Maturity onset diabetes of the young due to HNF1A variants in Croatia

Introduction Maturity onset diabetes of the young due to HNF1A mutations (HNF1A-MODY) is the most frequent form of monogenic diabetes in adults. It is often misdiagnosed as type 1 or type 2 diabetes, but establishing genetic diagnosis is important, as treatment differs from the common types of diabetes. HNF1A-MODY has not been investigated in Croatia before due to limited access to genetic testing. In this study we aimed to describe the characteristics of young adults diagnosed with diabetes before the age of 45 years, who have rare HNF1A allele variants, and estimate the prevalence of HNF1A-MODY in Croatia. Materials and methods We recruited 477 C-peptide positive and beta cell antibody negative subjects through the Croatian Diabetes Registry. HNF1A was sequenced for all participants and systematic assessment of the variants found was performed. The prevalence of HNF1A-MODY was calculated in the study group and results extrapolated to estimate the proportion of diabetic individuals with HNF1A-MODY in Croatia and the population prevalence. Results Our study identified 13 individuals harbouring rare HNF1A allelic variants. After systematic assessment, 8 were assigned a diagnosis of HNF1A-MODY. Two individuals were able to discontinue insulin treatment following the diagnosis. We estimated that HNF1A-MODY in Croatia has a prevalence of 66 (95% CI 61 - 72) cases per million. Conclusions The estimated prevalence of HNF1A-MODY in Croatia is similar to that reported in other European countries. Finding cases lead to important treatment changes for patients. This strongly supports the introduction of diagnostic genetic testing for monogenic diabetes in Croatia.


Introduction
Maturity-onset diabetes of the young (MODY) is a clinically heterogeneous group of disorders, caused by mutations in one of 13 different genes (1). One of the most frequent forms of MODY is as-sociated with variants in the HNF1A (hepatocyte nuclear factor 1-alpha) gene on chromosome 12, also known as HNF1A-MODY. HNF1A is expressed in beta cells, where it encodes a crucial member of the auto-regulatory transcriptional network during embryonic development of pancreas (2). In adult life, protein HNF1A regulates the expression of numerous beta cell genes, such as insulin, pyruvate kinase and the glucose transporter molecule Glut2 (3).
Subjects harbouring disease-causing HNF1A allelic variants typically present with prominent family history of diabetes (autosomal dominant mode of inheritance), early onset of diabetes (classically before the age of 25 years) and insulin independence evidenced by presence of C-peptide, often despite a long duration of diabetes (4). The majority of subjects with MODY do not show signs of insulin resistance and their body mass index (BMI) is in the normal range. The disease is characterized by progressive deterioration of beta cell function, due to an inability to increase insulin secretion in response to hyperglycaemia (5). Disease-causing variant alleles in HNF1A demonstrate a high penetrance, so 63% of variant carriers develop diabetes before 25 years of age, and 96% before the age of 55 years (6).
MODY is frequently misdiagnosed as type 1 or type 2 diabetes due to low awareness among primary care physicians, lack of widely accepted diagnostic protocols and/or unavailability of genetic testing. Definitive diagnosis of HNF1A-MODY is established by sequencing of HNF1A, although confirmation that allelic variants are disease-causing remains a challenge. A correct molecular diagnosis is of utmost importance, since subjects with HN-F1A-MODY are usually very sensitive to oral treatment with sulphonylurea derivatives (SU) such as gliclazide, which can provide excellent diabetes control for decades (7). Glinides have similar mechanism of action to SU, therefore are an alternative treatment for HNF1A-MODY (8). Establishing a genetic diagnosis of HNF1A-MODY may lead to treatment changes for previously misdiagnosed patients, even to the discontinuation of assumed lifelong insulin treatment. Correctly identifying MODY subjects also allows the recognition of affected family members, thereby enabling better disease management and clinical care for these individuals too.
The current diagnostic pathways for MODY, which mostly include clinical criteria, miss many cases through low sensitivity (9,10). Maturity-onset diabetes of the young is rarely diagnosed in Croatia because genetic testing is not available to most clinicians. Therefore, the prevalence of MODY in Croatia has never been studied. The goal of the study was to describe the clinical characteristics of Croatian subjects with HNF1A-MODY, estimate the prevalence of HNF1A-MODY within Croatian subjects diagnosed with diabetes as young adults, increase the awareness of this specific type of diabetes and thus provide support for introduction of genetic testing for monogenic diabetes in Croatia.

Subjects
Initial recruitment was performed using data from the Croatian national registry of patients with diabetes (CroDiab); a registry established to improve health care, to determine the prevalence and incidence of diabetes and its complications, as well as to monitor morbidity and mortality at the national level (11). All primary and secondary care physicians who provide diabetes care were obliged to supply data on their patients to the Vuk Vrhovac University Clinic for Diabetes, Endocrinology and Metabolic Diseases on an annual basis until 2015 (12). Data supplied constitutes a basic informatic sheet, recognised by the international diabetology community as optimal data collection form for the follow-up and improvement of diabetes care.
All adult subjects with diabetes onset under age of 45 years and currently older than 18 years were eligible to take part in the study. Other inclusion criteria encompassed the following: evidence of endogenous insulin secretion (fasting or random Cpeptide ≥ 0.2 nmol/L) and negative glutamic acid decarboxylase antibodies (GADA) and islet cell autoantibodies (ICA). Letters of invitation, describing the protocol and aims of the study, were sent to the home address of eligible subjects. Subjects newly diagnosed with diabetes who met inclusion criteria were also invited to participate. Study participants were recruited at Vuk Vrhovac University Clinic for Diabetes, Endocrinology and Metabolic Diseases, Merkur University Hospital, Zagreb; the Reference Centre for Diabetes of the Croatian Ministry of Health, in the period between June 2013 and June 2015. Whole blood and plasma samples were collected from every patient. Blood was collected into EDTA-treated tubes and plasma was separated by centrifugation. All samples were stored at -20 °C until further analysis.
Data collection included the following information: age, sex, age of diabetes onset, treatment history, current medication, family history of diabetes, anthropometrical data (weight, height, body mass index (BMI), blood pressure) and biochemical data (fasting glucose, glycated haemoglobin (HbA 1c ), C-peptide, lipid profile, creatinine). All recruited participants gave written informed consent. The study was approved by Ethics Committee of Merkur University Hospital and Faculty of Pharmacy and Biochemistry, University of Zagreb, Croatia. The study was conducted in accordance with the Declaration of Helsinki. Figure 1 illustrates the study recruitment strategy. Study participants were mainly recruited through CroDiab registry. According to the study inclusion criteria there were 13,035 adult subjects with dia- Figure 1. Flowchart

Biochemical and immunological assays
Plasma glucose and serum lipid profile were assayed using routine enzymatic methods (AU680, Beckman Coulter, Brea, USA) in the accredited (ISO15 189:2012) laboratory. HbA 1c was measured by turbidimetric immuno-inhibition traceable to the IFCC-reference system and reported in DCCTaligned units (AU680, Beckman Coulter, Brea, USA), with a total imprecision (CV) of 1.8%. C-peptide was analysed using a chemiluminescence immunoassay on the Advia Centaur XP analyser (Siemens Healthcare Diagnostics, Eschborn, Germany) with a CV of 5.6%. Islet cell antibodies were measured by indirect immunofluorescence. Rate of positivity was calculated by determining end-point titres of samples that were converted to the units of Juvenile Diabetes Foundation (JDF-U) by comparison with a standard curve of log 2 JDF units with log 2 of end-point titre of standard sera. The threshold of detection was > 5 JDF units. Glutamic acid decarboxylase antibodies were measured by enzyme-linked immunosorbent assay (ELISA) (Euroimmun AG, Lübeck, Germany). The cut-off for the positive result was set to 5 WHO Units/mL. Analysis were performed in a laboratory participating in the international harmonisation programme (DASP; Diabetes Antibody Standardization Program) (13).

HNF1A sequencing and analysis of the HNF1A variants
Deoxyribonucleic acid (DNA) was extracted from the whole blood sample using QIAmp DNA Mini Blood kit (Qiagen, Hilden, Germany), according to the manufacturer's protocol (14).
Quantity and quality of the extracted DNA were evaluated using Nanodrop 8000 UV-Vis spectrophotometer (Thermo Scientific, Waltham, USA). Both were satisfactory, with an average DNA concentration of 38.69 (6.49 -107.60) ng/µL and average A260/A280 ratio of 1

Statistical analysis
Non-parametric summary measures were used to describe the groups (subjects with a rare HNF1A allelic variant (N = 13) and without the HNF1A variants (N = 464)). Medians with ranges (min -max) were calculated for non-parametric continuous variables (current age, age at diagnosis, duration of diabetes, BMI, HbA 1c and C-peptide, lipid profile) and percentages for categorical measures (gender and current treatment). Non-parametric statistics (Mann-Whitney test) were used for comparison of continuous variables among groups. Differences of frequencies for categorical variables were tested using the Chi-squared (treatment) and Fisher's test (gender). Significant differences were assumed for P < 0.05. The prevalence of HNF1A-MO-DY among subjects with diabetes was calculated as the proportion of individuals with HNF1A-MO-DY out of the total number of participants and extrapolated to calculate estimated population prevalence using the latest available data (year 2015) from the Croatian Institute of Public Health. Statistical analysis was performed using SPSS v23.0 (IBM, Armonk, USA).

Characteristics of subjects
Basic anthropometric and clinical data of the recruited subjects are given in Table 1. None of the participants were known to have monogenic diabetes before the study. Subjects with diabetes and a rare HNF1A allelic variant were predominantly females (P = 0.02). Their median (min -max) age (including both probands and affected relatives) was 48 (18 -72) years, their median age at diagnosis of diabetes was 35 (16 -50) years and their median diabetes duration was 10 (2 -31) years. We found no significant differences between diabetic subjects with or without a rare HNF1A allelic variant regarding their current age, age at onset and duration of diabetes. The subjects with a rare HNF1A variant had lower median BMI compared to subjects without a rare HNF1A variant (P = 0.003). Lipid profile also showed differences between the groups. Median HDL-cholesterol was higher (1.4 vs.   Continuous variables are given as median and interquartile range, age is given as median (min -max), while categorical variables are given as percentages. *The number excludes one recruited relative with a rare HNF1A allelic variant, but without diabetes. † P-values were calculated using Mann-Whitney U test (significance level α = 0.05); for categorical variables Chi-square (treatment) and Fisher's (gender) tests were used. NP -statistical analysis was not performed due to too small sample size. BMI -body mass index. HbA 1cglycated haemoglobin. We observed no significant difference in HbA 1c values between the groups, but subjects with diabetes and an HNF1A variant had lower C-peptide values (P = 0.006).
Among subjects with an HNF1A variant, 12 had at least a two-generational family history of diabetes, whilst detailed family histories were unavailable for one proband due to loss of contact with relatives.
The PTVs and the protein variant p.P379R were previously reported in subjects with HNF1A-MO-DY, while the protein variant p.A251T was identified previously in a patient with SU-sensitive diabetes (16,17). Protein variants p.G288W and p. T515M have not been previously reported in MODY families. However, variants with a different amino acid change at the same position (namely, p.G288C and p.T515K) were previously reported as causing MODY (16). Two HNF1A protein variants (p.S3C, p.G151S) were novel.
All three bioinformatics tools (SIFT, PPH2 and Provean) predicted both novel variants and p.  Two out of nine identified allelic variants were present in the individuals from ExAC (c.862G>T, p.G288W and c.1544C>T, p.T515M) and none of the allelic variants were present in the non-diabetic individuals from T2D-GENES project.
Family data was available for 3 families: those with protein variants p.A251T, p.P379R and p.G288W (Figure 3). The p.A251T variant co-segregated with a milder form of dysglycaemia not typical for MODY. The diabetic mother of the proband with p.P379R has the same variant, while the other nondiabetic family members do not carry the variant. The mother of the proband with p.G288W carries the same variant, but does not have diabetes at age of 66 years, despite being obese. Unfortunately, it was not possible to perform co-segregation studies of the novel variants with diabetes due to the unavailability of other family members.  Based on the evidence discussed above, we showed that assigning disease-causality of the identified rare variant is complex and it may not be possible to provide a definitive answer. Taking all above into account, we concluded that six probands have disease-causing or likely diseasecausing protein variants (p.S3C, p.G151S, p.R229X, p.G292fs, p.P379fs and p.P379R). The protein variants p.G288W and p.T515M have weak evidence of causing a MODY phenotype. Protein variant p. A251T remains of unknown significance, and may be associated with reduced penetrance of dysglycaemia.

The prevalence of HNF1A-MODY in Croatia
In this study 13 probands and family members out of 477 individuals with C-peptide positive and beta cell antibody negative diabetes, diagnosed before 45 years of age, had a rare HNF1A allelic variant.

Discussion
This is the first study to identify individuals with monogenic diabetes in Croatia. We identified 13 subjects with 9 different rare HNF1A allelic variants in 477 individuals with C-peptide positive and beta cell antibody negative diabetes, diagnosed before 45 years of age. Eight of them have HNF1A-MODY. On this basis, we have calculated that 1.7% (95% CI 0.54 -2.86%) of subjects in our study group have HNF1A-MODY, which corresponds to a prevalence of 66 (95% CI 61 -72) cases per million of population.
In the literature, estimations of the proportion of all MODY cases (including other causative genes) vary between 0.14% and 4.2% of all reported diabetes cases (9,(18)(19)(20)(21)(22)(23). In 1981, a German study reported a prevalence of clinical MODY of 0.14% of total 40,927 diabetes cases, which corresponded to 70 cases per million of the population (19). Within the Norwegian HUNT2 study of 1972 subjects with diabetes, a minimum prevalence of HNF1A-MODY of 0.4% was calculated, corresponding to 63 cases per million (21). UK studies reported a minimum MODY prevalence of 108 per million using data from the diagnostic sequencing laboratory (half of the cases had HNF1A-MODY) and 84 cases per million from a population-based approach (only HNF1A-MODY) (9,22). These results from other European countries, calculated with various clinical or sequencing approaches, are comparable to our estimations.
The advantage of our study versus most of the so far published is the fact that all C-peptide positive and beta cell antibody negative participants diagnosed up to age of 45 years had HNF1A sequencing, so fewer cases were likely to be missed by narrowing of selection criteria for genetic testing. The "classic" clinical criteria for MODY diagnosis include onset of diabetes before age of 25 years, insulin-independence and at least two-generational family history of diabetes (24)(25)(26). Using this generally accepted cut off for diabetes onset we would miss more than three quarters of the HNF1A-MO-DY subjects identified in our study.
Despite having no difference in age of onset from the young adults without an HNF1A allelic variant, those with rare HNF1A allelic variants otherwise had typical features of MODY: they were leaner, with lower C-peptide in keeping with beta cell defect and had less dyslipidaemia.
A limitation of our study is the uncertainty that we selected a representative sample of all young adult diabetes onset individuals in Croatia. Although individuals throughout Croatia were invited, they needed to travel to Zagreb to be recruited, so the participation of non-Zagreb residents was limited. A further limitation of this study is that we did not use a gene-dosage method such as multiplex ligation-dependent probe amplification (MLPA) to detect larger deletions which are missed in Sanger sequencing. However, this is relatively uncommon cause (estimated 3%) of HNF1A-MODY (27). Another limitation is difficulty in absolutely assigning disease causality to an identified allelic variant. This is a challenge in many genetic disorders. For instance, allelic variants that were previously perceived as disease-causing have now been identified in populations comprising mostly healthy individuals, which leads us to question their pathogenicity. Careful assessment of identified variants will inform this process. Formal functional assessment of the variants we identified would provide further information about the likely effects, but may not completely define the wide range of phenotypes, which can be associated with the same variant (28). We also had difficulty in recruiting family members of the probands to establish co-segregation of the variant with diabetes for all the HNF1A variants found. This is likely to reflect the "real world" experience of clinicians, attempting to assess variants found through genetic testing, and reinforces the importance of diagnostic laboratories in collating data on variants of uncertain significance.
A final limitation is that we have investigated the prevalence of HNF1A-MODY only, as the most common form of monogenic diabetes, so the prevalence of other MODY subtypes, such as GCK-, HN-F4A-or HNF1B-MODY in Croatia remains unknown.

Pavić T. et al. HNF1A-MODY in Croatia
The exact prevalence of MODY can be determined only by population-based sequencing data, however, this is unlikely to be available in the near future.
In conclusion, in this study we calculated a minimum prevalence of 66 (95% CI 61 -72) HNF1A-MO-DY cases per million of population in Croatia, which is consistent with the estimated prevalence of this form of monogenic diabetes in other European countries. Currently, the vast majority of HN-F1A-MODY cases in Croatia remain unidentified, therefore the affected individuals are not receiving optimal management. Further advantage of establishing a molecular diagnosis is cascade screening of relatives, with potential change in the treatment of those with diabetes and giving a prognostic advice to currently non-diabetic carriers. In this study two individuals were able to discontinue insulin for oral hypoglycaemia treatment, demonstrating the potential benefits of making a molecular diagnosis. We believe that our study strongly supports the introduction of genetic testing for MODY in Croatia. Further work including health economic evaluation would be required to establish an appropriate pathway.