Metabolites related to purine catabolism and risk of type 2 diabetes incidence; modifying effects of the TCF7L2-rs7903146 polymorphism

Studies examining associations between purine metabolites and type 2 diabetes (T2D) are limited. We prospectively examined associations between plasma levels of purine metabolites with T2D risk and the modifying effects of transcription factor-7-like-2 (TCF7L2) rs7903146 polymorphism on these associations. This is a case-cohort design study within the PREDIMED study, with 251 incident T2D cases and a random sample of 694 participants (641 non-cases and 53 overlapping cases) without T2D at baseline (median follow-up: 3.8 years). Metabolites were semi-quantitatively profiled with LC-MS/MS. Cox regression analysis revealed that high plasma allantoin levels, including allantoin-to-uric acid ratio and high xanthine-to-hypoxanthine ratio were inversely and positively associated with T2D risk, respectively, independently of classical risk factors. Elevated plasma xanthine and inosine levels were associated with a higher T2D risk in homozygous carriers of the TCF7L2-rs7903146 T-allele. The potential mechanisms linking the aforementioned purine metabolites and T2D risk must be also further investigated.

www.nature.com/scientificreports www.nature.com/scientificreports/ questionnaire was filled in for each participant 18 . Body mass index (BMI) was calculated as weight divided by height squared (kg/m 2 ). Participants' triacylglycerol (TAG), total cholesterol, high-density lipoprotein (HDL) and low-density lipoprotein (LDL) levels were measured using fasting plasma at baseline. Blood glucose and insulin levels were centrally assessed at baseline and at the end of 1-year of follow-up. Insulin resistance was estimated by the HOMA method using the following equation: HOMA-IR was calculated using the following equation: HOMA-IR = [fasting insulin (μIU/mL) × fasting glucose (mmol/L)]/ 22.5. Statistical analysis. Baseline characteristics of cases and non-cases were described as means and standard deviations (SD) for quantitative variables, and percentages or numbers for categorical variables. We applied a natural logarithmic transformation to approximate a normal distribution of metabolites levels. We also examined product-to-precursor ratios of metabolites, as described in the pathway of purine catabolism, as a metabolite trait (by dividing the raw values and then taking natural logarithmic transformations). Person-time of follow-up was calculated as the interval between the randomization date and date of T2D event, death, or date of the last contact, whichever came first. We used Cox proportional hazard models, with Barlow weights (to account for the over-representation of cases), to estimate hazard ratios (HRs) and their 95% confidence intervals (CIs) for risk of T2D. A crude model and two multivariable-adjusted Cox regression models were fitted as follows: a) multivariable model 1 (MV1), adjusted for age (years), sex (male, female), body mass index (kg/m 2 ), intervention group and baseline fasting glucose (mg/dl) (adding a quadratic term to account for the departure from linearity; b) MV2, additionally adjusted for the TCF7L2-rs7903146 polymorphism (assuming an additive genetic model), smoking (never, current, former), leisure-time physical activity (metabolic equivalent tasks in minutes/day), baseline dyslipidaemia (yes/no), and hypertension (yes/no). We stratified the models according to recruitment center. Baseline metabolites levels and their ratios were analysed as both continuous variables (1-SD increment in their transformed levels) and using quartiles (using cut-points defined among non-cases). To appraise the linear trend across quartiles, the median metabolites levels and their ratios within each quartile was included in the Cox regression models as a continuous variable. To account for multiple testing, we adjusted p-values of the multivariable adjusted associations between quartiles or 1-SD increment in metabolites levels and T2D risk, using Benjamin-Hochberg false discovery rate (FDR) procedure 19 . A FDR-p-value < 0.05 was considered to be statistically significant. We also examined the associations of 1-year changes in metabolites and their ratios with T2D risk. We used the same models as in the baseline analyses but further adjusted for baseline metabolites levels or their ratios. With respect to metabolites or their ratios, we first calculated the ratio between 1-year and baseline levels and then normalized this ratio with the natural logarithmic transformation. To test the robustness of our findings we conducted one sensitivity analysis on the associations of baseline metabolites or their ratios with risk for T2D by including two variables (HDL and TAG) in the multivariable model 2 that were significantly different between cases and non-cases, leaving 71 incident cases due to the high percentage of missing values (HDL: 41%, TAG: 33%). To examine whether the aforementioned genotypes modified the association between metabolites or relevant ratios and T2D incidence, interactions were tested by including a multiplicative interaction term (metabolite x TCF7L2-rs7903146 genotypes) or (ratio x TCF7L2-rs7903146 genotypes) and also a TCF7L2-rs7903146 genotypes main effect in the MV2. Stratified analyses by TCF7L2-rs7903146 genotypes were also carried out assuming an additive genetic model. The associations were adjusted for multiple testing as described above. We also estimated the joint association of plasma metabolites levels and rs7903146 polymorphisms (TT or CC genotypes) with T2D. We considered as the reference group those participants with CC genotype and metabolites levels (lower than the median). Statistical analyses were performed using Stata 13.1 (Stata Corp., College Station, Texas, USA). A two-sided p value less than 0.05 was considered significant.

Results
Participants' characteristics. The median follow-up of the study population was 3.8 years. A total of 251 incident cases and 641 control participants were included in the study. Participants' characteristics are summarized in Table 1. The mean age of participants at baseline was 66.5 years and the mean BMI was 30.1 (3.5) kg/m 2 . Briefly, a higher proportion of incident cases were men and current smokers compared to non-cases. As compared with non-cases, those participants who developed T2D were also more likely to have a higher prevalence of hypertension and dyslipidaemia in addition to higher BMI, fasting glucose and triacylglycerol levels while lower HDL cholesterol levels ( Table 1). Table 2 shows the associations of baseline levels of individual metabolites and relevant ratios with T2D risk. In fully-adjusted model (MV2), allantoin when modelled as quartile, was significantly associated with lower T2D risk [HR in the highest versus lowest quartile was 0.44 (95% CI 0.25-0.80). Regarding ratios of allantoin-to-uric acid and xanthine-to-hypoxanthine, the estimated HR for incident T2D in the highest versus lowest quartile was 0.42 (95% CI 0.22-0.82) and 2.53 (95% CI 1.40-4.58), respectively. These associations remained significant after accounting for multiple comparisons. There were no significant associations between other baseline metabolites or ratios and incident T2D, either they were modelled continuously (per 1 SD) or as quartiles.

Associations of baseline metabolites and relevant ratios (precursor-product) with risk for T2D.
One year changes in levels of metabolites, relevant ratios (precursor-product) and risk of T2D. Associations between 1-year changes in metabolites levels and relevant ratios across quartiles with the risk of T2D are shown in Supplemental Table 1. In the highest quartile of increase in inosine-to-adenosine and xanthine-to-guanosine ratio a significant inverse association with T2D risk was found, with HR 0.43 (95% CI 0.19-0.95), and HR 0.42 (95% CI 0.20-0.88), respectively. We repeated the analyses using 1-SD increment in 1-year changes of metabolite levels and relevant ratios and found that per 1-SD increase in xanthine-to-guanosine www.nature.com/scientificreports www.nature.com/scientificreports/ ratio, the risk of T2D was associated with a decrease of 22%, with HR 0.78 (95% CI 0.62-0.98) (Supplemental Table 2). After adjusting for multiple testing, none of these associations remained statistically significant. Table 3 shows the associations of baseline levels of individual metabolites and relevant ratios with T2D risk after including HDL cholesterol and TAG in the fully adjusted model. The associations between allantoin and allantoin-to-uric acid ratio with T2D risk remained significant.
Interaction between the TCF7L2-rs7903146 polymorphism and metabolites on T2D risk. As expected, TT individuals had a higher risk for T2D compared with CC homozygotes, with HR 2.03 (95% CI 1.12-3.70), p = 0.020, after controlling for several potential confounding factors. No excess risk was conferred by the heterozygous genotype [CT vs. CC: HR 1.19 (95% CI 0.79-1.79), p = 0.387]. We found statistically significant interactions between metabolites levels, a relevant ratio and the TCF7L2-rs7903146 polymorphism in determining T2D risk in an additive model (Table 4). We found that per 1-SD increase in plasma xanthine and inosine levels, the risk of T2D significantly increased in TT individuals, with HR 2.34 (95% CI 1.23-4.45), and HR 3.09 (95% CI 1.23-7.73), respectively. On the other hand, per 1-SD increase in hypoxanthine-to-inosine ratio, a 74% lower risk of T2D was found in TT subjects [HR 0.26 (95% CI 0.10-0.70)]. After adjustment for multiple comparisons the associations between these metabolites, the ratio and T2D risk remained statistically significant. When the joint effects were examined, individuals with TT genotype and plasma xanthine levels above/equal to the median value had significantly higher risk of T2D [HR 2.64 (95% CI 1.02-6.83)] than those with CC genotype and xanthine levels below the median (reference) (Fig. 1a). Similarly, higher risk was found in TT individuals when levels of inosine were higher or equal to the median [HR 2.65 (95% CI 1.05-6.69)] (Fig. 1b). The levels of the aforementioned metabolites did not significantly differ across the genotypes in the additive model.

Discussion
In the present case-cohort design study within the PREDIMED trial, after adjusting for recognized T2D risk-factors and multiple testing, we found inverse and positive associations between high baseline levels of allantoin, including allantoin-to-uric acid ratio and high xanthine-to-hypoxanthine ratio with T2D risk, respectively. We also found that elevated plasma xanthine and inosine levels were associated with a higher T2D risk in individuals with TT alleles on rs7903146. This is the first time, as far as we know 10,11 , that this genetic variant of the TCF7L2 gene, which mediates high susceptibility to T2D, is reported to interact with purine-catabolism metabolites.
Allantoin is produced from the non-enzymatic oxidation of uric acid (Fig. 2) in humans and is considered to be a specific biomarker of oxidative stress 3,20 . We found that higher allantoin levels were associated with lower risk of T2D incidence as was the allantoin-to-uric acid ratio. This ratio was also negatively correlated with HOMA-IR at baseline. Our results are in accordance with the literature. When administered through injection, allantoin decreased blood glucose levels and increased blood insulin in a dose-dependent manner, in normal rats 21 . Allantoin has also been found to lower plasma glucose in diabetic rats by activating imidazoline-receptors I2 or I3 21 , which improve insulin action. Furthermore, in diabetic rats, allantoin offered through diet had antidiabetic effects by modulating antioxidant activities and lipid profiles, promoting glucagon-like peptide-1 (GLP-1) release, thereby facilitating β-cells to maintain insulin and glucose levels 22 . Finally, in a cohort study, allantoin measured in urine of humans was also inversely associated with the risk of T2D 20 .
Uric acid is the end product of purine metabolism in humans. In our study, plasma uric acid was not associated with future risk of T2D, but correlated positively both with HOMA-IR at baseline and at 1-year follow-up. The causal association between serum uric acid levels and T2D remains controversial 23 . Uric acid is generally considered to be a feature of hyperinsulinemia or/and insulin resistance 23 ; however, in patients with diabetes, its level may be low, due to increased urate clearance that can be associated with glycosuria, increased sodium (Na) secretion, and overall decreased metabolic control 24 .
Uric acid can be produced from xanthine via the enzymatic action of XO, which also participates in the conversion of hypoxanthine-to-xanthine (Fig. 2). In our study, baseline xanthine levels and xanthine-to-hypoxanthine ratio were correlated with higher HOMA-IR levels. In addition, we found that this ratio was associated with increased risk of T2D. An increase in this ratio may reflect higher XO activity, which is involved in free radical production 25 and has been found, increased in T2D patients 5 . Whether xanthine production from hypoxanthine, plays a role in T2D development due to increased XO activity or due to xanthine production in parallel with the consequent decrease in hypoxanthine, requires further investigation.
In the present study, guanosine was also positively correlated to HOMA-IR. Cyclic Guanosine monophosphate (cGMP) is a second messenger that mediates incretin effects; potentiates glucose-stimulated insulin secretion; promotes proper beta-cells differentiation, and prevents beta-cells apoptosis, cooperating with biotin 26 . In addition, cGMP is involved in various signal-transduction pathways, mediating messages of insulin itself 27 . The interactions between the extracellular and intracellular guanosine metabolites, or/and the possible modulations of the latter in T2D, remain an open issue.   www.nature.com/scientificreports www.nature.com/scientificreports/ Inosine, a precursor of xanthine that was not associated to diabetes incidence in the overall population of our study, has notable anti-inflammatory effects, which may be mediated, at least in part, by activation of the adenosine A2a receptor 28 . Both, adenosine and inosine have been suggested to play a protective role against diabetes development 28 . We observed that inosine-to-adenosine ratio was negatively correlated with baseline HOMA-IR. A working hypothesis is that the production of inosine is more favourable in improving glucose homeostasis, as compared to adenosine.
The TCF7L2-rs7903146 polymorphism is one of the strongest and most widely replicated locus associated with T2D 29 , with the homozygous individuals (TT) being those who present a higher prevalence and fasting   glucose; the mechanism has yet to be determined 14 , although various hypotheses, affecting regulation of the Wnt signalling-pathway 30 have been proposed. The key effector of Wnt pathway, the bipartite transcription factor β-cat/TCF, is formed by free β-catenin (β-cat) and a TCF protein, including TCF7L2 31 . Wnt signalling and TCF7L2 appear to exercise a very complex effect in metabolic homeostasis, affecting not only the pancreatic islets, but also other organs (liver gluconeogenesis), and crosstalk with stress, aging as well as tumorigenesis signalling pathways/cascades 31 . Several signalling components of the Wnt signal transduction pathway have been identified but a clear understanding of the Wnt signalling's diverse function, integration and specificity is lacking. On the  www.nature.com/scientificreports www.nature.com/scientificreports/ other hand, there is strong evidence for a direct association between dysregulated Wnt signalling and chronic diseases 32 . In due course, choosing the rs7903146-allele, was an innovative approach in our study.
In this study, using a case-cohort design, we confirmed the association of the TCF7L2-rs7903146 TT genotype, with the risk of T2D in PREDIMED study participants 14 . Changes in the blood of TT homozygotes reported until now mainly concerned increased levels of plasma glucose in response to a meal challenge of proinsulin and elevated glucose-dependent insulinotropic peptide secretion 33 ; reduced secretion of insulin/glucagon, and reduced insulinotropic effect of incretin hormones 34 ; altered postprandial triglyceride response, mainly influencing VLDL and HDL subclasses 35 , as well as non-significant increase of plasma sphingomyelins, phosphatidylcholines and lyso-phosphatidylcholines species 36 . To our knowledge, this is the first study to examine the interaction between plasma purine-catabolism metabolites levels and the TCF7L2-rs7903146 genetic variation, focusing on T2D risk.
We observed that xanthine and inosine were associated with increased T2D risk only in individuals with TT alleles on rs7903146. Moreover, the combination of TT genotype and high plasma xanthine and inosine levels was associated with a higher risk of T2D than the combined CC genotype and their low levels, which confirmed that both the rs7903146 T allele and increased plasma xanthine and inosine are associated with T2D. The protective role of inosine is well-documented 37 , while xanthine may simply increase due to increased XO activity in the blood 38 . A working hypothesis is that xanthine and inosine play a compensatory role, competing or cooperating; since in both, when their levels were low, no significant association was found between TT and T2D, as compared to their high levels (see joint analysis). Among purine-catabolism related ratios, only hypoxanthine-to-inosine ratio was inversely associated with T2D risk in individuals with TT alleles on rs7903146 with a significant test of interaction. Whether hypoxanthine production from inosine confers higher protection from T2D in individuals carrying these polymorphisms, by inhibiting the activation of poly(ADP-ribose) polymerase 39 and thus increasing TCF7L2-mediated GLP-1 production and activity 40 , needs to be further explored.
The results of the present study should be interpreted in the context of its limitations and strengths. First, participants were elderly Mediterranean individuals at high cardiovascular risk and this may limit the generalizability of the findings to other age-groups or populations. Second, even though we adjusted for several potential confounders, residual confounding may exist. Regarding strengths, the prospective evaluation of the association between metabolites levels and well-documented incident T2D, in the frame of a case-cohort design, minimizes biases that can affect case-control studies.
In conclusion, our prospective study documented, for the first time, an inverse and a positive association between high plasma allantoin levels, including allantoin-to-uric acid ratio and high xanthine-to-hypoxanthine ratio with incident T2D risk, respectively, in an elderly population at high cardiovascular risk, independently of the TCF7L2-rs7903146 polymorphism. Elevated plasma levels of xanthine and inosine appeared to be associated with higher T2D risk only in TT individuals. These results must be interpreted cautiously and need to be replicated in other populations. The potential mechanisms linking the aforementioned purine metabolites and T2D risk must be also further investigated.