Abstract

Introduction

Genome-wide association studies identified variations in the first intron of the FTO (fat mass and obesity associated) gene as genetic risk factors for obesity in a wide range of populations [1–4]. The FTO gene (ID 79068, OMIM acc. N. 610966) codes a protein, the exact role of which remains unknown, but the FTO gene is widely expressed in human tissues, and some experiments suggest that FTO gene products exhibit slight and different enzymatic activities, namely 2-oxoglutarate-dependent nucleic acid demethylase activity [5] and non-haeme dioxygenase activity [6]. The research on FTO has primarily concentrated on its association with obesity and body mass index (BMI) values, food intake and energy expenditure [7].

More recently, a large cohort of Danish men reported an association between the FTO single nucleotide polymorphism and all-cause mortality [8]. The effect of the FTO gene appeared similar in obese and lean individuals, its magnitude was comparable with that of smoking, and the gene was not associated with a particular cause of death. Other recent reports suggest that the FTO variants are associated with a range of conditions, such as cancer [9], diabetes [10], hypertension [11], polycystic ovaria [12], acute coronary syndrome [13], metabolic syndrome [14], Alzheimer disease [15] and reduced brain volume [16]. Taken together, these recent reports imply a pleiotropic effect of the FTO gene (i.e. that its effects are not limited to body weight or obesity).

Chronic kidney disease (CKD) has become a major public health problem due to its high prevalence, enormous cost and reduction in patients’ life expectancy and quality of life [17]. The majority of CKD, leading to end-stage renal disease (ESRD), is closely linked to diabetes and hypertension; virtually, all patients with ESRD have diabetes, hypertension or both [17, 18]. Other risk factors for ESRD include history of cardiovascular disease (CVD), which has led to a proposition that CKD primarily reflects vascular disease [19]. Besides environmental/external factors, inherited/genetic dispositions have been suggested to influence the progression of CKD and ESRD development. Previous studies detected a strong genetic component of ESRD, with heritability ranging from 0.30 to 0.44 [20]. Indeed, several genetic polymorphisms affecting CKD/ESRD development have already been detected [21], but the current knowledge about the genetic determinants of CKD/ESRD is far from complete.

Given the fact that majority of ESRD is caused by diabetes and hypertension, which in turn is related to conditions linked with the FTO gene, an association between ESRD and FTO is plausible. To expand our understanding of the morbidities associated with the FTO gene, we have investigated whether there is a relationship between CKD/ESRD and the FTO polymorphism and whether the FTO genotype was associated with the age of onset of ESRD in patients with CKD.

Materials and methods

We conducted two independent case–control studies in the Czech Republic, using two different groups of patients with CKD/ESRD and two different sets of population-based controls. The local Ethical Committees of the Institute for Clinical and Experimental Medicine in Prague, Czech National Institute of Public Health and University College London approved the protocol of these studies. All subjects included in the study were Caucasian, and all gave their informed consent to participate in the study.

Results

In both sets of healthy controls, the FTO genotype was associated with BMI in the expected direction; i.e. the carriers of the GG genotype had significantly higher BMI than carriers of GT and TT alleles [25, 26]. Among cases, data on BMI before the onset of CKD were not available. However, BMI was available for a subset of cases at the time of the inclusion into the haemodialysis/transplantation programme. Similar to the controls, the FTO G allele was positively associated with BMI both in haemodialysed and transplant patients.

The main results are shown in Table 1. There are several notable findings. Firstly, the frequencies of the FTO genotypes differ significantly between cases and controls in both studies, and in both studies, the OR of ESRD/CKD increases with the number of G alleles. In Study 1, the OR of ESRD/CKD for GG homozygotes compared with TT homozygotes was 1.56 (95% CI 1.27–1.96), and trends of increasing OR with increasing number of G alleles were highly significant (P = 0.00004). In Study 2, the results were similar, although the association was weaker than in Study 1; the OR of ESRD for GG versus TT genotype was 1.27 (95% CI 1.07–1.51), and the P-value for trends was 0.006.

Secondly, there were no significant differences in the frequencies of the FTO genotypes between the two sets of cases (P = 0.60) and between the two sets of controls (P = 0.07).

Finally, cross analyses produced results consistent with those found in the primary analysis; the ORs of ESRD for GG versus TT genotype were 1.42 (95% CI 1.16–1.72, P for trend 0.00099) for comparing cases from Study 1 with controls in Study 2 and 1.40 (95% CI 1.15–1.70, P for trend 0.00036) for comparing cases from Study 2 with controls in Study 1. Given these results, it is not surprising that there was a highly significant association between the FTO genotype and ESRD/CKD in data pooled from both studies (2196 cases and 9048 controls). The OR for GG versus TT genotype was 1.37 (1.20–1.56), and the trend for increasing OR with increasing number of G alleles was highly significant (P = 0.000011). Further adjustment for age and sex did not materially change these estimates; the ORs for GG and GT versus TT genotypes were 1.34 (1.16–1.54) and 1.13 (1.00–1.47), respectively.

In addition to the case–control comparisons, we have also examined whether the age of onset of ESRD differs by FTO genotype in a subset of 753 haemodialysed and 1095 kidney transplantation patients of whom the age of onset was available (Table 2). We found that in haemodialysed patients (cases from Study 1), the age at onset of ESRD was significantly associated with FTO genotype; the onset of the disease in carriers of the G allele was ~3 years earlier than in TT homozygotes. The association in kidney transplant patients (cases from Study 2) was similar: GG homozygotes had onset of ESRD ~2.5 years earlier than TT homozygotes. Additional adjustment for presence of diabetes and sex did not change these results significantly (not shown in table).

Discussion

It is known that both genetic and environmental factors contribute to CKD/ESRD risk and both are of significant importance [20, 21]. As in other diseases, the vast majority of CKD/ESRD risk is under polygenic control, with many variants in a number of genes contributing to the total effect, each with a small, but detectable effect, in studies with sufficient power. The genes previously studied in association with CKD/ESRD and its complications include apolipoprotein E, methylentetrahydropholate reductase, interleukins and tumour necrosis factor alpha [27]. This is the first report that examines the possible role of the FTO gene. Our results, based on two large independent sets of cases and controls selected from general central European Slavonic population, comprising in total >2000 cases and almost 10000 controls, suggest a robust association between the FTO genotype and the risk of CKD/ESRD. We also found a suggestion of an earlier onset of ESRD in carriers of the G allele.

To our knowledge, this is the first study explicitly investigating this topic, although there are some indirect indications from other studies (see below). This study has several important strengths. Firstly, the large sample size means that the study had a large enough statistical power to detect modest effects. Secondly, we analysed two different sets of cases and controls and effectively replicated the main finding in an independent investigation. This greatly reduces the possibility that the results are due to chance (false positive).

Thirdly, this study was population based. Although the cases came from medical centres, the nature of ESRD (requiring referral for haemodialysis or transplantation) means that the cases were population based rather than drawn from specific sub-populations. Controls were random population samples. While, strictly speaking, controls and cases were not recruited from identical geographical areas, the fact that the controls were not selected on the basis of their health status reduces the scope for selection bias.

There are several limitations of this study. Firstly, we did not have more details about the cases. To recruit large numbers of cases, we had to compromise on the amount of data about them e.g. details about their health behaviours, other risk factors and clinical and laboratory data. Some of this information was available for subsets of both sets of cases, but because it was not collected in a comparable fashion with controls, these data do not allow detailed statistical analyses. Moreover, even if such data were available, they would reflect the patients’ status at a very late stage of CKD; this would be insufficient to estimate the role of, for example, diabetes and hypertension, in the development of the disease.

Secondly, ~50% of the cases did not undergo renal biopsy in the phase of ESRD; therefore, exclusion of potential exogenous causes of the CKD may not be reliable. However, it has been estimated that at least 75% of CKD cases are due to diabetes and hypertension [17], while other causes, such as cancer or poisoning, are relatively uncommon (and patients suffering from cancer or poisoning were not included in our study). While we could not reliably exclude all such secondary CKD from our study, their presence would probably lead to underestimation of the association between FTO and ESRD (assuming that the secondary CKD is not associated with the FTO gene). Our estimates of the effect of the FTO genotype are therefore likely to be realistic.

Finally, the distribution of the genotypes among controls in Study 2 was not in HWE. This can be partly explained by chance and high statistical power (even modest differences between expected and observed frequencies are statistically significant with sufficiently large numbers of subjects, and the P-value in our study was of a borderline significance). The literature suggests that the most common reason of Hardy-Weinberg disequilibrium in genetic association studies is genotyping error [28]. Our quality control did not suggest any problem with genotyping, and the relationship between FTO and BMI was similar in both sets of controls. The main difference between the observed and HWE expected genotype frequencies among the controls in Study 2 was the proportion of GT heterozygotes (47.7% in observed data versus 49.1% in expected data), while the observed and expected proportions of GG and TT homozygotes were similar to the expected. Moreover, the OR of ESRD based on the expected frequencies among controls in Study 2 was 1.29, which is slightly higher than the observed OR of 1.27. These sensitivity calculations suggest that the association between FTO and ESRD was not overestimated in Study 2.

The finding that, among haemodialysis patients, the GG homozygotes had lower age of CKD/ESRD onset than the TT homozygotes is consistent with the results of our two case–control studies. This was replicated in transplant patients, although in this group the difference in age of onset was somewhat smaller. This may be due to the fact that transplant patients are generally younger and have lower rates of diabetes, CVD and other comorbidities than patients with ESRD not selected for transplantation. This was also the case in our study; transplant patients represent a healthier pre-selected group of CKD/ESRD patients, and this may explain the weaker association between the FTO polymorphism and the age of onset of ESRD.

It is difficult to assess the consistency of our findings with previous studies, because the existing literature on FTO and CKD/ESRD is sparse. We identified only two sets of reports in the literature, and they are only partly relevant. Firstly, using two Korean studies analysing the FTO variant rs8050136, 629 healthy controls from a study by Cho et al. [29] were compared with 583 kidney transplantation patients from a separate independent study [30]. In this comparison, kidney transplant patients had a frequency of the less common allele of 25%, which is slightly higher than the 22.7% among healthy non-diabetic subjects; however, the OR 1.14 is not statistically significant, mainly because of the relatively small numbers of individuals in both groups. However, these Korean data are based on a population with a different ethnic background and using different FTO gene polymorphisms with different allelic frequencies and cannot be directly compared with our study. The second study investigated the role of the FTO gene variant (rs9939609) in the development of diabetic nephropathy in diabetic patients [31]. While the FTO genotype was associated with BMI, it was not related to the risk of nephropathy. This study, however, did not focus on the relation between ESRD and the FTO gene.

While the association between the FTO genotype and ESRD in our data is statistically robust, we can only speculate about the biological basis for this link. It is known that FTO is associated with obesity and diabetes mellitus and both these conditions are risk factors of CKD/ESRD development. It is possible that the FTO effect on CKD/ESRD is at least partly mediated through these conditions. However, pathways independent from diabetes and obesity are also likely, in analogy to earlier studies that showed that the FTO genotype was related to the risk of acute coronary syndrome after controlling for BMI and diabetes [13] or with increased mortality independently of BMI [8].

The gene translation product of FTO exhibits more enzymatic activities, but the real function of FTO remains largely unknown and is a matter of speculation. As FTO was originally identified as a determinant of obesity, functional studies have focused on the possible associations of the FTO gene variants with the preference of energy-dense food, with higher fat intake or with markers of physical activity. While the association of FTO with higher energy intake has been replicated in children and adolescents, it seems unlikely that the FTO gene affects the development of CVD or ESRD through this mechanism.

Alternatively, the FTO polymorphisms may enhance the risk of ESRD through its possible effect on DNA methylation (i.e. the epigenetic status). There is strong experimental evidence that the FTO gene could be involved in demethylation of the 1-methyladenine and 3-methylcitosine [32]. In addition, animal experiments suggest that nutrition and other life style factors significantly affect the methylation (epigenetic) status of many genes, which play a significant regulatory role in wide spectrum of human diseases [33, 34] and which may include genes influencing ESRD. Since the FTO gene codes protein with slight 2-oxoglutarate-dependent nucleic acid demethylase activity [5], it could provide the link between life style and risk of chronic diseases. Finally, at least in some conditions (newborns with low weight for gestational age; other individuals were so far not studied), it was proven that FTO variants within the first intron are associated with different methylation status of the peroxisome proliferator-activated receptor gamma gene promoter [35].

Despite the uncertainty about biological mechanisms, there is strong evidence that the FTO genotype is associated with obesity and there is emerging evidence that the FTO gene is related to several other health outcomes. This well-powered population-based case–control study suggests that the FTO polymorphism is also associated with CKD. Further studies are required to replicate this association in other populations and to clarify the molecular and pathophysiological mechanisms.

Acknowledgments

References