short report

Common variants within 6p21.31 locus are associated with chronic lymphocytic leukaemia and, potentially, other non-Hodgkin lymphoma subtypes

Susan L. Slager,1 Nicola J. Camp,2 Lucia Conde,3 Tait D. Shanafelt,1 Sara J. Achenbach,1 Kari G. Rabe,1 Neil E. Kay,1 Anne J. Novak,1 Timothy G. Call,1 Paige M. Bracci,4 Fenna M. C. Sille,3 Sylvia Sanchez,3 Nicholas K. Akers,3 Julie M. Cunningham,1 Daniel J. Serie,1 Shannon K. McDonnell,1 Jose F. Leis,1 Alice H. Wang,1 J. Brice Weinberg,5,6 Martha Glenn,2 Brian Link,7 Celine M. Vachon,1 Mark C. Lanasa,5 Christine F. Skibola3 and James R. Cerhan1 1

Mayo Clinic College of Medicine, Rochester,

MN, 2University of Utah School of Medicine,

Summary A recent meta-analysis of three genome-wide association studies of chronic lymphocytic leukaemia (CLL) identified two common variants at the 6p21.31 locus that are associated with CLL risk. To verify and further explore the association of these variants with other non-Hodgkin lymphoma (NHL) subtypes, we genotyped 1196 CLL cases, 1699 NHL cases, and 2410 controls. We found significant associations between the 6p21.31 variants and CLL risk (rs210134: P = 0
01; rs210142: P = 6
8 9 10 3). These variants also showed a trend towards association with some of the other NHL subtypes. Our results validate the prior work and support specific genetic pathways for risk among NHL subtypes.

Keywords: chronic lymphocytic leukaemia, non-Hodgkin lymphoma, single nucleotide polymorphisms, BAK1, risk locus.

Salt Lake City, UT, 3Division of Environmental Health Sciences, School of Public Health, University of California Berkeley, Berkeley, CA, 4

Department of Epidemiology and Biostatistics,

School of Medicine, University of California San Francisco, San Francisco, CA, 5Duke University Medical Center, Durham, NC, 6VA Medical Center, Durham, NC and 7University of Iowa College of Medicine, Iowa City, IA, USA Received 12 July 2012; accepted for publication 27 August 2012 Correspondence: Dr Susan L. Slager, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA. E-mail: [email protected]

Chronic lymphocytic leukaemia (CLL) is a B-cell malignancy and one of the most common non-Hodgkin lymphomas (NHL). Through a meta-analysis of three genome-wide association studies (GWAS) of CLL, we provided evidence that two common variants (rs210134, rs210142) at the 6p21.31 locus contribute to the heritability of CLL (Slager et al, 2012). These two variants are highly correlated with each other (r2 = 0
8) with rs210142 located within intron 1 of BAK1 and rs210134 located 100 kb telomeric to BAK1. Given that replication is important for validation of the findings, we evaluated the association of these two variants with CLL risk in additional independent samples from First published online 1 October 2012 doi: 10.1111/bjh.12070

1196 CLL cases and 2142 controls collected from five CLL studies. Further, because NHL comprises a group of closely related B- and T- cell neoplasms, we explored the association of these CLL-associated variants in 1699 patients with other common NHL subtypes [583 diffuse large B-cell lymphomas (DLBCL), 585 follicular lymphomas (FL), 229 marginal zone lymphomas (MZL), 156 T-cell lymphomas (TCL), and 135 mantle cell lymphomas (MCL)]. Finally, we sequenced the BAK1 gene in germline DNA from 67 CLL cases and 39 controls to identify functional variants that are correlated with either rs210134 or rs210142. ª 2012 Blackwell Publishing Ltd British Journal of Haematology, 2012, 159, 572–576

Short Report

Methods Study participants Five studies contributed Caucasian samples for genotyping (Table SI). The Mayo Clinic Lymphoma SPORE case-control study is a clinic-based study of incident cases and frequencymatched controls (based on age, sex, and residence) conducted in Rochester, Minnesota, between 2002 and 2008. The University of Utah CLL case-control study ascertained cases through the Huntsman Cancer Hospital clinics and the Utah Cancer Registry between 1979 and 2011. Controls were frequency matched (based on age, sex, and residence) and ascertained via the Utah Population Database. The San Francisco (SF) Bay Area study is a population-based case-control study of NHL and included incident cases diagnosed between 2001 and 2006 and controls that were frequency matched to cases by age in five-year groups, sex, and county. Duke University ascertained incident CLL cases through the Duke Haematology clinic between 1999 and 2011. Finally, the Genetic Epidemiology of CLL (GEC) Consortium started in 2004 and is an on-going, family-based study in which families with two or more members with prevalent CLL are recruited through haematology clinics or through the internet. The NHL diagnoses were confirmed by study pathologists and classified according to the World Health Organization classification (Harris et al, 2008; Swerdlow, 2008). Each study was approved by its respective institutional review boards; all participants provided written informed consent.

Genotyping Genotyping of the GEC/Mayo Clinic samples was performed at Mayo Clinic as part of a larger genotyping project using a custom Illumina Infinium array (Illumina Inc., San Diego, CA, USA). Genotyping of the University of Utah/Duke University samples was performed using Illumina Goldengate assay (Illumina Inc.). Genotyping of the SF samples was done using Taqman (Applied Biosystems, Carlsbad, CA, USA) and was corroborated using Sanger sequencing. Standard genotyping quality control procedures were performed at each genotyping centre and included duplicate samples within centre, dropping samples with call rates 99% genotyping concordance among duplicate samples within genotyping centres.

Statistical analysis Main analyses used SASv9.2. The association between each single nucleotide polymorphism (SNP) and NHL risk was assessed by the Cochran-Armitage trend test. Odds ratios (OR) and 95% confidence intervals (CI) were calculated using logistic regression with and without adjustment for age and sex covariates. Meta-analysis of the results reported here ª 2012 Blackwell Publishing Ltd British Journal of Haematology, 2012, 159, 572–576

and our previously reported results (Slager et al, 2012) was conducted under a fixed-effects model. Cochran’s Q statistic, to test for heterogeneity (P.het), and the I2 statistic, to quantify the proportion of the total variation due to heterogeneity, were calculated (Higgins & Thompson, 2002). Linkage disequilibrium (LD) metric (r2) between rs210134/rs210142 and other variants in the region were calculated using the CEU [Utah residents (Centre d’Etude du Polymorphisme Humain) with Northern and Western European ancestry] samples from version 2 of 1000 Genome data. Case-only analysis of rs210142 genotypes with either CLL platelet counts or CLL Rai stage was conducted using Kruskal–Wallis test or chi-square test, respectively.

Sequencing To identify potential functional variants, we sequenced the BAK1 gene in 67 CLL cases from CLL pedigrees obtained in the GEC consortium and 39 controls from the Mayo Clinic biobank (Slager et al, 2011). Sequencing of the exons was performed using Agilent SureSelect capture 50 Mb kit (Agilent Technologies, Inc., Santa Clara, CA, USA). Briefly, 100 bp paired-end sequencing reads were aligned to Build 37 using Novoalign software (Novocraft Technologies, Selangor, Malaysia) and realignment was performed using genome analysis toolkit (GATK)(McKenna et al, 2010). Single-sample variant calling was performed using the GATK unified genotyper. Variants with total read-depth >9 and high-quality scores (>Q20) were included in analyses. Frequency of the variants was compared between CLL cases and controls using a chi-square test.

Results and discussion A total of 1198 CLL cases, 1699 other NHL cases, and 2410 controls were available for genotyping. However, two genotyping teams (Duke/Utah and Mayo/GEC) were only able to successfully genotype one of the two SNPs. rs210142 had low call rate for the Duke/Utah samples and was dropped from statistical analyses; rs210134 could not be multiplexed with the other variants genotyped on the Mayo/GEC Infinium array. For the SNPs that were successfully genotyped, the SNP call rates were >92% and did not differ significantly between cases and controls. Further, because these two variants are highly correlated with each other, the results were expected to be similar across these two variants; this was confirmed in our results (Table I). For both variants, we found an association with CLL risk (P < 0
01 for the combined samples) and similar effect sizes to each other and to our previously reported findings (Slager et al, 2012). We report the ordinal OR = 0
82 (95% CI: 0
71, 0
95) and OR = 0
78 (95% CI: 0
65, 0
93) for rs210142 and rs210134, respectively (Table I). These results did not change after adjusting for age and sex (results not shown). In a metaanalysis of our previous results (Slager et al, 2012) with the 573

Short Report Table I. Association between rs210142 and rs210134 with risk of NHL/CLL.

SNP

Position

Minor Allele

rs210142

33546837

T

rs210134

33540209

A

Number of samples genotyped

Minor allele frequency

NHL subtype (study)

Cases

Controls

Cases

Controls

OR (95% CI)

P-value

CLL (Mayo/GEC) CLL (SF) CLL (combined Mayo/GEC/SF) FL (Mayo) DLBCL (Mayo) MZL (Mayo) MCL (Mayo) TCL (Mayo) CLL (SF) CLL (Duke/Utah) CLL (combined SF/Duke/Utah)

524 116 640 585 583 229 135 156 116 543 659

1253 494 1747 1522 1522 1522 1522 1522 250 389 639

0
23 0
28 0
24 0
28 0
27 0
29 0
24 0
25 0
29 0
26 0
27

0
28 0
27 0
28 0
28 0
28 0
28 0
28 0
28 0
32 0
32 0
32

0
78 1
02 0
82 1
00 0
94 1
01 0
82 0
84 0
88 0
76 0
78

0
003 0
917 0
007 0
997 0
380 0
931 0
167 0
191 0
488 0
011 0
005

(0
66, (0
72, (0
71, (0
86, (0
81, (0
82, (0
61, (0
64, (0
62, (0
61, (0
65,

0
92) 1
43) 0
95) 1
16) 1
09) 1
25) 1
09) 1
09) 1
26) 0
94) 0
93)

OR, Odds Ratio; OR is ordinal OR. CI, Confidence Interval; SNP, single nucleotide polymorphism; NHL, non-Hodgkin lymphoma; FL, follicular lymphoma; DLBCL, diffuse large B-cell lymphoma; MZL, marginal zone lymphoma; MCL, mantle cell lymphoma; TCL, T-cell lymphoma; CLL, chronic lymphocytic leukaemia; GEC, Genetic Epidemiology of CLL Consortium; SF, San Francisco.

new data reported herein, there was continued strong support for an association between these two SNPs and CLL risk (Fig. 1). We next assessed the association of rs210142 across other NHL subtypes (Table I). Clearly, no association was observed for FL, DLBCL, and MZL (P > 0
05); although not statistically significant given the limited sample sizes and statistical power for these two subtypes, MCL and TCL showed similar effect sizes to that of CLL. MCL and CLL lymphomas have very similar immunophenotype patterns, but MCL overexpresses cyclin D1 due to an invariant t(11:14) chromosomal translocation. In contrast, TCL and CLL are most likely from a different cell-of-origin, and this suggestive association of BAK1 variants with TCL is of potential interest. Using the 1000 Genomes CEU data, we computed the LD among all available variants at the 6p21.31 locus. Two variants (rs511515, rs210143) were in high LD (r2 > 0
85) with the two GWAS SNPs. rs511515 is located within the 3′ UTR of BAK1, and rs210143 is in intron 1 and is 107 bp from rs210142. Sequencing the exons of BAK1 in 67 CLL cases and 39 controls identified four variants in our CLL cases, including one of the two correlated SNPs (rs511515), rs561276, and two novel rare variants. The two novel variants were not seen in our controls, and had an allele frequency of 1% and 3% in our cases. The allele frequencies of rs511515 and rs561276 did not statistically differ between the 67 cases and 39 controls (P = 0
56 and P = 0
26, respectively). Neither of these novel variants nor the two exonic SNPs changed the amino acid of the protein based on the scale-invariant feature transform (SIFT) algorithm (Kumar et al, 2009). Interestingly, variants in or near the BAK1 gene have been shown to be associated with platelet counts (Soranzo et al, 574

2009; Lo et al, 2011) with decreasing platelet counts as the number of major alleles increases. CLL staging at diagnosis includes determining the absolute levels of the platelet count. Thus, a CLL patient presenting with thrombocytopenia (i.e.,