Emerging Applications of Metabolomic and Genomic Profiling in Diabetic Clinical Medicine

Reviews/Commentaries/ADA Statements REVIEW

Emerging Applications of Metabolomic and Genomic Proﬁling in Diabetic Clinical Medicine

AINE M. MCKILLOP, PHD however, powerful methodologies are now PETER R. FLATT, PHD available for exploitation of novel quantitative and qualitative disease-related biomarkers. Novel biomarkers are needed that are Clinical and epidemiological metabolomics provides a unique opportunity to look at genotype- independent of known clinical risk factors. phenotype relationships as well as the body’s responses to environmental and lifestyle factors. fl Fundamentally, metabolomics aims to Fundamentally, it provides information on the universal outcome of in uencing factors on monitor changes in products of metabo- disease states and has great potential in the early diagnosis, therapy monitoring, and understand- lism and provide valuable information on ing of the pathogenesis of disease. Diseases, such as diabetes, with a complex set of interactions fl between genetic and environmental factors, produce changes in the body’s biochemical profile, arangeofin uencing factors and gene- thereby providing potential markers for diagnosis and initiation of therapies. There is clearly related outcomes. Exploitation of genomic a need to discover new ways to aid diagnosis and assessment of glycemic status to help reduce technology in recent times has resulted in diabetes complications and improve the quality of life. Many factors, including peptides, proteins, many technical advances, and genomic metabolites, nucleic acids, and polymorphisms, have been proposed as putative biomarkers for analysis has now emerged as a valuable diabetes. Metabolomics is an approach used to identify and assess metabolic characteristics, changes, tool in predicting the body’s response to and phenotypes in response to influencing factors, such as environment, diet, lifestyle, and path- fi stimuli caused by disease or injury. Indeed, ophysiological states. The speci city and sensitivity using metabolomics to identify biomarkers of methodologies such as epigenetic profil- disease have become increasingly feasible because of advances in analytical and information tech- ing, sequencing technologies, microarrays, nologies. Likewise, the emergence of high-throughput genotyping technologies and genome-wide fi association studies has prompted the search for genetic markers of diabetes predisposition or sus- functional ngerprinting, and analysis of ceptibility. In this review, we consider the application of key metabolomic and genomic method- genomic alternations are all well-established ologies in diabetes and summarize the established, new, and emerging metabolomic and genomic methodologies in practice. Complementing biomarkers for the disease. We conclude by summarizing future insights into the search for im- these technologies with computational proved biomarkers for diabetes research and human diagnostics. methods/bioinformatics that integrate large – amounts of heterogeneous genetic and Diabetes Care 34:2624 2630, 2011 genomic information has helped provide meaningful results to aid our understanding iabetes is a rapidly increasing meta- accompanied by the characterization and of the complex changes of genes and macro- Dbolic disorder precipitated by complex quantification of small molecules. It can often molecules. There is now a clear need to dis- and poorly understood interactions be confused with the term metabonomics, cover novel and effective clinical biomarkers between multiple environmental and genetic which represents the global, dynamic met- using technologies that encompass an array factors. The consequences of diabetes are far abolic response of living systems to biolog- of different methodologies. Chromatogra- reaching, and disturbances in both the se- ical stimuli or genetic manipulation. Both phy, two-dimensional electrophoresis, cretion and action of insulin impact on the terms are closely affiliated with each other mass spectrometry, functional magnetic global regulation of metabolism, affecting the owing to the analytical and experimental resonance, positron emission tomography, composition of blood and other body fluids. technologies used in each field. The obser- and protein/gene sequencing are some Understanding of this process and identifi- vation of the characteristics and changes in examples being used to unravel the body’s cation of potential disease biomarkers have metabolism by metabolomics allow the re- complex biological systems. Sensitive and been greatly facilitated in recent years by the sulting data to be merged with data from the high-resolution techniques used in clinical upsurge in new technologies for comprehen- other “-omic” technologies. Genomic, meta- metabolomics, such as nuclear magnetic sive metabolic profiling, which are often bolomic, and proteomic state-of-the-art resonance, gas chromatography–mass collectively termed metabolomics. technologies are now used increasingly by spectrometry, and liquid chromatography– researchers to identify clinical methodolo- mass spectrometry, are sensitive and robust fi Metabolomic pro ling in clinical gies for the early diagnosis and monitoring and have the capacity to process large vol- medicine of human degenerative diseases such as di- umes of data from population studies (1,2). fi Metabolomics is de ned as the analyti- abetes. Classical risk factors still have an im- However, overinterpretation of data remains cal description of biological samples portant role to play in diabetes assessment; one of the key limitations to be overcome ccccccccccccccccccccccccccccccccccccccccccccccccc for successful exploitation of metabolomics From the SAAD Centre for Pharmacy and Diabetes, School of Biomedical Sciences, University of Ulster, and metabonomics. Coleraine, Northern Ireland, U.K. In this brief review, we consider recent Corresponding author: Aine M. McKillop, [email protected]. applications of metabolomic and related Received 3 May 2011 and accepted 6 September 2011. technologies in diabetes together with their DOI: 10.2337/dc11-0837 © 2011 by the American Diabetes Association. Readers may use this article as long as the work is properly use in relation to clinical diagnostics. Tech- cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ nical details of the methodologies involved licenses/by-nc-nd/3.0/ for details. and their use in basic diabetes research have

2624 DIABETES CARE, VOLUME 34, DECEMBER 2011 care.diabetesjournals.org McKillop and Flatt

Table 1dEstablished, new, and emerging metabolomic biomarkers for type 2 diabetes discover new markers, as illustrated in ges- tational diabetes mellitus, where there is a Predictor Abbreviation Reference no. drive to reconsider diagnostic criteria recog- nizing the possibility of adverse pregnancy Metabolic markers outcomes of milder levels of glucose intol- Insulin 7 and 8 erance than hitherto appreciated. Glucose 7 and 8 Metabolic markers of type 2 diabetes. g-Glutamyl transferase GGT 4 and 8 Whereas glucose and insulin are the most Alanine aminotransferase ALT 5 and 6 well-established biomarkers, there are many Ferritin FTH1 7 and 8 new and emerging biomarkers of diabetes Pancreatic polypeptide PP 9 (Table 1). Positive associations have been Fibronectin 10 reported between serum g-glutamyl trans- Fetuin A 11 ferase andincident type 2 diabetes (4). Path- Sex hormone–binding globulin SHBG 7 and 12 ophysiological mechanisms underlying Free testosterone 12 how serum g-glutamyl transferase relates Insulin-like growth factor I IGF-I 13 to type 2 diabetes risk have not been elu- Insulin receptor 8 cidated, but insulin resistance, oxidative Creatine kinase-MB CKMB 8 stress, and chronic low-grade systemic in- MR-Pro atrial natriuretic peptide MR_PRO_ANP 8 flammation may be involved. Alanine ami- NT-Pro B-type natriuretic peptide NT_PRO_BNP 8 notransferase (ALT) is elevated in some B-type natriuretic peptide BNP 8 patients with type 2 diabetes indepen- Biomarkers of glycemia dently of confounding factors such as Glycated hemoglobin HbA1c obesity (5,6), and evidence now indicates Fructosamine 14 that markers associated with fatty liver 1,5-Anhydroglucitol 1,5AG 15 may predict future development of type 2 Glycated albumin 16 diabetes. In most cases of nonalcoholic fatty Glycated insulin 17 liver disease, the hepatic component of Glycosylated amylin 18 metabolic syndrome, ALT is elevated, and Glycated LDL 19 studies have associated raised ALT with Markers of oxidative stress and nutrient status metabolic syndrome and type 2 diabetes (5). Glutathione GSH 20 A strong association has been found Advanced glycated end products receptor RAGE 24 between raised ferritin levels (below the Ascorbic acid Vitamin C 21 range indicative of clinical hemochroma- 25-Hydroxyvitamin D Vitamin D 22 tosis) and development of incident di- Homocysteine 8 abetes (7,8). Ferritin was associated with Branched-chain and aromatic amino acids Leu, Ile, Val, Tyr, Phe 23 diabetes independently of established risk Lipid-related markers factors (age, BMI, sex, family history, Leptin LEP 6 and 25 physical inactivity, and smoking), as Adiponectin ADIPOQ 6, 8, and 25 well as dietary factors and alcohol intake. Apolipoprotein B ApoB 8 The mechanism is thought to involve in- Apolipoprotein A ApoA 8 sulin resistance, free radical damage, and Endothelial and inflammatory markers accumulation of iron in hepatocytes. C-reactive protein CRP 6–8and26 Pancreatic polypeptide, believed to act Interleukin-18 IL-18 8 and 27 as a regulator of pancreatic and gastroin- Interleukin-1 receptor antagonist IL-1ra 28 testinal functions, has been proposed as a Interleukin-2 receptor antagonist IL-2ra 7 possible marker of b-cell failure in diabetes Interleukin-6 IL-6 6–8 (9). Likewise, fibronectin levels change in Plasminogenactivatorinhibitor-1 PAI-1 6and29 insulin resistance, and Amrein et al. (10) Cell adhesion molecule CAM 30 proposed that levels could be used in the Tissue plasminogen activator antigen t-PA antigen, PLAT 31 diagnosis of insulin resistance and monitor- Neopterin 8 ing of disease progression. Fetuin-A, a he- Von Willebrand factor vWF 31 patic secretory protein that binds the insulin receptor and inhibits insulin action, has been shown to be associated with been covered in several excellent articles the global comparison of HbA1c values is incident diabetes independent of other and reviews (1,3). now possible as a result of the International markers of insulin resistance (11). Sex Federation of Clinical Chemistry and Lab- hormone–binding globulin (SHBG) is Metabolomics applied to the clinical oratory Medicine establishing true interna- known to be downregulated by insulin, fl diagnosis and prognosis of diabetes tional reference methods for HbA1c (in and low levels have been reported to re ect The American Diabetes Association offi- millimoles per mole) and the successful insulin resistance and incident diabetes in cially recommends HbA1c testing for the preparation of pure HbA1c calibration ma- women (7,12). Population studies have diagnosis and monitoring of diabetes, and terial. However, there is clearly a need to shown that low testosterone levels are care.diabetesjournals.org DIABETES CARE, VOLUME 34, DECEMBER 2011 2625 Metabolic and genomic profiling in diabetes commonly associated with the prediction of monoclonal antibodies against glycated risk of type 2 diabetes even after adjustment type 2 diabetes and the metabolic syn- LDL for monitoring glycemic control (19). for BMI, lifestyle factors, and cardiovascu- drome. Although the inverse association of Markers of oxidative stress and nutrient lar disease (6,25). It is well recognized that testosterone with diabetes is partially medi- status in type 2 diabetes. Since diabetes is adiponectin increases insulin sensitivity, ated by SHBG, low testosterone is linked to associated with overproduction of different regulates glucose and lipid metabolism, diabetes via a bidirectional relationship with reactive oxygen species leading to long-term and enhances insulin action in the liver. visceral fat, muscle, and possibly bone (12). development of diabetes complications, a Serum levels of adiponectin have been IGF-I,whichisinvolvedinsomaticgrowth, number of candidate biomarkers have shown to decrease with increasing obesity, cellular differentiation, and regulation of emerged (Table 1). Reduced levels of anti- and interestingly, elevated adiponectin has metabolism, is potentially another marker oxidants such as glutathione, vitamin C, been associated with a lower incidence of of diabetes, given its insulin-like effects and vitamin E (20–22) and changes in se- diabetes (6,25). Leptin is already a marker and involvement in glucose homeostasis. rum malondialdehyde and activities of su- of percentage fat mass in healthy individu- Large-scale gene association and prospec- peroxide dismutase and glutathione als and regulates body weight by effects on tive observational studies are needed to fully peroxidize have been found in diabetic pa- food intake and metabolism. The associa- elucidate the involvement of IGF-I (13). tients as well as changes in other oxidative tion between leptin and incident diabetes In a recent study of 31 novel biomark- stress biomarkers such as catalase, glutathi- has been difficult to determine but may also ers, Salomaa et al. (8) demonstrated an asso- one reductase, lipid peroxidation, and nitrite reflect insulin resistance. Adiponectin is ciation between clinically incident diabetes concentration (20). more strongly associated with type 2 dia- and insulin receptor, creatine kinase-MB, Ascorbic acid (vitamin C) (21) and 25- betes risk than leptin (25). Apolipoprotein MR-Pro atrial natriuretic peptide, NT-Pro hydroxyvitamin D (vitamin D) (22) are (Apo)B, and to a lesser extent ApoA, was a B-type natriuretic peptide, and B-type na- both associated with diabetes risk, but be- particularly strong predictor of diabetes triuretic peptide. The utility of these as po- cause of the many confounding determi- even when controlling for BMI and waist- tential biomarkers and the nature of their nants, levels need further investigation. to-hip ratio (8). links to diabetes clearly deserve further Homocysteine also has promising links to Endothelial and inflammatory markers study. diabetes (8). These and other biomarkers in of type 2 diabetes. C-reactive protein Biomarkers of glycemia in diabetes. Table 1 have yielded promising results, but (CRP) is a predictor of diabetes indepen- HbA1c is the most widely known glycosy- most have been tested one at a time, with dent of other clinical indicators such as lated protein in diabetes, and its assay has lack of independent validations. Many of BMI, fasting triglyceride, and glucose (26), been the gold standard for the evaluation of these apparently “independent” risk factors but circulating levels correlate with lipids, glycemic status for many years. Limitations may in fact be related by virtue of their SHBG, and adiponectin. The value of CRP is of the HbA1c measurement do exist, but it common origins or shared metabolic path- promising, but further evaluation is needed. remains an important tool in the manage- ways (6). There may even be different pat- Elevated levels of the cytokine interleukin ment of diabetes along with self-monitored terns of biomarkers of diabetes associated (IL)-18 are linked with an increased risk of blood glucose profile data. Fructosamine is with early or late-stage diabetes. Recently, type 2 diabetes, independent of a general- another marker used in practice (14) and amino acid profiles were proposed as im- ized proinflammatory state (27). Studies refers to the ketoamine rearrangement portant in assessing diabetes risk as ele- have also reported upregulation of anti- product formed by the interaction of glu- vated levels of five amino acids were inflammatory cytokine IL-1 receptor antag- cose with the ´-amino group on lysine res- shown to predict the development of dia- onist (IL-1ra) in individuals with obesity idues of albumin. The assay is thought to betes at early stages (23). Combinations of and insulin resistance (28). These studies be less accurate than HbA1c because of fac- the five branched-chain and aromatic indicate that individuals with high risk of tors affecting the half-life of its many com- amino acidsdleucine, isoleucine, valine, type 2 diabetes can be characterized by the ponents and is thus considered of less tyrosine, and phenylalaninedrather than presence of an early compensatory, anti- clinical value. Other markers of glycemic a single amino acid, served as a more accu- inflammatory response that precedes the control that have been considered but not rate predictor of diabetes risk (23). development of the disease and inflam- widely used are 1,5-anhydroglucitol (15) In diabetes, advanced glycation end matory markers. Like IL-1ra and CRP, and glycated albumin (16). products form as a consequence of long- IL-2ra is involved in inflammatory path- Much interest has surrounded the role of term hyperglycemia, and a number of ways; however, only one study to date glycosylated regulatory proteins as biomark- truncated forms of the advanced glyca- has identified IL-2ra as a diabetes marker ers. Because insulin glycation is dependent tion end product receptor (RAGE) have (7), which may be due to oxidative stress on the degree and duration of hyperglyce- been identified. The C-terminally trun- in diabetes culminating in T lymphocyte mia, monitoring of glycemic status in di- cated form, named endogenous secretory activation. abetic patients using glycated insulin could RAGE,haspotentialasabiomarkerand IL-6 has been reported to be elevated aid approaches to the diagnosis, manage- in the estimation of the risk of atheroscle- in incident diabetes, independent of obe- ment, and treatment of diabetes (17). Other rotic disorders and occurrence of metabolic sity and fasting glucose (6–8). Studies are peptides such as amylin and amylin-like syndrome (24). needed to determine the relationship be- peptides have been disclosed as potentially Lipid-related markers of type 2 diabetes. tween IL-6 and diabetes and whether useful in the detection and/or evaluation of Adipokines are involved in a broad range there is a causal link. Plasminogen activa- diabetes. Glycosylated amylin (18) has of physiological processes such as insulin tor inhibitor 1 levels reflect an acute phase been proposed as a predictor of the onset sensitivity, lipid metabolism, vascular he- response, and elevated levels are found of diabetes in patients who otherwise mostasis, blood pressure regulation, angio- with incident diabetes independent of show normal glycemic control, and an- genesis, and appetite control. Leptin and obesity and insulin resistance (29). The other group has filed a patent based on adiponectin are associated with increased association between high plasminogen

2626 DIABETES CARE, VOLUME 34, DECEMBER 2011 care.diabetesjournals.org McKillop and Flatt activator inhibitor 1 and incident diabetes Genomic variations and DNA (42), GCK (glucokinase) (42), and GCKR may be due to associations with liver fat profiling of those at risk for (glucokinase receptor) (42); however, fur- (6). Circulating levels of several other in- type 2 diabetes ther investigation of genetic loci on glucose flammatory endothelial-derived factors Despite many candidate gene studies and homeostasis and their impact on type 2 di- such as cell adhesion molecules (30), genome-wide linkage studies, very few sus- abetes is needed. Indeed, a recent study by tissue-plasminogen activator antigen (31), ceptibility loci for type 2 diabetes have been Soranzo et al. (42) using GWA studies iden- neopterin (8), and von Willebrand factor identified until the recent emergence of tified ten genetic loci associated with (31) have been linked to diabetes risk. genomic-wide association (GWA) data and HbA1c. Genetic factors affecting expression, Recent studies such as the MONICA/ large-scale replication studies (Table 2). turnover, and abnormal glycation of hemo- KORA study have shown that when a Meta-analysis of GWA studies provides globin may be associated with changes in risk prediction model of multiple inflam- theuniqueopportunitytoinvestigatethe levels of HbA1c. mation markers is used, the prediction of heterogeneity or consistency of genomic as- Significant effects of many suscepti- incident type 2 diabetes and coronary sociations across diverse datasets and study bility loci are still to be determined and events is significantly improved com- populations. Recently, Voight et al. (32), replicated, and further large-scale associ- pared with cardiometabolic risk factors using large-scale association analyses ation studies will be required. Recently, (25,27). combining the data from eight GWA stud- Schleinitz et al. (43) found some of the ies, identified 12 new susceptibility loci for type 2 diabetes risk alleles or related sub- Increased clinical value of type 2 diabetes. phenotypes to be weak, including those of evaluating panels composed of Despite identification of many puta- JAZF1, CDC123/CAMK1D, NOTCH2, different biomarkers tive causative genetic variants, few have ADAMTS9, THADA, and TSPAN8-LGR5. Advances in technology plus awareness of generated credible susceptibility variants for The TNF/LTA locus has been a long- an increasing number of diabetes-related type 2 diabetes. Indeed, the most important standing type 2 diabetes candidate gene, metabolomic analytes are likely to facili- finding using linkage studies is the discov- whereas a recent study found no evidence tate use of panels of combined biomarkers ery that the alteration of TCF7L2 (TCF-4) of an association between TNF/LTA region rather than reliance on single biomarkers gene expression or function (33) disrupts variation and type 2 diabetes (44). The as- for diabetes. If these are selected from dif- pancreatic islet function and results in en- sociation of polymorphisms in TNFa and ferent tissue origins/pathways, their ability hanced risk of type 2 diabetes. Candidate type 2 diabetes has been extensively repor- to predict diabetes risk is likely to be gene studies have also reported many type ted. Recently, the TNFa variant rs3093662, increased, thereby facilitating earlier inter- 2diabetes–associated loci and the coding linked to higher serum levels of tumor ne- ventions. This view is supported by two variants in the nuclear receptor peroxisome crosis factor-a, was shown to be associated comprehensive studies (7,8). Kolberg et al. proliferator–activated receptor-g (34), the with elevated insulin (45). Mutated tran- (7) evaluated the potential of 58 candidate potassium channel KCNJ11 (34), WFS1 scription factors, hepatocyte nuclear factor diabetes-related biomarkers plus six clini- (35), and HNF1B (TCF2) (36) are among (HNF)1A and HNF4A, have received sub- cal factors for predicting 5-year risk of di- the few that have been replicated (Table 2). stantial attention, and there is evidence for abetes in 160 of 632 individuals from the Recently, there have been great advances susceptibility of the variants to maturity- Danish Inter99 cohort who went on to de- in the analysis of associated variants in onset diabetes of the young (MODY) and velop type 2 diabetes. A six-biomarker GWA and replication studies due to high- type 2 diabetes. Recently, high-sensitivity model (adiponectin, CRP, ferritin, IL-2ra, throughput genotyping technologies, the CRP was shown to discriminate HNF1A- glucose, and insulin) showed improved International HapMap Project, and the MODY from other subtypes of diabetes performance over single markers such as Human Genome Project. Type 2 susceptibility (46). JAZF1 CDC123-CAMK1D HbA1c and fasting glucose, being equiv- loci such as , , Interestingly, many of the established alent to a 2-h oral glucose tolerance test TSPAN8-LGR5, THADA, ADAMTS9, susceptibility loci are involved in insulin (7). Similarly, Salomaa et al. (8) evaluated NOTCH2, and ADCY5 (37,38) are among secretion signaling, supporting an important the potential of 31 novel biomarkers as some of the established loci (Table 2). role for defects in b-cell function and b-cell predictors for clinically incident diabetes CDKN2A/B, CDKAL1, SLC30A8, IGF2BP2, mass in type 2 diabetes. The exciting po- in a combined total of 12,804 individuals HHEX/IDE,andFTO are other established tential of genetic testing for susceptibility of fromtheFINRISK97andHealth2000co- susceptibility loci for diabetes (Table 2) diabetes appears to be some way off, apart horts of whom 596 later developed diabetes (34,39,40). GWA studies have also identi- from rare forms of monogenic diabetes in 10-year follow-up. This study revealed fied the potassium voltage-gated channel (44). Moreover, it is well known that non- that adiponectin, ApoB, CRP, and ferritin KCNQ1 (32) as an associated gene variant genetic factors such as obesity and lifestyle improved diabetes prediction even after tak- for diabetes. A recent GWA study reporting a factors play an important role in the disease. ing BMI, glucose, and other classical risk genetic variant with a strong association with New phenotyping approaches to studying factors into account (8). Sex-specificanaly- insulin resistance, hyperinsulinemia, and metabolite and protein abundance and sis further showed potential value of includ- type 2 diabetes, located adjacent to the in- data integration are needed to bring genomic ing IL-1ra and insulin as biomarkers. These sulin receptor substrate 1 (IRS1) gene, is the and metabolomic goals together. In this data suggest that biomarker scores reflect- C allele of rs2943641 (41). Interestingly, context, the Human Metabolome Project ing different pathological processes may the parental origin of the single nucleotide in Canada (47), aimed at providing a have significant potential for improving fu- polymorphism is of importance because linkage between the human metabolome ture prediction of diabetes. Similarly, eval- the allele that confers risk when paternally and the human genome, has identified and uation of amino acid profiles appears to be inherited is protected when maternally quantified normal concentration ranges more effective in prediction than are single transmitted. GWA studies for glycemic for a large number of metabolites in cere- amino acids (24). traits have identified loci such as MTNR1B brospinal fluid, serum, urine, and other care.diabetesjournals.org DIABETES CARE, VOLUME 34, DECEMBER 2011 2627 Metabolic and genomic profiling in diabetes

Table 2dType 2 diabetes susceptibility loci established through candidate-gene, genome-wide linkage, and GWA studies

Gene/region Gene name Chromosomal location Identification Reference no. TCF7L2 Transcription factor 7-like 2 10q25.3 Linkage study 33 and 39 PPARg Peroxisome proliferator–activated receptor g 3q25 Candidate gene 34 KCNJ11 Potassium channel, inwardly rectifying 11p15.5 Candidate gene 34 subfamily J, member 11 WFS1 Wolfram syndrome 1 (wolframin) 4p16.1 Candidate gene 35 HNF1B HNF1 homeobox B 17q12 Candidate gene 36 JAZF1 Juxtaposed with another zinc finger gene 1 7p15 GWA 37 CDC123-CAMK1D Cell division cycle protein 123 homolog/ 10p13-p14 GWA 37 calcium/calmodulin-dependent protein kinase 1D TSPAN8-LGR5 Tetraspanin 8 and leucine-rich-repeat- 12q21 GWA 37 containing G-protein coupled THADA Thyroid adenoma-associated 2p21 GWA 37 ADAMS9 ADAM metallopeptidase with 3p14 GWA 37 thrombospondin type 1 motif, 9 NOTCH2 Notch homolog 2, Drosophila 1p12 GWA 37 ADCY5 Adenylate cyclase 3 GWA 38 CDKN2A/B Cyclin-dependent kinase inhibitor 2A/B 9p21 GWA 40 CDKAL1 CDK5 regulatory subunit associated 6p22.2 GWA 34 and 40 protein 1-like 1 SLC30A8 Solute carrier family 30, member 8 8q24.11 GWA 34 and 40 IGF2BP2 Insulin-like growth factor 2 mRNA 3q28 GWA 34 and 40 binding protein 2 HHEX/IDE Hematopoietically expressed homeobox 10q23-q25 GWA 34 and 40 and insulin-degrading enzyme FTO Fat mass and obesity associated 16q12.2 GWA 34 MTNR1B Melatonin receptor 1B 11q21-q22 GWA 42 KCNQ1 Potassium channel, voltage-gated, KQT-like 12q21 GWA 32 subfamily, member 1 IRS1 Insulin receptor substrate 1 2q36 GWA 41 GCK Glucokinase 7p15.3-p15.1 GWA 42 GCKR Glucokinase regulator 2p23 GWA 42 C6PC2 Glucose-6-phosphatase, catalytic 2 2q24.3 GWA 42 TNFa Tumor necrosis factor-a 6p21.3 Candidate gene 45 HNF1A Hepatocyte nuclear factor 1a 12q24.2 Candidate gene 46 HNF4A Hepatocyte nuclear factor 4a 20q13.12 Candidate gene 46 tissues and biofluids. There are currently for up to 30–50% of genetic type 1 dia- The highest-risk DR/DQ haplotypes, 7,900 entries in the Human Metabolome betes risk (48). The association of genes DR3 and DR4, exhibit a spectrum of risk Database (http://www.hmdb.ca), which of the class II region is thought to reflect from increased to neutral to protective (50). contains detailed information about small their role in the T-cell immune response. Type 1 diabetes incidence is increasing molecule metabolites and will be useful for The major susceptibility class II loci are worldwide each year, and it appears that applications in metabolomics, clinical HLA-DRB1 and HLA-DQB1/DQA1 on as the disease increases the percentage of chemistry, and biomarker discovery. chromosome 6p21 and, to a lesser extent, cases with the high-risk HLA DR3/4 geno- HLA-DPB1/DPA1 (48,49). Association type is decreasing, suggesting an increased Susceptibility gene markers in studies are complicated by the high poly- environmental pressure or contribution of type 1 diabetes morphism of the HLA DPA1 and DPB1 other non-HLA class II alleles to diabetes In type 1 diabetes, the study of suscepti- loci. The HLA loci, DRB1, DQA1, DQB1, risk (51). Type 1 diabetes risk is also linked bility genes has been facilitated by the DPA1, DPB1, A, B, and C, have repeatedly to the major histocompatibility complex availability of large collections of families been shown to be involved in type 1 diabe- independently of HLA-DR/DQ such as with affected sibling pairs as seen in the tes susceptibility. However, conflicting re- HLA class I alleles (52). Studies have Type 1 Diabetes Genetics Consortium sults have been obtained for the HLA loci supported a role for HLA class I alleles in (48). Type 1 diabetes is a multifactorial involved in susceptibility or protection as type 1 diabetes susceptibility including disease where loci within the HLA account a result of coinherited loci, population- B*3906 and B*5701 (53). The large data- for most of the genetic susceptibility. The specific differences, typing approaches sets generated by studies such as the Type 1 major susceptibility locus maps to the used, differences in study design, or low- Diabetes Genetics Consortium are crucial HLA class II genes at 6p21, accounting powered studies (49). for the generation of sufficient class I data

2628 DIABETES CARE, VOLUME 34, DECEMBER 2011 care.diabetesjournals.org McKillop and Flatt for disease association studies (48). Studies associated with additional features of the model from a panel of serum biomarkers are ongoing, investigating HLA class I and type 2 diabetes phenotype such as impaired from the Inter99 cohort. Diabetes Care class III alleles. glucose tolerance, defective first-phase 2009;32:1207–1212 More than 40 non-HLA susceptibility insulin release, and insulin resistance. Com- 8. Salomaa V, Havulinna A, Saarela O, et al. gene markers have been identified that bination of genetic and biomarkers screens Thirty-one novel biomarkers as predictors for clinically incident diabetes. PLoS ONE contribute to type 1 diabetes risk. However, may provide further opportunities. The 2010;5:e10100 for many of these genetic predictors of risk challenges in harnessing the potential of 9. Christ A, Evers S, Krapfenbauer K, Sebokova the effect is small, even for the strongest loci new biomarkers should be alleviated by E. Pancreatic polypeptide as target/marker (54). These non-MHC loci include the in- new and exciting collaborations between of beta cell failure. US patent 2009/0215069. sulin gene (INS) on chromosome 11p15 pharmaceutical agencies, diagnostic com- 27 August 2009 (55), which confers ~10% of the genetic sus- panies, and academic institutions, with the 10. Amrein K, Berndt P, Evers S, Foster S, ceptibility to type 1 diabetes. The cytotoxic harnessing of skills from the different Fountoulakis M, Sebokova E. Fibronectin Tcell–associated protein 4 (CTLA4)geneon clinical, biomedical, diagnostic, and phar- as target/marker for insulin resistance. chromosome 2q33 is associated with type 1 macological areas. International patent WO/2007/06278. diabetes (56). Also, the protein tyrosine 7 June 2007 11. Stefan N, Fritsche A, Weikert C, et al. phosphatase, nonreceptor type 22 (lym- d PTPN22 Acknowledgments Work cited from the au- Plasma fetuin-A levels and the risk of type phoid) ( ) gene on chromosome thors’ laboratory has been supported by the 2 diabetes. Diabetes 2008;57:2762–2767 1p13 (57), involved in preventing sponta- Wellcome Trust, Diabetes UK, the Depart- 12. Ding EL, Song Y, Malik VS, Liu S. Sex dif- neous T-cell activation, is linked to type 1 ment of Health and Social Sciences (Northern ferences of endogenous sex hormones and diabetes risk (57). A number of other asso- Ireland), and the University of Ulster Strategic risk of type 2 diabetes: a systematic review and ciations have been proposed such as IL-2 Funding. meta-analysis. JAMA 2006;295:1288–1299 receptor, -a (IL2RA), and interferon- The authors are also supported by Innovation 13. Sandhu MS. Insulin-like growth factor-I induced with helicase C domain 1 (IFIH1) Ulster insofar as they hold several patents in the and risk of type 2 diabetes and coronary genes (58) and KIAA0350 (59) and small areas of diabetes diagnostics and therapeutics. heart disease: molecular epidemiology. fl – ubiquitin-like modifier 4 (SUM04) (60). No other potential con icts of interest relevant Endocr Dev 2005;9:44 54 Many new candidate genes are emerging to this article were reported. 14. Armbruster DA. Fructosamine: structure, A.M.M. researched data and wrote the man- analysis, and clinical usefulness. Clin Chem such as IL10, IL19, IL20, GLIS3, CD69, and – IL27 uscript. P.R.F. researched data and reviewed and 1987;33:2153 2163 (58), but further genotyping and edited the manuscript. 15. Nowatzke W, Sarno MJ, Birch NC, Stickle functional studies are needed to determine DF, Eden T, Cole TG. Evaluation of an assay whether the genes are causal. With impor- for serum 1,5-anhydroglucitol (GlycoMark) tant breakthroughs in DNA sequencing References and determination of reference intervals on technology and mapping of diabetes cases, 1. Lanza IR, Zhang S, Ward LE, Karakelides the Hitachi 917 analyzer. Clin Chim Acta the determination of extreme genetic risk H, Raftery D, Nair KS. Quantitative me- 2004;350:201–209 of type 1 diabetes in the general population tabolomics by H-NMR and LC-MS/MS 16. Kouzuma T, Usami T, Yamakoshi M, could eventually lead to intervention or confirms altered metabolic pathways in Takahashi M, Imamura S. An enzymatic prevention trials. diabetes. PLoS ONE 2010;5:e10538 method for the measurement of glycated 2. Li LO, Hu YF, Wang L, Mitchell M, Berger albumin in biological samples. Clin Chim A, Coleman RA. Early hepatic insulin re- Acta 2002;324:61–71 Future developments in the search sistance in mice: a metabolomics analysis. 17. Lindsay JR, McKillop AM, Mooney MH, for improved biomarkers for the Mol Endocrinol 2010;24:657–666 O’Harte FPM, Bell PM, Flatt PR. Demon- diagnosis and treatment 3. Bain JR, Stevens RD, Wenner BR, Ilkayeva stration of increased concentrations of of diabetes O, Muoio DM, Newgard CB. Metabolomics circulating glycated insulin in human type Translating research findings to useful and applied to diabetes research: moving from 2 diabetes using a novel and specific ra- reliable clinical tests has been challenging; information to knowledge. Diabetes 2009; dioimmunoassay. Diabetologia 2003;46: however, the discovery of ideal biomarkers 58:2429–2443 475–478 for diabetes is improving along with the 4. Meisinger C, Löwel H, Heier M, Schneider 18. Rittenhouse J, Koda J, Fineman M, Percy A. development of biomarker panels and new A, Thorand B; KORA Study Group. Serum Glycosylated amylins. International patent WO/1997/035600. 2 October 1997 methodologies. In the future, diagnostic gamma-glutamyltransferase and risk of type 2 diabetes mellitus in men and women from 19. Cohen MP. Monoclonal antibodies against tests may be used to select individuals who fi the general population. J Intern Med 2005; glycated low density lipoprotein. US patent are likely to bene t from treatment or those 258:527–535 5,494,791. 27 February 1996 who demonstrate an objective indication of 5. Schindhelm RK, Diamant M, Dekker JM, 20. Maritim AC, Sanders RA, Watkins JB 3rd. treatment efficacy. Emergence of diabetes- Tushuizen ME, Teerlink T, Heine RJ. Ala- Diabetes, oxidative stress, and antioxidants: associated genetic variants represents a pow- nine aminotransferase as a marker of non- a review. J Biochem Mol Toxicol 2003;17: erful tool for improving our understanding alcoholic fatty liver disease in relation to 24–38 of the pathogenesis of diabetes. However, type 2 diabetes mellitus and cardiovascular 21. Sargeant LA, Wareham NJ, Bingham S, translation of these novel findings to genetic disease. Diabetes Metab Res Rev 2006;22: et al. Vitamin C and hyperglycemia in the 437–443 European Prospective Investigation into screening and personalized medicine is still d at an early stage. Characterization of func- 6. Sattar N, Wannamethee SG, Forouhi NG. Cancer Norfolk (EPIC-Norfolk) study: Novel biochemical risk factors for type a population-based study. Diabetes Care tional variants and an understanding of the 2 diabetes: pathogenic insights or pre- 2000;23:726–732 mechanisms by which these loci confer diction possibilities? Diabetologia 2008; 22. Zhao G, Ford ES, Li C. Associations of se- susceptibility to disease are needed. With 51:926–940 rum concentrations of 25-hydroxyvitamin discovery of genes linked to fasting glucose, 7. Kolberg JA, Jørgensen T, Gerwien RW, D and parathyroid hormone with surrogate itmaybepossibletoidentifyotherloci et al. Development of a type 2 diabetes risk markers of insulin resistance among U.S. care.diabetesjournals.org DIABETES CARE, VOLUME 34, DECEMBER 2011 2629 Metabolic and genomic profiling in diabetes

adults without physician-diagnosed dia- 35. Sandhu MS, Weedon MN, Fawcett KA, 48. Erlich H, Valdes AM, Noble J, et al.; Type betes: NHANES, 2003-2006. Diabetes Care et al. Common variants in WFS1 confer 1 Diabetes Genetics Consortium. HLA 2010;33:344–347 risk of type 2 diabetes. Nat Genet 2007; DR-DQ haplotypes and genotypes and 23. Wang TJ, Larson MG, Vasan RS, et al. Me- 39:951–953 type 1 diabetes risk: analysis of the type 1 tabolite profiles and the risk of developing 36. Winckler W, Weedon MN, Graham RR, diabetes genetics consortium families. diabetes. Nat Med 2011;17:448–453 et al. Evaluation of common variants in Diabetes 2008;57:1084–1092 24. Koyama H, Yamamoto H, Nishizawa Y. the six known maturity-onset diabetes of 49. Varney MD, Valdes AM, Carlson JA, et al.; Endogenous secretory RAGE as a novel the young (MODY) genes for association Type 1 Diabetes Genetics Consortium. biomarker for metabolic syndrome and with type 2 diabetes. Diabetes 2007;56: HLA DPA1, DPB1 alleles and haplotypes cardiovascular diseases. Biomark Insights 685–693 contribute to the risk associated with type 2007;2:331–339 37. Zeggini E, Scott LJ, Saxena R, et al.; 1 diabetes: analysis of the type 1 diabetes 25. Thorand B, Zierer A, Baumert J, Meisinger Wellcome Trust Case Control Consor- genetics consortium families. Diabetes C, Herder C, Koenig W. Associations be- tium. Meta-analysis of genome-wide 2010;59:2055–2062 tween leptin and the leptin/adiponectin association data and large-scale replica- 50. Pugliese A, Gianani R, Moromisato R, ratio and incident type 2 diabetes in mid- tion identifies additional susceptibility et al. HLA-DQB1*0602 is associated dle-aged men and women: results from the loci for type 2 diabetes. Nat Genet 2008; with dominant protection from diabetes MONICA/KORA Augsburg study 1984- 40:638–645 even among islet cell antibody-positive 2002. Diabet Med 2010;27:1004–1011 38. Saxena R, Hivert MF, Langenberg C, first-degree relatives of patients with IDDM. 26. Freeman DJ, Norrie J, Caslake MJ, et al.; West et al; GIANT consortium; MAGIC inves- Diabetes 1995;44:608–613 of Scotland Coronary Prevention Study. tigators. Genetic variation in GIPR in- 51. Steck AK, Armstrong TK, Babu SR, C-reactive protein is an independent pre- fluences the glucose and insulin responses Eisenbarth GS; Type 1 Diabetes Genetics dictor of risk for the development of di- to an oral glucose challenge. Nat Genet Consortium. Stepwise or linear decrease abetes in the West of Scotland Coronary 2010;42:142–148 in penetrance of type 1 diabetes with Prevention Study. Diabetes 2002;51: 39. Salonen JT, Uimari P, Aalto JM, et al. Type lower-risk HLA genotypes over the past 1596–1600 2 diabetes whole-genome association 40 years. Diabetes 2011;60:1045–1049 27. Thorand B, Kolb H, Baumert J, et al. Ele- study in four populations: the DiaGen 52. Noble JA, Martin A, Valdes AM, et al. vated levels of interleukin-18 predict consortium. Am J Hum Genet 2007;81: Type 1 diabetes risk for human leukocyte the development of type 2 diabetes: re- 338–345 antigen (HLA)-DR3 haplotypes depends sults from the MONICA/KORA Augsburg 40. Zeggini E, Weedon MN, Lindgren CM, on genotypic context: association of Study, 1984-2002. Diabetes 2005;54: et al; Wellcome Trust Case Control Con- DPB1 and HLA class I loci among DR3- 2932–2938 sortium (WTCCC). Replication of ge- and DR4-matched Italian patients and 28. Herder C, Brunner EJ, Rathmann W, et al. nome-wide association signals in UK controls. Hum Immunol 2008;69:291– Elevated levels of the anti-inflammatory samples reveals risk loci for type 2 di- 300 interleukin-1 receptor antagonist precede abetes. Science 2007;316:1336–1341 53. Valdes AM, Erlich HA, Noble JA. Human the onset of type 2 diabetes: the Whitehall 41. Rung J, Cauchi S, Albrechtsen A, et al. leukocyte antigen class I B and C loci II study. Diabetes Care 2009;32:421–423 Genetic variant near IRS1 is associated contribute to Type 1 Diabetes (T1D) sus- 29. Festa A, D’Agostino R Jr, Tracy RP, Haffner with type 2 diabetes, insulin resistance ceptibility and age at T1D onset. Hum SM; Insulin Resistance Atherosclerosis and hyperinsulinemia. Nat Genet 2009; Immunol 2005;66:301–313 Study. Elevated levels of acute-phase pro- 41:1110–1115 54. Steck AK, Rewers MJ. Genetics of type 1 teins and plasminogen activator inhibitor-1 42. Soranzo N, Sanna S, Wheeler E, et al.; diabetes. Clin Chem 2011;57:176–185 predict the development of type 2 dia- WTCCC. Common variants at 10 genomic 55. Bell GI, Horita S, Karam JH. A polymorphic betes: the insulin resistance atherosclerosis loci influence hemoglobin A1c levels via locus near the human insulin gene is as- study. Diabetes 2002;51:1131–1137 glycemic and nonglycemic pathways. Di- sociated with insulin-dependent diabetes 30. Song Y, Manson JE, Tinker L, et al. Cir- abetes 2010;59:3229–3239 mellitus. Diabetes 1984;33:176–183 culating levels of endothelial adhesion 43. Schleinitz D, Tönjes A, Böttcher Y, et al. 56. Concannon P, Chen WM, Julier C, et al.; molecules and risk of diabetes in an eth- Lack of significant effects of the type Type 1 Diabetes Genetics Consortium. nically diverse cohort of women. Diabetes 2 diabetes susceptibility loci JAZF1, Genome-wide scan for linkage to type 1 2007;56:1898–1904 CDC123/CAMK1D, NOTCH2, ADAMTS9, diabetes in 2,496 multiplex families from 31. Wannamethee SG, Sattar N, Rumley A, THADA, and TSPAN8/LGR5 on diabetes the Type 1 Diabetes Genetics Consortium. Whincup PH, Lennon L, Lowe GD. Tissue and quantitative metabolic traits. Horm Diabetes 2009;58:1018–1022 plasminogen activator, von Willebrand Metab Res 2010;42:14–22 57. Bottini N, Musumeci L, Alonso A, et al. A factor, and risk of type 2 diabetes in older 44. Boraska V, Rayner NW, Groves CJ, et al. functional variant of lymphoid tyrosine men. Diabetes Care 2008;31:995–1000 Large-scale association analysis of TNF/ phosphatase is associated with type I di- 32. Voight BF, Scott LJ, Steinthorsdottir V, LTA gene region polymorphisms in type 2 abetes. Nat Genet 2004;36:337–338 et al.; MAGIC investigators; GIANT Con- diabetes. BMC Med Genet 2010;11:69 58. Barrett JC, Clayton DG, Concannon P, sortium. Twelve type 2 diabetes suscep- 45. Arora P, Garcia-Bailo B, Dastani Z, et al. et al. Genome-wide association study and tibility loci identified through large-scale Genetic polymorphisms of innate immunity- meta-analysis find that over 40 loci affect association analysis. Nat Genet 2010;42: related inflammatory pathways and their risk of type 1 diabetes. Nat Genet 2009; 579–589 association with factors related to type 2 41:703–707 33. Grant SF, Thorleifsson G, Reynisdottir I, diabetes. BMC Med Genet 2011;12:95 59. Concannon P, Onengut-Gumuscu S, et al. Variant of transcription factor 7-like 46. McDonald TJ, Shields BM, Lawry J, Owen Todd JA, et al.; Type 1 Diabetes Genetics 2 (TCF7L2) gene confers risk of type 2 KR, Gloyn AL, Ellard S, Hattersley AT. Consortium. A human type 1 diabetes diabetes. Nat Genet 2006;38:320–323 High-sensitivity CRP discriminates susceptibility locus maps to chromosome 34. Scott LJ, Mohlke KL, Bonnycastle LL, et al. HNF1A-MODY from other subtypes of di- 21q22.3. Diabetes 2008;57:2858–2861 A genome-wide association study of type abetes. Diabetes Care 2011;34:1860–1862 60. Guo D, Li M, Zhang Y, et al. A functional 2 diabetes in Finns detects multiple 47. Psychogios N, Hau DD, Peng J, et al. The variant of SUMO4, a new I kappa B alpha susceptibility variants. Science 2007;316: human serum metabolome. PLoS ONE modifier, is associated with type 1 diabetes. 1341–1345 2011;6:e16957 Nat Genet 2004;36:837–841

2630 DIABETES CARE, VOLUME 34, DECEMBER 2011 care.diabetesjournals.org