MODY is prevalent in later-onset diabetes, has potential for targeted therapy but is challenging to identify

Abstract

Maturity Onset Diabetes of the Young (MODY) can present after the age of 40 years, but its prevalence, clinical characteristics, and the utility of simple clinical features for selecting cases in this age group remain poorly defined. We analysed whole-exome and clinical data from 51,619 individuals with diabetes diagnosed after age 40 from one UK and one US cohort. The prevalence of MODY due to pathogenic variant in the 10 most common MODY genes was 1 in 191 (0.52%) in the UK cohort and 1 in 633 (0.16%) in the US cohort. For subtypes with treatment implications (GCK, HNF1A, HNF4A, ABCC8, KCNJ11), prevalence was 1 in 234 and 1 in 935, respectively. GCK-MODY was most common, followed by HNF4A and lower-penetrance RFX6-MODY. Clinical features of MODY largely overlapped with non-MODY either insulin-treated from diagnosis or not insulin-treated from diagnosis. Only BMI, HbA1c and HDL were statistically different in MODY from non-MODY in both cohorts (all P<0.0018). Applying strict clinical criteria (BMI<25 and noninsulin treated and parent with diabetes) only increased the MODY diagnosis to 2.64% and 0.87% but missed over 86% of cases. MODY is prevalent in later-onset diabetes, has potential for targeted therapy but is challenging to identify.

Introduction

Current genetic testing for Maturity Onset Diabetes of the Young (MODY) focuses primarily on individuals diagnosed with diabetes before the age of 40 years (1–4). MODY is a familial form of diabetes caused by heterozygous pathogenic variants in one of 11 genes and typically presents before age 30 (1). This focus on early-onset diabetes reflects the higher prevalence of MODY in younger individuals (~4%) and the clear clinical benefits of targeted treatment in this group (1; 5). For example, GCK-MODY usually requires no pharmacological therapy, while HNF1A-, HNF4A-, KCNJ11-, and ABCC8-MODY respond better to sulphonylureas than to insulin (1; 6–8).

It is increasingly recognised that MODY can also present after the age of 40 (9). For example, in a large multigenerational study of HNF1A-MODY, approximately 35% of cases developed diabetes after age 40 (10). Identifying these individuals could potentially provide the same benefits from tailored treatment as in younger-onset cases. In younger individuals, clinical features such as age, BMI, parent diabetes and HbA1c help distinguish MODY from more common forms of diabetes and are used in selecting individuals for genetic testing (11). However, it remains unclear whether clinical features can similarly identify MODY in those diagnosed after age 40.

Robust prevalence data and a better understanding of their clinical features are needed before widespread genomic screening for MODY can be considered in this age group. A previous study by Bansal et al. investigated genetic aetiologies in this age group but included only 2,670 individuals with diabetes diagnosed after age 40 limiting precision of the estimates (12). Larger studies, such as those by Billings et al. (13) and Bonnefond et al. (14), included individuals with diabetes across all ages (n = 14,622 and 25,699, respectively), but did not specifically report on individuals diagnosed after age 40 or provide the clinical features of this group (13; 14). None of these studies assessed whether simple clinical criteria can be used to select individuals for genetic testing in this age group. Therefore, in this study, using whole-exome sequencing and clinical data from 51,619 individuals with diabetes diagnosed after age 40 from two large cohorts, we aimed to determine the prevalence, genetic causes, and clinical features of MODY. We also assessed whether simple clinical criteria could help identify individuals for genetic testing in this age group.

Research Design and Methods

Study Population

We used data from the UK Biobank, a population-based cohort of over 500,000 individuals with, whole-exome sequencing, and extensive phenotypic data (15). The dataset includes baseline demographics, age at diabetes diagnosis, and HbA1c measurements, which allow for robust identification of individuals with diabetes (15). For this study, we included 25,012 individuals with diabetes diagnosed at ≥ 40 years of age. The clinical characteristics of these individuals are detailed in Supplementary Table 1. Ethics approval for the UK Biobank study was obtained from the North West Centre for Research Ethics Committee (11/NW/0382). Written informed consent was obtained from all participants.

We also used data from an independent US health system–based Geisinger MyCode cohort of 173,247 individuals from Pennsylvania (16–18). This cohort is enriched for metabolic and cardiovascular diseases and provides an alternatively ascertained cohort more representative of the clinical setting where clinicians may come across these individuals in routine clinical practice (17; 18). This cohort includes detailed genetic data from whole-exome sequencing, along with deep phenotyping derived from electronic health records. We identified 26,607 individuals with diabetes diagnosed ≥ 40 years of age in our study. Supplementary Table 1 provides their clinical characteristics. This study was approved by the General Institutional Review Board; participants in MyCode have provided broad consent for research use of their exome and EHR data.

In both cohorts, we defined diabetes based on at least one of the following criteria: self-report by participants, presence of an ICD-9 or ICD-10 code for diabetes, use of diabetes medication, or HbA1c ≥48 mmol/mol prior to recruitment (19). Age at diagnosis was defined as the age at the first recorded evidence of diabetes (19). We used self-reported data to identify individuals with a parental history of diabetes. Those with ‘no’ or missing information were considered not to have a parental history of diabetes.

Genetic Data

We used whole-exome sequencing data from the 450,000-participant release of the UK Biobank to identify individuals with MODY. The exome sequencing methodology, including quality control and sample or variant filtering, has been described by Szustakowski et al. (20) and is available at: https://biobank.ctsu.ox.ac.uk/showcase/label.cgi?id=170. Individuals from the Geisinger cohort underwent whole-exome sequencing as part of the DiscovEHR collaboration with the Regeneron Genetics Center. The methods for this process have been reported previously by Mirshahi et al. (10). We reviewed missense and protein-truncating variants in the ten most common MODY genes, which together account for over 99% of autosomal dominant MODY cases: ABCC8, GCK, HNF1A, HNF1B, HNF4A, INS, KCNJ11, NEUROD1, PDX1, and RFX6 (21). We excluded CEL because exome sequencing cannot reliably detect pathogenic variants in this gene (22). We also included HNF1B 17q12 deletions in the study as they contribute to >50% of all HNF1B-MODY cases (23). We identified individuals with HNF1B 17q12 deletions using array data, as described by Cannon et al. (24). We included variants classified as pathogenic or likely pathogenic according to ACMG/AMP guidelines and confirmed their quality through manual review of the sequencing read data (25; 26).

Statistical Analysis

We compared categorical variables using Fisher’s exact test and continuous variables using the Mann–Whitney U test. We calculated exact 95% confidence intervals for proportions and prevalence using the Clopper–Pearson method. We assessed clinical features of individuals with MODY in comparison with two treatment-defined groups: those treated with insulin from diagnosis (defined as starting insulin within one year and remaining on it, with or without adjunctive therapies), and those not treated with insulin at diagnosis. These categories reflect those used in the widely adopted MODY calculator (11; 27). We conducted all analyses using STATA 18 (StataCorp, USA), R, and Python, including the SciPy and statsmodels packages.

Results

MODY is prevalent in individuals with diabetes diagnosed after age 40

Among 25,012 UK Biobank participants with diabetes diagnosed after 40 years of age, the prevalence of MODY was 0.52% (1 in 191; 95% CI, 0.44–0.62%; n = 131) (Figure 1A). In the US health system–based Geisinger MyCode cohort of 26,607 individuals with diabetes diagnosed after 40 years of age, the prevalence was lower at 0.16% (1 in 633; 95% CI, 0.11–0.21%; n = 42) (Figure 1A). The combined prevalence of GCK, HNF1A, HNF4A, and ABCC8-MODY subtypes with important treatment implications was 0.43% in the UK Biobank (1 in 234; 95% CI, 0.35–0.52%; n = 107) and 0.10% in the US cohort (1 in 985; 95% CI, 0.063–0.14%; n = 27) (Figure 1B). In both cohorts, GCK was the most frequent cause, followed by HNF4A and the lower-penetrance gene RFX6 (Figure 1B). All pathogenic variants identified are listed in Supplementary table 2 and 3.

Figure 1. Prevalence and genetic aetiologies of MODY in individuals with diabetes onset at or after 40 years of age.

A) Prevalence of MODY among 25,012 diabetes individuals from UK Biobank population cohort and 26,607 individuals from the US health system–based Geisinger MyCode cohort with diabetes diagnosed at after 40 years of age. Prevalence is shown for all MODY genes (pathogenic variants in 10 MODY genes) and for MODY subtypes with treatment-modifying implications (ABCC8, KCNJ11, GCK, HNF1A, and HNF4A).

B) Genetic aetiology of MODY cases in the UK Biobank and MyCode cohort (UK Biobank, GCK = 77, HNF4A = 14, RFX6 = 12, HNF1A = 11, ABCC8 = 5, Other (PDX1, NEUROD1, HNF1B = 12. MyCode cohort, GCK = 15, HNF4A = 6, RFX6 = 7, HNF1A = 5, ABCC8 = 1, Other (PDX1, HNF1B= 8)

Individuals with MODY were on inappropriate treatment for their genotype

Of the 92 individuals with GCK-MODY from both cohorts, 37 (40.2%) were receiving pharmacological treatment for diabetes (28 non-insulin and 9 insulin). Of the 42 individuals with HNF1A, HNF4A, and ABCC8-MODY 73.8% (n = 31) did not receive sulphonylureas (12 non-sulphonylureas, 8 insulin, 11 no pharmacological treatment).

The clinical features of MODY in individuals diagnosed at and after the age of 40 overlap substantially with those of non-MODY diabetes

To identify features that might guide genetic testing in this age group, we compared individuals with MODY separately against individuals treated with insulin from diagnosis and those not treated with insulin at diagnosis. This approach allowed us to assess which clinical features may support case selection for genetic testing, particularly among insulin-treated individuals, who are likely to benefit most from a genetic diagnosis and targeted therapy. In the UK Biobank, individuals with MODY differed significantly from insulin-treated non-MODY individuals in age at diagnosis, BMI, HbA1c, LDL cholesterol, and parental history of diabetes (Table 1). Among these, HbA1c showed the greatest difference (median 50.6 mmol/mol, IQR 48.5–54.8 vs 59.8, IQR 51.0–69.7; p = 3.6 × 10⁻¹⁵), followed by age at diagnosis (median 54.5 years, IQR 49.4–60.4 vs 50.5, IQR 45.5–57.0; p = 0.00012) (Table 1). We then compared non-insulin-treated at diagnosis individuals to individuals with MODY and found differences in age at diagnosis, BMI, triglycerides, and HDL cholesterol. Of these, BMI and triglycerides showed the most marked differences (median BMI 27.9 kg/m², IQR 25.1–30.2 vs 30.9, IQR 27.8–34.9; p = 6.93×10⁻¹⁴; median triglycerides 1.4 mmol/L, IQR 1.0–2.0 vs 2.0, IQR 1.4–2.8; p = 2.3×10⁻⁹) (Table 1).

Table 1. Clinical characteristics of MODY and Non-MODY diabetes in cases with diabetes onset after 40 years in UK Biobank population cohort.

Clinical Features	MODY	Insulin treated from diagnosis	Not insulin treated from diagnosis	P-Value MODY vs insulin treated from diagnosis	P-Value MODY vs not insulin treated from diagnosis
N	131	1,176	23,705	-	-
Age at recruitment, y	60.2 (55.1-64.9)	61.4 (55.8-65.2)	62.2 (56.8-66.1)	0.36	0.01
Age at diabetes diagnosis, y	54.5 (49.4-60.4)	50.5 (45.5-57)	56.5 (50.5-61.5)	0.00012*	0.0028*
Females	63 (48.1)	430 (36.6)	8,896 (37.5)	0.01	0.01
European Ancestry	121 (92.4)	996 (84.7)	19,990 (84.3)	0.02	0.01
Parent with diabetes	56 (42.7)	335 (28.5)	7,865 (33.2)	0.0012*	0.03
BMI, kg/m2	27.9 (25.1-30.2)	29.4 (26.1-33.7)	30.9 (27.8-34.9)	0.00014*	6.93×10-14*
HbA1c, mmol/mol	50.6 (48.5-54.8)	59.8 (51-69.7)	51.1 (46.6-58.3)	3.6×10-15*	0.76
Triglycerides, mmol/l	1.4 (1-2)	1.5 (1-2.4)	2 (1.4-2.8)	0.45	2.3×10-9*
HDL, mmol/l	1.4 (1-1.6)	1.2 (1-1.5)	1.1 (1-1.3)	0.03	1.9×10-7*
LDL, mmol/l	2.7 (2.3-3.5)	2.5 (2.1-3)	2.7 (2.2-3.3)	0.00039*	0.7
Diabetes Treatment:				3.02×10-148*	0.0032*
Diet/No Treatment	73 (55.7)	0 (0)	10,752 (45.4)
Insulin +/- Other	16 (12.2)	1,176 (100)	1,991 (8.4)
Non-insulin only	42 (32.1)	0(0)	10,962 (46.2)

N(%) for categorical data and median (interquartile range) for continuous data. Insulin treated from diagnosis defined as starting insulin within one year and remaining on insulin, with or without adjunctive therapies. * Below Bonferroni corrected threshold (0.004).

In the US health system–based Geisinger MyCode which is enriched for metabolic disorders, the clinical features of MODY overlapped even more with those of both non-MODY groups. Only BMI, HbA1c, and HDL cholesterol differed significantly (median BMI 31.2 vs 34.9 vs 34.3 kg/m², median HbA1c 55.2 vs 71.6mmol/mol, median HDL 1.3 vs 1.1 vs 1.1mmol/L), and the magnitude of these differences was small (Table 3). Taken together, these findings suggest that although certain clinical features differ statistically between MODY and non-MODY diabetes after age 40, most differences are modest in clinical terms and show substantial overlap.

Table 2. Clinical characteristics of MODY and Non-MODY diabetes in cases with diabetes onset after 40 years in the US health system–based Geisinger MyCode cohort.

Clinical Features	MODY	Insulin treated from diagnosis	Not insulin treated from diagnosis	P-Value MODY vs insulin treated from diagnosis	P-Value MODY vs not insulin treated from diagnosis
N	42	3,656	22,909	-	-
Age at recruitment, y	66.2 (52.8-72.7)	61.3 (54.3-68.6)	64.2 (56.8-71.2)	0.21	0.85
Age at diabetes diagnosis, y	57.9 (45.6-65.2)	56.9 (50.0-64.3)	58.0 (50.8-65.7)	0.79	0.34
Females	18 (42.9)	1,846 (50.5)	11,625 (50.7)	0.0084	0.0089
European Ancestry	42 (100)	3,302 (90.3)	21,730 (94.9)	0.030	0.28
Parent with diabetes	22 (52.4)	1,603 (43.8)	8,668 (37.8)	0.53	0.057
BMI, kg/m2	31.2 (27.4-34.0)	34.9 (29.8-41.1)	34.3(29.9-39.8)	0.0016*	0.0018*
HbA1c, mmol/mol	55.2 (48.6-67.2)	71.6 (57.4-91.3)	49.7 (43.2-59.6)	4.3×10-6*	0.18
Triglycerides, mmol/l	1.7 (0.7-2.1)	1.9 (1.2-2.9)	1.9 (1.3-2.7)	0.22	0.25
HDL, mmol/l	1.3 (1.1-1.4)	1.1 (0.9-1.3)	1.1 (1-1.3)	0.00086*	1.9×10-7*
LDL, mmol/l	3.0 (2.2-3.5)	2.6 (1.9-3.3)	2.7 (2.2-3.4)	0.23	0.63
Diabetes Treatment:				8.51×10-63	0.024
Diet/No Treatment	7 (16.7)	0 (0)	5,861 (25.6)
Insulin +/- Others	13 (31.0)	3,656 (100)	3,488 (15.2)
Non-insulin only	22 (52.4)	0 (0)	13,560 (59.2)

N(%) for categorical data and median (interquartile range) for continuous data. Insulin treated from diagnosis defined as starting insulin within one year and remaining on insulin, with or without adjunctive therapies. * Below Bonferroni corrected threshold (0.004).

Table 3. Clinical selection criteria and their utility at identifying individuals with MODY in individuals with diabetes diagnosed after age 40.

Clinical Criteria	Number of MODY	Total number of individuals	% MODY	Number (%) MODY missed
UK Biobank population cohort
No criteria	131	25,012	0.52	0 (0)
BMI < 30	95	10,777	0.88	36 (27.5)
Not insulin treated	115	21,829	0.53	16 (12.2)
Parent with diabetes	56	8,256	0.68	75 (57.3)
BMI < 30 & not insulin treated & parent with diabetes	36	3,142	1.13	95 (72.5)
BMI < 25 & not insulin treated & parent with diabetes	18	682	2.64	113 (86.3)
US health-system based Geisinger MyCode cohort
No criteria	42	26,607	0.16	0 (0)
BMI < 30	15	4506	0.33	27 (64.3)
Not insulin treated	29	19,450	0.15	13 (31.0)
Parent with diabetes	22	10,293	0.21	20 (47.6)
BMI < 30 & not insulin treated & parent with diabetes	7	1180	0.59	35 (83.3)
BMI < 25 & not insulin treated & parent with diabetes	2	230	0.87	40 (95.2)

In sensitivity subgroup analyses, comparisons of insulin-treated MODY cases with Insulin treated non-MODY groups, and of non–insulin-treated MODY cases with non-insulin treated non-MODY groups, were consistent with the main analysis (Supplementary Tables 4 and 5). This is likely because MODY cases treated with insulin from diagnosis were largely similar to the rest of the MODY cohort (Supplementary Tables 4 and 5). Comparisons of MODY cases with either screen-detected or clinically diagnosed non-MODY diabetes cases showed directionally consistent results in both cohorts (Supplementary Table 6 and 7).

Simple clinical criteria alone are insufficient to identify MODY in individuals diagnosed with diabetes at and after the age of 40

We next evaluated a range of clinical criteria, from broad to more stringent, based on classical MODY definitions, to assess their performance in this age group (Table 3). Applying a BMI threshold of <30 kg/m² increased the observed prevalence from 0.52% to 0.88% (95 out of 10,777) in the UK Biobank but missed 27.5% of individuals with MODY. In the US health system–based Geisinger MyCode cohort, the observed prevalence rose from 0.16% to 0.33%, yet 64.3% of individuals with MODY remained undetected. Adding further criteria of no insulin treatment at recruitment and at least one parent with diabetes with BMI raised the observed prevalence to 1.13% (36 out of 3,142) in the UK Biobank and to 0.59% (7 out of 1,180) in the US health system–based Geisinger MyCode cohort. However, these stricter criteria still failed to detect 72.5% and 83.3% of individuals with MODY, respectively.

Discussion

Our large study of 51,619 individuals with diabetes diagnosed after 40 years of age demonstrates that MODY is relatively common (1 in 191 – 1 in 633) in this age group, has potential for precision therapy but has overlapping clinical features with non-MODY diabetes and it will be difficult to identify the majority of individuals with MODY based on clinical features alone.

MODY prevalence was lower in our US health-system based cohort than in the UK Biobank population cohort. This is likely due to recruitment differences. The UK Biobank has a healthy volunteer bias (15), while the MyCode cohort was enriched for metabolic disease (17; 18), as reflected by a higher average BMI (30.9 vs 33.6kg/m², respectively). These contrasting designs are a strength, enabling prevalence estimates across diverse settings. The true prevalence likely lies between the two estimates (0.16 – 0.52%). Our findings align with Bansal et al., who reported a 0.6% prevalence in this age group (12), and with Gjesing et al. (28), who found similar rates of GCK-MODY in late-onset diabetes (UK Biobank: 0.31% vs Gjesing et al.: 0.8%). Half our cohort was drawn from the UK Biobank, with partial overlap with studies by Billings et al. and Bonnefond et al. (13; 14). Billings et al. analysed 14,622 individuals with diabetes of any age using the 200K exome release from UK Biobank. Bonnefond et al. combined 2,151 individuals with diabetes from the UK Biobank (50K release) with 23,548 from other cohorts (n = 25,699). In contrast, we used the 450K exome release and an independent US health system–based cohort, Geisinger Mycode, focusing specifically on diabetes diagnosed after age 40. This targeted approach offers a largely non-overlapping, age-specific view of MODY in later-onset diabetes.

MODY is less common in individuals diagnosed after age 40 than in those with young-onset diabetes and has distinctly different genetic causes. The prevalence of MODY in early-onset diabetes is approximately 4%, significantly higher than individuals with later-onset diabetes (5). This difference reflects the distinct genetic aetiologies in these groups. For example, variants in the highly penetrant HNF1A gene are the most common cause of progressive diabetes, but it ranked fourth among those diagnosed after age 40 in the UK Biobank and MyCode cohort (21; 29). In contrast to individuals with early-onset MODY, late-onset MODY includes a higher proportion of lower-penetrance genes. RFX6, NEUROD1, and PDX1-MODY together accounted for 15.3% of late-onset MODY in the UK Biobank, compared to only 4.2% in young-onset cases (21). GCK-MODY is an exception, with similar prevalence across age groups. This reflects its distinct pathophysiology. GCK-MODY causes mild, lifelong fasting hyperglycaemia that begins at birth and gradually worsens with age (7; 30; 31). As a result, age at diagnosis reflects when hyperglycaemia is detected rather than when the variant becomes penetrant. Alongside increased diabetes screening in older adults, this explains the consistent frequency of GCK-MODY across age groups. The proportion of GCK-MODY in the UK Biobank cohort was higher than in the Geisinger MyCode cohort (0.31% vs 0.056% prevalence) (58.8% vs 35.7% of MODY cases). This difference is likely explained by differences in recruitment. The UK Biobank has a well-described ‘healthy volunteer’ bias, whereas the Geisinger MyCode cohort is recruited through a health care system, where the prevalence of metabolic disease is higher (17; 18). As GCK-MODY is typically asymptomatic and rarely causes serious complications, it is more likely to be present in a relatively healthy population and under-represented in a health care system cohort. Our estimate from the UK Biobank is higher than the prevalence suggested by Chakera et al.(32) of 0.11% (1.1 in 1000). However, this is based on a relatively small cohort of young pregnant women without diabetes (n=356) whereas our cohorts include older individuals and focus on diabetes cases only.

Our results show substantial overlap in clinical features between MODY and other forms of diabetes diagnosed after age 40, making clinical recognition difficult. Combined with its lower prevalence in this age group, this limits the ability to select individuals with sufficient prior probability for routine genetic testing. For example, in young-onset referrals, clinical suspicion yields a MODY positivity rate of approximately 20% (21). In contrast, even under ideal criteria in individuals with diabetes diagnosed after 40 years of age (BMI < 25, not insulin-treated, and with a parent affected by diabetes), the positivity rate was only 1–2.6%, and 87% of the individuals with MODY would be missed. Adding biomarkers such as islet autoantibodies or C-peptide in insulin-treated patients may improve detection (1). However, the overall low yield, predominance of low-penetrance variants, and limited impact on treatment decisions argue against routine genomic screening for MODY in this age group. Instead, genetic testing can be considered on a case-by-case basis, where it is supported by additional evidence—for example, a younger affected family member or specific clinical features suggesting an underlying aetiology, such as renal cysts in HNF1B-MODY (22).

Our study has limitations. We used whole-exome data, which does not provide coverage of promoter regions, and will therefore miss some pathogenic variants in HNF1A, HNF4A and GCK. However, these variants contribute to <1.7% (24/1420 cases for HNF1A/HNF4A and 6/1441 GCK) (33; 34) of MODY cases in these genes. This suggest that only a small minority of cases are missed by not capturing these variants. It is therefore important to note that our study presents a minimum prevalence estimate for MODY in these populations. Whole-exome sequencing also does not capture the mitochondrial genome and therefore cannot be used to identify mitochondrial variants causing late-onset diabetes. Future studies using whole-genome data, such as our previous work (35), are needed to better define the contribution of mitochondrial variants to late-onset diabetes. We lacked data on type 1 diabetes biomarkers, such as islet autoantibodies and C-peptide. Both factors would have help selecting individuals for MODY testing. Although we included participants of all ethnicities, the cohorts were predominantly of European ancestry, which may limit generalisability. We focused on the ten most common MODY genes and did not assess CEL so our prevalence estimates may be slightly conservative. Rare syndromic or recessive forms of monogenic diabetes were also excluded, though these account for less than 7% of monogenic diabetes cases and are unlikely to substantially alter overall prevalence estimates (21). Finally, we restricted our analysis to genes with definite or limited evidence of causality according to ClinGen (36) and did not include refuted genes.

In summary, MODY is present in individuals diagnosed with diabetes after age 40, but its genetic aetiology differs from that seen in younger-onset cases. The substantial overlap in clinical features with other forms of later-onset diabetes makes implementing routine genomic screening in this group challenging.

Highlights.

Why did we undertake this study?

MODY can present later in life, and diagnosis can enable precision treatment. However, individuals with later-onset diabetes are rarely tested.

What specific question did we ask?

How common is MODY in people diagnosed with diabetes after 40 years, and can they be identified clinically?

What did we find?

MODY affects 1 in 191 to 633 individuals with diabetes onset after 40 years, but clinical features alone cannot reliably identify them.

What are the implications?

MODY is relatively common in later-onset diabetes but difficult to detect clinically, limiting routine genetic testing in this group.

Data and Resource Availability

The UK Biobank data used in this study is freely accessible from the UK Biobank https://www.ukbiobank.ac.uk. Data from the Geisinger MyCode cohort are available within the article and its supplemental information. Additional information is available upon request, subject to data user agreement.