Prevalence of maturity-onset diabetes of the young in phenotypic type 2 diabetes in young adults: a nationwide, multi-center, cross-sectional survey in China

Abstract

Background:

Maturity-onset diabetes of the young (MODY) is the most common monogenic diabetes. The aim of this study was to assess the prevalence of MODY in phenotypic type 2 diabetes (T2DM) among Chinese young adults.

Methods:

From April 2015 to October 2017, this cross-sectional study involved 2429 consecutive patients from 46 hospitals in China, newly diagnosed between 15 years and 45 years, with T2DM phenotype and negative for standardized glutamic acid decarboxylase antibody at the core laboratory. Sequencing using a custom monogenic diabetes gene panel was performed, and variants of 14 MODY genes were interpreted as per current guidelines.

Results:

The survey determined 18 patients having genetic variants causing MODY (6 HNF1A, 5 GCK, 3 HNF4A, 2 INS, 1 PDX1, and 1 PAX4). The prevalence of MODY was 0.74% (95% confidence interval [CI]: 0.40–1.08%). The clinical characteristics of MODY patients were not specific, 72.2% (13/18) of them were diagnosed after 35 years, 47.1% (8/17) had metabolic syndrome, and only 38.9% (7/18) had a family history of diabetes. No significant difference in manifestations except for hemoglobin A1c levels was found between MODY and non-MODY patients.

Conclusion:

The prevalence of MODY in young adults with phenotypic T2DM was 0.74%, among which HNF1A-, GCK-, and HNF4A-MODY were the most common subtypes. Clinical features played a limited role in the recognition of MODY.

Introduction

Monogenic diabetes plays a key role in the implementation of precision diabetes, being one of the few forms of diabetes with a specific etiology. Considering the utmost importance of genetic etiology for pharmacologic treatment, disease progression prognosis, family counseling, and efficient screening, the accurate diagnosis of monogenic diabetes is vital for clinical practice.^[1–3]

Maturity-onset diabetes of the young (MODY) is the most common monogenic diabetes; patients with hepatocyte nuclear factor-1 alpha (HNF1A)- and 4 alpha (HNF4A)-MODY are sensitive to low-dose sulfonylureas, while glucokinase (GCK)-MODY could be treated with diets, except during pregnancy.^[3,4] Several studies have attempted to determine the prevalence of MODY in different populations. Carlsson et al^[5] elucidated that the minimum prevalence of MODY was 1.2% in Swedish patients, diagnosed with diabetes before 18 years of age. The Treatment Options for Type 2 Diabetes in Adolescents and Youth study discovered that in an overweight/obese cohort, 4.5% of participants aged 10 to 17 years should be diagnosed with MODY.^[6] Ma et al^[7] reported that the prevalence of GCK-MODY was approximately 1.3% in the Chinese population with diabetes between 25 and 75 years of age. However, the majority of studies in available literature concentrated on pediatric clinics or specific MODY subtypes. There are limited relevant studies focusing on all subtypes of MODY in unselected young adults with phenotypic type 2 diabetes (T2DM). Therefore, conducting a nationwide and multi-center survey is critical.

In addition to the sparse preliminary data on prevalence in young adults, another barrier for the application of precision diabetes for MODY is the unclear role of clinical features in identifying MODY patients. MODY is characterized by an early onset, autosomal dominant mode of inheritance, and a primary defect in pancreatic β-cell function.^[8,9] Overlaps in clinical presentations caused misdiagnosis as type 1 diabetes or T2DM in about 80% of MODY patients, with a mean delay of >10 years from diabetes diagnosis to the correct genetic diagnosis, leading to years of unnecessary treatment.^[10,11] To help identify young patients with diabetes who might have MODY, Shields et al^[12] developed a MODY probability calculator based on the clinical characteristics of Caucasian patients, which showed clear discrimination between MODY and type 1 diabetes or T2DM. However, this calculator was only applicable to patients diagnosed ≤35 years and its performance in Asian populations was conflicting.^[13,14]

To address these questions, we conducted a nationwide, multicenter, cross-sectional study in young adults with diabetes to investigate the prevalence of MODY and to evaluate the diagnostic values of clinical parameters in discriminating MODY.

Methods

Ethical approval

This study was approved by the Ethics Committee of the Second Xiangya Hospital, Central South University in China (No. 2014032). The ethics review committee/institutional review board of every participating hospital approved the study protocol. Participants aged ≥18 years were asked to provide informed consent for themselves, while for individuals <18 years, approval was obtained from their parents.

Research design and subjects

This study was conducted from April 2015 to October 2017 and included patients diagnosed with diabetes at ≥15 years. They were recruited consecutively from 46 tertiary care hospitals in 24 provincial administration areas to participate in this survey; the 46 participating hospitals were distributed across all of the seven geographic regions of China (four Northeast, eight North, three Northwest, nine Central, three Southwest, seven South, and 12 East), thereby representing the diversity in climates, cultures, and ethnicities of the Chinese population.^[15] The inclusion criteria were as follows: (a) diagnosed with diabetes according to the American Diabetes Association (ADA) recommendations,^[16] (b) new-onset patients with <1 year disease duration, (c) negative glutamic acid decarboxylase antibody (GADA), to exclude latent autoimmune diabetes in adults, and (d) outpatients attending clinics in the Department of Endocrinology of the participating hospitals. Among 19,163 enrolled participants, we included 5687 young adults with phenotypic T2DM diagnosed ≤45 years. Finally, 2429 samples were available with genetic testing [Figure 1].

Figure 1.

Flow chart of the study participants. Age_diag: Age at diagnosis; T1DM: Type 1 diabetes mellitus; T2DM: Type 2 diabetes mellitus.

Data collection

Demographic characteristics, clinical features, medical history, and lifestyle information were collected using a standard questionnaire via face-to-face interviews by research nurses. Patient height, weight, waist circumference, hip circumference, and blood pressure were measured using standard procedures. Body mass index (BMI) was calculated as weight in kilograms divided by squared height in meters. Medication information was retrieved from medical notes during diagnosis. For confirmation, the completed questionnaires were sent to a second trained nurse. Finally, the completed databases were uploaded onto a centralized database periodically.

Laboratory assays

Fasting venous blood samples were collected for measurements of hemoglobin A1c (HbA1c), plasma glucose, fasting C-peptide (FCP), triglycerides (TGs), total cholesterol levels, high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C), whereas postprandial venous blood samples were tested for C-peptide and plasma glucose. All were assayed by standard methods at the study sites.

Serum samples for the GADA assay were transported to the core laboratory in Central South University on ice within the day and stored at −80°C before analysis. The cutoff value of positivity for GADA was 0.05 index/mL, and the positivity was confirmed by repeated assays. The sensitivity and specificity of GADA testing were 82% and 98%, respectively, as shown in the 2016 Islet Autoantibody Standardization Program.^[17]

Molecular analyses

Genomic DNA was extracted from ethylenediaminetetraacetic acid (EDTA)-anticoagulated blood using the phenol-chloroform method. A multiplex polymerase chain reaction (PCR) panel was designed to detect nucleotide substitutions by amplifying the coding regions ± 50 bp of 36 genes and mitochondrial 3243 A > G, which are known to cause monogenic diabetes, while our study focused on 14 genes identified to cause MODY. The total amplicon number was 738 and the target size was 102.2 kb and with a theoretical coverage of 97.49% for the targeted regions. The library was constructed using a two-step PCR method. In the first PCR, the targeted DNA region was amplified using specific primers flanked by tails. These tails allow for a second PCR to add Illumina adaptor sequences and indexes to multiplex samples. Thereon, the multiplex libraries were pooled and run on NovaSeq 6000 Sequencing System (Macrogen Inc., Korea). The Cutadpter 3.0 program was used to remove adapters and primers before mapping. GRCh38/HG38 was selected as a reference sequence and the BWA 0.7.12 software package was used to map reads against the reference genome at the default settings. To recalibrate base quality scores and perform local realignment around known indels, Picard was used. Genetic variants were called with GATA4.1, using the default setting. Single-nucleotide variants with genotype quality scores <50 and coverage depth <30 × were filtered from the analysis. The average sequencing depth of the target region was >1000 × , and samples with ≥30 × mean base coverage depth of ≥99% of the target region were used for analysis. Variants were annotated by ANNOVAR software using multiple large population databases, in silico prediction tools, and disease databases.

All non-synonymous substitutions and nucleotide changes at a splice site were evaluated by database searches. Non-common (<5% minor allele frequency) coding or splice-site variants were analyzed for pathogenicity according to American College of Medical Genetics/Association for Molecular Pathology (ACMG/AMP) guidelines for variant interpretation.^[18] The considerations of population data, computational data, clinical phenotypes, and functional assessments of the variants, were classified into five categories. Thereon, patients were diagnosed with MODY when they carried a pathogenic (P) or likely pathogenic (LP) variant. Modified from the ACMG/AMP guidelines, maximum mutant allele frequency ≤0.1% and pathogenicity predicted by a set of bioinformatic tools were used to redefine the variants of uncertain significance (VUS), which could be summarized as PM2 and PP3.

To classify the identified variants, the following available online bioinformatic tools were used: Exome Aggregation Consortium Browser, 1000 Genomes, the Genome Aggregation Database, RefSeq, dbSNP, Sorting Intolerant From Tolerant, likelihood ratio test, Mutation Taster, Mutation Assessor, Functional Analysis Through Hidden Markov Models, Protein Variation Effect Analyzer, Genomic Evolutionary Rate Profiling, and ClinVar.

Definition of metabolic syndrome (MetS)

As per the 2017 Chinese Diabetes Society's criteria, MetS was diagnosed when three or more of the following criteria were met: (1) central obesity: waist circumference ≥90 cm in men and ≥85 cm in women; (2) TGs ≥1.70 mmol/L; (3) HDL-C <1.04 mmol/L; (4) hypertension: blood pressure ≥130/85 mmHg or currently under antihypertension therapy; (5) hyperglycemia: fasting blood glucose ≥6.1 mmol/L, or postprandial blood glucose ≥7.8 mmol/L or currently under antihyperglycemic therapy.^[19] All patients in the study were defined to fulfill the criteria for hyperglycemia.

Tools for MODY identification

To discriminate MODY from other subtypes of diabetes, the following parameters were used to build a MODY clinical risk score: (1) age of onset ≤30 years, (2) without MetS, (3) a family history of diabetes (FHD), (4) FCP >200 pmol/L, and (5) without diabetic ketoacidosis (DKA). As possible factors for clinical diagnosis of MODY, meeting one parameter was counted as 1 point, and the full score was 5.

In addition to the MODY clinical risk score, the diagnostic value of “3-1-2” Clinical Diagnostic Criteria of MODY^[20] and MODY probability calculator^[12] were evaluated in this study.

Statistical analysis

Statistical analysis was performed with SPSS 22.0 software (IBM Corporation, Chicago, IL, USA). Measurement data were expressed as mean ± standard deviation or median (Q₁–Q₃). Enumeration data were expressed as a number of cases, rate (%). Independent Student's t-test and Mann–Whitney U test were used to compare means and medians between two groups; differences of means and medians among three groups were compared using analysis of variance and Kruskal–Wallis H test, respectively. The categorical variables among two or three groups were compared using chi-squared tests or Fisher's statistics. A two-sided P < 0.05 was considered statistically significant.

Results

Study population characteristics

No significant differences in gender, HbA1c, and BMI were observed between patients who were qualified for the study (n = 5687) and those sequenced (n = 2429). However, an older onset age was found in individuals who had a genetic testing [Supplementary Table 1]. The clinical features of patients sequenced in our study with all indicators are summarized in Table 1; samples with all indicators had a higher proportion of FHD when compared with those with missing indicators [Supplementary Table 2]. The mean age at diagnosis was 36.45 ± 6.88 years, 1460 (67.1%) were >35 years, and 743 (34.2%) had an FHD.

Table 1.

Clinical features of subjects sequenced by panel test.

Variables	Response rate (N = 2429) (%)	Patients with all indicators (n = 2176)
Male, n (%)	100.0	1582 (72.7)
Duration (years)	100.0	0.03 (0.01–0.23)
Age_diag (years)	100.0	36.45 ± 6.88
15.00–25.00		180 (8.3)
25.01–35.00		536 (24.6)
35.01–45.00		1460 (67.1)
HbA1c (%)	93.2	9.64 ± 2.71
≤7.5		581 (26.7)
>7.5		1595 (73.3)
BMI (kg/m2)	94.2	25.41 ± 3.97
≤23.9		837 (38.5)
24.0–27.9		835 (38.4)
≥28.0		504 (23.2)
FPG (mmol/L)	95.3	9.51 ± 3.78
FCP (pmol/L)	91.3	556.1 (346.6–821.0)
PCP (pmol/L)	89.8	1349.9 (826.0–2101.2)
TG (mmol/L)	92.2	1.90 (1.24–3.15)
TC (mmol/L)	93.1	9.51 ± 3.78
LDL-C (mmol/L)	92.7	2.89 ± 0.99
HDL-C (mmol/L)	91.6	1.11 ± 0.35
FHD∗	100.0	743 (34.1)
FDR		554 (25.5)
SDR		209 (9.6)
Current smoking	100.0	723 (33.2)
Current drinking	100.0	430 (19.8)

Data are expressed as mean ± standard deviation, median (Q₁–Q₃) or n (%).

^∗20 patients with all indicators had both FDR and SDR.

Age_diag: Age at diagnosis; BMI: Body mass index; FCP: Fasting C-peptide; FDR: First-degree relative; FHD: Family history of diabetes; FPG: Fasting plasma glucose; HbA1c: Hemoglobin A1c; HDL-C: High-density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol; PCP: Postprandial C-peptide; SDR: Second-degree relative; TC: Total cholesterol; TG: Triacylglyceride.

Prevalence of MODY in young adults with T2DM

We identified 15 different variants in known MODY genes classified as LP or P in 18 participants, and the prevalence of MODY in young adults with phenotypic T2DM was 0.74% (18/2429, 95% confidence interval [CI]: 0.40–1.08%). Due to a significant difference in age at diagnosis between patients qualified and sequenced, we adjusted this covariate and found that the age-standardized prevalence was also 0.74%. Among all participants with MODY, 33.3% (6/18) had variants in HNF1A, 27.8% (5/18) in GCK, 16.7% (3/18) in HNF4A, 11.1% (2/18) in insulin (INS) gene, 5.6% (1/18) in pancreatic and duodenal homeobox 1 (PDX1) gene, and 5.6% (1/18) in paired homeobox 4 (PAX4) gene. Crucially, the specific diagnosis in 77.8% of MODY (HNF1A, GCK, and HNF4A) could guide their clinical management [Table 2].

Table 2.

Characteristics of the 18 patients identified as MODY.

ID	Gene	Mutation	Protein effect	Age_diag (years)	Gender	BMI (kg/m2)	FCP (pmol/L)	PCP (pmol/L)	HbA1c (%)	FPG (mmol/L)	MetS	FHD	Treatment	MODY PPV (%)	Previous studies
1546	HNF4A	c.268delA	p.K90Rfs∗36	34.00	Female	30.47	319.7	1078.9	7.5	8.58	Yes	No	Insulin	6.4	Novel
2037	HNF4A	c.T706A	p.S236T	37.90	Male	18.07	422.9	1175.5	7.2	13.90	No	No	Insulin	–	Novel
90	HNF4A	c.892 + 2T > C	–	43.60	Male	17.15	–	–	–	–	No	Father	Insulin	–	Novel
219	GCK	c.483 + 1G > C	–	24.40	Female	22.27	346.3	1258.7	6.9	5.80	–	No	Insulin	62.4	Novel
34	GCK	c.C683T	p.T228M	37.40	Male	20.76	–	–	7.2	7.30	No	Mother	Diet	–	[39]
12	GCK	c.G1340A	p.R447Q	21.80	Female	20.05	576.6	1246.5	5.8	7.20	No	No	Diet	75.5	[40]
196	GCK	c.G1345A	p.A449T	41.90	Male	29.38	459.5	2404.3	6.7	6.61	Yes	No	Diet	–	[41]
245	GCK	c.G1386A	p.M462I	44.10	Female	24.03	–	–	–	–	Yes	Mother	Diet	–	[42]
1078	HNF1A	c.C29T	p.T10M	39.50	Male	21.95	592.7	1638.4	8.1	10.60	No	Father	Diet	–	[43]
1175	HNF1A	c.C29T	p.T10M	41.20	Female	27.59	359.6	652.7	7.6	8.73	Yes	Mother	OHA	–	[43]
1826	HNF1A	c.C29T	p.T10M	36.50	Male	21.45	246.6	303.3	9.8	10.53	No	Grandma	Diet	–	[43]
1227	HNF1A	c.C1135A	p.P379T	42.00	Male	24.09	751.0	1950.0	8.5	8.40	No	Brother	Insulin	–	[44]
981	HNF1A	c.1624-2A > T	–	32.30	Male	–	923.2	4249.6	5.3	6.11	Yes	No	OHA	–	[45]
1140	HNF1A	c.1624-2A > T	–	43.00	Female	28.62	436.2	1595.1	8.4	13.35	Yes	No	Diet	–	[45]
96	PDX1	c.495delC	p.L166Yfs∗18	42.65	Male	30.80	749.0	1480.0	10.0	12.10	Yes	No	Insulin	–	Novel
1193	PAX4	c.772-1G > A	–	42.10	Male	24.21	843.2	1336.5	12.4	10.92	Yes	No	Insulin	–	[46]
2245	INS	c.C37A	p.L13M	43.60	Male	24.09	462.5	1414.8	11.9	6.13	No	No	OHA	–	Novel
2524	INS	c.C155T	p.P52L	15.30	Male	19.37	366.6	866.6	9.1	7.70	No	No	Insulin	75.5	Novel

BMI: Body mass index; FCP: Fasting C-peptide; FHD: Family history of diabetes; FPG: Fasting plasma glucose; HbA1c: Hemoglobin A1c; MetS: Metabolic syndrome; MODY: Maturity-onset diabetes of the young; OHA: Oral hypoglycemic agent; PCP: Postprandial C-peptide; PPV: Positive predictive value; –: Not applicable.

When subjects were divided into several subgroups according to gender, age at diagnosis, DKA status, BMI, MetS status, FCP, HbA1c, or FHD, the prevalence of MODY ranged from 0 in patients whose FCP ≤200 pmol/L to 1.5% in young adults diagnosed between 15 and 25 years [Figure 2]. When multiple parameters were combined, the prevalence of MODY could reach 4.76% (3/63) in individuals aged 15 to 25 years, whose BMI <24 kg/m² and FCP >200 pmol/L.

Figure 2.

Forest plot for prevalence and 95% CI of MODY. Age_diag: Age at diagnosis; BMI: Body mass index; CI: Confidence interval; DKA: Diabetic ketoacidosis; FCP: Fasting C-peptide; FHD: Family history of diabetes; HbA1c: Hemoglobin A1c; MetS: Metabolic syndrome; MODY: Maturity-onset diabetes of the young; PPV: Positive predictive value.

Clinical features of MODY

The mean age at diagnosis and BMI of MODY individuals were 36.86 ± 8.40 years and 23.79 ± 4.29 kg/m², with a median FCP level of 459.5 (Q₁–Q₃: 359.6–749.0) pmol/L. Surprisingly, only 16.7% (3/18) of MODY patients were <30 years of age, only 38.9% (7/18) had an FHD, and 47.1% (8/17) of MODY patients had MetS [Table 2].

When compared with patients carrying P/LP variants, except for higher HbA1c, no significant difference in clinical characteristics was seen irrespective of being a carrier of non-P/LP variants (non-MODY) or VUS variants [Supplementary Tables 3 and 4]. We divided subjects into P/LP, VUS, and others, and we found no statistical difference between these subgroups [Supplementary Table 5].

The diagnostic value of tools to help identify MODY

The MODY clinical risk score was built in our study and the full score was 5; combining five parameters reported to assist the diagnosis of MODY. The sensitivity and specificity were 100.0% and 5.6%, 57.1% and 43.7%, 35.7% and 85.1% when MODY clinical risk score was no less than 2, 3, and 4, respectively [Supplementary Figure 1].

As for the Clinical Diagnostic Criteria “3-1-2”, widely used in clinical practice, the criteria included an autosomal-dominant pattern of inheritance in at least three generations, diabetes onset ≤25 years in at least one patient in a family, and insulin independence within 2 years after diabetes onset. However, none of the 18 cases met the above three diagnostic criteria.

As a prediction model proposed by the UK Exeter diabetes group, only 672 cases could use the MODY probability calculator, including four MODY patients, whose positive predictive values (PPV) ranged from 6.4% to 75.5%. At a PPV threshold of 62.4%, the prevalence of MODY in patients whose PPV ≥ 62.4% was significantly higher than that in subjects with PPV < 62.4% (4.41% vs. 0.17%, P = 0.004) [Figure 2].

Discussion

Our data showed that 18 of 2429 (0.74%) carried a P or an LP MODY variant in phenotypic T2DM in young Chinese adults aged 15 to 45 years, nearly 80% of whom would have personalized management. In addition, the role of clinical parameters in diagnosing MODY was limited. To our knowledge, this is the first study to investigate the prevalence of MODY in young adults with phenotypic T2DM. The participating hospitals in this study were across all of the seven geographic regions of China and were considered representative of various populations.

The data analysis showed that individuals sequenced by panel test were older, but the prevalence rates were similar before and after adjustment of age at diagnosis. Globally, the prevalence of MODY would be lower than that reported in many previous studies, which might be associated with the relatively old age at diagnosis (36.45 ± 6.88 years) of our subjects sequenced. Nevertheless, the number of MODY cases misdiagnosed as T2DM was considerable due to a large population base. As per the data from the Sixth National Population Census in 2010,^[21,22] we can deduce that at least 76,000 MODY cases were misdiagnosed as T2DM, 60,000 of which could receive precision management, by treating with low doses of sulfonylureas or only diets. The prevalence of MODY in patients aged >35 years was 0.80%, which was nearly half lower than that in the 15 to 25 years age group. However, it accounted for the largest amount of MODY, with up to 52,500 cases, due to the relatively high prevalence of T2DM, reminding that MODY in T2DM diagnosed between the age group of 35 and 45 years also needs attention, and not just pediatrics. As the prevalence of diabetes diagnosed by the World Health Organization criteria in China increased from 9.7% in 2010 to 11.2% in 2020, and reaching 12.8% when estimated by the ADA criteria, it is concerning that the number of MODY might be largely underestimated.^[23,24]

Compared with other studies focusing on children, the lower prevalence of MODY could partly be attributed to the population with a broad range of characteristics, especially the age at diagnosis. Multiple studies investigating the prevalence were concentrated on pediatric patients, owing to that, the age period of youth is an important feature of MODY. In Poland, Fendler et al^[25] found that 3.1% to 4.2% of children with patients had monogenic diabetes, which were mostly MODY. Among Italian patients <18 years, the prevalence of monogenic diabetes was 4.9%, ranking second as the cause of diabetes in the Italian youth.^[26] The SEARCH for Diabetes in Youth study identified a prevalence of at least 1.2% in the pediatric population with diabetes.^[27] In a Korean study, among 40 MODY patients or early onset T2DM, only 10% had known MODY gene defects.^[28] Recently, MODY was identified in 2.8% of patients ≤20 years with clinically diagnosed T2DM in a Progress in Diabetes Genetics in Youth Collaboration.^[29] A prevalence of 6.5% of MODY was found in a large pediatric population in Southern Italy, newly diagnosed with diabetes; in this study, GCK-MODY was the most common type of MODY.^[30] However, young adults should also have been paid attention to, as a population with a relatively high proportion of MODY. Thanabalasingham et al^[31] sequenced HNF1A and HNF4A in 80 patients with clinically labeled as T2DM, diagnosed ≤30 years and/or diagnosed ≤45 years without MetS; and surprisingly, 15% (12/80) carried HNF1A or HNF4A mutations. The considerable number of MODY patients misdiagnosed with T2DM aged 35 to 45 years showed in our study also proved the need for conducting related research in young adults.

Considering the low prevalence of MODY and expensive cost of implementing genetic testing across the entire population, it was crucial to identify suspected MODY cases, who needed to be tested based on manifestations or biomarkers.^[32] Unfortunately, our data showed no significant difference in the parameters of MODY and non-MODY, except for HbA1c, as the HbA1c in MODY was lower than that in non-MODY. This difference is attributable to the patients carrying GCK mutations, which accounts for around one third of MODY cases and having mild elevated HbA1c.^[33] Furthermore, not only clinical risk score but the common clinical criteria “3-1-2” did not show satisfactory performance in discriminating MODY, as well. These results are not difficult to explain. Considering BMI or MetS and FHD as examples, previous studies demonstrated that BMI or the proportion of MetS was lower in MODY patients than that in those with T2DM.^[20,34] However, considering the rising prevalence of obesity, the usefulness of BMI or MetS as a distinguishing feature between MODY and T2DM is minimized in all populations.^[29] As for FHD, a powerful discriminated parameter for MODY, underestimation of self-report, de novo variants and a high proportion of FHD in T2DM, would limit its role in identifying MODY from T2DM. Furthermore, epidemiological studies have demonstrated that de novo mutations of the major MODY genes could be more frequent than what was previously assumed,^[35] Stanik et al^[36] described 11 de novo mutations of GCK, HNF1A, and HNF4A in 150 probands fulfilled all MODY criteria. In addition, potential biomarkers such as high-sensitivity C-reactive protein and urine C-peptide creatinine ratio have shown promising results in specifically separating the most common types of MODY from other subtypes.^[11] However, different cutoffs with the absence of uniformity had an influence on translating them into clinical practice.^[37] As a prediction model for facilitating the nomination of potential probands, among those who should have a genetic test, the positive capacity of the MODY probability calculator was supported in this study. Nevertheless, only 672 of 2429 individuals met the criteria for utilization, suggesting that these use conditions limited the application in patients with atypical MODY, such as age at diagnosis >35 years.

Our study had several strengths. First, being a large and multi-center study, the recruited subjects were newly diagnosed. Second, the antibody assay was performed in a single well-equipped laboratory, which ensured high-quality measurement of the antibody. Last but not least, all MODY genes that have been reported so far were tested, which can help us to learn the distribution of subtypes. The study also had some limitations. First, most of the subjects sequenced were male; however, there was no statistical difference when compared with all T2DM aged from 15 to 45 years. This might be caused by the higher prevalence of diabetes in male populations of this age range.^[24,38] Second, panel sequencing was focused on single-nucleotide variants in exons, which might miss potential non-exonic or splicing causal variants. Third, we only used GADA, and not five antibodies to evaluate diabetes-specific autoimmunity. This might result in a small amount of latent autoimmune diabetes of adults not being excluded in our study. Finally, no longitudinal data could be obtained.

In this study, we defined wider criteria to select subjects for genetic testing compared with current recommendations. This highlights that even with this selection, the minimum prevalence of MODY still reached 0.74%. Regrettably, no clinical characteristics could reliably distinguish MODY from T2DM in young adults. It was prohibitively expensive and almost impossible to carry out MODY genetic testing across the entire population with diabetes. Further research is warranted to find or create an ideal criterion for identifying individuals who should undergo genetic testing.

Acknowledgment

The authors thank all the patients, nurses, doctors, investigators, and technicians involved at the 46 participating centers of the National Clinical Research Center for Metabolic Diseases for their efforts in data and sample collection. The members of National Clinical Research Center for Metabolic Diseases (investigators and hospitals); Linong Ji, Xueyao Han, Ling Chen, Xiaoling Chen, Peking University People's Hospital; Lixin Guo, Xiaofan Jia, Shan Ding, Beijing Hospital; Xinhua Xiao, Cuijuan Qi, Xiaojing Wang, Peking Union Medical College Hospital; Zhongyan Shan, Yaxin Lai, Zhuo Zhang, The First Hospital of China Medical University; Yu Liu, Yan Cheng, Hanqing Cai, The Second Hospital of Jilin University; Yadong Sun, Yan Ma, Haiying Wang, People's Hospital of Jilin Province; Yiming Li, Chaoyun Zhang, Shuo Zhang, Hua Shan Hospital, Fudan University; Tao Yang, Hao Dai, Mei Zhang, The First Affiliated Hospital with Nanjing Medical University; Liyong Yang, Peiwen Wu, Xiaofang Yan, The First Affiliated Hospital of Fujian Medical University; Yangang Wang, Fang Wang, Hong Chen, The Affiliated Hospital of Qingdao University; Qifu Li, Rong Li, The First Affiliated Hospital of Chongqing Medical University; Qiuhe Ji, Li Wang, Xiangyang Liu, Xijing Hospital, Fourth Military Medical University; Jing Liu, Suhong Wei, Gansu Province People's Hospital; Yun Zhu, Rui Ma, The First Affiliated Hospital of Xinjiang Medical University; Gebo Wen, Xinhua Xiao, Jianping Qin, The First Affiliated Hospital of University of South China; Jian Kuang, Yan Lin, Guangdong General Hospital; Shaoda Lin, Kun Lin, the First Affiliated Hospital of Shantou University Medical College; Xiaohong Niu, Li Li, Heji Hospital Affiliated to Changzhi Medical College; Gan Huang, Shuoming Luo, The Second Xiangya Hospital of Central South University; Huibiao Quan, Leweihua Lin, Hainan General Hospital; Hongyu Kuang, Weihua Wu, The First Affiliated Hospital of Harbin Medical University; Yuling He, The First Affiliated Hospital of Guangxi Medical University; Xiaoyan Chen, Yuyu Tan, The First Affiliated Hospital of Guangzhou Medical University; Ling He, Guangzhou First People's Hospital; Chao Zheng, The Second Affiliated Hospital of Wenzhou Medical University; Jianying Liu, Zhifang Yang, The First Affiliated Hospital of Nanchang University; Xiaoyang Lai, The Second Affiliated Hospital of Nanchang University; Ling Hu, Yan Zhu, Ying Hu, The Third Affiliated Hospital of Nanchang University; Xuqing Li, Henan Provincial People's Hospital; Hong Li, Yushan Xu, The First Affiliated Hospital of Kunming Medical University; Heng Su, Yang Ou, The First People's Hospital of Yunnan Province; Jianping Wang, The Second Hospital University of South China; Changqing Luo, Xiaoyue Wang, The First People's Hospital of Yueyang; Zhiming Deng, Shenglian Gan, The First People's Hospital of Changde City; Zhaohui Mo, Ping Jin, Honghui He, The Third Xiangya Hospital of Central South University; Qiuxia Huang, Dongguan People's Hospital; Fang Wang, Heping Hospital Affiliated to Changzhi Medical College; Yi Zhang, Zhenzhen Hong, First Hospital of Quanzhou Affiliated to Fujian Medical University; Yuezhong Ren, Pengfei Shan, The Second Affiliated Hospital of Zhejiang University School of Medicine; Caifeng Yan, Hui Zhang, Northern Jiangsu People's Hospital; Zhiwen Liu, Shanghai Xuhui District Central Hospital; Meibiao Zhang, The First People's Hospital of Huaihua; Ming Liu, Heting Wang, Tianjin Medical University General Hospital; Hongwei Jiang, Liujun Fu, The First Affiliated Hospital of the Henan University of Science and Technology; Hui Fang, Tangshan Gongren Hospital; Hui Sun, The Affiliated Hospital of Inner Mongolia Medical University.

We thank Yue Liu, from Central South University, for her help in knowing MODY; Dr. Beverley Shields and Dr. Kevin Colclough, from the University of Exeter Medical School, for their kind help in giving the excel version of the MODY probability calculator, and for providing suggestions on the pathogenicity assessment of a few MODY variants; Xiaohan Tang, from Central South University, for her assistance in the calculation of age-standardized prevalence.

Funding

This study was supported by grants from the Science and Technology Innovation Program of Hunan Province (No. 2020RC4044) and the National Science and Technology Infrastructure Program (No. 2013BAI09B12).

Conflicts of interest

None.