Genetic Variants in Interleukin-2 and Risk of Lymphoma among Children in Korea

To estimate the genetic susceptibility for childhood lymphoma, we conducted an association study for 23 cases and 148 controls. Total 1536 tag single nucleotide polymorphisms (SNPs) were selected in 138 candidate gene regions related to immune responses, apoptosis, the cell cycle, and DNA repair. Twelve SNPs were significantly associated with the risk of lymphoma (Ptrend<0.05) in six genes (IL1RN, IL2, IL12RB1, JAK3, TNFRSF13B, and XRCC3). The most significant association was seen for IL2 variant rs2069762 (ORTG+GG vs. TT=3.43 (1.29-9.11), Ptrend=0.002, minP=0.005). These findings suggest that common genetic variants in IL2 might play a role in the pathogenesis of childhood lymphoma.


Introduction
Lymphoma is the third most common cancer of children around the world (Howlader et al., 2011).Although it is known that immune system disorders are main causes of lymphoma, little is known regarding genetic factors involved not only in the immune system but in others like apoptosis, cell cycle, DNA repair (Seidemann et al., 2005).
Although some studies have reported SNPs associated with lymphoma, there are a few studies about immune response, apoptosis, cell cycle and DNA repair genes aimed among children.As gene variations of immune system, TNF and lymphotoxin-a (LT-a) polymorphisms were potential prognostic factors in childhood Non-Hodgkin lymphoma (NHL) (Seidemann et al., 2005)., a DNA repair related gene, variants were associated with decreased Burkitt lymphoma and B-cell lymphoma in children (Baris et al., 2009).
We hypothesized that gene variations of immune response and DNA repair gene may affect childhood lymphoma risk.In order to estimate the genetic susceptibility for childhood lymphoma, we conducted an association study using 1536 SNPs selected in 138 candidate gene regions related to immune response, apoptosis, cell cycle, and DNA repair among Korean children.

Subjects
Forty cases and 254 controls were recruited in three hospitals (Seoul National University Hospital, Samsung, Samsung Medical Center, ASAN Medical Center, Boramae Hospital) in Seoul, Korea independently between May 2003 and August 2006.Cases were lymphoma patients aged under 23 and controls were non-cancer patients aged under 23 who recruited from the departments of pediatrics, pediatric surgery, pediatric orthopedic surgery, pediatric urology, and pediatric otolaryngology without a medical history of cancer at the same hospitals.From all cases and controls, informed consent was obtained and questionnaires administered by trained interviewers and peripheral blood samples were collected.The patients comprised both Hodgkin lymphoma (n=4) and non-Hodgkin lymphoma (n=36).

SNP selection and genotyping
Tag SNPs were selected from common SNPs in candidate gene regions by Cancer Genome Anatomy Project (CGAP) and SNP500 database using the Tagzilla algorithm (Packer et al. 2006)http://taqzilla.nci.nih.gov/.In this process, the chosen tag SNPs were based on minor allele frequency >5% and threshold of r 2 >0.8.Analyzed gene regions are comprised of immune response, apoptosis, cell-cycle, DNA repair, and the other genes, and the numbers of tag SNPs are 932, 304, 295, and 5 respectively among total 1536 tag SNPs in each gene region (supplementary table 1).More than half of SNPs (55%) were located in intron, 22% in promoter (flanking region, UTR), 15% in 3' UTR, and 9% in exon among 1536 tag SNPs.
The DNA was extracted from the peripheral blood of cases and controls using the Gentra Puregene Blood Kit (Gentra, USA) and it was quantified using PicoGreen.For genotyping, the GoldenGate™ oligonucleotide pool assay (OPA) was performed among Korean children (Illumina®, San Diego, CA).

Statistical analysis
A Pearson χ 2 test and Fisher's exact test were used to test whether the genotype conformed to the HWE (P<0.001).Age adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by unconditional logistic regression model.The sex, age at diagnosis, birth weight, breast feeding, education levels for parents, and family history of cancer were assessed as potential confounding factors of lymphoma risk.Numerical variables like age at diagnosis (<5, 5-9, 10-14, and >14) and birth weight (<3.25 and ≥3.25) were categorized.To calculate the P trend of each SNP, the genotypes was regarded as a continuous variable after coding as 0, 1, and 2. We used the minP test to decrease type I error rate (Chen et al. 2006).The minP test was used to calculate the proportion of p-values that were smaller than the lowest P trend within each gene region with resampling based on 10,000 times permutations of case-control status.The homozygote of the most common allele was set as the reference group of each genotype in dominant and co-dominant models.
All statistical analyses were conducted using the SAS® software version 9.2 (Cary, North Carolina).

Results
The patients consist of 19 male (82.6%) and 4 female (17.4%), and the controls consist of 99 male (66.9%) and 49 female (33.1%).Only age at diagnosis was statistically significant between cases and controls.The mean age of controls was younger than that of cases (P<0.01).Cases and controls were similar in sex, birth weight, breast feeding, education levels for parents, and family history of cancer (P>0.05).

Discussion
In our association study, there were 12 SNPs in immune response genes (IL1RN, IL2, IL12RB1, JAK3, and TNFRSF13B) and DNA repair gene (XRCC3) to contribute to childhood lymphoma risk.The most significant association with childhood lymphoma was seen in IL2.
As a growth factor for T cells and natural killer (NK) cells, IL2 activated lymphoma cells and IL2 levels affected the Hodgkin lymphoma risk (Skinnider et al. 2002;Cozen et al., 2008;Chan et al., 2010).Although IL2 rs2069762 in our study was not associated with increased risk of NHL in a pooled analysis, the rs2069762 was associated with follicular lymphoma (FL) survival in adult (Rothman et al., 2006;Cerhan et al., 2007).Up to date, most studies about lymphoma and genetic factors were focused on adults.Still there were only a few studies about genetic factors and risk of lymphoma, further study could be needed to understand immunological etiology of lymphoma.
One of the limitations of this study is a small sample size.In order to decrease the possibility of false-positive results, we conducted 10,000 permutation test with lowest P trend at the gene level.We could not perform subtype analysis according to lymphoma subtypes because of the small sample sizes.Although only a proportion of the collected samples were able to be analyzed at the quality control step, there were no significant differences in selected characteristics between subjects included and excluded in the analysis Our study has a comprehensive candidate gene approach related to immune response, apoptosis, cell cycle, DNA repair with 1536 SNPs in 138 genes for childhood lymphoma.In conclusion, our findings suggest that IL2 genetic variation might play a role in the pathogenesis of childhood lymphoma.However, additional larger studies are needed to replicate these findings and understand the mechanisms of these genes in childhood lymphoma patients.
Asian Pacific Journal of Cancer Prevention, Vol 13, 2012 623 had