impact of type 2 diabetes susceptibility variants on...
Post on 12-Jun-2018
215 Views
Preview:
TRANSCRIPT
1
Impact of type 2 diabetes susceptibility variants on quantitative glycemic traits reveals
mechanistic heterogeneity
Running title: Physiological characterization of known type 2 diabetes loci
Antigone S. Dimas1,2
*, Vasiliki Lagou1,3
*, Adam Barker4*, Joshua W. Knowles
5*, Reedik
Mägi1,3,6
, Marie-France Hivert7,8
, Andrea Benazzo9, Denis Rybin
10, Anne U. Jackson
11,
Heather M. Stringham11
, Ci Song12,13
, Antje Fischer-Rosinsky14
, Trine Welløv Boesgaard15
,
Niels Grarup16
, Fahim A. Abbasi5, Themistocles L. Assimes
5, Ke Hao
17, Xia Yang
18, Cécile
Lecoeur19
, Inês Barroso20,21
, Lori L. Bonnycastle22
, Yvonne Böttcher23
, Suzannah
Bumpstead20
, Peter S. Chines22
, Michael R. Erdos22
, Jurgen Graessler24
, Peter Kovacs25
,
Mario A. Morken22
, Narisu Narisu22
, Felicity Payne20
, Alena Stancakova26
, Amy J. Swift22
,
Anke Tönjes23,27
, Stefan R. Bornstein24
, Stéphane Cauchi19
, Philippe Froguel19,28
, David
Meyre19,29
, Peter E.H. Schwarz24
, Hans-Ulrich Häring30
, Ulf Smith31
, Michael Boehnke11
,
Richard N. Bergman32
, Francis S. Collins22
, Karen L. Mohlke33
, Jaakko Tuomilehto34-36
,
Thomas Quertemous5, Lars Lind
37, Torben Hansen
16,38, Oluf Pedersen
16,39-41, Mark Walker
42,
Andreas F.H. Pfeiffer14,43
, Joachim Spranger14
, Michael Stumvoll23,27
, James B. Meigs8,44
,
Nicholas J. Wareham4, Johanna Kuusisto
26, Markku Laakso
26, Claudia Langenberg
4, Josée
Dupuis45,46
, Richard M. Watanabe*47
, Jose C. Florez*44,48,49
, Erik Ingelsson*1,12
, Mark I.
McCarthy*1,3,50
, Inga Prokopenko*1,3,28
on behalf of the MAGIC investigators
* These authors contributed equally to this work.
Correspondence should be addressed to:
Inga Prokopenko, MSc, PhD
Department of Genomics of Common Disease
School of Public Health
Imperial College London
Burlington Danes Building,
Hammersmith Hospital,
Du Cane Road, London,
W12 0NN, UK
Phone: +4420 759 46501
E-mail: i.prokopenko@imperial.ac.uk
Prof. Mark I. McCarthy
OCDEM, Churchill Hospital, University of Oxford
Old Road, Headington, OX3 7LJ
Email: mark.mccarthy@drl.ox.ac.uk
Phone: +44 1865 857298
Erik Ingelsson, MD, PhD, FAHA
Professor of Molecular Epidemiology
Department of Medical Sciences
Page 3 of 73 Diabetes
Diabetes Publish Ahead of Print, published online December 2, 2013
2
Molecular Epidemiology and Science for Life Laboratory
UCR/MTC
Dag Hammarskjölds väg 14B
Uppsala Science Park
SE-752 37 Uppsala
Sweden
Phone: +46-70-7569422
Fax: +46-18-515570
E-mail: erik.ingelsson@medsci.uu.se
Jose C. Florez, MD, PhD
Diabetes Unit / Center for Human Genetic Research
Simches Research Building - CPZN 5.250
Massachusetts General Hospital
185 Cambridge Street
Boston, MA 02114
Phone: 617.643.3308
Fax: 617.643-6630
Richard M. Watanabe, PhD
Departments of Preventive Medicine and Physiology & Biophysics
Diabetes and Obesity Research Institute of USC
Keck School of Medicine of USC
2250 Alcazar Street, CSC-204
Los Angeles, CA 90089-9073
Phone: 1-323-442-2053
Affiliations
1. Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, OX3
7BN, UK.
2. Alexander Fleming, Biomedical Sciences Research Center, 34 Fleming Street, Vari,
16672 Athens, Greece.
3. Oxford Centre for Diabetes, Endocrinology and Metabolism, University of Oxford,
UK, OX3 7LJ.
4. MRC Epidemiology Unit, Institute of Metabolic Science, Addenbrooke’s Hospital,
Cambridge, UK.
5. Department of Medicine and Cardiovascular Institute, Stanford University School of
Medicine, Stanford, CA 94305, USA.
6. Estonian Genome Center, University of Tartu, Tartu, 51010, Estonia.
7. Department of Medicine, Université de Sherbrooke, Sherbrooke (Quebec), Canada.
8. General Medicine Division, Massachusetts General Hospital, Boston,Massachusetts,
USA.
9. Department of Biology and Evolution, University of Ferrara, Ferrara, Italy.
10. Boston University Data Coordinating Center, Boston, Massachusetts, MA 02118,
USA.
11. Department of Biostatistics and Center for Statistical Genetics, University of
Michigan School of Public Health, Ann Arbor, Michigan 48109, USA.
Page 4 of 73Diabetes
3
12. Department of Medical Sciences, Molecular Epidemiology and Science for Life
Laboratory, Uppsala University, Uppsala, Sweden.
13. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet,
Stockholm, Sweden.
14. Charité-Universitätsmedizin Berlin, Department of Endocrinology and Metabolism.
15. Steno Diabetes Center, Gentofte, Denmark.
16. The Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of
Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark.
17. Department of Genetics and Genomic Sciences, Institute of Genomics and Multiscale
Biology, Mount Sinai School of Medicine, New York, NY 10029-6574, USA.
18. Department of Integrative Biology and Physiology, University of California, 610
Charles E. Young Dr. East, Los Angeles, CA 90095, USA.
19. CNRS UMR8199-Institute of Biology, Pasteur Institute, Lille 2-Droit et Santé
University, Lille, France.
20. Wellcome Trust Sanger Institute, Hinxton, CB10 1SA, UK.
21. University of Cambridge Metabolic Research Laboratories and NIHR Cambridge
Biomedical Research Centre, Level 4, Institute of Metabolic Science, Box 289,
Addenbrooke’s Hospital, Cambridge CB2 OQQ, UK.
22. Genome Technology Branch, National Human Genome Research Institute, Bethesda,
MD.
23. Leipzig University Medical Center, IFB AdiposityDiseases, Liebigstr. 21, 04103
Leipzig, Germany.
24. Department of Medicine III, Division of Prevention and Care of Diabetes, University
of Dresden, 01307 Dresden, Germany.
25. Interdisciplinary Center for Clinical Research (IZKF) Leipzig, Liebigstr. 21, 04103
Leipzig, Germany.
26. Department of Medicine, University of Eastern Finland and Kuopio University
Hospital, 70210 Kuopio, Finland.
27. Department of Medicine, University of Leipzig, Liebigstr. 18, 04103 Leipzig,
Germany.
28. Department of Genomics of common diseases, Imperial College London, London,
UK.
29. Department of Clinical Epidemiology & Biostatistics, McMaster University,
Hamilton, Canada.
30. Department of Internal Medicine, Division of Endocrinology, Diabetology, Vascular
Medicine, Nephrology and Clinical Chemistry, University of Tübingen, Tübingen,
Germany.
31. Lundberg Laboratory for Diabetes Research, Center of Excellence for Metabolic and
Cardiovascular Research, Department of Molecular and Clinical Medicine,
Sahlgrenska Academy, University of Gothenburg, SE-413 45 Gothenburg, Sweden.
32. Department of Physiology & Biophysics, Keck School of Medicine, University of
Southern California, Los Angeles, California 90033, USA.
33. Department of Genetics, University of North Carolina Chapel Hill, North Carolina
27599, USA.
34. Diabetes Prevention Unit, National Institute for Health and Welfare, 00271 Helsinki,
Finland.
35. Centre for Vascular Prevention, Danube-University Krems, 3500 Krems, Austria.
36. King Abdulaziz University, Jeddah 21589, Saudi Arabia.
Page 5 of 73 Diabetes
4
37. Department of Medical Sciences, Uppsala University, Akademiska sjukhuset,
Uppsala, Sweden.
38. Faculty of Health Sciences, University of Southern Denmark, Odense, Denmark.
39. Hagedorn Research Institute, Copenhagen, Denmark.
40. Institute of Biomedical Science, Faculty of Health Sciences, University of
Copenhagen, Copenhagen, Denmark.
41. Faculty of Health Sciences, University of Aarhus, Aarhus, Denmark.
42. Institute of Cellular Medicine, Newcastle University, UK.
43. German Institute of Human Nutrition, Department of Clinical Nutrition.
44. Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115,
USA.
45. Department of Biostatistics, Boston University School of Public Health, Boston,
Massachusetts, MA 02118, USA.
46. The National Heart, Lung, and Blood Institute’s Framingham Heart Study,
Framingham, Massachusetts, USA.
47. Departments of Preventive Medicine and Physiology & Biophysics, Keck School of
Medicine of USC, Los Angeles, CA 90033, USA.
48. Center for Human Genetic Research and Diabetes Research Center (Diabetes Unit),
Massachusetts General Hospital, Boston, Massachusetts 02114, USA.
49. Program in Medical and Population Genetics, Broad Institute, Cambridge,
Massachusetts 02142, USA.
50. Oxford NIHR Biomedical Research Centre, Churchill Hospital, Oxford OX3 7LJ,
UK.
Page 6 of 73Diabetes
5
ABSTRACT
Patients with established type 2 diabetes display both beta-cell dysfunction and insulin
resistance. To define fundamental processes leading to the diabetic state, we examined
the relationship between type 2 diabetes risk variants at 37 established susceptibility
loci and indices of proinsulin processing, insulin secretion and insulin sensitivity. We
included data from up to 58,614 non-diabetic subjects with basal measures, and 17,327
with dynamic measures. We employed additive genetic models with adjustment for sex,
age and BMI, followed by fixed-effects inverse variance meta-analyses. Cluster analyses
grouped risk loci into five major categories based on their relationship to these
continuous glycemic phenotypes. The first cluster (PPARG, KLF14, IRS1, GCKR) was
characterized by primary effects on insulin sensitivity. The second (MTNR1B, GCK)
featured risk alleles associated with reduced insulin secretion and fasting
hyperglycemia. ARAP1 constituted a third cluster characterized by defects in insulin
processing. A fourth cluster (including TCF7L2, SLC30A8, HHEX/IDE, CDKAL1,
CDKN2A/2B) was defined by loci influencing insulin processing and secretion without
detectable change in fasting glucose. The final group contained twenty risk loci with no
clear-cut associations to continuous glycemic traits. By assembling extensive data on
continuous glycemic traits, we have exposed the diverse mechanisms whereby type 2
diabetes risk variants impact disease predisposition.
Page 7 of 73 Diabetes
6
MAIN TEXT
Type 2 diabetes is a metabolic disorder characterized by impaired insulin secretion and
reduced sensitivity to the peripheral actions of insulin. Both genetic and environmental
factors contribute to the development of type 2 diabetes [1] but the fundamental mechanistic
defects contributing to the evolution of disease remain far from clear. Recently, genome-wide
association efforts have extended the number of loci robustly implicated in type 2 diabetes
risk to more than 60 [2-5]. Each of these loci contains sequence variants that are causal for
disease risk and elucidation of the mechanisms through which these operate has the potential
to reveal processes fundamental to disease pathogenesis.
To date, systematic review of the effects of disease risk variants on processes contributing to
the diabetic state has mostly been restricted to the examination of basal indices of beta-cell
function or insulin sensitivity [2, 3]. These studies have demonstrated that most, but not all,
of these loci exert their primary effects on disease risk through deficient insulin secretion
rather than insulin resistance [2, 4, 6].
Further dissection of these mechanisms requires more intensive phenotyping in risk allele
carriers and controls. In principle, such studies, particularly if performed in non-diabetic
individuals, can provide readouts of the status of various aspects of intermediary metabolism.
For example, disproportionately raised levels of circulating fasting proinsulin (PI), as
compared to those of fasting insulin (FI), reflect beta-cell stress and impairment in early
insulin processing. Fasting 32,33 split PI provides a more detailed assessment of the impact
of genetic variants on insulin processing [7]. Dynamic tests of insulin secretion following oral
and/or intravenous (IV) glucose administration can provide insights into early beta-cell
dysfunction and loss of early insulin release [8]. Studies of selected diabetes risk loci
conducted using more refined dynamic measures of insulin secretion and/or sensitivity have,
Page 8 of 73Diabetes
7
in the main, corroborated the broad inferences obtained from basal measures [9-13], but have
often been underpowered with respect to more detailed mechanistic insights. To date, the
most substantial effort to characterize the physiology of risk allele carriers involved a survey
of 19 loci defined by their primary associations with quantitative glycemic traits, such as
fasting glucose (FG) and FI, eight of which were also type 2 diabetes risk loci [14]. This
study underscored the heterogeneity of genetic effects on glucose homeostasis.
Here we expand upon that study to examine the effect of a total of 37 type 2 diabetes risk loci
across a broad range of quantitative measures of glycemic metabolism. To do this, within the
Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC), we have gathered
basal and dynamic testing data from 15 studies providing detailed measures of insulin
processing (fasting PI; 32,33 split PI), insulin secretion (insulinogenic index; acute insulin
response [AIR]), insulin sensitivity (indices derived from oral and IV tests of glucose-
stimulated insulin secretion [15]) and insulin clearance (C-peptide), and combined these with
existing published data for FG, FI and homeostasis model assessments (HOMA) data from
the MAGIC meta-analysis of fasting traits [16]. In doing so, we are able to demonstrate that
the mechanisms of action of these various disease risk loci can be grouped into a number of
specific categories.
Research Design and Methods
Contributing studies. Three partially overlapping collections of samples were used (Table 1,
Supplementary Table 1): (i) nine studies with detailed physiologic basal and/or dynamic
measures after oral glucose stimulation, from a total of 23,443 individuals; (ii) twenty-nine
studies, including up to 58,614 individuals, with fasting trait data available from the MAGIC
genome-wide meta-analysis [16]; and (iii) seven studies, including 4,180 subjects, with IV-
Page 9 of 73 Diabetes
8
derived measures of insulin sensitivity. All participants were adult, non-diabetic (diabetes
defined by clinical diagnosis, diabetes treatment or fasting plasma glucose levels ≥7.0
mmol/l), and of European ancestry. Individuals with impaired FG or impaired glucose
tolerance were maintained in the analyses. Subjects provided informed consent, and all
studies were approved by local ethics committees.
Phenotypes. We collected available data from participating studies for any of 14
intermediate quantitative glycemic phenotypes, grouped as follows: (i) fasting glycemic traits
including FG (Nmax=58,614), FI (Nmax=52,379), and derived HOMA indices of beta-cell
function (HOMA-B) and insulin resistance (HOMA-IR) (Nmax=50,908) [16]; (ii) oral glucose
tolerance test (OGTT)-derived measures including insulinogenic index (pmol/mmol,
N=11,268), and/or insulin sensitivity indices (ISIs, Supplementary Table 2) including
Belfiore ISI [17] (Nmax=10,348), Stumvoll ISI [18] (Nmax=10,239), Matsuda ISI [19]
(Nmax=10,364) and Gutt ISI [20] (Nmax=13,158); (iii) circulating levels of fasting intact PI
(pmol/l) adjusted for concomitant FI (pmol/l), measured in plasma or serum (Nmax=13,912);
(iv) IV measures (up to 4,180 individuals from seven studies) including M/I derived from
euglycemic hyperinsulinemic clamp (Nmax=2,626), SI from the frequently sampled IV glucose
tolerance test (FSIGT, Nmax=1,173), and steady state plasma glucose (SSPG) from the insulin
suppression test (Nmax=381); (v) AIR analysed by the incremental area under the insulin
curve from 0-10 minutes, (Nmax=1,135); (vi) C-peptide (Nmax=5,059) and 32,33 split PI
(pmol/l) adjusted for concomitant FI (pmol/l, Nmax=2,568), ((ii)-(vi) summarized in Table 1).
Given the wide range of sample sizes available for different traits, we divided these 14
phenotypes into two groups based on a sample size cut-off of 10,000. This resulted in ten
“principal” traits with data from >10,000 individuals ((i)-(iii) above) and four traits with data
on fewer individuals ((iv)-(vi)). We focused initial analyses on principal traits.
Page 10 of 73Diabetes
9
Trait normalisation. Insulinogenic index, ISIs, PI, AIR, C-peptide, and 32,33 split PI were
natural log transformed. IV insulin sensitivity measures (M/I, SI and SSPG) were z-score
transformed to enable meta-analysis as a single IV trait.
SNP definition and proxies. We included 37 of the 38 type 2 diabetes-associated loci
documented in the DIAGRAM consortium meta-analysis [2]. This subset of known risk
variants was selected given their relatively large effects on type 2 diabetes risk (earliest
discovered variants), identification in individuals of European descent, and good
representation on genome-wide association study (GWAS) or custom genotype panels within
the contributing studies. Variants at FTO were excluded given the well-documented primary
association with BMI, which mediates the effect of FTO on type 2 diabetes risk [21, 22].
None of the other 37 loci has evidence for primary BMI associations. At these 37 loci, we
included data for the lead SNP and a total of 126 alternative proxy SNPs. In studies where
data for the lead SNP were not available, we chose the best proxy SNP for each locus on a
study-specific basis, using r2 measures from CEU HapMap [23] (Supplementary Table 3).
QC and exclusion criteria were as previously described [14] (Supplementary Note,
Supplementary Table 3). Not all samples had called genotypes on the sex chromosomes, and
as a result the maximum sample size for chromosome X locus DUSP9 was 7,642 individuals
(i.e. <10,000). We thus excluded this locus from the main analyses.
Statistical analysis. Linear regression was performed to test for association, under an
additive genetic model, between SNPs and quantitative glycemic traits adjusting for age, sex,
and BMI within each cohort. Cohort-specific effect estimates and standard errors derived
from the regression models were then combined in an inverse variance-weighted fixed effects
meta-analysis using GWAMA [24] or METAL [25]. Association P values are reported
without correction for multiple testing. Two-sided P values<0.05 were considered significant
Page 11 of 73 Diabetes
10
given high prior probabilities for association of established type 2 diabetes risk loci (reported
previously at genome-wide significance level P<5x10-8
) with glycemic traits [2, 4, 26, 27].
To investigate the impact of risk variants on physiologic traits, for selected trait pairs, we
plotted the standardised beta-coefficient estimates of the effects to account for differences in
trait transformations and power of individual meta-analyses.
Cluster analysis of physiologic traits and type 2 diabetes loci. To explore the
physiological basis of type 2 diabetes associations, we performed a primary cluster analysis
using principal traits only, and a subsidiary analysis that included all 14 traits. We also
performed cluster analysis of the 36 loci (excluding DUSP9) which grouped their effects on
principal traits. We used meta-analysis z-scores to perform complete linkage hierarchical
clustering and aligned all effects to the disease risk-increasing allele. In this type of cluster
analysis, the distance between two clusters is computed as the maximum distance between a
pair of traits/SNPs that map in separate clusters [28]. Locus clusters were defined by L2, a
Euclidean distance dissimilarity measure. The uncertainty of hierarchical clustering was
evaluated via multiscale bootstrap resampling [29]. Ten thousand bootstrap replicates were
generated to compute a probability for the strength of support for each dendrogram node, and
to evaluate topology sensitivity to sample size for each phenotype. We subsequently
performed a centroid-based clustering analysis to identify the most supported number of
clusters, where the full set of SNPs could be structured. In this clustering method orthogonal
transformation results in a reduced set of observations for each locus, translating ten
phenotypes to two linearly uncorrelated principal components. Dendrograms were created
where markers were forced into k groups (from two to eight), and the Calinski index [30] was
computed as a measure of clustering support. We then performed principal component
analysis (PCA) to centroid-based clustering results to visualise graphically the assignment of
SNPs to inferred clusters.
Page 12 of 73Diabetes
11
Results
Association meta-analysis. After excluding DUSP9, we first examined the pattern of trait
associations across the 36 remaining loci. First, we highlight those specific trait-SNP
associations with the strongest statistical support (Table 2, Supplementary Tables 4 and 5).
The strongest association was seen between type 2 diabetes risk alleles at the HHEX/IDE
locus and reduced insulinogenic index (P=3.6×10-21
, Supplementary Fig. 1A). We also
confirmed to genome-wide levels of significance, associations between: (i) 32,33 split PI and
ARAP1; and (ii) insulinogenic index and MTNR1B, in line with previous findings in partially-
overlapping data sets [7, 14, 31-35] and in addition to the established overlapping
associations with basal FG, FI and PI. Furthermore, we uncovered strong associations (which
here we define as P<5×10-5
) between: (i) insulinogenic index and CDKAL1 (Supplementary
Fig. 1B); (ii) 32,33 split PI and HNF1A (Supplementary Fig. 1C); and (iii) AIR and
MTNR1B, KCNQ1 and CDKN2A/B (Supplementary Fig. 1D-F). For all variants showing at
least nominal association (P<0.05) with AIR, the diabetes risk allele reduced AIR. In the joint
analysis of samples with IV-derived indices of insulin sensitivity, nominally significant
associations were observed between the disease risk allele and reduced insulin sensitivity for
the IRS1 (P=7×10-4
) and ADCY5 (P=6×10-3
) loci (Table 2, Supplementary Fig. 2A and B).
Cluster analyses. Complete linkage cluster analysis focusing on the structure of trait
relationships (Supplementary Fig. 3 principal traits; Supplementary Fig. 4 all traits)
confirmed that traits grouped in a meaningful manner consistent with expected physiological
relationships. We observed, for example, grouping of the four ISI measures, and of HOMA-B
and insulinogenic index. The Calinski index, which determines best partitioning and optimal
number of these trait clusters, was maximal at eight, indicating no large sub-clusters
(Supplementary Fig. 3B).
Page 13 of 73 Diabetes
12
We next re-clustered data, using principal traits only, to define relationships amongst the 36
type 2 diabetes loci (Fig. 1A). The unimodal distribution of the Calinski index was maximal
at five, with the same number of groups emerging from an alternative clustering analysis
based on centroids (Supplementary Fig. 5). Group assignment for the optimal Nclusters
determined using the Calinski index was coherent with results from linkage clustering. On
this basis, we defined four locus clusters with distinctive phenotypic features (Fig. 1B) which
we term: insulin resistance (IR); hyperglycemic (HG); proinsulin processing (PI); and beta
cell (BC). The twenty remaining loci not included in these phenotypic clusters we describe as
forming an unclassified group (UC). In bootstrap resampling (Fig. 1A), the baseline
branching nodes were highly supported (strength P≥0.84, with best support for a node
defined as Pmax=1.00 and absence of support Pmin=0) with the only exception of the PI group
where there was slightly less evidence of separation from the BC-UC clade (strength
P=0.64).
Four clusters of type 2 diabetes risk loci with distinctive phenotypic features. Here we
describe the features of each of the four clusters. To visualise some of these groupings, Fig. 2
and Supplementary Fig. 6 present selected scatterplots of physiological trait pairs.
Association P values reported in the text are selected for illustrative purposes with full data
detailed in Table 2 and Supplementary Tables 4 and 5.
The IR cluster contains four loci (IRS1, GCKR, PPARG, KLF14) and is characterized by the
association between type 2 diabetes risk alleles and reduced insulin sensitivity, as evidenced
by higher HOMA-IR (IRS1 [P=7×10-10
], GCKR [P=4×10-16
], PPARG [P=4×10-6
], KLF14
[P=7×10-7
]) and by reduced ISIs, with the strongest associations seen for decreased ISI-
Matsuda (IRS1 [P=2×10-7
], GCKR [P=0.008], PPARG [P=0.004]). The IR loci also tended to
show elevation of FG, FI, PI (IRS1 [P=0.006], GCKR [P=0.050]) and C-peptide levels (IRS1
Page 14 of 73Diabetes
13
[P=0.01], GCKR [P=0.008]). We observed some evidence that carriers of risk alleles at these
loci had higher values for HOMA-B (IRS1 [P=3×10-7
]) and insulinogenic index (IRS1
[P=4×10-4
], Fig. 2A, Supplementary Fig. 6A), which we interpret as likely to reflect, at least
in truncated ascertainment amongst non-diabetic individuals (see Discussion). Given access
to data from the largest set of individuals yet characterized for IV indices of insulin
sensitivity (Nmax=4,180), we specifically evaluated the effects of IR loci on these measures.
We observed an association between the disease risk variant at IRS1 and reduced IV
measures of insulin sensitivity (P=7×10-4
, Table 2) but no nominally significant associations
were seen at GCKR, PPARG or KLF14. Dropping either FI or HOMA-IR measures from the
cluster analyses (these traits are highly correlated) had only subtle effects on cluster
definitions and relationships: for example in analyses that retain FI but exclude HOMA-IR,
three loci, KCNQ1 [rs231362], JAZF1 and HMGA2 moved from the UC group to the IR
cluster. Adjustment for BMI improved estimates of both basal and dynamic measures of
insulin sensitivity (data not shown).
The HG cluster comprises the risk loci mapping near MTNR1B and GCK. Type 2 diabetes
risk alleles at these loci are associated with markedly reduced insulin secretion, combined
with fasting hyperglycemia, even in non-diabetic individuals (Fig. 2B). The insulin secretory
defect is manifest in reduced HOMA-B (MTNR1B [P=4×10-26], GCK [P=1×10
-11]),
insulinogenic index (MTNR1B [P=1×10-14
], GCK [P=0.03]) (Fig. 2B), and AIR (MTNR1B
[P=4×10-6
], GCK [P=0.051]). There was some evidence for inverse associations with insulin
sensitivity, with type 2 diabetes risk alleles associated with higher HOMA-IR (MTNR1B
[P=0.002], GCK [P=4×10-4
]) (Fig. 2C), but no consistent associations were found with ISIs
or IV measures of insulin sensitivity.
Page 15 of 73 Diabetes
14
The PI cluster actually consists of a single locus, ARAP1 (also named CENTD2). The key
feature of this locus is that the type 2 diabetes risk allele is associated with lower levels of
intact PI (P=2×10-50
) and 32,33 split PI (P=1x10-8
) (Supplementary Fig. 6B and C),
combined with reduced insulin secretion (insulinogenic index [P=2×10-5
], HOMA-B
[P=3x10-6
]) (Supplementary Fig. 6B and D) [7]. Insulin sensitivity was also slightly reduced,
as shown by lower ISIs, Matsuda and Gutt ISIs being nominally significant (P=0.033 and
0.050, respectively). Although excluded from the cluster analysis due to smaller sample size,
risk alleles at DUSP9 were also associated with lower PI values (P=0.035, Table 2) and
reduced insulinogenic index (P=3×10-4
), indicating that this locus may share some
mechanistic overlap with ARAP1. However, in contrast to ARAP1, the risk allele at DUSP9
was associated with increased insulin sensitivity (e.g. Matsuda ISI, P=9×10-4
).
The nine loci in the BC cluster (TCF7L2, SLC30A8, HHEX/IDE, CDKAL1, CDKN2A/2B,
THADA, DGKB, PROX1, ADCY5) are characterised by reduced insulin secretion in the
presence of increased PI and, in comparison with the HG cluster, only modest increases in
fasting glycemia. For all nine loci, we observed associations with lower HOMA-B (Fig. 2C),
higher FG, and lower FI (Table 2). The strongest associations were seen at TCF7L2, but all
loci also showed similar patterns of reduced insulinogenic index (Fig. 2B) and AIR
(Supplementary Fig. 6E), as well as increased PI (Supplementary Fig. 6C). There was little
systematic evidence for effects on HOMA or IV measures of insulin sensitivity
(Supplementary Fig. 6F), though the type 2 diabetes risk allele at ADCY5 displayed reduced
sensitivity on IV measures (P=6×10-3
, Supplementary Fig. 6F).
The UC group includes 20 loci for which, despite large sample sizes, detailed phenotyping
and established effects on risk for type 2 diabetes, no systematic evidence of association with
basal or dynamic glycemic phenotypes was detected. The only nominal associations observed
Page 16 of 73Diabetes
15
were at HNF1B (TCF2) (lower FI [P=8×10-4
], Table 2) and HNF1A (lower 32,33 split PI
[P=8×10-6
], insulinogenic index [P=0.005], Table 2).
We observed no material difference in the overall picture of trait associations when we
compared results from this study, with those previously described by Ingelsson et al. [14] in
partially overlapping samples, at the eight loci examined in both.
Discussion
We present here the most comprehensive report to date concerning the role of type 2 diabetes
risk loci on physiologic trait variability in non-diabetic individuals. In addition, this study
includes the largest number of individuals phenotyped using “gold standard” IV-derived
measures of insulin sensitivity. Through cluster analysis, we have shown that, in terms of
diabetes risk allele effect size on intermediate glycemic traits, type 2 diabetes risk loci fall
into five broad categories.
The IR cluster loci, including those mapping near PPARG, KLF14, IRS1 and GCKR, are
characterized by consistent effects across multiple indices of insulin sensitivity, in both the
basal and stimulated state. Three of the four IR loci (PPARG excluded) ranked in the lower
half of type 2 diabetes risk loci in terms of effect size: this may indicate that some of the loci
in the UC group act through weak effects on insulin sensitivity (as has indeed been suggested
for HMGA2) [2]. The observation that disease risk alleles at IR loci are associated with some
degree of improved beta-cell function (such as measured by HOMA-B) is likely to reflect a
combination of complementary factors. An increase in insulin secretion as a result of higher
insulin resistance is a known physiological phenomenon, which occurs until individuals
decompensate and shift towards development of diabetes. Therefore, alleles that primarily
raise insulin resistance may be secondarily associated with improved insulin secretion
Page 17 of 73 Diabetes
16
through a compensatory mechanism. In addition, the apparent association with insulin
sensitivity may in part reflect the fact that analysis was restricted to non-diabetic subjects:
this can lead to a truncation effect, whereby individuals with genotypes that favor both high
insulin resistance and poor insulin secretion are preferentially depleted from the analysis due
to their diabetes predisposition [6].
The PI cluster is limited to the type 2 diabetes risk variant at ARAP1, the cardinal feature
being a marked reduction in fasting PI levels in carriers of the disease risk allele. The
combination of diabetes risk reduced basal and stimulated insulin secretion and reduced PI
levels runs counter to the usual epidemiological associations [7]. Our data therefore support
previous assertions that the ARAP1 risk variant increases risk of type 2 diabetes through
defects in early steps of insulin production [7].
The HG cluster includes loci characterised by a prominent reduction in basal and stimulated
beta-cell function [35-37], resulting in a marked increase in fasting glucose levels. This
glycemic effect is consistent with the fact that the type 2 diabetes risk alleles at these loci are
associated with modest reductions in basal insulin sensitivity (as measured by HOMA-IR).
These associations with insulin action phenotypes did not however extend to the stimulated
measures, and may simply reflect limitations in the HOMA model. This cluster includes both
GCK and MTNR1B, and whilst they share similar features in terms of their effects on FG,
HOMA-B, and insulin/glucagon ratios [11], external data suggest that these genes act through
different mechanisms. In individuals with MODY2 [38], severe loss-of-function mutations in
GCK lead to impaired glucose sensing, and a higher homeostatic set point for glucose. At
MTNR1B, several studies have shown that lower insulin secretion is accompanied by
markedly reduced insulin sensitivity [11] and increased insulin response to GLP-1 [39]. The
Page 18 of 73Diabetes
17
apparent lack of effects on PI measures for MTNR1B, as for GCK, points to an effect on the
secretory function of beta-cells, rather than on PI processing [14].
The BC cluster includes a number of loci associated with defective insulin secretion but
without the marked glycemic effects seen in the HG cluster. These include several of the loci
with the strongest allelic effects on diabetes risk, such as those at TCF7L2, SLC30A8,
CDKAL1 and CDKN2A/2B, suggesting that the distinction from the loci in HG cluster is not
simply a consequence of the degree of allelic impact. Although the BC loci map to a single
cluster, there is a degree of heterogeneity with respect to other traits. For example, disease
risk alleles at TCF7L2 and SLC30A8 are associated with increased PI, reduced insulinogenic
index and lower FI levels, underscoring defects in insulin processing and secretion [7]. In
contrast, the risk alleles at HHEX/IDE, DGKB, CDKAL1, PROX1 and CDKN2A/2B display
reduced insulinogenic index and AIR, with no effect on fasting PI, suggesting defects during
first phase insulin response and early insulin secretion. These phenotypic associations within
the BC cluster highlight the potential for further subdivision of loci according to
pathophysiological patterns.
The UC group combines all remaining loci, which although genome-wide significant for type
2 diabetes, display no discernible impact on glycemic measures across the large sample sets
assembled here. Most of these loci rank in the lower half of signals in terms of type 2
diabetes effect size and modest functional impact may offer a partial explanation for this
observation. However, it is notable that the UC group includes many of the loci where the
common variant effects are highly likely to be mediated by transcripts implicated in
monogenic and syndromic forms of diabetes (including HNF1A, HNF1B [40], KCNJ11 [41]
and WFS1 [42]). Thus, even in the setting where rare coding mutations result in severe
abrogation of islet function, and common variants acting through those same genes influence
Page 19 of 73 Diabetes
18
type 2 diabetes risk to stringent levels of significance, those same common variants can also
be compatible with normal glucose homeostasis [2].
This study combined information from a large number of individuals phenotyped using IV
measures of insulin sensitivity, including over 2,600 subjects examined using the “gold
standard” euglycemic hyperinsulinemic clamp. Overall, the data do not support the view that
the combination of IV measures used in this study offers sufficient boost in precision to
compensate for the reduction in sample size when compared to the numbers available for the
basal or OGTT-derived measures. Except at IRS1 and ADCY5, neither the IV measures nor
the OGTT-derived ISIs generated powerful signals of association with deficient insulin
action, even at loci where the existing evidence for an effect on insulin sensitivity is
compelling (e.g. PPARG, KLF14 [2, 43]).
The study presented here is the largest investigation of its kind published to date: we set out,
to maximise the sample size for each trait included. One consequence of this strategy is that
the characteristics of the particular samples informative for each trait may differ, raising the
possibility of artefacts when comparing genotypic associations across traits. However, the
extent to which our findings are both internally consistent, and in broad agreement with
published data describing more detailed phenotypic studies of individual variants, provides
considerable reassurance. An alternative strategy, which restricted analyses to a small core of
individuals with data available for all traits, would have resulted in a substantial loss of
power.
We recognise some other limitations of this study, and of others of its kind. First, the
complexity of the metabolic phenotypes examined, the longitudinal dimension of diabetes
development and progression, and the relative imprecision of the experimental tools at our
disposal, limit the inferences that are possible. This is most obviously seen in the large
Page 20 of 73Diabetes
19
numbers of loci, which though genome-wide significant for type 2 diabetes, show no
evidence of relationship with intermediate metabolic traits. We also note, for example, that
the measures of early insulin secretion employed (insulinogenic index, AIR) may, in part,
reflect insulin clearance, a trait that was not evaluated directly. Second, in the interests of
avoiding the secondary effects of diabetes and its treatment, analyses were restricted to non-
diabetic individuals: this is likely to have introduced some truncation effects such as the
apparent enhancement of beta-cell function for disease risk alleles disrupting insulin
sensitivity. These have the potential to lead to incorrect inference, if not recognised as such
[6, 44]. Third, the variants we have examined are not, in most cases, known to be causal for
the association signals detected, leaving open the possibility that, at some loci, the causal
alleles, once identified, may have more pronounced effects on intermediate traits than the
variants studied here. Finally additional recently published disease associations [3] of variants
with lesser genetic effects were not considered. Future efforts to extend our study to
incorporate these variants will likely require substantially larger sample sizes.
In summary, in this examination of the intermediate metabolic phenotypic associations of
proven type 2 diabetes risk alleles, we have demonstrated that the loci fall into a limited
number of broad mechanistic categories. These highlight the diverse mechanisms
contributing to individual risk of disease. These data will guide future experimental studies
which use more specific, tailored tests to further dissect these key pathogenetic processes.
Page 21 of 73 Diabetes
20
Table 1. Studies and sample sizes of physiologic glycemic traits studied.
Study
Number of
loci
covered
Number of individuals included in analysis by study
Insulinogenic
Index
ISI
Belfiore
ISI
Stumvoll
ISI
Matsuda
ISI
Gutt Proinsulin M/I SI SSPG AIR
C-
peptide
32,33 Split
Proinsulin
ELY 31 1,435 1,441 1,425 1,311 1,476 1,600 - - - - 1,602 1,598
EUGENE2 36 - - - - - - 595 - - - - -
FHS 35 - - - - 2,599 5,752 - - - - - -
FUSION 10 - - - - - - - 538 - 557 925 -
MESYBEPO 30 883 887 894 885 1,060 - - - - - - -
METSIM 30 6,847 6,898 6,794 6,898 6,933 5,076 - - - - - -
NHANES 12 - - - - - - - - - - 1,221 -
PIVUS 29 - - - - - 911 - - - - - -
QTL FAMILIES 34 262 265 267 274 268 - - 261 - 214 274 -
RISC 36 1,015 - - - - - 1,042 - - - - -
PARTNERS/ROCHE 7 583 608 600 619 612 - - - - - - -
SORBS 36 756 758 765 758 763 - - - - - 718 -
STANFORD 36 - - - - - - - - 381 - - -
ULSAM 29 978 987 976 984 989 979 989 - - - - 979
UNG92 35 - - - - - - - 374 - 373 374 -
Total N * 11,268 10,348 10,239 10,364 13,158 13,912 2,626 1,173 381 1,135 5,059 2,568
All data have been adjusted for age, sex and BMI.
* Total number of individuals with phenotype, non-missing covariates and genetic data available.
Page 22 of 73Diabetes
21
Table 2. Effects of 37 SNPs previously associated with type 2 diabetes on physiologic glycemic measures.
Principal traits
Effectsignificance
Other traits†
Effectsignificance
Dynamic traits Fasting traits
Nearest gene (SNP) Alleles
(effect/
other)
Proinsulin
(n=13,912)
Insulinoge
nic Index
(n=11,268)
ISI
Belfiore
(n=10,348)
ISI
Stumvoll
(n=10,239)
ISI
Matsuda
(n=10,364)
ISI
Gutt
(n=13,158)
Fasting
Glucose^
(n=58,614)
Fasting
Insulin^
(n=52,379)
HOMA-
B^
(n=50,908)
HOMA-
IR^
(n=50,908)
M/I, SI, SSPG
, (n=4,180)
Insulin Resistance (IR)
PPARG (rs13081389) A/G 0.008 0.009 -0.019* -0.021* -0.035* -0.015* 0.010 0.025*** 0.017** 0.025*** -0.025
IRS1 (rs7578326) A/G 0.014* 0.037** -0.023*** -0.013* -0.040*** -0.020*** 0.005 0.019**** 0.014*** 0.019**** -0.070***
GCKR (rs780094) C/T 0.010* 0.009 -0.008 -0.015* -0.019* 0.007 0.033**** 0.019**** 0.006* 0.023**** -0.019
KLF14 (rs972283) G/A 0.004 -0.002 -0.003 -0.008 -0.001 -0.008 0.007* 0.013*** 0.009** 0.014*** -0.002
Hyperglycemic (HG)
MTNR1B (rs10830963) G/C 0.009 -0.083**** 0.006 -0.007 -0.004 -0.006 0.079**** 4x10-4 -0.033**** 0.011* 0.012
GCK (rs4607517) A/G -7x10-4 -0.036* -0.005 -0.015 -0.008 -0.016* 0.065**** 0.006 -0.024**** 0.014** 0.015
Proinsulin (PI)
DUSP9 (rs5945326) A/G -0.013* -0.037* 0.012* 0.010 0.022** 0.006 0.011 -0.013 -2.890 * -0.066 0.010
ARAP1 (rs1552224) A/C -0.093**** -0.054** -4x10-4 -0.013* -0.013 -0.010* 0.024**** -0.007* -0.015*** -0.005 -0.010
Beta Cell (BC)
TCF7L2 (rs7903146) T/C 0.044**** -0.063*** 0.009 -0.009 0.010 -0.017** 0.026**** -0.014*** -0.019**** -0.010* 0.014
CDKAL1 (rs10440833) A/T 0.008 -0.102*** 0.005 -0.011 -0.010 -0.001 0.014** -0.006* -0.009* -0.004 0.047*
CDKN2A/B (rs10965250) G/A -4x10-4 -0.045* 0.013* -7x10-4 0.012 0.005 0.017** -0.005 -0.012** -0.002 0.047
THADA (rs11899863) C/T 0.015 0.008 0.010 -0.004 0.016 -0.007 0.020** -0.010* -0.020** -0.008 -0.053
HHEX/IDE (rs5015480) C/T 0.006 -0.102**** 0.020*** -0.010 0.017* -0.005 0.010* -0.001 -0.005* 0.001 -0.004
SLC30A8 (rs3802177) G/A 0.038**** -0.041** 0.001 -0.004 -0.001 -0.011* 0.033**** -0.003 -0.016*** 2x10-4 -0.014
ADCY5 (rs11708067) A/G 0.016* -0.009 -0.008 0.001 -0.003 -0.014* 0.022*** -0.006 -0.015*** -0.001 -0.063*
PROX1 (rs340874) C/T 0.009 -0.028* 0.008* 0.008 0.009 -0.001 0.017*** -0.002 -0.009** -3x10-4 -0.002
DGKB/TMEM195 (rs2191349) T/G -0.001 -0.028* 0.011* 0.007 0.027** 0.006 0.032**** -0.002 -0.017**** 0.002 -5x10-4
Unclassified (UC)
CHCHD9 (rs13292136) C/T -5x10-4 -0.025 0.012 -0.001 0.026* 0.006 -0.003 -0.005 -0.002 -0.005 0.034
HMGA2 (rs1531343) C/G 0.010 0.030 -0.013 0.002 -0.027* 0.001 0.013* 0.007 -0.001 0.011* -0.034
HNF1B (TCF2) (rs4430796) G/A -5x10-4 -0.044 0.007 -0.011 0.008 -0.013 -0.001 -0.012** -0.011** -0.012** 0.025
Page 23 of 73 Diabetes
22
IGF2BP2 (rs1470579) C/A -0.011 -0.038* 0.013 0.010 0.010 -8x10-4 0.016*** -0.001 -0.006* 1x10-4 0.013
JAZF1 (rs849134) A/G 0.003 0.013 -0.005 -0.005 -0.014 -0.006 0.007* -1x10-4 -0.001 -2x10-4 0.015
WFS1 (rs1801214) T/C -0.019** -0.017 0.002 -8x10-4 0.008 -0.008 0.008* -0.001 -0.005* 3x10-4 -0.016
TSPAN8/LGR5 (rs4760790) A/G 0.006 -0.016 0.001 0.010 -0.003 -4x10-4 0.004 -5x10-4 -0.002 5x10-4 0.016
ZBED3 (rs4457053) G/A -0.008 -0.014 4x10-4 0.002 4x10-4 -0.004 0.019*** 0.002 -0.005 0.003 -0.008
BCL11A (rs243021) A/G 0.007 0.018 0.004 0.008 0.013 0.009* 0.004 0.001 3x10-4 0.000 -0.006
KCNQ1 (rs163184) G/T 0.016 -0.011 0.002 0.009 0.009 -0.004 0.012** 0.002 -0.004 0.004 0.015
HNF1A (rs7957197) T/A -0.011 -0.036* 0.009 -0.006 -0.003 5x10-4 -0.001 -0.004 -0.003 -0.004 0.005
NOTCH2 (rs10923931) T/G -1x10-5 -0.012 0.011 0.003 0.019 0.001 0.002 -0.003 -0.004 -0.001 0.038
CDC123/CAMK1D (rs12779790) G/A -0.008 -0.029* 0.001 -0.002 -0.006 0.002 0.014** -0.002 -0.007* -0.001 0.020
KCNQ1 (rs231362) G/A -0.004 0.005 -8x10-5 -0.004 -0.003 -0.005 0.017*** 0.007* -0.001 0.007* -0.021
ADAMTS9 (rs6795735) C/T 6x10-4 0.009 0.006 0.003 0.011 0.001 0.008* 0.002 -0.001 0.003 -0.004
KCNJ11 (rs5215) C/T 0.001 -0.024 0.002 -0.015 9x10-4 0.004 -0.004 -0.001 0.003 -0.002 -0.018
PRC1 (rs8042680) A/C -0.008 -0.016 0.003 0.007 0.011 -0.002 0.011* -0.002 -0.004 -3x10-4 0.003
TP53INP1 (rs896854) T/C -0.003 -0.003 -0.001 -0.006 -0.002 3x10-4 0.007* -0.002 -0.004 -0.001 0.007
ZFAND6 (rs11634397) G/A 0.008 -0.010 -0.001 -0.005 -0.007 -0.005 0.001 0.003 0.002 0.003 0.032
HCCA2 (rs2334499) T/C -0.017* -0.018 0.010 0.016 0.015 -0.006 0.001 -0.004 -0.004 -0.004 0.031
Within each cluster, loci were ordered by the previously described size of their effects on T2D risk. These risk estimates were derived from stage
2 metabochip data from Morris et al [3], with the exception of DUPS9 for which we used the OR reported in Voight et al [2] (DUSP9 was not
included in the Metabochip design). Effects represent beta-coefficients. Stars next to per allele effect correspond to P values equal or less than: *
0.05, ** 10-3
, *** 10-5
, **** 10-8
. ^ Fasting trait results have been previously reported in Manning et al. [16]. DUSP9 was grouped together with
ARAP1 as risk alleles at both loci were associated with lower PI values and reduced insulinogenic index. However, in contrast to ARAP1, the risk
allele at DUSP9 was associated with increased insulin sensitivity.
Page 24 of 73Diabetes
23
AUTHOR CONTRIBUTIONS
A.Ba., A.Be., A.S.D., J.F., M-F.H., E.I., J.K., V.L., C.La., M.I.M, R.M., I.P., R.W. wrote the
manuscript.
A.Ba., A.Be., A.S.D., J.D., V.L., C.La., M.I.M, R.M., I.P., D.R. carried out meta-analysis,
clustering, and higher level analysis of the data.
(EUGENE2) H-U.H., M.L., U.S.; (Ely) N.J.W.; (Framingham Heart Study) J.B.M.; (FUSION)
M.B., R.N.B., F.S.C., K.L.M., J.T., R.M.W.; (MeSyBePo) J.S., A.F.H.P.; (METSIM) J.K., M.L.;
(NHANES III)/(Partners/Roche) J.B.M.; (PIVUS) E.I., L.L.; (QTL/UNG92) T.H., O.P.; (RISC)
M.W.; (Sorbs) M.S.; (Stanford IST) T.Q.; (ULSAM) E.I. designed, managed and coordinated the
project.
(EUGENE2) H-U.H., M.L., U.S.; (Ely) N.J.W.; (Framingham Heart Study) J.B.M., J.C.F.;
(FUSION) R.N.B., J.T.; (MeSyBePo) J.S., A.F-R.; (METSIM) A.S., J.K., M.L.; (NHANES
III)/(Partners/Roche) J.B.M.; (PIVUS) E.I., L.L.; (QTL/UNG92) T.W.B., T.H.; (RISC) M.W.;
(Sorbs) A.T.; (Stanford IST) T.Q., F.A.A.; (ULSAM) E.I. conducted phenotyping.
(EUGENE2) H-U.H., M.L., U.S.; (Ely) I.B., S.B., F.P.; (Framingham Heart Study) J.D., J.B.M.;
(FUSION) L.L.B, M.R.E., A.J.S., N.N., M.A.M., P.S.C.; (M-A) M.I.M.; (METSIM) L.L.B, M.R.E.,
A.J.S., N.N., M.A.M., P.S.C.; (NHANES III)/(Partners/Roche) J.B.M.; (PIVUS) E.I.;
(QTL/UNG92) N.G.; (RISC) M.W.; (Sorbs) P.K., Y.B.; (Stanford IST) T.L.A., J.W.K, K.H., X.Y.;
(ULSAM) E.I. carried out genotyping.
(Ely) C.La., A.Ba.; (Framingham Heart Study) J.D., D.R., M-F.H., J.C.F.; (FUSION) A.U.J.,
H.M.S.; (M-A) A.S.D.; (MeSyBePo) J.S., A.F-R.; (METSIM) A.U.J., H.M.S.; (NHANES
III)/(Partners/Roche) J.B.M.; (PIVUS) E.I., C.S.; (QTL/UNG92) N.G., T.W.B.; (Sorbs) A.S.D.,
I.P., R.M., V.L.; (Stanford IST) T.L.A., J.W.K, F.A.A., K.H., X.Y.; (ULSAM) E.I., C.S. carried out
initial data analysis.
All authors have read and approved the manuscript.
A.S.D. and I.P are the guarantors of this work and, as such, had full access to all the data in the
study and take responsibility for the integrity of the data and the accuracy of the data analysis
Page 25 of 73 Diabetes
24
ACKNOWLEDGEMENTS
The major funding for this meta-analysis was provided through the European Community's Seventh
Framework Programme (FP7/2007-2013), ENGAGE project, grant agreement HEALTH-F4-2007-
201413.
Ely: I.B. acknowledges funding from the Wellcome Trust grant WT098051, United Kingdom NIHR
Cambridge Biomedical Research Centre and the MRC Centre for Obesity and Related Metabolic
Diseases.
EUGENE 2: EUGENE2 was supported by a grant from the European Community’s FP6 EUGENE2
no. LSHM-CT-2004-512013.
Framingham Heart Study: This research was conducted in part using data and resources from the
Framingham Heart Study of the National Heart Lung and Blood Institute of the National Institutes
of Health and Boston University School of Medicine. The analyses reflect intellectual input and
resource development from the Framingham Heart Study investigators participating in the SNP
Health Association Resource (SHARe) project. This work was partially supported by the National
Heart, Lung and Blood Institute's Framingham Heart Study (Contract No. N01-HC-25195) and its
contract with Affymetrix. Inc. for genotyping services (Contract No. N02-HL-6-4278). A portion of
this research utilized the Linux Cluster for Genetic Analysis (LinGA-II) funded by the Robert
Dawson Evans Endowment of the Department of Medicine at Boston University School of
Medicine and Boston Medical Center. Also supported by National Institute for Diabetes and
Digestive and Kidney Diseases (NIDDK) R01 DK078616 to J.B.M., J.D. and J.C.F., NIDDK K24
DK080140 to J.B.M., and a Doris Duke Charitable Foundation Clinical Scientist Development
Award to J.C.F.
FUSION: We would like to thank the many Finnish volunteers who generously participated in the
FUSION, D2D, Health 2000, Finrisk 1987, Finrisk 2002, and Savitaipale studies from which we
chose our FUSION GWA and replication cohorts (no overlap with Health 2000 replication cohort).
We also thank Terry Gliedt and Peggy White for informatics and analysis support. The Center for
Inherited Disease Research performed the GWA genotyping. Support for this study was provided
by the following: NIH grants DK069922 (R.M.W.), U54 DA021519 (R.M.W.), DK062370 (M.B.),
and DK072193 (K.L.M.). Additional support comes from the National Human Genome Research
Institute intramural project number 1 Z01 HG000024 (F.S.C.).
MESYBEPO: We are indebted to all subjects who participated to this study. This work was
supported by grants from "Agence Nationale de la Recherche" and "Conseil Regional Nord-Pas de
Calais/Fonds européen de développement économique et régional". We thank Marianne Deweider
and Fredéric Allegaert for the DNA bank management, Philippe Gallina, Philippe Delfosse and
Sylvie Poulain for the recruitment of most obese adults, "Department of Nutrition of Paris Hotel
Dieu Hospital" for the contribution to the recruitment effort and Stéphane Lobbens for genotyping.
METSIM: Support was provided by grant 124243 from the Academy of Finland.
MeSyBePo: J.S. was supported by a Clinical Research Group (KFO218/1) and a Heisenberg-
Professorship (SP716/2-1) of the Deutsche Forschungsgemeinschaft (DFG). J.S. was also funded by
a research group on molecular nutrition of the Bundesministerium für Bildung und Forschung
(BMBF).
Page 26 of 73Diabetes
25
NHANES: We thank Dr. Sekar Kathiresan at the Massachusetts General Hospital and Broad
Institute of Harvard and MIT for his collaboration in assembling the NHANES DNA samples, and
Jody E. McLean at the Division of Health and Nutrition Examination Surveys, National Center for
Health Statistics, Centers for Disease Control and Prevention, for her assistance with the analysis of
NHANES genetic data.
Partners/Roche: We thank Roche Pharmaceuticals for its support and collaboration in assembling
the Partners/Roche cohort. Supported by the Mallinckrodt General Clinical Research Program,
National Center for Research Resources, NIH RR01066.
PIVUS and ULSAM: Genotyping was performed by the SNP&SEQ Technology Platform in
Uppsala (www.genotyping.se). We thank Tomas Axelsson, Ann-Christine Wiman and Caisa
Pöntinen for their excellent assistance with genotyping. The SNP Technology Platform is supported
by Uppsala University, Uppsala University Hospital and the Swedish Research Council for
Infrastructures. E.I. was supported by grants from the Swedish Research Council, the Swedish
Heart-Lung Foundation, the Swedish Foundation for Strategic Research, and the Royal Swedish
Academy of Science.
QTL families and UNG92: Lundbeck Foundation Centre of Applied Medical Genomics for
Personalized Disease Prediction, Prevention and Care (LuCAMP), The Danish Diabetes
Association and The Danish Research Council.
RISC: The RISC Study is supported by European Union grant QLG1-CT-2001-01252 and
AstraZeneca.
Sorbs: This work was supported by grants from the German Research Council (SFB- 1052 “Obesity
mechanisms”), from the German Diabetes Association and from the DHFD (Diabetes Hilfs- und
Forschungsfonds Deutschland). P.K. is funded by the Boehringer Ingelheim Foundation. IFB
Adiposity Diseases is supported by the Federal Ministry of Education and Research (BMBF),
Germany, FKZ: 01EO1001.
We would like to thank Knut Krohn (Microarray Core Facility of the Interdisciplinary Centre for
Clinical Research, University of Leipzig) for the genotyping/analytical support and Joachim Thiery
(Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University of
Leipzig) for clinical chemistry services. We thank Nigel W. Rayner (WTCHG, University of
Oxford, UK) for the excellent bioinformatics support. I.P. and V.L. were partial funded through the
European Community's Seventh Framework Programme (FP7/2007-2013), ENGAGE project, grant
agreement HEALTH-F4-2007-201413. R.M. is funded by European Commission under the Marie
Curie Intra-European Fellowship and an EFSD New Horizons grant. M.I.M. is supported by
Wellcome Trust grant 098381.
Stanford: Sample collection for the Stanford IST cohort was carried out in part in the Clinical and
Translational Research Unit, Stanford University, with funds provided by the National Center for
Research Resources, 2 M01 RR000070, U.S. Public Health Service. J.W.K. is supported by an
AHA Fellow to Faculty Transition award, 10FTF3360005.
Page 27 of 73 Diabetes
26
CONFLICTS OF INTEREST
I.B. and spouse own stock in GlaxoSmithKline and Incyte. J.B.M. currently serves on a consultancy
board for Interleukin Genetics. R.M.W. has received consulting honoraria from Merck & Co.,
receives research funding from Merck & Co., and receives research material support from Takeda
Pharmaceuticals North America. J.C.F. has received consulting honoraria from Daiichi-Sankyo and
AstraZeneca. A.S.D., V.L., A.Ba., J.W.K., R.M., M-F.H., A.Be., D.R., A.U.J., H.M.S., C.S.,A.F-R.,
T.W.B., N.G., F.A.A., T.L.A., K.H., X.Y., C.Le., L.L.B., Y.B., S.B., P.S.C., M.R.E., J.G., P.K.,
M.A.M., N.N., F.P., A.S., A.J.S., A.T., S.R.B., S.C., P.F., D.M., P.E.H.S., H-U.H., U.S., M.B.,
R.N.B., F.S.C., K.L.M., J.T., T.Q., L.L., T.H., O.P., M.W., A.F.H.P., J.S., M.S., N.J.W., J.K., M.L.,
C.La., J.D., E.I., M.I.M., I.P. declare no conflict of interest.
Page 28 of 73Diabetes
27
REFERENCES
1. Stumvoll, M., B.J. Goldstein, and T.W. van Haeften, Type 2 diabetes: principles of
pathogenesis and therapy. Lancet, 2005. 365(9467): p. 1333-46.
2. Voight, B.F., et al., Twelve type 2 diabetes susceptibility loci identified through large-scale
association analysis. Nat Genet, 2010. 42(7): p. 579-89.
3. Morris, A.P., et al., Large-scale association analysis provides insights into the genetic
architecture and pathophysiology of type 2 diabetes. Nat Genet, 2012. 44(9): p. 981-990.
4. Dupuis, J., et al., New genetic loci implicated in fasting glucose homeostasis and their
impact on type 2 diabetes risk. Nat Genet, 2010. 42(2): p. 105-16.
5. Billings, L.K. and J.C. Florez, The genetics of type 2 diabetes: what have we learned from
GWAS? Ann N Y Acad Sci, 2010. 1212: p. 59-77.
6. Florez, J.C., Newly identified loci highlight beta cell dysfunction as a key cause of type 2
diabetes: where are the insulin resistance genes? Diabetologia, 2008. 51(7): p. 1100-10.
7. Strawbridge, R.D., J; Prokopenko, I; et al., Genome-wide association identifies nine
common variants associated with fasting proinsulin levels and provides new insights into
the pathophysiology of type 2 diabetes. (in press). Diabetes, 2011.
8. Ashcroft, F.M. and P. Rorsman, Diabetes mellitus and the beta cell: the last ten years. Cell,
2012. 148(6): p. 1160-71.
9. Banasik, K., et al., The effect of FOXA2 rs1209523 on glucose-related phenotypes and risk
of type 2 diabetes in Danish individuals. BMC Med Genet, 2012. 13: p. 10.
10. Grarup, N., et al., The diabetogenic VPS13C/C2CD4A/C2CD4B rs7172432 variant impairs
glucose-stimulated insulin response in 5,722 non-diabetic Danish individuals. Diabetologia,
2011. 54(4): p. 789-94.
11. Jonsson, A., et al., Effect of Common Genetic Variants Associated with Type 2 Diabetes and
Glycemic Traits on alpha- and beta-cell Function and Insulin Action in Man. Diabetes,
2013.
12. Nielsen, T., et al., Type 2 diabetes risk allele near CENTD2 is associated with decreased
glucose-stimulated insulin release. Diabetologia, 2011. 54(5): p. 1052-6.
13. Wagner, R., et al., Glucose-raising genetic variants in MADD and ADCY5 impair
conversion of proinsulin to insulin. PLoS One, 2011. 6(8): p. e23639.
14. Ingelsson, E., et al., Detailed physiologic characterization reveals diverse mechanisms for
novel genetic Loci regulating glucose and insulin metabolism in humans. Diabetes, 2010.
59(5): p. 1266-75.
15. Ferrannini, E. and A. Mari, How to measure insulin sensitivity. Journal of hypertension,
1998. 16(7): p. 895-906.
16. Manning, A.K., et al., A genome-wide approach accounting for body mass index identifies
genetic variants influencing fasting glycemic traits and insulin resistance. Nat Genet, 2012.
44(6): p. 659-669.
17. Belfiore, F., S. Iannello, and G. Volpicelli, Insulin sensitivity indices calculated from basal
and OGTT-induced insulin, glucose, and FFA levels. Molecular genetics and metabolism,
1998. 63(2): p. 134-41.
18. Stumvoll, M., et al., Use of the oral glucose tolerance test to assess insulin release and
insulin sensitivity. Diabetes care, 2000. 23(3): p. 295-301.
19. Matsuda, M. and R.A. DeFronzo, Insulin sensitivity indices obtained from oral glucose
tolerance testing: comparison with the euglycemic insulin clamp. Diabetes care, 1999.
22(9): p. 1462-70.
20. Gutt, M., et al., Validation of the insulin sensitivity index (ISI(0,120)): comparison with
other measures. Diabetes research and clinical practice, 2000. 47(3): p. 177-84.
Page 29 of 73 Diabetes
28
21. Fall, T., et al., The role of adiposity in cardiometabolic traits: a mendelian randomization
analysis. PLoS Med, 2013. 10(6): p. e1001474.
22. Frayling, T.M., et al., A common variant in the FTO gene is associated with body mass
index and predisposes to childhood and adult obesity. Science, 2007. 316(5826): p. 889-94.
23. International HapMap, C., A haplotype map of the human genome. Nature, 2005. 437(7063):
p. 1299-320.
24. Magi, R. and A.P. Morris, GWAMA: software for genome-wide association meta-analysis.
BMC Bioinformatics, 2010. 11: p. 288.
25. Willer, C.J., Y. Li, and G.R. Abecasis, METAL: fast and efficient meta-analysis of
genomewide association scans. Bioinformatics, 2010. 26(17): p. 2190-1.
26. Prokopenko, I., et al., Variants in MTNR1B influence fasting glucose levels. Nat Genet,
2009. 41(1): p. 77-81.
27. Zeggini, E., et al., Meta-analysis of genome-wide association data and large-scale
replication identifies additional susceptibility loci for type 2 diabetes. Nat Genet, 2008.
40(5): p. 638-45.
28. Everitt, B.S., S. Landau, and M. Leese, Cluster Analysis (Fourth ed.), ed. Arnold. 2001,
London.
29. Shimodaira, H., Approximately unbiased tests of regions using multistep-multiscale
bootstrap resampling. Annals of Statistics, 2004. 32: p. 2616-2641.
30. Caliński, T. and J. Harabasz, A dendrite method for cluster analysis. Communications in
Statistics, 1974. 3(1): p. 1-27.
31. Gonzalez-Sanchez, J.L., et al., Association of variants of the TCF7L2 gene with increases in
the risk of type 2 diabetes and the proinsulin:insulin ratio in the Spanish population.
Diabetologia, 2008. 51(11): p. 1993-7.
32. Kirchhoff, K., et al., Polymorphisms in the TCF7L2, CDKAL1 and SLC30A8 genes are
associated with impaired proinsulin conversion. Diabetologia, 2008. 51(4): p. 597-601.
33. Loos, R.J., et al., TCF7L2 polymorphisms modulate proinsulin levels and beta-cell function
in a British Europid population. Diabetes, 2007. 56(7): p. 1943-7.
34. Stolerman, E.S., et al., TCF7L2 variants are associated with increased proinsulin/insulin
ratios but not obesity traits in the Framingham Heart Study. Diabetologia, 2009. 52(4): p.
614-20.
35. Lyssenko, V., et al., Common variant in MTNR1B associated with increased risk of type 2
diabetes and impaired early insulin secretion. Nature genetics, 2009. 41(1): p. 82-8.
36. Langenberg, C., et al., Common genetic variation in the melatonin receptor 1B gene
(MTNR1B) is associated with decreased early-phase insulin response. Diabetologia, 2009.
52(8): p. 1537-42.
37. Matschinsky, F.M., Regulation of pancreatic beta-cell glucokinase: from basics to
therapeutics. Diabetes, 2002. 51 Suppl 3: p. S394-404.
38. Fajans, S.S., G.I. Bell, and K.S. Polonsky, Molecular mechanisms and clinical
pathophysiology of maturity-onset diabetes of the young. N Engl J Med, 2001. 345(13): p.
971-80.
39. Simonis-Bik, A.M., et al., Gene variants in the novel type 2 diabetes loci
CDC123/CAMK1D, THADA, ADAMTS9, BCL11A, and MTNR1B affect different aspects of
pancreatic beta-cell function. Diabetes, 2010. 59(1): p. 293-301.
40. Bell, G.I. and K.S. Polonsky, Diabetes mellitus and genetically programmed defects in beta-
cell function. Nature, 2001. 414(6865): p. 788-91.
41. Gloyn, A.L., et al., Activating mutations in the gene encoding the ATP-sensitive potassium-
channel subunit Kir6.2 and permanent neonatal diabetes. N Engl J Med, 2004. 350(18): p.
1838-49.
Page 30 of 73Diabetes
29
42. Awata, T., et al., Missense variations of the gene responsible for Wolfram syndrome
(WFS1/wolframin) in Japanese: possible contribution of the Arg456His mutation to type 1
diabetes as a nonautoimmune genetic basis. Biochem Biophys Res Commun, 2000. 268(2):
p. 612-6.
43. Small, K.S., et al., Identification of an imprinted master trans regulator at the KLF14 locus
related to multiple metabolic phenotypes. Nat Genet, 2011. 43(6): p. 561-4.
44. Helgason, A., et al., Refining the impact of TCF7L2 gene variants on type 2 diabetes and
adaptive evolution. Nat Genet, 2007. 39(2): p. 218-25.
Page 31 of 73 Diabetes
30
Figure legends
Figure 1. Cluster analysis of effects of 36 type 2 diabetes loci on principal physiologic traits.
Clustering of traits with meta-analysis results from at least 10,000 individuals (principal traits). The
existence of five clusters was revealed using two clustering approaches: (A) Complete linkage
dendrogram of type 2 diabetes SNPs with P values (%) indicating the robustness of each branching
event (shown in red).We named the clusters as hyperglycemic (HG) loci linked to reduced beta-cell
function following glucose stimulation, insulin resistance (IR) loci with a primary effect on IR at
basal measurements, proinsulin (PI) locus linked to decreased fasting proinsulin, beta cell (BC) loci
associated with defective beta cell function, and unclassified (UC) loci with no apparent impact on
glycemic measures. Strong support exists for the baseline branching notes (strength P≥0.84)
whereas branching of IR from the BC-UC clade shows lesser evidence for support (strength
P=0.64). (B) Calinski index computed on the centroid-based clustering of type 2 diabetes SNPs
provides further evidence for the existence of five locus groups.
Figure 2. Scatter plots of allelic effect size estimates for selected trait pairs.
In each scatter plot loci were assigned to the groups defined from the cluster analysis of principal
traits (groups highlighted by different colors). (A) Insulinogenic index vs. FI: this plot highlights
the effects of loci linked to IR (PPARG, KLF14, IRS1, GCKR) with respect to FI and insulinogenic
index. (B) Insulinogenic index vs. FG: the plot reveals the largest impact of HG loci (MTNR1B and
GCK) on FG driven by reduced beta-cell function. Large negative effects on insulinogenic index are
also seen for CDKAL1 and HHEX/IDE, but with very modest effects on FG. (C) HOMA-B vs.
HOMA-IR: the plot shows the separation of the BC, HG and IR clusters. Cluster group colors are
HG: orange, IR: green, PI: pink, BC: red, UC: blue. Loci named in the box are coded numerically
within the respective scatter plot.
Page 32 of 73Diabetes
Figure 1. Cluster analysis of effects of 36 type 2 diabetes loci on principal physiologic traits. Clustering of traits with meta-analysis results from at least 10,000 individuals (principal traits). The
existence of five clusters was revealed using two clustering approaches: (A) Complete linkage dendrogram of type 2 diabetes SNPs with P values (%) indicating the robustness of each branching event (shown in
red).We named the clusters as hyperglycemic (HG) loci linked to reduced beta-cell function following glucose stimulation, insulin resistance (IR) loci with a primary effect on IR at basal measurements, proinsulin (PI) locus linked to decreased fasting proinsulin, beta cell (BC) loci associated with defective beta cell function,
and unclassified (UC) loci with no apparent impact on glycemic measures. Strong support exists for the
baseline branching notes (strength P≥0.84) whereas branching of IR from the BC-UC clade shows lesser evidence for support (strength P=0.64). (B) Calinski index computed on the centroid-based clustering of
type 2 diabetes SNPs provides as further evidence for the existence of five locus groups. 414x435mm (96 x 96 DPI)
Page 33 of 73 Diabetes
305x523mm (96 x 96 DPI)
Page 34 of 73Diabetes
Figure 2. Scatter plots of allelic effect size estimates for selected trait pairs. In each scatter plot loci were assigned to the groups defined from the cluster analysis of principal traits
(groups highlighted by different colors). (A) Insulinogenic index vs. FI: this plot highlights the effects of loci linked to IR (PPARG, KLF14, IRS1, GCKR) with respect to FI and insulinogenic index. (B) Insulinogenic index vs. FG: the plot reveals the largest impact of HG loci (MTNR1B and GCK) on FG driven by reduced beta-cell function. Large negative effects on insulinogenic index are also seen for CDKAL1 and HHEX/IDE, but with very modest effects on FG. (C) HOMA-B vs. HOMA-IR: the plot shows the separation of the BC, HG and IR clusters. Cluster group colors are HG: orange, IR: green, PI: pink, BC: red, UC: blue. Loci named in the box
are coded numerically within the respective scatter plot. 557x398mm (96 x 96 DPI)
Page 35 of 73 Diabetes
542x398mm (96 x 96 DPI)
Page 36 of 73Diabetes
572x398mm (96 x 96 DPI)
Page 37 of 73 Diabetes
1
Impact of type 2 diabetes susceptibility variants on quantitative glycemic traits reveals
mechanistic heterogeneity
Running title: Physiological characterization of known type 2 diabetes loci
Antigone S. Dimas1,2*, Vasiliki Lagou1,3*, Adam Barker4*, Joshua W. Knowles5*, Reedik
Mägi1,3,6, Marie-France Hivert7,8, Andrea Benazzo9, Denis Rybin10, Anne U. Jackson11, Heather M. Stringham11, Ci Song12,13, Antje Fischer-Rosinsky14, Trine Welløv Boesgaard15, Niels
Grarup16, Fahim A. Abbasi5, Themistocles L. Assimes5, Ke Hao17, Xia Yang18, Cécile Lecoeur19, Inês Barroso20,21, Lori L. Bonnycastle22, Yvonne Böttcher23, Suzannah Bumpstead20, Peter S. Chines22, Michael R. Erdos22, Jurgen Graessler24, Peter Kovacs25, Mario A. Morken22, Narisu Narisu22, Felicity Payne20, Alena Stancakova26, Amy J. Swift22, Anke Tönjes23,27, Stefan
R. Bornstein24, Stéphane Cauchi19, Philippe Froguel19,28, David Meyre19,29, Peter E.H. Schwarz24, Hans-Ulrich Häring30, Ulf Smith31, Michael Boehnke11, Richard N. Bergman32,
Francis S. Collins22, Karen L. Mohlke33, Jaakko Tuomilehto34-36, Thomas Quertemous5, Lars Lind37, Torben Hansen16,38, Oluf Pedersen16,39-41, Mark Walker42, Andreas F.H. Pfeiffer14,43,
Joachim Spranger14, Michael Stumvoll23,27, James B. Meigs8,44, Nicholas J. Wareham4, Johanna Kuusisto26, Markku Laakso26, Claudia Langenberg4, Josée Dupuis45,46, Richard M.
Watanabe*47, Jose C. Florez*44,48,49, Erik Ingelsson*1,12, Mark I. McCarthy*1,3,50, Inga Prokopenko*1,3,28 on behalf of the MAGIC investigators
* These authors contributed equally to this work.
Correspondence should be addressed to:
Inga Prokopenko, MSc, PhD Department of Genomics of Common Disease School of Public Health Imperial College London Burlington Danes Building, Hammersmith Hospital, Du Cane Road, London, W12 0NN, UK Phone: +4420 759 46501 E-mail: i.prokopenko@imperial.ac.uk Prof. Mark I. McCarthy OCDEM, Churchill Hospital, University of Oxford Old Road, Headington, OX3 7LJ Email: mark.mccarthy@drl.ox.ac.uk Phone: +44 1865 857298 Erik Ingelsson, MD, PhD, FAHA Professor of Molecular Epidemiology Department of Medical Sciences Molecular Epidemiology and Science for Life Laboratory UCR/MTC Dag Hammarskjölds väg 14B Uppsala Science Park SE-752 37 Uppsala Sweden
Page 38 of 73Diabetes
2
Phone: +46-70-7569422 Fax: +46-18-515570 E-mail: erik.ingelsson@medsci.uu.se Jose C. Florez, MD, PhD Diabetes Unit / Center for Human Genetic Research Simches Research Building - CPZN 5.250 Massachusetts General Hospital 185 Cambridge Street Boston, MA 02114 Phone: 617.643.3308 Fax: 617.643-6630 Richard M. Watanabe, PhD Departments of Preventive Medicine and Physiology & Biophysics Diabetes and Obesity Research Institute of USC Keck School of Medicine of USC 2250 Alcazar Street, CSC-204 Los Angeles, CA 90089-9073 Phone: 1-323-442-2053
Affiliations
1. Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, OX3 7BN, UK.
2. Alexander Fleming, Biomedical Sciences Research Center, 34 Fleming Street, Vari, 16672 Athens, Greece.
3. Oxford Centre for Diabetes, Endocrinology and Metabolism, University of Oxford, UK, OX3 7LJ.
4. MRC Epidemiology Unit, Institute of Metabolic Science, Addenbrooke’s Hospital, Cambridge, UK.
5. Department of Medicine and Cardiovascular Institute, Stanford University School of Medicine, Stanford, CA 94305, USA.
6. Estonian Genome Center, University of Tartu, Tartu, 51010, Estonia. 7. Department of Medicine, Université de Sherbrooke, Sherbrooke (Quebec), Canada. 8. General Medicine Division, Massachusetts General Hospital, Boston,Massachusetts,
USA. 9. Department of Biology and Evolution, University of Ferrara, Ferrara, Italy. 10. Boston University Data Coordinating Center, Boston, Massachusetts, MA 02118, USA. 11. Department of Biostatistics and Center for Statistical Genetics, University of Michigan
School of Public Health, Ann Arbor, Michigan 48109, USA. 12. Department of Medical Sciences, Molecular Epidemiology and Science for Life
Laboratory, Uppsala University, Uppsala, Sweden. 13. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet,
Stockholm, Sweden. 14. Charité-Universitätsmedizin Berlin, Department of Endocrinology and Metabolism. 15. Steno Diabetes Center, Gentofte, Denmark. 16. The Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health
and Medical Sciences, University of Copenhagen, Copenhagen, Denmark.
Page 39 of 73 Diabetes
3
17. Department of Genetics and Genomic Sciences, Institute of Genomics and Multiscale Biology, Mount Sinai School of Medicine, New York, NY 10029-6574, USA.
18. Department of Integrative Biology and Physiology, University of California, 610 Charles E. Young Dr. East, Los Angeles, CA 90095, USA.
19. CNRS UMR8199-Institute of Biology, Pasteur Institute, Lille 2-Droit et Santé University, Lille, France.
20. Wellcome Trust Sanger Institute, Hinxton, CB10 1SA, UK. 21. University of Cambridge Metabolic Research Laboratories and NIHR Cambridge
Biomedical Research Centre, Level 4, Institute of Metabolic Science, Box 289, Addenbrooke’s Hospital, Cambridge CB2 OQQ, UK.
22. Genome Technology Branch, National Human Genome Research Institute, Bethesda, MD.
23. Leipzig University Medical Center, IFB AdiposityDiseases, Liebigstr. 21, 04103 Leipzig, Germany.
24. Department of Medicine III, Division of Prevention and Care of Diabetes, University of Dresden, 01307 Dresden, Germany.
25. Interdisciplinary Center for Clinical Research (IZKF) Leipzig, Liebigstr. 21, 04103 Leipzig, Germany.
26. Department of Medicine, University of Eastern Finland and Kuopio University Hospital, 70210 Kuopio, Finland.
27. Department of Medicine, University of Leipzig, Liebigstr. 18, 04103 Leipzig, Germany. 28. Department of Genomics of common diseases, Imperial College London, London, UK. 29. Department of Clinical Epidemiology & Biostatistics, McMaster University, Hamilton,
Canada. 30. Department of Internal Medicine, Division of Endocrinology, Diabetology, Vascular
Medicine, Nephrology and Clinical Chemistry, University of Tübingen, Tübingen, Germany.
31. Lundberg Laboratory for Diabetes Research, Center of Excellence for Metabolic and Cardiovascular Research, Department of Molecular and Clinical Medicine, Sahlgrenska Academy, University of Gothenburg, SE-413 45 Gothenburg, Sweden.
32. Department of Physiology & Biophysics, Keck School of Medicine, University of Southern California, Los Angeles, California 90033, USA.
33. Department of Genetics, University of North Carolina Chapel Hill, North Carolina 27599, USA.
34. Diabetes Prevention Unit, National Institute for Health and Welfare, 00271 Helsinki, Finland.
35. Centre for Vascular Prevention, Danube-University Krems, 3500 Krems, Austria. 36. King Abdulaziz University, Jeddah 21589, Saudi Arabia. 37. Department of Medical Sciences, Uppsala University, Akademiska sjukhuset, Uppsala,
Sweden. 38. Faculty of Health Sciences, University of Southern Denmark, Odense, Denmark. 39. Hagedorn Research Institute, Copenhagen, Denmark. 40. Institute of Biomedical Science, Faculty of Health Sciences, University of Copenhagen,
Copenhagen, Denmark. 41. Faculty of Health Sciences, University of Aarhus, Aarhus, Denmark. 42. Institute of Cellular Medicine, Newcastle University, UK. 43. German Institute of Human Nutrition, Department of Clinical Nutrition. 44. Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115, USA. 45. Department of Biostatistics, Boston University School of Public Health, Boston,
Massachusetts, MA 02118, USA.
Page 40 of 73Diabetes
4
46. The National Heart, Lung, and Blood Institute’s Framingham Heart Study, Framingham, Massachusetts, USA.
47. Departments of Preventive Medicine and Physiology & Biophysics, Keck School of Medicine of USC, Los Angeles, CA 90033, USA.
48. Center for Human Genetic Research and Diabetes Research Center (Diabetes Unit), Massachusetts General Hospital, Boston, Massachusetts 02114, USA.
49. Program in Medical and Population Genetics, Broad Institute, Cambridge, Massachusetts 02142, USA.
50. Oxford NIHR Biomedical Research Centre, Churchill Hospital, Oxford OX3 7LJ, UK.
SUPPLEMENTARY NOTES AND FIGURES
Supplementary Note. Genotyping and quality control.
Supplementary Table 1. Cohort descriptions and baseline characteristics (Excel file attached).
Supplementary Table 2. Definitions of physiological measurements.
Supplementary Table 3. Genetic markers investigated.
Supplementary Table 4. Effects of 37 SNPs previously associated with type 2 diabetes on
physiologic glycemic measures with genetic data available in >10,000 non-diabetic individuals
(principal traits).
Supplementary Table 5. Effects of 37 SNPs previously associated with type 2 diabetes on
physiologic glycemic measures with genetic data available in <10,000 non-diabetic individuals.
Supplementary Figure 1. Forest plots for meta-analyses of insulinogenic index, split
proinsulin and acute insulin response (AIR). (A) insulinogenic index at HHEX/IDE, (B)
insulinogenic index at CDKAL1, (C) split proinsulin at HNF1A, (D) AIR at MTNR1B, (E) AIR
at KCNQ1, (F) AIR at CDKN2A/B.
Supplementary Figure 2. Forest plots for meta-analyses of intravenous (IV) insulin sensitivity
traits. (A) IRS1, (B) ADCY5.
Supplementary Figure 3. Results of cluster analysis using principal traits.
Supplementary Figure 4. Results of cluster analysis using all traits.
Supplementary Figure 5. Principal component analysis (PCA) of type 2 diabetes SNPs.
Supplementary Figure 6. Scatter plots of allelic effect size estimates for given trait pairs.
Page 41 of 73 Diabetes
5
Supplementary note
Genotyping and quality control
Genotyping data was obtained de novo for eleven studies and in silico (HapMap imputed) for four studies. Initial association testing for each cohort was performed using Merlin, STATA 10.1, R (LMEKIN package, R 2.10.0 and R 2.9.2), SNPTEST v1.1.4, SPSS 18.0, and SAS (9.1.3 and 9.1.9). Prior to meta-analysis, low quality SNPs were filtered in individual cohorts using the following criteria: (1) call rate for directly genotyped SNPs <0.70, (2) SNP Hardy-Weinberg equilibrium P<10-6, (3) SNPs minor allele count (MAC) ≥10 (calculated as total number of observed alleles [twice the sample size] at each SNP multiplied by minor allele frequency [MAF]). In samples where initial genotyping of an index SNP failed, a proxy SNP in strong linkage disequilibrium (LD) with the original SNP was genotyped and reported wherever possible.
Supplementary Table 2. Definitions of physiological measurements.
Index Equation Reference
Insulinogenic
Index
[(30-min insulin-fasting insulin)]/[(30-min glucose-fasting
glucose)]
ISI Belfiore
2/[({[0.5*fasting PG (mmol/L)] + OGTT 1-h PG + (0.5*OGTT
2-h PG)} / 11.36) * ({[0.5*fasting PI (pmol/L)] + OGTT 1-h PI
+ (0.5*OGTT 2-h PI)} / 638) + 1]
(1)
ISI Gutt
{75000 + [fasting PG (mg/L) - OGTT 2-h PG] * 0.19 * body
weight (kg)/120} / [(fasting PG + OGTT 2-h PG) / 2] /
log10{[fasting PI (mU/L) + OGTT 2-h PI] / 2}
(2)
ISI Matsuda 10 000√{fasting PG (mg/dL) * fasting PI (µU/mL) * [mean
OGTT PG * mean OGTT PI]} (3)
ISI Stumvoll 0.226 - [0.0032 * BMI (kg/m2)] - [0.0000645 * OGTT 2-h PI
(pmol/L)] - [0.00375 * OGTT 1.5-h PG (mmol/L)] (4)
Abbreviations: PG: plasma glucose, OGTT: oral glucose tolerance test, PI: plasma insulin.
Page 42 of 73Diabetes
6
Supplementary Table 3. Genetic markers investigated.
Lead SNP Chr
Position
(b36)
Nearest
gene
Effect
allele*/
Other MAF Proxy used
r2
with
lead
SNP†
rs10830963 11 92348358 MNTR1B G/C 0.30 rs7112766 0.861 rs11020124 0.725 rs1387153 0.703 rs10830956 0.703
rs4607517 7 44202193 GCK A/G 0.20 rs730497 1 rs2908289 1 rs1799884 1 rs6975024 1
rs13081389 3 12264800 PPARG A/G 0.04 rs1596417 1 rs17036130 1 rs1562040 1 rs1801282 0.536
rs972283 7 130117394 KLF14 G/A 0.45 rs4731702 1 rs3996352 1 rs11765979 1 rs10954284 1
rs7578326 2 226728897 IRS1 A/G 0.36 rs2943651 0.964 rs2943632 0.964 rs13423088 0.929 rs2943641 0.74
rs780094 2 27594741 GCKR C/T 0.38 rs780093 1 rs1260326 0.932 rs2911711 0.87 rs1260333 0.87
rs1552224 11 72110746 ARAP1 A/C 0.13 rs11603334 1 rs11605166 0.853 rs613937 0.736
rs7903146 10 114748339 TCF7L2 T/C 0.25 rs4506565 0.917 rs4132670 0.917
rs3802177 8 118254206 SLC30A8 G/A 0.25 rs13266634 1 rs11558471 0.957
rs5015480 10 94455539 HHEX/
IDE C/T 0.45 rs10882102 1 rs1111875 1 rs12778642 0.904
rs2191349 7 15030834 DGKB/
TMEM195 T/G 0.47 rs2191348 1 rs4719433 1 rs6947830 1 rs2215383 1 rs10244051 1
rs340874 1 212225879 PROX1 C/T 0.49 rs340835 0.765
Page 43 of 73 Diabetes
7
rs340883 0.548
rs10965250 9 22123284 CDKN2A/
B G/A 0.19 rs10965250 1 rs10811661 0.95 rs2383208 0.95
rs10440833 6 20796100 CDKAL1 A/T 0.25 rs9368222 1 rs7766070 1 rs7756992 1 rs7754840 0.677
rs11899863 2 43472323 THADA C/T 0.07 rs7607777 1 rs6743071 1 rs13408002 1 rs17406646 1 rs7578597 0.786
rs11708067 3 124548468 ADCY5 A/G 0.22 rs11717195 0.86 rs2124500 0.819 rs6798189 0.795 rs7613951 0.753
rs4430796 17 33172153 HNF1B
(TCF2) G/A 0.47 rs2005705 0.632 rs757210 0.615
rs2334499 11 1653425 HCCA2 T/C 0.44 rs4752780 0.964 rs4417225 0.934 rs4752781 0.931 rs10839220 0.583
rs7957197 12 119945069 HNF1A T/A 0.15 rs12427353 0.832 rs7965349 0.818 rs7305618 0.55
rs10923931 1 120319482 NOTCH2 T/G 0.12 rs2641317 1 rs2934381 1 rs1493694 1 rs2453040 1 rs2793831 1
rs13292136 9 81141948 CHCHD9 C/T 0.07 rs17791555 1 rs13297387 1 rs1328405 1 rs1328405 1
rs1801214 4 6353923 WFS1 T/C 0.27 rs6446480 0.917 rs10937720 0.917 rs13108780 0.917 rs13128674 0.917 rs10010131 0.917
rs4457053 5 76460705 ZBED3 G/A 0.26 rs6878122 0.864 rs7708285 0.862
rs8042680 15 89322341 PRC1 A/C 0.22 rs6496743 1 rs6496742 0.811 rs11853073 0.811 rs8028856 0.81
rs1470579 3 187011774 IGF2BP2 C/A 0.29 rs1470579 1
Page 44 of 73Diabetes
8
rs6767484 1 rs6769511 1 rs11711477 1 rs9859406 1
rs896854 8 96029687 TP53INP1 T/C 0.48 rs13255935 1 rs13257021 0.967 rs7845219 0.846 rs12548874 0.743
rs12779790 10 12368016 CDC123/
CAMK1D G/A 0.23 rs11257655 0.802 rs7069060 0.617 rs10906111 0.562
rs11634397 15 78219277 ZFAND6 G/A 0.41 rs870185 0.897 rs12441266 0.897 rs1357335 0.897 rs4778582 0.896
rs4760790 12 69921061 TSPAN8/
LGR5 A/G 0.23 rs4760915 1 rs7306184 1 rs1353362 0.955 rs7961581 0.909
rs5215 11 17365206 KCNJ11 C/T 0.41 rs1557765 0.902 rs757110 0.899 rs7124355 0.745
rs243021 2 60438323 BCL11A A/G 0.46 rs243020 1 rs243019 1 rs243018 0.967
rs849134 7 28162747 JAZF1 A/G 0.47 rs849135 1 rs849133 1 rs1635852 1 rs864745 0.967
rs1531343 12 64461161 HMGA2 C/G 0.10 rs2612067 1 rs2583922 1 rs2358949 1 rs1122590 1 rs2884592 1
rs231362 11 2648047 KCNQ1 G/A 0.49 rs163184 11 2803645 KCNQ1 G/T 0.44
rs6795735 3 64680405 ADAMTS9 C/T 0.45 rs7428936 1 rs4611812 1 rs17727064 1 rs4607103 0.28
rs5945326 X 152553116 DUSP9 A/G 0.21 rs3020789 0.939 rs2301142 0.548
*-Effect allele defined as the type 2 diabetes risk increasing allele. †-r2 values based on HapMap CEU, release 22
Page 45 of 73 Diabetes
9
Supplementary Table 4. Effects of 37 SNPs previously associated with type 2 diabetes on physiologic glycemic measures with genetic data
available in more than 10,000 non-diabetic individuals.
Dynamic traits
Fasting traits Insulino- genic index (n=11,268)
ISI
Belfiore
(n=10,34)
ISI
Stumvoll
(n=10,23
9)
ISI
Matsuda
(n=10,364)
ISI Gutt
(n=13,158)
Proinsulin
(n=13,912)
Fasting
Glucose*
(n=58,614)
Fasting
Insulin*
(n=52,379) HOMA-B*
(n=50,908)
HOMA-
IR*
(n=50,908)
SNP Nearest
gene
Alleles
(effect
/
other)
Effect (SE) P
Effect
(SE) P
Effect
(SE) P
Effect (SE) P
Effect (SE) P
Effect (SE) P
Effect (SE) P
Effect (SE) P
Effect (SE) P
Effect (SE) P
Insulin Resistance (IR)
rs13081389 PPARG A/G 0.0092 (0.017)
-0.019 (0.0069)
-0.021 (0.0082)
-0.035 (0.012)
-0.015 (0.0071)
0.0082 (0.0088)
0.0097 (0.0062)
0.025 (0.0052)
0.017 (0.0048)
0.025 (0.0055)
0.59 0.0055 0.012 0.0037 0.039 0.35 0.12 1.6×10-6 4.96×10-4 4.37×10-6
rs972283 KLF14 G/A -0.0020 (0.011)
-0.0030 (0.0042)
-0.0083 (0.0052)
-0.0013 (0.0077)
-0.0079 (0.0041)
0.0045 (0.0052)
0.0066 (0.0032)
0.013 (0.0027)
0.0091 (0.0026)
0.014 (0.0029)
0.85 0.47 0.11 0.86 0.055 0.39 0.040 4.41×10-6 4.16×10-4 6.84×10-7
rs7578326 IRS1 A/G 0.037 (0.010)
-0.023 (0.0043)
-0.013 (0.0050)
-0.040 (0.0075)
-0.020 (0.0042)
0.014 (0.0052)
0.0045 (0.0034)
0.019 (0.0029)
0.014 (0.0027)
0.019 (0.0031)
0.00038 1.19×10-7 0.0090 1.46×10-7 2.47×10-6 0.0063 0.19 2.68×10-11 2.65×10-7 6.95×10-10
rs780094 GCKR C/T 0.0090 (0.010)
-0.0077 (0.0041)
-0.015 (0.0048)
-0.019 (0.0073)
0.0072 (0.0039)
0.010 (0.0053)
0.033 (0.0032)
0.019 (0.0027)
0.0058 (0.0026)
0.023 (0.0029)
0.38 0.061 0.0020 0.0077 0.061 0.050 3.30×10-24 3.36×10-12 0.022 4.68×10-16 Hyperglycemic (HG)
rs10830963 MTNR1B G/C -0.083 (0.011)
0.0057 (0.0043)
-0.0073 (0.0051)
-0.0043 (0.0076)
-0.0059 (0.0043)
0.0093 (0.0058)
0.079 (0.0038)
0.0004 (0.0033)
-0.033 (0.0031)
0.011 (0.0035)
1.23×10-14 0.18 0.15 0.57 0.17 0.11 8.72×10-95 0.91 4.23×10-26 0.0018
rs4607517 GCK A/G -0.036 (0.017)
-0.0048 (0.0070)
-0.015 (0.0083)
-0.0078 (0.012)
-0.016 (0.0060)
-0.00068 (0.0080)
0.065 (0.0043)
0.0064 (0.0037)
-0.024 (0.0035)
0.014 (0.0039)
0.031 0.49 0.077 0.52 0.0069 0.93 1.39×10-51 0.081 1.32×10-11 3.89×10-4 Proinsulin (PI)
Page 46 of 73Diabetes
10
rs1552224 ARAP1 A/C -0.054 (0.012)
-0.00038 (0.0048)
-0.013 (0.0062)
-0.013 (0.0089)
-0.010 (0.0052)
-0.093 (0.0062)
0.024 (0.0042)
-0.0075 (0.0034)
-0.015 (0.0032)
-0.0051 (0.0037)
1.54×10-5 0.94 0.033 0.15 0.050 2.16×10-50 9.40×10-9 0.029 3.16×10-6 0.16 Beta Cell (BC)
rs7903146 TCF7L2 T/C -0.063 (0.012)
0.0094 (0.0051)
-0.0085 (0.0058)
0.010 (0.0085)
-0.017 (0.0046)
0.044 (0.0061)
0.026 (0.0036)
-0.014 (0.0031)
-0.019 (0.0029)
-0.010 (0.0033)
1.40×10-7 0.064 0.14 0.24 0.00021 4.20×10-13 9.55×10-13 4.81×10-6 4.25×10-11 0.0022
rs3802177 SLC30A8 G/A -0.041 (0.011)
0.00066 (0.0043)
-0.0039 (0.0055)
-0.0012 (0.0081)
-0.011 (0.0050)
0.038 (0.0062)
0.033 (0.0035)
-0.0035 (0.0031)
-0.016 (0.0028)
0.0002 (0.0033)
0.00024 0.88 0.47 0.88 0.034 1.25×10-9 4.32×10-21 0.26 3.41×10-8 0.96
rs5015480 HHEX/ID
E C/T -0.10 (0.011) 0.020 (0.0042)
-0.010 (0.0053)
0.017 (0.0078)
-0.0052 (0.0041)
0.0063 (0.0053)
0.010 (0.0032)
-0.0008 (0.0027)
-0.0051 (0.0025)
0.0007 (0.0028)
3.63×10-21 1.46×10-6 0.054 0.030 0.21 0.23 0.0011 0.77 0.042 0.81
rs2191349 DGKB/TMEM195 T/G
-0.028 (0.010)
0.011 (0.0040)
0.0071 (0.0047)
0.027 (0.0072)
0.0060 (0.0038)
-0.0011 (0.0049)
0.032 (0.0032)
-0.0017 (0.0027)
-0.017 (0.0025)
0.0018 (0.0028)
0.0058 0.0059 0.13 0.00018 0.12 0.83 1.15×10-24 0.51 2.07×10-11 0.52
rs340874 PROX1 C/T -0.028 (0.010)
0.0082 (0.0041)
0.0079 (0.0047)
0.0090 (0.0072)
-0.00066 (0.0040)
0.0086 (0.0055)
0.017 (0.0032)
-0.0019 (0.0028)
-0.0094 (0.0026)
-0.0003 (0.003)
0.0060 0.044 0.096 0.21 0.87 0.12 6.80×10-8 0.50 2.45×10-4 0.91
rs10965250 CDKN2A/
B G/A -0.045 (0.014)
0.013 (0.0057)
-0.00072 (0.0070)
0.012 (0.010)
0.0046 (0.0054)
-0.00039 (0.0071)
0.017 (0.0042)
-0.0049 (0.0037)
-0.012 (0.0035)
-0.0023 (0.0039)
0.0013 0.023 0.92 0.26 0.39 0.96 5.17×10-5 0.18 4.27×10-4 0.55
rs10440833 CDKAL1 A/T -0.10 (0.023) 0.0047 (0.0090)
-0.011 (0.010)
-0.0095 (0.017)
-0.0013 (0.0066)
0.0077 (0.0085)
0.014 (0.0035)
-0.0061 (0.0029)
-0.0088 (0.0027)
-0.0043 (0.0031)
6.84×10-6 0.60 0.30 0.58 0.84 0.37 5.19×10-5 0.036 0.0012 0.17
rs11899863 THADA C/T 0.0078 (0.020)
0.0096 (0.0080)
-0.0036 (0.0095)
0.016 (0.015)
-0.0071 (0.0072)
0.015 (0.0097)
0.020 (0.0054)
-0.010 (0.0048)
-0.020 (0.0046)
-0.0080 (0.0051)
0.70 0.23 0.70 0.28 0.32 0.12 2.28×10-4 0.036 1.30×10-5 0.12
rs11708067 ADCY5 A/G -0.0089 (0.013)
-0.0079 (0.0051)
0.0012 (0.0062)
-0.0025 (0.0093)
-0.014 (0.0047)
0.016 (0.0064)
0.022 (0.0039)
-0.0060 (0.0034)
-0.015 (0.0032)
-0.0011 (0.0036)
0.49 0.13 0.85 0.79 0.0024 0.012 1.89×10-8 0.076 5.39×10-6 0.76 Unclassified (UC)
rs4430796 HNF1B
(TCF2) G/A -0.044 (0.023)
0.0072 (0.0091)
-0.011 (0.011)
0.0082 (0.018)
-0.013 (0.0095)
-0.00053 (0.016)
-0.0007 (0.0039)
-0.012 (0.0035)
-0.011 (0.0031)
-0.012 (0.0037)
0.061 0.43 0.32 0.64 0.17 0.97 0.85 8.25×10-4 5.14×10-4 9.08×10-4
Page 47 of 73 Diabetes
11
rs2334499 HCCA2 T/C -0.018 (0.023)
0.0098 (0.0089)
0.016 (0.0099)
0.015 (0.017)
-0.0057 (0.0068)
-0.017 (0.0086)
0.0011 (0.0032)
-0.0044 (0.0027)
-0.0036 (0.0026)
-0.0037 (0.0029)
0.43 0.27 0.11 0.38 0.40 0.043 0.73 0.11 0.16 0.21
rs7957197 HNF1A T/A -0.036 (0.018)
0.0086 (0.0075)
-0.0062 (0.0091)
-0.0029 (0.014)
0.00049 (0.0061)
-0.011 (0.0087)
-0.0014 (0.004)
-0.0043 (0.0033)
-0.0028 (0.0031)
-0.0045 (0.0035)
0.049 0.25 0.49 0.83 0.94 0.22 0.73 0.20 0.36 0.20
rs10923931 NOTCH2 T/G -0.012 (0.016)
0.011 (0.0065)
0.0026 (0.0078)
0.019 (0.012)
0.0011 (0.0065)
-0.000015 (0.0084)
0.0025 (0.005)
-0.0034 (0.0042)
-0.0043 (0.0039)
-0.0014 (0.0044)
0.45 0.085 0.74 0.11 0.87 1.00 0.63 0.41 0.28 0.76
rs13292136 CHCHD9 C/T -0.025 (0.016)
0.012 (0.0069)
-0.0013 (0.0084)
0.026 (0.012)
0.0063 (0.0069)
-0.00054 (0.0086)
-0.0028 (0.0057)
-0.0054 (0.0046)
-0.0022 (0.0042)
-0.0050 (0.0049)
0.13 0.078 0.87 0.028 0.36 0.95 0.63 0.24 0.61 0.30
rs1801214 WFS1 T/C -0.017 (0.011)
0.0016 (0.0045)
-0.00084 (0.0052)
0.0082 (0.0080)
-0.0076 (0.0042)
-0.019 (0.0053)
0.0081 (0.0033)
-0.0011 (0.0028)
-0.0054 (0.0027)
0.0003 (0.003)
0.12 0.72 0.87 0.30 0.071 0.00023 0.016 0.67 0.042 0.92
rs4457053 ZBED3 G/A -0.014 (0.012)
0.00038 (0.0049)
0.0021 (0.0058)
0.00039 (0.0083)
-0.0040 (0.0045)
-0.0078 (0.0055)
0.019 (0.004)
0.0016 (0.0033)
-0.0053 (0.0031)
0.0031 (0.0035)
0.21 0.94 0.72 0.96 0.36 0.16 2.38×10-6 0.61 0.084 0.38
rs8042680 PRC1 A/C -0.016 (0.011)
0.0030 (0.0046)
0.0074 (0.0053)
0.011 (0.0081)
-0.0021 (0.0043)
-0.0082 (0.0057)
0.011 (0.0034)
-0.0016 (0.0028)
-0.0036 (0.0027)
-0.0003 (0.003)
0.15 0.51 0.16 0.19 0.62 0.15 0.0015 0.56 0.17 0.93
rs1470579 IGF2BP2 C/A -0.038 (0.018)
0.013 (0.0068)
0.0096 (0.0079)
0.0096 (0.013)
-0.00083 (0.0054)
-0.011 (0.0073)
0.016 (0.0034)
-0.0012 (0.0028)
-0.0062 (0.0027)
0.0001 (0.003)
0.030 0.062 0.23 0.46 0.88 0.15 2.34×10-6 0.66 0.022 0.96
rs896854 TP53INP
1 T/C -0.0035 (0.010)
-0.0013 (0.0038)
-0.0056 (0.0049)
-0.0025 (0.0069)
0.00026 (0.0040)
-0.0031 (0.0049)
0.0069 (0.0031)
-0.0023 (0.0026)
-0.0045 (0.0025)
-0.0013 (0.0028)
0.73 0.74 0.26 0.72 0.95 0.53 0.028 0.38 0.070 0.65
rs12779790 CDC123/
CAMK1D G/A -0.029 (0.013)
0.0014 (0.0055)
-0.0021 (0.0064)
-0.0060 (0.0096)
0.0016 (0.0053)
-0.0082 (0.0068)
0.014 (0.0042)
-0.0018 (0.0036)
-0.0068 (0.0034)
-0.0009 (0.0038)
0.029 0.80 0.74 0.53 0.76 0.23 5.61×10-4 0.62 0.046 0.81
rs11634397 ZFAND6 G/A -0.0098 (0.010)
-0.00098 (0.0043)
-0.0055 (0.0052)
-0.0068 (0.0071)
-0.0054 (0.0042)
0.0081 (0.0051)
0.0014 (0.0035)
0.0030 (0.0029)
0.0023 (0.0028)
0.0029 (0.0031)
0.34 0.82 0.29 0.34 0.19 0.11 0.68 0.31 0.41 0.35
rs4760790 TSPAN8/
LGR5 A/G -0.016 (0.013)
0.00096 (0.0050)
0.0096 (0.0061)
-0.0035 (0.0092)
-0.00039 (0.0046)
0.0061 (0.0060)
0.0040 (0.0036)
-0.0005 (0.003)
-0.0021 (0.0029)
0.0005 (0.0032)
Page 48 of 73Diabetes
12
0.21 0.85 0.11 0.70 0.93 0.31 0.27 0.88 0.47 0.89
rs5215 KCNJ11 C/T -0.024 (0.017)
0.0017 (0.0065)
-0.015 (0.0076)
0.00091 (0.012)
0.0035 (0.0052)
0.0010 (0.0071)
-0.0040 (0.0032)
-0.0010 (0.0027)
0.0027 (0.0025)
-0.0020 (0.0029)
0.15 0.79 0.057 0.94 0.50 0.89 0.22 0.72 0.28 0.48
rs243021 BCL11A A/G 0.018 (0.011)
0.0045 (0.0040)
0.0084 (0.0051)
0.013 (0.0074)
0.0090 (0.0041)
0.0070 (0.0052)
0.0038 (0.0032)
0.0009 (0.0027)
0.0003 (0.0025)
0.00 (0.0029)
0.093 0.26 0.10 0.067 0.029 0.18 0.23 0.72 0.89 0.99
rs849134 JAZF1 A/G 0.013 (0.011)
-0.0052 (0.0042)
-0.0047 (0.0052)
-0.014 (0.0077)
-0.0061 (0.0041)
0.0035 (0.0052)
0.0074 (0.0032)
-0.0001 (0.0027)
-0.0014 (0.0025)
-0.0002 (0.0029)
0.23 0.21 0.37 0.068 0.14 0.50 0.019 0.97 0.57 0.943
rs1531343 HMGA2 C/G 0.030 (0.019)
-0.013 (0.0076)
0.0017 (0.0087)
-0.027 (0.014)
0.0012 (0.0067)
0.010 (0.0090)
0.013 (0.0054)
0.0074 (0.0045)
-0.0013 (0.0043)
0.011 (0.0048)
0.11 0.084 0.84 0.050 0.86 0.26 0.018 0.10 0.76 0.029
rs231362 KCNQ1 G/A 0.0045 (0.011)
-0.000085 (0.0045)
-0.0039 (0.0053)
-0.0033 (0.0079)
-0.0051 (0.0044)
-0.0037 (0.0052)
0.017 (0.0035)
0.0069 (0.003)
-0.0010 (0.0028)
0.0066 (0.0032)
0.68 0.98 0.46 0.67 0.24 0.48 2.7×10-6 0.020 0.72 0.038
rs163184 KCNQ1 G/T -0.011 (0.022)
0.0024 (0.0088)
0.0092 (0.0096)
0.0087 (0.017)
-0.0037 (0.0079)
0.016 (0.010)
0.012 (0.0032)
0.0019 (0.0028)
-0.0045 (0.0027)
0.0039 (0.003)
0.63 0.79 0.34 0.60 0.64 0.11 1.90×10-4 0.51 0.091 0.20
rs6795735 ADAMTS
9 C/T 0.0088 (0.010)
0.0060 (0.0043)
0.0027 (0.0051)
0.011 (0.0076)
0.0012 (0.0041)
0.00056 (0.0050)
0.0076 (0.0032)
0.0025 (0.0027)
-0.0014 (0.0025)
0.0032 (0.0029)
0.40 0.16 0.60 0.14 0.77 0.91 0.017 0.35 0.57 0.26 Other
rs5945326 DUSP9 A/G -0.037 (0.010)
0.012 (0.0038)
0.010 (0.0048)
0.022 (0.0067)
0.0059 (0.0047)
-0.013 (0.0060)
0.011 (0.013)
-0.013 (0.0138)
-2.89 (1.3345)
-0.066 (0.041)
0.00032 0.0017 0.033 0.00085 0.21 0.030 0.40 0.35 0.030 0.11
Page 49 of 73 Diabetes
13
Supplementary Table 5. Effects of 37 SNPs previously associated with type 2 diabetes on physiologic glycemic measures with genetic data
available in fewer than 10,000 non-diabetic individuals. IR: insulin resistance, HG: hyperglycemic, PI: proinsulin, BC: beta cell, UC: unclassified.
SNP Nearest gene Alleles
(effect/other)
M/I, SI, SSPG
(n=4,180) Acute Insulin
Response
(n=1,135)
C-Peptide
(n=5,059) 32,33 Split
Proinsulin
(n=2,568)
Effect (SE) P
Effect (SE) P
Effect (SE) P
Effect (SE) P
IR
rs13081389 PPARG A/G -0.025 (0.049) -0.025 (0.066) 0.015 (0.012) 0.0050 (0.032)
0.60 0.70 0.22 0.88 rs972283 KLF14 G/A -0.0023 (0.020) 0.019 (0.043) 0.0068 (0.0084) 0.0024 (0.017)
0.91 0.65 0.42 0.88 rs7578326 IRS1 A/G -0.070 (0.021) -0.0061 (0.043) 0.021 (0.0083) 0.021 (0.018)
0.00065 0.89 0.013 0.24 rs780094 GCKR C/T -0.019 (0.019) 0.034 (0.033) 0.020 (0.0074) 0.027 (0.017)
0.32 0.31 0.0079 0.12
HG
rs10830963 MTNR1B G/C 0.012 (0.021) -0.16 (0.035) -0.0069 (0.0080) -0.0034 (0.019)
0.55 4.81×10-6 0.39 0.86 rs4607517 GCK A/G 0.015 (0.025) -0.097 (0.050) -0.014 (0.014) -0.0097 (0.038)
0.55 0.051 0.32 0.80
PI
rs1552224
ARAP1
A/C -0.010 (0.026) 0.69
-0.0064 (0.054)
0.91
0.019 (0.012)
0.11
-0.13 (0.023)
1.39×10-8
BC
rs7903146 TCF7L2 T/C 0.014 (0.021) -0.054 (0.039) -0.021 (0.011) 0.012 (0.030) 0.50 0.17 0.051 0.68
rs3802177 SLC30A8 G/A -0.014 (0.020) -0.11 (0.033) 0.0096 (0.0077) 0.066 (0.018)
Page 50 of 73Diabetes
14
0.47 0.0016 0.21 0.00019 rs5015480 HHEX/IDE C/T -0.0040 (0.020) -0.11 (0.044) 0.0011 (0.0087) 0.017 (0.017)
0.84 0.013 0.90 0.33 rs2191349 DGKB/TMEM195 T/G -0.00054 (0.018) -0.013 (0.032) -0.0055 (0.0072) -0.00057 (0.017)
0.98 0.67 0.44 0.97 rs340874 PROX1 C/T -0.0020 (0.018) -0.043 (0.032) -0.00058 (0.0072) 0.00035 (0.016)
0.91 0.18 0.94 0.98 rs10965250 CDKN2A/B G/A 0.047 (0.024) -0.18 (0.044) -0.0022 (0.011) 0.011 (0.022)
0.051 3.57×10-5 0.84 0.60
rs10440833 CDKAL1 A/T 0.047 (0.021) -0.15 (0.048) 0.00083 (0.013) -0.042 (0.029) 0.026 0.0020 0.95 0.15
rs11899863 THADA C/T -0.053 (0.032) 0.013 (0.066) 0.012 (0.013) 0.038 (0.028)
0.10 0.84 0.38 0.17 rs11708067 ADCY5 A/G -0.063 (0.023) -0.013 (0.040) -0.0018 (0.0088) 0.062 (0.019)
0.0061 0.75 0.84 0.0012
UC
rs4430796 HNF1B (TCF2) G/A 0.025 (0.032) -0.021 (0.042) 0.00040 (0.0087) 0.030 (0.022)
0.45 0.61 0.96 0.18 rs2334499 HCCA2 T/C 0.031 (0.021) - 0.0043 (0.018) -0.016 (0.026)
0.14 - 0.81 0.53 rs7957197 HNF1A T/A 0.0053 (0.024) -0.079 (0.054) 0.0094 (0.011) -0.094 (0.021)
0.82 0.14 0.39 8.08×10-6
rs10923931 NOTCH2 T/G 0.038 (0.030) -0.083 (0.066) -0.017 (0.014) -0.0073 (0.027) 0.21 0.21 0.22 0.78
rs13292136 CHCHD9 C/T 0.034 (0.042) 0.040 (0.068) -0.0091 (0.017) -0.010 (0.044)
0.43 0.56 0.58 0.82 rs1801214 WFS1 T/C -0.016 (0.023) -0.063 (0.034) -0.0083 (0.0079) -0.0048 (0.021)
0.48 0.063 0.29 0.82
Page 51 of 73 Diabetes
15
rs4457053 ZBED3 G/A -0.0084 (0.023) -0.013 (0.046) -0.017 (0.0098) -0.014 (0.018)
0.72 0.77 0.084 0.43 rs8042680 PRC1 A/C 0.0034 (0.025) 0.00053 (0.043) -0.014 (0.0092) -0.013 (0.024)
0.89 0.99 0.12 0.57 rs1470579 IGF2BP2 C/A 0.013 (0.021) -0.15 (0.046) -0.0026 (0.0091) -0.0040 (0.018)
0.54 0.0014 0.77 0.82 rs896854 TP53INP1 T/C 0.0071 (0.019) 0.030 (0.041) -0.012 (0.0086) -0.022 (0.017)
0.70 0.46 0.16 0.19 rs12779790 CDC123/CAMK1D G/A 0.020 (0.024) -0.015 (0.053) -0.028 (0.011) -0.016 (0.021)
0.40 0.78 0.015 0.46 rs11634397 ZFAND6 G/A 0.032 (0.021) 0.075 (0.045) -0.0070 (0.0093) -0.022 (0.017)
0.12 0.096 0.45 0.21 rs4760790 TSPAN8/LGR5 A/G 0.016 (0.022) 0.019 (0.046) -0.012 (0.0087) -0.026 (0.018)
0.45 0.68 0.18 0.15 rs5215 KCNJ11 C/T -0.018 (0.020) -0.099 (0.042) -0.0053 (0.0088) 0.035 (0.017)
0.37 0.018 0.54 0.042 rs243021 BCL11A A/G -0.0056 (0.023) 0.0037 (0.041) -0.0053 (0.0086) 0.027 (0.022)
0.81 0.93 0.54 0.22 rs849134 JAZF1 A/G 0.015 (0.019) -0.0017 (0.041) 0.0092 (0.0084) 0.021 (0.017)
0.44 0.97 0.27 0.20 rs1531343 HMGA2 C/G -0.034 (0.036) -0.12 (0.075) 0.0062 (0.014) -0.0042 (0.034)
0.35 0.11 0.65 0.90 rs231362 KCNQ1 G/A -0.021 (0.020) -0.18 (0.041) 0.0065 (0.0091) 0.0096 (0.017)
0.28 1.63×10-5 0.47 0.58
rs163184 KCNQ1 G/T 0.015 (0.021) - 0.0076 (0.017) 0.041 (0.027)
0.48 - 0.66 0.12 rs6795735 ADAMTS9 C/T -0.0045 (0.020) 0.031 (0.051) -0.0017 (0.0095) 0.012 (0.018)
0.82 0.54 0.86 0.52
Page 52 of 73Diabetes
16
Other rs5945326 DUSP9 A/G 0.069 (0.047) -0.019 (0.038) -0.018 (0.015) - 0.15 0.63 0.23 -
Page 53 of 73 Diabetes
17
(A) Insulinogenic index at HHEX/IDE
(B) Insulinogenic index atCDKAL1
Page 54 of 73Diabetes
18
(C) 32,33 split proinsulin at HNF1A
(D) AIR at MTNR1B
Page 55 of 73 Diabetes
19
(E) AIR at KCNQ1
(F) AIR at CDKN2A/B
Page 56 of 73Diabetes
20
Supplementary Figure 1. Forest plots for meta-analyses of insulinogenic index, split
proinsulin and acute insulin response (AIR). (A) insulinogenic index at HHEX/IDE, (B) insulinogenic index at CDKAL1, (C) 32,33 split proinsulin at HNF1A, (D) AIR at MTNR1B, (E) AIR at KCNQ1, (F) AIR at CDKN2A/B.
Page 57 of 73 Diabetes
21
(A) IV insulin sensitivity traits at IRS1
(B) IV insulin sensitivity traits at ADCY5
Supplementary Figure 2. Forest plots for meta-analyses of intravenous (IV) insulin sensitivity traits (A) IRS1, (B) ADCY5.
Page 58 of 73Diabetes
22
(A) (B)
Supplementary Figure 3. Results of cluster analysis using principal traits. (A) Clustering of ten traits with meta-analysis results from at least 10,000 individuals (principal traits). Traits include: insulinogenic index, proinsulin, ISI Belfiore, ISI Gutt, ISI Matsuda, ISI Stumvoll and four fasting traits: fasting glucose (FG), fasting insulin (FI), HOMA-B, HOMA-IR. Closest links are observed between the four ISIs, which are connected to the
Page 59 of 73 Diabetes
23
HOMA-IR/FI branch and constitute a larger clade of traits linked to insulin sensitivity and resistance. This clade is linked to proinsulin, making up a cluster bearing insulin secretion and sensitivity traits. In turn, this is connected to the HOMA-B/insulinogenic index clade representing effects of type 2 diabetes loci on beta-cell function and early insulin secretion. Finally, the effects of type 2 diabetes loci on FG cluster separately from all other traits. (B) Calinski index computed as a measure of clustering support. The most robust set contained eight trait groups underscoring a clear distinction between most of the investigated physiologic phenotypes.
Page 60 of 73Diabetes
24
Supplementary Figure 4. Results of cluster analysis using all 14 traits. (A) Complete linkage dendrogram of the 14 traits. Cluster analysis confirmed the trait connectivity observed for principal traits and revealed a direct link between AIR and insulinogenic index. C-peptide and split proinsulin were connected directly and linked subsequently to the clade of insulin secretion measurements. IV measures cluster with the ISIs (Matsuda/Belfiore clade). The C-peptide/split proinsulin cluster was connected directly to ISIs/ IV measures and this whole clade was linked subsequently to the group of insulin secretion measurements. (B) Calinski index computed as a measure of clustering support. The most robust set contained twelve trait groups underscoring a clear distinction between the investigated physiologic phenotypes.
(A) (B)
Page 61 of 73 Diabetes
25
Supplementary Figure 5. Principal component analysis (PCA) of type 2 diabetes SNPs. Loci are grouped using centroid-based clustering. Each group is represented using a specific color and group name is shown next to each ellipse (see main text for loci belonging to each group). The two components explain the 64.29% of the total phenotypic variance. Numbers on ellipses correspond to locus groups: 1: hyperglycemic (HG), 2: insulin resistance (IR), 3: proinsulin (PI), 4: beta cell (BC), 5: unclassified (UC).
Page 62 of 73Diabetes
26
(A)
Page 63 of 73 Diabetes
27
(B)
Page 64 of 73Diabetes
28
(C)
Page 65 of 73 Diabetes
29
(D)
Page 66 of 73Diabetes
30
(E)
Page 67 of 73 Diabetes
31
(F)
Supplementary Figure 6. Scatter plots of allelic effect size estimates for selected trait pairs. For each scatter plot, loci are assigned into the groups defined from the cluster analysis of principal traits (groups highlighted by different colours). (A) Insulinogenic index vs. M/I SI SSPG reveals elevated insulinogenic index at IRS1 and a negative correlation of insulinogenic index with the compensatory effects IV measures. (B) Proinsulin vs. HOMA-B, this plot highlights the effect of ARAP1 on proinsulin. (C) Proinsulin vs. 32,33 split proinsulin shows the diminished proinsulin levels at ARAP1. (D)
Page 68 of 73Diabetes
32
Insulinogenic index vs. 32,33 split proinsulin reveals a similar effect for ARAP1 on insulinogenic index. (E) Acute insulin response vs. M/I SI SSPG (combined IV measures): this plot illustrates a weak negative correlation of acute insulin response with compensatory effects IV measures. (F) ISI Belfiore vs. M/I SI SSPG plot reflects good correlation between the two measures of insulin sensitivity. Cluster group colors are HG: orange, IR: green, PI: pink, BC: red, UC: blue.
Page 69 of 73 Diabetes
33
REFERENCES
1. Belfiore F, Iannello S, Volpicelli G: Insulin sensitivity indices calculated from basal and OGTT-induced insulin, glucose, and FFA levels. Mol Genet Metab 1998;63:134-141
2. Gutt M, Davis CL, Spitzer SB, Llabre MM, Kumar M, Czarnecki EM, Schneiderman N, Skyler JS, Marks JB: Validation of the insulin sensitivity index (ISI(0,120)): comparison with other measures. Diabetes Res Clin Pract 2000;47:177-184
3. Matsuda M, DeFronzo RA: Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes care 1999;22:1462-1470
4. Stumvoll M, Mitrakou A, Pimenta W, Jenssen T, Yki-Jarvinen H, Van Haeften T, Renn W, Gerich J: Use of the oral glucose tolerance test to assess insulin release and insulin sensitivity. Diabetes care 2000;23:295-301
Page 70 of 73Diabetes
COHORT ELY EUGENE2 FHS (proinsulin) FHS (Gutt) FUSION MESYBEPO METSIM NHANES PIVUS QTL FAMILIES RISCPARTNERS/ROCH
ESORBS STANFORD ULSAM UNG92
CONTACT PERSON/ANALYST CL ML JF JF RW JS AUJ JBM EI NG CL JF AT/IP
Ke Hao, Fahim
Abbasi, Tim
Assimes, Josh
Knowles
EI NG
Ethnicity European
Northern
European, German
(Finland, Sweden,
Denmark,
germany)
White WhiteEuropean
descentEuropean
Northern
European
(white,
Caucasian)
European descent
Northern
European (white,
Caucasian)
Northern
European (white,
Caucasian)
European
Northern
European (white,
Caucasian)
Sorbs (Slavonic
origin)White, Caucasian
Northern
European (white,
Caucasian)
Northern European
(white, Caucasian)
Country (Sample source) UK USA USA Finland Germany Finland USA Sweden Denmark Multi-Centre USA Germany
United States,
San Francisco
Bay Area
Sweden Denmark
Type (Population-based or case-
control)Population-based Offspring of T2DM
Population-
based
Population-
based
Family-based:
nondiabetic
offspring and
spouses of T2D
affected sibs
population-
based
Population-
basedPopulation-based
Population-
based
Type 2 diabetes
offspringPopulation-based Case-Control Population-based Population based Population-based Population-based
OGTT MEASUREMENTS
Sample (Plasma? Blood?)
Fresh venous
plasma in lithium
heparin or
sodium fluoride
with heparin
(glucose); frozen
at -80C until
measurement
(insulin)
Plasma NA NA NA Plasma Plasma NA NAGlucose: plasma;
insulin: serum
Fresh venous
plasma in lithium
heparin or sodium
fluoride with
heparin
Fasting Plasma Serum NA PlasmaGlucose: plasma;
insulin: serum
Time points for collection0, 30, 60, 120
mins0, 30, 60, 90, 120 NA NA NA 0,30,60,90,120 0, 30, 120 min NA NA
0, 10, 20, 30, 40,
50, 60, 75, 90,
105, 120 mins
0, 30, 60, 90, 120
mins0, 30, 120 mins 0, 30, 120 min NA
0, 30, 60, 90, 120
minsNA
Glucose assay Hexokinase assayGlucose oxidase
colorimetric assayNA NA NA
Hexokinase,
ABX Pentra
Reagents
Glucose
hexokinase
(Konelab
System
Reagents,
Glucose (HK),
Thermo Fisher
Scientific,
Vantaa, Finland)
NA NA
hexokinase/G6P-
DH technique
(Boehringer
Mannheim,
Germany).
Glucose oxidase
method (Cobas
Integra, Roche)
Hexokinase
Hexokinase
method
(Automated
analyser Modular,
Roche Diagnostics,
Mannheim,
Germany)
NA
Glucose
dehydrogenase
method (Gluc-DH,
Merck, Darmstadt,
Germany)
hexokinase/G6P-
DH technique
(Boehringer
Mannheim,
Germany).
Insulin assayImmunometric
assay
Immunoassay
(ADVIA Centaur
Insulin IRI,
Siemens Medical
Solutions
Diagnostics)
NA NA NA ELISA, Mercodia
Immunoassay
(ADVIA Centaur
Insulin IRI,
Siemens
Medical
Solutions
Diagnostics)
NA NA
ELISA (Dako
Diagnostics, Ely,
UK)
Specific time-
resolved
fluroimmunoassa
y (AutoDELFIA
Insulin kit; Wallac
Oy, Turku,
Finland)
Human specific
insulin RIA, Linco
Research Inc., St
Louis MO, USA.
The cross-
reactivity of
Pharmacia insulin
antibody with
proinsulin is 0%
AutoDELFIA Insulin
assay (PerkinElmer
Life and Analytical
Sciences, Turku,
Finland)
NA
Immunoreactive
insulin: Enzymatic-
immunological
assay (Enzymun,
Boehringer
Mannheim)
ELISA (Dako
Diagnostics, Ely,
UK)
Page 71 of 73 Diabetes
Reference
Addenbrooke's
Hospital,
Cambridge
Laakso M, et al:
Insulin sensitivity,
insulin release and
glucagon-like
peptide-1 levels in
persons with
impaired …
Diabetologia
51:502-11, 2008
NA NA NA NA NA NA NA NA
Hills SA, Balkau B,
Coppack SW,
Dekker JM, Mari
A, Natali A,
Walker M,
Ferrannini E; EGIR-
RISC Study Group.
The EGIR-RISC
STUDY (The
European group
for the study of
insulin resistance:
relationship
between insulin
sensitivity and
cardiovascular
disease risk): I.
Methodology and
objectives.
Diabetologia.
2004
Mar;47(3):566-70.
NA NA
Ingelsson E,
Sundström J,
Ärnlöv J, Zethelius
B, Lind L. Insulin
Resistance and
Risk of Congestive
Heart Failure.
JAMA 2005;
294(3):334-341
NA
PROINSULIN MEASUREMENTS
Sample (Fasting? Plasma?
Blood?)Fasting plasma NA Fasting plasma NA NA Fasting plasma NA Fasting plasma NA Fasting plasma NA NA NA Fasting plasma NA
Assay
Monoclonal
antibody-based
two-site
immunoradiomet
ric assay
NALinco humna
proinsulin RIA
Linco humna
proinsulin RIANA NA
RIA (human
proinsulin Ria
kit, Linco
Research, St.
Charles, MO,
USA)
NA
Two-site
immunometric
assay technique
NA
Monoclonal
antibody-based
two-site
immunoradiometr
ic assay
NA NA NA
Two-site
immunometric
assay technique
NA
Reference 2669734 NA
http://www.mill
ipore.com/catal
ogue/item/hpi-
15k
http://www.mill
ipore.com/catal
ogue/item/hpi-
15k
NA NA NA NA
Sobey WJ, Beer
SF, Carrington
CA, Clark PMS,
Hales CN.
Sensitive and
specific two-site
immunoradiome
tric assays for
human insulin,
proinsulin, 65-
66split and 32-
33 split
proinsulins.
Biochem J 1989;
260: 535-541
NA PMID: 17846745 NA NA NA
Sobey WJ, Beer SF,
Carrington CA,
Clark PMS, Hales
CN. Sensitive and
specific two-site
immunoradiometri
c assays for human
insulin, proinsulin,
65-66split and 32-
33 split
proinsulins.
Biochem J 1989;
260: 535-541
NA
C-PEPTIDE MEASUREMENTS NA NA
Sample (Fasting? Blood?
Serum?)Fasting plasma NA NA NA Fasting plasma NA NA Fasting serum NA fasting serum Fasting plasma NA Fasting serum NA NA fasting serum
Assay
A two-step time
resolved
fluorometric
assay with
samples assayed
in singleton on a
1235 AutoDELFIA
automatic
immunoassay
system (Perkin
Elmer, UK)
NA NA NA Novo RIA NA NA RIA NA
Monoclonal
antibody-based
two-site
immunoradiometr
ic assay
NA LIAISON (DiaSorin) NA NA RIA
Page 72 of 73Diabetes
Reference
Christopher P,
Hattersley AT,
Sutton PJ. C-
peptide
measurement:
clinical
applications and
sample storage
using a new wide
range assay.
Proc. ACB Natl.
Meeting, p.69
(1991)
NA NA NA NA NA NA
http://www.cdc.go
v/nchs/data/nhan
es/nhanes3/cdrom
/nchs/manuals/lab
man.pdf
NA NA PMID: 14968294 NA NA NA NA PMID: 1239396
EXCLUSIONS AND NUMBERS OF
SAMPLES PER PHENOTYPE
Exclusions
Fasting plasma
glucose>=7.0
mmol/l
Diabetes (known
diabetes or fasting
plasma glucose >=
7.0 mmol/l or 120
min. glucose
during OGTT >=
11.1 mmol/l)
Non-fasting
individuals,
other diabetic
treatment,
fasting glucose
≥ 7 mmol/L
Non-fasting
individuals,
other diabetic
treatment,
fasting glucose
≥ 7 mmol/L
Type 2 diabetes,
on metab.
medication
Diabetics
(fasting plasma
glucose <=7.0 or
diabetes
treatment)
Diabetics
(fasting plasma
glucose>=7.0 or
diabetes
treatment) or 2-
h glucose at
least 11.1
mmol/l
Diabetics (fasting
plasma
glucose>=7.0 or
diabetes
treatment)
Diabetics (fasting
plasma
glucose>=7.0 or
diabetes
treatment)
Diabetes (known
diabetes or
fasting plasma
glucose > 7.0
mmol/l or 120
min. glucose
during OGTT >
11.1 mmol/l)
Fasting plasma
glucose>=7.0
mmol/l
Diabetics
Diabetics (fasting
plasma
glucose>=7.0 or
diabetes
treatment)
Diabetics (known
diabetes or
fasting plasma
glucose > 7.0
mmol/l ) or
diabetes
treatment
Diabetics (fasting
plasma
glucose>=7.0 or
diabetes
treatment)
Fasting glucose > 7
Samples with M-value: N all
(%males / %females)0 874 (41/59) 0 0 0 0 0 0 0 0 1319 (44.8/55.2) 0 0 0 989(100/0) 0
Samples with SSPG: N all
(%males / %females)0 0 0 0 0 0 0 0 0 0 0 0 0
381 (42.3% male/
57.7% female)0 0
Samples with FSIGT: N all
(%males / %females)0 0 0 0
545 (46.97 /
53.03)0 0 0 0 267 (44.2/55.8) 0 0 0 0 0 376 (49.2/50.8)
Samples with ISI-Stumvoll: N all
(%males / %females)1454 (45.9/54.1) 861 (41/59) 0 0 0
919 (30.69 /
69.31)6888 (100/0) 0 0 272 (43.8/56.2) 1322 (44.8/55.2) 619 (45.2 / 54.8) 820 (40.4/59.6) 0 976 (100/0) 0
Samples with insulinogenic
index: N all (%males / %females)1440 (46.3/53.7) 861 (41/59) 0 0 0
905 (30.28 /
69.72)6941 (100/0) 0 0 268 (44/56) 970 (45.4/54.6) 621 (45.1 / 54.9) 810 (40.4/59.6) 0 979 (100/0) 0
Samples with proinsulin: N all
(%males / %females)1604 (46.3/53.4) 0
5759
(46.7/53.3)0 0 0 5128 (100/0) 0 911 (48.9/51.1) NA 900 (43.8/56.2) 0 0 0 989 (100/0) 0
Samples with C-peptide: N all
(%males / %females)1606 (44.9/55.0) 0 0 0
1005
(44.7/55.3)0 0 1217 (38.4 / 61.6) 0 281 (43.1/56.9) 1465 (44.7/55.3) 0 747 (41.1/58.9) 0 0 376 (49.2/50.8)
Samples with ISI-Belfiore: N all
(%males / %females)1446 (45.9/54.1) 0 0 0 0
912 (30.48 /
69.52)6996 (100/0) 0 0 271 (43.9/56.1) 1278 (44.5/55.5) 616 (45.1 / 54.9) 812 (40.4/59.6) 0 989(100/0) 0
Samples with ISI-Matsuda: N all
(%males / %females)1315 (46.0/54.0) 853 (41/59) 0 0 0
902 (30.27 /
69.73)6996 (100/0) 0 0 281 (42.7/57.3) 1158 (44.9/55.1) 628 (44.9 / 55.1) 812 (40.4/59.6) 0 986 (100/0) 0
Samples with ISI-Gutt: N all
(%males / %females)1480 (45.5/54.5) 860 (41/59) 0
2604
(45.8/54.2)0
1090 (32.39 /
67.61)7028 (100/0) 0 0 273 (43.6/56.4) 1289 (44.6/55.4) 612 (44.9 / 55.1) 816 (40.3/59.7) 0 991 (100/0) 0
PHENOTYPE DISTRIBU TIONS(ALLVALUESINORIGINALUNITSBEFORETRANSFORMATIONS)
Age [Mean (sd) males / Mean
(sd) females], years
61.23 (9.18) /
60.55 (9.08)
40.4 (10.5) / 39.9
(10.3)
48.5 (13.4) /
48.9 (13.8)
54.0 (9.9) / 54.1
(9.8)
44.37 (13.37) /
47.25 (13.46)
51.28 (14.64) /
49.75 (13.75)
57.07 (6.98) /
NA
51.2 (20.6) / 51.5
(20.3)
70.13 (0.17) /
70.26 (0.15)
41.4 (11.8) / 44.3
(12.6)
43.41 (8.48) /
44.60 (8.19)
52.68 (12.96) /
52.53 (12.80)
47.8 (16.6) / 48.1
(16.1)52 (9) / 49 (10) 70.99 (0.63) / NA
25.5 (3.45) / 25
(3.55)
BMI [Mean (sd) males / Mean
(sd) females], kg/m2
27.13 (3.86)
/27.06 (5.36)
26.9 (4.26) / 26.4
(5.35)
28.2 (4.3) / 26.8
(5.8)
27.8 (3.9) / 26.1
(4.9)
27.39 (4.19) /
27.28 (5.50)
28.40 (5.50) /
29.44 (6.51)
26.79 (3.76) /
NA
26.9 (4.8) / 26.4
(5.8)
26.82 (3.64) /
26.85 (4.66)
26.3 (4.1) / 26.3
(5.07)
26.52 (3.53) /
24.93 (4.40)
27.76 (5.19) /
27.08 (7.23)
26.5 (4.8) / 27.2
(5.1)
30.3 (5.1) / 30.1
(5.6)25.99(3.20) / NA
24.1 (3.38) / 23
(3.91)
Fasting plasma glucose [Mean
(sd) males / Mean (sd) females],
mmol/l
5.10 (0.53) / 4.88
(0.53)
5.29 (0.52) / 4.95
(0.50)
5.46 (0.48) /
5.15 (0.50)
5.36 (0.47) /
5.12 (0.50)
5.36 (0.54) /
5.10 (0.53)
5.75 (0.57) /
5.68 (0.58)5.70 (0.48) / NA
5.39 (0.49) / 5.15
(0.55)
5.02 (0.55) / 4.95
(0.56)
5.26 (0.501) /
5.07 (0.538)
5.24 (0.53) / 4.95
(0.59)
4.97 (0.47) / 4.77
(0.49)5.4 (0.5) / 5.4 (1.1) 100 (16) / 96 (12) 5.37 (0.56) / NA
5.15 (0.466) / 4.82
(0.368)
Fasting insulin [Mean (sd) males
/ Mean (sd) females], pmol/l
58.75 (34.04) /
54.21 (34.78)
52.0 (57.5) / 49.9
(54.8)
25.7 (19.2) /
22.5 (16.9)
30.3 (11.2) /
27.2 (8.7)
76.83 (46.30) /
72.48 (53.06)
54.13 (39.16) /
49.31 (30.55)
49.09 (33.9) /
NA
62.4 (44.0) / 57.9
(36.9)
52.73 (32.66) /
50.24 (30.61)
42.5 (30.2) / 42.5
(33.3)
38.21 (20.83) /
34.26 (19.42)
88.49 (62.33) /
81.12 (77.59)
42.3 (32.7) / 42.3
(29.1)NA 73.25 (40.25) / NA
35 (21.1) / 39.3
(23)
M-value [Mean (sd) males /
Mean (sd) females]NA
39.5 (15.8) / 43.2
(16.8)NA NA NA NA NA NA NA NA
124.1 (64.2) /
155.1 (74.4)NA NA NA 5.46 (1.94) / NA NA
Original units NA umol/kg/min NA NA NA NA NA NA NA NAmmol*min-1*kg
ffm-1*nM-1NA NA NA mg/kg bw/min NA
SSPG [Mean (sd) males / Mean
(sd) females]NA NA NA NA NA NA NA NA NA NA NA NA NA
163 (71) / 155
(74)NA NA
Original units NA NA NA NA NA NA NA NA NA NA NA NA NA mg/dl NA NA
Si from FSIGT [Mean (sd) males /
Mean (sd) females]NA NA NA NA
7.04 (4.61) /
7.52 (4.34)NA NA NA NA
9.64 (6.18) / 10.7
(6.1)NA NA NA NA NA
15.3 (8.84) / 15.2
(9.71)
Original units NA NA NA NA x10-5 min-1/pM NA NA NA NA x10-5 min-1/pM NA NA NA NA NA x10-5 min-1/pM
ISI-Stumvoll [Mean (sd) males /
Mean (sd) females]
0.085 (0.029) /
0.089 (0.033)
0.092 (0.027) /
0.091 (0.028)NA NA NA
0.0781 (0.0338)
/ 0.0749
(0.0324)
0.089 (0.024) /
NANA NA
0.0941 (0.0258)
/ 0.0963
(0.0285)
0.096 (0.024) /
0.103 (0.026)
0.085 (0.031) /
0.088 (0.040)
0.1 (0.02) / 0.1
(0.03)NA 0.082 (0.028) / NA NA
Page 73 of 73 Diabetes
Insulinogenic index [Mean (sd)
males / Mean (sd) females],
pmol/mmol
111.4 (88.3) /
139.4 (205.3)
117.0 (114.7) /
132.7 (484.3)NA NA NA
126.33 (209.45)
/ 102.23
(128.18)
136.5 (190.2) /
NANA NA
86.3 (55.1) / 113
(98.4)
7.92 (10.42) /
9.81 (14.75)
114.8 (386.9) /
163.3 (383.8)
80.2 (67.9) / 145.6
(357.4)NA 88.88 (70.26) / NA NA
Proinsulin [Mean (sd) males /
Mean (sd) females], pmol/l
4.82 (3.52) / 3.77
(2.75)NA
12.3 (10.9) / 9.2
(7.5)NA NA NA
13.31 (6.16) /
NANA
10.33 (7.82) /
8.48 (7.03)NA
9.76 (11.30) /
7.27 (9.64)NA NA NA 7.16 (5.61) / NA NA
C-peptide [Mean (sd) males /
Mean (sd) females], pmol/l
705.0 (282.0) /
639.5 (282.0)NA NA NA
1.77 (0.83) /
1.65 (0.77)NA NA
740.0 (406.6) /
676.6 (420.1)NA
521 (196) / 523
(200)
636.1 (299.1) /
561.3 (257.7)NA
6.5 (0.36) / 6.48
(0.35)NA NA
454 (156) / 493
(160)
ISI-Belfiore [Mean (sd) males /
Mean (sd) females]
1.03 (0.28) / 1.09
(0.28)NA NA NA NA
0.9692 (0.3051)
/ 0.9885
(0.2670)
1.02 (0.30) /
NANA NA
1.12 (0.29) / 1.15
(0.273)
1.55 (0.25) / 1.55
(0.23)
0.98 (0.30) / 1.03
(0.30)
1.2 (.27) / 1.18
(.26)NA 0.96 (0.27) / NA NA
ISI-Matsuda [Mean (sd) males /
Mean (sd) females]
5.78 (3.38) / 6.58
(4.08)
2.64 (1.98) / 2.80
(1.79)NA NA NA
6.015 (4.26) /
6.012 (3.53)
6.96 (4.13) /
NANA NA
7.01 (4.03) / 7.6
(4.86)
8.21 (5.14) / 9.25
(5.20)
5.13 (3.30) / 5.78
(3.16)
10.18 (6.9) / 9.3
(5.31)NA 4.21 (2.60) / NA NA
ISI-Gutt [Mean (sd) males /
Mean (sd) females]
14.37 (4.41) /
14.76 (4.30)
81.2 (222.8) / 86.1
(29.4)NA
26.8 (7.2) / 26.2
(6.7)NA
53.77 (27.10) /
46.67 (24.12)
50.13 (22.26) /
MANA NA
62.3 (48.1) / 51.3
(18.1)
16.30 (5.37) /
15.95 (4.36)
26.96 (8.38) /
26.99 (7.33)
80.01 (50.88) /
60.76 (31.82)NA 39.31 (13.82 / NA NA
GENOTYPING
In silico/De novo De novo In silico In silico In silico De novo De novo De novo De novo De novo De novo De novo De novo In silico In silico De novo De novo
Genotyping platform & SNP
panelSequenom iPLEX
Illumina Infinium
HumanHap550
BeadChip
Affymetrix 500K
and MIPS 50K
Affymetrix 500K
and MIPS 50K
iPLEX
Sequenom
MassARRAY
Oxford
iPLEX
Sequenom
MassARRAY
Sequenom iPLEXIllumina Golden
Gate assay
KASPar SNP
genotypingKASPar Chemistry Sequenom iPLEX
500K Affymetrix
GeneChip (250K
Sty and 250K Nsp
arrays, Affymetrix,
Inc) and Affymetrix
Genome-Wide
Human SNP Array
6.0
Affymetrix SNP
Array 6.0
Illumina Golden
Gate assay
KASPar SNP
genotyping
Genotyping centreWTSI/ MRC
Epidemiology
Finnish Genome
CenterAffymetrix Affymetrix NHGRI Oxford NHGRI Broad Institute
Uppsala SNP
Technology
Platform
Kbiosciences KBioscience Broad Institute
Microarray Core
Facility of the
Interdisciplinary
Centre for Clinical
Research,
University of
Leipzig, Germany
and ATLAS Biolabs
GmbH, Berlin,
Germany
Rosetta
Uppsala SNP
Technology
Platform
Kbiosciences
Genotyping calling algorithm Sequenom
Illumina
Beadstation
Genotyping
Solution
BRLMM BRLMM
Sequenom
MassArray
Typer 3.4
NA
Sequenom
MassArray
Typer 3.4
SequenomGenCall
(Illumina)KASPar KASPar Sequenom
BRLMM algorithm
(Affymetrix, Inc)
for 500K and
Birdseed Algorithm
for Genome-Wide
Human SNP Array
6.0
BirdSuite GenCall (Illumina) KASPar
SAMPLE QC
Call rate [filter detail / N
individuals excluded]94%-99.2% > 0.98 ≥ 97% / 79 ≥ 97% / 77 NA 94.3% - 97.3% > 96-97% 98.7% (60% / 86) ≥ 90% / 0 ≥95% / 0 99.7% (96% / 14) >0.94 > 0.98 ≥ 90% / 7
Ethnic outliers excluded - Yes - - NA NA NA - - - - - Yes Yes - -
Other exclusions -Duplicates, gender
mismatch
Heterozygosity
5 SD from mean
(< 25.758% or >
29.958%) or
excess
mendelian
inconsistencies
/ n=10
Heterozygosity
5 SD from mean
(< 25.758% or >
29.958%) or
excess
mendelian
inconsistencies
/ n=13
NA NA NA - - - - -Duplicates, gender
mismatch
Duplicates,
gender mismatch- -
Individuals for analysis 1612 595 5759 2604 1546 1130 7043 912 911 286 1428 648 921 381 1030 376
SNP QC (prior to imputation)
MAF [filter detail / N SNPs
excluded]1% / 0 < 2% 1% / 68 953 1% / 68 953 1% / 0 1% / 0 1% / 0 1% / 0 1% / 0 1% / 0 1% / 0 1% / 0 1% < 2% 1% / 0 1% / 0
HWE [filter detail / N SNPs
excluded]0.01 / 0 1 x 10-6 10-6 / 20 999 10-6 / 20 999 0.01 / 1 0.01 / 1 0.01 / 2 0.01 / 0 0.001 / 0 0.001 / 0 0.01 / 0 0.01 / 0 10-4 1 x 10-6 0.001 / 0 0.001 / 0
Call rate [filter detail / N SNPs
excluded]90% / 0 < 90% 95% / 23 312 95% / 23 312 95% / 0 92.5% / 0 90% / 0 90% / 0 95% / 0 95% / 0 90% / 0 90% / 0 0.95 < 90% 95% / 0 95% / 0
Other NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA
SNP number in QC'd dataset NA 378,163 378,163 NA NA NA NA NA NA NA NA 378,513 NA NA
IMPUTATION STATS
Imputation software NA MACH MACH MACH NA NA NA NA NA NA NA NA IMPUTE MACH NA NA
Page 74 of 73Diabetes
Imputation quality metrics NASNPs of MACH r^2
< 0.3 are excludedr2hat > 0.3 r2hat > 0.3 NA NA NA NA NA NA NA NA Proper-info > 0.4
SNPs of MACH
r^2 < 0.3 are
excluded
NA NA
Other SNP QC filters applied? NA
SNPs of MAF < 5%
and HWE<1E-6 are
excluded
MAF > 1% MAF > 1% NA NA NA NA NA NA NA NAMAF>1%, HWE<10-
4
SNPs of MAF <
5% and HWE<1E-
6 are excluded
NA NA
DATA ANALYSIS
Number of SNPs in analysis / N
imputed16 / 0 38 /0 19 / 17 19 / 17 16 / 0 32 / 0 18 / 0 18 / 0 61 / 0 36 / 0 16 / 0 18 / 0 19 / 17 38 / 0 61 / 0 36 / 0
Trait transformation M-value NAZ score
transformationNA NA NA NA NA NA NA NA
Z-score
transformationNA NA NA
Z-score
transformationNA
Trait transformation SSPG NA NA NA NA NA NA NA NA NA NA NA NA NAZ-score
transformationNA NA
Trait transformation FSIGT NA NA NA NAZ-score
transformationNA NA NA NA Natural log NA NA NA NA NA Natural log
Trait transformation ISI-Stumvoll Natural log NA NA NA NA Natural log Natural log NA NA Natural log Natural log natural log Natural log NA Natural log NA
Trait transformation
insulinogenic indexNatural log NA NA NA NA Natural log Natural log NA NA Natural log Natural log natural log Natural log NA Natural log NA
Trait transformation proinsulin Natural log NA Natural log Natural log NA Natural log Natural log Natural log Natural log NA NA NA NA Natural log NA
Trait transformation C-peptide Natural log NA NA NA Natural log Natural log NA NA NA Natural log Natural log NA Natural log NA NA Natural log
Trait transformation ISI-Belfiore Natural log NA NA NA NA Natural log Natural log NA NA Natural log Natural log natural log Natural log NA Natural log NA
Trait transformation ISI-Matsuda Natural log NA NA NA NA Natural log Natural log NA NA Natural log Natural log natural log Natural log NA Natural log NA
Trait transformation ISI-Gutt Natural log NA NA NA NA Natural log Natural log NA NA Natural log Natural log natural log Natural log NA Natural log NA
Adjustments in first model Age, sexage, sex and age,
sex, BMI
Age, sex, natural
log of fasting
insulin
Age, sex, natural
log of fasting
insulin
Age, sex Age, sex Age Age, sex Age, sex Age, sex, family Age, sex, centre Age + Sex Age, sexage, sex and age,
sex, BMIAge Age, sex
Analysis method Linear regression Linear regressionLinear mixed
effect model
Linear mixed
effect model
Linear
regression
Linear
regression
Linear
regressionGLM Linear regression Mixed model Linear regression GLM Linear regression Linear regression Linear regression Linear regression
Software for analysis STATA 10.1 R 2.9.2R LMEKIN
package
R LMEKIN
packageMerlin SPSS 18.0 Merlin SAS 9.1.3 STATA 10.1 SPSS STATA 10.1 SAS 9.1.9 GWAMA R 2.9.2 STATA 10.1 R 2.10.0
REFERENCES
Study sample reference PMID: 17257284
Laakso M, et al:
Insulin sensitivity,
insulin release and
glucagon-like
peptide-1 levels in
persons with
impaired …
Diabetologia
51:502-11, 2008
– -
Fischer A, Fisher
E, Mˆhlig M,
Schulze M,
Hoffmann K,
Weickert MO,
Schueler R,
Osterhoff M,
Pfeiffer AFH,
Boeing H,
Spranger J:
KCNJ11 E23K
Affects Diabetes
Risk and Is
Associated With
the Disposition
Index. Diabetes
Care 31:87-89,
2008
Stancakova A,
Javorsky M,
Kuulasmaa T,
Haffner SM,
Kuusisto J,
Laakso M;
Changes in
insulin
sensitivity and
insulin release
in relation to
glycemia and
glucose
tolerance in
6414 Finnish
men. DIabetes
2009
May;58(5):1212-
21. Epub 2009
Feb 17
Ingelsson E,
Hulthe J, Lind L.
Inflammatory
markers in
relation to
insulin resistance
and the
metabolic
syndrome. Eur J
Clin Invest
2008;38(7):502-
9
Pubmed ID:
14968294
Ai M et al Clin
Chim Acta. 2009
May 22. [Epub
ahead of print]
Tönjes et al., EJHG,
2009 (in press -
PMID: 19584900),
Tönjes et al.,
HumMolGenet,
2009 (in press)
Ryan MC, Fenster
Farin HM, Abbasi
F, Reaven GM:
Comparison of
waist
circumference
versus body mass
index in
diagnosing
metabolic
syndrome and
identifying
apparently
healthy subjects
at increased risk
of cardiovascular
disease. Am J
Cardiol 102:40-
46, 2008
Zethelius B, Byberg
L, Hales CN, Lithell
H, Berne C.
Proinsulin and
acute insulin
response
independently
predict Type 2
diabetes mellitus
in men--report
from 27 years of
follow-up study.
Diabetologia
2003;46(1):20-6
PMID: 8787683
Website
http://www.mrc-
epid.cam.ac.uk/S
tudies/Ely/
NA
http://www.ncb
i.nlm.nih.gov/pr
ojects/gap/cgi-
bin/study.cgi?st
udy_id=phs000
007.v4.p2
http://www.ncb
i.nlm.nih.gov/pr
ojects/gap/cgi-
bin/study.cgi?st
udy_id=phs000
007.v4.p2
- NA NA
http://www.cdc.go
v/nchs/nhanes.ht
m
http://www.med
sci.uu.se/pivus/p
ivus.htm
NAhttp://www.egir.o
rg/egirrisc/-
http://innere.unikli
nikum-
leipzig.de/_forschu
ng/schwerpunkte/
sorbs.html
NAhttp://www.pubca
re.uu.se/ULSAMNA
Page 75 of 73 Diabetes
top related