The Genetic Spectrum of Maturity-Onset Diabetes of the Young (MODY) in Qatar, a Population-Based Study

Maturity-onset diabetes of the young (MODY) is a rare monogenic form of diabetes mellitus. In this study, we estimated the prevalence and genetic spectrum of MODY in the Middle Eastern population of Qatar using whole-genome sequencing (WGS) of 14,364 subjects from the population-based Qatar biobank (QBB) cohort. We focused our investigations on 14 previously identified genes ascribed to the cause of MODY and two potentially novel MODY-causing genes, RFX6 and NKX6-1. Genetic variations within the 16 MODY-related genes were assessed for their pathogenicity to identify disease-causing mutations. Analysis of QBB phenotype data revealed 72 subjects (0.5%) with type 1 diabetes, 2915 subjects (20.3%) with type 2 diabetes and 11,377 (79.2%) without diabetes. We identified 22 mutations in 67 subjects that were previously reported in the Human Genetic Mutation Database (HGMD) as disease-causing (DM) or likely disease causing (DM?) for MODY. We also identified 28 potentially novel MODY-causing mutations, predicted to be among the top 1% most deleterious mutations in the human genome, which showed complete (100%) disease penetrance in 34 subjects. Overall, we estimated that MODY accounts for around 2.2–3.4% of diabetes patients in Qatar. This is the first population-based study to determine the genetic spectrum and estimate the prevalence of MODY in the Middle East. Further research to characterize the newly identified mutations is warranted.


Introduction
Maturity-onset diabetes of the young (MODY) is a rare monogenic form of diabetes mellitus, accounting for approximately 1-5% of diabetes cases worldwide [1]. MODY is an autosomal dominant condition, clinically characterized by a multigenerational family history of diabetes and absence of pancreatic β-cell autoantibodies, and disease onset before the age of 25 years in most cases. A prominent feature of MODY includes progressive β-cell dysfunction, while in some cases it could be associated with other organ abnormalities, such as kidney abnormalities and fat malabsorption [2]. Patients with MODY are often misdiagnosed due to similarities in their clinical manifestations to those with other forms of diabetes [3]. To date, 14 genes are ascribed to the cause of MODY, each encoding for a different subtype [2]. Recent studies have identified novel variants in two genes encoding regulatory factor X6 (RFX6) and NK6 Homeobox 1 (NKX6-1) as possible MODY contributors [4,5].
The overview of the study design is illustrated in Figure 1. This study was based on WGS and phenotypic data of 14,364 participants from the population-based QBB cohort [26]. Analysis of phenotype data showed that 72 participants (0.5%) of the study cohort were classified as T1DM, 2915 (20.3%) as T2DM, while 11,377 (79.2%) were classified as nondiabetes. The clinical characteristics of the study population are listed in Table 1. Female participants constituted 55.8% of the study population, while the age of the overall study cohort ranged from 18 to 89 years. Notably, the prevalence of obesity in Qatar is remarkably high (77.6% of the total study population were overweight or obese). Additionally, 74.1% of non-diabetes participants and 91.0% of diabetes patients in our study were overweight or obese. Only 35 subjects (0.24%) met the clinical criteria for diagnosing MODY based on being diabetic with normal BMI and the age of onset before <25 (only 63% of diabetes patients declared age of disease onset). Therefore, the widely accepted clinical criteria for MODY diagnosis was not accurate when applied to our study cohort, given that most patients were overweight or obese, and genetic testing is critical in MODY diagnosis. obese. Only 35 subjects (0.24%) met the clinical criteria for diagnosing MODY based on being diabetic with normal BMI and the age of onset before <25 (only 63% of diabetes patients declared age of disease onset). Therefore, the widely accepted clinical criteria for MODY diagnosis was not accurate when applied to our study cohort, given that most patients were overweight or obese, and genetic testing is critical in MODY diagnosis. Figure 1. Study Design. This study was based on whole-genome sequencing (WGS) data of the Qatar Biobank (QBB) participants. The analysis cohort comprised n = 14,364 subjects, classified into diabetes mellitus subtypes (T1DM; n = 72 and T2DM; n = 2915) and non-diabetes subjects (n = 11,377) based on phenotypic data. The detected mutations were classified into previously known or potentially novel MODY-causing mutations based on pathogenicity annotations in the human gene mutation database (HGMD), QIAGEN Clinical Insight and ClinVar databases or Combined Annotation-Dependent Depletion (CADD) scores, as well as on disease (diabetes mellitus) penetrance.  . Study Design. This study was based on whole-genome sequencing (WGS) data of the Qatar Biobank (QBB) participants. The analysis cohort comprised n = 14,364 subjects, classified into diabetes mellitus subtypes (T1DM; n = 72 and T2DM; n = 2915) and non-diabetes subjects (n = 11,377) based on phenotypic data. The detected mutations were classified into previously known or potentially novel MODY-causing mutations based on pathogenicity annotations in the human gene mutation database (HGMD), QIAGEN Clinical Insight and ClinVar databases or Combined Annotation-Dependent Depletion (CADD) scores, as well as on disease (diabetes mellitus) penetrance. Study cohort comprised of 14,364 subjects. * Statistically significant (p < 0.001) compared to non-diabetes controls. BMI: body mass index. ** Subjects with BMI below 18.5 kg/m 2 were classified as underweight, between 18.5 to 24.9 kg/m 2 as normal weight, between 25 to 29.9 kg/m 2 as overweight and greater than or equal to 30 kg/m 2 as obese [27]. N/A, not available.

Identification of Previously Known MODY Causative Variants
We first investigated previously known variants associated with MODY onset. We investigated variants that were previously reported as disease-causing (DM) or likely diseasecausing (DM?) mutations for MODY in the human gene mutation database (HGMD) and classified as pathogenic or likely pathogenic in Qiagen Clinical Insight (QCI) and/or Clin-Var databases ( Table 2). QCI classification criteria are based on the 2015/2016 ACMG/AMP Guidelines. Variants were also annotated by SIFT and PolyPhen-2 databases to indicate the potential functional effect of each mutation. We detected 10 previously identified missense MODY mutations in 24 participants ( Table 2). The majority of detected variants were located in HNF1A. However, rs587778397 (p.Arg177Trp) in HNF1A and rs766934515 (p.Val113Gly) in BLK were the most frequently occurring mutations. Additional variants in HNF4A, GCK, KLF11 and ABCC8 were also recorded. Moreover, 10 subjects carried mutations in the HNF1A gene, 5 in BLK and 4 in GCK. Complete penetrance was observed in HNF4A (rs1555817727; p.Arg312Cys) and HNF1A (rs137853238; p.Arg272His).
In addition, we also identified variants of uncertain significance (VUS) that are predicted to rank in the top 1% of most deleterious mutations in the human genome based on CADD Phred scores (>20). We identified 12 missense VUS in 43 subjects (Table 3). Most subjects were heterozygous carriers of variants in HNF1A (n = 26), followed by GCK (n = 5) and BLK (n = 5) genes. HNF1A (rs202039659; p.His514Arg) was the most frequent mutation, recorded in 11 subjects. However, complete penetrance was only recorded for HNF4A (rs371124358; p.Arg310Gln and rs377151067; p.V404I). Combined, we found that previously known MODY mutations are mostly detected in the HNF1A gene, accounting for 53.7% of cases, followed by BLK and GCK mutations accounting for approximately 15% and 13% of cases, respectively. Overall, we identified 22 MODY-causing mutations in 8 genes, accounting for 2.2% of diabetes participants (Tables 2 and 3).

Identifying Potentially Novel MODY-Causing Mutations
Next, we investigated potentially novel MODY-causing mutations based on complete disease penetrance (100%) and a CADD Phred score >20. We detected 28 mutations in 14 genes in 34 diabetes subjects ( Table 4). None of these variants are reported as MODYcausing mutations in the HGMD or ClinVar databases, while the majority are predicted to be damaging by Polyphen-2 and deleterious by SIFT. In addition, 18 mutations are not reported in dbSNP or the Genome Aggregation Database (gnomAD). All detected novel mutations were missense, while variants in HNF1A (MODY3) and CEL (MODY8) were the most common, detected in eight subjects. Moreover, we also detected variants in rare MODY-related genes, including NEUROD1 (MODY6), KCNJ11 (MODY13) and APPL1 (MODY14), among other MODY subtypes. Notably, we also identified three variants in RFX6 and two variants in NKX6-1 in seven subjects ( Figure 2). Importantly, all potentially novel mutations were detected in patients classified as T2DM based on phenotype data.

Estimating the Prevalence of MODY in Qatar
We detected 22 previously known MODY-causing mutations in 67 subjects, yielding a prevalence of MODY at 2.2% in individuals with diabetes ( Figure 1). In addition, we identified 28 potentially novel MODY-causing mutations, accounting for ~56% of the overall reported mutations in 34 subjects and 1.14% of diabetes cases. Of note, combining subjects carrying mutations in both known and potentially novel MODY-causing mutations in our study population yielded an overall prevalence of 3.4% of diabetes patients in Qatar (Figure 1). Overall, previously reported and potentially novel MODY mutations were detected in 101 subjects. All detected mutations were heterozygous. The proportions of subjects carrying mutations in each MODY subtype are illustrated in Figure 3. We found that HNF1A-MODY3 is the most prevalent MODY subtype in the

Estimating the Prevalence of MODY in Qatar
We detected 22 previously known MODY-causing mutations in 67 subjects, yielding a prevalence of MODY at 2.2% in individuals with diabetes ( Figure 1). In addition, we identified 28 potentially novel MODY-causing mutations, accounting for~56% of the overall reported mutations in 34 subjects and 1.14% of diabetes cases. Of note, combining subjects carrying mutations in both known and potentially novel MODY-causing mutations in our study population yielded an overall prevalence of 3.4% of diabetes patients in Qatar (Figure 1). Overall, previously reported and potentially novel MODY mutations were detected in 101 subjects. All detected mutations were heterozygous. The proportions of subjects carrying mutations in each MODY subtype are illustrated in Figure 3. We found that HNF1A-MODY3 is the most prevalent MODY subtype in the Qatari population, accounting for approximately 40% of cases. In addition, we also compared the clinical characteristics of subjects carrying MODY-related mutations (Supplementary Table S1). The identified MODY patients were predominately overweight and obese (83.1%), while they also showed strong diabetic medical history (95%). Moreover, the gender distribution of MODY cases showed higher prevalence in females (54.5%). In addition, we also compared the clinical parameters and genetic matrices of MODY-related genes between non-diabetes, T1DM and T2DM patients (Supplementary Table S2 and Supplementary Figure S1). Based on clinical parameters, half of the participants who carried mutations in MODY-related genes were classified as T2DM patients and half were non-diabetes based on phenotype data, while one patient was classified as T1DM (Supplementary Table S2). The genetic profile of these subjects also showed that the majority of them carried mutations in HNF1A (Supplementary Figure S1).

Discussion
The etiopathogenesis and clinical manifestations of MODY are broadly analogous to other forms of diabetes, such as youth-onset T2DM, which renders clinical risk assessment and genetic testing critical for accurate diagnosis and therapy selection. However, due to the challenges associated with genetic screening, MODY is underreported in the MENA region. In Europe, MODY accounts for 1-5% of all diabetes mellitus cases [28]. We estimated the prevalence of MODY in Qatar at 2.2% of diabetes patients, while inclusion of potentially novel mutations yielded a higher proportion of MODY, accounting for 3.4% of cases. The MENA region has one of the highest and fastest increasing levels of diabetes prevalence compared to global estimates [29]. We estimated the prevalence of diabetes in the population-based QBB cohort as ~21%, in agreement with previous reports (~17-20%) [30][31][32].

Discussion
The etiopathogenesis and clinical manifestations of MODY are broadly analogous to other forms of diabetes, such as youth-onset T2DM, which renders clinical risk assessment and genetic testing critical for accurate diagnosis and therapy selection. However, due to the challenges associated with genetic screening, MODY is underreported in the MENA region. In Europe, MODY accounts for 1-5% of all diabetes mellitus cases [28]. We estimated the prevalence of MODY in Qatar at 2.2% of diabetes patients, while inclusion of potentially novel mutations yielded a higher proportion of MODY, accounting for 3.4% of cases. The MENA region has one of the highest and fastest increasing levels of diabetes prevalence compared to global estimates [29]. We estimated the prevalence of diabetes in the population-based QBB cohort as~21%, in agreement with previous reports (~17-20%) [30][31][32].
The prevalence of MODY mutations among children with diabetes from different ethnicities was estimated at 4.56 per 100,000 individuals (~1.5% of diabetes cases) according to a recent single clinic-based study conducted in Qatar [25]. Among the 12 mutations detected in children, 9 mutations were also detected in our study cohort. These included one previously reported mutation in HNF1A (rs202039659; p.His514Arg) and a previously known mutation in GCK (rs1375656631; p.Ala259Thr). However, seven additional mutations were detected in our study cohort but they did not meet our analysis criteria, such as HGMD and Clinvar classifications. Overall, the differences in reported variants among MODY cases in children compared to Qatari adults could be attributed to differences in study design, given that our study was a large population-based cohort of Qatari adults (18-89 years), whereas the previous study was a small study of children with different ethnicities attending a single diabetes clinic in Qatar.
We identified 22 previously reported and 28 potentially novel MODY-causing mutations in 101 participants, which accounted for 3.4% of diabetes patients. The estimated prevalence is in accordance with the widely accepted overall prevalence of MODY, accounting for 1-5% of all diabetes cases. Combined, mutations in HNF1A were the most common, detected in 40 cases (39.6%); followed by BLK (10.9%) and GCK (9.9%) genes. HNF1A MODY-causing mutations are the most prevalent in European, North American, and Asian populations; accounting for 20-50% of all MODY cases [3,6,[33][34][35]. We also detected several known but rare MODY-causative mutations in different genes including PDX1-MODY4, KLF11-MODY7 and ABBC8-MODY12. However, many subjects carrying previously reported mutations did not present the clinical symptoms of diabetes, which could be attributed to incomplete penetrance of MODY mutations, which varies with differences in age and in genetic architecture between populations [6,36]. MODY is well known for its clinical and genetic heterogeneity with variable disease penetrance. Notably, the associations between BLK mutations and MODY penetrance has been indecisive. Five MODY causing mutations in BLK were identified in three families including p.A71T [37]. However, p.A71T was detected in 52 normoglycemic subjects in a European cohort and showed weak associations with increased T2DM risk [38].
We identified 28 potentially novel MODY-causing mutations in 34 participants (~1.1% of diabetes cases), accounting for~56% of the overall reported mutations. Twenty-three mutations were in MODY-causing genes, while five mutations were detected in potentially novel MODY-causing genes-three variants in RFX6 and two variants in NKX6-1 in seven subjects. Several studies have associated variants in these genes with MODY progression [4]. Notably, the transcription factor RFX6 plays a vital role in the development of endocrine pancreas and in the differentiation of islet cells, including insulin-producing cells [39]. To date, only a few potential MODY-causing variants of RFX6 have been identified [4,5,40]. Mutations in the RFX6 gene, for instance, c.1954C>T (p.Arg652X), have been found to reduce insulin secretion and insulin-dependent insulinotropic polypeptide responses [5,40]. Moreover, GWAS have associated variants of RFX6 with blood glucose levels [41] and T2DM [42]. Similarly, NKX6.1 has been identified as a critical transcription factor in the early and late development of pancreatic β-cells and in maintaining β-cells' identity, and it regulates many genes involved in glucose uptake, metabolism and insulin biosynthesis [43]. Moreover, variants in NKX6-1 have also been associated with T2DM [44]. Notably, only two mutations in NKX6-1, P329L and S317L have been described in South India and associated with MODY development [4]. These genes could have a profound effect on MODY onset, but further genetic and functional investigations are warranted.
Despite recent advances in technology, the genetic and molecular basis of MODY onset remains to be fully elucidated. The complexity and broad spectrum of phenotypes associated with MODY, including de novo mutations and genetic background, make it challenging to conduct a clinical manifestation profile for all subtypes. However, identification of population-specific mutations attributed to MODY has the potential for the development of diagnostic or prognostic microarrays as a more feasible option. For instance, multiplex ligation-dependent probe amplification led to the identification of two de novo mutations (deletions) in the HNF1B gene in suspected MODY cases in a Greek cohort [45].
Our findings highlight the genetic anomalies associated with MODY in Qatar based on a substantially large cohort. Importantly, our findings also reiterate the importance of genetic testing for MODY diagnosis given that our cohort predominantly comprised subjects classified as overweight or obese. Moreover, of the 101 suspected MODY subjects who carried mutations in MODY-related genes, 84 patients (83.2%) were overweight or obese, consistent with high obesity prevalence in Qatar. Therefore, the widely accepted clinical criteria for MODY diagnosis is not accurate when applied to our study cohort given that most patients were overweight or obese, and genetic testing is critical in MODY diagnosis.
Our study was, however, limited by the lack of detection for structural variants of MODY-associated genes, although MODY is predominantly caused by missense mutations. Moreover, additional clinicopathological data to investigate associations between specific mutations and disease severity, age of diagnosis and mode of inheritance (inherited or de novo) were not available. In addition, details about autoantibody detection to stratify patients into diabetes subtypes were also unavailable. Overall, the reported variants and mutations may be explored further in functional analysis to confirm their pathogenicity in other populations and in the development of diagnostic and prognostic tools for MODY.

Study Participants
Qatar Biobank (QBB) was the primary source of study participants. QBB aims to gather information from the native Qatari population, as well as from long-term residents (≥15 years), with follow-ups scheduled every 5 years. The study cohort analyzed in this study consisted of 14,364 Qatari participants aged between 18 to 89 years. Informed consent was obtained from all participants prior to inclusion in the study. The study was approved by the institutional review boards of QBB (E-2019-QF-QBB-RES-ACC-0179-0104) and Hamad Bin Khalifa University (approval no. 2021-03-078). Participants underwent extensive medical examination, filled in a standardized questionnaire and self-reported past medical histories, diet, physical activity and lifestyle [26]. Physical measurements, such as the body mass index (BMI), and biological samples, including blood, urine and saliva, were also collected. Measurements relevant to diabetes were investigated by the Qatar National Health Service (Hamad Medical Corporation, Doha, Qatar) to assess the participants' health status by evaluating clinical diagnostic biomarkers, including glucose, hemoglobin A1c (HbA1c) and C-peptide levels.

Phenotypic Data and Patient Classification
Study participants were identified as diabetes or non-diabetes based on blood biomarkers (C-peptide, HbA1c, glucose), self-reported diabetes and self-reported administration of anti-diabetes medications. Serum C-peptide levels were determined by the sandwich electrochemiluminescence immunoassay using Elecsys C-Peptide kit (Roche, Basel, Switzerland), while HbA1c in blood was measured using turbidimetric inhibition immunoassay (TINIA) utilizing Tina-quant HbA1c Gen. 3 kit (Roche) on hemolyzed blood samples and random glucose levels in serum were measured using the enzymatic reference method with GLUC3 glucose hexokinase kit (Roche), utilizing a COBAS instrument (Roche).
We classified our study cohort based on the phenotypic data provided by QBB. We first identified T1DM as self-declared diabetes subjects receiving insulin treatment exclusively and with serum C-peptide levels < 0.5 ng/mL. Subjects who did not meet the criteria for T1DM were classified as T2DM if they met any of the following criteria: self-declared diabetes, self-reported diabetes medication or with HbA1c > 6.5%. Subjects who did not meet the T1DM or T2DM criteria were classified as non-diabetes subjects.

Whole-Genome Sequencing
The WGS data of the QBB study participants was provided by the Qatar Genome Project (QGP) [46]. Details of WGS and quality control measures have been described previously [47]. Briefly, genomic DNA from peripheral blood samples was extracted using Qiagen MIDI kit (Qiagen, Hilden, Germany), following the manufacturer's protocol and using the automated QIASymphony SP instrument (Qiagen). DNA integrity was evaluated using Caliper Labchip GXII (Perkin Elmer, Waltham, MA, USA) Genomic DNA assay and was quantified using the Quant-iT dsDNA Assay (Invitrogen, Waltham, MA, USA). Whole-genome libraries were prepared using Illumina TruSeq DNA Nano kit (Illumina, San Diego, CA, USA). HiSeq X Ten (Illumina) was used to sequence genomic libraries to achieve a minimum average coverage of 30×. Library construction and sequencing was performed at Sidra Clinical Genomics Laboratory Sequencing Facility (Sidra Medicine, Doha, Qatar). The generated files were subjected to quality control using FastQC (v0.11.2) and reads were aligned to the CRCh38 reference genome. Quality control on mapped reads was performed using Picard (v1.117). A combined variant call file (gVCF) was generated for all study subjects, which contained all genetic variations detected in the QBB study participants.

Bioinformatics Analysis to Identify MODY-Causing Mutations
Genetic variants of the following known MODY-related genes-HNF4A, GCK, HNF1A, PDX1, HNF1B, NEUROD1, KLF11, CEL, PAX4, INS, BLK, ABCC8, KCNJ11, APPL1, RFX6 and NKX6-1-were extracted from the gVCF file, based on genomic coordinates obtained from the University of California Santa Cruz (UCSC) Genome Browser [48]. We included genetic variants with minor allele frequency < 0.001 and mean sequencing depth > 20×. The average sequencing depth of MODY-related variants was 32.1×. We extracted genetic variants based on their location in each of the MODY-causing genes and related the genotype data to the clinical phenotype data for stratification into MODY subtypes. Variant annotations were performed using Human Gene Mutation Databases (HGMD) [49] to identify previously reported disease-causing mutations (DM) or likely disease-causing mutations (DM?). The clinical significance of variants was investigated using the QIAGEN Clinical Insight (Qiagen) and ClinVar database [50]. Potentially novel MODY-related mutations were determined based on a Combined Annotation-Dependent Depletion (CADD) [51] Phred score > 20 and complete disease penetrance (100%), implying that the subject had diabetes and the genetic variant showed a CADD Phred score > 20 (predicated to be among the top 1% most deleterious mutations in the human genome). Penetrance for each variant was calculated by dividing the number of carriers showing the phenotype (T1DM or T2DM as described above) by the overall number of mutation carriers (N). The functional annotations of the identified variants was performed using Sorting Intolerant From Tolerant (SIFT) [52] and Polymorphism Phenotyping v2 (Polyphen-2) [53] tools.

Statistical Analysis
Statistical analyses were performed using GraphPad Prism 9 software (GraphPad Software, San Diego, CA, USA). Shapiro-Wilk normality test was carried out to determine data distribution. Unpaired t-test was used for data comparisons between groups on normally distributed data, while Mann-Whitney test was used for datasets that did not show normal distribution. Chi-square (χ 2 ) test was performed to determine the differences between binary variables within study cohorts. A p value of < 0.05 was considered statistically significant. Data are presented as mean ± standard deviation (SD). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data analyzed in this study are subject to the following licenses/ restrictions: the raw whole-genome sequence data from Qatar Biobank are protected and are not available for deposition into public databases due to data privacy laws. Access to QBB/QGP phenotype and whole-genome sequence data can be obtained through an ISO-certified protocol, which involves submitting a project request at https://www.qatarbiobank.org.qa/research/how-apply, subject to approval by the Institutional Review Board of the QBB. Requests to access these datasets should be directed to https://www.qatarbiobank.org.qa/research/how-apply.