Autoantibodies against zinc transporter 8 are related to age and metabolic state in patients with newly diagnosed autoimmune diabetes

We studied the prevalence of GAD-ab, IA2-ab and ZnT8-ab in children and adults with newly diagnosed T1DM between 2010 and 2014. Moreover, IAA were measured in pediatric group, and ICA testing was performed in all patients with adult onset of diabetes. All participants were an ethnically homogenous population from Poland. The cohort comprised 367 European Caucasian patients, including: 218 children [girls: 95; boys: 123; median age: 9 years (interquartile range: 6–13 years)] and 149 adults [women: 78; men: 71; median age: 34 years (interquartile range: 27–43) years)], admitted to the Department of Pediatric Diabetes and Obesity, and to the Department of Internal Medicine and Diabetology of the Poznan University of Medical Sciences. The study was approved by the Poznan University of Medical Sciences Ethics Committee (decisions No 1623/05, No 739/09 and No 960/12). Informed consent was obtained from all individual participants included in the study.

Diabetes was diagnosed based upon WHO criteria, including fasting plasma glucose ≥ 7.0 mmol/l, 2-h postprandial plasma glucose ≥ 11.1 mmol/l, during an oral glucose tolerance test, classic symptoms of hyperglycemia or hyperglycemic crisis, as well as random plasma glucose concentration of ≥ 11.1 mmol/l. The first day of insulin administration was at T1DM diagnosis.

Standard laboratory tests, including blood glucose, capillary blood gases and blood or urinary ketones, were performed in every patient. DKA was defined as blood glucose concentration > 13.9 mmol/l, blood pH < 7.30, concentration of HCO₃ < 15 mmol/l, detection of ketones in the urine or elevated ketones in the serum, anion gap > 12. It was considered as mild, moderate and severe if pH was < 7.3 ≥ 7.2, < 7.2 ≥ 7.1 and < 7.1, respectively. Level of C-peptide was assessed by radioimmunoassay [C-PEP II-RIA-CT, DIA source Immunoassay, S.A, Belgium]. Glycated hemoglobin (HbA1c) value was evaluated by Chemiluminescent Microparticlr Immunoassey [HbA1C ARCHITECT System, Abbott Laboratories, U.S.A.]. To confirm autoimmune diabetes origin, typical autoantibodies were tested. The blood samples were collected on the day of diagnosis, before an initial insulin therapy, centrifuged to obtain serum and frozen until analyzed. GAD-ab and IA2-ab were measured by the RIA test kits, GAD-ab assay (positivity: > 1 U/ml) and IA-2-ab assay (positivity: > 1 U/ml), respectively [GAD and IA2 RIA. EUROIMMUN, Germany]. Level of ZnT8-ab was determined by the enzyme-linked immunosorbent assay [ElisaRSR ZnT8 Ab^™, UK]. The upper limit of the normal range was 15 U/mL, according to the manufacturer’s recommendation. Although IAA were checked in children, these antibodies are less likely to be present in older individuals and hence were not screened in adults and are not considered in the study.

The results are displayed as means and standard deviations (± SD), medians and interquartile ranges (IQR) or as numbers and percentages (%). Normality of the data distributions was tested using D’Agostino-Pearson test. We compared groups with different profiles of diabetes-associated autoantibodies using Student’s t test or Mann–Whitney U test for continuous variables and Fisher’s exact test for categorical variables. We calculated Pearson’s correlation coefficients to determine the association between selected continuous variables and the titer of diabetes-related autoantibodies.

We performed all tests at a significance level of 0.05 (two sided). Statistical analyses were carried out using Statistica version 10 (StatSoft Inc., Tulsa, OK, USA) and MedCalc version 17.5.3 (MedCalc Statistical Software, Ostend, Belgium).

Children had higher occurrence of all the three of the assessed antibodies (49.5%), whereas single antibody was rarely detected (7.8%). Adults were mostly positive for just 1 or 2 antibodies, and only 18.1% of them had three autoantibodies. This may suggest different intensity of β-cell autoimmunity. In subjects with only one autoantibody, isolated GAD-ab were the most common antibody found in adults, while in children the occurrence of any of three assessed antibodies was similar. In those with 2 positive antibodies, the combination of IA2-ZnT8-abs was the most common in children, but rare in adults. In adults, the combination of GAD-IA2-abs and slightly less frequent GAD-ZnT8 abs were observed. Similarly, as reported by Salonen et al., children were mostly positive for multiple antibodies with predominance of paired ZnT8-ab and IA2-ab [23]. Such an observation might reflect a positive relationship between GAD-ab occurrence and mild form of T1DM onset in adults. The majority of Chinese adults with LADA had only one positive autoantibody (85.9%), and the most detected single autoantibody was GAD-ab (67%) [24]. However, our analysis was performed on Caucasian ethnic group.

We observed higher prevalence and titer of ZnT8-ab in children (81.1%) than in adults (34.8%). Additionally, higher prevalence of IA-2-ab, but surprisingly lower titer of IA-2-ab and GAD-ab, was found in children compared to adults. In contrast, adults had mostly GAD-ab, with higher titer of those autoantibodies. Moreover, a lower prevalence of GAD-ab and a more aggressive diabetes onset in younger children might suggest that with an increase in T1DM prevalence not only a clinical phenotype has changed, but also immunophenotype reflected by a change of GAD-ab occurrence with time. Previously, Long et al. [25] reported the rise of IA2-ab and ZnT8-ab prevalence, while GAD-ab and IAA remained stable.

In our study, the prevalence of ZnT8-ab and other autoantibodies in adults was in concordance with that found in previous studies [26‐29]. Compared to these findings in Caucasian populations using internationally validated assays, other studies had found an increased prevalence of ZnT8-ab in LADA only in patients with a Japanese ethnic background (19%) or when the cumulative reactivity to alternative ZnT8 constructs was considered [7, 30]. In children, we found more than 80% positivity for ZnT8-ab and IA2-ab, but less for GAD-ab. This proportion is slightly different from data reported by the Finnish Pediatric Diabetes Register, where children had mostly IA2-ab (76%), followed by GAD-ab (67%) and ZnT8-ab (62.7%) [23]. Głowińska-Olszewska et al. [31] noticed high prevalence of ZnT8-ab (72%) in Polish children with T1DM duration shorter than 5 years. This finding demonstrates that ZnT8-ab could identify heterogeneity in the age of diabetes onset and are useful markers of childhood onset T1DM. In the Italian multicenter study, the same sensitivity of ZnT8-ab and GAD-ab at diabetes diagnosis was observed in children [32]. Moreover, ZnT8-ab-positive individuals were younger, while GAD-ab positive older at diagnosis [31]. Those observations justify introduction of the three screen assay which detects ZnT8-ab, IA2-ab and GAD-ab, as a useful first steep screening of the β-cell autoimmunity in children [33].

There is a question about the diagnostic value of ZnT8-ab in adults. Mainly, if highest prevalence of ZnT8-ab was observed in children from 3 years old up to late adolescence, a clear tendency for gradual decline in ZnT8-ab was demonstrated thereafter. However, ZnT8-ab are highly β-cells specific, exclusively expressed in insulin containing secretory granules of β-cells, and it is a target of the autoimmune process [20, 34]. This high specificity emphasizes a diagnostic value of ZnT8-ab among adults mainly in patients with low-titer or absent GAD-ab or in acute onset of T1DM [26]. Particularly, we showed that addition of ZnT8-ab to GAD-ab and IA-2-ab measurements could improve the diagnostic accuracy of autoimmune diabetes in adults, but should be obligatory in children.

In children, an aggressive autoimmune process may result in overt disease within a few months after the appearance of autoantibodies, whereas in older subjects the preclinical phase may continue for several years. The majority of children had IAA and GAD-ab as their first autoantibodies detectable in preclinical phase [35]. We demonstrated that children had very low titer of GAD-ab compared to adults. Based on our results, it might be also hypothesized that the level of GAD-ab may drop significantly at diagnosis. While, in adults, autoantibodies persisting long after diagnosis are usually high-titer GAD-ab, and much less frequently ZnT8-ab or IA2-ab [36]. Similarly, Sosenko et al. [37] indicated that in children GAD-ab titer tend to decline at diagnosis. ZnT8-ab positivity identified subjects at a higher risk of symptomatic diabetes. ZnT8 is located within β-cells secretory granules, and ZnT8-ab expression may not occur until there is enough β-cells damage to make this autoantigen perceptible for the immune system. Overall, the number and titer of autoantibodies decline after progression to T1DM. The type of persistent antibodies can be influenced by age at onset and could add information about clinical phenotypes. Perhaps, ZnT8-ab and GAD-ab might be considered as predictive markers for T1DM development, if they were measured in prediabetes phase. It seems that GAD-ab are more useful in such a situation, because they are positive prior to a mild autoimmune diabetes onset. However, in children with potentially short prediabetes phase, the predictive value of ZnT8-ab is not excluded. As, ZnT8-ab might be one of the key players in mediating both insulin secretion and β-cells mass in T1DM [38].

We confirmed that children are at greater risk of DKA at T1DM diagnosis. Yet, factors involved in DKA development and reasons why children are more prone to be diagnosed with DKA are unclear. DKA is either a consequence of delayed diagnosis, or it reflects intensity of destruction of β-cells. Nevertheless, it is proved that the occurrence of DKA has impact on long-term clinical course, probably through some deleterious influence on lower residual β-cell function and frequency of remission [39‐42]. In our study, subjects with DKA were more likely to be positive for ZnT8-ab and IA-2-ab, but their titer of IA-2-ab and GAD-ab were lower. However, we found a correlation between the number of positive antibodies and the severity of DKA. Furthermore, ZnT8-ab positivity was associated with a higher risk of DKA independent of age and the number of autoantibodies. The association of ZnT8-ab and DKA has been analyzed before with conflicting results. Salonen et al. described that DKA at diagnosis in young children was less common among ZnT8-ab-positive subjects. In concordance with previous studies, ZnT8-ab was uncommon among children with the age of diabetes onset below 5 years. There was a steady increase in proportion of ZnT8-ab positivity in patients older than 5 years and a gradual decrease after 10 years of age [23]. Another study in Turkish children showed a high ZnT8-ab prevalence at T1DM onset (58%), but this was not associated with DKA [43]. However, our results are consistent with those in the Childhood Diabetes in Finland Study Group, in which positivity for ZnT8-ab was associated with older age and more frequent DKA in children at diagnosis as well as lower serum C-peptide concentrations and higher insulin doses over time compared to ZnT8-ab-negative peers [44]. Likewise, Kawasaki et al. [45] have shown that about two-thirds of adult-onset patients had the slow onset form, and some 20% present DKA at diagnosis. ZnT8-ab was identified in 58% patients with acute-onset and in 20% with slow-onset T1DM among the Japanese.

Overall in our study, ZnT8-ab, IA2-ab and GAD-ab were found in the majority of children, while GAD-ab was the main antibody presented in adults at diagnosis. ZnT8-ab emerged as an important marker for DKA at T1DM onset.