Genetically Guided Mediterranean Diet for the Personalized Nutritional Management of Type 2 Diabetes Mellitus

The current consensus for the prevention and management of type 2 diabetes mellitus (T2DM) is that high-quality diets and adherence to a healthy lifestyle provide significant health benefits. Remarkably, however, there is little agreement on the proportions of macronutrients in the diet that should be recommended to people suffering from pre-diabetes or T2DM. We herein discuss emerging evidence that underscores the importance of gene-diet interactions in the improvement of glycemic biomarkers in T2DM. We propose that we can achieve better glycemic control in T2DM patients by coupling Mediterranean diets to genetic information as a predictor for optimal diet macronutrient composition in a personalized manner. We provide evidence to support this concept by presenting a case study of a T2DM patient who achieved rapid glycemic control when adhered to a personalized, genetically-guided Mediterranean Diet.


Introduction
Long considered a disease of minor significance, type 2 diabetes mellitus (T2DM) currently represents a major threat to human health with an estimated number of 425 million adult patients and four million deaths globally in 2017. Substantial evidence indicates that T2DM can be largely prevented or managed through adherence to a healthy lifestyle and a high-quality diet (as assessed by Healthy Eating Index-HEI [1]) is recommended to people at risk for diabetes or to T2DM patients as a therapy [2][3][4][5]. Surprisingly, however, there are no specific recommendations for the proportions of macronutrient intake that should be applied for people suffering from pre-diabetes or diabetes [6]. The Obesity Society, for example, recommends a diet comprising approximately 30% energy from total fat [15-20% from monounsaturated fatty acids (MUFA), 10% from polyunsaturated fatty acid (PUFA), 7% from saturated fatty acids (SFA)], 15-35% from protein and 45-65% energy from carbohydrates for the management of T2DM [7]. On the other hand, the Mediterranean diet (Med Diet) which demonstrates benefits over several metabolic diseases including T2DM, is considered a high fat diet (40% total fat) but rich in MUFA and poor in saturated fat [8][9][10].
Notably, the traditional Med Diet is rich in fibers and high-quality foods that have been proven a key strategy to achieve glycaemic control [11]. For example, grains, whole grain flours and breads from traditional Greek wheat varieties of Triticum monococcum and Triticum dicoccum, contain very high levels of alkylresorcinols [12] which have been associated with increased insulin sensitivity in metabolic syndrome patients [13] but are absent from white flour and its products.
It is therefore evident that there are no "golden rules" guiding the proportion of macronutrients for T2DM dietary therapy [14]. Moreover, even though the Med Diet has been proven successful in reducing chronic disease burden at population level [15], it does not take into account individualized needs and, as many other diets, it is based on population average. As a result, the search for an ideal high-quality diet for diabetes patients continues to grow.
In addition, T2DM is a highly heterogeneous disease influenced by genetic factors and nutrient-gene interactions, as indicated by twin and family studies [16][17][18]. With the emergence of genome wide association studies (GWAS) and more recently exome sequencing, a plethora of data has provided additional evidence for a genetic basis for T2DM. More than 140 genomic loci have been associated with predisposition to T2DM, improving our understanding of the genetic architecture and biology of the disease [19]. Genetics also influences individual responses to both macro-and micro-nutrients [20,21] and emerging evidence underscores the importance of gene-diet interactions in the improvement of glycaemic biomarkers.
We herein review the published evidence linking genetic variation to the selection of specific macronutrient composition for the management of T2DM (Table 1). Our overarching concept is to bridge the Med Diet with genetic information that can be used as predictor for optimal diet macronutrient composition in a personalized manner. We propose that by taking into account genetic information when formulating Med Diet-based nutritional recommendations, we can achieve better glycaemic control in T2DM patients. Herein we provide evidence to support this concept by presenting a case study of a T2DM patient who achieved rapid glycaemic control when adhered to a personalized, genetically-guided Med Diet. In high-fat diet, the T allele was associated with a higher reduction of insulin and HOMA-B. # The opposite effect was observed in the low-fat diet group, although in this group the T allele was marginally associated with insulin and HOMA-B, 757 nondiabetic individuals who were randomly assigned to 1 of 2 energy-restricted diets that differed in macronutrient composition [29] Genetic score of SNPs related to habitual coffee consumption-

SNP
Low-fat diet (20% fat) and high-fat diet (40% fat) Participants genetically prone to high coffee consumption may benefit more by eating a low-fat diet in improving fasting insulin and HOMA-IR in a short term (Actual coffee consumption was not taken into account) TCF7L2 Transcription Factor 7-Like 2 rs7903146 (C/T) T risk allele for T2D; C is the common allele

High intake of desserts and milk
The T2D risk was greater in T allele carriers with high dessert and milk. In subjects with a high dessert intake, the T allele was also associated with higher fasting plasma glucose concentrations 787 subjects (378 men and 409women, aged between 30 and 64). [46]

Genetic Variants Guiding the Quantity of Carbohydrate Intake
Even though carbohydrates (CHO) have a major dietary influence on postprandial blood glucose, evidence on total carbohydrate needs and T2DM is largely conflicting [26]. Surprisingly, even the anticipated protective effect of total fruit intake on T2DM could not be replicated in a meta-analysis contacted in the context of the European Prospective Investigation into Cancer (EPIC)-InterAct study [47]. Genetic variation is likely to account for the heterogeneity of T2DM patient responses to carbohydrates (Table 1). Indeed, several polymorphisms have recently been identified to influence glucose levels in response to carbohydrate intake. Among them, transcription factor 7-like 2 (TCF7L2) gene variant rs7903146 has attracted significant attention as it is one of the strongest genetic markers of predisposition to diabetes, with the TT genotype increasing disease risk by 40-50% [48]. Interestingly, even though this high risk polymorphism is associated with lower BMI ( [49] and our unpublished data on Greek population), it is also associated with significantly higher increase of fasting glucose when consuming increased amounts of desserts compared to non-carriers [46].
The melatonin receptor 1b (MTNR1B) variant rs1387153 is another example of genetic variation influencing T2DM patient responses to CHO. An analysis of 5 cohort studies including more than 28,000 participants has demonstrated that the minor T allele of rs1387153 strongly interacts with CHO in modulating fasting glucose, to the extent that every 1% increase in CHO intake exacerbates the fasting glucose-raising effect of the T allele [22]. Cryptochrome 1 (CRY1) variant rs2287161, which has been associated with the regulation of circadian rhythms and linked to depression and sleep disturbance [50], also modifies the CHO effect on glycaemic indices. In this case, increased carbohydrate intake (as a percentage of energy intake) was associated with elevated HOMA-IR and fasting insulin only among individuals homozygous for the risk allele C [24]. In contrast, G homozygotes for pro-protein convertase subtilisin/kexin type 7 (PCSK7) gene variant rs236918 who were assigned to a high-carbohydrate diet displayed a greater decrease in fasting insulin levels and HOMA-IR than non-carriers [25]. Interestingly, both MTNR1B and PCSK7 polymorphisms have been associated with increased risk of diabetes [51,52].

Genetic Variants Guiding the Quality of Carbohydrate Intake
The quality of carbohydrates appears to play important roles in the management of T2DM and the results of relevant prospective studies and clinical trials are much more coherent compared to those addressing the impact of the quantity of carbohydrates on T2DM [53]. People with diabetes and those at risk for diabetes are encouraged to consume at least the amount and quality of dietary fibre recommended for the general public that includes vegetables, pulses and fruit [54].
However, emerging evidence suggests that genetic variation may influence individual responses to different sources of fibre. Among them, TCF7L2 gene variant rs7903146 has attracted significant attention [48]. Interestingly, two independent studies have reported that whole grain as a dietary source of CHO is associated with protection from T2DM only among non-risk CC genotype carriers [26,55]. This suggests that carriers of a T allele of rs7903146 would not benefit from a whole grain-enriched diet and alternative dietary approaches should be explored.
Likewise, the glucokinase regulatory protein (GCKR) gene variant rs780094, which has been explored as a component of polygenic risk for T2DM and dyslipidemia [56] diminishes the beneficial effects of whole-grain foods on insulin homeostasis, possibly via the reported effect of GCKR variant on both triglyceride and glucose levels [27].
Another gene polymorphism associated with the response of T2DM patients to particular sources of carbohydrates is apolipoprotein-A5 (APOA5) SNP rs964184. It has been reported that compared to rs964184 TT homozygotes, T2DM patients with impaired fasting glucose who carry the C allele of rs964184 could be more susceptible to the adverse effects of a high carbohydrate diet based on refined rice by having an elevation of triglycerides [23]. Therefore, the replacement of refined rice with whole grains and legumes in a high carbohydrate diet should be considered for individuals with impaired fasting glycemia who carry the APOA5 rs964184 C variant, to prevent diabetic hypertriglyceridemia.

Genetic Variations Guiding Fat Intake in T2DM
Even though fat consumption has broadly been associated with elevated risk for the development of metabolic diseases, there is still a debate about the recommended daily intake of fat in T2DM patients, reportedly ranging from 10% up to 40% [14]. Conflicting data may be due to variations in food sources of fat [57] or the proportion of fat in the context of the overall dietary pattern. Indeed, according to the PREDIMED clinical study, individuals who adhered to a Med Diet plan supplemented with extra virgin olive oil had on average 40% lower risk to develop diabetes compared to those following a lowfat diet [58].
Several genes have been identified to impact responses to fat in T2DM. One of them is PPM1K (phosphatase, Mg 2+ /Mn 2+ -dependent 1 K) gene variant rs1440581. Individuals who are homozygous for the T allele of this SNP benefit most by a high-fat diet (40-45% fat, 15% protein, and 40-45% carbohydrate) in reducing insulin and homeostatic model assessment for β-cell function (HOMA-B) during weight loss whereas the opposite effect is observed for a low fat diet (20-25% fat, 15% protein, and 60-65% carbohydrate) [29]. Similarly, in another intervention study, the C allele was related to smaller decreases in serum insulin and homeostatic model assessment for insulin resistance (HOMA-IR) in obese and overweight individuals following an energy-restricted high-fat diet plan (40% energy intake from fat), whereas an opposite genotype effect on changes in insulin and HOMA-IR was observed in a low-fat diet group (20% energy intake from fat [28]). Together, these reports suggest that pro-diabetes or T2DM patients with PPM1K rs1440581 TT genotype would benefit from an energy-restricted high-fat diet whereas carriers of the C allele could be advised to follow a low-fat diet.
Genetic risk scores (GRS), generated by combining the additive effects of several genetic variants, have also been explored to predict optimal dietary fat composition for T2DM patients. For example, a high genetic risk score (GRS) of 14 fasting glucose-associated SNPs could predict health benefit for a diet low in fat [32]. Furthermore, in another study [30] a significant interaction between a genetic score calculated on the basis of eight variants previously associated with habitual coffee consumption [59] and dietary fat intake was observed. It was shown that the group with the highest genetic score for coffee consumption had significantly greater reduction in fasting insulin when following a low fat diet. Interestingly, even though the initial idea of this study was based on the fact that coffee consumption is associated with improved insulin sensitivity and reduced risk for diabetes [60], actual coffee consumption was not considered [30].
Variations in food sources of fat may play an important role in modulating fasting glucose and other bioclinical variables in T2DM patients. In general, replacing saturated with unsaturated fat in the diet reduces total cholesterol, LDL-C and CVD risk [61]. Surprisingly, a meta-analysis of observational studies has not identified high dietary saturated fatty acid (SFA) intake as a risk factor for T2DM [62] However, this conclusion should be considered through the prism of genetic variation. Indeed, high dietary SFA intake (≥15.5% energy) impairs insulin sensitivity and increases the risk for metabolic syndrome in carriers of the T risk allele of TCF7L2 rs7903146 polymorphism relative to CC homozygotes [36] Likewise, high dietary SFA intake (≥11.8% energy) leads to increased waist circumference and consequently increases the risk for metabolic syndrome in carriers of the C risk allele of CLOCK rs1801260 [31]. Furthermore, diets rich in MUFA and PUFA may have a dual role as they can either reveal the protective effect that some polymorphisms have towards T2DM [35] or to abolish the detrimental effects of some others [34].

Genetic Variations Guiding Protein Intake in T2DM
Typically, the average daily level of protein intake for people with diabetes is 1-1.5 g/kg body weight/day or 15-20% of total calories. However, two published meta-analyses have indicated that high protein diets confer several health benefits in diabetes patients [63,64].
We reason that genetic variation may affect individual responses to different protein percentages in the diet (Table 1). Several studies support this notion. For example, polygenic scores, based on combinations of multiple susceptibility loci identified by GWAS, are increasing explored because they reflect the polygenic nature of T2DM and are more relevant to gene-diet interactions. A recent study demonstrated interaction between a polygenic score of 31 T2DM risk alleles and protein percentage in the diet. In particular, individuals with a lower polygenic score for T2DM showed more favourable responses to low-protein diets, including greater decreases in fasting insulin, HbA1c, and homeostatic model assessment for insulin resistance (HOMA-IR), and a lesser increase in and homeostatic model assessment for β-cell (HOMA-B) function, than did individuals with a higher risk score within 2 years of follow-up [38]. Another study similarly explored 159 obesity genes combined with phenotypic characteristics such as waist hip ratio and BMI to create different adiposity subtypes which were found to differentially modify the effect of protein intake on improving glucose metabolism [39].
An interaction between vitamin D metabolism-related variants and macronutrient responses in the context of T2DM was demonstrated by the Pounds Lost Trial group; patients with the vitamin D-increasing T allele of the DHCR7 variant rs12785878 had a greater improvement of glycaemic parameters when assigned to high-protein diets, whereas improvement was not evident in those who followed a low protein diet [40].

Genetic Variations and Mixed Dietary Patterns for the Management of T2DM
A variety of dietary patterns are acceptable for the management of diabetes. Still, the evidence surrounding comparative benefits of different macronutrient ratios or dietary patterns in T2DM patients and the influence of genetics thereof is scarce [14].
One example of genetic variation influencing responses to mixed dietary patterns is GIPR variant rs2287019 (Table 1). T2DM patients who carry the T allele of rs2287019 and choose a low-fat, high-carbohydrate and high-fibre diet, show greater improvement in glucose homeostasis [42]. Likewise, it has been reported that T2DM patients carrying the IRS1 variant rs2943641 CC genotype might obtain more benefits in weight loss and improvement of insulin resistance than those without this genotype when they adhere to a high-carbohydrate and low-fat diet [43]. Additionally, T2DM patients who are homozygous for the C allele of S100A9 (S100 calcium-binding protein A9) rs3014866 variant are more likely to benefit from a low SFA: carbohydrate ratio intake to improve insulin sensitivity, based on HOMA-IR levels [41]. A western type diet, characterised by high intake of processed meat, red meat and heme iron, increases the risk of T2DM in individuals with a high polygenic score defined by ten T2DM risk alleles [45].

Discussion
Rapidly evolving technologies have offered unparalleled opportunities to assess bioclinical, dietary and genetic data that can improve the management of metabolic diseases such as T2DM. As indicated in this review, high-quality diets rich in n-3 PUFA and MUFA and low in saturated fat are universally proposed for the prevention and management of T2DM. This concept aligns with the Mediterranean dietary patterns typified by high-quality nutritional elements offering several health benefits. Notably, there is significant variation in the macronutrient distribution among different Mediterranean diets (Cretan/Greek, Italian, Spanish or Mediterranean-like) [65]. Likewise, dietary recommendations for T2DM patients largely vary with respect to macronutrient distribution which may impair the efficacy and goals of nutritional therapy for diabetes [14].
We propose that prevention and management of T2DM could be improved by combining genetically-guided quantitative and qualitative macronutrient recommendations (Table 1) with the high quality nutrition patterns of Med Diets. We have successfully applied this concept to several T2DM patients. As an example, a normal weight male T2DM patient, homozygous for MTNR1B rs1387153 risk allele T, with a high genetic risk score for diabetes and a high genetic score for coffee consumption was advised to change the macronutrient distribution of his diet from 54% carbohydrate, 32% fat and 16% protein to lower carbohydrate (50%), low fat (25%) and high protein (25%) diet based on Mediterranean high quality foods and olive oil (rich in MUFA and low in saturated fat) [66][67][68]. Some life style changes were also recommended based on his genetic makeup, such as not to consume dinner late at night and breakfast early in the morning and to avoid early rising (this individual was homozygous to MTNR1B rs10830963 risk allele G which is in linkage disequilibrium with rs1387153 in Greek people) [69][70][71]. Two months later he had significantly lower plasma insulin concentrations and higher insulin sensitivity.
The emerging prospects of precision nutrition call for re-evaluation of dietary therapies for T2DM taking into account genetic information that could enable stratification of patients towards specific macronutrient intakes. Intervention and observational studies should involve collection of extensive dietary exposure data and, importantly, validation in independent populations to provide solid evidence for interactions between genotype, diet and disease [72].
Author Contributions: Conceptualization, K.G. and A.G.E.; methodology and data curation, K.G., E.L. and E.S.; writing-original draft preparation, K.G. and A.G.E.; writing-review and editing, all authors. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest with the exception of K.G. who is founder and CEO of Embiodiagnostics S.A.