Introduction

Type 1 diabetes (TID) is a chronic and immune-mediated disease, characterized by destruction of β-cells, the insulin-producing cells.^.(1

Consequently, there is a deficiency in secretion of this hormone, crucial in the metabolic regulation of carbohydrates, enhancing a permanent state of hyperglycemia.² About 90% of initial TID cases arise in children and adolescents, reflecting approximately 500 000 pediatric patients worldwide, 26% of whom are European.^3,4,5,6 The incidence rate of TID has been increasing in recent decades, by around 3%-4% per year, especially in younger children.⁷ Some studies report a recent disproportionate increase in children under 5 years of age, with a predisposition to double in the future.^8,9,10 In 2015 in Portugal, children aged 0-14 years had a prevalence of TID of 0,13% and an incidence of 13,3 new cases per 100.000 children.¹¹

The etiology of DM1 is multifactorial, to which genetic, environmental, and immune systems contributes. The potential trigger underlying pancreatic endocrine failure remains unknown, however, viral infections are currently one of the most consensual hypotheses.^12,13 The HLA-DR-DQ haplotype appears to be the most important genetic determinant, as it contributes more than 50% to the individual genetic risk for developing TID.^6,7,8,9,10

About 70%-90% of patients had an abnormal activation of the immune system, which leads to autoantibodies whose target are β-cell proteins.⁷ More than 90% of individuals newly diagnosed have at least two positive autoantibodies (AA+) and the order of appearance seems to be related to the HLA-DR-DQ haplotype.^14,15 Thus, autoimmunity arises months to years before diagnosis, with a mean time of 3.5 years from seroconversion to the manifestation of the disease.^{6,7,8,9,10,11,12,13,14,15,16}

Autoantibodies are biological indicators of β-cells damage, although they are not directly responsible for tissue destruction. This chronic inflammatory process is triggered by mononuclear cells and TCD8+ lymphocytes.^1,10,17 When this destruction rises above 90%, the symptoms of hyperglycemia appear.¹⁸ The autoantibodies involved in this process are islet cell antibodies (ICA, against cytoplasmic proteins in the β-cell), antibodies to glutamic acid decarboxylase (GAD-65), anti-insulin (IAA) and anti-zinc transporter 8 (ZnT8).^18,19,20 The risk of progression to a symptomatic stage is related to the number of existing autoantibodies and with the age of seroconversion. In case of two AA+, the risk of developing TID during childhood or adolescence is around 80%.^12,15,16

The presentation of TID in young people may present with symptoms characteristic of hyperglycemia, such as polyuria, polydipsia, polyphagia, asthenia and unintentional weight loss.^5,19 In more severe cases, diabetic ketoacidosis may be the first sign of TID, secondary to an absolute or relative insulin deficiency combined with an excess of counter-regulatory hormones. At an early stage, as the bicarbonate ion is still found normal, it could neutralize ketone bodies, so ketosis may occur without acidosis.^19,21

According to international consensus, if a child has glycated hemoglobin ≥6.5% or classic signs and symptoms with serum glucose ≥ 200 mg/dL, we are facing a diagnosis of diabetes^5,19,22. In TID, it is expected that there will be positivity for at least one autoantibody against β-cells¹⁹. Glycated hemoglobin is an indicator of great clinical utility, as it translates average blood glucose over the past 4-12 weeks. Taking into account that the half-life of erythrocytes is approximately 120 days, glycated hemoglobin should not be used as a diagnostic criterion if we are facing clinical conditions that affect the normal turnover of erythrocytes.^22,23,24

Nowadays there are few epidemiological data regarding the immune scenario underlying TID in the pediatric population worldwide, especially in Portugal.¹⁰ Thus, it is highly relevant to carry out studies that allow us to understand which autoantibodies are affected, their association with clinical and laboratory parameters (glycemia and glycated hemoglobin) and with the age of disease presentation. Given the prevalence, chronicity and high burden of disease associated with TID, it is important to expand the knowledge of autoimmunity to allow us to understand the pathological mechanisms and further that to explore potential therapeutic targets in the future.

Our major aim is to assess the influence of autoantibody positivity in the clinical symptoms, HbA1c and glucose at diagnosis. A secondary aim is to clarify if gender and age at diagnosis have a role in the clinical symptoms and laboratory parameters in children with inaugural TID up to 10 years of age.

Material and Methods

An observational, retrospective and analytical study was performed in the Pediatric Endocrinology and Diabetology Unit at Hospital de Braga including all children with inaugural diagnosis of TID aged 10 years and below. The exclusion criteria were: children diagnosed with TID beyond 10 years of age; children with neonatal diabetes; children whose clinical record does not contain information relating to any of the four autoantibodies under study; children with other clinical conditions that may interfere with the real value of the glycated hemoglobin, either by reducing or increasing it, namely: hemoglobinopathies, anemia hemolytic, hematologic neoplasms, nutritional deficiency of folic acid, vitamins B12 or B6, hyperthyroidism, chronic kidney disease, nutritional iron deficiency or any other conditions that promote an increase in the number of erythrocytes and/or hematocrit.^22,23,24,25

All data collection for the study was carried out based on clinical records, through Glintt^® software, detained by Hospital de Braga and included:

The antibodies and laboratory parameters were measured in the same laboratory.

For the statistical study, the Statistical Package for the Social Sciences program in version 26.0 was used for the Window operating system (SPSS, Inc).²⁶⁾

Primarily, the sample was characterized using descriptive statistics. At continuous quantitative variables that fulfilled the normality assumptions were represented through the mean±standard deviation (M±SD), while the rest were presented using the median (Mdn) and interquartile range (IQR). Regarding the categorical variables, these were described using absolute frequencies and percentages. The normality of the quantitative variables was tested using the Shapiro-Wilk test, from histogram and asymmetry and kurtosis values. When quantitative variables did not show a normal distribution, non-parametric tests were performed. For each individual type of autoantibody (ICA, GAD-65, IAA and ZnT8), patients were divided into two groups (AA+ vs AA-), making a total of 4 pairs of different groups. To assess the existence of differences between each two groups of autoantibodies (positive/negative), in concern age, gender and clinical and laboratory characteristics of the patients (glycated hemoglobin values, glycemia at admission and symptoms), comparisons were made between these groups. Additionally, patients were divided into two age groups: age ≤60 months and age >60 months. To assess differences between the two groups, these were compared regarding the type and number of AA+, gender and clinical and laboratory characteristics. The type of statistical test applied was suitable for the type of variable dependent (VD).

Thus, for continuous quantitative variables with a normal distribution, the test was performed T-Student for independent samples (t), the Levene’s test was applied to assess homogeneity of variances and the d-Cohen (d) was calculated as a measure of effect size. For the quantitative variables that did not meet the normality assumptions, the Mann-Whitney (U) test was used and r as measure of effect size. Additionally, the Chi-Square (X) test was performed for categorical VDs. As a measure of effect size, the Phi (φ) was presented when the variables were dichotomous and Cramer's V (φc) when the variables had more than two categories.

The magnitude of the effect was classified as "weak", "intermediate", "strong" or "very strong", according to what is established for the effect size of each test.^27,28 In order to determine the impact of the autoantibody(s), age and gender in the presence of ketoacidosis at diagnosis, glycemia and glycated hemoglobin values at admission, regressions were performed. Considering the VD ketoacidosis (present/absent), the adjusted odds ratio (OR) was calculated through the realization of a binary logistic regression (BLR). To predict quantitative outcomes (glycemia and glycated hemoglobin), the 𝛽 (standardized coefficient) was presented through the execution of a multiple linear regression (MLR).

Ethical Committee of Hospital de Braga and Ethical Committee of School of Medicine of University of Minho approved this study.

Results

This study included 95 patients whose demographic, immune, clinical characteristics and laboratory tests are represented in Table 1. There was a 51.6% of gender male, with a median age of 71 months. Regarding the autoantibodies, the majority were negative; related to the minority with positive antibodies, there was a predominance of ICA (82.1%) and GAD-65 (32.6%). At diagnosis, 41.4% of the children presented with ketosis without acidosis. The median glycemia on admission was 484 mg/dL and the mean glycated hemoglobin was 11%.

Discussion and Conclusion

In our study, 48.4% of the children had at least one AA+ and the male gender was the one that expressed the greatest global positivity, findings not corroborated by the literature, which reports approximately twice as many patients with AA+ at diagnosis,^1,29,30,31 with female preference.²⁹

Of the four types of autoantibodies analyzed, GAD-65 was the most prevalent (32.6%), followed by the IAA (23.2%). On the other hand, the largest number of negatives was verified in the ICA autoantibody (82.1%). The literature classifies the GAD-65 as the most prevalent autoantibody in TID patients, followed by IAA.^20,30,32,33 The detection of autoantibodies tends to decrease with age, and ICA autoantibody appearing to be the fastest to become negative with the progression of TID. Its frequency drops to 25% at 5 years and <5% at 10 years of disease, so this may be a plausible justification for our majority negativity.^1,20,34 In this sample, the antibodies panel may not have been performed exactly at the time of inaugural diagnosis, which may have conditioning a decrease of its levels to undetectable values when harvesting, something expected to happen with the progression of the disease.^29,35 However, it can be speculated that in 51.6% of our patients, autoantibodies will develop after diagnosis, translating a TID currently with no humoral response or, alternatively, with humoral activity but against antigens not officially recognized as part of this pathology, as is the case of tetrasparin-7 recently identified.^29,36 Another hypothesis to be considered is the possibility that some of these children may have another subtype of diabetes where the destruction of 𝛽-cells does not appear to be humoral, but eventually related to another cellular pathway or genetics.^19,29,33

As stated by Bravis V et al²⁹⁾ and Perchard R et al,³¹ our study also showed that having a certain type of autoantibody did not determine statistically significant differences regarding demographic, clinical parameters and laboratory tests of these children. According to the most recent international guidelines, ICA autoantibodies are no longer included as markers of autoimmunity for the diagnosis of TID, so its dosage is no longer in use.^19,22 The analysis of this autoantibody becomes more relevant in cases of negativity for the remaining autoantibodies.³⁷ In the present study, no statistically significant differences were found between the GAD-65 groups regarding the presence of ketoacidosis, a result similar to that described by Sabbah E et al.³³ However, as it is stated in other studies, GAD-65 appears to be more common in older and female patients.^32,33,38,39 Additionally, according to the literature, the IAA seems to be the first autoantibody to emerge and is significantly positive at younger ages, something that was not verified in our study.^{1,10,32,33,40} Despite the ZnT8 autoantibodies not showing significant differences between groups, a very important impact in the clinic at admission was detected (φc=0.42). Probably if there were more data regarding this autoantibody (78.6% omitted), statistically significant differences would be expected proven, as the size of the effect reflects an apparent relevance of ZnT8 in clinical parameters, as was evident in the study performed by Niechciał E et al.⁴¹

Regarding the clinical and laboratory characterization of this sample, there was a slight predominance of males (51.6%), an aspect consistent with the literature.^{10,11,29,30,39} Between our 95 patients evaluated, 36.4% manifested ketoacidosis at admission. This represents one of the complications with the highest risk of morbidity and mortality from TID, although it has a highly variable frequency around the world, around 15% to 67%, possibly due to the heterogeneity of the criteria implemented in its definition.^{20,30,40,42,43} In our sample, ketosis without acidosis was the main form of presentation of inaugural TID (41.1%), so it is assumed that there is a greater perception of hyperglycemic symptoms by the caregivers, preventing more severe metabolic decompensation in the form of ketoacidosis. Another important finding is the significant differences in the percentage of glycated hemoglobin value at diagnosis between age groups, as patients diagnosed over 60 months presenting a percentage superior of this laboratory parameter (association of intermediate effect), finding in agreement with the literature.^44,45 Thus, older patients probably have worse metabolic control in 4-12 weeks prior to diagnosis, translating a longer process of pancreatic failure and possibly, similarly to that described by Hanberger L et al,⁴⁵ with glycated hemoglobin values also higher during follow-up and facing higher risk of future complications.^46,47 A plausible explanation for these facts is that younger children, through a destructive autoimmune or cellular process, present a more rapid and extensive destruction of 𝛽-cells. This results into an abrupt insulin deficit, with a shorter duration of hyperglycemic symptoms and a tendency to severe and acute presentations in the form of ketoacidosis.^43,44,45,48 In contrast, no differences were reported between groups regarding glycemia at admission and immune characteristics, but most of the literature reports a significantly higher presence of the IAA autoantibody as an early marker of pancreatic destruction in younger children^{6,10,32,40,44)} and of GAD-65 in older children.^29,48 Similarly to our study, Chen YC et al⁴⁴ and Komulainen J et al⁴⁰⁾ did not report age differences regarding ICA and GAD-65 autoantibodies.

The percentage of glycated hemoglobin proved to be the only independent predictor statistically significant in the occurrence of ketoacidosis at diagnosis, adjusting sample for age, gender and serology of the ICA. For every percentage increase in glycated hemoglobin value, there is a 66% increase chance of ketoacidosis. Cherubini V et al⁴³ concluded that children under 4 years are twice as likely to have ketoacidosis than older girls. Niechciał E et al⁴¹ reported that AA ZnT8+ increases the risk of ketoacidosis, not corroborating by your study.

Considering children with the same age, gender and serology as the ICA, for each increase in 1 month of age, a significant increase in glycated hemoglobin value is expected and this finding is according to the most recent literature. In this sample, male sex and higher percentages of glycated hemoglobin values are related to higher glycemic values and older children have, in general, lower glycemia values. A higher incidence, reported in the literature, of metabolic decompensation in younger children may justify this last finding.^43,44,45,48

As this is a retrospective study, it was carried out based on pre-existing data, incomplete in some cases, especially about the serology of autoantibodies, which exponentially limited the results underlying. The small size of this sample and the high amount of missing data regarding autoantibodies may justify the lack of documented statistical significance. However, it is important to emphasize that the AA ZnT8 only started to be routinely dosed recently, which may justify its majority percentage of missing data.

The recent increase in inaugural diagnoses in younger age groups encourages carrying out larger studies aimed at understanding the influence of the immune context in the clinical and laboratory characteristics of the pediatric population diagnosed with TID. As a conclusion, authors would like to emphasize the importance of standardized and protocoled studies in the context of autoimmunity that allow to stratify patients into potential immune phenotypes and promote the discovery of new therapeutic targets, encouraging a future based on Precision Medicine.