Acc

epte

d A

rtic

le

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/sji.12325 This article is protected by copyright. All rights reserved.

Article Type: Regular Manuscript

Doubly reactive INS-IGF2 autoantibodies in children with newly diagnosed

autoimmune (type 1) diabetes.

Norio Kanatsunaa,*, Ahmed Dellia, Cecilia Anderssona, Anna-Lena Nilssona,b, Fariba

Vaziri-Sania, Karin Larssonc, Annelie Carlssona, Elisabeth Cedervalld, Björn Jönssone, Jan

Neiderudf, Helena Elding Larssona, Sten-Anders Ivarssona, Carina Törna, Malin Fexa, Åke

Lernmarka

aDepartment of Clinical Sciences, Lund University /CRC, Skåne University Hospital SUS,

Malmö, Sweden bDepartment of Pediatrics, Östersund Hospital, Östersund, Sweden cDepartment of Pediatrics, Kristianstad Hospital, Kristianstad, Sweden dDepartment of Pediatrics, Ängelholm Hospital, Ängelholm, Sweden eDepartment of Pediatrics, Ystad Hospital, Ystad, Sweden fDepartment of Pediatrics, Helsingborg Hospital, Helsingborg, Sweden

*Corresponding author. Norio Kanatsuna MD PhD, Department of Clinical Sciences, Lund

University Diabetes Center, Lund University, Jan Waldenströms gata 35, Skåne University

Hospital SUS, SE-205 02 Malmö, Sweden, TEL.: +46 40 39 19 01; FAX: +46 40 39 11 22;

E-mail: norio.kanatsuna@med.lu.se

Running head: INS-IGF2 autoantibodies in type 1 diabetes.

Clinical Immunology

Abstract

The splice variant INS-IGF2 entails the preproinsulin signal peptide, the insulin B-chain,

eight amino acids of the C-peptide and 138 unique amino acids from an ORF in the IGF2

gene. The aim was to determine whether levels of specific INS-IGF2 autoantibodies

(INS-IGF2A) were related to age at diagnosis, islet autoantibodies, HLA-DQ, or both, in

newly diagnosed type 1 diabetes patients and controls. Patients (n=676), 0-18 years of age,

diagnosed with type 1 diabetes in 1996-2005 and controls (n=363) were analyzed for specific

INS-IGF2A after displacement with both cold insulin and INS-IGF2 to correct for

non-specific binding and identify double reactive sera. GADA, IA-2A, IAA, ICA, ZnT8RA,

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

ZnT8WA, and ZnT8QA, and HLA-DQ genotypes were also determined. The median level of

specific INS-IGF2A was higher in patients than controls (p<0.001). Irrespective of age at

diagnosis, 19 % (126/676) of the patients had INS-IGF2A when the cut-off was the 95th

percentile of the controls (p<0.001). The risk of INS-IGF2A was increased among

HLA-DQ2/8 (OR=1.509; 95th CI 1.011, 2.252; p=0.045) but not in 2/2, 2/X, 8/8, 8/X or X/X

(X is neither 2 nor 8) patients. The association with HLA-DQ2/8 suggests that this

autoantigen may be presented on HLA-DQ trans, rather than cis heterodimers.

Autoantibodies reactive with both insulin and INS-IGF2A at diagnosis support the notion that

INS-IGF2 autoimmunity contributes to type 1 diabetes.

Introduction

The clinical onset of type 1 diabetes is strongly associated with islet autoimmunity against major

autoantigens such as insulin, GAD65, IA-2 and ZnT8 reactive with both autoantibodies and T cells

(for a review see [1, 2]). Standardized islet autoantibody tests have been developed and two or more

islet autoantibodies predict type 1 diabetes [3]. Insulin autoantibodies as a first autoantibody may

appear already at 1-3 years of age [4]. Autoantibodies against insulin are detected in radiobinding

assays [5] [6] using 125I-labeled insulin and an excess of cold insulin to detect specific insulin

autoantibodies (IAA) [7]. IAA are particularly common among patients with onset of type 1 diabetes

in early childhood but their frequency in newly diagnosed patients are decreasing with increasing age

[8-10]. It has been suggested that the insulin B-chain may be critical to insulin autoimmunity in type 1

diabetes [11-14]. The insulin B-chain is also present in INS-IGF2, a transcriptional splice variant of

INS and the coding sequences of the two proximal open reading frame (ORF) of the IGF2 gene [15].

The two ORFs of the IGF2 gene are normally non-coding exons but insert a novel 138 amino acid

c-terminal region unrelated to preproIGF2 [15, 16]. Therefore, INS-IGF2 consists of the preproinsulin

signal peptide, the insulin B-chain and eight amino acids of the C-peptide in addition to 138 amino

acids from the unemployed ORFs in the IGF2 gene [15, 16].

We recently reported that INS-IGF2 was expressed primarily in human islet beta cells and showed

higher levels of expression in islets from normal compared to donors with either type 2 diabetes or

high HbA1c levels [16]. Newly diagnosed type 1 diabetes patients selected to be either very young or

young adults had autoantibodies able to recognize both insulin and INS-IGF2 (doubly reactive

autoantibodies) [16]. INS-IGF2 and insulin autoantibodies may therefore share binding sites on both

INS-IGF2 and insulin, respectively, thus complicating the understanding of these two autoantigens in

type 1 diabetes [17, 18]. It was therefore necessary to determine doubly reactive specific INS-IGF2

autoantibodies (INS-IGF2A), defined by displacement with both cold insulin and cold INS-IGF2, in

incident newly diagnosed type 1 diabetes patients at 1-18 years of age ascertained in a

population-based study [9, 19]. The aims were to determine levels and frequencies of doubly reactive

specific INS-IGF2A in relation to 1) age at clinical diagnosis, 2) other islet autoantibodies (IAA,

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

GAD65 autoantibodies (GADA), islet antigen-2 autoantibodies (IA-2A), zinc T8 transporter

autoantibodies (ZnT8A) and islet cell antibodies (ICA)) and 3) HLA-DQ [19].

Materials and methods

Subjects

The subjects enrolled in the present study were diagnosed with type 1 diabetes, at 1–18

years of age, between February 1996 and April 2005 in the province of Skåne (1,200,000

inhabitants) in southern Sweden [9, 19]. A total of 676 serum samples were available for

INS-IGF2A analyses (Table 1). There were 366 (54%) boys, median age at diagnosis was

10.2 years (Table 1). Serum samples were stored at -20oC until analyzed.

A total of 363 healthy subjects detailed in Table 1 served as controls. The majority of the

controls were 12 year old school children previously described [20] and analyzed for multiple

autoantibodies [21, 22]. An additional 55 healthy adult blood donors (54.5% males; n=30)

with a median age of 23.0 years (range 19.0-25.0) were also included in the control group

(Table 1). None of the controls were typed for HLA.

The Research Ethics Committee of Lund University, Lund, Sweden approved the study.

Specific INS-IGF2 autoantibody radiobinding assay (RBA)

Preparation of 35S-INS-IGF2 including the pThINS-IGF2 vector, the coupled in vitro

transcription - translation of pThINS-IGF2 and the preparation of both labeled and unlabeled

INS-IGF2 were previously described in detail [16].

The RBA for INS-IGF2A was carried out as described in detail [16] with the following

additions. All samples were tested in duplicates with either serum alone or serum with both

11 international unit (IU)/ml insulin (Actrapid®, Novo Nordisk A/S, Bagsvaerd, Denmark)

and non-radioactive INS-IGF2, diluted to represent a concentration that was twice the amount

of 35S-INS-IGF2 added to each well of V-formed 96-well plates (Nunc V96 MicroWell™

plates, Nunc A/S, Roskilde, Denmark). Each well with 5 μl serum was incubated for 16 h

with 60 μl 35S-INS-IGF2 at around 400 cpm/μl diluted in antigen buffer (150 mmol/l NaCl,

20 mmol/l Tris, 0.15% (v/v) Tween 20 (MP Biomedicals, LLC. Santa Ana, CA, USA), 1%

(v/v), TritonX-100 (AppliChem GmbH, Darmstadt, Germany), 0.1% (w/v) BSA (MP

Biomedicals, LLC.), pH 7.4). All serum samples were diluted 1:8. A total of 50 μl reaction

mixture was next incubated for 1.5 h at 4 °C in 96-well filtration plate (MultiScreenHTS-DV

Plate, Millipore AB, Solna, Sweden) with 50 μl Protein A conjugated Sepharose 4B (20%)

washed four times at 4 °C by sedimentation in antigen buffer. The plate was then washed 8

times in wash buffer (150 mmol/l NaCl, 20 mmol/l Tris, 0.15% Tween 20 (MP Biomedicals,

LLC.), pH 7.4) using a microplate washer (ELx50TM Microplate Strip Washer, Biotek

Instruments, Inc. Winooski, VT, USA). Antibody-bound radioactivity was counted in a

β-counter (1450 MicroBeta® TriLux, Perkin Elmer, Boston, MA). Sepharose-bound

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

radioactivity was converted into units per ml (U/ml) using a standard curve generated by six

step doubling dilutions (1:128 – 1:4 dilutions) of a high-titer type 1 diabetes patient’s serum

with high reactivity for INS-IGF2A. Three internal quality control samples, two from healthy

subjects and one from a long-term diabetes subject were used in every assay.

Intra-assay coefficient of variation (CV) for duplicate determinations was 4.0% without

and 4.2% with both cold antigens added. The inter-assay CV without cold antigens was

4.4 %.

Preparation of serum from EDTA-plasma

At total of 52 EDTA plasma samples were converted to serum by incubating 30.7 µl

EDTA-plasma over night at room temperature with 5 µl CaCl2 (0.2 mol/l) at a final

concentration of 28 mmol/l Ca2+ [23]. After centrifugation for 10 min at 10,000xg, the

supernatant now representing serum was used in the INS-IGF2A RBA.

RBA for GAD65A and IA-2A

GAD65A and IA-2A were analyzed as described in detail elsewhere [9, 19]. Levels were

expressed as units per milliliter derived from the World Health Organization standard 97/550

[24]. Samples were considered positive if GAD65A levels were > 50 U/ml and IA-2A levels

> 10 U/ml. The intra-assay CV for duplicate determinations was 7% for GAD65A and 11%

for IA-2A 11%. In the Diabetes Autoantibody Standardization Program (DASP) 2010

workshop, the workshop sensitivity was 80% and specificity 99% for GAD65A and 60%

sensitivity and 99% specificity for IA-2A.

RBA for Zinc T8 transporter autoantibodies against either arginine (R), tryptophan (W)

or glutamine (Q) at amino acid position 325 (ZnT8 R, W and Q A)

Autoantibodies against all three variants were analyzed individually as described [9, 19,

25]. The 2010 DASP workshop [26] showed that our laboratory had a workshop sensitivity of

52% for ZnT8RA, 50% for ZnT8WA and 38% for ZnT8QA. The workshop specificity was

100% for all three variant autoantibodies.

RBA for insulin autoantibodies (IAA)

IAA were determined as described in detail [9, 16, 19]. Briefly, all samples were analyzed

without and with cold insulin (2 IU/ml; Actrapid®, Novo Nordisk A/S) to control for

non-specific binding. Three internal quality control samples, two from healthy subjects and

one from a long term diabetes subjects were used in every assay. Sepharose-bound

radioactivity was converted into U/ml using a standard curve in every run generated by six

step doubling dilutions of an IAA high-titer type 1 diabetes serum. The inter-assay CV was

13 % and the intra-assay CV 6 %. Our laboratory participated in DASP [6] to show 26%

sensitivity and 100% specificity on the DASP samples.

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

Islet cell cytoplasmic autoantibodies (ICA)

ICA were determined in a two-color indirect immunofluorescence assay performed on

sections of frozen human pancreas, as described previously [27]. Our laboratory participated

in the 13th Immunology of Diabetes Workshop Standardization and showed 100% sensitivity

and specificity. Levels of ICA are expressed in JDF-U, using the WHO reference standard

[28].

HLA-DQ genotyping

HLA-DQB1 and DQA1 genotypes were typed by sequence-specific oligonucleotide probes

in a DELFIA Hybridization assay (Perkin Elmer, Boston, MA) as detailed elsewhere [29].

The following HLA-DQ genotypes were determined (abbreviations are shown within

parenthesis): DQA1*05:01-DQB1*02:01/DQA1*03:01-DQB1*03:02 (HLA-DQ2/8);

DQA1*03:01-DQB1*03:02/DQA1*03:01-DQB1*03:02 (HLA-DQ8/8);

DQA1*05:01-DQB1*02:01/DQA1*05:01-DQB1*02:01 (HLA-DQ2/2); X/X which are all

other genotypes containing neither DQ2 nor DQ8 (HLA-DQX/X).

Statistical analyses

The IBM® SPSS® Statistics, Version 20 (IBM Corp. Armonk, NY, USA) and GraphPad

Prism version 6 for Windows (GraphPad Software, Inc. La Jolla, CA, USA), were used.

Differences in levels of INS-IGF2A between controls and patients were assessed by

Mann-Whitney U test (two-tailed). Correlations were assessed by Spearman’s rho and

Fisher's exact test to compare frequencies. P values less than 0.05 were considered significant.

A Q–Q normality plot was used to show the quantiles of the INS-IGF2A distribution (y axis)

against the quantiles of a standard normal distribution (x axis). The Receivers Operating

Characteristics (ROC) curve was computed by using the GraphPad Prism version 6.

Results

INS-IGF2A showed higher binding levels in the patients compared to the controls (Figure

1). The Q-Q plot revealed that the Area Under the Curve (AUC) was increased in the patients

(AUC 83,070) compared to the controls (AUC 42,196; Mann-Whitney test; p<0.001).

Already at the 10th percentile, the binding level in the patients (231 U/ml) was twice as high

compared with the controls (105 U/ml). At the 50th percentile the patients (827 U/ml) exceed

the binding of the controls (300 U/ml) by a factor of 2.8 and at the 75th percentile (patients

1132 U/ml and controls 602 U/ml) the factor was 1.9 (Figure 1).

The INS-IGF2A Q-Q plot differed markedly from that of IAA (Figure 2) as well as those

of GADA, IA-2A and all three ZnT8RWQA (Figure 3). Below the 95th percentile cut-off, the

INS-IGF2A AUC was 73528 U/ml in the patients compared to 35114 U/ml in the controls

(p<0.001; Figure 1, panel A). The IAA AUC below the 95th percentile also showed a

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

difference between patients (175 U/ml) and controls (3.0; p<0.001; Figure 2, panel A),

however, the binding levels were low.

Using the 95th percentile as the cut-off for both INS-IGF2A (Figure 1, panel B) and IAA

(Figure 2, panel B), the receiver operating characteristics (ROC) curves generated showed an

AUC for INS-IGF2A of 0.771 and for IAA of 0.816, respectively. The IAA ROC curve

indicates that a cut-off of 0.37 U/ml would yield 72% sensitivity at a specificity of 92%

(Figure 2, panel B). The INS-IGF2A ROC curve (Figure 1, panel B) does not immediately

lend itself to a specific cut-off to achieve the highest diagnostic sensitivity and specificity.

The median levels of specific INS-IGF2A were higher in patients than controls (p<0.001)

(Figure 4). Levels of specific INS-IGF2A were not related to either gender or age at diagnosis

(data not shown). Specific INS-IGF2A levels did not correlate with the levels of any of the

other islet autoantibodies except for a weak negative correlation with GADA (r2 =0.005;

p=0.024) (data not shown).

Using the 95th percentile (1200 U/ml) as the cut-off for specific INS-IGF2A, 19% of the

patients were considered positive (p<0.001). The frequency was not affected by the age at

clinical diagnosis. The specific INS-IGF2A frequency at the 95th percentile was not related to

the frequency of any of GADA, IA-2A, IAA or ZnT8A (all three variants) (data not shown).

It was next tested if specific INS-IGF2A at the 95th percentile cut-off was related to

HLA-DQ genotypes (Table 2). HLA-DQ2/8 was related to an increased risk for INS-IGF2A

(p=0.045). The expected increased risks for IAA (p=0.026) and ZnT8WA (p=0.039) among

HLA-DQ2/8 patients are also shown (Table 2). HLA-DQ8/8, 8/X showed the expected

increased risk for IA-2A (p<0.001) as well as for ZnT8RA (p=0.043) and ZnT8WA (p=0.024).

Also, the expected negative associations between HLA-DQ2/2, 2/X and IAA (p<0.001),

IA-2A (p<0.001) as well as ICA (p=0.005) are shown for comparison. The remaining

HLA-DQ X/X (X is neither DQ2, nor DQ8) did not show any association with INS-IGF2A or

any of the other islet autoantibodies (Table 2).

Discussion

The autoantibodies against INS-IGF2 in newly diagnosed type 1 diabetes patients differ in

character from the autoantibodies directed against the major autoantigens in the following

respects. INS-IGF2 has a unique structure, which is not at all related to IGF2. The name is

somewhat misleading as INS-IGF2 contains the preproinsulin signal peptide, the B-chain and

8 amino acids of the C-peptide. In addition, the remaining 138 amino acids constituting the

c-terminal end of the protein bear no similarity to IGF2. Therefore, autoreactivity to

INS-IGF2 has two major features. One is autoreactivity with the preproinsulin part of the

molecule. The other is autoreactivity with the unique c-terminal end. There will be

autoantibodies, which are doubly reactive. This means that they will recognize both

preproinsulin and INS-IGF2 if the autoantibody for example will exclusively recognize the

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

insulin B-chain.

Our data also suggests that type 1 diabetes patients have autoantibodies, which may

recognize the c-terminal end of INS-IGF2 [16]. In the present study, care was taken to block

both of the potential binding sites, the insulin part and the unique c-terminal end, with cold

insulin and cold INS-IGF2, respectively. Therefore we only report specific INS-IGF2A to

account for autoantibodies that are doubly reactive (reacts with both insulin and INS-IGF2)

and mono reactive (reacts only with the c-terminal end of INS-IGF2). Specific INS-IGF2A is

the difference between total binding and the binding after displacement with both cold insulin

and INS-IGF2. To our knowledge none of the major autoantigens in type 1 diabetes, GAD65,

IA-2, or ZnT8 share the feature of doubly reactive autoantibodies. The fact that INS-IGF2 is

expressed in human islet beta cells to share the signal peptide, the B-chain and 8 amino acids

of the C-peptide complicates the understanding of insulin autoreactivity in type 1 diabetes

[16].

The present data in newly diagnosed type 1 diabetes patients and controls confirm our

previous study [16] of a binding characteristics that also deviate from that of the

autoantibodies to major autoantigens. The major difference is that INS-IGF2A binding levels

deviate from that of controls already at the 10th percentile as shown in our Q-Q plots (Figure

1). This type of binding was not observed for IAA, GADA, IA-2A, ZnT8A (Figure 3).

Autoantibody-binding to insulin, GAD65, IA-2 and ZnT8 is characterized by a lack of

deviation from controls up to the 75th-99th percentile (Figure 3). INS-IGF2 autoantibody

binding deviates from controls already at the 10th percentile but did show less of a difference

at the 95th percentile albeit statistical significant (Figure 1). We speculate therefore that

subjects with type 1 diabetes have been sensitized to either insulin, INS-IGF2, or both, to

acquire specific INS-IGF2A representing primarily low titer sera while high-titer INS-IGF2A

are less prevalent. Regardless, high titer INS-IGF2A was found to be more common in newly

diagnosed type 1 diabetes patients (19%) than in the controls (5%).

The frequency, 19% of "positive" INS-IGF2A would therefore suggest that INS-IGF2

should be classified as a minor autoantigen [30, 31]. However, this conclusion may be

misleading as the low level binding suggests that subjects who eventually were diagnosed

with type 1 diabetes had been sensitized to INS-IGF2 more often than controls. The increasing

slope of INS-IGF2 autoreactivity at the 95th percentile of the controls indicates that also

control subjects may have developed autoreactivity against INS-IGF2. This self-reactivity to

INS-IGF2 seems unique to this autoantigen and needs further analyses.

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

Several of the autoantibodies against the major autoantigens, GAD65, IA-2, ZnT8 and

insulin tend to occur together when examined at the time of clinical diagnosis [9]. It is

speculated that autoreactivity to INS-IGF2 differ from that of IAA as is amply illustrated by

comparing the two ROC curves shown as inserts to Figure 1. At the time of clinical diagnosis,

IAAs are detected primarily in children below 5-10 years of age [8-10]. Seroconversion to

IAA only was the first sign of islet autoimmunity with a peak at 1-3 years of age [4]. In

contrast, the finding that specific INS-IGF2A were not related to age at onset suggest that

studies from the time of birth will be needed to identify when children may become

autoreactive to INS-IGF2.

Specific INS-IGF2A showed the highest OR (Table 2) in children with the HLA-DQ2/8

genotype. This genotype is also associated with an early age at diagnosis [32, 33]. The strong

relation between INS-IGF2A and HLA-DQ2/8, but neither with DQ2/2 nor DQ8/8, allow us

to speculate that trans-heterodimers of DQ2 and DQ8 may be important to INS-IGF2

autoreactivity (Figure 5). DQ2/8 transheterodimers have long been recognized in type 1

diabetes [34] and T cells restricted by DQ encoded in trans reported [35]. Deamidaded

proinsulin peptides were specifically presented on HLA-DQ8 in trans [36]. It can therefore

not be excluded that INS-IGF2 presentation is dependent on HLA-DQ heterodimers in trans-

rather than cis.

The present investigation comprising 676 type 1 diabetes patients consecutively diagnosed

during 10 years would seem to be fully representative of pediatric patients, however, a

potential weakness is the lack of age and gender matched controls sampled over the same

time period. The Q-Q plot and the ROC curve analyses suggest that further studies in the

general population of INS-IGF2 autoreactivity is warranted.

Similar to the standard analysis of IAA that requires the assay to be run with and without

an excess of cold insulin [7], the present study was carried with an excess of both cold insulin

and cold INS-IGF2. This approach may have precluded detection of autoantibodies that react

with certain epitopes of INS-IGF2. Further studies and novel approaches to detect

INS-IGF2A perhaps with a different assay format [37, 38] should make it possible to detect

autoantibodies that react with insulin alone, both insulin and INS-IGF2 as well as INS-IGF2

alone. It would also be of interest to label INS-IGF2 with 125I in an attempt to more directly

compare IAA with INS-IGF2A.

The observation that INS-IGF2 is expressed in human beta cells and perhaps as a secreted

molecule will be the subject for future studies. More important to the understanding of

INS-IGF2 as an autoantigen, -involved in the etiology of type 1 diabetes, would be

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

investigations of the reactivity of INS-IGF2 with CD4+ and CD8+ T cells. It needs to be

tested if INS-IGF2 peptides are better presented on DQ2/8 heterodimers than on DQ2/2 and

DQ8/8 heterodimers (Figure 5). Previous studies of peptides from both preproinsulin and the

insulin B-chain have demonstrated specific epitopes related to these molecules [13]. It is

important to note that such epitopes would also be present in the INS-IGF2 molecule.

In conclusion, autoantibodies reactive with both insulin and INS-IGF2 at diagnosis support

the notion that INS-IGF2 is an autoantigen that may contribute to autoimmune type 1

diabetes.

Acknowledgements

We thank Anita Ramelius, Ingrid Wigheden and Ida Jönsson for expert technical

assistance.

This work was supported by the Swedish Research Council grant K2001-54X-15312-07-6,

SUS Funds & Donations, Diabetesfonden, Childhood Diabetes Fund, and the Skåne County

Council for Research and Development.

Disclosure

The authors declare no conflict of interest.

Author statement

N.K. researched the data, performed the experiments, analysed the data and prepared the

first draft of the manuscript. Å.L. researched the data, reviewed and edited the manuscript. A.

D., C.A., A-L.N., F.V-S. assisted in the analyses of islet autoantibodies and in the data

analysis, K.L., A.C., E.C., B.J., J.N., H.E.L. S-A.I., C.T. and M.F. recruited patients and

controls. All authors approved the manuscript before submission.

References

1 Kanatsuna N, Papadopoulos GK, Moustakas AK, Lernmark A. Etiopathogenesis of insulin

autoimmunity. Anat Res Int. 2012;2012:457546.

2 Bonifacio E, Ziegler AG. Advances in the prediction and natural history of type 1 diabetes.

Endocrinol Metab Clin North Am. 2010 Sep;39:513-25.

3 Ziegler AG, Rewers M, Simell O et al. Seroconversion to multiple islet autoantibodies and risk of

progression to diabetes in children. JAMA. 2013 Jun 19;309:2473-9.

4 Ilonen J, Hammais A, Laine AP et al. Patterns of beta-cell autoantibody appearance and genetic

associations during the first years of life. Diabetes. 2013 Oct;62:3636-40.

5 Palmer JP, Asplin CM, Clemons P et al. Insulin antibodies in insulin-dependent diabetics before

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

insulin treatment. Science. 1983 Dec 23;222:1337-9.

6 Schlosser M, Mueller PW, Torn C, Bonifacio E, Bingley PJ. Diabetes Antibody Standardization

Program: evaluation of assays for insulin autoantibodies. Diabetologia. 2010 Dec;53:2611-20.

7 Williams AJ, Bingley PJ, Bonifacio E, Palmer JP, Gale EA. A novel micro-assay for insulin

autoantibodies. Journal of Autoimmunity. 1997 Oct;10:473-8.

8 Vardi P, Ziegler AG, Mathews JH et al. Concentration of insulin autoantibodies at onset of type I

diabetes. Inverse log-linear correlation with age. Diabetes Care. 1988 Oct;11:736-9.

9 Andersson C, Larsson K, Vaziri-Sani F et al. The three ZNT8 autoantibody variants together improve

the diagnostic sensitivity of childhood and adolescent type 1 diabetes. Autoimmunity. 2011 Jan 19;44:394-405.

10 Andersson C, Vaziri-Sani F, Delli A et al. Triple specificity of ZnT8 autoantibodies in relation to HLA

and other islet autoantibodies in childhood and adolescent type 1 diabetes. Pediatr Diabetes. 2013

Mar;14:97-105.

11 Alleva DG, Crowe PD, Jin L et al. A disease-associated cellular immune response in type 1 diabetics

to an immunodominant epitope of insulin. The Journal of Clinical Investigation. 2001 Jan;107:173-80.

12 Eerligh P, van Lummel M, Zaldumbide A et al. Functional consequences of HLA-DQ8 homozygosity

versus heterozygosity for islet autoimmunity in type 1 diabetes. Genes and Immunity. 2011 Sep;12:415-27.

13 Velthuis JH, Unger WW, Abreu JR et al. Simultaneous detection of circulating autoreactive CD8+

T-cells specific for different islet cell-associated epitopes using combinatorial MHC multimers. Diabetes. 2010

Jul;59:1721-30.

14 Griffin AC, Zhao W, Wegmann KW, Hickley WF. Experimental autoimmune insulitis. Induction by T

lymphocytes specific for a peptide of proinsulin. Am J Pathol. 1995 Sep;147:845-57.

15 Monk D, Sanches R, Arnaud P et al. Imprinting of IGF2 P0 transcript and novel alternatively spliced

INS-IGF2 isoforms show differences between mouse and human. Hum Mol Genet. 2006 Apr 15;15:1259-69.

16 Kanatsuna N, Taneera J, Vaziri-Sani F et al. Autoimmunity against INS-IGF2 expressed in human

pancreatic islets. J Biol Chem. 2013 Oct 4;288:29013-23.

17 Srikanta S, Ricker AT, McCulloch DK, Soeldner JS, Eisenbarth GS, Palmer JP. Autoimmunity to

insulin, beta cell dysfunction, and development of insulin-dependent diabetes mellitus. Diabetes. 1986

Feb;35:139-42.

18 Eisenbarth GS, Jackson RA, Pugliese A. Insulin autoimmunity: the rate limiting factor in pre-type I

diabetes. J Autoimmun. 1992 Apr;5 Suppl A:241-6.

19 Nilsson AL, Vaziri-Sani F, Andersson C et al. Relationship Between Ljungan Virus Antibodies,

HLA-DQ8, and Insulin Autoantibodies in Newly Diagnosed Type 1 Diabetes Children. Viral Immunol. 2013

Jun;26:207-15.

20 Carlsson AK, Axelsson IE, Borulf SK et al. Prevalence of IgA-antiendomysium and IgA-antigliadin

autoantibodies at diagnosis of insulin-dependent diabetes mellitus in Swedish children and adolescents.

Pediatrics. 1999 Jun;103:1248-52.

21 Lindberg B, Ahlfors K, Carlsson A et al. Previous exposure to measles, mumps, and rubella--but not

vaccination during adolescence--correlates to the prevalence of pancreatic and thyroid autoantibodies.

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

Pediatrics. 1999 Jul;104:e12.

22 Agardh D, Nilsson A, Tuomi T et al. Prediction of silent celiac disease at diagnosis of childhood type

1 diabetes by tissue transglutaminase autoantibodies and HLA. Pediatr Diabetes. 2001 Jun;2:58-65.

23 Rahmati K, Lernmark A, Becker C et al. A comparison of serum and EDTA plasma in the

measurement of glutamic acid decarboxylase autoantibodies (GADA) and autoantibodies to islet antigen-2

(IA-2A) using the RSR radioimmunoassay (RIA) and enzyme linked immunosorbent assay (ELISA) kits. Clin

Lab. 2008;54:227-35.

24 Mire-Sluis AR, Gaines Das R, Lernmark A. The World Health Organization International

Collaborative Study for islet cell antibodies. Diabetologia. 2000 Oct;43:1282-92.

25 Vaziri-Sani F, Delli AJ, Elding-Larsson H et al. A novel triple mix radiobinding assay for the three

ZnT8 (ZnT8-RWQ) autoantibody variants in children with newly diagnosed diabetes. J Immunol Methods. 2011

Aug 31;371:25-37.

26 Lampasona V, Schlosser M, Mueller PW et al. Diabetes antibody standardization program: first

proficiency evaluation of assays for autoantibodies to zinc transporter 8. Clin Chem. 2011 Dec;57:1693-702.

27 Landin-Olsson M, Arnqvist HJ, Blohme G et al. Appearance of islet cell autoantibodies after clinical

diagnosis of diabetes mellitus. Autoimmunity. 1999;29:57-63.

28 Greenbaum CJ, Palmer JP, Nagataki S et al. Improved specificity of ICA assays in the Fourth

International Immunology of Diabetes Serum Exchange Workshop. Diabetes. 1992 Dec;41:1570-4.

29 Kiviniemi M, Hermann R, Nurmi J et al. A high-throughput population screening system for the

estimation of genetic risk for type 1 diabetes: an application for the TEDDY (the Environmental Determinants

of Diabetes in the Young) study. Diabetes Technol Ther. 2007 Oct;9:460-72.

30 Hirai H, Miura J, Hu Y et al. Selective screening of secretory vesicle-associated proteins for

autoantigens in type 1 diabetes: VAMP2 and NPY are new minor autoantigens. Clin Immunol. 2008

Jun;127:366-74.

31 Skarstrand H, Lernmark A, Vaziri-Sani F. Antigenicity and epitope specificity of ZnT8 autoantibodies

in type 1 diabetes. Scand J Immunol. 2012 Nov 6.

32 Karjalainen J, Salmela P, Ilonen J, Surcel HM, Knip M. A comparison of childhood and adult type I

diabetes mellitus. The New England Journal of Medicine. 1989 Apr 6;320:881-6.

33 Graham J, Kockum I, Sanjeevi CB et al. Negative association between type 1 diabetes and HLA

DQB1*0602-DQA1*0102 is attenuated with age at onset. Swedish Childhood Diabetes Study Group. European

Journal of Immunogenetics. 1999 Apr-Jun;26:117-27.

34 Ronningen KS, Markussen G, Iwe T, Thorsby E. An increased risk of insulin-dependent diabetes

mellitus (IDDM) among HLA-DR4,DQw8/DRw8,DQw4 heterozygotes. Hum Immunol. 1989 Mar;24:165-73.

35 Lundin KE, Bosnes V, Gjertsen HA, Qvigstad E, Thorsby E. Recognition of allogeneic cis- or

trans-encoded HLA-DQ molecules both by alpha beta and gamma delta T-cell receptors. Transplant Proc. 1991

Feb;23:22-4.

36 van Lummel M, Duinkerken G, van Veelen PA et al. Posttranslational modification of HLA-DQ

binding islet autoantigens in type 1 diabetes. Diabetes. 2014 Jan;63:237-47.

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

37 Yu L, Miao D, Scrimgeour L, Johnson K, Rewers M, Eisenbarth GS. Distinguishing persistent insulin

autoantibodies with differential risk: nonradioactive bivalent proinsulin/insulin autoantibody assay. Diabetes.

2012 Jan;61:179-86.

38 Miao D, Guyer KM, Dong F et al. GAD65 autoantibodies detected by electrochemiluminescence

assay identify high risk for type 1 diabetes. Diabetes. 2013 Dec;62:4174-8.

Table 1 Characteristics of patient serum samples, frequencies of islet autoantibodies and

HLA

Type 1 diabetes Controls

n 676 363

Median age (years) 10.2 12.7

(range) (1.0-18.8) (11.6-25.0)

Gender (% males) 54.1 47.9

HLA-DQ

2/8 217 (32.1%) ND

8/8, 8/X (X is not 2) 233 (34.5%) ND

2/2, 2/X (X is not 8) 115 (17.0%) ND

X/X (X is neither 2 nor 8) 111 (16.4%) ND

Islet autoantibodies

GADA 441 (65.2%) 3/351 (0.9%)

IA-2A 508 (75.1%) 3/351 (0.9%)

IAA 309 (45.7%) 2/179 (1.1%)

ICA 523 (77.4%) ND

ZnT8WA 336 (49.7%) 3/351 (0.9%)

ZnT8RA 334 (49.4%) 3/351 (0.9%)

ZnT8QA 245 (36.2%) 3/351 (0.9%)

ND is not determined.

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

Table 2 HLA-DQ genotypes in relation to islet autoantibodies.

HLA-DQ Islet autoantibody HLA DQ OR 95% CI p value

Positive

(%)

Negative

(%)

DQ 2/8 INS-IGF2A

positive 50 (7.4) 76 (11.2) 1.509 1.011, 2.252 0.045

negative 167 (24.7) 383 (56.7)

GADA positive 151 (22.3) 290 (42.9) 1.333 0.943, 1.884 0.119

negative 66 (28.1) 169 (25.0)

IAA positive 113 (52.1) 196 (29.0) 1.458 1.054, 2.016 0.026

negative 104 (15.4) 263 (38.9)

IA-2A positive 164 (24.3) 344 (50.9) 1.034 0.711, 1.505 0.924

negative 53 (7.8) 115 (17.0)

ICA positive 174 (25.7) 349 (51.6) 1.275 0.858, 1.896 0.239

negative 43 (6.4) 110 (16.3)

ZnT8RA positive 103 (15.2) 231 (34.2) 0.892 0.645, 1.232 0.510

negative 114 (16.9) 228 (33.7)

ZnT8WA positive 95 (14.1) 241 (35.7) 0.704 0.509, 0.975 0.039

negative 122 (18.0) 218 (32.2)

ZnT8QA positive 69 (10.2) 176 (26.0) 0.750 0.532 ,1.055 0.104

negative 148 (21.9) 283 (41.9)

DQ 8/X

or 8/8

INS-IGF2A positive 43 (6.4) 83 (12.3) 0.982 0.653, 1.476 1.000

negative 190 (28.1) 360 (53.3)

GADA positive 145 (21.4) 296 (43.8) 0.818 0.588, 1.139 0.236

negative 88 (13.0) 147 (22.7)

IAA positive 111 (16.4) 198 (29.3) 1.126 0.819, 1.547 0.466

negative 122 (18.0) 245 (36.2)

IA-2A positive 195 (28.8) 313 (46.3) 2.131 1.424, 3.190 <0.001

negative 38 (5.6) 130 (19.2)

ICA positive 188 (27.8) 335 (49.6) 1.347 0.911, 1.991 0.147

negative 45 (6.7) 108 (16.0)

ZnT8RA positive 128 (18.9) 206 (30.5) 1.402 1.020, 1.929 0.043

negative 105 (15.5) 237 (35.1)

ZnT8WA positive 130 (19.2) 206 (30.5) 1.452 1.056, 1.998 0.024

negative 103 (15.2) 237 (35.1)

ZnT8QA positive 93 (13.8) 152 (22.5) 1.272 0.917, 1.764 0.153

negative 140 (20.7) 291 (43.0)

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

DQ 2/X

or 2/2

INS-IGF2A positive 14 (2.1) 112 (16.6) 0.556 0.306, 1.009 0.065

negative 101 (14.9) 449 (66.4)

GADA positive 77 (11.4) 364 (53.8) 1.097 0.717, 1.678 0.747

negative 38 (5.6) 197 (29.1)

IAA positive 35 (5.2) 274 (40.5) 0.458 0.298, 0.705 <0.001

negative 80 (11.8) 287 (42.5)

IA-2A positive 66 (9.8) 442 (65.4) 0.363 0.238, 0.553 <0.001

negative 49 (7.2) 119 (17.6)

ICA positive 77 (11.4) 446 (66.0) 0.522 0.337, 0.811 0.005

negative 38 (5.6) 115 (17.0)

ZnT8RA positive 48 (7.1) 286 (42.3) 0.689 0.459, 1.034 0.082

negative 67 (9.9) 275 (40.7)

ZnT8WA positive 55 (8.1) 281 (41.6) 0.913 0.611, 1.365 0.683

negative 60 (8.9) 280 (41.4)

ZnT8QA positive 40 (5.9) 205 (30.3) 0.926 0.608, 1.410 0.750

negative 75 (11.1) 356 (52.7)

DQ X/X INS-IGF2A positive 19 (2.8) 107 (15.8) 0.884 0.517, 1.512 0.790

negative 92 (13.6) 458 (67.8)

GADA positive 68 (10.1) 373 (55.2) 0.814 0.535, 1.239 0.383

negative 43 (18.3) 192 (28.4)

IAA positive 50 (7.4) 259 (38.3) 0.968 0.643, 1.458 0.917

negative 61 (9.0) 306 (45.3)

IA-2A positive 83 (12.3) 425 (62.9) 0.976 0.611, 1.561 0.905

negative 28 (4.1) 140 (20.7)

ICA positive 84 (12.4) 439 (64.9) 0.893 0.554, 1.438 0.622

negative 27 (4.0) 126 (18.6)

ZnT8RA positive 55 (8.1) 279 (41.3) 1.007 0.670, 1.512 1.000

negative 56 (8.3) 286 (42.3)

ZnT8WA positive 56 (8.3) 280 (41.4) 1.036 0.690, 1.557 0.917

negative 55 (8.1) 285 (42.2)

ZnT8QA positive 43 (6.4) 202 (29.9) 1.136 0.748, 1.727 0.590

negative 68 (10.1) 363 (53.7)

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.

Acc

epte

d A

rtic

le

This article is protected by copyright. All rights reserved.