
2
0.0% 1 0.4 2 i 3 4.3% 21 0.2 2A1 i 4 6.9% 1 0.2 2L1 (BCL-XL) An ro 4 2.9% 2 0.9 2L10 (Diva) i 1 5.6% 1 0.6 2L11 (BIM) 5 5.2% 2 0.0 2L12 2 5.0% 2 0.4 2L13 3 1.6% 4 0.1 2L14 (BCL-G) 6 4.9% 9 0.2 2L2 (BCL-W) i 2 6.7% 2 0.1 AF1 (BTF) i 2 7.5% 1 0.0
0
3.2% 6 0.4 6 7.4% 3 0.7
P2
3 0.9% 2 0.2 Pro 55 7 0 4P3 Pro 16 2 75.0% 0 2 0.19
1
0.0% 1 0.9 se family members
i
3 5.0% 3 0.4 F1 (CED4) 3 2.0% 4 0.0 C3 (AIP1) i 2 8.9% 2 0.3 PI (IL1BC) 1 6.7% 1 0.3 P10 9 0.0% 4 0.1 P2 3 0.0% 3 0.2 P3 1 9.2% 1 0.9 P4 6 2.7% 3 0.4 P5 2 6.4% 2 0.1 P6 2 0.0% 2 0.2 P7 0 1.4% 4 0.3 P8 (MACH) 2 7.6% 5 0.8 AS Pro 34 1 6 0 8
ASP9 (APAF3) Pro 58 7 96.6% 3 4 0.0022FFB Pro 19 2 10.5% 1 2 0.097

reviations: Anti, antiapoptotic; Pro, proapoptotic; DF, degrees of freedom.defined in Entrez Gene.tal number of SNPs (MAF > 0.05) from HapMap version Build 36 dbSNP 126.

isns,

on


caspaulateesis. SCASPmenteNHLtial exmembThe

on geapoptlatingneticComm(22),DNAmore,tifiedare asTo ousivelythe apcludinmembcase-ccontrociatioNHL

Hergermlpathwand cpoolethis pevalupoptothe in

Mate

StudyThi

Subjeand aFull dstudysota hbasedpants2005.theirsidendiagnconfir

TabStu

Gen

BAGBCLBCLBCLBCLBCLBCLBCLCAPCAPCAPCAPCAPCAPCAP

*Ref†SNgen‡OR

Apoptosis Pathway Variants and NHL Risk

www.a

Do


se family proteases, and genes that encode and reg-their transcription are important in lymphomagen-omatic mutations in many caspase genes, including3, CASP7, CASP8, and CASP10, have been docu-d in a wide variety of human cancers including(14, 15). Furthermore, there is evidence of differen-pression of both caspase genes and Bcl-2 familyer genes among the NHL subtypes (16-18).above-mentioned studies have primarily focusednetic events that affect expression or function ofotic proteins within the tumor. However, accumu-epidemiologic evidence suggests that germline ge-variation also plays a role in NHL etiology (19-21).on variation related to B-cell growth and survival

inflammation and immune function (23-28), andrepair (29) has been linked to NHL risk. Further-recent genome-wide association studies have iden-novel single-nucleotide polymorphisms (SNP) thatsociated with risk of developing CLL and FL (30, 31).r knowledge, only one other group has comprehen-evaluated the role of germline genetic variation inoptosis pathway with regard to NHL etiology, in-g 8 BCL2 family members and 12 caspase familyers, in a pooled analysis of three independentontrol studies with a total of 1,946 cases and 1,808ls (32, 33). Statistically significant gene-level asso-
ns of BCL2L11, CASP1, CASP8, and CASP9 withrisk were identified.
ing to(65%)

le gdy

e SNP ID* Type†Major/ MAF

a

5 (02L (02L (02L (02L (02L (0AF (1AF (1S9 (1S9 (0S9 (0S9 (0

s adjusted for age and sex.

Caacrjournals.org


e, we independently evaluated the hypothesis thatine genetic variation in genes from the apoptosisay is associated with risk of developing NHL,ompare these results with those reported in thed study in an attempt to validate the relevance ofathway in NHL etiology. The 36 candidate genesated (Table 1; Supplementary Table S1) are proa-tic or antiapoptotic, and are represented in bothtrinsic and extrinsic apoptotic pathways.

rials and Methods

population and data collections study was reviewed and approved by the Humancts Institutional Review Board at the Mayo Clinic,ll participants provided written informed consent.etails of this ongoing, clinic-based, case-controlconducted at the Mayo Clinic in Rochester, Minne-ave been previously reported (27). This analysis ison phase 1 of the study, which includes partici-enrolled from September 1, 2002 to September 30,Briefly, eligible patients were within 9 months offirst NHL diagnosis, ages 20 years or older, and re-ts of Minnesota, Iowa, or Wisconsin at the time ofosis. All cases were reviewed and histologicallymed by a hematopathologist and classified accord-
the WHO criteria (34). Of the 956 eligible cases, 626participated in the study. Clinic-based controls
- l
2. SNPof NHL
-level asso(2002-200

ciation5)

from
enes with P ≤ 0.05 from the gene-level test, Mayo Case
O d

aca

(0 (1(0 (0(0 7(0 (0(0 (0(0 (0(1 (0(1 (0 3(0 (1 3(0 (0(0(0 (0

ncer Epidemiol Biomarkers Prev; 19(11) Nove

10 American Association for Cancer Res

Contro

q

mber 201

earch.

Adjusted
R‡ (95% CI) Ptren
minor
Case Cont Pervari
copy ofnt allele

Onevari

copy ofnt allele

Twovari

opies ofnt allele

rs76
93 mrna-u tr C/T 0.39 0.34 1.24 .02-1.50) 1.15 .87-1.52) 1.62 .07-2.45) 0.031 0.384 11 rs674 6608 int G/A 0.39 0.44 0.82 .68-1.00) 0.77 .57-1.03) 0.70 .47-1.04) 0.049 0.384 11 rs1261 3243 int T/C 0.04 0.07 0.58 .38-0.87) 0.60 .40-0.92) — 0.008 0.332 11 rs724 710 c-s C/T 0.28 0.31 0.89 .72-1.09) 0.88 .67-1.15) 0.81 .49-1.33) 0.26 0.622 11 rs616 130 int A/C 0.41 0.45 0.87 .72-1.05) 0.84 .62-1.13) 0.84 .62-1.13) 0.15 0.542 11 rs675 3785 mrna-u tr C/A 0.39 0.44 0.84 .69-1.02) 0.80 .60-1.08) 0.72 .48-1.08) 0.078 0.453 1 rs797 558 int C/G 0.18 0.14 1.38 .07-1.80) 1.44 .07-1.93) 1.55 .60-3.99) 0.015 0.363 1 rs703 193 int C/T 0.18 0.14 1.42 .10-1.84) 1.50 .12-2.00) 1.54 .63-3.77) 0.007 0.332
rs668
5648 int T/C 0.33 0.26 1.41 .14-1.73) 1.29 .98-1.69) 2.31 .37-3.91) 0.001 0.219 rs464 6077 int G/A 0.23 0.26 0.87 .70-1.08) 0.91 .69-1.20) 0.68 .39-1.21) 0.20 0.561 rs230 8950 int G/A 0.02 0.02 0.73 .35-1.49) 0.73 .35-1.49) — 0.38 0.699 rs202 0902 int T/C 0.14 0.18 0.74 .57-0.95) 0.68 .51-0.92) 0.76 .33-1.75) 0.019 0.363 rs464 6018 int G/A 0.44 0.48 0.84 .69-1.02) 0.91 .67-1.24) 0.69 .47-1.03) 0.078 0.453 S9 (0 (0 (0
S9 rs2042370 int T/C 0.43 0.47 0.82 (0.68-1.00) 0.88 (0.65-1.19) 0.67 (0.46-0.98) 0.044 0.384S9 rs2308941 c-ns C/T 0.02 0.02 0.72 (0.36-1.45) 0.72 (0.36-1.45) — 0.36 0.675

erence sequence ID from dbSNP.P function as defined in dbSNP: c-ns, coding nonsynonymous; c-s, coding synonymous; int, intronic; I-r, variation in region ofe, but not in transcript; mrna-utr, variation in mRNA transcript, but not in coding region interval.

0 2849


werepatienMinnfor amediPatienma, lematchgraphRocheeligibAll paadminperiphextracdure (

GenoGen

typincandi(27).fromflamm9,412was susingincludtypingthesescribeCEPH

ples fand 2(37). Tstreamof ≥0platfothe as93.5%ple dthe 71forms475 cpassemastetheirthe ldroppwereFor thtarygenes

StatisAll

matedquencfreque(HWEexactevalu

Table sted CI) f SNP CasNHL 05)

Gene SNP Major ) )

Ordi Ptrend Ordi Ptrend

BAG5 rs76 C 1.24 (1 0.031 1.25 (0 0.13BCL2L rs674 G 0.82 (0 0.049 0.74 (0 0.049BCL2L rs1261 T/ 0.58 (0 0.0087 0.41 (0 0.022BCL2L rs724 C 0.89 (0 0.26 0.78 (0 0.14BCL2L rs616 A 0.87 (0 0.15 0.79 (0 0.13BCL2L rs675 C 0.84 (0 0.078 0.74 (0 0.057BCLAF rs797 C 1.38 (1 0.015 1.39 (0 0.10BCLAF rs703 C 1.42 (1 0.0073 1.43 (0 0.066CASP9 rs668 T/ 1.41 (1 0.0013 1.60 (1 0.0036CASP9 rs464 G 0.87 (0 0.20 1.07 (0 0.68CASP9 rs230 G 0.73 (0 0.38 0.21 (0 0.13CASP9 rs202 T/ 0.74 (0 0.019 0.85 (0 0.40CASP9 rs464 G 0.84 (0 0.078 0.65 (0 0.0049CASP9CASP9

llowin

Kelly et al.

Cance2850

Do


randomly selected from Mayo Clinic Rochesterts ages 20 years or older, who were residents ofesota, Iowa, or Wisconsin, and were being seenprescheduled medical examination in the generalcine divisions of the Department of Medicine.ts were not eligible if they had a history of lympho-ukemia, or HIV infection. Controls were frequencyed to cases by 5-year age group, gender, and geo-ic region (county groupings based on distance fromster, Minnesota and urban/rural status). Of the 818le controls, 572 (70%) participated in the study.rticipating subjects were asked to complete a self-istered risk-factor questionnaire and to provide aeral blood sample for genetic studies. DNA wasted from blood samples using a standard proce-Gentra, Inc.).

typingotyping reported here was part of a larger geno-g project to assess the role of immune and otherdate genes in the etiology and prognosis of NHLMost of the genes and SNPs reported here werethe ParAllele (now Affymetrix) Immune and In-ation SNP panel that included 1,253 genes andSNPs (35). The Immune and Inflammation panelupplemented by a second round of genotypinga custom Illumina Goldengate (36) OPA thated 384 SNPs from 100 candidate genes. Full geno-details and quality control measures for both of

genotyping platforms have been previously de-

(Continued on the fo

d (27, 28). Briefly, tagging SNPs were selected using(European-American) and Yoruba (African) sam-

Tabletified,



rom release 16 (Immune and Inflammation panel)1 (Illumina panel) of the HapMap Consortiumagging SNPs covered 5 kb upstream and down-of each gene with minor allele frequency (MAF)

.05 and pairwise r2 threshold of 0.8. Across bothrms, the overall sample success rate was >98%,say call rate was >93% (99.1% for ParAllele andfor Illumina), and the concordance rate of sam-

uplicates was >98%; the concordance rate amongSNPs that were duplicated across the two plat-was 99.7%. A total of 916 people (441 cases andontrols) were genotyped in both assays andd all quality control measures (28). This combinedr data set was restricted to subjects who reportedrace as Caucasian. After the duplicate SNPs withower platform-specific SNP call rates wereed and SNPs that had a MAF of <1% (n = 935)excluded, 8,034 SNPs remained in the data set.is analysis, we evaluated 226 SNPs (Supplemen-Table S1) from 22 BCL2 and 14 caspase family(Table 1).

tical analysisele frequencies from cases and controls were esti-using observed genotype frequencies. The fre-

ies in the controls were compared with genotypencies expected under Hardy-Weinberg equilibrium) using a Pearson goodness-of-fit test or Fisher'stest (MAF < 0.05). In this analysis, 14 of the 226ated SNPs had a HWE P < 0.05 (Supplementary

g page)

3. Adju(2002-20

ORs (95%
or selected s by NHL subtype, Mayo
S2); because no genotypethese SNPs were not excl

Cancer Epidemiology, B

10 American Association f

e-Control Study of

calling errors were iduded from analysis. W

iomarkers & Preventi

or Cancer Research.

ID*
/Minor All NHL (n = 441
nal OR†

CLL/SLL (n = 123

nal OR†

93
/T .02-1.50) .94-1.67) 11 6608 /A .68-1.00) .54-1.00) 11 3243 C .38-0.87) .19-0.88) 11 710 /T .72-1.09) .56-1.09) 11 130 /C .72-1.05) .59-1.07) 11 3785 /A .69-1.02) .55-1.01) 1 558 /G .07-1.80) .94-2.05) 1 193 /T .10-1.84) .98-2.10)
5648
C .14-1.73) .17-2.19) 6077 /A .70-1.08) .78-1.46) 8950 /A .35-1.49) .03-1.57) 0902 C .57-0.95) .58-1.24) 6018 /A .69-1.02) .49-0.88)
rs2042370 T/C 0.82 (0.68-1.00) 0.044 0.62 (0.46-0.83) 0.0015rs2308941 C/T 0.72 (0.36-1.45) 0.36 0.83 (0.28-2.49) 0.74

en-e

on


previoin ourTwo

tion bapproorthogSNP mnate,countwereSNP vthat athe SNsion mprincitiple ddecreacorrelgreessecontestedimplethe haponentests wInd

logistcorresfor hehomo

and cof varassumship.geneevaluest wand Cto simsubtyerogeSNP oTo

multimethoand Ssis thavariatstrengnumbpecteded q vassoci(pFDRTo

publi(32) fcontr

Table selected SN -Control StuNHL

3) 69) Phetero

Ordi Ptrend Ordi Ptrend

1.31 (0 0.084 1.67 (1 0.0074 0.40.87 (0 0.38 0.91 (0 0.63 0.50.57 (0 0.10 1.24 (0 0.52 0.00.94 (0 0.69 1.07 (0 0.73 0.30.88 (0 0.42 0.98 (0 0.93 0.60.94 (0 0.71 0.87 (0 0.49 0.41.24 (0 0.30 1.32 (0 0.27 0.81.32 (0 0.17 1.40 (0 0.16 0.91.21 (0 0.27 1.26 (0 0.27 0.30.79 (0 0.18 0.55 (0 0.016 0.00.92 (0 0.88 0.34 (0 0.30 0.30.72 (0 0.12 0.43 (0 0.0076 0.11.031.071.49

*Ref†OR‡P vregr


www.a

Do


usly found no evidence of population stratificationdata (27).methods were used when analyzing the associa-

etween each gene and case-control status. The firstach used a principal components analysis to createonal (e.g., uncorrelated) linear combinations of theinor allele count variables that provide an alter-

and equivalent, representation of the SNP genotypevariables. These component linear combinationsthen ranked according to the amount of the totalariance explained. The resulting smallest subsetccounted for at least 90% of the variability amongPs was included in a multivariable logistic regres-odel. A gene-specific global test using the resultantpal components was then carried out using a mul-egree-of-freedom likelihood ratio test. This methodses the dimensionality of the data when SNPs areated by reducing the number of independent de-of freedom that comprise the statistical test. Thed method in which the gene-level association wasused the global score test of Schaid et al. (38) as

mented in the S-Plus program Haplo.stats. Becauseplotype results were similar to the principal com-ts analysis, we only report these results. Gene-levelith P < 0.05 were declared of interest.

ividual SNPs were examined using unconditionalic regression to estimate odds ratios (OR) andponding 95% confidence intervals (CI) separately

ession.

terozygotes and minor allele homozygotes, usingzygotes for the major allele as the reference. ORs

genotpopul

Caacrjournals.org


orresponding 95% CIs were also estimated per copyiant allele for each SNP, and Ptrend was calculateding an ordinal (log-additive) genotypic relation-SNPs with Ptrend < 0.05 in the setting of a globaltest of P < 0.05 were declared of interest. We alsoated the association between SNPs in genes of inter-ith NHL risk by major NHL subtype (DLBCL, FL,LL/SLL). We used polytomous logistic regressionultaneously calculate ORs and 95% CIs for eachpe relative to controls, and to formally test for het-neity of the estimated association between eachf interest and lymphoma subtype (39).assess the robustness of our results in the setting ofple hypothesis testing, we used the tail strengthds of Taylor and Tibshirani (40) at both the geneNP level. This method tests the global null hypothe-t the distribution of P values from a large set of uni-e tests is uniformly distributed. As such, positive tailth values significantly >0 indicate that the observeder of small P values is greater than would be ex-by chance alone. In addition, we have also estimat-

alues at the SNP level to estimate the strength of theationwith respect to the positive false discovery rate; ref. 41).allow for SNP-level comparison with previouslyshed associations between BCL2 (33) and caspaseamily genes from a pooled analysis of three case-ol studies, we used the MACH 1.0.14 to impute

3. Adjusted ORs (95% CI) for(2002-2005) (Cont'd)

Ps by NHL subtype, Mayo Case

ypes for SNPs not directly observed in oation (42). The 60 unrelated HapMap CEU

ncer Epidemiol Biomarkers Prev; 19(11) Novem

10 American Association for Cancer Rese

dy of

FL (n = 11

nal OR†

DLBCL (n =

nal OR†
geneity
‡

.97-1.77)
.15-2.43) 8
.65-1.18)
.62-1.34) 7
.28-1.12)
.64-2.39) 47
.68-1.29)
.72-1.59) 9
.66-1.29)
.68-1.43) 5
.70-1.28)
.60-1.28) 5
.82-1.88)
.81-2.14) 9
.88-1.97)
.88-2.25) 4
.86-1.69)
.84-1.88) 6
.55-1.12)
.34-0.89) 49
.31-2.78)
.04-2.60) 4
.47-1.10)
.23-0.80) 7
.76-1.39)
.85-1.79) 098 (0 0.84 1.23 (0 0.27 0.0(0.80-1.44) 0.65 1.14 (0.79-1.64) 0.48 0.0046(0.61-3.63) 0.38 0.30 (0.04-2.31) 0.25 0.27
erence sequence ID from dbSNP.s adjusted for age and sex.alue for test of heterogeneity between the CLL/SLL, FL, and DLBCL NHL subtypes as determined by polytomous logistic

ur studysamples

ber 2010 2851

arch.


(fromdbSNmosomputedimputqualitNHLfromAn

SAS Isoftwand g

Resu

ParticThe

analy

58% w61.7 yactericomm113),

GeneWe

the 3all subysis, wed wiBCL2L0.009BCL2DFFB

Tableresul ublishe SNP- ts fromSouth HL Ca l Studie

Gene SNP ID* Position ctual res

W/V Varianfreq†

OR(95

Ptrend

BCL2L rs756 11157 Not e —BCL2L rs1338 11158 Not e —BCL2L rs1168 11158 Not e —BCL2L rs228 11158 Not e —BCL2L rs674 11160 G/A 0.42 0.82 (0 0.049BCL2L rs1261 11161 T/C 0.06 0.58 (0 0.0087BCL2L rs1748 11161 Not e —BCL2L rs376 11162 Not e —BCL2L rs724 11162 C/T 0.29 0.89 (0 0.27BCL2L rs378 11162 Not e —BCL2L rs1704 11162 Not e —BCL2L rs616 11162 A/C 0.43 0.87 (0 0.15BCL2L rs147 11163 Not e —BCL2L rs686 11163 Not e —BCL2L rs675 11164 C/A 0.42 0.84 (0 0.078BCL2L rs726 11164 Not e —BCL2L rs676 11164 Not e —CAPS9 rs464 15694 Not e —CAPS9 rs466 15695 Not e —CAPS9 rs668 15697 T/C 0.29 1.41 (1 0.0013CAPS9 rs464 15699 G/A 0.21 0.87 (0 0.20CAPS9 rs464 15704 Not e —CAPS9 rs230 15706 G/A 0.02 0.73 (0 0.38CAPS9 rs202 15706 A/G 0.16 0.74 (0 0.019CAPS9 rs464 15712 G/A 0.49 0.84 (0 0.078CAPS9CAPS9

llowin

Kelly et al.

Cance2852

Do


release 23a/phase II Mar08, NCBI Build 36,P build 12) were used to obtain the phased chro-es, and the expected genotype dosage was com-based on the posterior probability. SNPs withation r2 larger than 0.30 were deemed of sufficienty and were examined for their association withcase-control status using the SNP dosage estimatedMACH.alyses were implemented using SAS (version 8;nstitute) and S-Plus (version 7.05; Insightful Corp.)are systems. All analyses were adjusted for ageender.

lts

ipant characteristics

(Continued on the fo

re were 441 cases and 475 controls available forsis. For cases, the mean age was 60.1 years and

the sewere



ere male, whereas for controls the mean age wasears and 55% were male. Additional patient char-stics have been previously published (28). The moston NHL subtypes were SLL/CLL (n = 123), FL (n =and DLBCL (n = 69).

-level analysisfirst evaluated gene-level associations between6 candidate apoptosis pathway genes and NHL,types combined. Using principal components anal-e observed four genes to be significantly associat-

th NHL risk at P < 0.05 (Table 1): BAG5 (P = 0.026),11 (also known as BIM; P = 0.0019), BCLAF1 (P =7), and CASP9 (P = 0.0022). In addition, BCL2,L13, BCL2L14, BID, APAF1, CASP7, CASP10, andeach had one or more SNPs at P < 0.05 but in

g page)

4. Comparison of observed and imputed Mayo Case-Control Study of NHL (2002-2005) SNP-level
ts with pWales N
d observedse-Contro

level resuls

the pooled NCI-SEER, Connecticut, and New

tting ofnot cons

Cancer

10 Amer

Mayo a

a gene-level teidered further. A

Epidemiology,

ican Association

ults

st of P ≥ 0.05, and thll SNP-level associatio

Biomarkers & Preventi

for Cancer Research.

t
Ordinal% CI)
11
7444 9909 valuated 11 8646 1386 valuated 11 1263 4481 valuated 11 9321 6691 valuated 11 6608 9455 .68-1.00) 11 3243 3977 .38-0.87) 11 4848 7031 valuated 11 1704 0169 valuated 11 710 4162 .72-1.09) 11 9068 5718 valuated 11 1883 6213 valuated 11 130 9152 .72-1.05) 11 0053 2417 valuated 11 952 5817 valuated 11 3785 0101 .69-1.02) 11 430 7892 valuated 11 0053 9468 valuated
6092
260 valuated 1636 648 valuated 5648 782 .14-1.73) 6077 723 .70-1.08) 6047 370 valuated 8950 093 .35-1.49) 0902 947 .57-0.95) 6018 961 .69-1.02)
rs2042370 15714329 T/C 0.49 0.82 (0.68-1.00) 0.044s2308941 15717305 C/T 0.01 0.72 (0.36-1.45) 0.36

usns

on


from nTable

SNP-Nex

withinanaly(rs769P < 0For Bat P <creaselele coalleleof cor0.046)signif

moderate Ntivelsignif1.38 p1.42 pwereour psignifregressignagenotOR, 1OR, 0OR, 0

Tableresul dSouth

M s Poole act

W/V rianeq

OR(9

trend alit V rianeq†

R(95

rend

C/T .44 2 (0 56 .94 7 T .46 8 (0 055C/T .05 0 (5 001 .91 9 T .08 2 (0 8C/A .27 3 (1 036 .64 3 A .23 5 (0 8T/C .06 9 (0 14 .99 0 .06 7 (0 6

— — t e— — t e

T/C .11 9 (1 092 .79 2 .10 9 (0 7A/G .14 5 (1 004 .75 6 .12 1 (0 3

— — t eA/G .52 5 (1 018 .99 9 .46 3 (1 093C/A .11 3 (2 000 .79 3 A .12 8 (0 8

— t eG/T .21 4 (1 052 .66 5 T .17 5 (0 3C/A .28 5 (0 5 .96 0 A .29 7 (0 10

— t eT/C .20 1 (1 004 .67 3 .18 1 (0 96C/G .42 5 (1 031 .76 9 .44 1 (1 31C/T .21 7 (0 057 .88 8 T .23 9 (0 4C/T .39 4 (0 3 .68 5 T .33 8 (0 11

— — t e— — t e

C/T .51 8 (0 6 .85 2 T .46 0 (0 33— — — — t e— — — — .14 2 (0 2———

Abb*Ref†Var


www.a

Do


onsignificant genes are available in SupplementaryS2.

level analysist, we formally evaluated SNP-level associationsthe four genes with P < 0.05 from the gene-level

sis (Table 2). The single genotyped BAG5 tagSNP3) was significantly associated with NHL risk at.05 (OR, 1.24 per T allele copy; 95% CI, 1.02-1.50).CL2L11, two of the five tagSNPs were significant0.05, and variant alleles were associated with de-d NHL risk for both (rs6746608: OR, 0.82 per A al-py; 95% CI, 0.68-1.00; rs12613243: OR, 0.58 per Ccopy; 95% CI, 0.38-0.87). There was little evidencerelation between these two BCL2L11 SNPs (r2 =

iant allele frequency among the controls.

, and both genotyped SNPs remained statisticallyicant when modeled in a single logistic regression

wise0.079

Caacrjournals.org


l, confirming that these two SNPs represent sepa-HL risk signals (P = 0.011 and P = 0.0015, respec-y). Both genotyped BCLAF1 tagSNPs wereicantly associated with NHL risk (rs797558: OR,er G allele copy; 95% CI, 1.07-1.80; rs703193: OR,er T allele copy; 95% CI, 1.10-1.84). These two SNPsin strong linkage disequilibrium with each other inopulation (r2 = 0.96). Neither SNP reached statisticalicance when both were modeled in the same logisticsion model, indicating that they represent the samel related to NHL risk. Finally, three of the sevenyped CASP9 tagSNPs were significant (rs6685648:.41 per C allele copy; 95% CI, 1.14-1.73; rs2020902:.74 per C allele copy; 95% CI, 0.57-0.95; rs2042370:.82 per C allele copy; 95% CI, 0.68-1.00). The pair-

4. Comparison of observed and imputed Mayo Case-Control Study of NHL (2002-2005) SNP-level
ts with published observed SNP-level results from the pooleWales NHL Case-Control Studies (Cont'd)
correlationand 0.34);

ncer Epide

10 Americ

NCI-SEER, Connecticut, and New

ayo imputed result

of theshowever

miol Biom

an Asso

d study

e SNPs is f, they seem

arkers Prev

ciation for C

ual results

airly low (r2 betweto represent the sa

; 19(11) November 20

ancer Research.

Vafr

t
Ordinal5% CI)
P
Qu y r2 W/ Vafr
t O
ordinal% CI)
Pt

0
0.8 .67-1.01) 0.0 0 0.8 C/ 0 0.8 .80-0.96) 0.0 0 26. .04-134.06) 0.0 0 0 0.0 C/ 0 1.1 .95-1.34) 0.1 0 1.7 .20-2.50) 0.0 0 0.3 C/ 0 1.0 .94-1.17) 0.3 0 0.5 .38-0.90) 0.0 0 0.9 T/C 0 1.0 .89-1.28) 0.4
—
— No valuated — — No valuated
0
3.1 .33-7.65) 0.0 0 0.1 T/C 0 1.0 .94-1.26) 0.2 0 3.3 .70-6.58) 0.0 6 0 0.1 A/G 0 1.1 .97-1.28) 0.1
—
— No valuated 0 1.3 .12-1.64) 0.0 0 0.9 A/G 0 1.1 .03-1.24) 0.0 0 5.8 .55-13.37) 0.0 3 0 0.1 C/ 0 1.0 .94-1 24) 0.2
—
— No valuated 0 1.9 .22-3.10) 0.0 0 0.2 G/ 0 1.0 .93-1.18) 0.4 0 0.8 .68-1.06) 0.1 0 0.9 C/ 0 0.8 .79-0.97) 0.0
—
— No valuated 0 2.4 .47-3.94) 0.0 7 0 0.2 T/C 0 1.1 .98-1.24) 0.0 0 1.4 .13-1.86) 0.0 0 0.5 C/G 0 1.1 .01-1.21) 0.0 0 0.6 .51-0.89) 0.0 0 0.6 C/ 0 1.0 .97-1.21) 0.1 0 0.8 .63-1.11) 0.2 0 0.4 C/ 0 0.8 .80-0.97) 0.0
—
— No valuated — — No valuated
0
0.8 .70-1.10) 0.2 0 0.7 C/ 0 0.9 .82-0.99) 0.0 — — No valuated — — A/G 0 1.0 .89-1.16) 0.8 — — — No valuated — — — — t e
— — — — — — — Not evaluated— — — — — — — Not evaluated

reviations: W, wild type allele; V, variant allele.erence sequence ID from dbSNP.

enme

10 2853


signaeithermodeaddedratio tnificathe exnonco

MultiTo e

bothgene-spectthe 36whereP valuestimaand Sbutionexpecassociworthgenesthe SNof this

SubtyIn e

sessedPhetero

sion)and BIn conbetwedifferBCL2LvariansubtywithlimitepattersubtyalleleDLBCdecreaallelelimiters2042SNPsrantedsamp

CompThe

studyies frothat w

(BAXCASPgenesNHLwithCASPwith(BCL2leveltwo sand imtrates(basedBCL2LFor

typedstudyto bestudyThe esimilaand rand rof thethe asthe MsubtyconsissociatThe

studieservedlele aPtrend

OR =studyto bepoolenitudMayoinversat bostudiein theThe

so wetwo gtypingablecompcontro

Discu

In tlevel

Kelly et al.

Cance2854

Do


l when jointly modeled. That is, when we addedrs2020902 or rs2042370 to the logistic regressionl with the most significant SNP, rs6685648, theSNP was nonsignificant based on the likelihood

est (P > 0.1). Of note, all of the SNPs that were sig-ntly associated with NHL risk were intronic, withception of the BAG5 SNP rs7693, which is in theding region interval of an mRNA transcript.

ple testingvaluate the effect of multiple testing, we estimatedthe tail strength of the P values generated in theand SNP-level analyses and the q values with re-to the pFDR for each SNP. The tail strength forgene-level P values was 0.38 (95% CI, 0.05-0.70),as the tail strength estimate for the 226 SNP-leveles was 0.20 (95% CI, 0.07-0.33). As the tail strengthtes and 95% CIs exclude the null in both the gene-NP-level analyses, we can conclude that the distri-of P values is more extreme than we would have

ted by chance alone, and thus, observed significantations at both the gene and SNP level remain note-y. However, individual SNP q values within theof interest (Table 2) ranged from 0.22 to 0.38 inPs we have considered significant for the purposeanalysis.

pe analysisxploratory analyses, similar associations (as as-by direction and magnitude of ordinal ORs and

geneity obtained from polytomous logistic regres-were observed for significant SNPs from BCLAF1AG5 for the subtypes of CLL/SLL, FL, and DLBCL.trast, there is some evidence that the associationsen individual tagSNPs within BCL2L11 and CASP9among the three NHL subtypes (Table 3). For11, the decreased risk of NHL with copies of thet A allele at rs6746608 seems similar across NHLpe (Pheterogeneity = 0.57), but the decreased NHL riskcopies of the variant C allele at rs12613243 seemsd to CLL/SLL and FL (Pheterogeneity = 0.047). Then for CASP9 SNPs seemed most differential bype. The decreased NHL risk with copies of the Aat both rs4646077 and rs2020902 was limited toL (Pheterogeneity = 0.049 and 0.17, respectively); thesed NHL risk with copies of the A and C variants at rs4646018 and rs2042370, respectively, wasd to CLL/SLL (rs4646018, Pheterogeneity = 0.0098;370, Pheterogeneity = 0.0046); the remaining CASP9had Pheterogeneity > 0.25. However, caution is war-in interpreting the subtype results due to small

le sizes, especially SNPs with lower MAFs.

arison with published estimatesre were 17 genes that were evaluated in both thisand the pooled analysis of three case-control stud-
m the United States and Australia (32, 33): 13 genesere not associated with NHL risk in either study
CASPcount



, BCL2, BCL2A1, BCL2L1, BCL2L2, BCL2L10, CASP2,3, CASP4, CASP5, CASP6, CASP7, and CASP10), 2(CASP1 and CASP8) that were associated withrisk in the pooled study but were not associatedNHL risk in our study, and 2 genes (BCL2L11 and9) that were significantly associated at the gene levelNHL risk in both studies. For these two genesL11 and CASP9), we compared individual SNP-significance (based on ordinal ORs) between thesetudies. The results for the overlapping observedputed SNPs are presented in Table 4. Figure 1 illus-the relative position and linkage disequilibriumon HapMap CEPH population genotypes) of each11 (A) and CASP9 (B) genotyped SNPs.BCL2L11, there was no overlap in the SNPs geno-in the two studies. For the Mayo case-control

, we imputed genotypes for the four SNPs observedsignificantly associated with NHL risk in the pooled(rs7567444, rs3789068, rs686952, and rs6760053).stimated NHL risk with variant allele copies wasr in direction and magnitude across all four SNPs,eached significance for rs3789038 (Ptrend = 0.0018)s6760053 (Ptrend = 0.0031). In exploratory analysesse four SNPs by NHL subtype, we observed thatsociation between the imputed SNP genotypes inayo population were strongest in the CLL/SLLpe but were also observed in FL at a magnitudetent with the pooled study, which observed an as-ion specific to the FL subtype (Table 5).CASP9 SNP rs2020902 was genotyped in boths, although the results were inconsistent. The ob-association between copies of the variant G al-

nd NHL risk was OR = 0.74 (95% CI, 0.57-0.95;= 0.019) in the Mayo case-control study and1.02 (95% CI, 0.89-1.16; Ptrend = 0.82) in the pooled. Alternatively, two CASP9 SNPS were observedsignificantly associated with NHL risk in the

d study (rs4661636 and rs4646047), and the mag-e and direction of the estimate associated withimputed genotypes was consistent for both. Ane association with the number of variant T allelesth rs4661636 and rs4646047 was observed in boths, although associations did not reach significanceMayo case-control study.pooled study did not genotype BCLAF or BAG5,were not able to compare our findings for theseenes. Further, we did not have the original geno-data from the pooled analysis, so we were not

to impute genotypes in their population forarison with SNPs significant in the Mayo case-l study.

ssion

his case-control study, we show gene- and SNP-association of BCL2L11 (BIM), BCLAF1, BAG5, and
9with NHL risk that remained noteworthy after ac-ing for multiple testing. Whereas the pFDR q values
Cancer Epidemiology, Biomarkers & Prevention



FigureConneccorrelatgenotyp


www.a

Do


1. Linkage disequilibrium plot of SNPs genotyped in BCL2L11 (A) and CASP9 (B), Mayo Clinic Case-Control Study of NHL (2002-2005) and SEER/ticut/New South Wales pooled case-control study of NHL. The numbers indicate D′ values; the darker shading indicates higher r2 values of
ion between SNPs. ORs and 95% CIs for significant SNPs genotyped in the SEER/Connecticut/New South Wales pooled study and eithered or imputed in the current Mayo NHL case-control study are indicated above each SNP rs number.
Cancer Epidemiol Biomarkers Prev; 19(11) November 2010acrjournals.org 2855

on August 15, 2019. © 2010 American Association for Cancer Research. cebp.aacrjournals.org wnloaded from


estimathereindivitimategroupwe wthe siCASPfromInstitu(NCI-cludetheregenotwerethat wtionsrectioof thrthe firsociatnot brequirStre

controgenotbasedto tho(Minneffectof usifrom tto tagjects,lationof an

eralizparticassociin this<70%theseaddititifiedgermlgenesand danalyinvolvThe

plausantiappoptoway (selectlencinservetumotein emorscolonciatioand othe SN74% gvariatbe covariattion fhypot

TableSNP- w ed P- poConn N Wa St

SNP I

Mayo resrs75 C 3 (0 59 7 (0 1 1 (0 4rs37rs68rs67 C 8 (1 077 1 (0 7 0 (0 9

Pooled edrs75 C 3 (0 4 9 (0 009 1 (0 5rsrsrs

*Ref

Kelly et al.

Cance2856

Do


ted for BCL2L11 and CASP9 SNPs do indicate thatis a moderate chance that any of these associationsdually may be false positives, the tail strength es-s indicated that the distribution of P values for theof SNPs from these genes is more extreme thanould have expected by chance alone. Moreover,gnificant gene-level associations for BCL2L11 and9 were consistent with previously published dataa pooled analysis of three studies [National Cancerte–Surveillance, Epidemiology, and End ResultsSEER), Connecticut, and New South Wales] that in-d 1,946 cases and 1,808 controls (32, 33). Althoughwas minimal overlap in the individual tagSNPsyped across our study and the pooled study, weable to impute genotypes for the SNPs in our studyere not directly observed, and we found associa-that were largely consistent in magnitude and di-n across all four significant BCL2L11 SNPs and twoee CASP9 SNPs from the pooled study, making thisst independent replication of these results. The as-ions from our study with BCLAF1 and BAG5 haveeen tested in an independent data set and thuse replication.ngths of our study include a carefully designed case-l study, central pathology review, and high qualityyping. Although this study was not population, both case and control participation was restrictedse residing in the region surrounding Mayo Clinicesota, Iowa, and Wisconsin), thus minimizing theof referral bias and increasing the internal validityng frequency-matched general medicine controlshe same region. CommonHapMap SNPs were usedgenes of interest, and through other genotyping pro-we have ruled out the presence of significant popu-

erence sequence ID from dbSNP.

stratification (27). The major limitations are the useexclusively Caucasian population, which limits gen-

maticby the



ability, and the relatively small sample size, which inular precludes robust estimation of NHL subtypeations. There are several apoptosis genes evaluatedstudy for which the genotyped tagSNPs providedgene coverage, and thus, an association betweengenes and NHL risk cannot yet be ruled out. Inon, there are >100 human genes more broadly iden-with apoptosis pathway involvement in whichine variants may play a role in NHL risk, includingwith caspase recruitment domains, death domains,eath effector domains (43). The genes in the currentsis represent only the core of a larger set of genesed in apoptosis-related pathways.two genes that replicated have strong biological

ibility in NHL pathogenesis. BCL2L11 balances theoptotic influence of BCL2 and coordinates proa-tic signaling through the intrinsic apoptosis path-4). In addition, BCL2L11 is required for negativeion of autoreactive lymphocytes (44). Functional si-g of BCL2L11 through methylation has been ob-d in Burkitt lymphoma cell lines and primaryr biopsies, and reduced BCL2L11 mRNA and pro-xpression has also been documented in other tu-including renal cell carcinoma, melanoma, andcancer (45). Of note, we did not identify any asso-n between BCL2 and NHL risk at the gene level,nly 1 of the 53 genotyped SNPs was significant atP level. Although these 53 SNPs comprised onlyene coverage, it does suggest that BCL2 germlineion may not play a role in NHL risk. This wouldnsistent with the hypothesis that most of the BCL2ion in lymphoma tumors is a result of hypermuta-ollowing the t(14;18) translocation event (4). Thishesis should be further explored by comparing so-

5. Comparison of observed and imputed Mayo Case-Control Study of NHL (2002-2005) BCL2L11
level resultsecticut, and
ith publishew South

observed BCL2L11 SNles NHL Case-Control

1

level results from theudies

and germline mutation amont(14;18) translocation.

Cancer Epidemiology, Biom

0 American Association for C

oled NCI-SEER,

g patients stratifie

arkers & Preventio

ancer Research.

D* Major/minor CLL/SLL FL DLBCL

Ordinal OR Ptrend Ordinal OR Ptrend Ordinal OR Ptrend

study (imputed
ults) 67444 /T 0.7 .53-1.01) 0.0 0.8 .64-1.20) 0.4 0.9 .61-1.36) 0.6 89068 G 1.5 .16-2.10) 0.0 1.2 .95-1.72) 0.1 1.0 .72-1.52) 0.7 A/ 6 (1 032 8 (0 0 5 (0 96952 C/A 0.74 (0.52-1.06) 0.097 0.91 (0.65-1.29) 0.60 1.01 (0.66-1.54) 0.98 60053 /G 1.6 .15-2.46) 0.0 1.3 .89-1.92) 0.1 1.1 .68-1.79) 0.6 study (observ results) 67444 /T 0.9 .74-1.17) 0.5 0.7 .68-0.91) 0.0 0.9 .79-1.04) 0.1 89068 G 1.0 .82-1.29) 0.8 1.2 .12-1.47) 0.0 1.1 .97-1.26) 0.1 37 A/ 3 (0 0 8 (1 004 0 (0 4
686952 C/A 0.93 (0.72-1.20) 0.56 0.78 (0.67-0.92) 0.0023 0.87 (0.75-1.01) 0.0786760053 C/G 1.00 (0.79-1.26) 0.99 1.19 (1.03-1.36) 0.015 1.09 (0.95-1.24) 0.22

d

n


CAtotic pand isoptosto cytphospptosisbeen son mitrinsicproteiturn areachalthousignifCASPvidualow-uTo

ationwithboth Btotic)geneshoulstudyAlth

data pspecif

ly BCtion owas ostudielargelFL anIn c

four gand tlightmankey p

Discl

No p

Ackn

We t

Grant

NIHported

Thepaymenadvertisthis fac

Refe1. Jem

2002. Ale

maSu

3. MalicuJ H

4. Rery

5. Rohea

6. Dymo77:

7. LinTraof16

8. Code

9. Bulym

10. LamCaDif

11. BalymAn

12. Hu


www.a

Do


SP9, the other gene to be replicated, is a proapop-rotease integral to the intrinsic apoptotic pathway,responsible for effector caspase activation and ap-

is execution following activation by Apaf-1 boundochrome c released from mitochondria (46). Hyper-horylation of caspase-9 may lead to aberrant apo-inhibition, and the relevance of this process hashown in several other cancer types (46). Also of note,tochondrial release of cytochrome c via both the in-and extrinsic apoptosis pathways, the cytoplasmicnApaf-1 binds caspase-9 to form the apoptosome, inctivating the caspase cascade (4). APAF1 did notgene-level statistical significance in our population,gh 6 of the 13 genotyped tagSNPs in this gene wereicant at P < 0.05. Given the gene-level significance of9, there may be some clinical relevance of the indi-l tagSNP significance in APAF1, and further fol-p on this gene is warranted.our knowledge, this is the first report of an associ-between germline variation in BAG5 and BCLAF1regard to lymphoma risk. BCLAF1 and BAG5 arecl-2 family members that suppress BAX (proapop-gene expression, in turn suppressing the APAF1and inhibiting apoptosis (4, 47). These associationsd be confirmed in follow-up with an independentpopulation.ough underpowered to assess NHL subtypes, our
rovide some evidence that there may be subtype-ic associations in the apoptosis pathway, particular-
Recepublish

phoma-associated genetic aberrations in healthy individuals.n Hematol 2002;81:64–75.i D, Satkunam N, Al Kaptan M, Reiman T, Lai R. Pathway-specific

apan

13. SclatInt

14. Soin

15. Shge

16. Yaan20

17. Ruap20

18. AliB-20

19. Waetipona20

20. Zhanph

21. SkHa

22. NoB-

Caacrjournals.org


L2L11 and CASP9. Evidence of differential associa-f germline variants in the BCL2 and CASP familiesbserved in the pooled analysis of three case-controls (32, 33), although the BCL2L11 association wasy limited to FL (33), a pattern we observed for bothd CLL/SLL in our study.onclusion, our results support an association ofenes from the apoptosis pathway, with NHL risk,hese associations may vary by NHL subtype. Inof the importance of the apoptosis pathway to hu-lymphomagenesis, further characterization of thelayers within this pathway is warranted.

osure of Potential Conflicts of Interest

otential conflicts of interest were disclosed.

owledgments

hank Sondra Buehler for her editorial assistance.

Support

, National Cancer Institute grant R01 CA92153. J.L. Kelly was sup-by NIH, National Heart Lung and Blood Institute grant HL007152.costs of publication of this article were defrayed in part by thet of page charges. This article must therefore be hereby markedement in accordance with 18 U.S.C. Section 1734 solely to indicatet.

ived 06/03/2010; revised 08/17/2010; accepted 08/30/2010;ed OnlineFirst 09/20/2010.

rencesal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics,9. CA Cancer J Clin 2009;59:225–49.xander DD, Mink PJ, Adami HO, et al. The non-Hodgkin lympho-s: a review of the epidemiologic literature. Int J Cancer 2007;120ppl 12:1–39.sir N, Campbell LJ, Goff LK, et al. BCL2 protein expression in fol-lar lymphomas with t(14;18) chromosomal translocations. Braematol 2009;144:716–25.ed JC. Bcl-2-family proteins and hematologic malignancies: histo-and future prospects. Blood 2008;111:3322–30.ssi D, Gaidano G. Messengers of cell death: apoptotic signaling inlth and disease. Haematologica 2003;88:212–8.er MJ. The pathogenetic role of oncogenes deregulated by chro-somal translocation in B-cell malignancies. Int J Hematol 2003;315–20.P, Jetly R, Lennon PA, Abruzzo LV, Prajapati S, Medeiros LJ.nslocation (18;22) (q21;q11) in B-cell lymphomas: a report4 cases and review of the literature. Hum Pathol 2008;39:64–72.ry S, Adams JM. The Bcl2 family: regulators of the cellular life-or-ath switch. Nat Rev Cancer 2002;2:647–56.ggins AG, Pepper CJ. The role of Bcl-2 family proteins in chronicphocytic leukaemia. Leuk Res;34:837–42.kanfiM, FestjensN,DeclercqW,VandenBergheT,VandenabeeleP.

spases in cell survival, proliferation and differentiation. Cell Deathfer 2007;14:44–55.secke J, Griesinger F, Trumper L, Brittinger G. Leukemia- and

optotic gene expression profiling in chronic lymphocytic leukemiad follicular lymphoma. Mod Pathol 2006;19:1192–202.huler F, Dolken L, Hirt C, et al. Prevalence and frequency of circu-ing t(14;18)-MBR translocation carrying cells in healthy individuals.J Cancer 2009;124:958–63.ung YH, Lee JW, Kim SY, et al. Somatic mutations of CASP3 genehuman cancers. Hum Genet 2004;115:112–5.in MS, Kim HS, Kang CS, et al. Inactivating mutations of CASP10ne in non-Hodgkin lymphomas. Blood 2002;99:4094–9.manaka K, Clark R, Dowgiert R, et al. Expression of interleukin-18d caspase-1 in cutaneous T-cell lymphoma. Clin Cancer Res06;12:376–82.mmel MJ, de Vos S, Hoelzer D, Koeffler HP, Hofmann WK. Alteredoptosis pathways in mantle cell lymphoma. Leuk Lymphoma04;45:49–54.zadeh AA, Eisen MB, Davis RE, et al. Distinct types of diffuse largecell lymphoma identified by gene expression profiling. Nature00;403:503–11.ng SS, Slager SL, Brennan P, et al. Family history of hematopoi-c malignancies and risk of non-Hodgkin lymphoma (NHL): aoled analysis of 10 211 cases and 11 905 controls from the Inter-tional Lymphoma Epidemiology Consortium (InterLymph). Blood07;109:3479–88.ang Y, Wang R, Holford TR, et al. Family history of hematopoieticd non-hematopoietic malignancies and risk of non-Hodgkin lym-oma. Cancer Causes Control 2007;18:351–9.ibola CF, Curry JD, Nieters A. Genetic susceptibility to lymphoma.
ematologica 2007;92:960–9.vak AJ, Slager SL, Fredericksen ZS, et al. Genetic variation incell-activating factor is associated with an increased risk of
ncer Epidemiol Biomarkers Prev; 19(11) November 2010 2857



de421

23. PutioNe200

24. RoIL1Lym

25. Purecgen

26. WatheNFOn

27. Ce12ph

28. CetumandPre

29. ScNBHo

30. Skare200

31. SlaforPre

32. Lanpaof264

33. Moma

cycBio

34. Jafclatop

35. Hainvin

36. Olinothr

37. ThPro

38. Sctesph

39. HoJo

40. Taall

41. Sto20

42. NoAGe

43. Kitap21

44. BomeNa

45. Pinsta

46. Alca

Kelly et al.

Cance2858

Do


veloping B-cell non-Hodgkin lymphoma. Cancer Res 2009;69:7–24.rdue MP, Lan Q, Kricker A, et al. Polymorphisms in immune func-n genes and risk of non-Hodgkin lymphoma: findings from thew South Wales non-Hodgkin Lymphoma Study. Carcinogenesis7;28:704–12.thman N, Skibola CF, Wang SS, et al. Genetic variation in TNF and0 and risk of non-Hodgkin lymphoma: a report from the Inter-ph Consortium. Lancet Oncol 2006;7:27–38.

rdue MP, Lan Q, Wang SS, et al. A pooled investigation of Toll-likeeptor gene variants and risk of non-Hodgkin lymphoma. Carcino-esis 2009;30:275–81.ng SS, Purdue MP, Cerhan JR, et al. Common gene variants intumor necrosis factor (TNF) and TNF receptor superfamilies and-κB transcription factors and non-Hodgkin lymphoma risk. PLoSe 2009;4:e5360.rhan JR, Ansell SM, Fredericksen ZS, et al. Genetic variation in53 immune and inflammation genes and risk of non-Hodgkin lym-oma. Blood 2007;110:4455–63.rhan JR, Liu-Mares W, Fredericksen ZS, et al. Genetic variation inor necrosis factor and the nuclear factor-κB canonical pathwayrisk of non-Hodgkin's lymphoma. Cancer Epidemiol Biomarkersv 2008;17:3161–9.huetz JM, MaCarthur AC, Leach S, et al. Genetic variation in theS1, MRE11, RAD50 and BLM genes and susceptibility to non-dgkin lymphoma. BMC Med Genet 2009;10:117.ibola CF, Bracci PM, Halperin E, et al. Genetic variants at 6p21.33associated with susceptibility to follicular lymphoma. Nat Genet9;41:873–5.ger SL, Goldin LR, Strom SS, et al. Genetic susceptibility variantschronic lymphocytic leukemia. Cancer Epidemiol Biomarkersv 2010;19:1098–102.Q, Morton LM, Armstrong B, et al. Genetic variation in cas-

se genes and risk of non-Hodgkin lymphoma: a pooled analysisthree population-based case-control studies. Blood 2009;114:
–7.rton LM, Purdue MP, Zheng T, et al. Risk of non-Hodgkin lympho-associated with germline variation in genes that regulate the cell
47. Acofmu



le, apoptosis, and lymphocyte development. Cancer Epidemiolmarkers Prev 2009;18:1259–70.fe ES, Harris N, Stein H, Vardiman J. World Health Organizationssification of tumours, pathology, and genetics: tumors of hema-oietic and lymphoid tissues. Lyon: IARC Press; 2001.rdenbol P, Yu F, Belmont J, et al. Highly multiplexed molecularersion probe genotyping: over 10,000 targeted SNPs genotypeda single tube assay. Genome Res 2005;15:269–75.phant A, Barker DL, Stuelpnagel JR, Chee MS. BeadArray tech-logy: enabling an accurate, cost-effective approach to high-oughput genotyping. Biotechniques 2002;Suppl:56–8, 60–1.e International HapMap Consortium. The International HapMapject. Nature 2003;426:789–96.haid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA. Scorets for association between traits and haplotypes when linkagease is ambiguous. Am J Hum Genet 2002;70:425–34.smer DW, Lemeshow S. Applied logistic regression. New York:hn Wiley & Sons, Inc.; 2000.ylor J, Tibshirani R. A tail strength measure for assessing the over-univariate significance in a dataset. Biostatistics 2006;7:167–81.rey JD. A direct approach to false discovery rates. J R Stat Soc B02;64:479–98.thnagel M, Ellinghaus D, Schreiber S, Krawczak M, Franke A.comprehensive evaluation of SNP genotype imputation. Humnet 2009;125:163–71.ada S, Pedersen IM, Schimmer AD, Reed JC. Dysregulation ofoptosis genes in hematopoietic malignancies. Oncogene 2002;:3459–74.uillet P, Purton JF, Godfrey DI, et al. BH3-only Bcl-2 familymber Bim is required for apoptosis of autoreactive thymocytes.ture 2002;415:922–6.on JD, Labi V, Egle A, Villunger A. Bim and Bmf in tissue homeo-sis and malignant disease. Oncogene 2009;27 Suppl 1:S41–52.lan LA, Clarke PR. Apoptosis and autophagy: regulation ofspase-9 by phosphorylation. FEBS J 2009;276:6063–73.hiron A, Feldman A, Mandel M, Gurevich M. Impaired expression
peripheral blood apoptotic-related gene transcripts in acuteltiple sclerosis relapse. Ann N Y Acad Sci 2007;1107:155–67.
Cancer Epidemiology, Biomarkers & Prevention



2010;19:2847-2858. Published OnlineFirst September 20, 2010.Cancer Epidemiol Biomarkers Prev Jennifer L. Kelly, Anne J. Novak, Zachary S. Fredericksen, et al.

Hodgkin's Lymphoma−Non Germline Variation in Apoptosis Pathway Genes and Risk of

Updated version

10.1158/1055-9965.EPI-10-0581doi:

Access the most recent version of this article at:

Material

Supplementary

http://cebp.aacrjournals.org/content/suppl/2010/09/20/1055-9965.EPI-10-0581.DC1

Access the most recent supplemental material at:

Cited articles

http://cebp.aacrjournals.org/content/19/11/2847.full#ref-list-1

This article cites 44 articles, 13 of which you can access for free at:

Citing articles

http://cebp.aacrjournals.org/content/19/11/2847.full#related-urls

This article has been cited by 4 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cebp.aacrjournals.org/content/19/11/2847To request permission to re-use all or part of this article, use this link

on August 15, 2019. © 2010 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from


http://cebp.aacrjournals.org/lookup/doi/10.1158/1055-9965.EPI-10-0581

http://cebp.aacrjournals.org/content/suppl/2010/09/20/1055-9965.EPI-10-0581.DC1

http://cebp.aacrjournals.org/content/19/11/2847.full#ref-list-1

http://cebp.aacrjournals.org/content/19/11/2847.full#related-urls

http://cebp.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cebp.aacrjournals.org/content/19/11/2847


& prevention germline variation in apoptosis pathway genes...

Documents