Personalized Therapy by Phenotype and Genotype

Upon the publication of the results of the 20-year-long UK Prospective Diabetes Study (UKPDS) (1), the general recommendation was to achieve, in as many as possible of individuals with type 2 diabetes (T2DM), a target HbA_1c value ≤7%, i.e., the average value associated with significant reduction in the risk of micro- and macrovascular complications. This “universal” glycemic target, however, has been challenged by the results of subsequent intervention trials (2–4). Given the critical analysis of the results of these trials, the American Diabetes Association (ADA) and the American Heart Association suggested that an individualized HbA_1c target should be identified based on the duration of diabetes, life expectancy, presence and severity of diabetes complications, and propensity for hypoglycemia (5). This suggestion has been endorsed by several international guidelines. They also suggest a selection of glucose-lowering strategies to be individualized in order to ensure the most appropriate risk-to-benefit ratio for all patients. According to the ADA/European Association for the Study of Diabetes position statement for T2DM management (6), this can be achieved through balanced assessment of efficacy, risk of hypoglycemia, effect on body weight, and costs of available glucose-lowering agents and their combination.

Although the concept of treatment individualization is appealing, how to pursue it in the clinical setting remains highly debated. To this purpose, careful phenotypic description of the person with T2DM may be appropriate. A few years ago, the ABCD approach was proposed to help physicians in individualizing glycemic target and treatment selection (7). “ABCD” stands for four phenotypic traits that are easy to collect and to factor in: age, body weight, complications and comorbidities, and duration of diabetes.

Age is the strongest risk factor for life expectancy and mortality. The number of elderly people with diabetes is growing, and >70% of the elderly with T2DM have some degree of impaired physical ability. Treatment of diabetes in these subjects can be challenging, as it may expose to the risk of accelerating the development of a geriatric syndrome. Treatment should aim at keeping the patient free of acute metabolic complications, preventing development and/or progression of diabetes complications, preserving cognitive function and physical ability, ensuring adequate quality of life, and avoiding drug reactions. The selection process of glucose-lowering agents should, therefore, take these recommendations into full consideration.

A critical appraisal of glucose-lowering therapy has recently been published (8). Though age per se is not a contraindication to metformin, its use can be limited by gastrointestinal side effects, which may be undesirable in older, frail patients. Metformin can lower vitamin B₁₂ levels, which might accelerate cognitive dysfunction. The main limitation for the use of the drug, however, remains impaired renal function and risk of lactic acidosis.

Pioglitazone is not contraindicated in people with reduced glomerular filtration rate (GFR), but its use is limited by increased risk of bone fractures, fluid retention, and heart failure.

The main concern with respect to sulfonylureas and glinides remains the increased risk of hypoglycemia. We have reported that 40% of patients accessing the emergency room because of hypoglycemia were on antidiabetes agents with sulfonylureas/glinides ± insulin sensitizers being used in 92% of the cases and that the average age was 79 ± 11 years (9). Glibenclamide was the most frequently used sulfonylurea (69%). Moreover, among patients taking sulfonylureas, 47% had an estimated GFR <60 mL/min/1.73 m² and 13.5% had <30 mL/min/1.73 m². This observation supports the recommendation that the risk associated with insulin secretagogues should be carefully assessed, particularly when prescribed in vulnerable T2DM patients.

Trials preformed in elderly patients with T2DM (rev. in 10) have confirmed efficacy, body weight neutrality, and low risk of hypoglycemia of dipeptidyl peptidase (DPP)-4 inhibitors. In patients with moderate-to-severe renal impairment, appropriate dosage adjustment, with the exception of linagliptin, which has no renal excretion, provides a good benefit-to-risk ratio with no further loss in renal function or increased rate of hypoglycemia.

Data on glucagon-like peptide 1 (GLP-1) analogs in older patients with T2DM are scantier. These agents have a favorable benefit-to-risk ratio, but their use in the elderly may be limited by concomitant impaired kidney function. The GLP-1 agonists, though with some difference within the class, can cause nausea and vomiting, which may contraindicate their use in elderly patients with lower body weight, weight loss, or erratic feeding.

Efficacy of sodium–glucose cotransporter (SGLT) 2 inhibitors is likely to be self-limited in patients with reduced estimated GFR, as this will reduce tubular load and, therefore, glucose excretion. Though these compounds are generally considered to have a satisfactory efficacy-to-safety ratio, the risk for dehydration and hypotension is higher in the elderly in those using a loop diuretic and in those with impaired kidney function.

Finally, insulin remains an effective form of treatment, although focused trials on insulin therapy in the elderly are needed. The risk of hypoglycemia is the highest with insulin, and it may become particularly challenging in the frail, unsupported elderly patients.

Body weight and adipose tissue distribution are common phenotypic traits and potential markers of insulin resistance and metabolic syndrome. Of 1,598 patients with T2DM referring to our clinic, 50% had a BMI >27 kg/m², 27% a BMI >30 kg/m², and 13% >30 kg/m². Moreover, 69% had a waist circumference >100 cm. In the overall population, it was quite apparent that the greater the BMI, the higher the HbA_1c, triglyceride levels, and systolic and diastolic blood pressure and the lower the HDL cholesterol concentration. Therefore, in individualizing treatment, body weight may provide some guidance (Fig. 1). In class II and III (i.e., BMI 35 to <40 and ≥40 kg/m²) obese patients, reduction of body weight is a major target along with restoration of glycemic control. Hence, treatment associated with weight loss, i.e., GLP-1 receptor agonists and SGLT2 inhibitors, particularly if used in combination with metformin, may be a preferred choice. On the opposite side of BMI distribution, the few leaner patients with T2DM, in whom a major defect of β-cell function can be hypothesized, may be better candidates for earlier insulin therapy. Finally, subjects with diabetes with intermediate obesity (BMI 30 to <35 kg/m²), in whom the risks associated with obesity are still a matter of debate (11), a main goal should be preventing weight gain or inducing moderate weight loss. This suggests avoiding drugs that may favor weight gain (e.g., sulfonylureas, pioglitazone, and insulin) while favoring weight-neutral agents such as DPP-4 inhibitors or agents that may cause a moderate though sustained weight loss such as SGLT2 inhibitors (12).

The existence of diabetes complications requires careful assessment of pharmacologic glucose-lowering therapy. A typical example is the presence of diabetic nephropathy and loss of kidney function, as a number of agents have to be either reduced in dose or not be used. Out of many therapeutic options, only insulin, pioglitazone, and DPP-4 inhibitors can be used through all stages of renal function. The risk of hypoglycemia is high with insulin, and careful dose adjustment is required. Pioglitazone is not eliminated through the kidney, and therefore it can be used with no limitation except for the risk of bone fracture and fluid retention, conditions for which people with low GFR are already at increased risk (8). DPP-4 inhibitors, with appropriate dosage adjustment with the exception of linagliptin, retain efficacy comparable with that of sulfonylureas with almost no risk of hypoglycemia in patients with moderate-to-severe renal failure (13). Similar results have been reported in patients with end-stage renal disease receiving dialysis (14).

Along with safety considerations, the physician may also consider added values a pharmacologic agent may have with respect to prevention of complications. For instance, in a number of trials, the use of DPP-4 inhibitors was found to be associated with reduction of albuminuria (rev. in 15). These observations find ground on the identification of DPP-4 in the kidney where it may be involved in neuropeptide Y and peptide YY metabolism, cell proliferation, and collagen production by preglomerular and mesangial cells (16). GLP-1 receptors also are expressed in the kidney where they may exert natriuretic and diuretic properties (16).

Even more interest exists with respect to cardiovascular (CV) risk, with metformin still believed to exert some CV protection. The PROspective pioglitAzone Clinical Trial In macroVascular Events (PROactive) provides the strongest evidence supporting beneficial effects of pioglitazone on CV outcomes (17). Insulin has been deemed free of risk as far as it is used in high-risk individuals with early-stage T2DM (18), though better focused trials may still be needed in light of recent analysis (19).

Great expectation was raised by the CV outcomes trials based on DPP-4 inhibitors and GLP-1 receptor agonists because of their efficacy and concomitant effects on the CV system (20). Preclinical studies with GLP-1 and its analogs have shown reduced infarct area upon ischemia (20). DPP-4 inhibition could provide CV protection not only by inhibiting GLP-1 degradation but also by blocking the degradation of chemokines and stromal cell–derived factor-1α with increased endothelial progenitor cells (15). The results of SAVOR-TIMI (Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus–Thrombolysis in Myocardial Infarction) (21), Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care (EXAMINE) (22), and TECOS (Trial Evaluating Cardiovascular Outcomes With Sitagliptin) (23), assessing, respectively, the effect of saxagliptin, alogliptin, and sitagliptin, and those of ELIXA (24), where lixisenatide was tested, have provided evidence for their safe use in T2DM patients with high CV risk. No reduction in CV events or mortality was detected, though it must be kept in mind that these were noninferiority studies, as required by regulatory agencies for all new diabetes drugs.

The CV outcome trials so far completed support the safe use of the incretin-based therapies in T2DM patients with high CV risk but leave unanswered the question of whether their use earlier in the course of the disease may also offer protection. Finally, the increased risk of hospitalization for heart failure observed in SAVOR-TIMI (21), but not in EXAMINE (22) and TECOS (23), will require further evaluation to establish whether this is a drug-specific effect or if it may be accounted for by differences in trial design or conduct.

SGLT2 inhibitors have the potential of reducing CV risk given their glucose-lowering efficacy associated with sustained moderate weight loss and blood pressure reduction and lowering of uric acid levels (11). In support of potential CV protection, the results of the first CV outcomes trial comparing empagliflozin with placebo on top of standard of care have shown a strong risk reduction of CV mortality and mortality for all causes in people with high-risk T2DM (25). In patients with T2DM with prior CV events, treatment with empagliflozin added on top of standard-of-care therapy was associated with a 16% risk reduction (95% CI 0.79–0.99, P < 0.04) for the composite end point of CV mortality, nonfatal myocardial infarction, and nonfatal stroke. Moreover, a significant risk reduction was documented for death from CV causes (−38%), hospitalization for heart failure (−35%), and death from any cause (−32%). Whether these results may apply to the general population with diabetes (i.e., individuals at lower CV risk) remains to be ascertained, while they should be kept in mind in dealing with patients with T2DM who already have experienced a CV event.

To conclude, in the individualization process, the item “complications” is complex, as it requires ascertainment of presence and severity of complications and relative safety of pharmacologic treatments and appreciation of their potential contribution to prevention of complications beyond their glucose-lowering effect.

The longer the duration of diabetes, the greater the chance of complications and the lower the life expectancy. From this point of view, the implications of long-standing diabetes have already been discussed. What remains for comment is the case of short duration and, by default, all cases of newly diagnosed T2DM. The results obtained in such patients in the UKPDS (1) support the recommendation of achieving strict and persistent glycemic control. However, to date, little information on demographic, clinical, and other factors that may influence response to medications is available, making it difficult to decide which drug to select. Metformin is universally accepted as first-line treatment. More uncertain is how to proceed upon metformin failure. Some answers will come from GRADE (Glycemia Reduction Approaches in Diabetes: A Comparative Effectiveness Study) comparing a sulfonylurea, DPP-4 inhibitor, GLP-1 receptor agonist, and basal insulin in patients with recently diagnosed T2DM treated with metformin to examine their effectiveness in maintaining the glycemic goal (HbA_1c <7%) over the time (26). Other outcomes will include relative effects on selected microvascular complications and CV risk factors, patient-centered outcomes, and cost-effectiveness. GRADE will also evaluate the phenotypic characteristics associated with success, failure, and adverse events of different drug combinations (26).

In summary, the ABCD approach can identify common phenotypes to be used in selecting pharmacologic treatment. Nonetheless, they may not be sufficient unless the next letter of the alphabet is added. In this case, E stands for “etiology,” i.e., the identification of the main pathophysiologic mechanisms responsible for the progression of the disease in a given individual.

A better understanding of the pathophysiologic mechanisms may help in defining the best strategy for managing T2DM. In the most recent years, along with insulin resistance and impaired β-cell function, a number of other mechanisms have been identified (27) accounting for an unexpected complexity in the etiology of the disease. Most of these mechanisms are now targeted for pharmacologic intervention with more to come as illustrated in Fig. 2. It is because of such complexity that the use of a single drug is unlikely to provide a sustained effect, while addressing more than one mechanism at diagnosis may be expected to result in a more sustained efficacy, particularly if rational combination is pursued.

Metformin is an insulin sensitizer with a prevalent effect in restraining hepatic glucose production. Metformin also increases GLP-1 gene expression in L cells and enhances β-cell response to GLP-1 (28). Therefore, metformin is a “natural” companion of DPP-4 inhibitors, which prevent degradation of endogenous GLP-1, improve β-cell sensitivity, and increase intraislet GLP-1 availability. A number of trials have proven efficacy and low rate of side effects of this combination in drug-naïve patients, though whether this translates into a durable effect remains to be verified. The VERIFY (Vildagliptin Efficacy in combination with metfoRmIn For earlY treatment of type 2 diabetes mellitus) trial is a 5-year multinational, double-blind, parallel-group study, investigating the long-term effects of initial combination of metformin plus vildagliptin in recently diagnosed patients with mild hyperglycemia, designed to assess durability of glycemic control (HbA_1c), changes in β-cell function and insulin sensitivity, and time to insulin initiation (29).

Glitazones have better durability than sulfonylureas and metformin (30), and preclinical studies have suggested a potential protection of β-cell function. Initial combination therapy with DPP-4 inhibitors and pioglitazone has led to improvement in several markers of β-cell function, suggesting that, if sustained, this combination may be beneficial for preservation of β-cell function in T2DM.

Early use of SGLT2 inhibitors can relieve glucose toxicity and, in turn, improve β-cell function and insulin sensitivity. However, their use is associated with concomitant increase in glucagon levels and hepatic glucose production (31,32), which may restrain some therapeutic effect. Concomitant use of metformin can prevent the increase in hepatic glucose production (33), while a DPP-4 inhibitor can prevent the increase in glucagon levels (34). The combination of the two inhibitors has shown a greater, though not additive, effect compared with the use of the single components in reducing HbA_1c, with almost twice as many patients reaching the HbA_1c target <7% and no increase in side effects (35).

In spite of potential rational combination of different glucose-lowering agents, the choice of the most appropriate combination remains challenging. Some information will come from GRADE (26), though the study will not include recent drugs such as SGLT2 inhibitors. DeFronzo and colleagues (36) have proposed a different approach. Since development and progression of hyperglycemia require the concomitance of hepatic and peripheral insulin resistance and impaired insulin secretion, a triple therapy with metformin, pioglitazone, and exenatide was initiated in subjects newly diagnosed with T2DM. Over the 2-year follow-up, the combination therapy, compared with classic sequential therapy with metformin, sulfonylurea, and insulin, was more effective (HbA_1c 5.95 vs. 6.50%; P < 0.001) and had a 7.5-fold lower rate of hypoglycemia and a mean weight loss of 1.2 kg vs. a mean weight gain of 4.1 kg (P < 0.01) with sequential add-on therapy (36). This approach, however, assumes that all patients share the same pathogenetic profile. Though reasonable, this may not be universally acceptable, suggesting the need for individual pathogenetic profiling. Direct assessment of pathogenetic mechanism is theoretically possible but unrealistic in the clinical setting. Recently, biomarkers have been identified that are associated with insulin sensitivity and risk of developing diabetes (37) while there is still an unmet need with respect to biomarkers for β-cell function and mass and other pathogenetic components.

In summary, the patient’s phenotype can be complex, and though some phenotype traits can be easily accessible, their interpretation may still rely on the physician’s personal knowledge and experience. Development of dedicated algorithm might help (38), but validation may require extensive work. We have recently proposed an algorithm (39) based on the result of a survey performed presenting a number of case vignettes to 244 top diabetologists from all world regions. The algorithm was then used to define individualized glycemic targets and validated by presenting three new cases to 57 leading diabetologists who suggested glycemic targets that were similar to those calculated by the algorithm. We concluded that the suggested algorithm could be an additional decision-making tool offered to the clinician to supplement clinical decision making in considering a glycemic target for the individual patient with diabetes. Available data on individual responses to drugs remain insufficient to be built into an efficient algorithm, and the phenotype description may remain too broad to provide sufficient information on the underlying pathogenic factors. Moreover, other factors such as early response to glucose-lowering treatment may be needed to considered as well, as recently suggested by Fu et al. (40). More recently, the development of more accessible genotyping techniques and the discovery of a number of genetic variants associated with the risk of T2DM have raised the expectation for a more precise personalized treatment. Whether combining phenotyping and genotyping may offer better opportunities, to the best our knowledge, has not been fully investigated. In the few studies where this was attempted (41–43) there was only a small improvement in the ability to predict future development of T2DM.

T2DM recognizes a complex pathogenesis, and each mechanism does recognize a genetic background. The great expectation is that, through genotyping, precision medicine could be generated. Pharmacogenetics is still young, with very few studies, most of which are of small sample size and different design. Nevertheless, despite current limitations, there are examples for a guided treatment choice for diabetes (Table 1).

MODY (maturity-onset diabetes of the young) is a monogenic form of diabetes with autosomal dominant inheritance. MODY is characterized by defects in insulin secretion, and there are at least six, and most likely more, different MODY types. MODY offers a good example for pharmacogenetic recommendations. MODY 3, the most prevalent MODY type (∼61%), is due to mutations in the HNF1A (hepatic nuclear factor 1-α) gene and is often misdiagnosed as type 1 diabetes requiring insulin treatment. However, these patients are very sensitive to sulfonylureas, requiring one-quarter or less of the dose usually prescribed in patients with T2DM (44). This hypersensitivity is due to decreased HNF1A expression causing reduced liver uptake of sulfonylurea and elevation in its circulating levels. MODY 2 (∼15%) is caused by mutations in the GCK (glucokinase) gene and is characterized by mild hyperglycemia usually not requiring treatment. The mutation leads to reduced activity of pancreatic glucokinase resulting in increased glucose sensing threshold (between 6 and 8 mmol/L) for stimulation of insulin secretion. Diagnosis of MODY 2 has important implication in the treatment of women with gestational diabetes mellitus. If the fetus carries the same mutation, treatment of the mother with insulin can lower glucose to a level not sufficient to sustain fetal insulin secretion resulting in intrauterine growth retardation (45).

Another example of genetics providing guidance for treatment is neonatal diabetes (NDM) caused by activating mutations in the KCNJ11 and ABCC8 genes. These genes encode for two subunits of the K_ATP channels of sulfonylurea receptor in the β-cells causing inability to close the K_ATP channels and defective insulin secretion. Patients carrying the KCNJ11/ABCC8 mutations are poorly regulated by insulin but respond very well to high doses of sulfonylureas (46). Children with NDM often present with DEND syndrome (developmental delay, epilepsy, and neonatal diabetes). Since the KCNJ11 gene is also expressed in the brain, switching the children to high doses of sulfonylureas can also positively affect developmental defects (46).

The advent of genotyping and sequencing technologies has contributed to discovery of many genetic variants contributing to T2DM pathogenetic complexity. These advances open attractive opportunity to determine the utility of genomic regions to predict treatment response. However, interpretation of pharmacogenetic findings is not always clear. Whereas most studies use improvement in HbA_1c levels as the main outcome, others define responders in terms of percentage of HbA_1c decline or reduction of fasting and postprandial glucose and, in some cases, changes in insulin secretion and sensitivity. These diversities might contribute to inconsistency and heterogeneity between findings. In the following section, pharmacogenomics studies with different glucose-lowering agents will be briefly examined.

A number of genetic variants have been associated with different responses to metformin. The drug is an organic cation, which is not metabolized in the body and requires transporters for absorption, distribution, metabolism, and elimination. Variants in the organic cation transporter, OCT1 (SLC22A1), have been associated in some (47), but not all (48), studies with lower metformin efficacy and increased renal clearance. Similarly, better efficacy and reduced risk of developing diabetes have been associated with genetic variants of MATE1 and MATE2 (multidrug extrusion proteins 1 and 2) (49). Finally, by genome-wide association, variants in the ATM (ataxia teleangiectasia mutation) have been associated with significant reduction in HbA_1c levels after metformin treatment (50).

Pharmacogenetic studies of sulfonylureas have been focused on genetic variants affecting pancreatic β-cell K_ATP channels and drug-metabolizing enzyme CYP2C9. Variants in the KCNJ11 and ABCC8 genes coding for K_ATP channel subunits have been found to be associated with T2DM risk and reduced response to sulfonylurea (51). Additionally, two variants in the CYP2C9 gene (CYP2C9*2 and CYP2C9*3) were found to be associated with increased drug levels and increased risk of hypoglycemia after treatment with sulfonylurea (52), fitting with a better HbA_1c response in the Genetics of Diabetes Audit Research in Tayside Scotland (GoDARTS) study (53). Finally, in GoDARTS, individuals carrying the TCF7L2 variant have been reported to have a lesser response to sulfonylurea compared with that in noncarriers (54).

Thiazolidinediones (TZDs) are attractive candidates for pharmacogenetic studies, as variants in the drug target PPARG gene are associated with T2DM risk and insulin sensitivity in populations without diabetes. Yet, pharmacogenetic studies in subjects with prediabetes and subjects with diabetes yielded inconsistent results as far response to TZD therapy is concerned (55,56). Genetic variants of the CYP2C8*3 enzyme have been associated with reduced TZD activity (57). Finally, genetic variants in the aquaporin 2 (AQP2) and sodium/potassium/chloride transporter (SLC12A1) genes have been associated with increased risk of edema in rosiglitazone-treated patients (58), offering opportunity for pharmacogenomics not only in terms of therapeutic response but also in identifying underlying mechanisms in subjects at risk for side effects.

Variants in the chymotrypsinogen gene (CTRB1/2) have been linked to impaired GLP-1–stimulated insulin secretion in 232 individuals undergoing hyperinsulinemic clamp. The variants also were associated with increased chymotrypsinogen activity and gene expression in human pancreatic islets (59). These findings found support in a GoDARTS study reporting greater HbA_1c reduction in carriers of the CTRB1/2 variant treated with gliptins (54). Many of the genetic variants (TCF7L2, WFS1, and KCNQ1) increasing susceptibility to T2DM are associated with impaired incretin action, whereas variants in the glucose-dependent insulinotropic peptide (GIP) receptor gene (GIPR) were associated with reduced incretin secretion (60). Recently, some differences in response to linagliptin have been reported in patients with diabetes with and without the TCF7L2 gene variants (61).

Metiglinides (nateglinide and repaglinide) close the K_ATP channels to stimulate insulin secretion through interaction with subunits other than those activated by sulfonylureas. Effects of genetic variants in transporters (SLCO1B1) and metabolizing enzymes (CYP2C9) have been associated with decreased rate of glinide metabolism, though no effect was observed in treatment response (62). Interestingly, a variant in the imprinted KCNQ1 gene was associated with better glucose and insulin sensitivity responses after 48-week repaglinide therapy but not when these effects were corrected for age, sex, and BMI (63). KCNQ1 has previously been implicated in T2DM pathogenesis, displaying parent-of-origin effects, suggesting the importance of taking into account whether alleles are inherited from the mother or from the father.

Pharmacogenetics possesses the potential for the discovery of novel drug targets. For instance, a recent study from the Botnia region in Finland comparing individuals with early-onset T2DM and subjects remaining diabetes free despite clustering of T2DM risk factors has led to the identification of a rare protective variant in the SLC30A8 gene (R138X) (64). Follow-up studies will shed light on underlying biology of the protective effect and hopefully facilitate development of drugs based on these mechanisms.

In addition to biologic factors, response to drug therapy also involves gene-environment interactions including dietary patterns and physical activity as well as psychosocial factors such as motivation, education, family support, and treatment adherence. The decline in cognitive function in T2DM patients may play a role in treatment adherence. Genomic regions associated with cognitive function have been shown to have positive evolutionary selection, but this phenomenon was not observed for the regions associated with risk factors for late-onset metabolic diseases (65). Therefore, it will be interesting to examine to what extent response to therapy and poor compliance are determined by the patient’s cognitive function. Furthermore, individual genetic susceptibility could be modified by epigenetic mechanisms involved in different tissues and organs, including several brain areas. Finally, multiethnic studies are warranted to study generalization of treatment responses in different populations. In summary, there is accumulating evidence that genetics might become an important tool for diabetes diagnosis, classification, and treatment, as it is the case for monogenic forms of diabetes. Similarly, pharmacogenetics has been successfully used to guide choice of therapy in monogenetic forms of diabetes, while so far having limited clinical utility in the polygenic T2DM phenotypes. However, genetic effects can be modified by different environmental factors with subsequent epigenetic modifications. Therefore, more and larger studies are needed to provide better classification of heterogeneous T2DM into different subgroups to appreciate the potential of pharmacogenomics for personalized treatment.

Precision medicine is not just an academic discussion, as it already belongs to the political agenda of the most industrialized countries. President Barack Obama, in his State of the Union Address in January 2015, announced the launch of a “Precision Medicine Initiative to bring us closer to curing diseases like cancer and diabetes.” This scenario has been clearly described in its challenges, expectations, research, and scientific approaches in a clear editorial by Collins and Varmus (66) where cancer was identified as the “near-term focus,” whereas a “whole range of diseases” will be targeted in a longer term. For people with diabetes, the promise of personalized therapy has the main objectives of providing effective, sustained, and safe treatment. On the basis of the patients’ individual—phenotypic and genetic—profile, precision medicine aims at predicting which patient is more likely to benefit and which one is more likely to experience side effects in response to therapeutic modalities. Though a number of phenotypic traits and genetic variants have been identified, their interpretation and translation into clinical guidance remain uncertain and insufficiently investigated and thus, of limited, if any, clinical use. More data and more clinical trials are needed. To this purpose, it sounds desirable that data generated in clinical trials for new drug classes could be pooled in order to gain sufficient power for identifying the main predictors of therapeutic success and risk of adverse events. Such an approach may require the industry to move out of the expectation of the “blockbuster drug,” that is, a drug that can be used in the vast majority of individuals with T2DM. On the contrary, it would be of great value to understand which drug(s) better suits a given individual in a given phase of disease. Collection of this information may not be limited to randomized clinical trials. Rather, postmarketing large-scale surveillance, electronic health registries, and even mobile health technology could be used to gain stronger information over much longer follow-up in larger and more diversified populations with diabetes. Currently, up to 13 classes of drugs are available for the treatment of T2DM targeting different pathogenetic mechanisms. The challenge is how to use the right drug (and the right combination of drugs) in the right patient. While we wait for a more scientific and less pragmatic guidance, a few things should be kept in mind. We must appreciate that T2DM is a complex disease with complex pathogenesis and, therefore, is not easy to treat, particularly in the pursuit of long-term glycemic control. From this point of view, the multiplicity of available pharmacologic agents should be welcome, though appropriate use requires expert knowledge of advantages and disadvantages of each available drug and the potential for rational combination. This complexity may sound at odds with the need for effective treatment in a growing number of individuals with T2DM worldwide and the need to develop treatment strategies that can be implemented in primary health care. Yet, inappropriate use of drugs may not provide the necessary response and, particularly in the case of novel agents, may result in unjustified expense excess. It is, therefore, the duty of the diabetes community to search for a new strategic model of interaction between the specialist and the other health care providers ensuring appropriate management of personalized treatment for the person with diabetes.

Funding. This work was in part supported by Progetti di ricerca di Rilevante Interesse Nazionale (PRIN) grant 2010 YK7Z5K_006.

Duality of Interest. V.L. is employed by Steno Diabetes Center A/S, a research hospital working in the Danish National Health Service and owned by Novo Nordisk A/S. C.B. has received honoraria for advisory work from Eli Lilly. S.D.P. has received honoraria for advisory work and lectures from AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, GlaxoSmithKline, Hanmi Pharmaceuticals, Intarcia, Janssen, Merck Sharp & Dohme, Novartis, Novo Nordisk, Sanofi, and Takeda as well as research support from Merck Sharp & Dohme, Novartis, and Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.