Received: 16 March 2017

Revised: 19 June 2017

Accepted: 23 June 2017

DOI: 10.1111/pedi.12563

ORIGINAL ARTICLE

Transglutaminase antibodies and celiac disease in children with type 1 diabetes and in their family members Anna Parkkola1,2,3 | Taina Härkönen1,3 | Samppa J Ryhänen1,3 | Raivo Uibo4 | Jorma Ilonen5

| Mikael Knip1,2,3,6

1 Scientific Laboratory, Children's Hospital, University of Helsinki and Helsinki University Hospital, Helsinki, Finland 2

Folkhälsan Research Center, Helsinki, Finland

3

Research Programs Unit, Diabetes and Obesity, University of Helsinki, Helsinki, Finland 4

Department of Immunology, Institute of Biomedicine and Translational Medicine, University of Tartu, Tartu, Estonia 5

Immunogenetics Laboratory, University of Turku, Turku, Finland 6

Tampere Center for Child Health Research, Tampere University Hospital, Tampere, Finland Correspondence Mikael Knip, MD, PhD, Children's Hospital, University of Helsinki, PO Box 22, FI-00014 Helsinki, Finland. Email: [email protected] Funding information Academy of Finland, Centre of Excellence in Molecular Systems Immunology and Physiology Research, Grant/Award number: 250114; Sigrid Jusélius Foundation; Novo Nordisk Foundation; Liv and Hälsa Fund; Finnish Medical Foundation; National Graduate School of Clinical Investigation; Biomedicum Helsinki Foundation; Estonian Ministry of Education and Research, Grant/ Award number: IUT20-43; European Regional Developmental Fund, University of Tartu.

| and the Finnish Pediatric Diabetes Register†

Objectives: We set out to determine the prevalence of tissue transglutaminase antibodies (anti-tTG) and celiac disease (CD) in children with newly diagnosed type 1 diabetes (T1D) and their first-degree relatives (FDR). The hypothesis was that the individuals with both diabetes and CD form a distinct subgroup in terms of human leukocyte antigen (HLA) class II genetics, islet autoantibodies, and clinical characteristics at diabetes diagnosis. Subjects and methods: This population-based observational study included 745 index children with T1D and their 2692 FDR from the Finnish Pediatric Diabetes Register. CD was ascertained by registers, patient records, and screening anti-tTG positive individuals for further testing. Results: Among the index children, 4.8% had anti-tTG at diabetes diagnosis, and at the end of the study 3.2% had CD. Among the relatives, 2.9% had anti-tTG (4.8% mothers, 2.4% fathers, and 2.1% siblings), and 2.5% had CD (4.6% mothers, 2.1% fathers, and 1.4% siblings). Anti-tTG and CD associated with the HLA DR3-DQ2 haplotype. The usual female predominance of CD patients was observed in relatives (70%) but not among index children (46%). The index children with both diseases had a lower number of detectable islet autoantibodies than those with diabetes alone. Conclusions: The children with double diagnosis differed from those with diabetes alone in HLA genetics, humoral islet autoimmunity directed against fewer antigens, and in the lack of usual female preponderance among CD patients. Compared with 61% of the anti-tTG positive relatives, only 36% of anti-tTG positive index children developed CD implicating transient antitTG positivity at diagnosis of T1D. KEYWORDS

endomysial antibodies, HLA, islet autoantibodies, pediatric

1 | I N T RO D UC T I O N

antibodies

(anti-tTG),

a

sensitive

marker

of

CD-related

autoimmunity.3–6 Compared with 1% in the general population,7 Type 1 diabetes (T1D) and celiac disease (CD) are immune-

the prevalence of CD in patients with T1D is around 6%.8 Reports

mediated diseases caused by genetic predisposition and environ-

on CD among FDR of T1D patients are fewer with prevalences

mental factors. Patients with these diseases are at increased risk

from 1% to 6%.1,9,10 The human leukocyte antigen (HLA) class II

for the other, as are their first-degree relatives (FDR).1,2 At diagno-

DRB1*03-DQA1*05-DQB1*02 (DR3-DQ2) haplotype predisposes to

sis of T1D, 4% to 17% test positive for tissue transglutaminase

both CD and T1D. This haplotype is present in 35% to 60% of children with T1D, approximately 90% of children with CD, and in

†

Investigators of Finnish Pediatric Diabetes Register are listed in the Appendix.

15% to 25% of the general population.11–14

© 2017 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd Pediatric Diabetes. 2017;1–9.

wileyonlinelibrary.com/journal/pedi

1

2

PARKKOLA ET AL.

The reported prevalences of CD in children with T1D in Finland 15–17

methods of ascertainment. The data analysis was carried out in March

and those among their FDR

2014, 7.0 to 12.1 years (median 9.6) after the diagnosis of T1D in the

largely unknown. We set out to assess the prevalence of CD and

index child denoting the end of study. An individual was classified as

anti-tTG positivity in a recent nationwide sample. We assumed that,

having CD, if the disease was reported in either of the registers or

among those with anti-tTG/CD, the HLA-DR3-DQ2 haplotype would

patient records (n = 56), if duodenal biopsy confirmed the diagnosis

are relatively low (2.1%-4.7%),

be overrepresented. Additionally, a broad activation of autoimmune

after current testing (n = 18), or if the probability of CD was estimated

processes—manifested as autoimmunity related to both T1D and

to be more than 50% due to an appropriate HLA genotype in combi-

CD—could lead to broader β-cell-specific autoimmunity, resulting in

nation with (1) high CD-related autoantibody levels (anti-tTG ≥10

more frequent and higher titers of β-cell autoantibodies.

times ULN or EMA at least 1:80) at screening of samples from the FPDR sample repository (n = 8), (2) or in a current sample (n = 4), or (3) findings suggestive of CD in duodenal biopsy in combination with

2 | SUBJECTS AND METHODS

high CD-related autoantibody levels (n = 4, Figure 1). The Ethics Committee of the Hospital District of Helsinki and

2.1 | Study population

Uusimaa approved the study protocol, and study participants (aged ≥18) or their guardians gave their written informed consent. Partici-

The study families were participants in the Finnish Pediatric Diabetes

pants aged 10 to 17 gave their written informed assent. The study

Register (FPDR). Since 2002, this nationwide project has invited

was conducted in accordance with the Declaration of Helsinki.

patients diagnosed with diabetes in pediatric units to participate, and it covers >90% of all children diagnosed.18 The FPDR includes a structured questionnaire on family history of diabetes and other

2.2 | Analysis of tissue transglutaminase antibodies

autoimmune diseases,19 as well as clinical data from the diagnosis. At

All 3437 samples were analyzed for both IgA- and total-anti-tTG

data collection in August 2007, the FPDR included 2758 index chil-

(representing mostly IgG). Our assays were essentially like those

dren (median age 8.2, 56.1% male). As we wanted to assess CD auto-

described for β-cell autoantibodies20 and IgA- isotype-specific β-cell

immunity both in FDR and index children, we included only families

autoantibodies21 (Appendix S1, Supporting Information).

with serum samples available from mother, father, at least 1 sibling,

Antibody levels were expressed in relative units (RU) based on a

and an index child diagnosed with T1D before the age of 15. In total,

standard curve from a serial dilution of a pool of highly positive sam-

745 index children (56.6% boys) fulfilled these inclusion criteria. The

ples. The cut-off limit for IgA-anti-tTG positivity was 1.56 RU and for

included children were diagnosed at a median age of 8.8 compared

total-anti-tTG

with 8.0 years in those excluded (P = .005), but the gender distribu-

216 healthy Finnish children. Based on the 2007 anti-tTG workshop,

tion (P = .074), and the reported prevalence of CD (0.7% vs 0.9%,

the disease sensitivity and specificity of the IgA-anti-tTG assay were

P = .74) were similar. In addition, we included 745 mothers (median

91% and 98%.22

1.75 RU,

representing

the

99th

percentiles

in

age 38.3), 745 fathers (median age 40.5), and 1202 siblings (median

Only samples at/above the cut-off for positivity were included

age 9.7, 49.8% male) of the index children. Blood samples were taken

when calculating the median antibody titers. The samples testing pos-

from the index children at a median of 6 (range 0-1290) days, and

itive for total-anti-tTG but negative for IgA-anti-tTG were analyzed

from the FDR, 20 days after the diagnosis of T1D in the index case.

with a photometric, immunochemical method for total IgA levels in

For the analysis of β-cell autoantibodies in the index children, the

the Helsinki University Hospital Diagnostic Laboratory (HUSLAB) to

samples taken more than 30 days after the diagnosis were excluded.

detect IgA deficiency. For subjects with CD suggestive findings on

Data from the questionnaires of the FPDR determined the prevalence of known CD diagnoses at T1D diagnosis. These data are self-

the initial anti-tTG/EMA screening, but no known diagnosis of CD in the registers, anti-tTG were reanalyzed in local laboratories.

reported, but considered to represent biopsy-confirmed CD as this is the accepted criterion for CD diagnosis in Finland. As follow-up data is not collected in the FPDR, we used different methods of ascertain-

2.3 | Endomysial antibodies

ment for the current CD status. First, the prevalence of CD was deter-

Samples positive in both or either of the anti-tTG assays were tested

mined by April 2012 (median of 7.7 years [range 5.1-10.2] after the

for IgA-specific EMA with an immunofluorescence method described

diagnosis of the index case) from the register of Social Insurance Insti-

previously.23 Cut-off for positivity was a serum dilution 1:5.

tution of Finland. This registration requires biopsy-confirmed CD and is mandatory for social insurance benefits, and thus we assumed that most diagnoses were included. Second, the subjects with no known

2.4 | Islet autoantibodies

diagnosis of CD in the registers (n = 51), but with high anti-tTG (≥10

Insulin autoantibodies (IAA), glutamic acid decarboxylase autoantibo-

times the upper limit of normal/cut-off [ULN]) in either the IgA- or the

dies (GADA), islet antigen-2 autoantibodies (IA-2A), and zinc trans-

total-anti-tTG assay or with both anti-tTG and endomysial antibodies

porter 8 autoantibodies (ZnT8A) were analyzed using specific

(EMA) positive in the FPDR sample repository sample, were contacted

radiobinding assays.18,24–26 The cut-off limits were 2.80, 5.36, 0.77,

through local hospitals to offer current autoantibody testing and/or

and 0.61 RU, respectively, determined as the 99th percentiles in

duodenal biopsy. Finally, we went through the patient records of chil-

Finnish non-diabetic children. According to the 2010 Diabetes Auto-

dren with T1D but without CD diagnosis after the above-mentioned

antibody Standardization Program the sensitivities of these assays

3

PARKKOLA ET AL.

3437 subjects

Self-reported CD in the pediatric diabetes register n=20/3437 (0.6%)

Index children n=5/745 (0.7%)

Mothers n=8/745 (1.1%)

Fathers n=4/745 (0.5%)

Siblings n=3/1202 (0.2%)

Additional CD in the registers of the Social Insurance Institution, n=30

n=6

n=16

n=1

n=7

Additional CD detected through anti-tTG/EMA screening or from patient records, n=41

n=13

n=10

n=11

n=7

Final CD prevalence

n=24/745 (3.2%)

n=34/745 (4.6%)

n=16/745 (2.1%)

n=17/1202 (1.4%)

CD at the end of the study n=91/3437 (2.6%)

Ascertained through the registers n=50 (54.9%)

Biopsy proven n=18 (52.9%)

Ascertained through patient records n=7 (7.7%)

Biopsy not specific for CD, high current CD antibodies n=4 (11.8%)

Ascertained by anti-tTG and EMA screening of diabetes register samples, all subjects carry predisposing HLA genotypes n=34 (37.4%)

High current CD antibodies, no biopsy n=4 (11.8%)

No current sample, high anti-tTG/EMA at initial screening n=8 (23.5%)

FIGURE 1

Case finding of celiac disease (CD) patients from the Finnish Pediatric Diabetes Register. anti-tTG, tissue transglutaminase antibodies; EMA, endomysial antibodies

were 50%, 88%, 64%, and 60% and specificities 96%, 94%, 99%, and 100%, respectively. Islet cell antibodies (ICA) were analyzed with indirect immunofluorescence on human group 0 donor pancreas with 2.5 Juvenile Diabetes Foundation (JDF) units as the detection limit.27 Only samples at/above the cut-off were included when calculating

2.6 | Markers of metabolic decompensation at diagnosis Blood pH, plasma glucose, and β-hydroxybutyrate concentrations in the index children were analyzed in local laboratories at diagnosis of T1D.

the median antibody titers.

2.5 | HLA typing

2.7 | Statistical analysis

Polymerase chain reaction (PCR) amplification followed by hybridiza-

IBM SPSS statistics 20 software package (IBM, Chigaco, Illinois) and

tion with lanthanide-labeled sequence-specific oligonucleotide probes

R 2.13.0 were used for statistical analyses. A 2-tailed P value