Distinguishing patients with monogenic diabetes from those with type 1 diabetes (T1D) is important for correct diagnosis, treatment, and selection of patients for gene discovery studies. We assessed whether a T1D genetic risk score (T1D-GRS) generated from T1D-associated common genetic variants provides a novel way to discriminate monogenic diabetes from T1D. The T1D-GRS was highly discriminative of proven maturity-onset diabetes of young (MODY) (n = 805) and T1D (n = 1,963) (receiver operating characteristic area under the curve 0.87). A T1D-GRS of >0.280 (>50th T1D centile) was indicative of T1D (94% specificity, 50% sensitivity). We then analyzed the T1D-GRS of 242 white European patients with neonatal diabetes (NDM) who had been tested for all known NDM genes. Monogenic NDM was confirmed in 90, 59, and 8% of patients with GRS <5th T1D centile, 50–75th T1D centile, and >75th T1D centile, respectively. Applying a GRS 50th T1D centile cutoff in 48 NDM patients with no known genetic cause identified those most likely to have a novel monogenic etiology by highlighting patients with probable early-onset T1D (GRS >50th T1D centile) who were diagnosed later and had less syndromic presentation but additional autoimmune features compared with those with proven monogenic NDM. The T1D-GRS is a novel tool to improve the use of biomarkers in the discrimination of monogenic diabetes from T1D.

Distinguishing patients with monogenic diabetes from those with the more common type 1 diabetes (T1D) has important clinical and scientific implications. Many subtypes of monogenic diabetes can be treated with sulfonylurea tablets, either low dose (HNF1A/HNF4A subtype) (1) or high dose (KCNJ11/ABCC8 subtype) (2), whereas patients with T1D are treated with lifelong insulin injections. Identifying new genetic causes of diabetes can lead to the discovery of biologically important pathways and may provide potential therapeutic targets. The genetic etiology is unknown in up to 20% of patients with maturity-onset diabetes of young (MODY) and neonatal diabetes (NDM) (3,4). Therefore, novel tools that facilitate the discovery of new genetic causes of diabetes are useful, particularly ones that discriminate monogenic diabetes from T1D.

T1D is a polygenic disease that shows well-characterized strong genetic predisposition from HLA and non-HLA loci (5). Large-scale genome-wide association studies have identified common genetic variants (single nucleotide polymorphisms [SNPs]) in HLA and more than 40 non-HLA genes that contribute to T1D genetic susceptibility (57). Genotyping these variants and combining the risk of each inherited allele to create a type 1 diabetes genetic risk score (T1D-GRS) allows us to accurately capture an individual’s polygenic susceptibility to T1D (8,9). The T1D-GRS has been shown to predict development of T1D and islet autoantibodies (8,10) and to predict the development of insulin deficiency in patients with diabetes diagnosed between 20 and 40 years of age (9).

In this study, we assessed whether a T1D-GRS generated from 30 common T1D-associated SNPs can provide a novel way to discriminate monogenic diabetes from T1D. We first assessed the ability of the GRS to discriminate T1D and confirmed MODY. We then assessed its utility in patients with NDM diagnosed before 6 months of age to identify 1) patients with a known monogenic etiology and 2) patients with a high likelihood of a novel monogenic etiology.

Study Populations

Patients With T1D

T1D patients were from the Wellcome Trust Case Control Consortium (WTCCC) and have been previously described in detail (11). In brief, the WTCCC T1D subjects (n = 1,963) were all clinically diagnosed with T1D before 17 years of age and treated with insulin from diagnosis. Patients with known MODY or NDM were excluded.

Patients With Confirmed MODY

We included 805 white European MODY patients with a confirmed monogenic etiology on genetic testing (415 patients with HNF1A MODY, 346 with GCK MODY, 42 with HNF4A MODY, and 2 with HNF1B MODY). The median age of diagnosis was 20 years (interquartile range 15, 30), and 532 patients were female.

Patients With NDM

We included 242 white European patients with NDM diagnosed before 6 months of age. They had undergone comprehensive genetic testing of all 23 known NDM genes using a targeted next-generation sequencing panel as previously described (3,12). A monogenic etiology was confirmed in 80% (n = 194) of patients. These patients had mutations in KCNJ11 (n = 79), ABCC8 (n = 35), INS (n = 30), 6q24 (methylation defect, n = 23), GATA6 (n = 15), EIF2AK3 (n = 5), and FOXP3 (n = 4) and one mutation each in GATA4, GCK, and PTF1A. This comprehensive genetic testing did not identify a mutation in the remaining 20% of subjects (n = 48).

Generating T1D-GRS for WTCCC, MODY, and NDM Patients

The T1D-GRS was generated using 30 SNPs as previously described (Supplementary Table 1) (9). We excluded samples where genotyping results were missing for any of the alleles that had the greatest influence on the GRS (DR3/DR4-DQ8 or HLA_DRB1_15) or >2 other SNPs.

Statistical Analysis

The appropriate parametric (Student t) and nonparametric (Mann-Whitney U and Kruskal-Wallis) tests were used to compare the continuous variables. Fisher exact test was used to compare categorical variables. Logistic regression and receiver operating characteristic (ROC) curve analysis were used to measure the discriminatory power of the T1D-GRS. Statistical analysis was carried out using Stata 13.1 (StataCorp LP, College Station, TX).

Ethics Approval

The study was approved by the Genetic Βeta Cell Research Bank, Exeter, U.K. with ethics approval from the North Wales Research Ethics Committee, U.K.

T1D-GRS Was Highly Discriminatory of MODY and T1D

T1D-GRS of the MODY cohort was lower (mean GRS 0.231, 95% CI 0.228–0.233) than that of the WTCCC T1D cohort (0.279, 95% CI 0.278–0.280, P = 4 × 10−136) but similar to that of the WTCCC control subjects (0.229, 95% CI 0.228–0.230, P = 0.26) (Fig. 1A). The analysis of the T1D-GRS using the ROC curve showed that the T1D-GRS was highly discriminatory between MODY and T1D (ROC area under the curve 0.87, 95% CI 0.86–0.89) (Fig. 1B).

Figure 1

T1D-GRS is higher in T1D patients than in confirmed MODY patients and control subjects without diabetes. A: Dot plot of T1D-GRS from 1,963 T1D patients, 805 MODY (subtype of monogenic diabetes) patients, and 2,938 control subjects without diabetes. The blue bar highlights the median T1D-GRS for each group. The upper red horizontal line shows a GRS equivalent to 50th centile for the T1D cohort, and the lower red horizontal line shows a GRS equivalent to 5th centile for the T1D cohort. The control subjects and T1D patients came from WTCCC (11). B: ROC curve (area under the curve 0.87) for T1D-GRS to discriminate MODY (n = 805) and T1D (n = 1,963).

Figure 1

T1D-GRS is higher in T1D patients than in confirmed MODY patients and control subjects without diabetes. A: Dot plot of T1D-GRS from 1,963 T1D patients, 805 MODY (subtype of monogenic diabetes) patients, and 2,938 control subjects without diabetes. The blue bar highlights the median T1D-GRS for each group. The upper red horizontal line shows a GRS equivalent to 50th centile for the T1D cohort, and the lower red horizontal line shows a GRS equivalent to 5th centile for the T1D cohort. The control subjects and T1D patients came from WTCCC (11). B: ROC curve (area under the curve 0.87) for T1D-GRS to discriminate MODY (n = 805) and T1D (n = 1,963).

High Specificity of T1D-GRS Cutoffs Based on T1D Centiles

To further understand the discriminatory power of the T1D-GRS, MODY and T1D patients were categorized using a T1D-GRS equivalent to the 5th, 25th, 50th, and 75th centile of the T1D cohort (Table 1). Only 6% of the MODY patients had T1D-GRS >0.280 (>50th T1D centile) compared with 50% of T1D patients, and thus this level of GRS was indicative of T1D with 94% specificity and 50% sensitivity (Table 1). Conversely, 53% of the MODY patients had T1D-GRS <0.234 (<5th T1D centile) compared with only 5% of the T1D patients, providing 50% sensitivity and 95% specificity to identify MODY (Table 1).

Table 1

T1D-GRS is discriminatory for T1D and MODY

Categories based on GRS centiles of T1D cohort (actual score value)T1D, n (%)Confirmed MODY, n (%)TotalOdds ratio for T1D (95% CI)
<5th centile (<0.234)
 
98 (5)
 
423 (53)
 
521
 
1
 
5–25th centile (0.234–0.262)
 
392 (20)
 
242 (30)
 
634
 
7 (5–9)
 
26–50th centile (0.263–0.280)
 
492 (25)
 
89 (11)
 
581
 
24 (16–36)
 
51–75th centile (0.281–0.299)
 
491 (25)
 
42 (5)
 
533
 
50 (29–89)
 
>75th centile (>0.299)
 
490 (25)
 
9 (1)
 
499
 
235 (73–755)
 
Total 1,963 (100) 805 (100) 2,768  
Categories based on GRS centiles of T1D cohort (actual score value)T1D, n (%)Confirmed MODY, n (%)TotalOdds ratio for T1D (95% CI)
<5th centile (<0.234)
 
98 (5)
 
423 (53)
 
521
 
1
 
5–25th centile (0.234–0.262)
 
392 (20)
 
242 (30)
 
634
 
7 (5–9)
 
26–50th centile (0.263–0.280)
 
492 (25)
 
89 (11)
 
581
 
24 (16–36)
 
51–75th centile (0.281–0.299)
 
491 (25)
 
42 (5)
 
533
 
50 (29–89)
 
>75th centile (>0.299)
 
490 (25)
 
9 (1)
 
499
 
235 (73–755)
 
Total 1,963 (100) 805 (100) 2,768  

The groups were categorized by GRS centiles of the T1D cohort. The odds ratios are relative to <5th centile of T1D-GRS category.

T1D-GRS Can Identify Patients With NDM Who Had Monogenic Etiology on Comprehensive Genetic Testing

To further assess the utility of the T1D-GRS, we compared the result of comprehensive genetic testing for NDM with different values of T1D-GRS. In the 242 patients with NDM, comprehensive genetic testing showed that 81% (194/242) had monogenic NDM. Similar to MODY patients, 54% of proven monogenic NDM patients had T1D-GRS <0.234 (<5th T1D centile) (Supplementary Fig. 1). Out of 117 patients with T1D-GRS <5th T1D centile, 90% (105/117) were shown to have monogenic NDM (Fig. 2A). The proportion of proven monogenic NDM was reduced to 59% (13/22) for patients with T1D-GRS 50–75th T1D centile and only 8% (1/13) in patients with T1D-GRS >75th T1D centile (P = 1.5 × 10−11) (Fig. 2A).

Figure 2

T1D-GRS identified proven and probable monogenic NDM. A: T1D-GRS identified proven monogenic NDM. A total of 242 white European patients with NDM (diagnosed before 6 months of age) who had comprehensive genetic testing for all 23 NDM genes were grouped by their T1D-GRS (12). The subgroups of T1D-GRS are based on centiles of the T1D cohort: low to high score groups (left to right). The bar chart shows the proportion of proven monogenic NDM (total n = 194, patients with positive genetic test, black bar) and NDM with unknown etiology (total n = 48, NDMX patients, white bar) for each score category. The total number of patients is also shown for each category. B: T1D-GRS identified probable monogenic NDM and probable T1D in NDMX patients. The T1D-GRS distribution of NDMX patients (n = 48) showed bimodal distribution with peaks at <5th T1D centile and >75th T1D centile with the nadir around 50th T1D centile. T1D-GRS 50th T1D centile categorized this group as probable T1D (T1D-GRS >50th T1D centile, n = 21) and probable monogenic diabetes due to mutations in as-yet undiscovered disease genes (T1D-GRS ≤50th T1D centile, n = 27).

Figure 2

T1D-GRS identified proven and probable monogenic NDM. A: T1D-GRS identified proven monogenic NDM. A total of 242 white European patients with NDM (diagnosed before 6 months of age) who had comprehensive genetic testing for all 23 NDM genes were grouped by their T1D-GRS (12). The subgroups of T1D-GRS are based on centiles of the T1D cohort: low to high score groups (left to right). The bar chart shows the proportion of proven monogenic NDM (total n = 194, patients with positive genetic test, black bar) and NDM with unknown etiology (total n = 48, NDMX patients, white bar) for each score category. The total number of patients is also shown for each category. B: T1D-GRS identified probable monogenic NDM and probable T1D in NDMX patients. The T1D-GRS distribution of NDMX patients (n = 48) showed bimodal distribution with peaks at <5th T1D centile and >75th T1D centile with the nadir around 50th T1D centile. T1D-GRS 50th T1D centile categorized this group as probable T1D (T1D-GRS >50th T1D centile, n = 21) and probable monogenic diabetes due to mutations in as-yet undiscovered disease genes (T1D-GRS ≤50th T1D centile, n = 27).

NDM With Unknown Etiology Group Showed Bimodal Distribution of T1D-GRS

We next assessed the utility of T1D-GRS to indicate the etiology of diabetes in the 48 patients in whom comprehensive genetic testing was negative (NDMX, unknown etiology). The T1D-GRS in this group (mean T1D-GRS 0.260, 95% CI 0.248–0.275) was intermediate between proven monogenic NDM (0.228, 95% CI 0.222–0.233) and T1D (0.279, 95% CI 0.278–0.280). There was a bimodal distribution of T1D-GRS in NDMX with peaks at <5th T1D centile and >75th T1D centile with a nadir close to T1D-GRS 0.280 (50th T1D centile) (Fig. 2B). This suggests that the NDMX group included patients with monogenic NDM due to mutations in as-yet undiscovered disease genes, as well as those with T1D.

T1D-GRS Identified Patients With Probable Monogenic NDM and Early-Onset T1D

By dividing the NDMX patients by T1D-GRS 50th T1D centile (the nadir), we identified 21/48 (44%) patients with a high genetic predisposition for T1D (T1D-GRS >50th T1D centile). This along with a negative genetic test suggested that these patients might have very early-onset T1D (probable T1D). In contrast, 27/48 (56%) patients with low T1D-GRS, ≤50th T1D centile, were likely to have monogenic NDM due to a novel genetic cause (probable monogenic NDM) (Fig. 2B).

These T1D-GRS–based etiological groups of NDMX were further supported by their clinical characteristics. The patients with probable T1D presented later (12 weeks vs. 1 week, P = 0.04) and had a higher proportion of other autoimmune disease (14% vs. 1%, P = 0.003) and fewer syndromic presentations (10% vs. 40%, P = 0.007) than patients with proven monogenic NDM. In contrast, the clinical characteristics of patients with probable monogenic NDM were similar (P > 0.05) to those of patients with proven monogenic NDM (Table 2).

Table 2

Clinical characteristics of patients with proven and probable monogenic NDM and probable T1D

Proven monogenic NDM <6 months, n = 194Probable monogenic NDM <6 months (T1D-GRS ≤50th T1D centile), n = 27Probable T1D <6 months (T1D-GRS >50th T1D centile), n = 21P value (proven monogenic vs. probable monogenic)P value (proven monogenic vs. probable T1D)
T1D-GRS, median (IQR)
 
0.231 (0.206–0.251)
 
0.237 (0.210–0.250)
 
0.299 (0.295–0.313)
 
0.86
 
<0.001
 
Age at diagnosis in weeks, median (IQR)
 
5 (1–12)
 
1 (0.6–6)
 
12 (3–20)
 
0.13
 
0.04
 
Male, n (%)
 
111 (57)
 
17 (63)
 
15 (71)
 
0.68
 
0.25
 
Presence of other autoimmune disorders, n (%)
 
1 (1)
 
1 (4)
 
3 (14)
 
0.23
 
0.003
 
Syndromic presentation, n (%)
 
77 (40)
 
9 (33)
 
2 (10)
 
0.67
 
0.007
 
Consanguineous parents, n (%) 3 (2) 1.00 1.00 
Proven monogenic NDM <6 months, n = 194Probable monogenic NDM <6 months (T1D-GRS ≤50th T1D centile), n = 27Probable T1D <6 months (T1D-GRS >50th T1D centile), n = 21P value (proven monogenic vs. probable monogenic)P value (proven monogenic vs. probable T1D)
T1D-GRS, median (IQR)
 
0.231 (0.206–0.251)
 
0.237 (0.210–0.250)
 
0.299 (0.295–0.313)
 
0.86
 
<0.001
 
Age at diagnosis in weeks, median (IQR)
 
5 (1–12)
 
1 (0.6–6)
 
12 (3–20)
 
0.13
 
0.04
 
Male, n (%)
 
111 (57)
 
17 (63)
 
15 (71)
 
0.68
 
0.25
 
Presence of other autoimmune disorders, n (%)
 
1 (1)
 
1 (4)
 
3 (14)
 
0.23
 
0.003
 
Syndromic presentation, n (%)
 
77 (40)
 
9 (33)
 
2 (10)
 
0.67
 
0.007
 
Consanguineous parents, n (%) 3 (2) 1.00 1.00 

NDMX patients were categorized into probable T1D and probable monogenic NDM by T1D-GRS equivalent to 50th centile of T1D cohort. Fisher exact test was used to compare proportions, and Mann-Whitney U test was used to compare continuous variables. IQR, interquartile range.

We describe a novel application of polygenic T1D genetic susceptibility to discriminate monogenic diabetes from type 1 diabetes. We have shown that a T1D-GRS aids the discrimination of MODY and T1D in large white European cohorts. Using this tool in patients with NDM revealed a lower T1D-GRS in those with a genetic diagnosis compared with those of unknown etiology and identified a subset of patients without a monogenic etiology who were likely to have very early-onset T1D.

The T1D-GRS provides an additional tool to discriminate monogenic diabetes from T1D. Islet autoantibody testing allows some discrimination (13); however, islet autoantibodies are absent in ∼10% of T1D patients at the time of diagnosis, and this proportion increases with diabetes duration (14,15). C-peptide can be useful to discriminate between MODY and T1D but may not be helpful close to diagnosis because of the honeymoon period (16). Approximately 8% of T1D patients have significant C-peptide (urine C-peptide–to–creatinine ratio >0.2 nmol/mmol) secretion >5 years after diagnosis, which further reduces the discriminatory power of C-peptide (17). C-peptide also cannot differentiate T1D from NDM, as the latter is usually C-peptide negative (3). A DNA-based test such as the T1D-GRS does not change with time and can provide independent and additive information to autoantibody or C-peptide status (9).

To assess the additional value of the T1D-GRS in the differentiation of MODY from type 1 diabetes, we will need a large prospective study comparing the relative value of clinical features, autoantibodies, and T1D-GRS in discriminating T1D and MODY. The integration of these factors could ultimately result in an improved MODY probability calculator, which presently only uses clinical features (18). Notably, the T1D-GRS does not discriminate between different non-T1D subtypes of diabetes such as type 2 diabetes and monogenic diabetes (Supplementary Fig. 2). Therefore, its utility is in the group of patients where differential diagnosis for monogenic diabetes is T1D and not type 2 diabetes.

This new tool should aid in the discovery of novel genes for monogenic diabetes. The main advantage of the T1D-GRS is its ability to discriminate monogenic diabetes from T1D regardless of subtypes of monogenic diabetes. Therefore, the application of the T1D-GRS in patients in whom known monogenic subtypes have been excluded can help to identify patients with the highest probability of a novel cause of monogenic diabetes. These patients can be prioritized for exome or genome sequencing to maximize the chance of novel discoveries. A similar strategy of selecting patients for genetic testing for monogenic hypercholesterolemia using GRS associated with polygenic hypercholesterolemia has been proposed (19).

We have confirmed that diabetes diagnosed before 6 months of age predominantly has monogenic etiology, with ∼91% of these patients likely to have a known (80%) or novel (11%) cause of monogenic diabetes. This is consistent with previous studies showing that patients diagnosed in the first 6 months of life have the high-risk HLA alleles associated with T1D at a frequency similar to that of control subjects and rarely have autoantibodies (20,21). Currently, all patients with NDM diagnosed before 6 months of age undergo targeted next-generation sequencing for 23 genes regardless of clinical presentation (3,12). This approach has been successful and confirmed a monogenic etiology in 82% of patients. Our data support this approach, and we do not recommend excluding any patients for genetic testing based on the T1D-GRS in this age-group because of the high prior probability of monogenic diabetes. We suggest that the TID-GRS be used to define which patients are likely to be T1D or monogenic NDM in a patient diagnosed before 6 months of age only when the known monogenic causes are excluded.

The finding of probable early onset of T1D prior to 6 months of age was unexpected. Autoimmune disease in neonates is extremely rare and normally due to either transfer of maternal autoantibodies or specific monogenic defects in the immune system (22,23). In addition, the current understanding of the pathophysiology of T1D suggests that autoimmune pathways are not mature enough to cause a full-blown attack on the pancreas before 6–12 months of age (24,25). T1D has a prolonged preclinical phase in the months to years prior to developing clinical T1D (26). This raises the possibility that T1D before 6 months of age is caused by a disproportionate increase in immune-mediated destruction of β-cells due to environmental triggers in either the prenatal or immediate postnatal period or multiple factors such as genetics, environment, and epigenetics all paying a role. Further studies of these rare patients that include islet autoantibodies measurements and detailed HLA assessments will be important and may provide novel understanding of the biology of T1D.

Our study has limitations. We only included patients of white European ethnicity, as this was the racial group with the most information about T1D genetic susceptibility. Further work is required to validate this method in patients of other ethnicities. We have captured the most common and high-risk alleles associated with T1D but not the recently published, less frequent, and lower-risk alleles (5). However, these new variants will only result in minor improvement, as the major susceptibility alleles are already included in our GRS. More fine-mapping and larger studies will identify new variants, and the addition of these variants can only strengthen our results through improved capture of T1D genetic susceptibility.

In conclusion, we have shown that the T1D-GRS discriminates monogenic diabetes of any etiology from T1D. Using the T1D-GRS, we have identified cases of very early-onset T1D presenting before 6 months of age. The T1D-GRS helps to identify patients with a potential novel monogenic etiology so that they can be prioritized for exome or genome sequencing analysis.

M.N.W. and A.T.H. contributed equally to this study.

Funding. K.A.P. and R.A.O. are supported by a National Institute for Health Research (NIHR) Clinical Lectureship award. S.E.F. has a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (105636/Z/14/Z). S.E. and A.T.H. are Wellcome Trust Senior Investigators, and A.T.H. is also supported by a NIHR Senior Investigator award. M.N.W. is supported by the Wellcome Trust Institutional Strategic Support Fund (WT097835MF) and the Medical Research Council (MR/M005070/1). Additional support came from the University of Exeter and the NIHR Exeter Clinical Research Facility. This study makes use of data generated by the WTCCC. A full list of the investigators who contributed to the generation of the data are available from www.wtccc.org.uk. Funding for the project was provided by the Wellcome Trust under award 076113.

The views expressed are those of the authors and not necessarily those of the Wellcome Trust, the National Health Service, the NIHR, or the Department of Health.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.A.P., R.A.O., and M.N.W. researched data and performed statistical analyses. K.A.P. and A.T.H. wrote the first draft of the manuscript, which was modified by all authors. All authors contributed to the discussion and reviewed or edited the manuscript. A.T.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Pearson
ER
,
Starkey
BJ
,
Powell
RJ
,
Gribble
FM
,
Clark
PM
,
Hattersley
AT
.
Genetic cause of hyperglycaemia and response to treatment in diabetes
.
Lancet
2003
;
362
:
1275
1281
[PubMed]
2.
Pearson
ER
,
Flechtner
I
,
Njølstad
PR
, et al.;
Neonatal Diabetes International Collaborative Group
.
Switching from insulin to oral sulfonylureas in patients with diabetes due to Kir6.2 mutations
.
N Engl J Med
2006
;
355
:
467
477
[PubMed]
3.
De Franco
E
,
Flanagan
SE
,
Houghton
JAL
, et al
.
The effect of early, comprehensive genomic testing on clinical care in neonatal diabetes: an international cohort study
.
Lancet
2015
;
386
:
957
963
[PubMed]
4.
McCarthy
MI
,
Hattersley
AT
.
Learning from molecular genetics: novel insights arising from the definition of genes for monogenic and type 2 diabetes
.
Diabetes
2008
;
57
:
2889
2898
[PubMed]
5.
Onengut-Gumuscu
S
,
Chen
WM
,
Burren
O
, et al.;
Type 1 Diabetes Genetics Consortium
.
Fine mapping of type 1 diabetes susceptibility loci and evidence for colocalization of causal variants with lymphoid gene enhancers
.
Nat Genet
2015
;
47
:
381
386
[PubMed]
6.
Barrett
JC
,
Clayton
DG
,
Concannon
P
, et al.;
Type 1 Diabetes Genetics Consortium
.
Genome-wide association study and meta-analysis find that over 40 loci affect risk of type 1 diabetes
.
Nat Genet
2009
;
41
:
703
707
[PubMed]
7.
Bradfield
JP
,
Qu
HQ
,
Wang
K
, et al
.
A genome-wide meta-analysis of six type 1 diabetes cohorts identifies multiple associated loci
.
PLoS Genet
2011
;
7
:
e1002293
[PubMed]
8.
Winkler
C
,
Krumsiek
J
,
Buettner
F
, et al
.
Feature ranking of type 1 diabetes susceptibility genes improves prediction of type 1 diabetes
.
Diabetologia
2014
;
57
:
2521
2529
[PubMed]
9.
Oram
RA
,
Patel
K
,
Hill
A
, et al
.
A type 1 diabetes genetic risk score can aid discrimination between type 1 and type 2 diabetes in young adults
.
Diabetes Care
2016
;39:337–344
[PubMed]
10.
Törn
C
,
Hadley
D
,
Lee
H-S
, et al
.; TEDDY Study Group.
Role of type 1 diabetes associated SNPs on risk of autoantibody positivity in the TEDDY Study
.
Diabetes
2015
;64:1818–1829
[PubMed]
11.
Wellcome Trust Case Control Consortium
.
Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls
.
Nature
2007
;
447
:
661
678
[PubMed]
12.
Ellard
S
,
Lango Allen
H
,
De Franco
E
, et al
.
Improved genetic testing for monogenic diabetes using targeted next-generation sequencing
.
Diabetologia
2013
;
56
:
1958
1963
[PubMed]
13.
McDonald
TJ
,
Colclough
K
,
Brown
R
, et al
.
Islet autoantibodies can discriminate maturity-onset diabetes of the young (MODY) from type 1 diabetes
.
Diabet Med
2011
;
28
:
1028
1033
[PubMed]
14.
Bingley
PJ
.
Clinical applications of diabetes antibody testing
.
J Clin Endocrinol Metab
2010
;
95
:
25
33
[PubMed]
15.
Decochez
K
,
Tits
J
,
Coolens
JL
, et al
.
High frequency of persisting or increasing islet-specific autoantibody levels after diagnosis of type 1 diabetes presenting before 40 years of age. The Belgian Diabetes Registry
.
Diabetes Care
2000
;
23
:
838
844
[PubMed]
16.
Besser
REJ
,
Shepherd
MH
,
McDonald
TJ
, et al
.
Urinary C-peptide creatinine ratio is a practical outpatient tool for identifying hepatocyte nuclear factor 1-α/hepatocyte nuclear factor 4-α maturity-onset diabetes of the young from long-duration type 1 diabetes
.
Diabetes Care
2011
;
34
:
286
291
[PubMed]
17.
Oram
RA
,
McDonald
TJ
,
Shields
BM
, et al.;
UNITED Team
.
Most people with long-duration type 1 diabetes in a large population-based study are insulin microsecretors
.
Diabetes Care
2015
;
38
:
323
328
[PubMed]
18.
Shields
BM
,
McDonald
TJ
,
Ellard
S
,
Campbell
MJ
,
Hyde
C
,
Hattersley
AT
.
The development and validation of a clinical prediction model to determine the probability of MODY in patients with young-onset diabetes
.
Diabetologia
2012
;
55
:
1265
1272
[PubMed]
19.
Talmud
PJ
,
Shah
S
,
Whittall
R
, et al
.
Use of low-density lipoprotein cholesterol gene score to distinguish patients with polygenic and monogenic familial hypercholesterolaemia: a case-control study
.
Lancet
2013
;
381
:
1293
1301
[PubMed]
20.
Edghill
EL
,
Dix
RJ
,
Flanagan
SE
, et al
.
HLA genotyping supports a nonautoimmune etiology in patients diagnosed with diabetes under the age of 6 months
.
Diabetes
2006
;
55
:
1895
1898
[PubMed]
21.
Iafusco
D
,
Stazi
MA
,
Cotichini
R
, et al.;
Early Onset Diabetes Study Group of the Italian Society of Paediatric Endocrinology and Diabetology
.
Permanent diabetes mellitus in the first year of life
.
Diabetologia
2002
;
45
:
798
804
[PubMed]
22.
Chang
C
.
Neonatal autoimmune diseases: a critical review
.
J Autoimmun
2012
;
38
:
J223
J238
[PubMed]
23.
Flanagan
SE
,
Haapaniemi
E
,
Russell
MA
, et al
.
Activating germline mutations in STAT3 cause early-onset multi-organ autoimmune disease
.
Nat Genet
2014
;
46
:
812
814
[PubMed]
24.
Krischer
JP
,
Lynch
KF
,
Schatz
DA
, et al.;
TEDDY Study Group
.
The 6 year incidence of diabetes-associated autoantibodies in genetically at-risk children: the TEDDY study
.
Diabetologia
2015
;
58
:
980
987
[PubMed]
25.
Todd
JA
.
Etiology of type 1 diabetes
.
Immunity
2010
;
32
:
457
467
[PubMed]
26.
Lernmark
A
,
Larsson
HE
.
Immune therapy in type 1 diabetes mellitus
.
Nat Rev Endocrinol
2013
;
9
:
92
103
[PubMed]

Supplementary data