Polymorphisms in the ANKS1B gene are associated with cancer, obesity and type 2 diabetes

Abstract Obesity and type 2 diabetes (T2D) are comorbidities with cancer which may be partially due to shared genetic variants. Genetic variants in the ankyrin repeat and sterile alpha motif domain containing (ANKS1B) gene may play a role in cancer, adiposity, body mass index (BMI), and body weight. However, few studies focused on the associations of ANKS1B with obesity and T2D. We examined genetic associations of 272 single nucleotide polymorphisms (SNPs) within the ANKS1B with the cancer (any diagnosed cancer omitting minor skin cancer), obesity and T2D using the Marshfield sample (716 individuals with cancers, 1442 individuals with obesity, and 878 individuals with T2D). The Health Aging and Body Composition (Health ABC) sample (305 obese and 1336 controls) was used for replication. Multiple logistic regression analysis was performed using the PLINK software. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. We identified 25 SNPs within the ANKS1B gene associated with cancer, 34 SNPs associated with obesity, and 12 SNPs associated with T2D (p < 0.05). The most significant SNPs associated with cancer, T2D, and obesity were rs2373013 (p = 2.21 × 10−4), rs10860548 (p = 1.92 × 10−3), and rs7139028 (p = 1.94 × 10−6), respectively. Interestingly, rs3759214 was identified for both cancer and T2D (p = 0.0161 and 0.044, respectively). Furthermore, seven SNPs were associated with both cancer and obesity (top SNP rs2372719 with p = 0.0161 and 0.0206, respectively); six SNPs were associated with both T2D and obesity (top SNP rs7139028 with p = 0.0231 and 1.94 × 10−6, respectively). In the Health ABC sample, 18 SNPs were associated with obesity, 5 of which were associated with cancer in the Marshfield sample. In addition, three SNPs (rs616804, rs7295102, and rs201421) were associated with obesity in meta-analysis using both samples. These findings provide evidence of common genetic variants in the ANKS1B gene influencing the risk of cancer, obesity, and T2D and will serve as a resource for replication in other populations.


Introduction
Obesity and type 2 diabetes (T2D) are associated with increased risks of developing cancer, and are both considerable public health burdens in the United States (US). Based on the 2009-2010 National Health and Nutrition Examination Survey (NHANES) data, the obesity prevalence in the US adults was 35.7% [1], with global projections of more than 1.12 billion obese individuals by 2030 [2]. The overall prevalence of T2D in the US was 7.7% based on the 2003-2004 NHANES [3] and 8.3% in 2011 (about 25.8 million) [4,5]. Worldwide, the incidence of T2D is rising rapidly. An estimated 284.6 million of people had diabetes in 2010 worldwide, and it is predicted that there will be 438.4 million in 2025, with about 90-95% being T2D [6]. In the US, it has been reported that more than 1.6 million new cancer cases were expected to be diagnosed in 2012 (Cancer Facts & Figures 2012, American Cancer Society). Based on the 2012 National Health Interview Survey (NHIS) data, the overall prevalence of cancer in the US is 8.6% (7.6% for males and 9.4% for females) [7]. The World Cancer Report 2014 reported that in 2012, the global incidence of cancer rose to an estimated 14 million new cases and the figure is expected to rise to an annual 19.3 million by 2025 [8].
It has been reported that obesity is associated with increasing rates of metabolic syndrome and T2D [9,10]; while overweight and obesity are associated with risk for some of the most common cancers [10]. Furthermore, meta-analyses of epidemiological data have linked diabetes to different types of cancers, along with an increased risk of cancer mortality, while obesity is associated with an increased risk of cancer and diabetes [11]. Observational studies have also demonstrated that obesity and diabetes are positively associated with an increased incidence of a number of cancers and cancer-related mortality [9,12,13]. However, the genetic mechanisms are still not clear.
The ankyrin repeat and sterile alpha motif domain containing 1B (ANKS1B) gene (also known EB1, ANKS2, AIDA1 and cajalin-2) is located at 12q23.1 and expressed in adult testis and brain and fetal brain [14]. Expression of this gene has been shown to be elevated in patients with pre-B cell acute lymphocytic leukemia associated with t(1;19) translocation [14,15]. Through a genome-wide association study (GWAS) of four quantitative traits related to body size and adiposity (BMI, weight, waist circumference, and height) in a cohort of 1792 adult Filipino women from the Cebu Longitudinal Health and Nutrition Survey (CLHNS), single nucleotide polymorphism (SNP) rs2373011 within the ANKS1B gene was associated with BMI (p = 9.00 × 10 −6 ) and waist circumference (p = 2.46 × 10 −6 ) [16]. Another study reported that some short tandem repeats (STRs) markers and SNPs and haplotypes of the EB1 gene were associated with genetic difference between diabetic patients with and without nephropathy [17]. The effect of ANKS1B gene on body weight was also found in animal model [18]. Using a case-control design, three SNPs (rs1549102, rs11110099, and rs10745877) within ANKS1B were associated with lung cancer [19]. A recent study provided the first evidence that ANKS1B expression was associated with smoking-related clear cell renal cell carcinoma (ccRCC) development [20].
However, no study has focused on the associations of ANKS1B with the risk of obesity and T2D. This study explored the associations of 272 SNPs within ANKS1B gene with the risk of cancer, obesity and T2D using a Caucasian sample and confirmed the results of obesity using another sample.

The Marshfield sample
The Marshfield sample is from the publicly available data from A Genome-Wide Association Study on Cataract and HDL in the Personalized Medicine Research Project Cohort-Study Accession: phs000170.v1.p1 (dbGaP). The primary goals of this project are to develop and validate electronic phenotyping algorithms, to accurately identify cases and controls while maintaining a positive predictive value (PPV) of > 95%, and to conduct a genome-wide association study that advances the understanding of two specific yet interrelated disease states, while simultaneously engaging the community in these research efforts. The details about these subjects were described elsewhere [21][22][23]. Obesity was determined as a body mass index (BMI) ≥ 30. Genotyping data using the ILLUMINA Human660W-Quad_v1_A are available for 3564 individuals (716 individuals with cancers, 1442 individuals with obesity and 878 individuals with T2D).

The Health ABC sample
The Health Aging and Body Composition (Health ABC) Study is a whole genome association study of visceral adiposity and is a NIA-sponsored cohort study of the factors that contribute to incident disability and the decline in function of healthier older persons, with a particular emphasis on changes in body composition in old age (dbGaP Study Accession: phs000169.v1.p1). The key components of Health ABC include a baseline examination, annual follow-up clinical examinations, and phone contacts every 6 months to identify major health events and document functional status between clinic visits. The details about these subjects were described elsewhere [23][24][25][26][27]. Obesity was determined as a body mass index (BMI) ≥ 30. The HABC dataset contains 1M Illumina Human SNPs for 1641 individuals (305 obese individuals and 1,336 controls).

Statistical analysis
Within the ANKS1B gene, 272 SNPs in the Marshfield sample and 310 SNPs in the Health ABC sample were available. Logistic regression analyses of cancer, obesity and T2D, adjusted for age and sex, were performed using PLINK v1.07 [28]. The asymptotic p-values were observed while the odds ratio (OR) and standard error (SE) of OR were estimated. The additive model was applied.
Hardy-Weinberg equilibrium was tested for all the SNPs in controls. Minor allele frequency (MAF) was determined for each SNP. In addition to obtaining nominal p-values, empirical p-values were generated by 100,000 permutation tests using Max (T) permutation procedure implemented in PLINK. In this procedure, pointwise estimate of an individual SNP's significance (empirical pointwise p-values) was calculated. Results of obesity from the two samples were directly meta-analyzed by combining the separate results of two samples (OR and standard error of OR) into one meta-analysis for overall effects. The basic meta-analysis function in PLINK was applied. Fixed-effect meta-analysis p-value and fixed-effect OR were reported. The between-study heterogeneity was tested by the χ 2 -based Cochrane's Q statistic [29].

Genotype quality control and descriptive statistics
We removed 4 of 272 SNPs with Hardy-Weinberg equilibrium test p < 10 −5 in the Marshfield sample. All 310 SNPs were in Hardy-Weinberg equilibrium in the controls of the Health ABC sample (p > 0.001). Participant characteristics for the two samples are presented in Table 1. The range of age was 46-90 years for the Marshfield sample and 69-80 years for the Health ABC sample. There were more females than males in both cases and controls in the Marshfield sample, and little more males than females in both cases and controls in the Health ABC sample.

Association with cancer
Single marker analysis showed that 25 SNPs were associated with cancer (p < 0.05) in the Marshfield sample (see supplement Table S1). The top seven SNPs (p < 0.01) are shown in Table 2. SNP rs2373013 revealed the strongest association with cancer (OR = 0.73, 95%CI = 0.61-0.86, p = 2.21 × 10 −4 ), while the next best signal was rs1275649 (p = 3.65 × 10 −3 ). All the seven SNPs in the Marshfield sample had empirical pointwise p-values p < 0.05 using a permutation procedure (Table 2).

Replication study and meta-analysis
In the Health ABC sample, 18 SNPs were associated with obesity (p < 0.05) ( Table 5). Five SNPs were associated with cancer in the Marshfield sample. Three SNPs (rs616804, rs7295102, and rs201421) were associated with obesity in the meta-analysis of the Marshfield and the Health ABC samples (p = 0.00479, 0.00152 and 0.00799, respectively). for cancer in the Marshfield sample; f p-value for cancer based on logistic regression; g Odds ratio for obesity in the Health ABC sample; h p-value for obesity based on logistic regression in the Health ABC sample; i empirical p-value for obesity generated by 100,000 permutation tests using Max (T) permutation procedure implemented in PLINK; j Odds ratio for the meta-analysis of the Marshfield and the Health ABC samples; k p-value for the meta-analysis of obesity; l p-value based on the χ2-based Cochrane's Q statistic.

Discussion
In this study, we found statistically significant associations of polymorphisms within the ANKS1B gene with cancer, obesity, and T2D. The most significant SNPs in the Marshfield sample for cancer, T2D, and obesity were rs2373013, rs10860548, and rs7139028, respectively. Interestingly, rs3759214 was linked with both cancer and T2D while seven SNPs were linked with both cancer and obesity. Furthermore, six SNPs were associated with both T2D and obesity. In the Health ABC sample, five cancer-related SNPs in the Marshfield sample were replicated. Finally, three SNPs were associated with obesity in meta-analysis using both the Marshfield and the Health ABC samples.
Previous studies have reported that expression of ANKS1B gene may be elevated in patients with pre-B cell acute lymphocytic leukemia associated with t(1;19) translocation [14,15]. A recent study demonstrated that it may play a role in a smoking-related molecular alteration in cancer [20]. Lin et al. (2012) reported that three SNPs (rs1549102, rs11110099, and rs10745877) of ANKS1B were associated with lung cancer using a case-control design [19]. In the present study, we identified 25 SNPs in the Marshfield sample associated with cancer (p < 0.05). However, we did not find associations of rs1549102, rs11110099 and rs10745877 with cancer in the Marshfield sample. It may due to the fact that the outcome in Lin et al. (2012) was restricted to lung cancer, while cancer status definition in the Marshfield sample was very broad. On the other hand, our results suggested that ANKS1B might play a role in various cancers.
To our knowledge, the present study is the first attempt to evaluate the associations of ANKS1B with obesity and T2D. The SNP rs2373011 within ANKS1B gene was associated with obesity in the Health ABC sample (p = 0.0416). Similarly, the same SNP was previously reported to be linked to BMI (p = 9.00 × 10 −6 ) and waist circumference (p = 2.46 × 10 −6 ) among Filipino women [16]. Another study reported that the polymorphisms of ANKS1B gene were associated with genetic difference between diabetic patients with and without nephropathy [17]. A more recent study revealed the effect of ANKS1B gene on body weight in an animal model [18]. We have identified 34 SNPs associated with obesity and 12 SNPs associated with T2D in the Marshfield, and 15 SNPs associated with obesity in the Health ABC sample. Interestingly, there are six SNPs associated with both T2D and obesity in the Marshfield sample. Furthermore, there are seven SNPs associated with both cancer and obesity in the Marshfield sample while five SNPs are associated with cancer in the Marshfield sample and with obesity in the Health ABC sample. The present study showed that the ANKS1B gene might play a role in the cancer, obesity and T2D.
Epidemiology studies have shown that obesity is associated with increasing rates of the metabolic syndrome and T2D [9][10][11]; while overweight, obesity and T2D are associated with increased risk of developing many different types of cancers [9][10][11][12][13]30,31]. However, the biological mechanisms underlying the relationship among obesity, T2D, and cancer are not well understood. Insulin resistance is common in obese individuals and is believed to be a key factor in the pathogenesis of the metabolic syndrome. For example, insulin, insulinlike growth factor 1, and insulinlike growth factor 2 signaling through the insulin receptor and the insulinlike growth factor 1 receptor can induce tumorigenesis, accounting to some extent for the link between diabetes, obesity, and cancer [30,31]. It has been suggested that such mechanisms may include modulation of energy balance and calorie restriction, growth factors, multiple signaling pathways, and inflammatory processes. For example, the PI3K/Akt/mTOR cascade, one of the signaling pathways linking obesity and cancer, is a target of many of the obesity-associated factors and regulates cell proliferation and survival [32]. Another study proposed that obesity, particularly abdominal obesity, is associated with insulin resistance and the development of dyslipidemia, hyperglycemia, and ultimately T2D. Although many metabolic abnormalities occur with obesity and T2D, insulin resistance and hyperinsulinemia appear to be central to these conditions and may contribute to dyslipidemia and alter levels of circulating estrogens and androgens; while hyperinsulinemia and dyslipidemia may contribute to cancer development [9]. Other mechanisms have been proposed such as hyperglycemia, hyperinsulinemia, hypertriglyceridemia, inflammatory cytokines, adipocytokines, and IGF-1 [31].
Previous findings and our present results suggested that ANKS1B was associated with cancer, obesity and diabetes. The ANKS1B gene encodes a multi-domain protein that is predominantly expressed in brain and testis [14]; while ANKS1B is involved in signals that regulate normal proliferation or differentiation of cells and therefore may play a role in cancer development [14,20,33]. In addition, the ANKS1B gene may be involved in some signaling pathways and multiple protein-protein interactions [14,34]. For example, ANKS1B is one of the genes in the apoptotic-pathway and was found to be associated with lung cancer [19]. However, the direct evidence of the role of ANKS1B in both obesity and T2D remain lacking. Previous studies showed that obesity may increase risk of cancer and diabetes, potentially caused by intracellular signaling pathways that involve in many essential cellular processes such as proliferation, differentiation, and apoptosis [35][36][37][38][39][40]. Therefore, ANKS1B may be one of the mechanisms linking obesity and T2D to cancer.
There are several strengths in this study. First, our sample sizes were relatively large and the two samples were relatively ethnically homogeneous. Second, we examined 272 SNPs in the Marshfield sample for discovery and used 310 SNPs in the Health ABC sample to replicate our findings. Third, we implemented a meta-analysis to increase precision in estimating effects by combining two study samples. Fourth, we detected pleiotropic effects of ANKS1B gene on three complex diseases-cancer, obesity and T2D.
We also realized some limitations in this study. First, the definition of cancer status in the Marshfield sample was broad (including any diagnosed cancer omitting minor skin cancer). It will be interesting to investigate the association of ANKS1B with single cancer type. Second, this study focused on obesity with BMI ≥ 30. It will be more informative to compare subtypes of obesity such as Class I obesity (BMI = 30-34.9), Class II obesity (BMI = 35-39.9), and Class III obesity (BMI ≥ 40). Third, our current findings might be subject to type I error. Therefore, these findings need to be replicated in additional samples.

Conclusion
These findings provide the first evidence of genetic variants in the ANKS1B gene influencing the risk of cancer, obesity and T2D and will serve as a resource for replication in other populations. Future functional study of this gene may help to better characterize the genetic architecture of these three comorbidities.