Professional Documents
Culture Documents
Original Articles: Comorbidity of Type 1 Diabetes Mellitus in Patients With Juvenile Idiopathic Arthritis
Original Articles: Comorbidity of Type 1 Diabetes Mellitus in Patients With Juvenile Idiopathic Arthritis
Objectives To determine the prevalence of type 1 diabetes mellitus (T1D) in patients with juvenile idiopathic ar-
thritis (JIA) and to characterize patients having both.
Study design Diabetes comorbidity was recorded in the National Pediatric Rheumatologic Database since 2012.
Data from the North Rhine-Westphalian diabetes registry served as the reference population for the prevalence of
diabetes in the general population. The National Pediatric Rheumatologic Database data were indirectly standard-
ized for age and sex for comparison with the general population. The diabetes prevalence ratio was calculated
using the Poisson regression model.
Results The analysis included 12 269 patients with JIA. A total of 58 patients had comorbid T1D, and the dia-
betes prevalence was 0.5%. The mean age was 11.6 years at the time of documentation, and the mean disease
duration was 4.2 years. Compared with the general population, the prevalence of diabetes in patients with JIA was
significantly increased (prevalence ratio 1.76 [95% CI 1.34; 2.28], P < .001). The onset of diabetes in patients with
JIA was earlier than that reported in the reference data. Sixty-three percent of patients developed T1D before JIA.
On average, diabetes onset was 56 months before the onset of JIA. Patients who first developed JIA developed
T1D on average 40 months later. The majority of patients had not received disease-modifying antirheumatic drugs
before diabetes onset.
Conclusions T1D occurs more frequently in patients with JIA than in the general population. The likelihood of
T1D occurrence appears to be slightly higher before JIA manifestation and without disease-modifying antirheu-
matic drug therapy after JIA onset. (J Pediatr 2018;192:196-203).
See editorial, p 6
J
uvenile idiopathic arthritis (JIA) is the most common chronic childhood inflammatory rheumatic disease in Europe,
with an estimated incidence of 2-20 cases per 100 000 children and adolescents.1 JIA comprises a clinically heteroge-
neous group of arthritides characterized by persistent joint inflammation and a disease onset before 16 years of age. Most
of the categories of JIA are related to autoimmunity.2
Twin and family studies have emphasized the importance of genetic predispo-
sition for autoimmune diseases (AIDs).3-5 Genetic analyses have shown an
association between mutations in certain susceptibility gene loci and AIDs. From the 1German Rheumatism Research Center Berlin,
a Leibniz Institute, Berlin, Berlin; 2Institute for Biometrics
Phenotypically different AIDs share genetic susceptibility factors.6,7 Several single and Epidemiology, German Diabetes Center, Leibniz
nucleotide polymorphisms associated with JIA have been identified.7-10 Patients Centre at Heinrich-Heine University Dusseldorf,
Dusseldorf; 3German Center for Diabetes Research
(DZD), Munich-Neuherberg; 4Institute for Social Medicine,
Epidemiology and Health Economics, Charité
Universitätsmedizin, Berlin; 5Department of
Rheumatology and Clinical Immunology, Charité
AID Autoimmune disease Universitätsmedizin Berlin, Berlin; 6Department of
AIT Autoimmune thyroiditis Pediatric Rheumatology, St. Josef-Stift, Sendenhorst;
7Hamburg Centre for Rheumatology for Children and
bDMARD Biological DMARD Adolescents, Clinical Center Eilbek; 8Asklepios Clinic
BMI Body mass index Sankt Augustin, Sankt Augustin, University Hospital of
CD Celiac disease Cologne, Cologne; 9Department of Pediatric
Rheumatology, Prof.-Hess-Kinderklinik, Bremen;
cJADAS Clinical Juvenile Arthritis Disease Activity Score 10Institute of Epidemiology and Medical Biometry, ZIBMT,
196
Volume 192 • January 2018
with 1 AID are at an increased risk for another AID.11 For the cumulative age distribution of T1D onset in the general
example, the genetic association between type 1 diabetes mel- population compared with the distribution of T1D onset in
litus (T1D) and JIA has been described.12,13 patients with JIA. This registry ascertains newly diagnosed
Epidemiologic knowledge about the comorbidity of JIA and cases of T1D from 3 data sources: a prospective hospital-
T1D is limited. Previous studies are based mostly on case based active surveillance system (Erhebungseinheit für Seltene
reports.14-16 A few observational studies have shown an in- Pädiatrische Erkrankungen in Deutschland [ESPED]), annual
creased frequency of diabetes mellitus (DM) in patients with inquiries among medical practices (pediatricians, general
JIA compared with the general population.17,18 Detailed in- practitioners, and internal specialists), and the diabetes pro-
formation about the comorbidity of T1D in patients with JIA spective follow-up (Diabetes-Patienten-Verlaufsdokumentation
has been limited. Two established prospective cohorts in [DPV]) database. DPV is a computer-based system for lon-
Germany, the National Pediatric Rheumatologic Database gitudinal documentation of treatment and outcomes in routine
(NPRD) for JIA and the North Rhine-Westphalian Diabetes diabetes care.26 The DPV software is used in more than 420
Registry for T1D, provide a unique opportunity to investi- diabetes centers in Germany. Completeness of the North
gate the co-occurrence of JIA and T1D. Rhine-Westphalian diabetes registry is estimated at 99%.
In 2015, Hermann et al published data on the prevalence Sociodemographic and clinical characteristics, and patient-
of JIA in patients with T1D. They showed an increased preva- reported outcome measures were annually recorded in the
lence of JIA compared with the general population and an NPRD. These included age, sex, body height, and weight, JIA
earlier onset of diabetes in patients with JIA.19 The present duration, age at JIA onset, JIA category, treatment, and others.
study, based on the NPRD data, aimed to determine the Body mass index (BMI) was calculated by dividing body weight
prevalence of T1D in patients with JIA and to characterize (in kg) by squared body height (in meters). Categories of un-
the features of patients with JIA with T1D vs patients with derweight, normal weight, overweight, and obesity were defined
JIA without T1D. according to sex- and age-specific BMI percentile curves for
German children as suggested by Kromeyer-Hauschild et al.27
Methods The physicians recorded treatments with conventional syn-
thetic and biological disease-modifying antirheumatic drugs
The NPRD of children and adolescents with rheumatic (DMARD) as well as glucocorticoids, including low dose
diseases was started in 1997 and includes incident and preva- (<0.2 mg/kg), high dose (≥0.2 mg/kg), and pulse therapy. In
lent cases of a wide spectrum of juvenile rheumatic diseases. addition, physicians recorded comorbidities of DM, AIT, and
The majority of enrolled patients (3 of 4 patients) have JIA. CD and their onset dates.
Enrolled patients are followed annually by a physician and The patient’s JIA disease activity was assessed by the pedi-
patient questionnaire. For more details, we refer to previous atric rheumatologist using a numeric rating scale (0-10, 0-no
NPRD publications.20-22 Patients with JIA according to the disease activity). The clinical Juvenile Arthritis Disease Activ-
International League of Associations for Rheumatology ity Score (cJADAS)28 is a composite index for assessment of
criteria23,24 who were 20 years of age or younger and re- JIA disease activity. The cJADAS-10 is the sum of the physi-
corded in the NPRD between 2012 and 2014, were included cian’s and patient’s/parent’s global assessment and the number
in this study. of active joints (maximum = 10). Functional ability was re-
Information about comorbid DM was recorded from 2012 ported by patients or parents using the Childhood Health As-
to 2014. The presence of autoimmune thyroiditis (AIT) and sessment Questionnaire, calculated as a disability index ranging
celiac disease (CD) was documented by physicians only in 2013 between 0 and 3 (0-no disability).29
and 2014. An additional short questionnaire on DM was used The DM questionnaire was sent to pediatric rheumatol-
to confirm the diagnosis of DM and to provide more details ogy centers that recorded a DM case between 2012 and 2014
regarding T1D. Patients recorded in multiple years were counted and included questions regarding the type of DM (T1D, type
only once. 2 diabetes mellitus [T2D], steroid-induced diabetes, maturity-
The prevalence of T1D in patients with JIA was compared onset diabetes of the young-type, and others), age at DM di-
with the prevalence of T1D in the general population of chil- agnosis, insulin treatment, diagnosis of other AIDs (including
dren and adolescents aged 0-20 years in the federal state of AIT and CD), and family history of AIDs.
North Rhine-Westphalia on December 31, 2014. Indepen-
dent registries for T1D exist in 4 federal German states. We Statistical Analyses
chose North Rhine-Westphalia for our comparison because it Cross-sectional data from the NPRD for the years 2012, 2013,
comprises 22%-23% of all German children and adoles- and 2014 were used to determine the diabetes rate in pa-
cents. The prevalence data of North Rhine-Westphalia were tients with JIA. Descriptive analyses included absolute and rela-
derived from the North Rhine-Westphalian diabetes registry tive frequencies for categorical variables, means and SDs or
maintained at the German Diabetes Center since 1993. The median and IQR, if appropriate, for continuously distrib-
prevalence of T1D in North Rhine-Westphalia is higher than uted variables.
in other parts of Germany.25 The frequency of T1D in patients with JIA was compared
Incident cases between the ages of 0 and 20 years for the with the population-based data of the North Rhine-Westphalian
period 2002-2014 were selected from the registry to estimate diabetes registry.25 Crude diabetes prevalence and respective
197
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 192
95% CIs were estimated assuming a Poisson distribution of patients. The T1D prevalence ratio differed among the various
cases. For comparison between the JIA cohort and the general age groups. The T1D rate ratio among male patients differed
population, prevalences were indirectly standardized with with age from 1.73 between 5 and 9 years and 2.18 between
respect to sex and/or age (age groups 0-4, 5-9, 10-14, and 15 and 20 years. Differences between age groups were more
15-20 years) to the general population of North Rhine- pronounced among female patients from 1.12 between 10 and
Westphalia. Standardized prevalence ratios (PRs) with 95% 14 years and 2.78 between 0 and 4 years. Differences between
CIs and P values of respective c2 tests were estimated.30 We the sexes were not statistically significant.
analyzed the cumulative distribution of the age at T1D onset The sociodemographic and clinical characteristics of pa-
for patients with JIA compared with the general population. tients with JIA with or without T1D are reported in Table I.
For a descriptive subanalysis of JIA treatment before T1D The differences in the JIA category distribution were not
onset, longitudinally collected data from 2000 to 2014 were statistically significant between the 2 groups. Patients with
considered in a subset of patients only, who had been docu- JIA without T1D were significantly younger at JIA onset
mented at least once before T1D onset. The overall than patients with T1D (P = .032). There were no significant
redocumentation rate was approximately 60%, which was the differences in the duration of JIA or in the therapy of both
reason longitudinal data were not available for all patients groups. The duration from symptom onset until the first
who developed JIA before T1D. Statistical analysis was visit to a rheumatologist did not differ between the patient
conducted using SPSS Statistics 23, (IBM SPSS; SPSS Inc, groups.
Chicago, Illinois) and SAS v 9.4 (SAS Institute, Cary, North A comparison of sex- and age-adjusted anthropometric SDS
Carolina). did not reveal differences for height between the groups. Pa-
tients without T1D had a lower weight than patients with T1D
(P = .042). Both groups had a lower weight than the refer-
Results ence data. The BMI SDS in patients with JIA and T1D (0.17)
were significantly higher than the BMI SDS in patients without
DM was recorded in 62 of 12 269 patients with JIA. Two pa- T1D (-0.25, P = .007). The proportion of overweight/obese
tients were diagnosed with T2D and 2 patients had maturity- patients in the group of patients with JIA and T1D was higher
onset diabetes of the young and, therefore, were excluded from than in those without T1D (14.3% vs 10.8%, respectively). The
analysis. Therefore, the prevalence of T1D in the NPRD was distribution of the outcome variables (functional disability, phy-
0.47% (95% CI 0.36-0.61). T1D prevalences for sex and age sician’s global assessment, and cJADAS-10) was comparable
groups are shown in Figure 1. T1D was significantly more between the groups.
common in patients with JIA than in the general population The mean age at diabetes onset was 7.1 years (SD 4.4
(PR 1.76 (CI 1.34; 2.28), P < .001). The T1D PR was 1.72 (CI years), and the mean duration of diabetes at documentation
1.21; 2.38, P = .003) for female patients and 1.83 (CI 1.15; 2.77, was 6.4 years (SD 4.5 years). Eleven (19%) of the patients
P = .012) for male patients. There was no statistically signifi- with JIA and T1D were documented with an additional AID.
cant difference in the T1D rate between male and female Comorbid AIT and CD were reported in 6 (10.3%) patients
Figure 1. Prevalence of T1D in patients with JIA compared with population data stratified by age and sex; reference data from
the North Rhine-Westphalian diabetes register.
198 Schenck et al
January 2018 ORIGINAL ARTICLES
Figure 2. Cumulative distribution of age at the onset of T1D in patients with JIA (n = 54) and in the general population (n = 9943);
reference data from the North Rhine-Westphalian diabetes register.
Comorbidity of Type 1 Diabetes Mellitus in Patients with Juvenile Idiopathic Arthritis 199
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 192
and antitumor necrosis factor therapy37 also decreased the In summary, several findings in our study support the
risk for development of DM in patients with rheumatoid hypothesis, that bDMARD therapy may be associated with a
arthritis. The investigations included patients with all DM lower incidence of T1D in patients with JIA. Patients with
diagnoses (International Classification of Diseases, Ninth Revision, JIA with T1D who had a low insulin requirement were more
Clinical Modification 250) including T1D. However, the ma- often treated with bDMARD than patients with JIA with
jority of patients in these studies probably had T2D. The key T1D who had a higher insulin requirement. The majority of
mechanism in T1D pathogenesis is inflammation of pancre- T1D cases occurred before the onset of JIA and therefore,
atic b cells and a linked expression of proinflammatory before JIA-indicated therapy. None of the cases of diabetes
cytokines.38,39 The immunomodulatory effect of several medi- in our cohort (with available therapy data) were treated with
cations was found to improve T1D outcomes37,40 and may a bDMARD before the onset of diabetes. The association
potentially prevent T1D.41-45 Asseldonk et al observed im- between DMARD therapy in general, and bDMARD in par-
proved insulin sensitivity and a reduced insulin requirement ticular, and a reduced risk for T1D should further be
in 14 patients with T1D who were treated with the interleukin-1 investigated in longitudinal analysis of a large, long-term
receptor antagonist anakinra for 1 week.46 The effect lasted 4 prospective cohorts.
weeks after the final application, which may be an explana- A limitation of our study is the small number of patients
tion for the low insulin requirement of patients with JIA in with T1D, especially for subgroup analysis such as insulin re-
our study cohort. The insulin requirement in patients with quirement. We may have underestimated the number of pa-
comorbidity of T1D and JIA was lower in the NPRD data tients with T1D because we required validation of the DM
than in the diabetes registry (0.55 IU/kg/day vs 1.03 IU/kg/ diagnosis by a positive response to that item in the physi-
day, respectively). The subanalysis of our data revealed that cian’s questionnaire. Information on comorbid DM was not
patients with JIA with T1D and low insulin requirement provided in one-fourth of the patients with JIA. Further-
were more frequently treated with bDMARDs, and patients more, the nationwide JIA data were compared with the ref-
with JIA and a higher insulin requirement were less often erence data of the federal state of North Rhine-Westphalia. The
treated with bDMARDs (33.0% vs 21.4%, respectively). Un- T1D prevalence in North Rhine-Westphalia is higher than the
fortunately, we did not have information on the treatment estimated nationwide T1D prevalence,25 which would result
of patients with JIA in the diabetes registry. Thus, the com- in a lower prevalence ratio.
parison of DMARD therapy between the registries remains Our analyses were based on diabetes data that were col-
open. lected in the NPRD only from 2012 to 2014. Cross-sectional
Compared with reference data from the general popula- data do not allow many assertions about the disease course
tion, patients in our study with JIA and T1D had an earlier and other factors. Moreover, because of the NPRD study
onset of diabetes. Hermann et al 19 found significantly design, patients were documented at an arbitrary time
earlier T1D manifestations in patients with comorbid JIA vs during the disease course, which may have led to a variation
patients without JIA (median: 7.2 vs 8.3 years, respectively, in the time between the onset of JIA and outcome
P = .04). Regarding the age distribution of the cumulative T1D assessments at documentation. Nevertheless, most of our
incidence (Figure 2), we observed that the peak incidence findings are supported by results of previous international
of diabetes cases in the older age groups found in the general studies.
population data was absent in the JIA data. We speculate that In conclusion, our study in the NPRD and the study of
the decreased occurrence of T1D in older patients with JIA may Hermann et al in the diabetes registry confirm the clustering
be associated with JIA-indicated DMARD therapy. Alterna- of JIA, T1D, AIT, and CD in large populations.19 If a child or
tively, the decreased T1D incidence may also be caused by adolescent has developed JIA or T1D, the treating pediatri-
“depletion of susceptibles.” Predisposed patients may develop cian should be aware of the increased risk for the develop-
T1D at an early age because of a strong autoimmune ment of the other AID. ■
phenotype.
Furthermore, we analyzed the treatment with glucocorti- We thank the participating patients and sites and the following pedi-
coids and DMARDs in patients with JIA diagnosed before atric rheumatologists, who enrolled at least 200 patients: Ralf Trauzeddel
T1D onset. Although the number of patients was small (Berlin), Tilmann Kallinich (Berlin), Rolf-Michael Küster (Wedel,
(n = 10), we observed that no patient in our cohort received current address: Hamburg), Gerd Ganser (Sendenhorst), Christoph
bDMARD therapy before T1D onset, which may be linked to Rietschel (Frankfurt), Markus Hufnagel (Freiburg), Johannes-Peter Haas
(Garmisch-Partenkirchen), Rainer Berendes (Landshut), Anton Hospach
an association between DMARD therapy and the risk of (Stuttgart), Thomas Lutz (Heidelberg), and Hans-Iko Huppertz
T1D. Three patients were documented as previously being (Bremen).
treated with MTX therapy; MTX alone does not seem to be The support of Tobias Schwarz, Ivan Foeldvari, Gerd Horneff, Frank
associated with a reduced risk for DM in patients with JIA.35 Weller-Heinemann in the NPRD is gratefully acknowledged.
Bongartz et al hypothesized that the risk-reducing effect of
DMARDs for DM is not based only on an anti-inflammatory Submitted for publication Apr 26, 2017; last revision received Jun 2, 2017;
accepted Jul 26, 2017
effect but may be amplified by pathway-specific effects of
Reprint requests: Sandra Schenck, MSc, German Rheumatism Research
immunosuppressive agents that induce an anti-hyperglycemic Centre Berlin, a Leibniz Institute, Berlin, Charitéplatz 1, 10117 Berlin, UK.
impact.47 E-mail: Sandra.Schenck@drfz.de
Comorbidity of Type 1 Diabetes Mellitus in Patients with Juvenile Idiopathic Arthritis 201
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 192
202 Schenck et al
January 2018 ORIGINAL ARTICLES
44. Koulmanda M, Budo E, Bonner-Weir S, Qipo A, Putheti P, Degauque N, 46. van Asseldonk EJ, van Poppel PC, Ballak DB, Stienstra R, Netea MG, Tack
et al. Modification of adverse inflammation is required to cure new- CJ. One week treatment with the IL-1 receptor antagonist anakinra leads
onset type 1 diabetic hosts. Proc Natl Acad Sci USA 2007;104:13074- to a sustained improvement in insulin sensitivity in insulin resistant pa-
9. tients with type 1 diabetes mellitus. Clin Immunol 2015;160:155-62.
45. Strom TB, Koulmanda M. Cytokine related therapies for autoimmune 47. Bongartz T, Kudva Y. Can treatment of chronic inflammatory diseases
disease. Curr Opin Immunol 2008;20:676-81. reduce the risk of diabetes mellitus? JAMA 2011;305:2573-4.
Comorbidity of Type 1 Diabetes Mellitus in Patients with Juvenile Idiopathic Arthritis 203