In this study of islet autoimmunity, we analyzed three islet AAb (ICA, GAD and IA2) in the group of children and adolescents with AITD (AT and GD) and in the control group. We found that 10.6% of AITD patients were positive for one or more islet AAb, which was significantly more than in controls (1.9%, p=0.002). This difference was entirely due to antibodies found in AT patients, in comparison to controls (11.8% vs 1.9%; p=0.001), as we found no significant difference in islet autoimmunity between the GD patients and control group. All of our patients were positive for thyroid AAb prior to inclusion in the study which allowed us to select and follow the patients that developed thyroid autoimmunity before the onset of T1D. When analyzing islet AAb separately, all three AAb were significantly more frequent in AITD patients compared to the control group (ICA p=0.002, IA-2 p= 0.001), and GAD AAb were found only in the AITD patients, but not in controls.
Few studies reported the frequency of ICA autoimmunity in patients with AITD and only four of them included children and adolescents [6-9]. Bright et al. found ICA AAb in 2.3% of children with AITD compared to 0% of controls . In studies conducted in adult AITD patients, ICA positivity ranged from 0 - 4.9% [12,13,15,17].
Only one study evaluated frequency of IA2 AAb in children with AT (but not with GD) and found them to be more common than in control subjects (3.39% vs. 1.16%, p=0.012) , as was confirmed in our study. In one study conducted in adult patients, IA2 AAb were found more frequently in patients with AT as well as GD .
GAD positivity was assessed in three studies conducted in children with AT [7,9]. In two of them [8,9], GAD AAb were found significantly more often in children with AT than in controls (9.8-10.6% vs. 0-3.3%, p= 0.003 and p=0.036, respectively), as was confirmed in our study. However, in the study by Pilia et al. the difference was not significant . Several studies analyzed GAD autoimmunity in adult patients with AT [10,11,14-16] and GD [12,13,15,16]. Relative frequency of GAD AAb ranged from 3.4-6.6% [10,11,14-16] in patients with AT, and although the positive correlation between GAD AAb and AT was found in some of the studies , it was not always statistically significant . In adult GD patients, GAD AAb were found in 6.1-13% of patients, [12,13,15,16], significantly more common compared to controls in some of the studies [15,16]. However, control subjects were not always included in the evaluation [10-14].
Recently, islet AAb and thyroid autoimmunity were analyzed in children [24,25] and in adult patients , but these results cannot be compared to results from our study because they did not differentiate patients with AITD and risk for development of T1D from those that already had T1D.
We did not find significant differences in islet autoimmunity between males and females with AITD. However, females with AITD were positive to islet AAb, particularly to GAD and IA-2 AAb, significantly more often than females in control group, but this difference was not observed for ICA AAb. On the other hand, in males with AITD, we did not find any difference in islet autoimmunity compared to controls. As thyroid autoimmunity and AITD are more common in females, female gender was proposed as a risk factor for positive association between islet autoimmunity and thyroid autoimmunity .
In 16 patients with AITD who were positive for islet AAb, we wanted to assess the susceptibility for T1D development. One patient was diagnosed with T1D upon initial evaluation and five developed T1D during the follow up period of six years (Table 3). However, we cannot exclude that more patients would develop diabetes if the follow-up was longer. In the study of Bright et al.  one of two children with AT and positive ICA AAb developed diabetes after one year, and two children with AT and negative ICA AAb after four and six years, respectively. Pilia et al.  reported that over two years of follow-up, 2/19 children with AT and islet autoimmunity developed T1D (one positive to GAD AAb and the other to GAD and IA2 AAb). Lethagen et al.  found reduced ability of insulin secretion in GAD AAb positive AT patients and concluded that GAD AAb may be a marker of subclinical insulitis. During the follow-up of 4 years, 2/15 of their GAD AAb positive patients (compared to 11/426 GAD AAb negative patients) were diagnosed with diabetes (p=0.08) . Hallengren et al.  followed nine GAD AAb positive patients (two of them also ICA positive) for 27-70 months. One patient, who was positive for both islet AAb, developed diabetes. Maugendre et al. , found high frequency of GAD AAb (16/150 GD patients) but a low progression towards diabetes (only one patient). Aksoy et al.  studied insulin sensitivity and secretion patterns in GAD AAb positive and GAD AAb negative AT patients and concluded that it does not seem likely that presence of GAD AAb per se is associated with disturbance in glucose metabolism. Significant relationship between the higher titer of GAD AAb and abnormalities of glucose metabolism was found by Marhawa et al.  and Moriguchi et.al , but Kawasaki et al  did not report similar findings. In our study, AITD patients who developed T1D had significantly higher titers of GAD AAb (p= 0.02), compared to AITD patients with islet autoimmunity and normal glucose metabolism. Moreover, we noticed significantly higher titers of ICA and IA-2 AAb (both p=0.01) in this patient group.
We further measured TPO and Tg AAb and found higher titers in AITD patients with islet autoimmunity, compared to AITD patients without islet autoimmunity, although the difference was not statistically significant. Marwacha et al.  found that GAD AAb levels increased with increasing titer of anti-TPO AAb, but Yamaguchi et al.  and Kawasaki et al.  did not find correlations for the titer level of ICA  or IAA and GAD AAb .
Also, the observed relative frequency of T1D development in our patients with AITD was compared to that in Croatian general population. In our cohort T1D was found in 3.7% of AITD patients, much more frequently than in general population in the same age groups (0.2%) [27; Croatian registry of diabetes in children and adolescents, unpublished data].
Islet autoimmunity and susceptibility to T1D was also investigated in 20 first-degree family members of patients with AITD and positive islet AAb. Positive islet AAb were found in 5/20 family members (3/9 fathers and 2/10 mothers). One of the mothers was already diagnosed with T1D, and one father developed T1D during the follow-up, suggesting that family members of patients affected with AITD and islet autoimmunity might have higher risk for development of T1D. However, we did not test family members of AITD patients without positive islet AAb. It would be necessary to confirm these results on larger number of family members of patients with AITD, both those with and without islet autoimmunity, in order to determine the risk for glucose metabolism impairment in relatives of patients with AITD.
As development of T1D cannot be prevented or delayed, the clinical significance of screening for islet AAb and continuous follow-up of antibody positive patients is uncertain . However, as new studies found decreased rate of diabetic ketoacidosis in individuals screened for islet AAb , it might be important to identify groups at risk for development of T1D. Nevertheless, some authors concluded that screening for T1D risk to reduce diabetic ketoacidosis is not economically viable .
In conclusion, we found that children and adolescents with AITD, in particular females with AT and patients with higher titer of two or more islet AAb, but also family members of AITD patients with positive islet AAb, have an increased risk for developing T1D. Prospective long-term studies on a larger number of subjects are required to examine the factors responsible for islet cell destruction, insulin deficiency and evolution towards diabetes in patients with AITD. Patients with GD are particularly underrepresented and not well studied and it would be of importance to determine the rate of islet cell autoimmunity and risk for T1D development in these patients.
To the best of our knowledge, our study was the first to evaluate islet autoimmunity and glucose metabolism in family members of AITD patients with islet AAb, indicating they are at risk for developing T1D, but this observation must be verified in larger studies.