Our data showed diverse responses to lifestyle intervention when analyzing the entire sample or subgroups stratified according to the SNPs. This may suggest a role for genotype in the response to lifestyle interventions. Contrasting with numerous reports on the frequencies of these three SNPs (TNF-α -308 G/A, IL6 -174 G/C and AdipoQ 45 T/G) associated with body adiposity, metabolic disturbances and inflammation, their association with response to intervention on lifestyle has been rarely investigated. Admixture of the Brazilian population does not allow adequate classification by ethnic group; this condition may limit the comparisons of our findings with the same polymorphisms studied in other populations. Despite the heterogeneity of our population, the frequencies of the variant alleles did not markedly differ from data previously reported in literature [39–41].
The TNF-α, IL-6 and AdipoQ polymorphisms were selected considering the participation of these cytokines in underlying mechanisms of obesity-related metabolic disturbances, such as inflammation and insulin resistance. It is consistently demonstrated the deleterious effect of TNF-α in insulin signaling pathways . Both, TNF-α and IL-6 were shown to promote liver secretion of inflammatory biomarkers like CRP . Also, insulin-sensitizing, anti-inflammatory and anti-atherogenic properties of adiponectin have been described . Although it is clear that TNF-α and adiponectin exert opposite roles, findings regarding interleukins action are not so homogeneous.
The presence of the TNF-α -308 G/A polymorphism was associated with more favorable response of glucose metabolism to intervention, despite non-significant change in adiposity. Considering that the entire sample lose weight, it is possible that the subset with this TNF-α variant allele may have been protected against glucose disturbance independent of the body adiposity. At baseline, variant allele carriers had higher mean fasting plasma glucose levels, and exhibited a greater decrease after intervention compared with the reference group, even after adjustments. This is in line with a case control study in a Tunisian population, in which this SNP was not associated with obesity or T2D . However, the results of a meta-analysis indicated no association with variant allele and fasting plasma glucose or T2D .
On the other hand, other intervention-induced cardiometabolic benefits, such as increases in HDL-c and Apo A1 concentrations, were not detected in variant allele carriers but in the reference genotype group. Deleterious effects of this TNF-α variant allele was previously suggested in a meta-analysis in which individuals carrying the -308 G/A polymorphism had greater risk of obesity and comorbidities .
It has been postulated that functional polymorphisms in promoter region could affect transcription rates and circulating levels of cytokines . However, in our sample, the presence of the -308 G/A SNP did not influence TNF-α concentration at baseline or the response pattern to the intervention, since both genotype groups exhibited reductions in mean values of this cytokine. This is in agreement with a recent meta-analysis including healthy individuals that reported lack of associations between this TNF-α SNP and circulating levels, refuting a functional consequence of its presence .
In our study, some findings suggest that C allele carriers of the IL-6 -174 G/C polymorphism could have obtained more benefits from the intervention on lipid metabolism than those with the reference genotype, but not in glucose metabolism. Only C allele carriers showed a significant decrease in total cholesterol and LDL-c and increase in HDL-c and Apo A1 concentrations. Cross-sectional studies, conducted in several populations, had already reported a worse lipid profile in the reference genotype individuals compared with variant carriers. In Caucasian populations, significantly higher plasma triglyceride, total cholesterol and LDL-c and slightly lower HDL-c were observed in individuals with the reference genotype than in C variant allele carriers [17, 47]. However, the variant allele carriers did not reduce significantly fasting plasma glucose but did those with reference genotype. These contrasting results might be at least in part to a decrease in central adiposity verified only in the reference genotype group. Our findings are in agreement with those obtained from a Finnish Caucasian population in which GG genotype had lower fasting plasma glucose than C allele carriers .
Both genotype groups IL-6 -174 G/C SNP responded to the intervention with reductions in plasma TNF-α and CRP, but with unexpected elevations in mean IL-6 concentrations. Additionally, variant allele carriers had significant increase adiponectin concentration after intervention. Attenuation of pro-inflammatory actions of TNF-α and CRP and the favorable effect of adiponectin on inflammation and insulin sensitivity, those findings are desirable for individuals at high cardiometabolic risk undergoing lifestyle interventions.
To our knowledge, increased circulating levels of IL-6, particularly in carriers of IL-6 -174 G/C SNP submitted to long-term lifestyle intervention, have not been reported. It is known that IL-6 is released from skeletal muscle during exercise proportionally to exercise intensity . However, the increased output found in the recovery phase of prolonged exercise is most probably from the skeletal muscle and, to a lesser extent, from the subcutaneous adipose tissue . Our intervention program included physical activity but the majority of the sample performed light-to-moderate exercises (data not shown). Given the fact that IL-6, more than any other cytokine, is produced locally in the skeletal muscle in response to exercise, it is likely that it plays a beneficial role in mediating exercise-related metabolic changes .
Our recommendation to enhance physical activity during the intervention was attended by the entire sample since the proportion of active individuals increased at the end of the intervention. Despite no correlation detected between changes in physical activity and changes in IL-6 concentration (data not shown), we speculate that the IL-6 elevation in our sample could be partially attributed to exercise. This finding does not seem to reflect a pro-inflammatory effect, since mean values of CRP reduced significantly. Further investigations, including a large spectrum of interleukins, are necessary to understand their role in interventions on life habits.
The group of carriers of variant allele AdipoQ 45 T/G SNP did not change fasting plasma glucose and adiponectin concentrations but the reference genotype group did. We speculate that the lack of increment in adiponectin concentration after intervention may have contributed to impede benefits in glucose metabolism in the variant allele carriers. Controversial data have been reported among populations [27, 29, 51]. Our result may be considered somehow different from a case–control study, since higher adiponectin concentration was found among variant allele homozygous in healthy Korean woman . In contrast, a cross-sectional survey in Chinese population showed lower plasma adiponectin in variant allele carriers compared with the TT genotype . Data from the STOP-NIDDM trial  and a case control study in obese Iranian population  are in line with ours, since they support that variant G allele might confer susceptibility to T2D.
Our findings indicated unfavorable effects of the variant allele of the AdipoQ 45 T/G in lipid profile, when compared to the entire sample and to the reference genotype group. Carriers of the G allele did not decrease Apo B nor increase HDL-c concentrations after intervention. Similarly to ours, in a Chinese cross-sectional study, G carriers had higher total and LDL-c levels than the reference genotype group . The beneficial response of CRP concentrations to our intervention seems having not depended on the AdipoQ 45 T/G SNP. This set of evidence suggests the presence of this SNP impedes metabolic response to lifestyle interventions.
The comparison of the group the individuals with the three variants associated with worse glycemic response with the remainder sample could suggest that the combination of polymorphisms would enhance the negative impact on glucose metabolism. Despite similar fasting plasma glucose at baseline, the former group finished the intervention with higher levels, which would be compatible with this hypothesis. This speculation deserves further investigation based on large samples.
Despite the strength of addressing important SNPs codifying for proteins known to affect inflammatory status and insulin sensitivity, and examining a great number of metabolic biomarkers, our study has the limitation of the sample size. The small sample size could be responsible for the absence of Hardy-Weinberg equilibrium for TNF -308 G/A. Actually, a study including a large sample of individuals from the same region of Brazil indicated that this SNP is in HWE . Another possible explanation for such finding could be that this SNP itself would be a risk factor for diabetes and cardiometabolic risk. Therefore, our results should be considered preliminary at best.
In summary, the TNFα -308 G/A SNP may predispose a better response of glucose metabolism to intervention. The IL-6 -174 G/C SNP may confer a beneficial effect on lipid but not in glucose metabolism. Our findings reinforce unfavorable effects of the AdipoQ 45 T/G SNP in lipid profile and glucose metabolism after lifestyle intervention in Brazilians at cardiometabolic risk. Further studies on SNPs of genes involved on the control of inflammatory status and insulin sensitivity should provide more subsidies to achieve the goals of preventing metabolic disturbances through lifestyle changes in individuals at high cardiometabolic risk.