Type 1 diabetes in Brazil: a narrative overview of the Brazilian Type 1 Diabetes Study Group

Abstract

Background

To evaluate the real-world care of Brazilian individuals with type 1 diabetes (T1D).

Methods

The Brazilian Type 1 Diabetes Study Group (BrazDiab1SG), was conducted in two waves, in all Brazilian geographical regions. The first enrolled 3,591 individuals with data obtained from medical records. The second evaluated 1,760 individuals with laboratory and genetic data performed at State University of Rio de Janeiro.

Results

In the first and second waves, the average HbA1c was 9.3% and 9.0%, HbA1c above 10%, 32.2% and 26.3%, and individuals with HbA1c < 7.0%, 13.2% and 13.5%, respectively. Intermediate/long plus short acting insulins was the commonest used treatment modality. Less than 30% of the individuals achieved good glycemic control. Individuals in pumps and insulin analogs had lower mean HbA1c. Diabetic retinopathy, chronic kidney disease, cardiac autonomic neuropathy, peripheral neuropathy and macrovascular diseases were found in 35.7%; 25.2%; 23.4%; 14.8% and 2,1%, respectively. The 10-year Steno type 1 risk engine (S1TRE) showing moderate and high risk was noted in 18.0% and 11.3% of individuals. Diabetic retinopathy was associated with all diabetes-related chronic complications. Severe hypoglycemia was reported by 19%, periodontal disease by 4. % and autoimmune diseases by 19.5% of individuals. European genomic ancestry prevailed in the whole sample even in those Black self-reported. HLA-DRB1*03:01 ~ DQA1*05:01 ~ DQB1*02:01 was the most frequent risk haplotype with the highest frequency in individuals who self-reported as White and the haplotype HLA-DRB1*09:01 ~ DQA1*03:01 g ~ DQB1*02:02 had the highest frequency among those self-reported as Black. European genomic ancestry was associated with the risk alleles: DRB1*03:01; DRB1*04:01; DRB1*04:02; DQA1*05:01 DQB1*02:01; DQB1*03:02 and African genomic ancestry with the protective alleles: DRB1*03:02; DRB1*11:01; and DRB1*15:03. Amerindian (43.6%, haplogroup C) and African (38.2%, haplogroup L3) were the most frequent matrilineal ancestries (MtDNA). European patrilineal ancestry was predominant, in approximately 85%, followed by African ancestry in approximately 9%. European haplogroups R1b and E1b were the most prevalent.

Conclusions

In both waves, Brazilian individuals with T1D presented poor glycemic control and high rates of diabetes-related acute and chronic complications. Our data has been used to improve T1D management driving public health policies in Brazil.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13098-025-01708-6.

Introduction

Type 1 diabetes mellitus (T1D) affects mostly children, adolescents and young adults and is associated with high morbidity and mortality rates. Due to the disease itself and its related complications, it is estimated that these individuals experience, a significant decrease in life expectancy, a worsening in quality of life and a significant decrease in work capacity, resulting in high direct and indirect costs for National Health Systems and the society.

In the last decades, a worldwide growing incidence of T1D has been observed in developed, developing and underdeveloped countries as well as among minorities [1]. Despite this, few national multicenter studies have been carried out in developing countries, such as Brazil, in underdeveloped countries and with minorities, which are characterized by vast socioeconomic, ethnic and genetic diversity that probably influence the diagnosis, treatment and the occurrence of T1D-related acute and chronic complications.

In Brazil, there was little data on the assessment of care for individuals with T1D, and when these data existed, they were limited to specific topics and few locations. In order to understand the reality of T1D in our country, the Brazilian Diabetes Society (SBD) conducted the first wave of a national multicenter study, the so called Brazilian Type 1 Diabetes Study Group (BrazDiab1SG) with partial grants from Farmanguinhos, aiming to evaluate demographic, personal and social characteristics of Brazilian individuals with T1D. Data regarding diagnosis, types of treatment and its costs, adherence to treatment, clinical-metabolic control and quality of life were collected. This wave was performed between December 2008 and December 2010 in 28 secondary and tertiary level of care public clinics located in 20 cities in all Brazilian geographic regions (North/Northeast, Midwest, Southeast and South). Overall, 3,591 individuals were enrolled (56% females, 57.1% Caucasians, with a median age of 19 years (1–66), median age at diagnosis of 10 years (< 1–44) and T1D duration of 7 years (< 1–50). The diagnoses were made between 1960 and 2010. The following variables were assessed through an interview during a clinical visit: current age, age at diagnosis, diabetes duration (years), height (m) and weight (kg), blood pressure (systolic and diastolic in mmHg), modalities of treatment for T1D and its comorbidities, frequency of self-monitoring of blood glucose (SBGM) and smoking status. The levels of HbA1c, fasting plasma glucose, fasting total cholesterol, LDL cholesterol, HDL cholesterol and triglycerides were obtained from medical records. It was also determined whether screening for diabetic retinopathy (DR) by fundoscopy, for nephropathy by microalbuminuria, and a feet examination were performed within one year of the study assessment, in those individuals with T1D duration longer than 5 years.

The second wave was conducted from August 2011 until August 2014 in 14 diabetes centers in 11 cities, with the participation of 1,760 individuals, 55.8% females, 54.4% Caucasians, mean age 30.9 ± 11.1 years, mean age at diagnosis 14.7 ± 8.9 years, with T1D duration of 15.4 ± 9.2 years. All individuals were diagnosed from 1960 to 2014. Among these individuals, 650 (37%) had also participated in the first wave. This second wave has received grants from Conselho Nacional de Pesquisa (CNPq), Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, (CAPES) which made possible to standardize all laboratory tests, that were performed in a single center (State University of Rio de Janeiro). HbA1c, creatinine, urea, triglycerides, total cholesterol, HDL cholesterol and LDL cholesterol, were measured using enzymatic techniques (BioSystem). Thyroid stimulating hormone (TSH), free levothyroxine (FT4), thyroid peroxidase antibodies (Anti-TPO) and vitamin B12 were measured by electrochemiluminescence (Cobas). HbA1c was measured using high-performance liquid chromatography (HPLC, Bio-Rad Laboratories, Hercules, California, USA). HbA1c at goal (good glycemic control) was defined as HbA1c < 7.0% (53 mmol/mol) [2–5]. Inadequate glycemic control was defined as HbA1c ≥ 7.0% (53 mmol/mol). Serum uric acid was measured using an uricase-based commercial kit (BioSystem) with results expressed in milligrams per deciliter (mg/dl) and normal values ranging between 3.5–7.2 mg/dl for men and 2.6–6.0 mg/dl for women. Friedewald’s equation was used to calculate LDL cholesterol values. Creatinine was measured using a colorimetric assay kit (Biosystems), corrected for standardized creatinine assay by mass spectrometry. C-reactive protein (CRP) was measured by immunoturbidimetry (Biosystems). In this wave the presence of all diabetes-related acute and chronic complications was investigated as well genetic markers for T1D. The complete list of received research funding and respective agencies for the BrazDiab1SG are described in supplementary Table 1.

In both waves, the included individuals were diagnosed with T1D by physicians according to the SBD [2] and American Diabetes Association (ADA) criteria [3] (i.e., continuous use of insulin since diagnosis), were receiving medical care for at least 6 months at the respective diabetes center, and were regularly treated at secondary and tertiary levels of care centers in each different city (population > 100,000 inhabitants). Each center provided data on at least 50 individuals. Those individuals with missing data and who did not fulfill these criteria were excluded, such as pregnant or lactating women and those who had an acute infection or ketoacidosis in the 3 months preceding the recruitment. Written informed consent was obtained from all individuals and/or from their parents where necessary.

The strength of this study was the coverage of a large sample of individuals, quite representative of the Brazilian population with T1D, with various racial backgrounds, from all geographic regions of Brazil, following a uniform recruitment protocol in different participating centers and with a uniform definition of T1D.

As limitations, we must mention that individuals from private clinics were not included, and that autoantibody and C-peptide levels were not measured, which could suggest that other types of diabetes could have been included in our sample. However, 93.1% of individuals were diagnosed < 30 years of age, which strongly suggests that they actually had T1D. Furthermore, the vast majority of individuals lived in large cities and only 14% lived in rural areas, which may have led to a negligence in the evaluation of individuals treated in primary care, which could also cause a selection bias. Finally, since the study design was cross-sectional, associations can be identified but causality cannot be inferred.

The purpose of this overview was to summarize all data that has been collected and published with the database obtained in the BrazDiab1SG.

Epidemiology

In parallel with the BrazDiab1SG, due to the scarcity of data on the incidence of T1D in Brazil, two surveys were carried out in Bauru, State of São Paulo, one of the BrazDiab1SG participating centers. The first was performed from 1986 to 2006, and the second, from 1986 to 2015. It was found that between these two surveys, the incidence increased from 10.4 to 12.8/100,000 in children < 15 years old/year, which categorized Bauru as a city with a medium/high incidence rate, according to the DiaMond project, with an overall annual increase of 3.1%. Most of the diagnosis occurred in cold months, mainly in the 10–14 years age group, with no significant difference between genders. Although no difference was observed according to self-reported color-race it is noteworthy that almost 20% of the individuals self-reported as being non-White [6, 7].

Diagnosis

Regarding T1D diagnosis, it was observed that the majority of individuals (42.3%) had their diagnoses made in ketoacidosis, 39.4% through fasting plasma glucose, 14.4% with a random blood glucose measurement, 1.8% through an oral glucose tolerance test and 2.1% through other different ways. Ketoacidosis was more frequently found among individuals with low socioeconomic status, mainly in the southeast region and in the 0–4.9 years age group. It was also observed that the incidence of ketoacidosis at diagnosis was 26.5% lower in those individuals diagnosed after the year 2000 [8]. A decrease in diagnoses in ketoacidosis was observed in the second wave (37.2%) with an increase in diagnoses by random plasma glucose (20.5%). These data are shown in Fig. 1. This could be due to an increase in awareness of clinical signs and symptoms of diabetes decompensation by general pediatricians and general practitioners as has been observed in other studies [9]. It is noteworthy that ketoacidosis at diagnosis is still an important cause of death and loss of beta cell function [10].

Fig. 1.

Diagnosis of Type 1 diabetes.Data from Brazilian Diabetes Type 1 Study Group l (BrazDiab1SG)

Glycemic control

The average HbA1c was 9.3% and 9.0% in the first and second waves, respectively, both quite far from the targets proposed by SBD [2] and ADA [3]. Overall, 54% of the participating centers in the second wave had an overall HbA1c mean below the overall mean of each enrolled center.

Another point to be highlighted was the number of individuals with HbA1c above 10%, which was 32.2% and 26.3%, in the first and second waves, respectively. No statistical difference between the two waves was found. On the other hand, the number of individuals with HbA1c < 7.0% was 13.2% and 13.5%, in the first and second waves, respectively, without difference between waves [4]. Approximately 13% of individuals did not have HbA1c determination in the previous year. These data are shown in Fig. 2.

Fig. 2.

Percentage of patients according to the level of HbA1c. Data from Brazilian Diabetes Type 1 Study Group l (BrazDiab1SG)

In the first wave, more children achieved glycemic control goals compared to adolescents and adults, being of 29.2%, 17.9%, and 11.6%, respectively [11]. In the observational T1D Exchange Registry, HbA1c mean was 8.4%, with more adults and children achieving good glycemic control compared to adolescents. It is noteworthy mentioning that in this study the majority of individuals (60%), were using insulin pumps, in contrast with the low number (3,5%) of Brazilian individuals on this therapeutic modality [12].

In the second wave, which did not include children, good glycemic control was found in 17.5% and 14.5% of adolescents and adult individuals, respectively [13, 14]. In general, good glycemic control was associated with insulin therapeutic regimen type, socioeconomic status, years of school attendance and adherence to the prescribed diet [5, 14]. A relationship was found between knowledge about diabetes management (e.g., what HbA1c means) and better glycemic control, mainly in individuals with higher length of school attendance, better socioeconomic status, longer follow-up in the respective diabetes centers and among those participating in educational programs [15].

Our findings show that achieving good glycemic control in individuals with T1D is still a challenge worldwide.

Self-monitoring of blood glucose

In the first wave, 88.4% of individuals used to perform SMBG. More than 80.0% used to do it three or more times daily. The daily average of SMBG in this study was 3.4 ± 1.8 times. Individuals who performed SMBG most frequently were females, < 19 years old, under more intensive treatment regimens, belonging to higher socioeconomic status and living in the southeast region [16].

No additional benefits were found in individuals who performed SMBG more than five times a day, since more than 40% of these individuals had HbA1c > 9% [16]. In the second wave, 94.5% of the individuals used to perform SMBG with an increase of ~ 7 points% compared with the first wave, but with a similar daily average of SMBG of 3.6 ± 1.4 times daily. In the second wave, individuals answered two questions regarding SMBG: whether they changed their treatment according to SMBG results and which was the most common adopted procedure. Almost 84% answered affirmatively, being changes in insulin doses the most common procedure, performed by 64.5% of the individuals [15].

Cardiovascular risk factors

In the first wave 31% of the overall sample (children, adolescents and adults) presented overweight/obesity [4, 11]. In the second wave, 37% of the overall sample presented overweight/obesity, but these rates decreased to 26% in adolescents [13], and reached 40.7% in adult individuals [17]. In both waves, in all age groups, the presence of overweight/obesity was associated with female gender, higher levels of total and LDL-cholesterol, hypertension and metabolic syndrome which were considered as cardiovascular risk factors (CVRF) [18].

Risk factors for the occurrence of metabolic syndrome were age, female gender, uric acid levels and the presence of acanthosis nigricans. It should be mentioned that in individuals with metabolic syndrome (32.3%), there was an increase in risk markers for non-alcoholic hepatic steatosis, such as transaminases, uric acid and triglycerides (> 129.5 mg/dL) [18]. Regarding overweight/obesity, an earlier age at menarche (< 11 years), that was found in 23.4% of our individuals, was a strong risk factor for its development in pubertal or adult life [19]. A linear association between parity was also found to be a risk factor for increased body mass index (BMI) [20].

It is noteworthy that adherence to the prescribed diet, adopted by 54.2% of individuals, was associated with a partial decrease in CVRF [14]. Those individuals who followed the prescribed diets at least 80% of the time, were those who used to perform carbohydrates count, followed diets set out in medical societies guidelines, followed the basal-bolus insulin protocol and used insulin pumps. Adherence to the prescribed diet decreased HbA1c levels by 1.1% [14].

In the first wave, hypertension was present in 19.2% of individuals, being more frequently found among adults than in children/adolescents, in 31.3% and 8.3%, respectively. Age, longer diabetes duration, male gender, higher BMI and higher levels of total cholesterol, LDL-cholesterol, triglycerides and ethnicity (non-Caucasians) were the factors more frequently associated with the presence of hypertension [21]. Only 53.7% were receiving treatment, and only 22.3% of those treated reached the targets for systolic and diastolic blood pressure. Moreover, 7.5% of these individuals had not been evaluated in the previous year and only 65.5% were aware of having hypertension and knew the importance of treating this condition [18].

In the second wave, the overall prevalence of hypertension was 17.7%, being also more frequent in adults than in adolescents, 22.2% and 1.0%, respectively. Differences in the prevalence of hypertension between the two waves could be due to active and standardized measurements performed in the second wave. The number of individuals using anti-hypertensive agents increased to 83.9%, while 16.3% of them did not receive any type of treatment [22].

In the first wave, 57.9% and 52.7% of children/adolescents and adults, respectively, had their LDL-cholesterol at goal, and the use of statins was found to be done in 3.3% and 14.8% of children/adolescents and adults, respectively [4, 11]. In the second wave, 54.1% of the individuals were out of the goal for LDL-cholesterol, being 20.2% adolescents and 79.8% adults, but only 5.2% and 26.1% of the adolescents and adults were using statins, respectively [13, 15]. It is important to highlight that the correlation between serum lipids and HbA1c was heterogeneous across the spectrum of glycemic control. However, a worsening of LDL-cholesterol and triglycerides was observed with an increase in HbA1c levels [22].

Regional differences in clinical parameters

Several regional differences were found in many evaluated parameters, in the first wave [23]. Overall, 18.4% of individuals had HbA1c < 7.0% and 47.5% ≥ 9%. Higher HbA1c values were not related to the adopted insulin therapeutic regimen nor to the Brazilian geographic region. Hypertension was more prevalent in midwest. In southeast, more individuals were found with LDL cholesterol within the target and using statins, but 30–50% did not reach the proposed goals. Screening for diabetes-related chronic complications, such as DR and neuropathy, was less frequently carried out in the north/northeast and midwest regions. Obesity and overweight were present in 1/3 of individuals, being less prevalent in the north/northeast region [23].

More intensive and complex insulin therapeutic regimens were found in south and southeast regions [23].

Psychological impact

The presence of T1D in a child or adolescent was associated with discomfort/anxiety/depression, especially in mothers, who were in general, the only person in the family involved in the treatment. Changes in family functioning were reported by most parents who felt overwhelmed and reported a lot of concern about the possibility of occurring diabetes-related complications. Many parents reported difficulties in imposing limits on their children with T1D and concerns about letting them sleep outside home [24].

Costs

The annual direct costs per capita for these individuals were more than US$1,300.00, being approximately six times higher for those using insulin pumps. Individuals with more than 15 years of diagnosis showed a 56% increase in costs compared to those with less than 5 years of diagnosis [25].

The highest costs were found in the southeast, south, north/northeast and Midwest regions, respectively, and were related to the presence of microvascular complications and high socioeconomic status [26]. The use of insulin pumps and of insulin analogs (basal and rapid) were different across the geographical regions of the country and could have some influence in the direct costs of the disease.

In addition to the direct financial costs, these individuals had a high frequency of indirect costs such as early retirement which occurred in 4.2%, regardless of gender, at an average age of 35.5 ± 9.3 years, which represented a loss of laboral force of 17.5 ± 9, 1 years. The risk of early retirement was approximately five times greater in individuals with microvascular complications and approximately four times greater in those with macrovascular complications [27]. Mortality rates among individuals from Bauru and Rio de Janeiro (986 individuals) were evaluated, and were found to be three times higher than in the general population, with an average age of 30 years at death, diabetes duration of 15 0.6 years, mainly due to chronic kidney disease (CKD) and ketoacidosis. These data are shown in Fig. 3. Both, early retirement and early mortality were considered as indirect costs for the Brazilian National Health System [28].

Fig. 3.

Causes of mortality. Data from Bauru and Rio de Janeiro

Quality of life

Reducing HbA1c by 1% increased 1.5 points on the general scale used to assess quality of life in the BrazDiab1SG. However, Brazilian individuals with T1D had worse quality of life than individuals from other countries, such as the Netherlands and the United Kingdom [29, 30].

Individuals from the north/northeast regions had better quality of life rates, especially those with shorter diabetes duration, possibly due to lower rates of diabetes-related chronic complications [27, 28].

African genomic ancestry (GA) was also associated with worse quality of life. Each 1% increase in African ancestry decreased 9.5 points in this score [31].

Treatment modalities

Different treatment modalities were analyzed, regarding the type and combinations of insulin used and its injection methods. In the first and second waves, the commonest therapeutic modality was the use of intermediate/long plus short acting insulins both in children/adolescents and adults. Human insulins and insulin analogs were used by almost 83.2 and 91.0% of the patients in the 1st and 2nd waves respectively [23, 32]. Although few patients were using only intermediate or long acting insulin, almost 15% and 5% were still using these insulins in the 1st and 2nd waves, respectively (23,32). Few patients were using insulin pumps both in the first and second waves, 1.5% and 3.5%, respectively (23,32).

The majority of individuals did not achieve HbA1c targets in any used regimen, and there was also no association between therapeutic regimen modality with overweight and obesity. Less than 30% of individuals in any therapeutic regimen achieved good glycemic control [5]. Although a similar number of individuals using conventional therapy, insulin pumps and rapid or long acting analogs achieved HbA1c targets, individuals using pumps and insulin analogs had lower mean HbA1c in comparison to those using conventional therapy [5]. The prescription of insulin analogs or insulin pumps was more frequently found in Caucasian individuals, with higher socioeconomic status and with more years of school attendance [33]. This latter variable was the most importantly associated with better glycemic control [5, 33].

The majority of individuals did not adhere to any type of prescribed therapeutic regimen. Good adherence to insulin treatment, was found in older individuals, with greater adherence to the prescribed diet, and less hypoglycemia, which led to a decrease in HbA1c levels by 0.6% [32].

Diabetes-related chronic complications

The assessment of diabetes-related chronic complications had different approaches between the two waves. In the first wave, data were obtained from medical records regarding the frequency that screening for diabetes-related complications was performed, according to SBD and ADA guidelines. These screenings had to have occurred up to one year before the beginning of BrazDiab1SG. In the second wave, all centers were trained to assess these complications, with support of the central committee. All laboratory evaluation was done in the same laboratory. Consequently, more reliable data regarding these complications were obtained. Data regarding diabetes-related chronic complications and overweight and obesity are shown in Fig. 4.

Fig. 4.

Diabetes chronic complication, cardiovascular disease (CVD) risk * and comorbidities: Data from Brazilian Diabetes Type 1 Study Group l (BrazDiab1SG). *Cardiovascular disease risk was calculated according to Steno type 1 risk engine

First wave (December 2008 until December 2010)

In the first wave, the overall missing data for annual screening for diabetes- related chronic complications in children and adolescents was of ~ 35% [11]. The screening for DR was of 66.2%, for CKD 69.7%, and 62.7% for foot alterations. The screening frequency was not gender-related but a higher rate of screening was observed in those individuals receiving care in tertiary level of care centers compared to those from secondary level of care (p < 0.001).

Overall, missing data for annual screening for diabetes-related complications for adult individuals in the previous year was of 37% [2]. The screening for DR was of 70.1%, for CKD 63.1%, and 63.1% for foot alterations, respectively. As observed in children and adolescents, the screening was not gender-related but a higher rate of screening was observed in individuals from tertiary level of care centers compared to individuals receiving care in secondary level of care centers (p < 0.004). No data regarding peripheral and autonomic neuropathy and macrovascular complications were described. These data are shown in Fig. 1

Second wave (August 2011 until August 2014)

In the second wave, the following diabetes-related chronic complications were evaluated: DR, CKD, periodontal disease, autonomic and peripheral neuropathy and macrovascular disease. The Steno type 1 risk engine was also calculated. In a subgroup of individuals from one participating center, non-alcoholic fatty liver disease (NAFLD) was also investigated. In this wave, traditional and non-traditional risk factors including serum and genetic biomarkers were also investigated. These data are shown in Fig. 2.

Diabetic retinopathy

Considering that DR is currently one of the main causes of blindness and visual impairment worldwide, its screening is absolutely necessary. We used and compared two methods for its approach, the binocular indirect ophthalmoscopy (BIO) and telemedicine protocols of digital retinography (mydriatic and non-mydriatic) in our sample of individuals with T1D [34]. In this latter analysis, 1,266 individuals from seven participating centers were enrolled. The results showed a good agreement (kappa 0.67–0.74) between BIO (DR absent or minimal) and referral DR (moderate or severe non-proliferative DR, proliferative DR, or apparently present diabetic macular edema) and a poor agreement (kappa 0.58) between BIO and non-mydriatic retinography. The agreement between the two methods was poor for macular edema screening.

According to these results DR screening strategy with conventional non-mydriatic retinography was not recommended. So, the choice of the central committee was to evaluate DR in the overall sample by BIO since the retinal digital camera was not available in all centers.

A total of 1,644 (93.4%) individuals were investigated. Overall, 35.7% presented DR and 12% presented vision-threatening DR which included also macular edema [35]. Referable DR was observed in 18% of individuals [36]. The most important risk factors for the presence of any type of DR were diabetes duration, current smoking, presence of hypertension and CKD, high levels of HbA1c and uric acid. The presence of proliferative DR and macular edema was higher in midwest [37]. The presence of any DR was associated with cardiovascular disease (CVD) with an odds ratio of 2.16 (p = 0.01)] [38].

The presence of DR was also observed in 8.5% of the adolescent individuals [39]. Among different types of social determinants it was noteworthy that individuals with only public health care insurance had a higher OR of 1.527, p = 0.002 for the presence of any type of DR compared with individuals with both private health care plus public health care assistance [40].

In the first wave, an assessment of the association between TSH levels and DR was carried out using data obtained from medical records. It was found that TSH levels of 0.4–2.5 mU/l were associated with a lower risk of DR, independently of glycemic control and diabetes duration [41].

Concerning inflammatory serum biomarkers involved in the pathophysiology of DR such as CRP, vascular endothelial growth factor (VEGF), tumor necrosis factor alpha (TNF-α) and interleukin-6 (IL-6), only C-reactve protein was associated with proliferative DR in a nested case–control analysis from our sample [42].

Considering the evaluated genetic markers, also in a nested case–control analysis, African GA was associated with severe non-proliferative or proliferative DR, even after adjustments for socioeconomic status and years of school attendance, both considered important social determinants in our country [43]. In another nested case–control analysis performed with our sample, no influence of genes of the HLA system was observed in association with severe non-proliferative or proliferative DR [44].

Diabetic chronic kidney disease

Considering that CKD is still an important cause of morbidity and mortality in individuals with T1D, its adequate screening is also needed. We have used urinary albumin concentration, determined by immunoturbidimetry, instead of albumin to creatinine ratio. We have adopted the SBD [4] and ADA [5] criteria, to define CKD. According to this criteria, individuals were classified into two groups, as having normal renal function or CKD. Individuals with normal renal function had an estimated glomerular filtration rate by CKD-EPI equation (eGFR) ≥ 60 ml/min and the absence of albuminuria, while CKD was defined as the presence of a GFR < 60 ml/min and/or the presence of albuminuria, ≥ 30 mg/dl in at least two morning urine samples.

Considering the overall sample, it was not possible to define and stratify CKD in 495 individuals (28.1%) due to albuminuria or eGFR missing data [45]. The most common missing data were related to albuminuria with discordant results in two morning urine samples. Absent results were found in 150 individuals (8.5%). The overall prevalence of CKD was 25.2%. The presence of CKD was associated with female gender, socioeconomic status, DR, older age, diabetes duration, time of follow- up in each center, BMI, systolic blood pressure, diastolic blood pressure, heart rate, hypertension, dyslipidemia, use of statins, use of renin-angiotensin system inhibitors, diuretics, high levels of HbA1c and uric acid. A separate subanalysis evaluating the relationship between uric acid and CKD showed an overall decrease of 4.11 ml/min and of 2.04 ml/min in the eGFR for every 1 mg/dl increase in serum uric acid, in the overall sample and in those individuals with normal renal function, respectively, independent of HbA1c levels and diabetes duration [46]. Self-reported color-race was not associated with CKD after adjustments. A tendency was observed between CKD and higher African GA ≥ 50% (OR = 1.956, p = 0.06) even after adjustments suggesting that further studies should be conducted to confirm this finding [47]. However, when studying eGFR, it was noted that in individuals with African GA ≥ 50% (n = 85) the use of ethnic correction factor in the CKD-EPI equation resulted in the redefinition from altered to normal renal function in 13 individuals (56.5%), mostly women with eGFR between 52 and 59.3 ml/min [48]. So, we can conclude that more studies are necessary to clarify the relationship between self-reported color-race, GA and CKD mainly in admixed, multi-ethnic populations such as the Brazilian.

It is important to highlight that CKD was also observed in 14.0% of adolescents [39]. Similar to what was observed for DR, TSH levels between 0.4 and 2.5 mU/l were also associated with a lower risk of CKD independently of glycemic control and diabetes duration [41]. The presence of self-reported periodontitis, found in 4.7% of individuals, was associated with higher risk of DR and CKD with an OR ~ 2.4 p < 0.05 [49].

Autonomic and peripheral neuropathy

The evaluation of autonomic (CAN) and diabetic peripheral neuropathy (DPN) was performed in accordance with SBD [4] and ADA [5] guidelines. Firstly, a study was carried out to assess the reproducibility of CAN assessment tests [50] and the results showed a good reproducibility of the three tests proposed by these guidelines 11%, 18.6% and 14.1%, for the E:I ratio, the Valsalva ratio and the max:min ratio, respectively, and for parameters obtained by spectral analyses, RRNN (6.4%) and SDNN (24.5%). Using these parameters in a pilot study in one of the participating centers, the prevalence of CAN was of 30.5% [51]. In the overall sample, 1,712 individuals (97.3%) were evaluated and the prevalence of CAN was of 23.4%, and was associated with age, smoking, lower socioeconomic status, more frequent yearly medical appointments, insulin therapeutic regimens, higher levels of HbA1c, total cholesterol, uric acid, diastolic blood pressure and heart rate, DR, CKD and a tendency to be associated with severe hypoglycemia [52].

Concerning the evaluation of DPN, among 1,732 evaluated individuals (98.4%), its prevalence was of 14.8% and was found to be associated with age, diabetes duration, increased percentage of African GA, HbA1c levels, type of health care insurance and insulin therapeutic regimen, number of yearly clinical visits, decreased exercise practice, hypertension, dyslipidemia, heart rate, statins use, uric acid levels, lower health-related quality of life, DR and amputations [53].

It is important to highlight that CAN and DPN were observed in 12.5% and 4.9% of the adolescents, respectively [39].

Macrovascular disease and Steno type 1 risk engine (S1TRE)

Data regarding clinical CVD were retrieved from medical records and defined as follows: ischemic heart disease (ICD-10- I20-I25), including angina (ICD-10-I20), ischemic stroke (ICD-10- I64), heart failure (ICD-10- I50), and peripheral artery disease (ICD-10-I73.9). After excluding adolescents (n = 239), 32 individuals (2.1%) were diagnosed as having CVD, mainly ischemic heart disease, n = 26 (81.3%). Peripheral vascular disease was observed in four individuals (12.5%) and heart failure in two (6.3%). Individuals with CVD were older, had higher age at T1D diagnosis, and longer diabetes duration. Most of them were male [(n = 19 (59.4%)], older than 40 years [(n = 24 (75%)] and had a diabetes duration at CVD diagnosis between 15 and < 30 years [n = 25 (78.1%)]. Moreover, they had longer follow-ups in their respective diabetes centers and presented larger waist circumferences, higher BMI, systolic blood pressure, fatty liver index (FLI), a serum biomarker and lower eGFR. Hypertension and DR were more commonly found in these individuals, as well as referable DR. Regarding medical treatment, these individuals were more frequently receiving statins, antihypertensive drugs, and metformin. No difference was found in HbA1C levels [54].

For Steno type 1 risk engine (S1TRE) calculation, adolescents (n = 239) and individuals with missing data for ST1RE calculation (n = 221; 12.5%), were excluded, with a final sample of 1,300 eligible individuals. The overall median, [inter quartile range] of 5-year and 10-year risk for CVD was 2.7% [4.41] and 5.44% [8.47], respectively. A 10-year moderate and high risk was noted in 234 (18.0%) and 147 (11.3%) individuals, respectively. Higher age at T1D diagnosis, presence of DR, and higher levels of hs-CRP and uric acid were predictors of 10-year moderate/high risk. Considering the ADA 2014 standards of medical care in diabetes [5] which was used in BrazDiab1SG, a significant number of individuals in the moderate/high cardiovascular risk group with blood pressure (~ 38%) or LDL-cholesterol (~ 58%) out of targets were not receiving antihypertensive medications or statins, respectively. This highlights the sub-optimal treatment of CVRF, including the management of glycemic control in these latter individuals. It is important to point out that a larger number of individuals in the high 10-year risk, 20.1% had a FLI ≥ 60 which is considered currently as possible indicator of NAFLD and/or metabolic dysfunction-associated steatotic liver disease (MASLD).

Non-alcoholic fatty liver disease (NAFLD) and/or metabolic dysfunction-associated steatotic liver disease (MASLD)

At the time of conceptualization of BrazDiab1SG the used nomenclature for this condition was NAFLD. In 2020, the nomenclature has changed to MASLD, with both definitions being quite similar but with subtle differences. NAFDL encompasses a wide range of alterations that include steatosis, steatohepatitis, fibrosis and cirrhosis in individuals with no alcohol consumption or a daily alcohol consumption of ≤ 20 g for females and ≤ 30 g for males in the absence of another cause of chronic liver disease. Meanwhile, MASLD encompasses the identification of hepatic steatosis (as assessed using blood-based biomarkers, imaging methods or liver biopsy) in combination with at least one common CVRF, including excess adiposity, the presence of prediabetes or diabetes, increased blood pressure or atherogenic dyslipidemia (high triglycerides or low HDL cholesterol). If we consider that the diagnosis of diabetes is already configured as a CVRF, all individuals included in the BrazDiab1SG would have MASLD. However, it is important to assess that in the BrazDiab1SG sample some individuals did not have metabolic syndrome and other important CVRF [55]. According to Ponticasa et al. [56], NAFDL/MASLD and MASLD “need further validation to facilitate global acceptance and streamlined transition to the new nomenclature”.

The assessment of NAFDL was initially approached within the context of metabolic syndrome and its risk factors. The overall prevalence of metabolic syndrome was 27.1%, being 32% and 8.4% in adults and adolescents, respectively, and was associated with female gender and the presence of acanthosis nigricans. In adults, metabolic syndrome was associated with higher levels of serum transaminases (aspartate aminotransferase-AST)/alanine aminotransferase-ALT); GGT, gamma glutamyl transferase; uric acid and CRP. ALT above the normal range (> 25U/L for women and > 33U/L for men) was associated with triglycerides levels > 129,5 mg/dl pointing out that these individuals would possibly require imaging investigation such as ultrasound. Self-reported color-race was not associated with metabolic syndrome but higher levels of European GA showed a tendency with an OR of 1.77 (p = 0.05) [57]. A nested case–control analysis with the overall sample did not show an association between inflammatory biomarkers (IL6/TNFα /interleukin 17 (IL17)) with metabolic syndrome [58]. In this study, higher levels of FLI were associated with metabolic syndrome and with DR [58]. One participating center performed NAFDL investigation by two imaging methods, ultrasound and transient elastography [59]. The prevalence of NAFLD was of 12% and 25% when ultrasound and transient elastography were used, respectively. Fibrosis was present in 8% of participants. A total of 36.8% of individuals had at least one hepatic imaging altered exam that was associated with metabolic syndrome which main risk factor was higher triglycerides levels, even if within the normal range.

Autoimmune diseases

Although not considered as a diabetes-related chronic complication, autoimmune diseases (AIDs) such as, autoimmune thyroid diseases (AITD) (hyper and hypothyroidism), celiac disease, skin diseases, and many others present a higher prevalence in individuals with T1D. These diseases can occur sporadically or in combination, either with an insidious presentation or as an important hormonal decompensation. In the BrazDiab1SG, AIDs were found in 344 individuals (19.5%). The majority of them presented autoimmune hypothyroidism, n = 258 (14.7%), followed by hyperthyroidism, n = 26 (1.5%) [60]. Overall, 18 individuals (1.0%) presented a combination of AIDs being the most frequent the combination of AITD with vitiligo (n = 6 p individuals and with rheumatoid arthritis (n = 6 individuals). AIDs were associated with female gender, older age, longer diabetes duration, self-reported color-race (White, OR = 4.188;p < 0.001 and Brown, OR = 3.018; p = 0.003), geographic region (north/northeast region, OR = 0.494;p = 0.01) and higher anti-TPO levels (≥ 35 UI/ml).

A higher prevalence of DR was observed in individuals with exclusively AITD and with other AIDs independent of the presence of AITDs in comparison to individuals without any AIDs, [ 118 (42.1%) vs 469 (34.1%), p = 0.01] and [30 (49.2%) vs 469 (34.1%), respectively, p = 0.02]. A higher prevalence of CKD was observed in individuals with exclusively AITDs in comparison to individuals without any AIDs.

Acute complications

Considering the overall sample, 1,319 individuals (74.9%) had at least one episode of hypoglycemia in the previous month. Comparing individuals with hypoglycemia with those without, the main risk factors for hypoglycemia were lower HbA1c levels (8.79 ± 1.92% vs. 9.65 ± 2.55%, p < 0.01, respectively); better adherence to SMBG (96.4% vs 88.9%, p < 0.001, respectively); and longer time of school attendance (12.5 ± 3.8 vs. 11.4 ± 9.2 years, p < 0.01, respectively). Severe hypoglycemia was reported by 251 (19%) subjects that, in comparison to those without, presented higher alcohol consumption; p < 0.01, more hospitalizations in the previous year y p < 0.001, and higher frequency of diabetes-related chronic complications such as DR, CKD, CAN and ketoacidosis [61]. The data related to hypoglycemia of BrazDiab1SG were included in Fig. 4. The Brazilian Hypoglycemia Assessment Tool study, has found a 25.7% rate of reported severe hypoglycemia [62].

A study that compared 24 individuals treated with autologous non-myeloablative hematopoietic stem-cell transplantation (AHST) matched with 144 individuals from the BrazDiab1SG, treated with conventional therapy, showed that individuals with an insulin-dose adjusted A1C (IDAA1c) ≤ 9 in comparison to individuals with an IDAA1c > 9, in both groups, had less diabetes-related chronic complications (DR and CKD) and fewer episodes of hypoglycemia, in the month prior to the enrollment [63].

Genetic data

The study of T1D genetic markers in Brazil must be done considering the complex contest in which our colonization occurred. The Brazilian population is highly diverse and heterogeneously admixed, primarily composed by three ancestral groups: Native Americans, Europeans, and Sub-Saharan Africans. Over centuries, these groups have intermixed through asymmetric mating patterns, notably between European men and Native American or African women [64].

Given this diverse ancestry, it is crucial to explore the association between T1D genetic markers, particularly genes of the HLA system, and the Brazilian population’s autosomal ancestry. Additionally, the study should consider the influence of both matrilineal (mitochondrial DNA, mtDNA) and patrilineal (Y chromosome, CRY) lineages. These genetic markers may provide insight into the origins and distribution of T1D in the Brazilian population, which is shaped by this mixed heritage.

To address this knowledge gap, the BrazDiab1SG study conducted an innovative investigation into the relationship between various genetic factors. For this purpose, a sample of individuals and controls from Maranhão, a northeastern state of Brazil, was added to the original sample and evaluated with the aim of providing valuable insights into the origins of T1D in Brazil.

Genomic ancestry

So far, most genetic studies in T1D have been conducted in homogeneous populations, i.e., Caucasians with European ancestry, currently considered as White, with few studies describing highly admixed populations like the Brazilian. These data are described in Fig. 5. Since 1991, self-reported color-race has been used for Brazilian population censuses, being divided into five color-race groups: White (branca), Black (preta), Brown (parda), Asian (amarela) and Indigenous (indígena). This classification has also been used in the majority of Brazilian epidemiological studies. However, the risk of misclassification is high, being then minimized by the use of methodologies that allow the inference of GA, like the Ancestry Informative Markers (AIMs). The study that evaluated GA in T1D in comparison to a control population showed that the former had a significantly higher median [IQR] individual European GA in comparison to controls 67.8 [31.2] vs. 56.3 [25.7]%, respectively [65]. As for self-reported color-race among T1D individuals, 923 (54.3%) participants reported to be White, 610 (35.9%) Brown, 132 (7.8%) Black, 18 (1.1%) Asian and 15 (0.9%) Indigenous. Although, European GA prevailed in those individuals who self-reported as being White (74.6%) and Brown (61.1%), it constituted up to 39.1% in those who self-reported as being Black. As expected, a great diversity was observed across the country but in all regions T1D individuals had a higher proportion of European GA than controls. Only individuals with T1D (N = 110, 6.3%) showed a European GA greater than 95%. No individual with T1D or controls had African or Native Amerindian GA greater than 90%. Native Amerindian GA was the lowest found across the country, except in North region. It is noteworthy that an increased African GA was associated with proliferative DR [43], metabolic syndrome [58]) and a tendency to be associated with DKD [47], without an association with glycemic control in young individuals [66]. The increase in African GA was associated with an increase in the transcription of genes that are related to innate immune response, cytokine production and bactericidal activity [67]. Currently, a close link between hyperglycemia, inflammation and diabetes-related complications is a key point for understanding its pathogenesis.

Fig. 5.

Percentage of European, African and Amerindian GA* according to Self-reported color-race Data from Brazilian Diabetes Type 1 Study group (l BraZDiab1SG). *GA genomic ancestry

Histocompatibility leukocyte antigen system (HLA)

The HLA system genes, located on chromosome 6p21.3, play a key role in the human immune response and are responsible for almost 50% of the risk for T1D, which is currently considered an autoimmune disease. Although HLA Class I genes and non-HLA genes also contribute to T1D risk, Class II alleles, such as DR and DQ, show the strongest associations with the disease [68]. Findings from the BrazDiab1SG study revealed that the HLA-DRB1* alleles associated with an increased risk of T1D were DR3 (OR = 4.03), DR4 (OR = 2.98), and DR9 (OR = 2.43). The most frequent protective HLA-DRB1 alleles were DR13 (OR = 0.56), DR7 (OR = 0.49), and DR11 (OR = 0.26). The two HLA-DQB1 alleles associated with T1D risk were DQB1*02 (OR = 2.93) and DQB1*03 (OR = 1.38) [68].

In Fig. 6, the most frequently found risk haplotypes for T1D can be seen, and were DRB1*03:01 ~ DQA1*05:01 g ~ DQB1*02:01 (OR = 5.8), DRB1*04:05 ~ DQA1*03:01 g ~ DQB1*03:02 (OR = 5.34), DRB1*04:02 ~ DQA1*03:01 g ~ DQB1*03:02 (OR = 3.43). The haplotype DRB1*03:01 ~ DQA1*05:01 ~ DQB102:01 was the most frequent risk haplotype across all self-reported racial groups, with the highest frequency observed in individuals who self-identified as White. In contrast, the haplotype HLA-DRB1*09:01 ~ DQA1*03:01 g ~ DQB1*02:02 had the highest frequency in individuals who self-identified as Black. Additionally, an increase in European ancestry percentage was associated with the risk alleles DRB1*03:01, DRB1*04:01, DRB1*04:02, DQA1*05:01, DQB1*02:01, and DQB1*03:02, while an increase in African ancestry percentage was mainly associated with the protective alleles DRB1*03:02, , DRB1*11:02, and DRB1*15:03 [69].

Fig. 6.

Percentage of Histocompatibility haplotypes (risk and protection) according to self-reported color-race. Data from Brazilian Diabetes Type 1 Study Group l (BrazDiab1SG). DR3 = DRB1*03:01 ~ DQA1*05:01 g ~ DQB1*02:01. DR4 = DRB1*04:05 ~ DQA1*03:01 g ~ DQB1*03:02. DR9 = DRB1*09:01 ~ DQA1*03:01 g ~ DQB1*02:02. DR7 = DRB2*07:01 ~ DQA1*02:01 ~ DQB1*02:02. DR13 = DRB1*13:01 ~ DQA1*01:03 ~ DQB1*06:03. T1D Type 1 diabetes

When analyzing the age of T1D onset, the combination of DRB1*03/ DRB1*04:05/04:02 was higher in individuals with earlier age at T1D onset, with a mean age of 6.8 years [68]. In contrast, the protective allele, mostly DQB1*06:02 g in haplotypes with the alleles DRB1*15:01 g or DRB1*15:03 g were more frequently found in individuals with disease onset ≥ 19 years [70]. These data are shown in Fig. 7.

Fig. 7.

Percentage of alleles of the Histocompatibility Antigens according to age at diagnosis of diabetes. Data from Brazilian Diabetes Type 1 Study Group l (BrazDiab1SG)

A sub-analysis was performed including data on family history, specifically relatives' birthplace and self-reported race/skin color [71]. These data were carefully obtained through interviews with participants or family members, covering three generations. The majority of individuals reported having a White relative (95.17%), and the most frequently found allele among them was DRB1*03:01. Increased odds (OR = 1.53) of presenting this allele was found only in individuals who reported having all White relatives.

In conclusion, data regarding risk alleles of the HLA system in individuals enrolled in the BRAZDiab1SG suggest that most of them are likely to belong to European, Caucasian ethnicity. However, it is important to emphasize that this remains a hypothesis that requires further investigation.

Mitochondrial DNA (mtDNA)

Mitochondria has many roles, including adaptive mechanisms to cope with environmental changes in cell physiology, growth characteristics, and inflammatory systems. All the above-mentioned actions have a great impact on a broad range of metabolic and degenerative diseases, such as cancer, diabetes, and aging which are related to the polymorphisms and mutations that are present in mtDNA (mitochondrial DNA). The mtDNA through the hypervariable regions (HVS), could be a good marker for inferring the possible maternal geographic origin. A pilot study with BRAZDiab1SG data showed the following percentage of haplogroups: Native Americans 43.6%; African 38.2% and European 18.1%. The most frequent haplogroup in the studied population was of Native Americans C (16.7%), followed by African origin L3 (16.2%) [72]. According to self-reported color-race, those who self-declared as White had predominantly Native American, followed by European and African matrilineal ancestries of 50.6%, 27.6% and 21.8%, respectively. Those who self-declared as being Brown had 12.5% of European matrilineal ancestry and a similar frequency of African and Native American matrilineal origins of 45.2% and 42.3%, respectively. Those who self-declared as Blacks had 90.9% of their matrilineal ancestry of African origin and 9.1% of Native American matrilineal ancestry. The ANOVA test was used to validate our hypothesis, finding a p-value of p = 0.0513 that revealed only a tendency.

Cromossome Y (CRY)

Chromosome Y has a non-recombinant portion that is transmitted practically intact through the paternal lines, allowing the determination of their ancestry. A pilot study of BRAZDiab1SG conducted in the State of Maranhão demonstrated that European patrilineal ancestry was predominant, in approximately 85%, followed by African ancestry of approximately 9%, with no difference between individuals with diabetes and controls [73]. Furthermore, in the Y chromosome haplogroups analysis, we observed that R1b was the most frequently found in both groups, followed by E1b1b in TD1 group and by E1b1a in the control group. There were significant differences in the distribution of the DRB1*/DRB1* genotype between European, African and Native American patrilineal ancestries. Individuals with only European patrilineal ancestry showed a higher frequency of DRB1*03 and DRB1*04homozygote. The frequency of DRB1*03 and DRB1*04 homozygote genotypes and DRB1*03/DRB1*04 heterozygote genotype amounted to 44.9% in European patrilineal ancestry in individuals with T1D.

Overall, BRAZDiab1SG data concerning the evaluation of mtDNA and CRY endorse the influence of Brazil’s historical colonization period in the formation of the genetic background of the Brazilian population with T1D.

Other genetic markers

A study was carried out, aiming to investigate the genetic markers that are associated with the autoimmune process associated with the pathogenesis of T1D. This study also aimed to determine the relationship between the age of T1D onset and two single nucleotide polymorphisms (SNPs) in the interleukin-18 (IL18) gene (rs187238, g.-137G > C;rs1946518, g.-607C > A) and one SNP of the IL12B gene (rs3212227 g._159A > C,3′UTR). Both cytokines have proinflammatory properties modulating T Helper1 lymphocytes. It was found a relationship between African GA, IL18 gene and earlier age at T1D onset [74].

Conclusion

The BrazDiab was the first and largest study that aimed to evaluate the real-world care of Brazilian individuals with T1D. It was conducted in two waves, in all Brazilian geographic regions. In this study, several aspects of T1D were focused, such as its diagnosis and management, clinical factors such as glycemic control, treatment, and comorbidities associated with the disease.

This study also sought to map the profile of individuals with T1D, investigating characteristics such as age, gender, GA, lifestyle, and health conditions that can affect the evolution of the disease. In addition, data were collected on individuals’ adherence to recommended treatments and the impact of T1D on the Brazilian health system.

The results of BrazDiab1SG have been used by SBD to show the reality of individuals with T1D to the Brazilian Ministry of Health and its several Health Secretaries and Departments, aiming to update treatment options for T1D, improve glycemic control, quality of life, treatment adherence and the frequency of hypoglycemia, diabetes-related complications and untreated comorbidities. Another important objective of the SBD was to reduce the discrepancies found in the treatment and management of T1D among different geographical regions and social classes in Brazil, that were found by the BraZDiab1SG. Finally, the data generated by this study provided support for the incorporation of insulin analogs for T1D individuals and for the evolution of the Clinical and Therapeutic Protocol for Type 1 Diabetes Mellitus of the Ministry of Health.

Abbreviations

T1D

Type 1 diabetes

SBD

Brazilian Diabetes Society

BrazDiab1SG

Brazilian Type 1 Diabetes Study Group

SMBG

Self-monitoring of blood glucose

FPG

Fasting plasma glucose

DR

Diabetic retinopathy

HbA₁C

Glycated hemoglobin

TSH

Thyroid stimulating hormone

FT4

Free levothyroxine

Anti-TPO

Thyroid peroxidase antibodies

HPLC

High-performance liquid chromatography

CRP

C-reactive protein

AD

American Diabetes Association

CVRF

Cardiovascular risk factors

BMI

Body mass index

GA

Genomic ancestry

VEGF

Vascular endothelial growth factor

TNF-α

Tumor necrosis factor alpha

IL-6

Interleukin-6

ICD

International classification of diseases

GGT

Gamma glutamyl transferase

AST

Aspartate aminotransferase

ALT

Alanine aminotransferase

IL-17

Interleukin-17

AHST

Autologous non-myeloablative hematopoietic stem-cell transplantation

IDAA1c

Insulin-dose adjusted A1C

Author contributions

MBG and CAN: Conceptualization, writing and manuscript review. BRAZDiab1SG had collected the data All the authors have approved the final version of this manuscript.

Funding

This work was supported by grants described in additional Table 2.

Availability of data and materials

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Each local ethics committee has approved the study. Written informed consent was obtained from all participants and/or from their parents where necessary.

Competing interests

The authors declare no competing interests.