Conceived and designed the experiments: JMS RPG JJS ABW. Performed the experiments: JMS. Analyzed the data: JMS DD JG JJS. Contributed reagents/materials/analysis tools: DD. Wrote the paper: JMS ABW. Did the pathology review: RDG BRB. Responsible for clinical review for all cases: JMC.
The authors have declared that no competing interests exist.
Non-Hodgkin lymphomas are a heterogeneous group of solid tumours that constitute the 5th highest cause of cancer mortality in the United States and Canada. Poor control of cell death in lymphocytes can lead to autoimmune disease or cancer, making genes involved in programmed cell death of lymphocytes logical candidate genes for lymphoma susceptibility.
We tested for genetic association with NHL and NHL subtypes, of SNPs in lymphocyte cell death genes using an established population-based study. 17 candidate genes were chosen based on biological function, with 123 SNPs tested. These included tagSNPs from HapMap and novel SNPs discovered by re-sequencing 47 cases in genes for which SNP representation was judged to be low. The main analysis, which estimated odds ratios by fitting data to an additive logistic regression model, used European ancestry samples that passed quality control measures (569 cases and 547 controls). A two-tiered approach for multiple testing correction was used: correction for number of tests within each gene by permutation-based methodology, followed by correction for the number of genes tested using the false discovery rate.
Variant rs928883, near miR-155, showed an association (OR per A-allele: 2.80 [95% CI: 1.63–4.82];
This is the first reported association between a germline polymorphism at a miRNA locus and lymphoma.
Non-Hodgkin lymphoma (NHL) is a collection of different subtypes, each with different clinical presentation, preferred treatment regimens and prognosis. These subtypes appear to be derived from progenitor cells at different stages of B- and T-cell development. For example, mantle cell lymphomas are derived from pre-germinal centre B cells while follicular lymphomas arise from germinal centre B-cells
Programmed cell death, the best characterized form of which is apoptosis, gives cells the ability to auto-destruct following specific triggers. Such triggers can be the presence or absence of a signal, depending on the cell context. Apoptosis has a central role in the development of the immune system, balancing the need for an effective immune response with the need to eliminate auto-reactive cells
The number of BCL2-homology domains (or BH domains) determines whether the proteins in this family have anti-apoptotic or pro-apoptotic roles. Members with 4 BH-domains (BCL2, MCL1, and BCL-xL) have anti-apoptotic roles. Both the BH3-only members (including BIM, PUMA/BBC3, NOXA/PMAIP1, BID, BAD, BMF, BIK/BLK/NBK, HRK/DP5), which have only a BH3 domain, and the members with three BH domains (BAX and BAK) have pro-apoptotic functions
Seventeen candidate genes were selected following literature review and were chosen based on their role in apoptosis in lymphocyte development:
The study population has been previously described
From the 1587 samples, DNA was derived from whole blood in 407 samples, lymphocytes isolated from blood in 782 samples, mouthwash in 24 samples and saliva in 48 samples. DNA extraction was done using the PureGene DNA isolation kit (Gentra Systems, Minneapolis, MN) following the manufacturer's instructions. DNA was quantified using PicoGreen (Molecular Probes, Invitrogen, Burlington, ON). Samples with limited amounts of DNA (326/1587 samples) were subjected to whole-genome amplification using the RepliG kit (QIAGEN, Mississauga, ON, Canada); these are hereafter referred to as “WGA” samples.
Forty-seven NHL case samples (
Genes were chosen for exon re-sequencing based on the depth of genetic variation already present in HapMap Phase II. For each gene, boundaries were selected by first capturing the gene with an extra 1.5 kb of sequence on either side of the gene within the browser window, and then extending the window boundaries until at least one more SNP in either direction was captured within the genomic region displayed. Genotypes were downloaded into Haploview
Uniform sequencing conditions were enabled by adding the -21M13F (
Sequencing results were exported from Mutation Surveyor and imported into Haploview
The SNPs selected for this study were part of a larger Golden Gate assay (Illumina, San Diego, CA), which also contained SNPs from candidate genes in other pathways related to other hypotheses. SNPs predicted to fail assay design (most often due to being too close together) were replaced when possible by equivalent tagSNPs. In addition to tagSNPs, 51 ancestry-informative markers (AIMs) selected from Halder
600 ng of each genomic or WGA DNA sample was arrayed into 18 96-well plates and sent to The Centre for Applied Genomics, the Hospital for Sick Children in Toronto, Canada. The resulting genotypes were assessed using Genome Studio version 2009.1 (Illumina, San Diego, CA). The first steps in genotype quality control (Q/C) were performed in Genome Studio; systems and databases in the laboratory of DD completed the final steps. Genotypes derived from WGA and genomic DNA were subjected to Q/C separately.
For all genotypes, a GenCall Score cutoff of 0.25 was used, following Illumina recommendations for GoldenGate technology. The following SNPs were excluded: SNPs with GenTrain scores <0.4; SNPs with GenTrain scores between 0.4 and 0.7 and poor clustering (unexpected number of clusters, or clusters that were not well defined); SNPs with more than 3 clusters (which potentially indicate copy number variants); mono-allelic SNPs; SNPs with any genotype discrepancies between 53 pairs of duplicate samples; SNPs with call rate <95% in any sample type category (blood, mouthwash or saliva) were excluded for all samples in that category; one X-chromosome SNP that showed heterozygous calls in confirmed male samples; and SNPs out of Hardy-Weinberg equilibrium (
Of 178 tagSNPs selected for genotyping; 28/178 failed in the design phase, leaving 150 SNPs ordered for genotyping. Twenty-seven SNPs were excluded at the genotype quality control stage (12 SNPs were rejected by the genotyping centre upon initial inspection, 2 for low GenTrain scores, 5 for being potential copy number variants, 3 for being monoallelic, 3 for having a call rate <0.95, and 2 for having 1 error between duplicate genotypes). An additional 16 SNPs failed quality control only in WGA samples (7 for low GenTrain score and 9 for call rate <0.95), and 2 SNPs failed Q/C only in mouthwash or saliva samples. This left 123 SNPs (82%), listed in
WGA samples give different genotype intensities than non-WGA samples
Samples with sex discrepancies (Qu
DNA samples from 1587 individuals were genotyped, 326/1587 being WGA samples. 58 non-WGA samples and 115 WGA samples were removed during quality control (23 non-WGA and 1 WGA sample due to gender discrepancies, and 35 non-WGA and 114 WGA samples due to call rate below 98%), leaving 1203 non-WGA samples+211 WGA samples, or 1414 samples in total.
45/51 AIMs passed quality control. Using genomic controls methodology, which has been shown to be appropriate when a modest number of candidate genes are assessed
Self reported ethnicities were confirmed using identity by state allele sharing and Multi-Dimensional Scaling (MDS) plots using HapMap data as the reference. These analyses identified two previously unidentified sets of sib-pairs, and three individuals who self-identified as Asian but clustered close to the HapMap CEU reference cluster. Sib-pair relationships were subsequently confirmed after review of study records; the youngest sibling of each pair (both were non-WGA samples) was excluded from analyses, as excess relatedness within a group of cases is not appropriate for a population-based association study. For the three subjects whose self-reported ethnicity was Asian, a review of study records indicated one subject self-reported being of Chinese ancestry, and two self-reported as being of Iranian descent. The subject of Chinese descent (non-WGA sample) was removed from the analyses due to concerns of a possible sample swap; and for the two subjects of Iranian descent the ethnicity was changed to “other”. This left 1200 non-WGA samples and 211 WGA samples, or 1411 samples in total (
Cases (%) | Controls (%) | |
|
||
Male | 416 (58%) | 360 (52%) |
Female | 301 (42%) | 334 (48%) |
|
||
20–49 | 131 (18%) | 172 (25%) |
50–59 | 173 (24%) | 153 (22%) |
60–69 | 196 (27%) | 185 (27%) |
70+ | 217 (30%) | 184 (27%) |
|
||
Caucasian | 569 (79%) | 547 (79%) |
Asian | 66 (9%) | 69 (10%) |
South Asian | 26 (4%) | 31 (4%) |
Mixed/Other | 33 (5%) | 29 (4%) |
Unknown/Refused | 23 (3%) | 18 (2%) |
|
||
|
||
DLBCL | 189 (26%) | - |
FL | 205 (29%) | - |
MZL/MALT | 78 (11%) | - |
MCL | 43 (6%) | - |
SLL/CLL | 39 (5%) | - |
LPL | 40 (6%) | - |
MISC BCL | 54 (8%) | - |
|
||
MF | 38 (5%) | - |
PTCL | 24 (3%) | - |
MISC TCL | 7 (1%) | - |
|
717 (100%) | 694 (100%) |
DLBCL = Diffuse Large B-Cell Lymphoma, FL = Follicular Lymphoma, MZ/MALT = Marginal Zone lymphoma/Mucosa-Associated Lymphoma Tissue lymphoma, MCL = Mantle Cell lymphoma, SLL = Small Lymphocytic Lymphoma, LPL = Lymphoplasmacytic Lymphoma, Misc. B-cell = Miscellaneous B-cell lymphoma, MF = Mycosis Fungoides, PTCL = Peripheral T-Cell Lymphoma, Misc. T-cell = Miscellaneous T-cell lymphoma.
After all quality control, 1116/1411 samples (569 cases and 547 controls) of European ancestry were included in statistical analysis.
Statistical analysis was performed using SVS Suite 7 (Golden Helix, Bozeman, MT). Logistic regressions under an additive model were fit for diffuse large B-cell lymphoma (DLBCL), follicular lymphoma (FL), marginal zone lymphoma (MZL), mantle cell lymphoma (MCL), all B-cell NHLs and all T-cell NHLs. In all subtype analyses, selected cases were compared to all controls. European-ancestry samples were used for analysis; other ethnicities (Asian, south-east Asian and “other”) were tested only for SNPs that showed association in European-ancestry samples. For each SNP, we calculated
To correct for multiple testing within each gene, full scan permutation carried out in SVS (10,000 permutations)
Candidate Gene | #HapMap SNPs/#tagSNPs |
# novel SNPs found by re-sequencing |
# SNPs genotyped | # SNPs analyzed |
|
1.40 | 4 | 8 | 7 |
|
|
0 | 21 | 21 |
|
1.54 | 12 | 12 | 11 |
|
|
2 | 9 | 7 |
|
|
1 | 11 | 7 |
|
1.00 | 11 | 7 | 3 |
|
1.00 | 3 | 2 | 1 |
|
|
13 | 2 | 2 |
|
|
9 | 17 | 14 |
|
1.00 | 7 | 11 | 7 |
|
1.64 | 8 | 8 | 7 |
|
1.00 | 9 | 9 | 6 |
|
|
6 | 10 | 10 |
|
|
1 | 14 | 11 |
|
|
2 | 3 | 3 |
|
|
4 | 3 | 1 |
|
|
0 | 3 | 3 |
*No SNPs were reported in HapMap.
Genes with ratio >2 are marked in bold and only had upstream regions re-sequenced.
Novel SNPs defined as not present in dbSNP build 127.
For a summary of all analysis results see
DLBCL (n = 148) | FL (n = 165) | MZL/MALT (N = 55) | MCL (n = 40) | All B-cell (n = 523) | All T-cell (n = 45) | |
|
rs4645900: 0.62 | rs4645900: 0.73 | rs704243: 1.34 | rs704243: 1.66 | rs4645900: 0.66 | rs11667229: 0.73 |
|
rs4987873: 0.48 | rs1026825: 1.35 | rs1564483: 1.85 | rs7226979: 1.37 | rs4987852: 1.44 | rs1801018: 0.57 |
|
rs1474326: 1.30 | BCL6_x10(3083)_A/T: 1.72 | rs1474326: 0.86 | rs2229362: 0.50 | rs3733017: 1.34 | rs1523474: 1.23 |
|
rs12134420: 0.91 | rs4949928: 1.25 | rs4949927: 1.24 | rs4949927: 1.63 | rs4949928: 1.17 | rs12134420: 1.19 |
|
rs3761704: 1.25 | rs10204044: 0.87 | rs17041883:0.29 |
rs17041868: 0.46 | rs10204044: 0.89 | rs17041868: 0.54 |
|
rs45474992: 0.54 | rs45474992: 0.68 | rs884171: 0.90 | rs45474992: 0.65 | rs45474992: 0.66 | rs884171: 0.73 |
|
rs10512488: 1.36 |
rs10512488: 0.93 | rs10512488: 1.38 | rs10512488: 1.52 | rs10512488: 1.13 | rs10512488: 1.16 |
|
BMI1_UPSTR(-809)_C/T: 1.55 | BMI1_UPSTR(-809)_C/T: 1.04 | BMI1_UPSTR(-809)_C/T: 1.23 | BMI1_UPSTR(-809)_C/T: 0.53 | BMI1_UPSTR(-809)_C/T: 1.27 | rs985000: 0.49 |
|
rs9658761: 1.34 | rs2147420: 0.81 | rs983751: 0.62 | rs2147420: 0.69 | rs978522: 1.19 | rs9658761: 1.61 |
|
rs6655975: 1.40 | rs6655975: 0.77 | rs35392872: 0.47 | rs35392872: 0.59 | rs35661734: 1.82 | rs35661734: 2.63 |
|
rs1695144: 0.45 | rs769412: 0.68 | rs3730536: 1.14 | rs1695147: 1.41 | rs769412: 0.85 | rs937283: 1.45 |
|
rs1041978: 1.22 | rs1041978: 1.53 | rs9957673: 0.67 | rs1041978: 1.84 | rs1041978: 1.35 | rs1041978: 1.42 |
|
rs11587785: 1.27 | rs12738115: 0.63 | rs12738115: 0.63 | rs6676805: 2.05 |
RFWD2_UPSTR(-184)_C/G: 0.70 | 11587785: 2.37 |
|
rs3804439: 0.82 | rs2362973: 1.2 | rs:17279275: 0.63 | rs:17279275: 0.61 | rs4440390: 1.16 | rs10071838: 0.63 |
|
rs9535416: 0.81 | rs9535416: 1.05 | rs9535416: 0.71 | rs2476391: 0.33 | rs9535416: 0.91 | rs2476391: 1.11 |
|
rs17642969: 0.89 | rs17642969: 0.98 | rs17642969: 0.74 | rs17642969: 0.85 | rs17642969: 1.03 | rs17642969: 0.41 |
|
rs2829803: 0.87 | rs928883: 1.35 | rs928883: 2.80 |
rs2829803: 0.78 | rs928883: 1.30 ( | rs2829803: 1.61 |
All analyses are done against 547 European-ancestry controls.
*:
:
DLBCL: Diffuse Large B-Cell Lymphoma, FL: Follicular Lymphoma, MZ/MALT: Marginal Zone lymphoma/Mucosa-Associated Lymphoma Tissue lymphoma, MCL: Mantle Cell lymphoma.
Four other results indicated an association when the permuted
miR-155 has many known targets, including p53BP1 (the product of
Association studies are often based on tagSNPs, which are usually selected from HapMap. TagSNPs derived from HapMap have been identified as non-transferable between populations in some cases
rs928883, which is associated with MZL in our study, is located 2.3 kb upstream of
The
miRNAs are ∼22 nucleotide RNAs that act as negative post-transcriptional regulators of gene expression by binding to imperfectly complementary sites on target mRNAs. 50% of miRNA genes are located in fragile sites or cancer-associated genomic regions
Although little is known about the role of miR-155 in MZL, its role has been investigated in other subtypes of NHL. High miR-155 expression is a marker of poor prognosis in HL and DLBCL
Morton
Some studies have found that further classifying DLBCLs according to cell-of-origin subtypes Germinal-Centre B-cell like (GCB) and Activated B-Cell like (ABC) can clarify associations; for example, rs80031434 (G397C) was associated only with the GCB subtype of DLBCL
In summary, we report an association for rs928883, located near miR-155, with marginal zone lymphoma. To our knowledge this is the first reported association between germline polymorphisms involving a miRNA gene and lymphoma, though the relevance of microRNAs to cancer in general is well established. We were not able to replicate some previously reported associations of SNPs in other apoptosis genes with subtypes of NHL. Replication of our finding in other studies is needed to confirm its significance.
(PDF)
(PDF)
(PDF)
(PDF)
(XLS)
(XLS)
(PDF)
The authors thank Stephen Leach for his longstanding lab assistance and Dr. Tara Paton of the Centre for Applied Genomics at The Hospital for Sick Children, Toronto, Canada for assistance with genotype data processing and quality control.