Abstract
Background
Epidemiologic studies have shown consistent associations between obesity and increased thyroid cancer risk, but, to date, no studies have investigated the relationship between thyroid cancer risk and obesity-related single nucleotide polymorphisms (SNPs).
Methods
We evaluated 575 tag SNPs in 23 obesity-related gene regions in a case-control study of 341 incident papillary thyroid cancer (PTC) cases and 444 controls of European ancestry. Logistic regression models, adjusted for attained age, year of birth, and sex were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) with SNP genotypes, coded as 0, 1, and 2 and modeled continuously to calculate P-trends.
Results
Nine out of 10 top-ranking SNPs (Ptrend<0.01) were located in the FTO (fat mass and obesity associated) gene region, while the other was located in INSR (insulin receptor). None of the associations were significant after correcting for multiple testing.
Conclusions
Our data do not support an important role of obesity-related genetic polymorphisms in determining the risk of PTC.
Impact
Factors other than selected genetic polymorphisms may be responsible for the observed associations between obesity and increased PTC risk.
Keywords: single nucleotide polymorphisms, case-control study, obesity, body mass index, thyroid neoplasms
INTRODUCTION
Obesity has consistently been associated with increased risk of thyroid cancer in epidemiologic studies [1], but the biological mechanisms underlying this association remain poorly understood. Evaluating genetic variation in obesity-related genes may help to identify pathways involved in thyroid cancer etiology, independent of, or mediated by, body size.
We examined associations between single nucleotide polymorphisms (SNPs) in 23 obesity-related candidate genes and papillary thyroid cancer (PTC), the most common histological type of thyroid cancer. These genes were chosen because of their role in body energy homeostasis and metabolism or previous associations with obesity or type 2 diabetes [2–5].
METHODS
The study population has been previously described [6]. In brief, cases included individuals diagnosed with incident, histologically confirmed PTC during follow-up of the US Radiologic Technologists (USRT) cohort (n=202) and individuals diagnosed and treated for PTC at the University of Texas M. D. Anderson Cancer Center (UTMDACC) (n=142). In USRT, controls (n=452) were frequency matched by race, year of birth (± two years), and sex to cases. Controls from USRT were then selected to match cases from UTMDACC. Analyses were restricted to non-Hispanic whites. Three cases and eight controls were excluded due to missing height or weight. The institutional review boards approved the use of these data, and all subjects provided written informed consent.
The 23 genes chosen for this analysis (listed in Supplemental Table 1) were selected a priori. Tag SNPs (n=575) were selected from the common SNPs (minor allele frequency >5%) genotyped by the HapMap Project in the Caucasian population using TagZilla, part of the GLU software package, with a binning threshold of r2>0.8. Genotyping was performed at the NCI Core Genotyping Facility using a custom-designed iSelect Infinium assay. SNPs were excluded if they failed quality-control measures: <95% concordance, <90% completion, or had evidence of a departure from Hardy-Weinberg equilibrium in controls (P<0.00001). Allele frequencies were largely similar between USRT and UTMDACC cases; thus, these groups were combined for analyses.
Data on demographics, medical history, anthropometry, and other health-related characteristics were collected by self-administered questionnaires or telephone interview in USRT and self-administered questionnaire at time of blood collection in UTMDACC.
We computed SNP-specific P-values for trend and odds ratios (ORs) and 95% confidence intervals (CIs) for each genotype, using logistic regression models adjusted for sex, attained age, and year of birth. Separate models additionally adjusted for body mass index (BMI). We also examined 138,605 two-way SNP-SNP interactions using allelic-based gene-gene interactions in models adjusted for sex, attained age, year of birth, and BMI [7]. We combined SNP-specific P-values of trend into region-based P-values using the adaptive rank truncated method [8]. P-values <0.05 were considered statistically significant, and tests were two-sided. While tables show uncorrected P-values, we also conducted correction for multiple comparisons controlling the false discovery rate (FDR). Statistical analyses were conducted using Stata/SE version 11.0 and R software.
RESULTS
Compared to controls, PTC cases were more likely to have a family history of thyroid cancer among first-degree relatives and less likely to be current smokers (Table 1). Cases had higher BMI compared to controls.
Table 1.
Select characteristics of the cases and controls
| Cases | Controls | |||
|---|---|---|---|---|
| Women (n=272) | Men (n=69) | Women (n=415) | Men (n=29) | |
|
| ||||
| Study, n (%) | ||||
| U.S. Radiologic Technologists | 182 (67) | 18 (26) | 415 (100) | 29 (100) |
| M. D. Anderson Cancer Center | 90 (33) | 51 (74) | 0 | 0 |
| Attained age, n (%) | ||||
| 19–25 | 23 (8) | 2 (3) | 30 (7) | 0 |
| 26–35 | 58 (21) | 15 (22) | 95 (23) | 5 (17) |
| 36–45 | 89 (33) | 24 (35) | 142 (34) | 13 (45) |
| 46–55 | 65 (24) | 11 (16) | 99 (24) | 6 (21) |
| 56–65 | 29 (11) | 13 (19) | 43 (10) | 5 (17) |
| 66–79 | 8 (3) | 4 (6) | 6 (1) | 0 |
| Smoking status, n (%) | ||||
| Never | 167 (62) | 38 (55) | 218 (53) | 12 (41) |
| Former | 67 (25) | 20 (29) | 89 (22) | 10 (34) |
| Current | 37 (14) | 11 (16) | 104 (25) | 7 (24) |
| Family history of thyroid cancer, n (%) | 13 (4.8) | 4 (5.8) | 5 (1.2) | 1 (3.5) |
| Body mass index (BMI)a, median (interquartile range) | 23.9 (21.1–28.3) | 28.8 (25.6–32.6) | 23.2 (21.0–25.8) | 25.0 (23.1–27.5) |
Weight in kilograms divided by height in meters squared
Of the ten SNPs identified with the lowest SNP-level P-values (Table 2), nine were located in FTO (fat mass and obesity associated) and one was located in INSR (insulin receptor). However, none remained statistically significant after FDR correction. Although BMI was associated with increased PTC risk (per 5 kg/m2, OR=1.18, 95% CI: 1.02–1.37), additional adjustment for BMI did not appreciably change the SNP-PTC associations. We did not observe statistically-significant SNP-SNP interactions after FDR correction. Also, at gene region level none was significantly associated with PTC risk (all region-based P-values >0.2).
Table 2.
Associations between obesity-related SNPs and papillary thyroid cancer risk
| SNP | Region | Location | Variant | Cases | Controls | ORa | ORb |
|---|---|---|---|---|---|---|---|
| RS17817288 | FTO | 52365265 | GG | 60 | 114 | 1.00 (Reference) | 1.00 (Reference) |
| AG | 181 | 235 | 1.57 (1.06–2.32) | 1.60 (1.08–2.38) | |||
| AA | 99 | 95 | 2.08 (1.33–3.25) | 2.14 (1.36–3.36) | |||
| P-trend=0.001* | P-trend=0.001* | ||||||
| RS11642841 | FTO | 52402988 | CC | 156 | 154 | 1.00 (Reference) | 1.00 (Reference) |
| AC | 140 | 217 | 0.65 (0.47–0.90) | 0.65 (0.47–0.91) | |||
| AA | 45 | 73 | 0.53 (0.34–0.84) | 0.52 (0.33–0.83) | |||
| P-trend=0.002 | P-trend=0.002 | ||||||
| RS8047395 | FTO | 52356024 | AA | 71 | 120 | 1.00 (Reference) | 1.00 (Reference) |
| AG | 170 | 234 | 1.27 (0.87–1.85) | 1.28 (0.88–1.88) | |||
| GG | 100 | 86 | 1.97 (1.28–3.05) | 2.03 (1.31–3.16) | |||
| P-trend=0.002 | P-trend=0.002 | ||||||
| RS1121980 | FTO | 52366748 | GG | 136 | 137 | 1.00 (Reference) | 1.00 (Reference) |
| AG | 155 | 223 | 0.72 (0.52–1.01) | 0.70 (0.50–0.97) | |||
| AA | 49 | 82 | 0.54 (0.35–0.85) | 0.52 (0.33–0.82) | |||
| P-trend=0.005 | P-trend=0.003 | ||||||
| RS8050136 | FTO | 52373776 | CC | 144 | 150 | 1.00 (Reference) | 1.00 (Reference) |
| AC | 156 | 222 | 0.77 (0.55–1.07) | 0.75 (0.54–1.04) | |||
| AA | 41 | 72 | 0.51 (0.32–0.81) | 0.49 (0.30–0.79) | |||
| P-trend=0.005 | P-trend=0.003 | ||||||
| RS9939609 | FTO | 52378028 | TT | 144 | 151 | 1.00 (Reference) | 1.00 (Reference) |
| AT | 155 | 220 | 0.77 (0.56–1.07) | 0.75 (0.54–1.05) | |||
| AA | 41 | 72 | 0.51 (0.32–0.82) | 0.49 (0.30–0.80) | |||
| P-trend=0.005 | P-trend=0.003 | ||||||
| RS1477196 | FTO | 52365759 | GG | 115 | 182 | 1.00 (Reference) | 1.00 (Reference) |
| AG | 164 | 203 | 1.33 (0.95–1.86) | 1.33 (0.95–1.86) | |||
| AA | 60 | 54 | 1.98 (1.24–3.13) | 2.00 (1.26–3.19) | |||
| P-trend=0.003 | P-trend=0.003 | ||||||
| RS7202116 | FTO | 52379116 | AA | 144 | 151 | 1.00 (Reference) | 1.00 (Reference) |
| AG | 156 | 220 | 0.78 (0.56–1.08) | 0.76 (0.55–1.06) | |||
| GG | 41 | 72 | 0.51 (0.32–0.82) | 0.49 (0.31–0.80) | |||
| P-trend=0.005 | P-trend=0.004 | ||||||
| RS1861867 | FTO | 52406062 | GG | 109 | 183 | 1.00 (Reference) | 1.00 (Reference) |
| AG | 163 | 187 | 1.61 (1.15–2.26) | 1.60 (1.13–2.25) | |||
| AA | 68 | 74 | 1.68 (1.09–2.59) | 1.73 (1.12–2.68) | |||
| P-trend=0.006 | P-trend=0.005 | ||||||
| RS919275 | INSR | 7212441 | TT | 118 | 116 | 1.00 (Reference) | 1.00 (Reference) |
| CT | 165 | 242 | 0.68 (0.48–0.96) | 0.69 (0.49–0.97) | |||
| CC | 53 | 82 | 0.54 (0.34–0.86) | 0.57 (0.36–0.91) | |||
| P-trend=0.006 | P-trend=0.011 |
SNP-based linear P-trends (unadjusted for multiple comparisons) based on modeling the three-level genotype (0, 1, 2) as continuous in logistic regression models
Adjusted for attained age (age at diagnosis for cases and referent age for controls; continuous), year of birth (<1940, 1940–1949, 1950+), and sex
Adjusted for attained age, year of birth, sex, and BMI (per 5 kg/m2)
DISCUSSION
In general, our results do not suggest an important role of selected obesity-related genetic variants in determining PTC risk. Certain polymorphisms in the FTO and INSR genes were weakly linked to PTC risk independent of BMI, but these associations were no longer significant after multiple comparisons correction.
Genes chosen for this analysis were a priori-selected based on their known functions or observed associations with obesity, thereby reducing the possibility that our findings were due solely to chance. Nonetheless, there may be other obesity-related genes that were not considered in our genotyping platform but may play an important role in papillary thyroid carcinogenesis. More agnostic approaches may be needed to discover important genetic risk factors for this disease. Additionally, while most individual SNPs and none of the two-way interactions were not significantly associated with PTC risk, certain combination of SNPs may have stronger effects, although larger studies are necessary to detect SNP-SNP interactions.
As the biological mechanisms underlying the observed obesity-thyroid cancer relationship remain unclear, the results of this study underscore the need to evaluate, directly, levels of various adipocytokines and other obesity-related biomarkers, as well as modifiable determinants of obesity, including over-nutrition and physical inactivity, as possible risk factors for this disease.
Supplementary Material
Acknowledgments
Financial support: This research was supported in part by the Intramural Research Program of the National Cancer Institute, National Institutes of Health. This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under Contract No. HHSN261200800001E.
This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under Contract No. HHSN261200800001E. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.
Footnotes
Conflict of interest: none
References
- 1.Renehan AG, Tyson M, Egger M, Heller RF, Zwahlen M. Body-mass index and incidence of cancer: a systematic review and meta-analysis of prospective observational studies. Lancet. 2008;371(9612):569–78. doi: 10.1016/S0140-6736(08)60269-X. [DOI] [PubMed] [Google Scholar]
- 2.Enns JE, Taylor CG, Zahradka P. Variations in adipokine genes AdipoQ, Lep, and LepR are associated with risk for obesity-related metabolic disease: the modulatory role of gene-nutrient interactions. Journal of Obesity. doi: 10.1155/2011/168659. (Epub 2011 Apr 19) [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Herbert A, Gerry NP, McQueen MB, Heid IM, Pfeufer A, Illig T, et al. A common genetic variant is associated with adult and childhood obesity. Science. 2006;312(5771):279–83. doi: 10.1126/science.1124779. [DOI] [PubMed] [Google Scholar]
- 4.Speliotes EK, Willer CJ, Berndt SI, Monda KL, Thorliefsson G, Jackson AU, et al. Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat Genet. 2010;42(11):937–48. doi: 10.1038/ng.686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Billings LK, Florez JC. The genetics of type 2 diabetes: what have we learned from GWAS? Ann N Y Acad Sci. 2010;1212:59–77. doi: 10.1111/j.1749-6632.2010.05838.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Neta G, Brenner AV, Sturgis EM, Pfeiffer RM, Hutchinson AA, Aschebrook-Kilfoy B, et al. Common genetic variants related to genomic integrity and risk of papillary thyroid cancer. Carcinogenesis. 2011;32(8):1231–7. doi: 10.1093/carcin/bgr100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Jung J, Song JJ, Kwon D. Allelic based gene-gene interactions in rheumatoid arthritis. BMC Proceedings. 2009;3 (Suppl 7):S76. doi: 10.1186/1753-6561-3-S7-S76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yu K, Li Q, Bergen AW, Pfeiffer RM, Rosenberg PS, Caporaso N, et al. Pathway analysis by adaptive combination of P-values. Genet Epidemiol. 2009;33(8):700–9. doi: 10.1002/gepi.20422. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.