¹Gaubius Laboratory, TNO Prevention and Health, Leiden, The Netherlands

²Gerontology and Geriatrics Section, Department of General Internal Medicine, Leiden University Medical Centre, Leiden, The Netherlands

³Clinical Epidemiology, Leiden University Medical Centre, Leiden, The Netherlands

Correspondence to: B Heijmans, PhD, Molecular Epidemiology, Leiden University Medical Centre, PO Box 9503, 2300 RA Leiden, The Netherlands. E-mail: b.t.heijmans@lumc.nl

^aCurrent affiliation BTH and PES: Molecular Epidemiology Section, Dept. Medical Statistics, Leiden University Medical Centre, Leiden, The Netherlands

This study was supported by grant no. 94.047 from The Netherlands Heart Foundation.

Ample evidence supports a role for tumour necrosis factor (TNF) in the development of type 2 diabetes and cardiovascular disease. TNF expression was found to be influenced by a -308G/A polymorphism in the promoter of the gene encoding TNF (TNF). We investigated the contribution of this polymorphism to diabetes and cardiovascular mortality in a population-based cohort of 664 subjects aged 85 years and over (Leiden 85-plus Study). The -308G/A TNF promoter polymorphism was associated with the prevalence of diabetes in old age (P = 0.006). The risk of diabetes among subjects homozygous for the A-allele was estimated to be 4.6-fold (95% CI, 1.6-13.3) higher than among subjects homozygous for the common G-allele. The promoter polymorphism did not, however, predict mortality from all causes, cardiovascular diseases, cancer or infectious diseases during a 10-year follow-up period. In addition to the promoter polymorphism, TNFa and TNFc microsatellite genotypes were determined but these polymorphisms were not associated with morbidity or mortality. In conclusion, the -308G/A polymorphism in the TNF promoter is strongly associated with the risk of diabetes but not cardiovascular mortality in old age.

Genes and Immunity (2002) 3, 225-228. doi:10.1038/sj.gene.6363859

tumour necrosis factor ; polymorphism; linkage disequilibrium; diabetes mellitus; cardiovascular disease; mortality

Introduction

Ample evidence supports a role of tumour necrosis factor (TNF) in the development of cardiovascular disease. TNF is expressed in atherosclerotic plaques but not in healthy vessels.¹ In atherosclerotic plaques, TNF may contribute to foam cell formation, to T-lymphocyte activation and to the expression of matrix metalloproteinases that may destabilise the plaque by degrading the extra-cellular matrix.² Detailed studies also implicated TNF in the aetiology of insulin resistance, a key feature of type 2 diabetes and a major risk factor for cardiovascular disease in the elderly. TNF mRNA expression is increased in adipose tissue of severely obese and insulin resistant fa/fa rats,³ while deficiency in the gene encoding TNF (TNF) results in an increased peripheral insulin sensitivity in obese mice.⁴ In humans, there is a strong positive association between levels of TNF mRNA in adipose tissue and the extent of hyperinsulinaemia,⁵ and TNF plasma levels are increased in patients with type 2 diabetes.⁶ The molecular mechanism underlying these correlations is that TNF inhibits the insulin induced tyrosine kinase activity of the insulin receptor.^4,7

The A allele of a common -308G/A polymorphism in the promoter region of the TNF gene is associated with higher reporter gene activity^8,9 and TNF production in whole blood cell cultures.¹⁰ In addition, several short tandem repeat polymorphisms have been identified at the TNF locus, of which TNFa and TNFc were suggested to be associated with differences in TNF secretion by human monocytes.¹¹ We have assessed the contribution these polymorphisms to diabetes and all-cause and cause-specific mortality in a population-based cohort of 664 subjects aged 85 years and over.

Results

The genotype distribution of TNF -308G/A polymorphism was 65.4% (G/G), 32.2% (G/A) and 2.4% (A/A) in the cohort of 664 subjects aged 85 years and over and was in the Hardy-Weinberg equilibrium. The polymorphism was significantly associated with the risk of diabetes (P = 0.006; Table 1). Adjusted for age and gender, the risk of diabetes associated with the G/A and the A/A genotypes were estimated at 0.9-fold (95% confidence interval (CI), 0.5-1.5) and 4.6-fold (95% CI, 1.6-13.3) increased, respectively. The risk estimate for the A/A genotype was 6.5 (95% CI, 1.3-33.1) among men and 3.0 (95% CI, 0.8-12.7) among women. There was no indication for heterogeneity of the association among subjects born in Leiden (55%) and elsewhere in The Netherlands (test for heterogeneity: P = 0.51 and P = 0.31 for the G/A and A/A genotypes, respectively).

The cohort was followed for mortality over a 10-year period. TNF -308G/A genotypes were not associated with all-cause mortality (89%), cardiovascular mortality (37%), or death from cancer (15%) and infectious diseases (9%) (Table 2). Similar risk estimates were obtained when men and women were analysed separately.

In addition to the -308G/A polymorphism, the TNFa (14 alleles) and TNFc (2 alleles) short tandem repeat polymorphisms were measured. The three polymorphisms were in linkage disequilibrium (all pairwise linkage disequilibria P < 0.00001). However, the TNFa and TNFc polymorphisms were not associated with diabetes (P = 0.53 and P = 0.54, respectively) nor were they associated with mortality from any cause or from a specific cause (data not shown). The absence of an association between the short tandem repeats and diabetes despite their strong linkage disequilibrium with the -308G/A polymorphism is explained by the fact that the -308A allele is distributed over several haplotypes (Table 3). For example, all the -308A alleles occur in combination with a TNFc1 allele, but about 70% of the TNFc1 alleles do not occur in combination with this allele; about 65% of the -308A alleles occur in combination with a TNFa2 allele, but about 56% of the TNFa2 alleles do not occur in combination with this allele.

The TNF gene is located in the HLA region, which is characterised by strong linkage disequilibrium. The association with diabetes found here might, therefore, have been the result of linkage disequilibrium between the -308G/A polymorphism and variation elsewhere in the HLA region influencing diabetes risk. The occurrence of diabetes was, however, independent of HLA-DR3, DR4 and B8 phenotypes (P = 0.71, P = 0.69 and P = 0.80, respectively), indicating that this was unlikely to be the case.

Discussion

The -308G/A polymorphism in the promoter of the gene encoding TNF strongly contributed to the risk of diabetes in a population-based cohort of elderly subjects aged 85 years and over. Homozygosity for A-allele conferred a more than four-fold increased risk of diabetes. This is in agreement with the extensive evidence for a direct role of TNF in the aetiology of insulin resistance and type 2 diabetes.^4,7

The TNF gene is located in the HLA region that is characterised by especially strong linkage disequilibrium. Although this makes it difficult if not impossible to definitely prove that the -308G/A polymorphism itself is the functional variation underlying the association with diabetes, it is an attractive candidate. First, the A-allele was associated with higher gene expression levels in several studies.^8,9,10 The absence of this effect in another study¹² may be related to the use of different reporter gene constructs, stimuli, cells or cell culture conditions. Furthermore, the association with diabetes in our study was independent of other genetic variation at the TNF locus as measured by two short tandem repeat polymorphisms and the -308A allele does not occur on the same haplotype as rare alleles of other promoter polymorphisms (-238G/A, -851C/T and -857C/A) in Caucasians.¹³ Finally, in view of the suggested shared genetic susceptibility of type 1 and type 2 diabetes^14,15 it is notable that linkage disequilibrium of the promoter polymorphism and HLA-DR3, DR4 and B8 was excluded as the underlying cause of the association.

It is interesting to note that the TNFa (14 alleles) and the TNFc (2 alleles) short tandem repeats were not associated with the risk of diabetes despite their strong linkage disequilibrium with the -308G/A promoter polymorphism. This illustrates that the complexity of haplotype structures may severely hamper the usefulness of linkage disequilibrium mapping as a tool in unravelling the genetic component of complex diseases.

The -308G/A polymorphism was not associated with type 2 diabetes in previous studies among patients with a mean age of 57 years¹⁶ and 39 years.¹⁷ Apart from differences in study design and environmental and genetic background of the subjects studied, this may suggest that other genetic factors contribute to the disease at younger ages. Possibly, the adverse effects of mild alterations in TNF expression become apparent only in old age. Age differences have also been suggested to underlie the inconsistent results from studies assessing the association of the promoter polymorphism and insulin sensitivity.^18,19 More extensive studies are warranted to more precisely characterise the role of the TNF promoter polymorphism in type 2 diabetes.

The -308G/A polymorphism in the promoter of the TNF gene was not related to all-cause and cardiovascular mortality during a 10-year follow-up period. In previous studies, the polymorphism was not associated with the risk of myocardial infarction¹³ or coronary artery disease.^20,21 This may reflect that the aetiology of cardiovascular diseases is much more heterogeneous than type 2 diabetes.

In conclusion, our study indicates that the TNF -308G/A polymorphism may be a potent risk factor for diabetes in old age. We did not find evidence that the polymorphism contributes to the risk of cardiovascular mortality.

Methods

Subjects

The Leiden 85-plus Study is a population-based study in which all inhabitants of Leiden, The Netherlands, aged 85 years and over were invited to take part.²² Out of a total of 1258 eligible subjects, 221 died before enrolment. Of the 1037 remaining subjects, 977 (94%) participated and were medically interviewed at home. Diabetes was diagnosed on the basis of a history, use of anti-diabetic medication and/or a glucose level over 11.0 mmol/l in a non-fasting blood sample. After the exclusion of subjects with a non-Dutch (n = 29) or unknown (n = 69) place of birth, sufficient cell material was available from 666 (188 men/478 women) subjects for the present genetic study. The study was approved by the Medical Ethics Committee of the Leiden University and informed consent was obtained from all participants.

Genotyping and HLA typing

The TNF -308G/A genotypes were determined by PCR-amplification followed by NcoI digestion.²³ Digestion products were separated on 7.5% polyacrylamide MADGE-gels (microtitre array diagonal gel electrophoresis).²⁴ Genomic regions containing the TNFa (located in the promoter of the gene LTA that encodes TNF and flanks the TNF gene) and TNFc (located in intron 1 of LTA) short tandem repeats were amplified in a multiplex PCR¹¹ and alleles were separated with an ALF-express automated sequencer (Amersham Pharmacia Biotech, Picataway, NJ, USA). Because of technical reasons two, four, and five subjects could not be typed for the -308G/A, TNFa, and TNFc polymorphism, respectively. All genotypes were independently assessed by two observers. As a standard laboratory procedure a randomly chosen 10% of the samples was reamplified.

Typing for HLA-DR antigens was performed with a two colour fluorescence test using a set of highly selected alloantisera to class II antigens.

Prospective study

All participants in the Leiden 85-plus Study were followed up for mortality until 1 October, 1996. Among the 666 subjects of the cohort studied, two were lost to follow-up. Primary causes of death were obtained from the Dutch Central Bureau of Statistics and categorised for cardiovascular disease (ICD-9 codes²⁵ 390-459), ischaemic heart disease (410-414), cerebrovascular disease (430-438) and cancer (140-239). Death from infection was coded as previously described.²²

Statistical analyses

Distributions of genotypes, alleles and HLA phenotypes were compared by the ²-test. Age and gender adjusted risks for diabetes and 95% CIs were estimated by odds ratios calculated using logistic regression analysis. Heterogeneity of associations was tested using Mantel-Haenszel's test for stratified analyses. Pairwise linkage disequilibria between the three TNF polymorphisms and maximum likelihood haplotype frequencies were estimated using Arlequin software version 2.000.²⁶

In the prospective study, survival times for subjects were computed from the date of the home visit to the date of one of the following events: death from a specific cause, death from any cause, or 1 October 1996. Age and gender adjusted mortality risks and 95% CIs were estimated with Cox proportional hazards models. Causes of death were assumed to be independent. P-values of less than 0.05 were considered to indicate statistical significance and all P-values were based on two-sided tests.

Acknowledgements

We wish to thank the Central Bureau of Statistics for generously making available the mortality statistics and database linking.

1 Barath P, Fishbein MC, Cao J, Berenson J, Helfant RH, Forrester JS. Detection and localization of tumor necrosis factor in human atheroma. Am J Cardiol 1990; 65: 297-302. MEDLINE

2 Lee RT, Libby P. The unstable atheroma. Arterioscler Thromb Vasc Biol 1997; 17: 1859-1867. MEDLINE

3 Hotamisligil GS, Shargill NS, Spiegelman BM. Adipose expression of tumor necrosis factor-: direct role in obesity-linked insulin resistance. Science 1993; 259: 87-91. MEDLINE

4 Uysal KT, Wiesbrock SM, Marino MW, Hotamisligil GS. Protection from obesity-induced induced insulin resistance in mice lacking TNF function. Nature 1997; 389: 610-614. Article MEDLINE

5 Hotamisligil GS, Arner P, Caro JF, Atkinson RL, Spiegelman BM. Increased adipose tissue expression of tumor necrosis factor- in human obesity and insulin resistance. J Clin Invest 1995; 95: 2409-2415. MEDLINE

6 Nilsson J, Jovinge S, Niemann A, Reneland R, Lithell H. Relation between plasma tumor necrosis factor- and insulin sensitivity in elderly men with non-insulin-dependent diabetes mellitus. Arterioscler Thromb Vasc Biol 1998; 18: 1199-1202. MEDLINE

7 Hotamisligil GS, Murray DL, Choy LN, Spiegelman BM. Tumor necrosis factor inhibits signaling from the insulin receptor. Proc Natl Acad Sci USA 1994; 91: 4854-4858. MEDLINE

8 Wilson AG, Symons JA, McDowell TL, McDevitt HO, Duff GW. Effects of a polymorphism in the human tumor necrosis factor promoter on transcriptional activation. Proc Natl Acad Sci USA 1997; 94: 3195-3199. Article MEDLINE

9 Kroeger KM, Carville KS, Abraham LJ. The -308 tumor necrosis factor- promoter polymorphism effects transcription. Mol Immunol 1997; 34: 391-399. MEDLINE

10 Louis E, Franchimont D, Piron A et al. Tumour necrosis factor (TNF) gene polymorphism influences TNF- production in lipopolysaccharide (LPS)-stimulated whole blood cell culture in healthy humans. Clin Exp Immunol 1998; 113: 401-406. MEDLINE

11 Pociot F, Briant L, Jongeneel CV et al. Association of tumor necrosis factor (TNF) and class II major histocompatibility complex alleles with the secretion of TNF- and TNF- by human mononuclear cells: a possible link to insulin-dependent diabetes mellitus. Eur J Immunol 1993; 23: 224-231. MEDLINE

12 Brinkman BM, Zuijdeest D, Kaijzel EL, Breedveld FC, Verweij CL. Relevance of the tumor necrosis factor (TNF ) -308 promoter polymorphism in TNF gene regulation. J Inflamm 1995; 46: 32-41. MEDLINE

13 Herrmann SM, Ricard S, Nicaud V et al. Polymorphisms of the tumour necrosis factor- gene, coronary heart disease and obesity. Eur J Clin Invest 1998; 28: 59-66. MEDLINE

14 Rich SS, Panter SS, Goetz FC, Hedlund B, Barbosa J. Shared genetic susceptibility of type 1 (insulin-dependent) and type 2 (non-insulin-dependent) diabetes mellitus: contributions of HLA and haptoglobin. Diabetologia 1991; 34: 350-355. MEDLINE

15 Tuomilehto-Wolf E, Tuomilehto J, Hitman GA et al. Genetic susceptibility to non-insulin dependent diabetes mellitus and glucose intolerance are located in HLA region. BMJ 1993; 307: 155-159. MEDLINE

16 Hamann A, Mantzoros C, Vidal-Puig A, Flier JS. Genetic variability in the TNF- promoter is not associated with type II diabetes mellitus (NIDDM). Biochem Biophys Res Commun 1995; 211: 833-839. Article MEDLINE

17 Altshuler D, Hirschhorn JN, Klannemark M et al. The common PPAR Pro12A1a polymorphism is associated with decreased risk of type 2 diabetes. Nat Genet 2000; 26: 76-80. Article MEDLINE

18 Fernández-Real JM, Gutierrez C, Ricart W et al. The TNF- gene Nco I polymorphism influences the relationship among insulin resistance, percent body fat, and increased serum leptin levels. Diabetes 1997; 46: 1468-1472. MEDLINE

19 Rasmussen SK, Urhammer SA, Jensen JN, Hansen T, Borch-Johnsen K, Pedersen O. The -238 and -308 GA polymorphisms of the tumor necrosis factor gene promoter are not associated with features of the insulin resistance syndrome or altered birth weight in Danish Caucasians. J Clin Endocrinol Metab 2000; 85: 1731-1734. MEDLINE

20 Francis SE, Camp NJ, Dewberry RM et al. Interleukin-1 Receptor Antagonist Gene Polymorphism and Coronary Artery Disease. Circulation 1999; 99: 861-866. MEDLINE

21 Wang XL, Oosterhof J. Tumour necrosis factor G-308A polymorphism and risk for coronary artery disease. Clin Sci (Colch) 2000; 98: 435-437. MEDLINE

22 Weverling-Rijnsburger AW, Blauw GJ, Lagaay AM, Knook DL, Meinders AE, Westendorp RGJ. Total cholesterol and risk of mortality in the oldest old. Lancet 1997; 350: 1119-1123. MEDLINE

23 Wilson AG, de Vries N, Pociot F, di Giovine FS, van der Putte LB, Duff GW. An allelic polymorphism within the human tumor necrosis factor promoter region is strongly associated with HLA A1, B8, and DR3 alleles. J Exp Med 1993; 177: 557-560. MEDLINE

24 O'Dell SD, Humphries SE, Day IN. Rapid methods for population-scale analysis for gene polymorphisms: the ACE gene as an example. Br Heart J 1995; 73: 368-371. MEDLINE

25 World Health Organization. International Classification of Diseases: Manual of the International Statistical Classification ofDiseases, Injuries, and Causes of Death World Health Organization: Geneva, 1977.

26 Schneider S, Kueffer JM, Roessli D, Excoffier L. Arlequin ver. 1.1: A Software for Population Genetic Data Analysis Genetics and Biometry Laboratory, University of Geneva: Geneva, 1997.

Table 1 TNF -308G/A genotype distributions in subjects aged 85 years and over with and without diabetes

Table 2 Ten-year all-cause and cause-specific mortality risks according to TNF -308G/A genotype in subjects aged 85 years and over

Table 3 Frequent TNF haplotypes in subjects aged 85 years and over