Type 1 diabetes is a T-cell mediated autoimmune disease leading to beta-cell destruction and loss of insulin secretion resulting in severe hyperglycemia. Type 1 diabetes results from a complex interaction between environmental and genetic factors. Several genes have been identified as causative in the development of type 1 diabetes [1, 2] and some of these genes as well as other genes are shown to exert an impact on the disease progression from onset in newly diagnosed type 1 diabetes children [3–6]. In a number of studies, the non-synonymous variant, C1858T, of the PTPN22 gene has been associated with development of type 1 diabetes as well as other autoimmune diseases [7–11]. Recently, this PTPN22 susceptibility variant was found to be significantly associated to lower fasting C-peptide levels, poorer glycemic control in recent onset type 1 diabetes subjects [6] and to higher GADA in type 1 diabetes patients with long disease duration [12]. The objective of the current longitudinal investigation was therefore to evaluate the impact of PTPN22 on disease progression as assessed by liquid meal-stimulated C-peptide and proinsulin, HbA_1c, daily insulin dose, insulin dose-adjusted HbA_1c (IDAA1C) [13], antibodies to the protein tyrosine phosphatase related IA-2 molecule (IA-2A), islet cell antibodies (ICA), insulin antibodies (IA), glutamic acid decarboxylase antibodies (GADA) and zinc transporter-8 antibodies (ZnT8Ab) in the Hvidoere Study Group on Childhood Diabetes (HSG) remission phase cohort [14].

The study population representing 15 countries in Europe and Japan was collected through HSG and is described in Mortensen et al 2009 [14]. The cohort included 126 girls and 131 boys, 84% of the patients were white Caucasian, and age at clinical diagnosis was 9.1 ± 3.7 years (mean ± SEM), BMI 16.5 ± 3.2 kg/m², and HbA_1c 11.2 ± 2.1% at the time of diagnosis. DKA (HCO₃ ≤ 15 mmol/l and/or pH ≤ 7.30) was present in 20.7% of the cases at the time of diagnosis.

Exclusion criteria were: suspected non-type 1 diabetes (type 2 diabetes, maturity-onset diabetes of the young (MODY) or secondary diabetes), decline of enrolment into the study by patients or parents, and patients initially treated outside of the centres for more than 5 days. There were no significant differences with respect to gender distribution, age, anthropometric data, HbA_1c at diagnosis, ethnicity or family history of diabetes between patients included and patients not included into the study (data not shown). The diagnosis of type 1 diabetes was according to the World Health Organization criteria. The study was performed according to the criteria of the Helsinki II Declaration and was approved by the local ethic committee in each centre. All patients, their parents or guardians gave informed consent.

In order to estimate the residual beta-cell function (C-peptide and proinsulin) a liquid-meal Boost™-test (6 ml/kg (max: 360 ml, Mead Johnson, Evansville, IN, USA; 237 ml = 8 FL OZ contains 33 g carbohydrate, 15 g protein and 6 g fat, a total of 240 kcal)) was carried out at 1, 6 and 12 months (± 1 week) after diagnosis in all 257 children with newly diagnosed type 1 diabetes. HbA_1c, IDAA_1c, insulin regimen, HLA typing, antibodies (except ZnT8Ab) and liquid-meal stimulated C-peptide levels were analyzed centrally [14]. Liquid-meal stimulated proinsulin was analysed by a sandwich ELISA assay using two monoclonal antibodies. The assay detects total proinsulin as well as the four metabolites: split(32-33), des(31-329, split(65-66) and des(64-65)-proinsulin. The detection limit is 0.3 pmol/l and the interassay coefficients of variation are <8.7%. This assay has no cross reactivity with insulin, C-peptide, IGF-I and IGF-II. ZnT8Ab measurements were done as described in [15]. C1858T genotyping (rs2476601) was done using an in-house KASPar system at KBioscience, UK.

Stimulated C-peptide (logarithmic) and stimulated proinsulin (logarithmic) at 1, 6, 12 months after diagnosis and HbA_1c, daily insulin dose and IDAA1C at 1, 3, 6, 9, 12 months after diagnosis were analyzed as dependent variables in two separate multiple regression repeated measurements models with unstructured variance with gender, age, genotype and IA as explanatory factors. The assumption of constant effect of genotype was checked by first allowing for interaction between genotype and disease duration. Results are given as the estimated factor between the T allele carriers (CT and TT) and the CC genotype. Autoantibodies were examined by the non-parametric Wilcoxon test for single time points. The relationship between IA and proinsulin was tested by multiple regression analysis at each time point with IA (logarithmic), age, gender and stimulated C-peptide (logarithmic) as explanatory factors. There were no statistical differences between the variables according to genotype groups whether the statistical analyses were performed with or without the four Japanese patients. P values below 0.05 were considered statistically significant.

When analyzing the effect of the PTPN22 gene variant on the residual beta-cell function in a repeated measurement model (including 1, 6 and 12 months) we did not find an effect on the stimulated C-peptide levels the first year after onset (est.: 1.018, p = 0.88) (Figure 1A). However, when analyzing the relation between proinsulin and carriers of the C1858T variant we find CT and TT carriers had significantly higher proinsulin (30%) levels over the 12 month period compared to the CC genotype group (est.: 1.30, p = 0.03) (Figure 1B). When adjusting for the IA levels in this analysis, we found a significant association between proinsulin and the IA levels (est.: 1.12, p = 0.002), which, however, did not influence the association between the PTPN22 gene variant and proinsulin (est.: 1.28, p = 0.03). Furthermore, the CT and TT genotype carriers had a borderline significant higher proinsulin/C-peptide ratio in comparison with the CC genotype carriers (est.: 1.25, p = 0.05).

There were no significant differences in the IDAA1C levels among the PTPN22 carriers, an HbA1c and insulin weighted indirect measure of the residual beta-cell function [13]. Nor did we find any differences between carriers of the PTPN22 variant and the glycemic control (as assessed by HbA_1c) or daily insulin dose.

Finally a significant association was observed between carriers of the CT and TT genotype groups and high IA levels at 12 months after onset (est.: 1.56, p = 0.05) (Figure 2), while there were no significant difference in the IA levels at 1 and 6 months. The prevalence of ZnT8Ab was 68% 1 month after disease diagnosis (time point nearest to disease onset) for either ZnT8RAb and/or ZnT8WAb. There was no association between the PTPN22 variant and the other diabetes-related autoantibodies (ICA, IA-2A, GADA and ZnT8Ab), neither close to onset or later during disease progression.

The PTPN22 gene is shown in numerous studies to be associated with the development of type 1 diabetes and other autoimmune diseases. The gene encodes a lymphoid tyrosine phosphatase (LYP) which by dephosphorylation of Src family kinases negatively regulates T cell receptor (TCR) signalling. The current working hypothesis suggest that the risk carrying allele, T1858, suppresses TCR signalling more efficiently during thymic development resulting in survival of autoreactive T-cells [16].

Bottini N and co-workers [16] suggested the use of PTPN22 SNPs as a prognostic factor for disease severity and variability in autoimmune diseases, but also requested further studies taking this point under investigation. One such study was recently reported in 120 new onset type 1 diabetes patients (from 5 to 36 years) [6]. The authors found an association between the T allele of the C1858T variant (using a dominant model) and low fasting C-peptide (as a surrogate marker of residual beta-cell mass) and poorer glycemic control (HbA_1c) from onset and during 12 months follow-up. The relationship between carriers of the T allele and proinsulin was not investigated. The present study, which was conducted in 257 children (from 2 to 16 years) (Table 1) with new onset type 1 diabetes, did not find a direct correlation between stimulated C-peptide and the PTPN22 variant (Figure 1A), but suggests an association between carriers of the T allele and high proinsulin throughout the first year after disease onset (Figure 1B). In accordance with this, an increased proinsulin/C-peptide ratio was evident in the T allele carriers. Previously, elevated proinsulin and proinsulin/C-peptide ratio was found in non-diabetic first-degree relatives positive for islets autoantibodies to identify individuals with increased risk of developing type 1 diabetes within 5 years [17]. Furthermore, elevated proinsulin levels were found to reflect an impaired beta-cell function in type 2 diabetes patients [18]. Thus, the increased proinsulin level and proinsulin/C-peptide ratio might either be due to impaired proinsulin processing or increased secretory demand on the beta-cells resulting from either autoimmunity or hyperglycemia induced residual beta-cell stress.

The discrepancies in significant/non-significant findings on C-peptide levels between this study and the study mentioned above [6] may relate to the use of fasting contra stimulated C-peptide. In the fasting state the differences between the genotype groups and C-peptide are apparent while during liquid-meal stimulated conditions the beta-cells are probably capable of compensating the hyperglycemia but with a rise in proinsulin indicating the stressed residual beta-cell function. Furthermore, the differences in the mean age between the patients from the two studies (14.9 yrs vs 9.1 yrs, Petrone et al vs. our study, respectively) might also partly explain the different results. Younger children have less residual beta-cell function as assessed by stimulated C-peptide the first year after diagnosis compared to older age groups [13]. GADA were not related to the PTPN22 variant, supporting previous findings that impact of this gene on the presence of GADA may only be observed in patients with long disease duration (> 10 years) [12].

Our results align with previous observations that PTPN22 gene variant may be associated with changes in residual beta-cell function and disease pathogenesis during the first year after onset of type 1 diabetes.