Systematic Genetic Study of Youth With Diabetes in a Single Country Reveals the Prevalence of Diabetes Subtypes, Novel Candidate Genes, and Response to Precision Therapy

Monogenic forms of diabetes are estimated to account for 1–2% of all diabetes cases, with a higher prevalence in the pediatric population (1–6). Monogenic diabetes (MD) includes all diabetes arising from a single gene defect, such as neonatal diabetes mellitus (NDM), defined as onset of diabetes within the first 6 months of life; maturity-onset diabetes of the young (MODY); and syndromic diabetes (7). Because of the complexity of the clinical phenotypes, a genetic test is mandatory to formally diagnose MD. Thanks to next-generation sequencing techniques, genetic testing has advanced considerably even as costs have dropped, yet most cases of MD remain undiagnosed or misdiagnosed worldwide (4). Recent data from a pediatric multinational registry show that 38% of HNF4A and HNF1B diabetes are misclassified as type 1 or 2 diabetes (8). Among young people in the U.K., 50% of MD cases may be misdiagnosed as type 1 diabetes. The diagnosis of MD is crucial because treatment tailored to the underlying gene defect optimizes metabolic control and quality of life.

The main aim of this study was to assess the prevalence and clinical manifestations of MD in youth with diabetes in Lithuania, a genetically quite homogenous population, and to search for additional relevant gene variants. For this purpose, we systematically screened a registry covering 100% of the pediatric and 70% of the young-adult population with diabetes, using high-throughput sequencing of 307 genes associated with diabetes, including MODY, NDM, and syndromic diabetes. We further prospectively assessed the predictive factors for successful treatment change from insulin to sulfonylurea according to the diabetes subtype.

This study included participants from a joint Lithuanian-Swiss project, consisting of the whole Lithuanian pediatric diabetes population (ages 0–18 years) and 70% of those ages >18–25 years with diabetes (the coverage was only 70% because many had left the country and some refused to participate). At study entry, all participants were screened for autoimmune markers for type 1 diabetes (antibodies to GAD65, tyrosine phosphatase [IA-2], and insulin [IAA]), as described earlier (9).

Data on diabetes manifestation (age at onset, glycated hemoglobin [HbA_1c], fasting C-peptide, glycemia levels), treatment at diabetes onset, birth weight, and gestational age were collected retrospectively from medical files. Ketosis was defined by the presence of diabetic ketoacidosis or urine ketone bodies ≥1+ at diabetes diagnosis.

All patients were investigated in a single research center at the Department of Endocrinology of the Lithuanian University of Health Sciences Kauno Klinikos. Clinical data and blood samples were collected during the first visit, and microvascular complications were assessed.

All autoimmune markers (GAD65, IA-2, IAA) were measured by radioimmunoassay. The cutoffs for positivity were >1.0 units/mL for GAD65 and IA-2 and ≥0.4 units/mL for IAA.

Genetic analyses were performed by high-throughput sequencing of DNA and selected for all coding and splicing region of 307 genes (14 MODY and 28 NDM genes, genes involved in glucose homeostasis, pancreas development, and β-cell function [Supplementary Table 1]). The DNA regions were captured by bait using the HaloPlex technology (Agilent Technologies, Santa Clara, CA) (for detailed variant analysis, see Supplementary Data).

The mitochondrial mutation (m.3243A>G) was analyzed by real-time PCR and melting curve analysis.

After molecular diagnosis, patients with variants in known MODY genes were invited to the clinic for a follow-up. Clinical data and blood samples were taken to measure fasting C-peptide levels and HbA_1c. In the context of the treatment change trial, patients with HNF1A, HNF4A, KCNJ11, and ABCC8 diabetes were switched from insulin injections to oral sulfonylurea therapy. Three months after the treatment change, glycemic control was again assessed.

SPSS software, version 22.0, was used for all statistical analyses. Data are presented as mean ± SD, with minimal and maximal values expressed in parentheses, unless stated otherwise. For normally distributed data, Student two-tailed t test (for continuous variables) or the χ² test (for categorical variables) was used. For nonnormal distributions, we used the Mann-Whitney U test. Fisher exact test was applied for contingency tables. P < 0.05 was defined as statistically significant, and all tests were two tailed.

The study was approved by the local and national ethics committees (no. BE-2–5), and all patients and their parents or legal guardians gave informed consent. The study was carried out in accordance with the 1964 Declaration of Helsinki and its later amendments.

The data sets generated during or analyzed during the current study are not publicly available due to patient confidentiality. Data sets will be available from the corresponding author upon reasonable request. No applicable resources were generated or analyzed during the current study.

We screened 1,209 patients with diabetes for autoimmune diabetes antibodies (GAD65, IA-2, and IAA) because antibody positivity is a good discriminatory marker for MD (10) and selected 153 patients for further genetic testing (Fig. 1). The selected patients were autoantibody negative (n = 86) or positive only for IAA (n = 65 [if they had been tested after the introduction of insulin treatment]). In addition, we included patients for whom we had a strong suspicion of MD (n = 2), regardless of positivity for antibodies (11) (Fig. 1). Basic clinical characteristics are represented in Table 1. A higher prevalence of variants in known MD genes was found in the antibody-negative group versus the IAA⁺ group (40.7% vs. 9.2%, P < 0.001) (Table 1).

We identified variants in known diabetes-associated genes in 42 participants, who then were classified as having MD. Of all participants with MD, 16.7% had positive antibodies (7 of 42). GCK diabetes was the most common form of all diabetes subtypes in the cohort, present in 54.8% (23 of 42) of the MD patients and in 25.6% (22 of 86) of those who were antibody negative (Supplementary Tables 2 and 3). HNF1A diabetes was the second most common form identified, with 19.0% (8 of 42) in the MD group and 8.1% (7 of 86) in the antibody-negative group (Supplementary Table 2). HNF4A variants were much rarer, identified in only 7.1% (3 of 42) of all patients with MD and in 1.2% (1 of 86) of the antibody-negative group. Four patients in the cohort had permanent NDM (Supplementary Table 2), all resulting from KCNJ11 changes (9.5% [4 of 42]) (Supplementary Table 3). We found two novel ABCC8 variants, representing 4.8% (2 of 42) of the patients with MD and 1.2% (1 of 86) of the antibody-negative group. A novel INS variant located in the A chain and introducing a stop codon was discovered in one patient (Supplementary Table 2). We also identified a novel KLF11 variant in one child with a mild developmental delay (Supplementary Table 2). We did not detect the mitochondrial mutation 3243A>G in any patients.

In this study, we detected likely pathogenic variants in 11 additional potential diabetes genes, with all variants classified according to the American College of Medical Genetics and Genomics (12). Six of the genes are expressed in human β-cells at the RNA level (13) (Table 2). Among these, four (DACH1, RFX2, RREB1, and ZBED3) encode transcription factors, SLC5A1 encodes a sodium–glucose cotransporter, and TMPRSS6 encodes a protease involved in iron metabolism.

Microvascular complications were present in both patients with variants in the ABCC8 gene, in three of four patients with KCNJ11 variants, and in two of three patients with HNF4A variants (Supplementary Table 2). Surprisingly, 17.4% of patients with GCK diabetes had microalbuminuria.

Of 42 patients with variants in known MD-related genes, 16 were offered precision therapy after the genetic diagnosis. Six patients with HNF1A diabetes and three patients with KCNJ11 NDM were successfully switched to sulfonylurea agents while the insulin substitution was stopped (Table 3).

In this study, we found that MD prevalence was 3.5% (42 of 1,209) and MODY prevalence was 3.14% (38 of 1,209) in the pediatric and young-adult population with diabetes in Lithuania. There was a high pickup rate of MD, with 40.7% (35 of 86) among patients who were antibody negative. These results are higher than reported in previous systematic studies, such as the Norwegian Childhood Diabetes Registry, where 6.5% of antibody-negative diabetes cases involved MODY (14). In that Norwegian study, however, patients diagnosed in an outpatient setting were largely missing from the registry, which was based only on hospital admissions.

The prevalence of MD identified here in the young population with diabetes in Lithuania also is higher than in other countries, including Norway (1.1%) (2), Germany (up to 2.3%) (3), the U.K. (2.5%) (4), New Zealand (2.5%) (5), and the U.S. (1.2%) (6). One explanation for the overall difference is the consistency of our cohort, which included all children with diabetes nationwide and most young adults. In addition, we performed genetic testing using a broad next-generation sequencing panel that included MODY and NDM genes for all patients with suspicion of MD and thus had a larger number of genes tested. Our results are consistent with a Polish nationwide study, however, that yielded a prevalence of MD of 3.1–4.2% (15), which possibly traces to similar ancestry among people living in these neighboring countries.

GCK diabetes was the most common MD form, followed by HNF1A and KCNJ11 in our cohort.

The prevalence of permanent NDM in Lithuania was 0.33% (4 of 1,209), which is slightly higher than the reported prevalence of 0.22% in the SEARCH for Diabetes in Youth (SEARCH) study, 0.15% in Norway (2), and 0.16% in Italy. No patient had diabetes associated with the mitochondrial 3243A>G mutation, possibly because the mean age of diabetes onset is rather after the third decade (16).

We report several novel, likely pathogenic gene variants that may affect diabetes development. Four genes—DACH1, RFX2, RREB1, and ZBED3—are transcriptional regulators and expressed in human β-cells. Dach1 gene deletion in mice reduces the numbers of β-cells and blocks perinatal β-cell proliferation (17). RFX2 is a member of the RFX protein family, which includes RFX3 and RFX6—both known diabetes-associated genes (18,19). RREB1 is a transcription factor that binds specifically to the RAS-responsive elements of gene promoters and potentiates the transcriptional activity of NEUROD1 (20), a known MODY-related gene. In general, RAS-responsive element binding augments RAS-mediated transcription response of that promoter. In addition, coding variants in RREB1 are associated with type 2 diabetes, and recent data suggest a role in β-cell function and development (21). Thus, RREB1 variants with more dramatic effects could be involved in childhood diabetes, and those with a smaller effect might have a role in type 2 diabetes. Such a role has been described for variants of KCNJ11 and HNF4A, which have been associated with MODY or type 2 diabetes, depending on how the variant affects protein function (22,23).

In line with the proposed approach of precision medicine, we performed a prospective treatment change in participants with HNF1A, HNF4A, KCNJ11, and ABCC8 diabetes. The most significant predictor for a positive response to sulfonylurea treatment was a higher C-peptide level, followed by age at treatment change and diabetes duration. The latter two factors have recently been associated with treatment switch success (24), and our findings confirm this link.

The main strength of our study is the coverage of 100% of pediatric patients with diabetes and the majority of young-adult patients in Lithuania. Moreover, we evaluated autoimmunity status for all participants and recruited for genetic testing only those with negative GAD65 and IA-2 or with strong suspicion for MD. In our genetically analyzed cohort, GAD65 and IA-2 positivity was 2.4%, while others have reported <1% positivity in patients with MODY (10).

Our pickup rate of 40.7% justifies screening for MD in autoimmune antibody–negative pediatric patients, even if the analysis is done after diabetes onset. Of note, in the IAA⁺ subgroup, 6 (9.2%) had a positive genetic result, so children who are positive for IAA antibodies and treated with insulin should not be excluded from genetic testing.

Follow-up studies in patients in whom we could not identify any suggestive genetic variant will include a genetic risk score analysis for type 1 diabetes (25). To further evaluate the impact of possible mutations on novel genes, we plan to perform functional analysis in vitro for selected variants to reveal their contribution to diabetes development.

Acknowledgments. The authors greatly acknowledge the participants of the study and their families. The authors thank Antoine Poncet, Division of Clinical Epidemiology, Geneva University Hospitals, Geneva, Switzerland, for help with the statistical analysis. The authors thank Jeremy Bevillard, Pediatric Endocrine and Diabetes Unit, Department of Pediatrics, Gynecology and Obstetrics, University Hospitals of Geneva, Geneva, Switzerland, for technical assistance in genetic analysis.

Funding. The study was jointly supported by a grant from Lithuanian Research Council Lithuanian-Swiss program “Research and development” and the Federal Department of Foreign Affairs of Switzerland (CH-3-ŠMM-01/09) and by the Swiss National Science Foundation (grants CR33I3_140655 and CR33I3_1166591 to V.M.S.).

Duality of Interest. This study was also supported by a Merck-sponsored educational program, European Society for Paediatric Endocrinology (ESPE) Clinical Fellowship, to I.S. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. I.S. participated in data collection and database creation, carried out statistical analyses, and wrote the manuscript. R.V. participated in the conception, design, and coordination of the study. J.-L.B. carried out the study design for genetic analysis, supervised the genetic analyses, and carried out data interpretation. P.K. helped to set up the database and participated in manuscript editing. R.D. participated in the conception and design of the study and coordinated pediatric patient data collection. E.D. participated in the design of the study and coordinated adult patient data collection. M.D. participated in manuscript editing. F.S. programmed the pipeline for genetic data analysis and analyzed the genetic data. D.R.-V. participated in adult patient enrollment and in data collection and database creation. D.M. participated in the conception, design, and coordination of the study. E.J. participated in pediatric patient enrollment. G.M. participated in pediatric patient enrollment. V.M.S. participated in the conception and design of the study, contributed to discussion, and edited and revised the manuscript. V.M.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at ENDO 2016, Boston, MA, 1–4 April 2016; at the 42nd Annual Meeting of the International Society for Pediatric and Adolescent Diabetes, Valencia, Spain, 26–29 October 2016; and at the 7th Meeting of the European Association for the Study of Diabetes Study Group on the Genetics of Diabetes (EASD-SGGD), Prague, Czech Republic, 16–18 May 2019.