Original article

Precision medicine for monogenic diabetes: from a survey to the development of a next-generation diagnostic panel

Sakina Kherra, Jean-Louis Blouin, Federico Santoni, Valerie Schwitzgebel

DOI: https://doi.org/10.4414/smw.2017.14535
Publication Date: 08.11.2017
Swiss Med Wkly. 2017;147:w14535

Monogenic diabetes (MD) accounts for 1–2% of all diabetes cases. Because of its wide phenotypic spectrum, MD is often misdiagnosed as type 1 or type 2 diabetes. While clinical and biochemical parameters can suggest MD, a definitive diagnosis requires genetic analysis. We conducted a survey among clinicians specialising in diabetes to document the cases with MD. Of 74 clinically suspected MD patients, 46% had undergone genetic analysis, which was mostly conducted using Sanger’s classical sequencing method. The most common recorded mutations were located in the GCK gene, followed by the mitochondrial genome (m.3243A>G mutation) and the HNF1B and HNF1A genes. The remaining 54% of patients only had a clinical diagnosis, mostly because genetic analysis was not easily accessible. Here, we designed a new diagnostic panel of 42 genes that was developed based on the survey. The panel was validated with an independent sample of nine known MD patients. Our survey confirms the need for a comprehensive analytical instrument for the diagnosis of MD, which will be met by the proposed panel. The diagnosis of MD is crucial because it dictates treatment and may improve metabolic control and reduce long-term complications as proposed by precision medicine.

Keywords: next-generation sequencing, pancreas, personalised medicine, diabetes, neonatal diabetes, precision medicine, genetic diabetes, autoimmune, type 1 diabetes, type 2 diabetes, monogenic diabetes

Introduction

Monogenic diabetes (MD), in contrast to polygenic type 1 and type 2 diabetes, is due to a single gene defect and has traditionally been referred to as maturity onset diabetes of the young (MODY). MODY has been defined as an autosomal dominant, non-insulin dependent form of diabetes that occurs before the age of 25 due to an underlying defect in beta cells [1]. At least 13 genes have now been discovered to cause MODY [2]. In neonatal diabetes, an additional form of MD, the genetic cause is now identified in over 85% of cases and involves over 20 genes [3].

Given that the clinical features of MD are often non-specific, more than 80% of MD cases remain undiagnosed or are misdiagnosed as type 1 or type 2 diabetes [4, 5]. Precision medicine through genetic analyses leads to the correct diabetes classification, which permits tailoring of treatment regimens and optimisation of health outcomes.

To estimate the need for such a tool, we conducted a survey among centres and private practices specialising in diabetes treatment to document how many patients had been diagnosed with MD, either clinically or genetically. Total diabetes prevalence in Switzerland is assessed at 6.5% [6]. Overall, MD is estimated to account for 1–2% of all diabetes cases. In countries with a widespread screening policy, such as the UK, the minimal prevalence of the most frequent MODY subtypes was 108 cases per 1 million inhabitants [4]. In the paediatric diabetes population, the prevalence of MD was 2.5% in the UK [7] and 1.1% in the Norwegian childhood diabetes registry [8].

We developed and validated a diagnostic tool using next-generation sequencing (NGS) technology to identify the genetic defect underlying suspected cases efficiently. This technology was not yet widely available.

Our study intends to improve the tools for clinicians to make a precise diagnosis of MD since treatment options may depend on the specific gene defect, ranging from diet-only treatment to oral anti-diabetic agents and the need for insulin replacement [9].

Methods

Questionnaire

We conducted a survey by sending a questionnaire to the members of the Swiss Society of Endocrinology and Diabetes (SGED/SSED) (n = 219) and to the members of the Swiss Society of Paediatric Endocrinology and Diabetes (SPGED/SSEDP) (n = 39) to collect anonymous data on diabetic subjects with either a clinical suspicion of MD or genetically confirmed MD. Subjects with mitochondrial diabetes were also included in our study. We requested the following data in the questionnaire: age at diagnosis; method of diagnosis of MD (clinical or genetic analysis); family history of diabetes; ethnic origin; birth weight; weight loss and body mass index (BMI) at diagnosis; glycosylated haemoglobin (HbA1c), glycaemia, C-peptide, ketosis and treatment at diagnosis; autoimmune anti-diabetes antibodies; lipid profile; liver enzymes; other health problems or congenital malformations; age, BMI, HbA1c, microvascular or macrovascular complications; and treatment at the last medical visit.

The data of the participants were then entered into the MODY probability calculator to establish a risk score for MD and to compare the results with the available genetic analyses [10].

No approval from the ethics committee was needed for this study. Informed consent was obtained from each patient for the use of anonymised DNA for the development of the NGS panel.

Diagnostic tool: Haloplex technology

This custom assay, designed based on liquid-phase capture (Haloplex HS, Agilent, Santa Clara, CA, USA), allows for the trapping of all coding regions of the 42 genes (CCDS exon reference sequence) and splicing regions (±10 nucleotides apart from each intron-exon junction). The sequence of the selected DNA fragments from each patient is then resolved with a next-generation sequencer (PGM, Ion Torrent, ThermoFisher, USA; or NextSEQ500, Illumina, La Jolla, USA). The raw sequencing data obtained are analysed through a series of processes that include the alignment of the readings (BWA), variant calling (SAMTOOL), and variant annotation (ANNOVAR). The last two steps of the process are automated using a locally developed bioinformatic pipeline (Pytline). The pipeline tracks all existing information to classify the variant (e.g., the frequency in the general population, report in a mutation database, and bioinformatic pathogenicity prediction (including PolyPhen2, SIFT, MutationTester). Each variant that is ultimately considered pathogenic is then validated by another sequencing method (Sanger) and reported to a physician. We did not incorporate mitochondrial genes in this panel; the request for mitochondrial DNA analysis will have to be done separately.

Results

Survey

We received a total of 74 answers corresponding to 74 subjects from hospitals, medical centres and private practices in different regions. The geographical locations are depicted in figure S1 (appendix 1). Among the subjects, there were 44 females and 30 males with a median age at diabetes onset of 24.5 years (range 0.03–49). Two subjects had neonatal diabetes (<6 months of age), 8 had childhood diabetes (≥6 months to <11 years), and 14 had adolescent diabetes (≥11 to <18 years). The remaining 50 patients were adults, and 19 (25.7%) were 35 years or older at diabetes onset. The clinical characteristics of the patient cohort are depicted in table 1.

Table 1

Clinical characteristics of the subjects at diabetes onset.

 All subjects
Number of patients74
Age at onset years (range)24.5 (0.03–49)
HbA1c mmol/mol (range) / % (range)63.9 ± 33.7 / 8.0 ± 3.08
 Subjects (n)Fraction (%)
Female gender44/7459.5
Age at onset years (range)  
     Neonatal onset <6 months22.7
     Childhood onset ≥6 months to <11 years810.8
     Adolescent onset ≥11 to <18 years1418.9
     Young adult onset ≥18 to <35 years3141.9
     Adult onset ≥35 years1925.7
Subjects with ketones5/4211.9
Family screening11/4226.2
Weight loss10/4223.8
Incidental diagnosis21/4250
Clinical diagnosis of MD40/7454
Genetic analysis34/7446

HbA1c = glycosylated haemoglobin; MD = monogenic diabetes
Data are presented as the mean (range) or ± SD. Number of counts: n.
Data on ketones, family screening, weight loss and incidental diagnosis were available for only 42 patients.

Mode of diabetes diagnosis

Diabetes was an incidental finding in 50% of the participants and was diagnosed during a familial screening in 26.2% of the subjects and after weight loss in 23.8% of the subjects. In 32 subjects (43.2%), the discovery mode was not reported (table 1). Forty-seven subjects (63.5%) had been tested for at least one diabetes autoantibody, and three subjects were positive (tables 2a and 2b). As an indicator of the persistence of endogenous insulin secretion, we used C-peptide levels (>200 pmol/l), which were analysed in 21 subjects (28.4%). Most values were measured fasting (in 86%), but no patient had a glucose level <4.7 mmol/l (range 4.7–16.2 mmol/l). All but one subject presented with levels >200 pmol/l. Five subjects presented with ketosis at the time of diagnosis.

Table 2a

Subject characteristics according to genotype.

 Age at diabetes diagnosis (years)HbA1c at diagnosis
(mmol/mol)
(%)
BMI at last visit
(kg/m2)
Family Hx
(%)
(n)
Auto-antibodies
positive/
tested
Treatment at last visitComplications
GCK25.6
(3–45)
46.9 ± 4.0
6.4 ± 0.4
22.7 ± 3.46
(15–24.7)
85.7
(12/14)
1/105 OHA (3 Glinides,1 Met,
1 SU+Met)
2 Insulin
7 Diet only
1 N (*)
HNF1B15.6
(10–22)
57.3 ± 17.6
7.4 ± 1.6
21.15 ± 4.4
(18.1–27.7)
60
(3/5)
0/41 OHA (SU)
4 insulin
1 N
HNF1A12.3
(10–14)
46.5 ± 6.4
6.4 ± 0.6
23.6 ± 7.7
(16–32)
100
(3/3)
0/21 OHA (Met) + Insulin
1 insulin
1 OHA (SU)
1 R
HNF4A2448
6.5
18.28100
(1/1)
1/11 OHA (SU+Met)0
KCNJ118.1
(0.03–16)
NA22.9 ± 0.7
(22.2–23.6)
100
(2/2)
0/11 OHA (SU)
1 Insulin
1 R + N + P
PDX10.03NA14100
(1/1)
0/1Insulin +
Exocrine enzymes
0

HbA1c = glycosylated haemoglobin; BMI = body mass index; NA = not available; Mt D = mitochondrial diabetes; MD = clinically diagnosed monogenic diabetes; Family Hx = family history; OHA = oral hypoglycaemic agent; Met = metformin; SU = Sulfonylurea; Pio = pioglitazone: DPP-4 = dipeptidyl peptidase-4. Microvascular complications: R = retinopathy; N = nephropathy; N* = microalbuminuria; P = polyneuropathy.
Data are presented as the mean (range) ± SD.
Since information for diabetic antibodies and for the family history was missing in several patients, we reported the data obtained for the total number of patients.
In the GCK group, 1 subject had positive IA-2 autoantibodies. In the HNF4A group, 1 subject had positive GAD autoantibodies.

Table 2b

Subject characteristics for mitochondrial diabetes and clinical diagnosis of MD.

 Age at diabetes diagnosis (years)HbA1c at diagnosis
(mmol/mol)
(%)
BMI at last visit
(kg/m2)
Family Hx
(%)
(n)
Auto-antibodies
positive/ tested
Treatment at last visitComplications
Mt D36.6
(29–34)
92.4 ± 50.5
10.6 ± 4.6
22.36
(16–24.8)
100
(8/8)
0/51 OHA (SU+ Pio+ DPP-4)
5 Insulin
2 OHA (Met) + Insulin
2 N
1 P
1 R + N
1 R + P
Clinical MD25.1
(10–48)
65.2 ± 34.7
8.1 ± 3.2
23.01 ± 4.2
(15.5–31)
73.7
(28/38)
1/2314 OHA
14 Insulin
1 OHA + Insulin
10 Diet only
1 NA
2 R
3 N
1 R + P
3 R + N + P

NA = not available; Mt D = mitochondrial diabetes; clinical MD = clinically diagnosed monogenic diabetes; Family Hx = family history; OHA = oral hypoglycaemic agent; Met = metformin; SU = sulfonylurea; Pio = pioglitazone: DPP-4 = dipeptidyl peptidase-4. Microvascular complications: R = retinopathy; N = nephropathy; P = polyneuropathy.
Data are presented as the mean (range) ± SD.
Since information for diabetic antibodies and for the family history was missing in several patients, we reported the data obtained for the total number of patients. One subject with a clinical diagnosis had positive GAD and IA-2 antibodies.

Genetic results

Overall, 34 participants (46%) of the survey already had a genetically confirmed diagnosis of MD, all obtained by classical Sanger sequencing. GCK gene mutations were the most frequent, followed by mutations in the HNF1A and HNF1B, HNF4A, KCNJ11 and PDX1 genes. 8 of the 34 subjects (23.5%) suffered from mitochondrial diabetes (fig. 1). The clinical characteristics of the different diabetes subtypes are listed in tables 2a and 2b. In the other 40 subjects (54%), the diagnosis was based only on clinical features and biochemical criteria, without genetic analysis (table 2b).

fullscreen
Figure 1
Types and proportions of mutations in surveyed patients with a genetically confirmed diagnosis of monogenic diabetes.
The genetic analysis was performed by Sanger sequencing in the 34 patients.

Diabetes treatment

Among the patients with identified genetic mutations, 29.4% were treated with oral hypoglycaemic agents only (OHA), 38.2% received insulin only, 8.8% were prescribed a combination therapy with insulin plus OHA, 2.9% had insulin plus exocrine enzymes, and 20.6% were on a diet only (tables 2a and 2b). In the GCK diabetes group, two subjects were treated with insulin injections, and five were treated with OHA. Insulin therapy was prescribed in two of the subjects with HNF1A diabetes and in all but one subject with HNF1B diabetes. One patient with a KCNJ11 mutation was put on insulin at diagnosis and was switched to oral treatment with sulfonylurea after obtaining the genetic results. The patient with pancreatic agenesis caused by a homozygous PDX1 mutation required exocrine pancreatic enzymes in addition to insulin [11].

All but one subject with mitochondrial diabetes were treated with insulin, and 2 patients received additional treatment with OHA (table 2b).

Overall, 35.9% of the clinically diagnosed subjects were managed with insulin only, 35.9% were treated with OHA, and 2.6% received a combination of both. 25.6% were on a diet only (table 2b). For one patient, the information was missing.

Diabetes complications

Microvascular complications were found in 15.4% (4 out of 26) of MODY patients, in 62.5% (5 out of 8 subjects) of patients with mitochondrial diabetes, and in 22.5% (9 out of 40) with a clinical suspicion of MD (tables 2a and 2b).

The likelihood of MD estimated by the MODY probability calculator

There is then the question of which selection criteria should be used for the genetic analysis. Recently, an algorithm called the MODY probability calculator has been proposed to estimate the likelihood of MD for subjects with diabetes onset before the age of 35 years [10]. The authors recommend the use of a positive predictive value of >20% as an indicator for MODY testing (http://www.diabetesgenes.org/content/mody-probability-calculator). All the genetically confirmed MODY subjects in this study showed a positive predictive value of >20%, except for 7 subjects in whom we did not perform the calculation because diabetes was diagnosed after 35 years of age (table 3).

Table 3

Positive predictive values calculated by the MODY probability calculator in subjects with a genetic diagnosis of MODY.

Gene defectPP >20%
<35 years (n)
PP ≤20%
<35 years (n)
GCK70
HNF1B50
HNF1A30
HNF4A10
KCNJ1110
Total170

PP = positive predictive value
Data are number of patients (n).
The probability for MODY was calculated for each patient with genetically confirmed MODY diabetes using the MODY probability calculator. The cut-off value for the positive predictive of 20% was used as an indicator for genetic testing (http://www.diabetesgenes.org/content/mody-probability-calculator). The calculator was developed for subjects with diabetes onset <35 years of age, which is why the results are depicted according to the age of diabetes onset.

To further characterise the clinically diagnosed MD group, we used the MODY probability calculator to determine the positive predictive value for MD in each subject, except for 5 subjects whose detailed information was missing (table S1 in appendix 1.). In the clinically diagnosed group younger than 35 years of age, 73.3% (22/30) had a positive predictive value >20%, and 26.7% (8/30) a positive predictive value <20%. In our survey, 19 subjects (25.7%) developed diabetes at a later age. We therefore reported the results according to the age of diabetes onset to include all subjects. For the patients of 35 years or older at diagnosis, an upper age limit of 35 years was put into the MODY probability calculator to obtain the positive predictive value. In the older age group, 40% (2/5) had a positive predictive value >20%, and 60% (3/5) a positive predictive value <20% (table S1).

Mitochondrial diabetes was diagnosed after 35 years of age in 50% of the patients. For the mitochondrial diabetes group, diagnosed younger than 35 years of age, 50% (2/4) had a positive predictive value >20%, but of those diagnosed at 35 years or older, only 25% (1/4) had a positive predictive value >20% (table S1).

Developing a novel NGS panel to diagnose MD

To date, more than 40 genes that cause MD have been identified, and every year new genes are discovered [12]. Our goal was to create an innovative diagnostic instrument that takes advantage of the power of NGS to offer a rapid and comprehensive analysis of patients with a suspected form of MD. In our first gene panel, we included 42 genes that have been reported to cause diabetes (table 4). We included all known MODY genes at that time, genes that cause neonatal diabetes, and genes that cause monogenic autoimmune and syndromic diabetes. Known enhancer regions and introns associated with diabetes were also included in the panel [3, 13, 14].

Table 4

Gene panel for the diagnosis of monogenic diabetes.

Gene nameRefSeq accession number
(GenBank)
Chromosome locationTheoretical coverage
(%)
Missing nucleotides
(n)
Non-covered (%)
HNF4ANM_000457Chr.20100
GCKNM_000162Chr.7100
HNF1ANM_000545Chr.12100
PDX1NM_000209Chr.13100
HNF1BNM_000458Chr.17100
NEUROD1NM_002500Chr.2100
KLF11NM_003597Chr.2100
CELNM_001807Chr.9100
PAX4NM_006193Chr.7100
INSNM_000207Chr.11100
BLKNM_001715Chr.8100
ABCC8NM_000352Chr.11100
KCNJ11NM_000525Chr.11100
SLC19A2NM_006996Chr.1100
DNAJC3NM_006260Chr.13100
PLAGL1NM_001080954Chr.6100
GATA6NM_005257Chr.18100
GATA4NM_002052Chr.8100
SLC2A2NM_000340Chr.3100
NKX2-2NM_002509Chr.20100
NEUROG3NM_020999Chr.10100
GLIS3NM_152629Chr.999.790.30%
RFX6NM_173560Chr.6100
MNX1NM_005515Chr.7100
EIF2AK3NM_004836Chr.2100
WFS1NM_006005Chr.4100
IER3IP1NM_016097Chr.18100
PAX6NM_000280Chr.11100
FOXP3NM_014009Chr. X100
STAT3NM_139276Chr.17100
PCBD1NM_000281Chr.10100
SIRT1NM_012238Chr.10100
LRBANM_001199282Chr.499.9820.02%
ZPF57NM_001109809Chr.6100
PTF1A enhancerhg19Chr.1096.6253.40%
INS intronhg19Chr.11100
PPP1R15BNM_032833Chr.1100
TMRT10ANM_152292Chr.4100
KMT2DNM_003482Chr.1298.87~2001.13%
KDM6ANM_021140Chr. X100
RAP1ANM_001010935Chr.1100
RAP1BNM_015646Chr.12100
CISD2NM_001008388Chr.4100
PTF1ANM_178161Chr.10100

Custom-designed gene panel with 42 diabetes genes and known enhancer regions and introns with coverage of 99.89% of the targets. Chromosome: Chr.

To offer the most robust sequencing tool in a clinical setting, we favoured the approach of custom-designed NGS restricted to 42 genes rather than performing whole exome sequencing using a commercial catalogue design that often harbours uncaptured regions. We optimised the efficiency of the probe design with the help of the capture kit provider to ultimately obtain a set of probes that covered 99.89% of the targets (table 4).

The validation of this assay was done in a blind manner as proposed by the national guidelines. Independent DNA samples, which were previously analysed by Sanger sequencing, were tested from nine patients with defects in six different genes, and all anomalies were properly identified. A mutation in the KCNJ11 gene required post-hoc analysis, followed by an adjustment of the pipeline for deletions. No mutations were identified in the negative control DNA. All 465 genomic regions corresponding to the 42 genes were thoroughly covered at an average of 500-fold in our setting (nine patient samples were loaded on a 316 chip, Ion Torrent PGM). In addition to identifying point variants (missense and nonsense), deletions or insertions/duplications of up to a few nucleotides, this assay also allows for the detection of larger deletions of one or even several exons as shown in four patients with large deletions (table 5). Another advantage of this approach is that the mutation search can be restricted to a subset of genes based on the clinical phenotype. If no mutation is identified, the investigation can be extended to more or even all 42 genes, since the raw sequencing data are securely conserved and can be reopened at any time for analysis.

We now propose the following selection criteria for genetic screening depicted in figure 2. Too strict criteria can miss a large proportion of people with MD [3437]. The counts in figure 2 reflect the number of patients from the survey with a genetically confirmed diagnosis of MD. The proposed flowchart is feasible with the documented cases of monogenic diabetes in this survey. An advanced genetic analysis will also contribute to the elucidation of even more complex forms of diabetes due to digenic or oligogenic defects. The knowledge gained will lead to novel drug development for specific mutations, further refining precision medicine in diabetes.

fullscreen
Figure 2
An updated pathway for clinical decision-making for monogenic diabetes screening.
The MODY probability calculator will use clinical features from the three different groups with either presumed MODY or type 1 or type 2 diabetes to calculate the probability for monogenic diabetes and therefore the indication for genetic testing. The calculator was developed for people with diabetes onset <35 years of age and should be used accordingly [10]. A direct molecular analysis is indicated for cases of neonatal and syndromic diabetes.
*For presumed autoantibody negative type 1 diabetes cases, an additional indicator for genetic screening are persisting C-peptide levels after the honeymoon period of >200 pmol/l with glucose >8 mmol/l, to avoid suppression of C-peptide levels by hypoglycemia [3133].
We show the numbers of the genetically confirmed diabetes cases from the survey (total number 34) in the different categories.
Type 1 diabetes: T1D; type 2 diabetes: T2D. Number of patients: N

Table 5

DNA used for the validation of the diagnostic gene panel.

PatientGene nameReference sequenceGene defectProtein effectPathogenicity classified according Richards **
1HNF4ANM_000457.4c.724G>Ap.Val242Metpathogenic
2GCKNM_000162.3c.608T>Cp.Val203Alapathogenic
3HNF1ANM_000545.6c.166G>Tp.Glu56*pathogenic
4HNF1BNM_000458.3c.1- ?_*+ ?delp. ?pathogenic
5KCNJ11NM_000525.3c.96_96delinsCTGp.Gln30fspathogenic
6EIF2AK3NM_004836.6c.2707C>Tp.Arg903*pathogenic
7Neg. control-No pathogenic variantp.?-
8HNF1ANM_000545.6c.327-?_526+?delp. ?pathogenic
9HNF1ANM_000545.6c.1- ?_*+ ?delp. ?pathogenic
10HNF1BNM_000458.3c.1- ?_1045+ ?delp. ?pathogenic

**Richards et al. [15]. We used the following analyses for the assessment of pathogenicity, 1. exonic silent variants, if not located in the first or in the last codon of an exon, were discarded; 2. all missense variants were evaluated according to their frequency in the general population (absent or very rare in the databases ExAC and gnomAD); 3. the pathogenic prediction was evaluated by different bioinformatics tools (SIFT, PolyPhen-2 and MutationTester); 4. the status regarding the pathogenicity according to ClinVar was sought; 5. the conservation score according to GERP was considered; 6. the literature was checked to ascertain whether the identified variants had been reported.
Deletions/Insertions: delins. Protein sequence: p. Coding DNA sequence: c.

Discussion

In Switzerland, as in most countries, MD remains under-diagnosed due to its clinical heterogeneity and the lack of comprehensive genetic analysis. In this survey, the classical Sanger method was the only method used for the genetic testing of MD. Traditionally, genetic testing for MD has focused on a few genes depending on the patient’s phenotype, but in our new NGS-based diagnostic tool, multiple genes (n = 42) are sequenced in parallel. Novel genes that are involved in the pathogenesis of MD will be incorporated into subsequent designs. Such methods have already been proposed by several research groups in the UK, Poland, France, Norway and the USA [7, 1621].

Over the last several years, many countries in Europe and across the globe have launched a new concept for the diagnosis and treatment of rare diseases (also called orphan diseases, http://www.orpha.net) and personalised medicine. Since MD belongs to this category, approval from health insurance for genetic testing will hopefully be obtained more easily. In Switzerland, a specific form is available in “documents” on the webpage of the Swiss Society of Medical Genetics (SGMG). The request for the genetic analysis for monogenic diabetes is available at the following website: http://www.hug-ge.ch/sites/interhug/files/structures/gr-demande-analyse/diagmol-std_e.pdf. The costs of the NGS analysis are based on the different national billing guidelines and the cost-effectiveness of testing for MD will improve as genetic testing becomes rapidly cheaper [2224]. For the time being, careful selection of patients is essential. Studies evaluating diagnostic strategies for MD are on-going [25].

In this survey, we identified 34 subjects with a genetically confirmed diagnosis of MD and 40 subjects with potential MD, but this number (74) represents only a fraction of the estimated number of subjects with MD. Furthermore, even when clinicians have identified subjects with a high clinical suspicion of MD, a genetic analysis was not conducted in 54% of the cases. These patients miss out on treatment optimisation with potentially increased metabolic control and decreased long-term complications and family counselling [26]. So far, the most commonly recognised mutations are located in the GCK gene, followed by mitochondrial and HNF1A and HNF1B mutations. HNF1B diabetes seems to be overrepresented in this study in comparison to internationally published data, where only 1–2% present with this form of diabetes [27]. The difference could be explained by the clinically easily recognisable renal phenotype, which may explain the increasing requests for genetic analysis for this diabetes subtype. Furthermore, the low number of reported cases may be leading to an overestimation in the results. Many subjects in the group with a genetic confirmation of MD were not treated according to international guidelines, and many patients still receive unnecessary treatment [26]. Patients with GCK mutations do not require pharmacological treatment, but 14.3% were receiving insulin treatment, and an additional 35.7% were getting OHA [28]. In the early course of HNF1A and HNF4A diabetes, glinides or low doses of sulfonylureas are more appropriate than insulin therapy [9, 29]. Most patients were given insulin in addition to OHA, and sulfonylureas were offered after molecular diagnosis in only a few cases. Mitochondrial diabetes usually requires insulin treatment, which was administered to 87.5% of the patients.

In clinical practice, the MODY probability calculator represents a useful tool for the selection of patients who should undergo genetic testing. The use of the MODY calculator in our study was very helpful for the majority of patients since all genetically confirmed MODY diabetes cases had a positive predictive value >20%. However, this method does not allow for a distinction between the different forms of MD. In the clinically diagnosed MD group, the positive predictive value was >20% in 73.3% of the subjects younger than 35 years of age, suggesting that genetic testing would be indicated. This probability calculator has not been developed for mitochondrial diabetes or patients older than 35 years at diabetes onset and should not be used. Increasing the cut-off of the positive predictive value could increase pick-up rate and increase cost-effectiveness. Another useful parameter for discriminating between type 1 and MODY diabetes is the urinary C-peptide/creatinine ratio (≥0.2 nmol/mmol), which has not yet been used [30].

1 Fajans SS, Bell GI, Polonsky KS. Molecular mechanisms and clinical pathophysiology of maturity-onset diabetes of the young. N Engl J Med. 2001;345(13):971–80. doi:. http://dx.doi.org/10.1056/NEJMra002168 PubMed

2 Schwitzgebel VM. Many faces of monogenic diabetes. J Diabetes Investig. 2014;5(2):121–33. doi:. http://dx.doi.org/10.1111/jdi.12197 PubMed

3 De Franco E, Flanagan SE, Houghton JA, Allen HL, Mackay DJ, Temple IK, et al.The effect of early, comprehensive genomic testing on clinical care in neonatal diabetes: an international cohort study. Lancet. 2015;386(9997):957–63. doi:. http://dx.doi.org/10.1016/S0140-6736(15)60098-8 PubMed

4 Shields BM, Hicks S, Shepherd MH, Colclough K, Hattersley AT, Ellard S. Maturity-onset diabetes of the young (MODY): how many cases are we missing?Diabetologia. 2010;53(12):2504–8. doi:. http://dx.doi.org/10.1007/s00125-010-1799-4 PubMed

5 Thomas CC, Philipson LH. Update on diabetes classification. Med Clin North Am. 2015;99(1):1–16. doi:. http://dx.doi.org/10.1016/j.mcna.2014.08.015 PubMed

6 Kaiser A, Vollenweider P, Waeber G, Marques-Vidal P. Prevalence, awareness and treatment of type 2 diabetes mellitus in Switzerland: the CoLaus study. Diabet Med. 2012;29(2):190–7. doi:. http://dx.doi.org/10.1111/j.1464-5491.2011.03422.x PubMed

7 Shepherd M, Shields B, Hammersley S, Hudson M, McDonald TJ, Colclough K, et al.; UNITED Team. Systematic Population Screening, Using Biomarkers and Genetic Testing, Identifies 2.5% of the U.K. Pediatric Diabetes Population With Monogenic Diabetes. Diabetes Care. 2016;39(11):1879–88. doi:. http://dx.doi.org/10.2337/dc16-0645 PubMed

8 Irgens HU, Molnes J, Johansson BB, Ringdal M, Skrivarhaug T, Undlien DE, et al.Prevalence of monogenic diabetes in the population-based Norwegian Childhood Diabetes Registry. Diabetologia. 2013;56(7):1512–9. doi:. http://dx.doi.org/10.1007/s00125-013-2916-y PubMed

9 Pearson ER, Starkey BJ, Powell RJ, Gribble FM, Clark PM, Hattersley AT. Genetic cause of hyperglycaemia and response to treatment in diabetes. Lancet. 2003;362(9392):1275–81. doi:. http://dx.doi.org/10.1016/S0140-6736(03)14571-0 PubMed

10 Shields BM, McDonald TJ, Ellard S, Campbell MJ, Hyde C, Hattersley AT. The development and validation of a clinical prediction model to determine the probability of MODY in patients with young-onset diabetes. Diabetologia. 2012;55(5):1265–72. doi:. http://dx.doi.org/10.1007/s00125-011-2418-8 PubMed

11 Schwitzgebel VM, Mamin A, Brun T, Ritz-Laser B, Zaiko M, Maret A, et al.Agenesis of human pancreas due to decreased half-life of insulin promoter factor 1. J Clin Endocrinol Metab. 2003;88(9):4398–406. doi:. http://dx.doi.org/10.1210/jc.2003-030046 PubMed

12 Stekelenburg CM, Schwitzgebel VM. Genetic Defects of the β-Cell That Cause Diabetes. Endocr Dev. 2016;31:179–202. doi:. http://dx.doi.org/10.1159/000439417 PubMed

13 Weedon MN, Cebola I, Patch AM, Flanagan SE, De Franco E, Caswell R, et al.; International Pancreatic Agenesis Consortium. Recessive mutations in a distal PTF1A enhancer cause isolated pancreatic agenesis. Nat Genet. 2014;46(1):61–4. doi:. http://dx.doi.org/10.1038/ng.2826 PubMed

14 Carmody D, Park SY, Ye H, Perrone ME, Alkorta-Aranburu G, Highland HM, et al.Continued lessons from the INS gene: an intronic mutation causing diabetes through a novel mechanism. J Med Genet. 2015;52(9):612–6. doi:. http://dx.doi.org/10.1136/jmedgenet-2015-103220 PubMed

15 Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, et al.; ACMG Laboratory Quality Assurance Committee. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405–24. doi:. http://dx.doi.org/10.1038/gim.2015.30 PubMed

16 Szopa M, Ludwig-Gałęzowska A, Radkowski P, Skupień J, Zapała B, Płatek T, et al.Genetic testing for monogenic diabetes using targeted next-generation sequencing in patients with maturity-onset diabetes of the young. Pol Arch Med Wewn. 2015;125(11):845–51. doi:. http://dx.doi.org/10.20452/pamw.3164 PubMed

17 Ellard S, Lango Allen H, De Franco E, Flanagan SE, Hysenaj G, Colclough K, et al.Improved genetic testing for monogenic diabetes using targeted next-generation sequencing. Diabetologia. 2013;56(9):1958–63. doi:. http://dx.doi.org/10.1007/s00125-013-2962-5 PubMed

18 Alkorta-Aranburu G, Carmody D, Cheng YW, Nelakuditi V, Ma L, Dickens JT, et al.Phenotypic heterogeneity in monogenic diabetes: the clinical and diagnostic utility of a gene panel-based next-generation sequencing approach. Mol Genet Metab. 2014;113(4):315–20. doi:. http://dx.doi.org/10.1016/j.ymgme.2014.09.007 PubMed

19 Bonnefond A, Durand E, Sand O, De Graeve F, Gallina S, Busiah K, et al.Molecular diagnosis of neonatal diabetes mellitus using next-generation sequencing of the whole exome. PLoS One. 2010;5(10):e13630. doi:. http://dx.doi.org/10.1371/journal.pone.0013630 PubMed

20 Bonnefond A, Philippe J, Durand E, Muller J, Saeed S, Arslan M, et al.Highly sensitive diagnosis of 43 monogenic forms of diabetes or obesity through one-step PCR-based enrichment in combination with next-generation sequencing. Diabetes Care. 2014;37(2):460–7. doi:. http://dx.doi.org/10.2337/dc13-0698 PubMed

21 Johansson BB, Irgens HU, Molnes J, Sztromwasser P, Aukrust I, Juliusson PB, et al.Targeted next-generation sequencing reveals MODY in up to 6.5% of antibody-negative diabetes cases listed in the Norwegian Childhood Diabetes Registry. Diabetologia. 2017;60(4):625–35. PubMed

22 Greeley SAW, John PM, Winn AN, Ornelas J, Lipton RB, Philipson LH, et al.The cost-effectiveness of personalized genetic medicine: the case of genetic testing in neonatal diabetes. Diabetes Care. 2011;34(3):622–7. doi:. http://dx.doi.org/10.2337/dc10-1616 PubMed

23 Naylor RN, John PM, Winn AN, Carmody D, Greeley SA, Philipson LH, et al.Cost-effectiveness of MODY genetic testing: translating genomic advances into practical health applications. Diabetes Care. 2014;37(1):202–9. doi:. http://dx.doi.org/10.2337/dc13-0410 PubMed

24 Schnyder S, Mullis PE, Ellard S, Hattersley AT, Flück CE. Genetic testing for glucokinase mutations in clinically selected patients with MODY: a worthwhile investment. Swiss Med Wkly. 2005;135(23-24):352–6. PubMed

25 Peters JL, Anderson R, Hyde C. Development of an economic evaluation of diagnostic strategies: the case of monogenic diabetes. BMJ Open. 2013;3(5):e002905–10. doi:. http://dx.doi.org/10.1136/bmjopen-2013-002905 PubMed

26 Hattersley AT, Patel KA. Precision diabetes: learning from monogenic diabetes. Diabetologia. 2017;60(5):769-777.

27 Edghill EL, Stals K, Oram RA, Shepherd MH. HNF1B deletions in patients with young‐onset diabetes but no known renal disease. Diabet Med. 2013;30(1):114-7.

28 Steele AM, Shields BM, Wensley KJ, Colclough K, Ellard S, Hattersley AT. Prevalence of vascular complications among patients with glucokinase mutations and prolonged, mild hyperglycemia. JAMA. 2014;311(3):279–86. doi:. http://dx.doi.org/10.1001/jama.2013.283980 PubMed

29 Becker M, Galler A, Raile K. Meglitinide analogues in adolescent patients with HNF1A-MODY (MODY 3). Pediatrics. 2014;133(3):e775–9. doi:. http://dx.doi.org/10.1542/peds.2012-2537 PubMed

30 Besser REJ, Shields BM, Hammersley SE, Colclough K, McDonald TJ, Gray Z, et al.Home urine C-peptide creatinine ratio (UCPCR) testing can identify type 2 and MODY in pediatric diabetes. Pediatr Diabetes. 2013;14(3):181–8. PubMed

31 Ellard S, Bellanné-Chantelot C, Hattersley AT; European Molecular Genetics Quality Network (EMQN) MODY group. Best practice guidelines for the molecular genetic diagnosis of maturity-onset diabetes of the young. Diabetologia. 2008;51(4):546–53. doi:. http://dx.doi.org/10.1007/s00125-008-0942-y PubMed

32 Berger B, Stenström G, Sundkvist G; B. Berger, G. Stenström, G. Sundkvi. Random C-peptide in the classification of diabetes. Scand J Clin Lab Invest. 2000;60(8):687–93. doi:. http://dx.doi.org/10.1080/00365510050216411 PubMed

33 Rubio-Cabezas O, Hattersley AT, Njølstad PR, Mlynarski W, Ellard S, White N, et al.; International Society for Pediatric and Adolescent Diabetes. The diagnosis and management of monogenic diabetes in children and adolescents. Pediatr Diabetes. 2014;15(S20, Suppl 20):47–64. doi:. http://dx.doi.org/10.1111/pedi.12192 PubMed

34 Thanabalasingham G, Pal A, Selwood MP, Dudley C, Fisher K, Bingley PJ, et al.Systematic assessment of etiology in adults with a clinical diagnosis of young-onset type 2 diabetes is a successful strategy for identifying maturity-onset diabetes of the young. Diabetes Care. 2012;35(6):1206–12. doi:. http://dx.doi.org/10.2337/dc11-1243 PubMed

35 Kropff J, Selwood MP, McCarthy MI, Farmer AJ, Owen KR. Prevalence of monogenic diabetes in young adults: a community-based, cross-sectional study in Oxfordshire, UK. Diabetologia. 2011;54(5):1261–3. doi:. http://dx.doi.org/10.1007/s00125-011-2090-z PubMed

36 Gandica RG, Chung WK, Deng L, Goland R, Gallagher MP. Identifying monogenic diabetes in a pediatric cohort with presumed type 1 diabetes. Pediatr Diabetes. 2015;16(3):227–33. doi:. http://dx.doi.org/10.1111/pedi.12150 PubMed

37 Shields B, Colclough K. Towards a systematic nationwide screening strategy for MODY. Diabetologia. 2017;60(4):609–12. doi:. http://dx.doi.org/10.1007/s00125-017-4213-7 PubMed

Appendix 1

Supplemental data

Table S1

Probability for MODY each patient with mitochondrial diabetes and in the group with clinical diagnosis.

 PP >20%
<35 years (n)
PP ≤20%
<35 years (n)
PP >20%
≥35 years (n)
PP ≤20%
≥35 years (n)
Mitochondrial Diabetes2213
Clinical MD22823

PP = positive predictive value
Data are number of patients (n). In the clinical group data, 5 patients could not be analysed because of missing data.
The probability for MODY was calculated for each patient with mitochondrial diabetes and in the group with clinical diagnosis using only the probability calculator. The cut-off value for the positive prediction of 20% was used as an indicator for genetic testing (http://www.diabetesgenes.org/content/mody-probability-calculator). The calculator was developed for subjects with diabetes onset <35 years of age, which is why the results are depicted according to the age of diabetes onset. To get a positive predictive value for the subjects with age of onset ≥35 years of age, we put 35 years of age into the calculator.

fullscreen
Figure S1
Geographical distribution of the medical centres that responded to the survey.
The geographical locations of the twelve medical centres and hospitals responding to the survey are depicted.

Kherra Sakinaa, Blouin Jean-Louisbc, Santoni Federicobc, Schwitzgebel Valerie M.ad

a Paediatric Endocrine and Diabetes Unit, University Hospitals of Geneva, Switzerland

b Department of Genetic Medicine and Development, Faculty of Medicine, University of Geneva, Switzerland

c Department of Genetic Medicine Laboratory and Pathology, University Hospitals of Geneva, Switzerland

d Diabetes Centre of the Faculty of Medicine, University of Geneva, Switzerland

SK and JLB: Both authors contributed equally

We acknowledge all members of SGED and SGPED who participated in the survey.

Our research was supported by the Swiss National Science Foundation (grant no. CR33I3_140655 to VMS). Funding was also provided by the Swiss Society for Endocrinology and Diabetes. SK is the recipient of a clinical fellowship from the European Society for Paediatric Endocrinology (ESPE). The funding agencies had no role in the study design, implementation, data collection and analysis, the decision to publish or the preparation of the manuscript.

No potential conflict of interest relevant to this article was reported.

Valerie M Schwitzgebel, MD, Head of Paediatric Endocrine and Diabetes Unit, Department of child and adolescent health, Children’s University Hospital, 6, Rue Willy Donze, CH-1211 Geneva, Valerie.schwitzgebel[at]unige.ch

next-generation sequencing, pancreas, personalised medicine, diabetes, neonatal diabetes, precision medicine, genetic diabetes, autoimmune, type 1 diabetes, type 2 diabetes, monogenic diabetes