FTOrs9939609 polymorphism is associated with metabolic disturbances and response to HCV therapy in HIV/HCV-coinfected patients

The combination antiretroviral therapy (cART) has made human immunodeficiency virus (HIV) infection a chronic manageable disease in high income countries [1]. In this setting, chronic hepatitis C (CHC) has turned into an important comorbidity and a major cause of death in HIV/hepatitis C virus (HCV) coinfected patients [2],[3], because HIV infection accelerates the natural history of CHC [4]-[6]. Moreover, although the published data suggest that cART might be beneficial for HIV/HCV-coinfected patients [4], the interactions among HIV, HCV and cART are also associated with several metabolic disorders [7], including dyslipidaemia, lipodystrophy, steatosis, insulin resistance and type 2 diabetes mellitus [7],[8].

For many years, dual therapy with pegylated-interferon-alpha plus ribavirin (pegIFNα/RBV) has been the standard anti-HCV therapy for HIV/HCV-coinfected patients [9], and is still used in combination with new direct-acting antivirals, such as telaprevir or boceprevir [10]. The rate of HCV clearance after pegIFNα/RBV therapy is around 20% to 40% for patients infected with HCV genotype 1 (HCV-GT1) and HCV genotype 4 (HCV-GT4), and 50% to 60% in HCV genotype 2 (HCV-GT2) and HCV genotype 3 (HCV-GT3) patients [11],[12]. To date, the best baseline predictors for HCV therapy are HCV genotype, HCV viremia, liver fibrosis and single-nucleotide polymorphisms around the interferon lambda 3 (IFNL3) gene, also known as interleukin 28B (IL28B) [13]. However, an unexplained variability in HCV treatment outcome still remains which suggests that other host genetic factors may play an important role in pegIFNα/ribavirin therapy [14]. Thus, the identification of predictors for HCV therapy might help to ensure an adequate selection of the best candidates and to minimize any undesirable toxicity.

At present, the new direct-acting antivirals (DAAs) are generally administered in combination with pegIFNα/ribavirin, particularly in difficult-to-treat patients infected with GT1/4 [15],[16]. Moreover, the new IFNα-free regimens with DAAs in combination with or without ribavirin are being developed for difficult-to-treat patients [17]. However, the potential use of these new DAAs in HIV/HCV coinfected patients is complicated due to the choice of patients to treat, potential for drug-drug interactions and overlapping toxicities between HIV and HCV therapy [16]. In addition, the new DAAs are more expensive and there are serious restrictions on their administration, and in many regions in the world these drugs are inaccessible. In fact, treatment with pegIFNα/ribavirin remains the only option of therapy for many patients throughout the world.

The Fat Mass and Obesity-Associated Protein, also known as FTO, is an alpha-ketoglutarate-dependent dioxygenase. The FTO gene plays an important role in the management of energy homeostasis and the regulation of body weight [18]. This gene is located on chromosome 16q12.2 and has nine exons which encode a 2-oxoglutarate-dependent nucleic acid demethylase highly conserved in vertebrates [18]. Single nucleotide polymorphisms (SNPs) that cluster in the first intron of the FTO gene were first discovered in a genome-wide association study (GWAS) for type 2 diabetes mellitus [19], and, subsequently, other GWAS reported that the FTO polymorphisms were associated with obesity [20]. Evidence from epidemiological and functional studies suggests that FTO confers an increased risk of obesity by subtly changing food intake and preference [21]. The FTO rs9939609 polymorphism, one of the SNPs that were reported, has been associated with obesity [22], metabolic syndrome [23], insulin resistance [24], type 2 diabetes mellitus [19] and cardiovascular disease [25].

Considering that obesity, insulin resistance and steatosis have been identified as important factors that promote metabolic syndrome progression and failure of HCV therapy in HCV-infected patients [26], the aim of this study was to examine the association of the FTO rs9939609 polymorphism with the metabolic disturbances and virologic response to HCV therapy with pegIFNα/RBV in HIV/HCV-coinfected patients.

The association between the FTO rs9939609 polymorphism and metabolic factors is well documented in the general population [19],[22]-[24]. However, to the best of our knowledge, this is the first description of the relation between the FTO rs9939609 polymorphism and metabolic disturbances in HIV/HCV-coinfected patients. In this study, we showed that patients with an rs9939609 AA genotype had increased odds of obesity, insulin resistance, significant fibrosis and liver steatosis; and the presence of the risk A allele (both AA and AT genotypes) was associated with a lower response rate to pegIFNα/ribavirin therapy. This effect does not appear to be dependent on the HCV genotype, although we could not confirm this with certainty due to a lack of statistical power when we performed the analysis stratified by HCV genotype.

Obesity is a complex metabolic disturbance in which genetic and environmental factors may interact to produce homeostatic abnormalities [21], but in addition, both the HIV/HCV coinfection and the cART also influence the development of other comorbidities, such as insulin resistance and steatosis [7]. Recent meta-analyses have reliably established that FTO rs9939609 is robustly associated with increased BMI and adiposity across different ages and populations [22],[33]. Evidence to date suggests that the association between rs9939609 and BMI is predominantly driven by an increased energy intake, and that patients with the rs9939609 AA genotype exhibit overall increased ad libitum food-intake and impaired satiety [34]-[37]. In our study, subjects homozygous for the `obesity-risk’ A allele also had an increased risk for overweight/obesity (both continuous BMI values and BMI ≥27.5 kg/m²) compared with carriers of the low-risk T allele. Nevertheless, we must not discount the fact that our patients had a relatively low BMI (only 6.5% of patients were overweight), possibly due to the fact that around 85% of them were IDUs. The HIV infection and chronic drug abuse both compromise the nutritional status of patients despite major advances in HIV treatment [38], allowing that HIV-positive IDUs had lower BMIs.

The relationship between HCV and obesity is clinically relevant due to the potential synergism on liver disease severity and the multifaceted interactions between HCV and glucose metabolism [26]. HCV infection is able to perturb glucose homeostasis through several direct and indirect mechanisms, leading to both hepatic and extrahepatic insulin resistance [26], which accelerates liver disease progression (development of steatosis, fibrosis and hepatocellular carcinoma), reduces response to HCV antivirals, and increases the risk of developing type 2 diabetes mellitus [26]. However, the mechanisms by which CHC leads to insulin resistance are not fully understood. There is growing evidence that DNA polymorphisms may contribute to differences in complex disease traits between individuals. In this setting, many studies have investigated the relationship of the FTO rs9939609 polymorphism with insulin resistance and type 2 diabetes mellitus in the general population (HIV and HCV seronegative subjects) [19],[24],[39]-[41]. According with this, our study revealed that coinfected patients carrying the rs9939609 AA genotype had higher HOMA-IR values and a frequency of HOMA-IR ≥2.5 than patients carrying the rs9939609 AT/TT genotype. However, our data revealed that the FTO rs9939609 polymorphism was not associated with HOMA-IR ≥3.0 and HOMA-IR ≥3.8 (data not shown), possibly due to the limited sample size, since a near-significant trend was observed. Moreover, this lack of association could be due to the possible distortive effect of direct and indirect factors related to both HIV and HCV infections, and cART [7].

Liver steatosis is a frequent finding in CHC (≥40% cases). In HCV infection with GT3, viral factors are implicated in the development of steatosis via activation of de novo lipogenesis, while for HCV infection with GT1 or GT4, metabolic host factors, such as obesity and insulin resistance, provide an increased risk of steatosis [26]. Additionally, oxidative damage induced by the HCV core protein may simultaneously induce steatosis and impair insulin signaling in the hepatocyte [26]. The FTO gene is highly expressed in the hypothalamus and pancreatic islets, and it is found in other tissues such as adipose tissue, liver, and skeletal muscle. The FTO over-expression level seems to be related to subcutaneous fat accumulation [42], obesity [43] and oxidative stress and lipid deposition in the liver [44]. Furthermore, the rs9939609 AA genotype exhibits increased FTO expression compared with the TT genotype [35]. In our study, the rs9939609 AA genotype was related to higher likelihoods of steatosis, a finding that is consistent with the relationship found between rs9939609, obesity and insulin resistance in these same patients. This fact could have important clinical implications because persistent fatty liver disease may be a problem for patients even with successful HCV clearance [45],[46].

Our analysis also shows that the rs9939609 polymorphism was associated with increased odds of significant liver fibrosis. Considering that the development of liver fibrosis is strongly associated with overweight/obesity, insulin resistance and steatosis [7],[47],[48], we think that the association between the FTO rs9939609 polymorphism and liver fibrosis might be mediated by the metabolic disorders related to CHC.

Considering all the factors discussed above, it could be possible that the FTO rs9939609 polymorphism might be related to response to HCV therapy through regulating obesity, insulin resistance and liver steatosis in CHC patients [26]. In our study, the rs9939609 A risk-allele showed an inverse significant association with SVR. These associations remained after adjusting for the most important predictive factors related to HCV treatment response, such as IFNL3 genotype, HOMA-IR, HCV genotype, HCV-RNA viral load, and so on [13]. However, when patients were stratified by HCV genotype, the trends were maintained but were not statistically significant, possibly due to the limited sample size. In any case, we did not observe an association that could be dependent on HCV genotype. Moreover, we made an algorithm based on the genotypes of HCV, IFNL3 and FTO, finding that the FTO genotype was able to improve the classification into responder and non-responder to pegIFNα/ribavirin therapy for difficult-to-treat patients (GT1/4) with an unfavourable IFNL3 genotype (rs12980275 AG/GG). Moreover, the classification of GT2/3 patients also improved without the assistance of the IFNL3 genotype. In this case, the rs12980275 genotype was not included in the GT2/3 decision tree due to two main reasons: 1) For a decision tree analysis, the sample size in the GT2/3 group (56 patients) was quite limited to obtain valid results; and thus, the second node was pruned because it had fewer than 10 patients. 2) IL28B SNPs are not been useless for predicting HCV therapy outcome in HIV/HCV-coinfected patients infected with HCV-GT2/3 [49]. This fact is may be due to the high SVR rate found in these IFN-sensitive genotypes, where a larger sample size would be required to establish statistical differences [50]. Thus, the FTO rs9939609 polymorphism might provide a new clinical value in patients who have no access to newer DAAs treatment.

To date, many articles have assessed the influence of IL28B polymorphisms on SVR in CHC patients, rs12979860, rs8099917, and rs12980275 being the most studied [51]. These IL28B polymorphisms are being used as predictive markers of response to pegIFNα/RBV therapy in clinical practice, especially in patients with HCV genotypes 1 and 4 [13]. Although rs12979860 is more likely to be correlated with SVR in the European white population, we have recently shown a strong association of rs12980275 and rs8099917 with SVR in HCV/HIV-coinfected patients [52]. In the current study, we analyzed rs12980275, which is also in high linkage disequilibrium with rs8099917 and rs12979860 in the European population [50]. In addition, rs12980275 has been less studied than rs12979860 in European white populations, and, therefore, additional results involving rs12980275 would be of interest.

At present, the new DAAs have achieved a very high response rate [17]. This fact might obscure the influence on treatment efficacy of IFNL3 polymorphisms and other SNPs, such as the FTO rs9939609 polymorphism. However, some authors have still suggested that the IFNL3 genotype plays a key role for certain IFN-free regimens since several clinical trials have revealed an association between IFNL3 polymorphisms and treatment efficacy [53]. Moreover, the role of tools for making pre-treatment decisions may still be relevant since DAAs are more expensive and carry a higher risk for side effects, while decision-making based on IFNL3 polymorphisms and other SNPs might permit non-DAA-based treatment algorithms. Finally, it should also be borne in mind that few data are available in HIV/HCV coinfected patients about the interaction of IFN-free regimens and metabolic disturbances, where FTO is a cornerstone. Thus, further analysis will be needed to determine whether FTO rs9939609 polymorphism could provide additional information to select patients with better or worse response to HCV treatment.

There are some issues that have to be considered for a correct interpretation of our data:

Firstly, this report has a cross-sectional design (analysis of metabolic disturbances) and a retrospective design (analysis of virological response to HCV therapy), both with a relatively small number of patients, which could limit the achievement of significant values between the rs9939609 polymorphism and the outcome variables (for example: BMI ≥25 kg/m², HOMA-IR ≥3.0, HOMA-IR ≥3.8, HCV-therapy response according to HCV genotype, and so on). In addition, our cohort had a population with mixed HCV genotypes (1, 2, 3 and 4), which complicates the interpretation of the data since, for example, GT1 and GT3 did not have exactly the same pathophysiology and response to HCV treatment.

Secondly, metabolic disturbances are caused by several interacting genetic and environmental determinants, making it complicated to find the true individual effects of each disease-associated factor. In this regard, we did not have data on some extra factors that may have an influence on lipid levels and insulin resistance, such as exercise habits, diet, lipodystrophy, and alcohol intake.

Thirdly, the patients selected for our study were patients who met a set of criteria for starting HCV treatment (for example, no alcohol abuse, high CD4 cell counts, controlled HIV replication and good treatment adherence), and it is possible that this may have introduced a selection bias. Furthermore, HCV therapy regimens were not identical since they varied in some characteristics, such as pegIFNα 2a or 2b and likely ribavirin dose. Instead, each physician administered the appropriate HCV therapy regimen according to his/her criteria and by following local and/or international guidelines.

Fourthly, we did not study any cohort of HCV-monoinfected patients in order to evaluate the influence of the FTO rs9939609 polymorphism on CHC without the presence of HIV infection. In addition, we did not study any cohort of HIV-monoinfected patients in order to evaluate the influence of the FTO rs9939609 polymorphism on the development of metabolic disturbances with the presence of HIV infection and cART. Moreover, since the study was carried out entirely in white Europeans, and the frequency of these alleles differs among different ethnicities, it would be necessary to perform an independent replication of this study for different ethnic groups.

Fifthly, we evaluated several outcome variables and could raise the need to adjust the `P-value’. However, we think that it is not necessary to adjust the `P-value’ after multiple tests on clinical-orientated studies [54],[55], because: 1) the outcome variables cannot be considered completely independent; 2) there was a hypothesis supported by theory and we were not doing a random search of a meaningful result; and 3) our results had a clear interpretation. Thus, our results, which always pointed in the same direction, should not be ruled out. However, we should not overlook the level of uncertainty of these data.

DPT performed statistical analysis, interpretation of the data and drafting the manuscript. JB participated in the study conception and design, patient selection, and acquisition of data. MAJS, SVM, MGA, MGF, PGB participated in sample preparation, DNA isolation and genotyping pre-procedure, and performed a critical revision for intellectual content. TAE, AC, CD, FT participated in patient selection, collection of samples and acquisition of data, and performed a critical revision for intellectual content. SR participated in the study conception and design, statistical analysis, interpretation of the data and drafting the manuscript. All authors read and approved the final manuscript.