Mendelian randomization studies of biomarkers and type 2 diabetes

in Endocrine Connections
Correspondence should be addressed to A Abbasi Email: ali.abbasi@mrc-epid.cam.ac.uk

Many biomarkers are associated with type 2 diabetes (T2D) risk in epidemiological observations. The aim of this study was to identify and summarize current evidence for causal effects of biomarkers on T2D. A systematic literature search in PubMed and EMBASE (until April 2015) was done to identify Mendelian randomization studies that examined potential causal effects of biomarkers on T2D. To replicate the findings of identified studies, data from two large-scale, genome-wide association studies (GWAS) were used: DIAbetes Genetics Replication And Meta-analysis (DIAGRAMv3) for T2D and the Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC) for glycaemic traits. GWAS summary statistics were extracted for the same genetic variants (or proxy variants), which were used in the original Mendelian randomization studies. Of the 21 biomarkers (from 28 studies), ten have been reported to be causally associated with T2D in Mendelian randomization. Most biomarkers were investigated in a single cohort study or population. Of the ten biomarkers that were identified, nominally significant associations with T2D or glycaemic traits were reached for those genetic variants related to bilirubin, pro-B-type natriuretic peptide, delta-6 desaturase and dimethylglycine based on the summary data from DIAGRAMv3 or MAGIC. Several Mendelian randomization studies investigated the nature of associations of biomarkers with T2D. However, there were only a few biomarkers that may have causal effects on T2D. Further research is needed to broadly evaluate the causal effects of multiple biomarkers on T2D and glycaemic traits using data from large-scale cohorts or GWAS including many different genetic variants.

Abstract

Many biomarkers are associated with type 2 diabetes (T2D) risk in epidemiological observations. The aim of this study was to identify and summarize current evidence for causal effects of biomarkers on T2D. A systematic literature search in PubMed and EMBASE (until April 2015) was done to identify Mendelian randomization studies that examined potential causal effects of biomarkers on T2D. To replicate the findings of identified studies, data from two large-scale, genome-wide association studies (GWAS) were used: DIAbetes Genetics Replication And Meta-analysis (DIAGRAMv3) for T2D and the Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC) for glycaemic traits. GWAS summary statistics were extracted for the same genetic variants (or proxy variants), which were used in the original Mendelian randomization studies. Of the 21 biomarkers (from 28 studies), ten have been reported to be causally associated with T2D in Mendelian randomization. Most biomarkers were investigated in a single cohort study or population. Of the ten biomarkers that were identified, nominally significant associations with T2D or glycaemic traits were reached for those genetic variants related to bilirubin, pro-B-type natriuretic peptide, delta-6 desaturase and dimethylglycine based on the summary data from DIAGRAMv3 or MAGIC. Several Mendelian randomization studies investigated the nature of associations of biomarkers with T2D. However, there were only a few biomarkers that may have causal effects on T2D. Further research is needed to broadly evaluate the causal effects of multiple biomarkers on T2D and glycaemic traits using data from large-scale cohorts or GWAS including many different genetic variants.

Introduction

Over the past decade, interest in studying biological markers (biomarkers) for type 2 diabetes (T2D) has increased intensely. This happened because multiple pathobiological processes may contribute to the disease progression, which provides an opportunity to introduce preventive and therapeutic interventions for T2D (1). In clinical practice, such biomarkers (e.g., glucose and glycated haemoglobin tests) are widely used for the diagnosis of diabetes or for the monitoring of therapeutic intervention (2, 3). Targeted intervention at the biomarker level would be useful where there is evidence for a causal relationship between an exposure (like a biomarker) and T2D (4).

Traditional epidemiological studies lack sufficient information to fill the evidence gap due to unmeasured confounding or reverse causality (4, 5, 6, 7). It has been successfully shown that a complementary analysis of genetic data, termed ‘Mendelian randomization,’ has additive value to infer a causal association (4, 5, 6, 7, 8). The main assumption for Mendelian randomization is that the genetic variants do not change over time and are inherited randomly (based on Mendel's laws). In other words, the genetic variants as proxy measures for exposures (e.g., biomarkers) are essentially considered free from confounding and reverse causation. Therefore, the analysis of integrated observational-genetic data is considered similar to that of the randomized trials (9, 10). In Mendelian randomization, if there is a causal association between a biomarker and T2D, the genetic variant(s) influencing the biomarker and the outcome of interest should be associated (5, 6, 8). In the current study, evidence for causal associations between biomarkers and the risk of T2D was updated via a systematic literature search to identify Mendelian randomization studies. Next, summary data from the two largest genome-wide association studies (GWAS) for T2D or glycaemic traits (10, 11) were used to examine the effect estimates for each genetic variant compared with that of the identified studies.

Methods

Search strategy for candidate biomarkers

PubMed and EMBASE were searched to identify Mendelian randomization studies examining the associations between biomarkers and T2D until April 2015. The overview of this systematic literature search was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines, when applicable (12). A manual search was also done for the references of included articles to identify other relevant studies. Because a review of published original studies was performed, the Declaration of Helsinki items related to ‘approval of medical ethics committee’ and ‘permission acquisition’ are not applicable to the current study.

Selection criteria

Studies were included if they formally quantified a causal association between one or more biomarkers (as main exposures) and T2D (as the main outcome); used data on biomarker-associated genetic variant; and classified/defined the exposure as a biomarker that has been objectively measured in serum, plasma or urine. The following MeSH keywords or related terms were used: Diabetes mellitus, Type 2[Mesh], Type 2 Diabetes[tiab], Diabetes[ti]; and Mendelian randomization, Mendelian randomisation, instrumental variable, genetic risk score or causal association.

Data extraction and quality assessment

A primary plan was made to extract necessary data from the full text of the original studies or to contact the corresponding author(s) when appropriate. Fig. 1 depicts the workflow of the literature search. T2D was determined as the main outcome if one or more of the following conditions were fulfilled: a physician diagnosed T2D as indicated by a self-report or in a primary-care database; fasting plasma glucose ≥7.0 mmol/l, a random sample plasma glucose ≥11.1 mmol/l; or the initiation of glucose-lowering medication as retrieved from a pharmacy registry or hospital records (11).

Figure 1
Figure 1

Flow diagram of the literature search.

Citation: Endocrine Connections 4, 4; 10.1530/EC-15-0087

Download Figure

Statistical analysis

All genetic variants that affect identified biomarkers were obtained from the original Mendelian randomization studies. To replicate the nature of the relationship between each biomarker and the outcome (i.e., T2D or glycaemic traits), a genetic approach using outcome-association data for the biomarker-related genetic variants was applied (5, 6, 8). In brief, to determine whether the same genetic variants (or a suitable proxy variant) were associated with T2D (or glycaemic traits), the corresponding summary association statistics from GWAS for T2D (DIAbetes Genetics Replication And Meta-analysis (DIAGRAMv3)) (13) and glycaemic traits (Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC)) (14) were extracted. DIAGRAMv3 is a meta-analysis of multiple GWAS with a total number of 12 171 diabetes cases and 56 862 controls of European descent (13). Previously, details regarding the use of GWAS data for Mendelian randomization were described (5, 6, 10, 11). These selected single-test association analyses in the GWAS data are equivalent to that of individual-level data analysis (5, 6, 11).

Extracted data were tabulated on Excel spreadsheets. All statistical analyses were conducted using Excel or Stata/SE version 13.1 for Windows (http://cran.r-project.org/). A two-sided P value <0.05 was considered nominally significant.

Results

Literature search and study characteristics

After scanning 812 titles and selected abstracts, 33 articles were selected for full text review (11, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46). The literature search process is shown in Fig. 1. Five studies were excluded as they did not measure a specific biomarker (n=1) or investigated a measure of insulin resistance as the main exposure (n=4). The characteristics of the 28 studies are summarized in Table 1.The studies were performed in different populations; most were cohort studies including middle-aged adults in the USA or Europe and were published between 2008 and 2015. Four studies were conducted only in Asian populations including Chinese or Taiwanese. In 12 studies, data from multiple GWAS were combined to examine the associations between genotypes and T2D. Six of these studies used data from DIAGRAMv2 (n=3) (47) or DIAGRAMv3 (n=3) (13), and the other six used at least two cohorts with genome-wide genotyping. DIAGRAMv2 was a meta-analysis of eight GWAS comprising 8130 individuals with T2D and 38 987 controls of European descent (47). All Mendelian randomization studies investigated only one biomarker as an exposure in relation to T2D. Some studies also examined the causal associations of a single biomarker with several other outcomes such as cardiovascular disease (18, 24), rheumatoid arthritis (18) or osteoporosis (24). In the final sample, studies included up to 28 144 T2D cases and 76 344 T2D controls or total participants.

Table 1

Mendelian randomization studies for biomarkers and type 2 diabetes.

BiomarkersNo. of studyStudyCountry/yearCases/total or controlsGenetic variant(s)/locusCausal estimateType of associations
Adiponectin3 (19, 35, 42)DIAGRAMv2, ARIC, FUSION, WTCCC, EPIC-InterActUS, Europe/201315 960/64 731rs17300539, rs1736653, rs3774261, rs3821799/ADIPOQNull
BHS, CUDAS, FDSAustralia/2013967/2355rs12637534, rs16861209, rs17366568/ADIPOQOR, 0.86Per allele
DIAGRAMv2US, Europe/2012–/22 044rs3001032/LYPLAL1, rs1108842/GNL3, rs1597466/TSC22D2, rs6810075/ADIPOQ, rs998584/VEGFA, rs2980879/TRIB1, rs7955516/PDE3A, rs601339/GPR109A, rs7133378/DNAH10, rs2925979/CMIP, rs12922394/CDH13, rs731839/PEPDβ, −0.30Average additive effect of adiponectin- raising alleles
Beta-carotene1 (33)WTCCC, DGI, FUSIONUSA, Europe/20094549/5579rs6564851/BCMO1Null
Bilirubin1 (11)PREVENDNetherlands/2015210/3381rs6742078/UGT1A1OR, 0.581-s.d.
N-terminal pro B-type natriuretic peptide1 (37)DIAGRAMv2, Norfolk Diabetes, Cambridgeshire, ADDITION-Ely, French, Swiss 1, Swiss 2, Austrian, MONICA, INCA, UKUS, Europe/201123 382/57 898rs198389/NPPBOR, 0.96C allele
C-reactive protein1 (16)Whitehall II UK/2008354/5274rs1130864, rs1205, rs3093077/ CRPNull
Delta-6 desaturase1 (30)EPIC-Potsdam Germany/2011673/2724rs174546/FADS1RR, 0.64TT genotype
Dimethylglycine1 (32)MDC-CC, MDC, MPPSweden/20154201/33 898rs2431332/ DMGDHOR, 1.1A allele
Ferritin/TMPRSS61 (22)BeijingChina/2012272/1574rs855791, rs4820268/ TMPRSS6OR, 0.79AG haplotype
Fetuin-a1 (27)CHSUS/2013259/3093rs4917, rs2248690/AHSGNull
Homocysteine2 (24, 25)Hangzhou China/2014774/500rs1801131/MTHFR, rs4646356/PEMTOR, 1.93, 1.52CC, CT+TT
Europe, Asia, Africa/20134011/4303rs12134663, rs1801133/MTHFROR, 1.295 μmol/l
IL-1Ra1 (26)DIAGRAMv3 and EPIC-InterActUS, Europe/201518 715/61 692rs6743376, rs1542176/IL1RNNull
Leukocyte telomere length1 (46)WHI-OSUS/20121675/2382rs34368910/TRF1, rs4888444/TRF2, rs4975605/TERT, rs938886, rs2228041,rs12880583/TPP1Null
Lipoprotein(a)2 (29, 44)Danish general populationDenmark/20142157/28 567rs10455872/LPANull
DIAGRAMv3,EPIC-NorfolkUSA, Europe/201410 088/68 346rs10455872/LPANull
Macrophage migration inhibitory factor1 (23)MONICA/KORA AugsburgGermany/2008502/1632rs1007888/MIFHR, 1.74C allele
miRNAs1 (41)HarbinChina/2015995/967rs895819/miR-27a, rs531564/miR-124a, rs2910164/miR-146aNull
Resistin1 (18)CVDFACTSTaiwan/2014 US, Europe/2010230/3400rs3745367,rs1423096/RETNOR, 1.38GG
Sex hormone-binding globulin2 (21, 34)WTCCC, Dundee, EFS-Y2TD, Danish, DGI, MCC, FUSION 1-2, KORA, Cambridgeshire, ADDITION-Ely, NDCCS, DeCODE, METSIM, DIAGEN27 657/58 481rs1799941/SHBGOR, 0.92G allele
WHSUSA/2009359/359rs6259/SHBGOR, 0.31-s.d.
Triglycerides1 (20)Go-DARTSUK/20115637/6860rs2954029/ TRIB1, rs714052/MLXIPL, rs7557067/APOB, rs17216525/NCAN, rs10889353/ANGPTL3, rs7679/PLTP, rs7819412/XKR6-AMAC1L2, rs328/LPL, rs3135506/APOA5, rs662799/APOA5Null
Uric acid1 (36)Cambridgeshire, ADDITION-Ely, Norfolk DiabetesUK/20117504/8560rs12129861/PDZK1, rs742132/LRRC16A, rs505802/SLC22A12, rs12356193/SLC16A9, rs17300741/SLC22A11, rs1165151/SLC17A1, rs2231142/ABCG2, rs734553/ SLC2A9Null
Vitamin D3 (17, 28, 45)DIAGRAMv3, ADDITION-Ely, Norfolk Diabetes, Cambridgeshire, EPIC-InterActUS, Europe/2015281 44/76 344rs12785878/DHCR7, rs10741657/CYP2R1, rs4588/DBP, rs17217119/CYP24A1Null
EPIC-PotsdamGermany/20131572/2121rs2282679, rs1155563/GC, rs12785878, rs3829251/DHCR7, rs10741657/CYP2R1, rs6013897/CYP24A1, rs6599638/Ch10orf88, rs10877012/CYP27B1Null
TromsøNorway/20121092/9528rs2298850/DBP GC, rs10741657/CYP2R1, rs3794060/NADSYN1, rs6013897/CYP24A1Null
Vitamin D binding protein1 (31)CaMosCanada/2014201/2254rs2282679/DBP GCNull

RR, relative risk; IL1-Ra, interleukin 1 receptor antagonist; DIAGRAM, DIAbetes Genetics Replication And Meta-analysis; EPIC-InterAct, European Prospective Investigation into Cancer and Nutrition (EPIC)-InterAct; ARIC, Atherosclerosis Risk In Communities; WTCCC, Welcome Trust Case Control Cohort; FUSION, Finland-US Investigation of NIDDM genetics; BHS, Busselton Population Health Survey; CUDAS, Carotid Ultrasound Disease Assessment Study; FDS, Fremantle Diabetes Study; PREVEND, Prevention of Renal and Vascular ENd-stage Disease; iNCA, iNsuffisance CAdiaque; MONICA, Multinational MONItoring of trends and determinants in CArdiovascular disease; MDC, Malmö Diet and Cancer Study; MDC-CC, Malmö Diet and Cancer Cardiovascular Cohort; MPP, Malmö Preventive Project; CVDFACTS, CardioVascular Disease risk FACtors Two-township Study; KORA, Kooperative Gesundheitsforschung in der Region Augsburg; EFS-YT2D, Exeter Family Study-Young-onset T2D; NDCCS, Norfolk Diabetes Case Control Study; METSIM, METabolic Syndrome In Men; DIAGEN, DIAbetes GENetic study; CHS, Cardiovascular Health Study; WHI-OS, Women's Health Initiative Observational Study Cohort; ADDITION, Anglo-Danish-Dutch Study of Intensive Treatment In People with Screen Detected Diabetes in Primary Care; Go-DARTS, Genetics of Diabetes Audit and Research in Tayside Scotland; CaMos, Canadian Multicentre Osteoporosis Study.

Biomarkers and Mendelian randomization

From 28 studies, data were retrieved on causal associations of biomarkers with T2D. In these studies, 21 unique biomarkers that were investigated at least once (16 biomarkers), twice (three biomarkers) and three times (two biomarkers) were identified (Table 1); 19 biomarkers were investigated in independent studies. However, the two biomarkers with three Mendelian randomization studies were investigated in combined studies where the same sample, e.g., DIAGRAMv2 for adiponectin (19, 43), or a part of whole cohort, e.g., European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam for vitamin D (17, 45), were used to make total cases/controls. Eleven studies used one genetic variant as a single instrumental variable in Mendelian randomization. Eight studies used at least two independent genetic variants in the same locus as instrumental variables. The rest of the nine studies used multiple genetic variants in different loci or created a multi-locus genetic risk score for each biomarker.

In main or sensitivity analyses, four studies (11, 26, 44, 45) examined causal associations of bilirubin, lipoprotein (a), vitamin D, interleukin-1 receptor antagonist (with corresponding biomarker-associated genotypes) with T2D using the DIAGRAMv3 summary data. In Table 1, the value for each causal estimate can be interpreted as odds ratio (OR), hazard ratio (HR) or log OR (β coefficient) for diabetes per one unit change in genetically determined biomarkers or biomarker-associated genotypes. Evidence of causal association was reported for adiponectin, bilirubin, N-terminal pro B-type natriuretic peptide (NT proBNP), delta-6 desaturase (D6D), dimethylglycine, ferritin (transmembrane protease serine 6), homocysteine, macrophage migration inhibitory factor (MIF), sex hormone-binding protein (SHBG) and resistin (Table 1). For adiponectin, causal estimates were calculated as an OR of 0.86 per allele (35) or a β coefficient of 0.3 for adiponectin multi-locus genotypic risk score (19). Causal estimate for bilirubin was reported as an OR of 0.58 per 1-s.d. genetically increased log-transformed bilirubin (11). For NT proBNP (37), D6D (30), dimethylglycine (32), ferritin (22), MIF (23) and resistin (18), the expected ORs for T2D were 0.96, 0.64, 1.1, 0.79, 1.74 and 1.38 per each copy of risk alleles or genotypes, respectively. For homocysteine, causal estimates were reported as an OR of 1.29/5 μmol/l genetically increased homocysteine (24) or ORs of 1.93 and 1.52 for CC and CC/TT genotypes (25) respectively. For SHBG, the expected ORs for T2D were 0.92 per each copy of the SHBG lowering allele (34) or 0.3/1-s.d. genetically increased natural log transformed SHBG (21).

Using the GWAS data from DIAGRAMv3 or MAGIC (13, 14), associations of the genetic variants influencing these biomarkers with T2D and glycaemic traits were tested. A nominally significance (P<0.05) was reached for the genetic variants that affect bilirubin (P=0.03 for glucose and Homeostatic model assessment (HOMA)-insulin resistance (IR)), NT proBNP (P=0.03 for T2D), D6D activity (P=0.003 for T2D; P=2.7×10−8 for glucose) and dimethylglycine (P=0.004 for glucose) in relation to T2D or glycaemic traits (Table 2). For adiponectin, a nominally significant association with T2D or glycaemic traits was observed for seven out of 19 genetic variants. All of these seven variants, except rs12637534, were non-ADIPOQ adiponectin genetic variants. In line with previous Mendelian randomization studies, the overall effect of non-ADIPOQ variants on T2D or glycaemic traits together with null association of ADIPOQ variants is compatible with pleiotropic effects of adiponectin on T2D (42).

Table 2

Nominally significance for each biomarker-associated genetic variants with glucose, HOMA-IR and the risk of type 2 diabetes.

P value*
BiomarkersGenetic variantsT2DM13Glucose14HOMA-IR14Nominally significance
Adiponectinrs1736653NA
rs29808790.960.960.93Null
rs168612090.890.850.51Null
rs6013390.890.830.03Yes
rs173005390.870.810.15Null
rs79555160.860.840.83Null
rs129223940.540.070.19Null
rs38217990.510.190.80Null
rs7318390.410.340.01Yes
rs37742610.320.440.76Null
rs173665680.310.970.40Null
rs71333780.160.210.33Null
rs126375340.150.380.00Yes
rs68100750.10.770.52Null
rs9985840.060.810.68Null
rs11088420.0320.470.05Yes
rs30010320.0290.270.00Yes
rs15974660.0090.060.47Yes
rs29259790.0020.050.18Yes
Beta-caroteners65648510.220.390.27Null
Bilirubinrs67420780.10.020.03Yes
NT proBNPrs1983890.030.370.39Yes
CRPrs12050.660.680.64Null
rs30930770.0480.950.39Yes
rs11308640.120.77Null
D6-desaturasers1745460.0032.7×10−80.89Yes
Dimethylglyciners24313320.510.0040.29Yes
Ferritin/TMPRSS6rs85579110.090.25Null
rs48202680.730.150.13Null
Fetuin-ars49170.080.110.67Null
rs22486900.0550.960.77Null
Homocysteiners46463560.880.430.35Null
rs121346630.780.690.62Null
rs18011310.220.610.24Null
rs18011330.100.460.43Null
IL-1Rars15421760.640.0080.71Yes
rs67433760.910.46Null
LTLrs12880583NA
rs2228041NA
rs34368910NA
rs9388860.880.0420.74Yes
rs49756050.660.450.61Null
rs48884440.090.910.34Null
Lp(a)rs104558720.360.420.32Null
MIFrs10078880.390.220.94Null
miRNAsrs126108730.0330.800.71Yes
rs5315640.660.530.97Null
rs29101640.520.010.36Yes
Resistinrs37453670.770.990.58Null
rs14230960.720.810.34Null
SHBGrs121506600.180.520.24Null
rs62590.940.420.35Null
Triglyceridesrs122850950.410.970.62Null
rs75570670.90.780.15Null
rs76790.880.900.29Null
rs29540290.720.640.87Null
rs6627990.280.040.55Yes
rs7140520.280.810.80Null
rs108893530.190.420.21Null
rs78194120.080.800.01Yes
rs3280.030.050.04Yes
rs172165250.000.980.81Yes
Uric acidrs11651510.740.420.86Null
rs22311420.730.830.36Null
rs5058020.620.600.62Null
rs7345530.570.200.72Null
rs7421320.250.500.91Null
rs123561930.140.950.47Null
rs121298610.080.940.50Null
rs173007410.030.340.36Yes
Vitamin-Drs10877012NA
rs37559670.740.850.19Null
rs11555630.930.770.06Null
rs22988500.660.890.15Null
rs38292510.540.090.26Null
rs37940600.340.280.10Null
rs65996380.300.170.43Null
rs107416570.230.370.08Null
rs127858780.140.320.17Null
rs172171190.060.360.21Null
rs60138970.060.330.12Null
Vitamin D-BPrs22826790.761.000.12Null

*P values for each of the biomarker variants were extracted from publicly available meta-analyses of genome-wide association studies (13, 14). NA, not applicable; CRP, C-reactive protein; BNP, brain natriuretic peptide; interleukin 1 receptor antagonist (IL1-Ra); Lp(a), Lipoprotein(a); LTL, leukocyte telomere length; MIF, macrophage migration inhibitory factor; NT proBNP, N-terminal pro B-type natriuretic peptide; SHBG, sex hormone binding globulin, Vitamin D-BP, Vitamin D binding protein.

Discussion

This literature search of Mendelian randomization studies shows here that ten out of 21 identified biomarkers were reported to be causally associated with T2D. In particular, the presence of potential causal associations (defined as nominally significant) between the biomarker-related variants and T2D and/or glycaemic traits can be confirmed for four biomarkers using the publically available GWAS data. The inconsistency between the identified studies and the summary data from DIAGRAMv3 or MAGIC can be to some extent explained by different design and populations applying across studies, the possibility of false positive associations, selection bias, the heterogeneity in T2D, the use of varied sources to ascertain T2D cases, differences in data quality control and pre-analysis preparations (6, 11).

Taken together, these findings support that the oxidative stress system (bilirubin or metabolites of bilirubin), the brain natriuretic peptide (BNP) hormone system, D6D activity and dimethylglycine may contribute to the development of diabetes or insulin resistance through secondary effects (i.e., pleiotropy) or direct mechanisms (11, 37, 42). Bilirubin that is the major end-product of heme catabolism has antioxidant properties and may compensate the oxidative stress (11, 48). Oxidative stress has been shown as an important factor in the pathophysiology of diabetes (11). At the cellular level, bilirubin can be oxidized to its precursor, biliverdin, to detoxify the excess of oxidants. Biliverdin is rapidly recycled to bilirubin via the action of biliverdin reductase, generating a physiologic cytoprotective cycle in several tissues (49). The underlying mechanism of a protective role of BNP in the aetiology of T2D is unknown in humans (37). In mice, overexpressed BNP signalling cascade can protect against diet-induced insulin resistance and obesity through promoting muscle mitochondrial biogenesis and fat oxidation (37, 50). The biological mechanisms of the relationship between D6D activity and T2D are not well understood (30). Although data of human experimental studies are scarce, observational studies have shown that D6D activity or lifestyle-induced changes in D6D activity was associated with insulin resistance (30, 51, 52). Because D6D catalyses the synthesis of fatty acids, one can speculate that the link between D6D activity and T2D is likely to be mediated by changes in fatty acid composition, which in turn may affect insulin signalling and receptor-binding affinities (30). Dimethylglycine is metabolized to glycine by dimethylglycine dehydrogenase (DMGDH) in mammals (32). Accordingly, a recent GWAS identified that the DMGDH genetic variants were strongly associated with blood-based dimethylglycine (53). Epidemiological studies have observed an inverse association between the precursor of dimethylglycine, betaine, and metabolic risk factors (54) but a positive association of elevated glycine with increased insulin sensitivity (32). In humans, cardiometabolic effects of the inhibition of DMHDH or the supplementation of dimethylglycine have not been investigated (32). However, experimental animal studies have suggested a protective role of dimethylglycine in glucose metabolism through a reduction in DMGDH function (32, 55, 56).

In the post-omics era, epidemiological studies basically suggest that the levels of a given biomarker differ between patients with T2D (or the individuals at high risk for developing diabetes) and individuals without diabetes (2, 57). If the biomarker is not causally related to the disease outcome, the process of developing diabetes may cause the increase or decrease in the levels of the biomarker, as one of the T2D consequences, called reverse causality (4, 26, 37). Unmeasured confounding or measured confounding factors with errors (for example, physical activity by self-report) is another explanation for the observed associations between most biomarkers and T2D. In this context, the use of genetic data (as in Mendelian randomization) can enhance the likelihood that the association between a biomarker and T2D is causal or not (4, 36, 37). The potential role of biomarkers in the development of T2D and the trajectories of glycaemic traits using longitudinal analysis remains to be further confirmed (6, 11). The latter analytical approach can provide insight into the potential value of biomarkers, which indicates pathobiological signals of metabolic changes in the aetiology of T2D. T2D is influenced by the interactions of multiple genes or a gene may have been mapped to multiple biomarkers rather than the biomarker of interest (4, 13, 14). Thus, an extensive knowledge of gene function and biological processes and that the genome interacts with environmental factors is needed to better understand how genetic variations in the human genome contributes to lifelong risk of T2D (4, 11, 13, 14, 57).

In this review, most studies only investigated a single biomarker-diabetes association, and statistically significant associations are reported more often. Here, publication bias should be considered. Other limitations include the lack of complementary analyses in the genetic associations for glycaemic traits and that set of genetic variants or large-scale GWAS for biomarkers were scarce. Similarly, the Mendelian randomization approach using GWAS summary statistics can be extended as a secondary analysis of several datasets that have data on the biomarker-associated variants (or their proxies) for several diseases or traits. This multidisciplinary research requires that a large group of consortia are contacted to obtain summary association statistics from GWAS consortia for outcomes of interest. Moreover, for the biomarkers linked to T2D, underlying biological mechanisms remain unknown. To uncover the underlying mechanisms, one needs to perform a complementary strand of experimental research. Before that, an in silico functional gene network (pathway) analysis can be used to speculate on the possible biological mechanisms of biomarker-disease associations (58, 59). Finally, Mendelian randomization cannot completely control for the possibility of developmental compensation, called canalization, confounding and pleiotropic effects (5, 6, 8, 11, 37).

In conclusion, this is the first study that updates evidence for causal associations between biomarkers and T2D to date. Several Mendelian randomization studies investigated the nature of associations of biomarkers with T2D. Most biomarkers were investigated in a single cohort study or population. However, there were only a few biomarkers that may have causal effects on T2D. Further research is warranted to broadly evaluate the causal effects of multiple biomarkers on T2D and glycaemic traits using data from large-scale cohorts or combined GWAS including many different genetic variants. This genetic approach may advance our understanding of the causes of T2D and potentially enable us to explore novel targets for the prevention and treatment of diabetes.

Declaration of interest

The author declares that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This work was supported by The Netherlands Organization for Scientific Research project (NWO) and the Medical Research Council UK (grant number MC_UU_12015/1). A Abbasi is supported by a Rubicon grant from the NWO (Project no. 825.13.004).

References

 

An official journal of

European Society of Endocrinology Society for Endocrinology

Sections

Figures

References

1

KnowlerWCBarrett-ConnorEFowlerSEHammanRFLachinJMWalkerEANathanDMDiabetes Prevention Program Research G. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. New England Journal of Medicine2002346393403. (doi:10.1056/NEJMoa012512).

2

AbbasiAStolkRPBakkerSJ. Identification of relevant biomarkers for type 2 diabetes. Lancet. Diabetes & Endocrinology20142106107. (doi:10.1016/S2213-8587(14)70005-1).

3

LyonsTJBasuA. Biomarkers in diabetes: hemoglobin A1c, vascular and tissue markers. Translational Research2012159303312. (doi:10.1016/j.trsl.2012.01.009).

4

SandhuMSDebenhamSLBarrosoILoosRJ. Mendelian randomisation studies of type 2 diabetes: future prospects. Diabetologia200851211213. (doi:10.1007/s00125-007-0903-x).

5

NelsonCPHambySESaleheenDHopewellJCZengLAssimesTLKanoniSWillenborgCBurgessSAmouyelP. Genetically Determined Height and Coronary Artery Disease. New England Journal of Medicine201537216081618. (doi:10.1056/NEJMoa1404881).

6

VanderWeeleTJTchetgen TchetgenEJCornelisMKraftP. Methodological challenges in mendelian randomization. Epidemiology201425427435. (doi:10.1097/EDE.0000000000000081).

7

NikolopoulosGKBagosPGTsangarisITsiaraCGKopteridesPVaiopoulosAKapsimaliVBonovasSTsantesAE. The association between plasminogen activator inhibitor type 1 (PAI-1) levels, PAI-1 4G/5G polymorphism, and myocardial infarction: a Mendelian randomization meta-analysis. Clinical Chemistry and Laboratory Medicine201452937950. (doi:10.1515/cclm-2013-1124).

8

HaaseCLTybjaerg-HansenAQayyumAASchouJNordestgaardBGFrikke-SchmidtR. LCAT, HDL cholesterol and ischemic cardiovascular disease: a Mendelian randomization study of HDL cholesterol in 54,500 individuals. Journal of Clinical Endocrinology and Metabolism201297E248E256. (doi:10.1210/jc.2011-1846).

9

SmithGDEbrahimS. Mendelian randomization: prospects, potentials, and limitations. International Journal of Epidemiology2004333042. (doi:10.1093/ije/dyh132).

10

BurgessSScottRATimpsonNJDavey SmithGThompsonSGEPIC- InterAct Consortium. Using published data in Mendelian randomization: a blueprint for efficient identification of causal risk factors. European Journal of Epidemiology201530543552. (doi:10.1007/s10654-015-0011-z).

11

AbbasiADeetmanPECorpeleijnEGansevoortRTGansROHillegeHLvan der HarstPStolkRPNavisGAlizadehBZ. Bilirubin as a potential causal factor in type 2 diabetes risk: a mendelian randomization study. Diabetes20156414591469. (doi:10.2337/db14-0228).

12

LiberatiAAltmanDGTetzlaffJMulrowCGotzschePCIoannidisJPClarkeMDevereauxPJKleijnenJMoherD. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. British Medical Journal2009339b2700. (doi:10.1136/bmj.b2700).

13

MorrisAPVoightBFTeslovichTMFerreiraTSegreAVSteinthorsdottirVStrawbridgeRJKhanHGrallertHMahajanA. Large-scale association analysis provides insights into the genetic architecture and pathophysiology of type 2 diabetes. Nature Genetics201244981990. (doi:10.1038/ng.2383).

14

DupuisJLangenbergCProkopenkoISaxenaRSoranzoNJacksonAUWheelerEGlazerNLBouatia-NajiNGloynAL. New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nature Genetics201042105116. (doi:10.1038/ng.520).

15

BesselingJKasteleinJJDefescheJCHuttenBAHovinghGK. Association between familial hypercholesterolemia and prevalence of type 2 diabetes mellitus. Journal of the American Medical Association201531310291036. (doi:10.1001/jama.2015.1206).

16

BrunnerEJKivimakiMWitteDRLawlorDADavey SmithGCooperJAMillerMLoweGDRumleyACasasJP. Inflammation, insulin resistance, and diabetes--Mendelian randomization using CRP haplotypes points upstream. PLoS Medicine20085e155. (doi:10.1371/journal.pmed.0050155).

17

BuijsseBBoeingHHircheFWeikertCSchulzeMBGottschaldMKuhnTKatzkeVATeucherBDierkesJ. Plasma 25-hydroxyvitamin D and its genetic determinants in relation to incident type 2 diabetes: a prospective case-cohort study. European Journal of Epidemiology201328743752. (doi:10.1007/s10654-013-9844-5).

18

ChungCMLinTHChenJWLeuHBYinWHHoHYSheuSHTsaiWCChenJHLinSJ. Common quantitative trait locus downstream of RETN gene identified by genome-wide association study is associated with risk of type 2 diabetes mellitus in Han Chinese: a Mendelian randomization effect. Diabetes/Metabolism Research and Reviews201430232240. (doi:10.1002/dmrr.2481).

19

DastaniZHivertMFTimpsonNPerryJRYuanXScottRAHennemanPHeidIMKizerJRLyytikainenLP. Novel loci for adiponectin levels and their influence on type 2 diabetes and metabolic traits: a multi-ethnic meta-analysis of 45,891 individuals. PLoS Genetics20128e1002607. (doi:10.1371/journal.pgen.1002607).

20

De SilvaNMFreathyRMPalmerTMDonnellyLALuanJGauntTLangenbergCWeedonMNShieldsBKnightBA. Mendelian randomization studies do not support a role for raised circulating triglyceride levels influencing type 2 diabetes, glucose levels, or insulin resistance. Diabetes20116010081018. (doi:10.2337/db10-1317).

21

DingELSongYMansonJEHunterDJLeeCCRifaiNBuringJEGazianoJMLiuS. Sex hormone-binding globulin and risk of type 2 diabetes in women and men. New England Journal of Medicine200936111521163. (doi:10.1056/NEJMoa0804381).

22

GanWGuanYWuQAnPZhuJLuLJingLYuYRuanSXieD. Association of TMPRSS6 polymorphisms with ferritin, hemoglobin, and type 2 diabetes risk in a Chinese Han population. American Journal of Clinical Nutrition201295626632. (doi:10.3945/ajcn.111.025684).

23

HerderCKloppNBaumertJMullerMKhuseyinovaNMeisingerCMartinSIlligTKoenigWThorandB. Effect of macrophage migration inhibitory factor (MIF) gene variants and MIF serum concentrations on the risk of type 2 diabetes: results from the MONICA/KORA Augsburg Case-Cohort Study, 1984-2002. Diabetologia200851276284. (doi:10.1007/s00125-007-0800-3).

24

HuangTRenJHuangJLiD. Association of homocysteine with type 2 diabetes: a meta-analysis implementing Mendelian randomization approach. BMC Genomics201314867. (doi:10.1186/1471-2164-14-867).

25

HuangTSunJChenYXieHXuDLiD. Associations of common variants in methionine metabolism pathway genes with plasma homocysteine and the risk of type 2 diabetes in Han Chinese. Journal of Nutrigenetics and Nutrigenomics201476374. (doi:10.1159/000365007).

26

Interleukin 1 Genetics CCardiometabolic effects of genetic upregulation of the interleukin 1 receptor antagonist: a Mendelian randomisation analysis. Lancet. Diabetes & Endocrinology20153243253. (doi:10.1016/S2213-8587(15)00034-0).

27

JensenMKBartzTMDjousseLKizerJRZiemanSJRimmEBSiscovickDSPsatyBMIxJHMukamalKJ. Genetically elevated fetuin-A levels, fasting glucose levels, and risk of type 2 diabetes: the cardiovascular health study. Diabetes Care20133631213127. (doi:10.2337/dc12-2323).

28

JordeRSchirmerHWilsgaardTJoakimsenRMMathiesenEBNjolstadILochenMLFigenschauYBergJPSvartbergJ. Polymorphisms related to the serum 25-hydroxyvitamin D level and risk of myocardial infarction, diabetes, cancer and mortality. The Tromso Study. PLoS ONE20127e37295. (doi:10.1371/journal.pone.0037295).

29

KamstrupPRNordestgaardBG. Lipoprotein(a) concentrations, isoform size, and risk of type 2 diabetes: a Mendelian randomisation study. Lancet. Diabetes & Endocrinology20131220227. (doi:10.1016/S2213-8587(13)70064-0).

30

KrogerJZietemannVEnzenbachCWeikertCJansenEHDoringFJoostHGBoeingHSchulzeMB. Erythrocyte membrane phospholipid fatty acids, desaturase activity, and dietary fatty acids in relation to risk of type 2 diabetes in the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study. American Journal of Clinical Nutrition201193127142. (doi:10.3945/ajcn.110.005447).

31

LeongARehmanWDastaniZGreenwoodCTimpsonNLangsetmoLBergerCMetastrokeFuLWongBY. The causal effect of vitamin D binding protein (DBP) levels on calcemic and cardiometabolic diseases: a Mendelian randomization study. PLoS Medicine201411e1001751. (doi:10.1371/journal.pmed.1001751).

32

MagnussonMWangTJClishCEngstromGNilssonPGersztenREMelanderO. Dimethylglycine deficiency and the development of diabetes mellitus. Diabetes20156430103016. (doi:10.2337/db14-1863).

33

PerryJRFerrucciLBandinelliSGuralnikJSembaRDRiceNMelzerDConsortiumDSaxenaRScottLJ. Circulating beta-carotene levels and type 2 diabetes-cause or effect?Diabetologia20095221172121. (doi:10.1007/s00125-009-1475-8).

34

PerryJRWeedonMNLangenbergCJacksonAULyssenkoVSparsoTThorleifssonGGrallertHFerrucciLMaggioM. Genetic evidence that raised sex hormone binding globulin (SHBG) levels reduce the risk of type 2 diabetes. Human Molecular Genetics201019535544. (doi:10.1093/hmg/ddp522).

35

PetersKEBeilbyJCadbyGWarringtonNMBruceDGDavisWADavisTMWiltshireSKnuimanMMcQuillanBM. A comprehensive investigation of variants in genes encoding adiponectin (ADIPOQ) and its receptors (ADIPOR1/R2), and their association with serum adiponectin, type 2 diabetes, insulin resistance and the metabolic syndrome. BMC Medical Genetics20131415. (doi:10.1186/1471-2350-14-15).

36

PfisterRBarnesDLubenRForouhiNGBochudMKhawKTWarehamNJLangenbergC. No evidence for a causal link between uric acid and type 2 diabetes: a Mendelian randomisation approach. Diabetologia20115425612569. (doi:10.1007/s00125-011-2235-0).

37

PfisterRSharpSLubenRWelshPBarrosoISalomaaVMeirhaegheAKhawKTSattarNLangenbergC. Mendelian randomization study of B-type natriuretic peptide and type 2 diabetes: evidence of causal association from population studies. PLoS Medicine20118e1001112. (doi:10.1371/journal.pmed.1001112).

38

SongYYeungELiuAVanderweeleTJChenLLuCLiuCSchistermanEFNingYZhangC. Pancreatic β-cell function and type 2 diabetes risk: quantify the causal effect using a Mendelian randomization approach based on meta-analyses. Human Molecular Genetics20122150105018. (doi:10.1093/hmg/dds339).

39

StrawbridgeRJDupuisJProkopenkoIBarkerAAhlqvistERybinDPetrieJRTraversMEBouatia-NajiNDimasAS. Genome-wide association identifies nine common variants associated with fasting proinsulin levels and provides new insights into the pathophysiology of type 2 diabetes. Diabetes20116026242634. (doi:10.2337/db11-0415).

40

TrombettaMBonettiSBoselliMLMiccoliRTrabettiEMalerbaGPignattiPFBonoraEDel PratoSBonadonnaRC. PPARG2 Pro12Ala and ADAMTS9 rs4607103 as "insulin resistance loci" and "insulin secretion loci" in Italian individuals. The GENFIEV study and the Verona Newly Diagnosed Type 2 Diabetes Study (VNDS) 4. Acta Diabetologica201350401408. (doi:10.1007/s00592-012-0443-9).

41

WangTTChenYJSunLLZhangSJZhouZYQiaoH. Affection of single-nucleotide polymorphisms in miR-27a, miR-124a, and miR-146a on susceptibility to type 2 diabetes mellitus in Chinese Han people. Chinese Medical Journal2015128533539. (doi:10.4103/0366-6999.151112).

42

YaghootkarHLaminaCScottRADastaniZHivertMFWarrenLLStancakovaABuxbaumSGLyytikainenLPHennemanP. Mendelian randomization studies do not support a causal role for reduced circulating adiponectin levels in insulin resistance and type 2 diabetes. Diabetes20136235893598. (doi:10.2337/db13-0128).

43

YaghootkarHScottRAWhiteCCZhangWSpeliotesEMunroePBEhretGBBisJCFoxCSWalkerM. Genetic evidence for a normal-weight "metabolically obese" phenotype linking insulin resistance, hypertension, coronary artery disease, and type 2 diabetes. Diabetes20146343694377. (doi:10.2337/db14-0318).

44

YeZHaycockPCGurdasaniDPomillaCBoekholdtSMTsimikasSKhawKTWarehamNJSandhuMSForouhiNG. The association between circulating lipoprotein(a) and type 2 diabetes: is it causal?Diabetes201463332342. (doi:10.2337/db13-1144).

45

YeZSharpSJBurgessSScottRAImamuraFInterActCLangenbergCWarehamNJForouhiNG. Association between circulating 25-hydroxyvitamin D and incident type 2 diabetes: a mendelian randomisation study. Lancet. Diabetes & Endocrinology201533542. (doi:10.1016/S2213-8587(14)70184-6).

46

YouNCChenBHSongYLuXChenYMansonJEKangMHowardBVMargolisKLCurbJD. A prospective study of leukocyte telomere length and risk of type 2 diabetes in postmenopausal women. Diabetes20126129983004. (doi:10.2337/db12-0241).

47

VoightBFScottLJSteinthorsdottirVMorrisAPDinaCWelchRPZegginiEHuthCAulchenkoYSThorleifssonG. Twelve type 2 diabetes susceptibility loci identified through large-scale association analysis. Nature Genetics201042579589. (doi:10.1038/ng.609).

48

Van CampenhoutAVan CampenhoutCLagrouARAbramsPMoorkensGVan GaalLManuel-y-KeenoyB. Impact of diabetes mellitus on the relationships between iron-, inflammatory- and oxidative stress status. Diabetes/Metabolism Research and Reviews200622444454. (doi:10.1002/dmrr.635).

49

SedlakTWSnyderSH. Messenger molecules and cell death: therapeutic implications. Journal of the American Medical Association20062958189. (doi:10.1001/jama.295.1.81).

50

MiyashitaKItohHTsujimotoHTamuraNFukunagaYSoneMYamaharaKTauraDInuzukaMSonoyamaT. Natriuretic peptides/cGMP/cGMP-dependent protein kinase cascades promote muscle mitochondrial biogenesis and prevent obesity. Diabetes20095828802892. (doi:10.2337/db09-0393).

51

CorpeleijnEFeskensEJJansenEHMensinkMSarisWHde BruinTWBlaakEE. Improvements in glucose tolerance and insulin sensitivity after lifestyle intervention are related to changes in serum fatty acid profile and desaturase activities: the SLIM study. Diabetologia20064923922401. (doi:10.1007/s00125-006-0383-4).

52

VessbyBGustafssonIBTengbladSBobergMAnderssonA. Desaturation and elongation of Fatty acids and insulin action. Annals of the New York Academy of Sciences2002967183195. (doi:10.1111/j.1749-6632.2002.tb04275.x).

53

RheeEPHoJEChenMHShenDChengSLarsonMGGhorbaniAShiXHeleniusITO'DonnellCJ. A genome-wide association study of the human metabolome in a community-based cohort. Cell Metabolism201318130143. (doi:10.1016/j.cmet.2013.06.013).

54

KonstantinovaSVTellGSVollsetSENygardOBleieOUelandPM. Divergent associations of plasma choline and betaine with components of metabolic syndrome in middle age and elderly men and women. Journal of Nutrition2008138914920.

55

HoJELarsonMGVasanRSGhorbaniAChengSRheeEPFlorezJCClishCBGersztenREWangTJ. Metabolite profiles during oral glucose challenge. Diabetes20136226892698. (doi:10.2337/db12-0754).

56

SheikhKCamejoGLanneBHalvarssonTLandergrenMROakesND. Beyond lipids, pharmacological PPARalpha activation has important effects on amino acid metabolism as studied in the rat. American Journal of Physiology. Endocrinology and Metabolism2007292E1157E1165. (doi:10.1152/ajpendo.00254.2006).

57

AhluwaliaTSAllinKHSandholtCHSparsoTHJorgensenMERoweMChristensenCBrandslundILauritzenTLinnebergA. Discovery of coding genetic variants influencing diabetes-related serum biomarkers and their impact on risk of type 2 diabetes. Journal of Clinical Endocrinology and Metabolism2015100E664E671. (doi:10.1210/jc.2014-3677).

58

KrajaATChasmanDINorthKEReinerAPYanekLRKilpelainenTOSmithJADehghanADupuisJJohnsonAD. Pleiotropic genes for metabolic syndrome and inflammation. Molecular Genetics and Metabolism2014112317338. (doi:10.1016/j.ymgme.2014.04.007).

59

VaezAJansenRPrinsBPHottengaJJde GeusEJBoomsmaDIPenninxBWNolteIMSniederHAlizadehBZ. In Silico Post Genome-Wide Association Studies Analysis of C-Reactive Protein Loci Suggests an Important Role for Interferons. Circulation. Cardiovascular Genetics20158487497. (doi:10.1161/CIRCGENETICS.114.000714).

Information

Cited By

PubMed

Google Scholar

Related Articles

Altmetrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 126 126 124
PDF Downloads 22 22 21