Precision diabetes: learning from monogenic diabetes
- First Online:
- 3.3k Downloads
The precision medicine approach of tailoring treatment to the individual characteristics of each patient or subgroup has been a great success in monogenic diabetes subtypes, MODY and neonatal diabetes. This review examines what has led to the success of a precision medicine approach in monogenic diabetes (precision diabetes) and outlines possible implications for type 2 diabetes. For monogenic diabetes, the molecular genetics can define discrete aetiological subtypes that have profound implications on diabetes treatment and can predict future development of associated clinical features, allowing early preventative or supportive treatment. In contrast, type 2 diabetes has overlapping polygenic susceptibility and underlying aetiologies, making it difficult to define discrete clinical subtypes with a dramatic implication for treatment. The implementation of precision medicine in neonatal diabetes was simple and rapid as it was based on single clinical criteria (diagnosed <6 months of age). In contrast, in MODY it was more complex and slow because of the lack of single criteria to identify patients, but it was greatly assisted by the development of a diagnostic probability calculator and associated smartphone app. Experience in monogenic diabetes suggests that successful adoption of a precision diabetes approach in type 2 diabetes will require simple, quick, easily accessible stratification that is based on a combination of routine clinical data, rather than relying on newer technologies. Analysing existing clinical data from routine clinical practice and trials may provide early success for precision medicine in type 2 diabetes.
KeywordsGCK HNF1A HNF4A KCNJ11 Maturity onset diabetes of the young MODY Monogenic diabetes Neonatal diabetes Precision diabetes Precision medicine Review Type 2 diabetes
Immunodysregulation polyendocrinopathy enteropathy X-linked
Permanent neonatal diabetes
Transient neonatal diabetes
Precision medicine refers to the tailoring of medical treatment to the individual characteristics of each patient or subpopulation . Precision diabetes is when a precision medicine approach is used to improve treatment of patients with diabetes. This review aims to examine how precision diabetes has been successfully applied in monogenic diabetes and to ask what this can teach us about the challenges of implementing a precision diabetes approach in type 2 diabetes. For a detailed discussion of precision diabetes for type 2 diabetes, please see review by Mark McCarthy in this issue of Diabetologia .
Overview of the precision medicine approach in monogenic diabetes
Overview of neonatal diabetes
Genetic causes of neonatal diabetes
Neonatal diabetes is defined as diabetes developed before 6 months of age. The knowledge that neonatal diabetes has a monogenic aetiology is based on two strong strands of evidence; first, patients with permanent diabetes diagnosed <6 months of age do not have an increased type 1 diabetes genetic susceptibility. This is in contrast with the high susceptibility seen when those diagnosed >6 months [3, 4]. Second, 96% of patients with known ‘neonatal’ monogenic genetic aetiology are diagnosed with diabetes < 6 months [5, 6]. Before genetic definition was possible, neonatal diabetes was classified solely on the clinical course of disease as transient neonatal diabetes (TNDM), permanent neonatal diabetes (PNDM), or by the specific syndrome when associated with other features e.g. Wolcott–Rallison syndrome or immunodysregulation polyendocrinopathy enteropathy X-linked (IPEX) syndrome. We now know of 23 different genetic causes of neonatal diabetes [7, 8].
Genetic diagnosis: impact on diabetes treatment
Genetic diagnosis: impact on clinical course
Another key benefit of precision medicine is the ability to explain additional clinical abnormalities that are associated with the underlying genetic cause (Fig. 1). These may be already present (e.g. cardiac defects in patients with GATA6 mutations, microencephaly in patients with IER3IP1 mutations and gall bladder and gut atresia in patients with RFX6 mutations), anticipated (e.g. exocrine pancreas deficiency in patients with mutations in GATA4, GATA6 or PDX1, and remission of transient diabetes in patients with 6q24 methylation abnormalities) or they may develop later (e.g. hepatic failure and bone abnormalities in Wolcott–Rallison syndrome or other autoimmune conditions with IPEX syndrome) .
Early comprehensive genetic testing: a paradigm shift for managing neonatal diabetes
The development in targeted next-generation DNA sequencing has allowed rapid and comprehensive testing of all known genetic aetiology in monogenic diabetes . In parallel, referral time from development of diabetes to genetic testing reduced from 4 years to 7 weeks between 2004 and 2013 . These two factors have led to a paradigm shift in the way that we manage neonatal diabetes as we can now make a rapid and precise genetic diagnosis before the development of all clinical features (Fig. 1). This can lead to early appropriate treatment of the diabetes and future planning for other clinical developments . For example, early diagnosis of TNDM allows remission to be predicted and planned for, and knowing that developmental delay is a feature of the genetic aetiology allows early developmental assessment and support. Furthermore, early treatment with sulfonylureas in KCNJ11- and ABCC8-neonatal diabetes probably results in less severe developmental delay . In addition, early treatment with thiamine in thiamine-responsive megaloblastic anaemia (TRMA) neonatal diabetes can improve glycaemic control . Finally, for IPEX and other severe monogenic autoimmune syndromes, early diagnosis allows consideration for early curative bone marrow transplantation before patients are too sick .
Overview of MODY
Genetic causes of MODY
MODY was originally defined as a clinical subgroup of familial diabetes that was diagnosed early (typically before 25 years of age) but despite this, this condition was not insulin dependent and showed autosomal dominant inheritance [15, 16]. The initial linkage analysis in large families led to discovery of the first MODY gene, encoding glucokinase (GCK) [17, 18]. This was rapidly followed by discovery of genes encoding hepatic nuclear factor 1 alpha (HNF1A) , hepatic nuclear factor 4 alpha (HNF4A)  and hepatic nuclear factor 1 beta (HNF1B) . Although other genetic causes have subsequently been described, none of these are as common as the four genetic causes initially described . MODY represents between 1.2% and 3.0% of diabetes diagnosed in children, at least in predominately white European populations ([11, 12] and reviewed in ).
Discrete clinical features of common MODY subtypes
Differential treatment response in MODY subtypes
Probably the most important clinical feature associated with precision diabetes in MODY patients has been the differential treatment response in discrete genetic groups (Fig. 2). GCK-MODY patients do not require treatment [24, 25] and do not respond to either oral agents or low-dose insulin [24, 26]. In contrast, HNF1A- and HNF4A-MODY patients can be treated with low-dose sulfonylureas [6, 27, 28]. Patients who require additional treatment can have dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonist and insulin in addition to sulfonylureas. Patients with HNF1B-MODY require insulin treatment as the response to sulfonylureas and other oral medication is limited .
Increasing genetic diagnosis of MODY throughout the world
MODY genetic testing is increasing throughout the world and most developed countries have at least one academic, health service or commercial laboratory providing monogenic diabetes testing. Within the UK, the Exeter laboratory have gone from ∼50 patients being diagnosed with MODY in 1996 to ∼5000 diagnoses in 2016.
Why has precision medicine in monogenic diabetes worked well?
Subgroups that are clearly defined by underlying aetiology
Large differences in treatment response in monogenic diabetes
Differences in treatment response can have a large impact in monogenic diabetes. The best example is the enhanced sensitivity to sulfonylureas in HNF1A-MODY, meaning that patients may become severely hypoglycaemic if standard doses are used, and that discontinuing sulfonylureas results in a marked deterioration in blood glucose (a 5% point reduction [31 mmol/mol] in HbA1c) [6, 34]. In a randomised trial sulfonylureas led to a fourfold greater reduction of fasting blood glucose in HNF1A-MODY patients compared with age, BMI and blood glucose level-matched type 2 diabetes patients . This sensitivity to sulfonylureas was initially identified from clinical observation and was not predicted from gene function . In contrast, there is a lack of glycaemic response with oral hypoglycaemic agents or low-dose insulin in patients with GCK-MODY . The lack of efficacy of insulin treatment at a median dose of 0.4 U kg−1 day−1 is also seen in pregnancy as the birthweight of offspring of GCK-MODY patients treated with insulin and without insulin are similar . Insulin is still recommended for individuals with GCK-MODY in some circumstances in pregnancy but even at very high doses, its ability to lower the mother’s blood glucose levels is limited (reviewed in ). The lack of response to therapy may be predicted because GCK-MODY patients have a regulated blood glucose that is set at a higher level, as a result of insulin and counter-regulatory hormones being regulated to maintain this elevated glucose level [24, 36].
High-dose sulfonylurea treatment in potassium channel-linked neonatal diabetes (ABCC8- and KCNJ11-neonatal diabetes) had a massive impact on endogenous insulin secretion (measured by C-peptide), which increased from an undetectable level to the level necessary to maintain glucose at near normal values . This resulted in an ∼2 percentage point (22 mmol/mol) improvement in HbA1c in the short term, which persisted for more than 5 years [9, 37]. Importantly, the basis for attempting this therapy arose from the knowledge that the potassium channel is the target of sulfonylureas.
The difficulties of bringing a precision diabetes approach into monogenic diabetes care
Neonatal diabetes: a success story of rapid implementation
The easy clinical recognition of neonatal diabetes combined with a dramatic treatment response following precise genetic diagnosis led to international guidelines being changed within 2 years after gene discovery . The simple clinical guidance was issued that diagnostic genetic testing is required for all patients who developed diabetes before 6 months of age. The simplicity of this guidance greatly helped towards its rapid dissemination worldwide. This was further helped by support from the Wellcome Trust, allowing the Exeter Molecular Genetics Laboratory to offer free rapid comprehensive genetic testing throughout the world for patients with neonatal diabetes until at least 2020. This has resulted in over 1700 patients from 87 countries being tested for neonatal diabetes .
MODY: slow uptake into clinical practice
A solution to diagnosing MODY when there is no single criterion or threshold
Diagnosing MODY requires a complex multi-dimensional assessment of probability based on more than one clinical criterion. This may be difficult for clinicians but can be easily done by use of a statistical calculator that uses multiple, but readily available, clinical information to assess the probability of a patient having MODY. The ‘MODY Probability Calculator’ was developed by B. Shields and is available without charge at www.diabetesgenes.org and on the ‘Diabetes Diagnostic’ app for iOS and Android mobile platforms . In a head-to-head competition, it proved to be as good as clinical experts with more than two decades of experience working with MODY (B. Shields, [University of Exeter Medical School, Exeter, UK] and A. T. Hattersley, personal communication). The probability calculator works best for patients who are not insulin treated. For patients who are insulin treated in whom the diagnosis of MODY is being considered, additional non-genetic tests (islet autoantibody testing and C-peptide analysis) should be considered as ‘rule-out tests’; the presence of islet autoantibodies and/or C-peptide <200 pmol/l effectively rules out MODY [44, 45].
The development of a MODY probability calculator has proved a very promising first step towards precision diabetes. The mobile app is widely used (>6000 downloads to date). This provides a good example of how sophisticated modelling of a complex diagnostic challenge can be simplified into a simple tool that uses readily available clinical information. This approach can greatly help rapid dissemination of precision diabetes.
New technology does not make clinical selection redundant
Next-generation sequencing has transformed our ability to perform genetic testing but it has not removed the need for clinical selection of patients with possible monogenic diabetes for the genetic test. It is now possible to test all genes involved in monogenic diabetes in a single gene panel test, both quickly and efficiently [7, 12]. This gene panel testing approach identifies approximately an additional 25% of monogenic patients with less common causes compared with selected testing of common genetic subtypes . It removes the need to define the likely genetic aetiology/subgroups prior to testing; however, extra care is needed when patients are not selected on phenotype as the prior likelihood of monogenic diabetes is greatly reduced and so false positive findings become more likely . The easy access to sequencing technology has led to laboratories (including commercial laboratories) offering diagnostic testing for monogenic diabetes even when they do not have experience of monogenic diabetes. This has resulted in benign polymorphisms frequently being reported as disease-causing mutations (e.g. 38% of reported cases of HNF1A-MODY in Germany had benign polymorphisms ). Clinical selection of patients with potential monogenic diabetes is still required since, even with improved technology, next-generation sequencing cannot find a genetic aetiology in patients who have type 1 or type 2 diabetes.
Precision medicine in type 2 diabetes: comparisons with monogenic diabetes
Difficulty in defining aetiological subgroups in type 2 diabetes
Defining subgroups using molecular genetic testing in type 2 diabetes is very unlikely to result in discrete aetiological subgroups because the genetic predisposition is polygenic rather than monogenic and the clinical phenotype reflects environmental as well as genetic influences (Fig. 3) . We can define aetiological subgroups based on physiological features, such as insulin resistance and beta cell failure. The main problems of using these categories are that these features change over time, there is a lack of agreement on optimum methods of assessment  and biochemical assays used in the definitions are not standardised between laboratories [50, 51]. Similar problems are seen in latent autoimmune diabetes in adults (LADA), in which there is a lack of agreement of which islet autoantibodies to study, variation in the assays used to measure a specific antibody and varying thresholds for a positive test . The difficulty in defining subgroups in type 2 diabetes has a major impact on the ability to optimise treatment.
The lack of marked differences in treatment response in type 2 diabetes
In type 2 diabetes, it is unlikely that differences in treatment will be as marked as in monogenic diabetes. On average, most glucose lowering therapies for type 2 diabetes reduce HbA1c by about 1% (11 mmol/mol) . It is known that there is considerable variation in treatment response to glucose lowering therapy in type 2 diabetes but, to date, there has been no description of any subgroups that respond with a dramatic 5 percentage point (31 mmol/mol) change in HbA1c, as observed in individuals with HNF1A-MODY. Pharmacogenetic impacts on treatment responses in type 2 diabetes exist but all have been small to date (<0.5% [5 mmol/mol] HbA1c) . An alternative approach may involve defining type 2 diabetes patients who are unlikely to respond to a specific therapy, with the best example to date being insulin-treated type 2 patients with islet autoantibodies or low C-peptide who do not respond to GLP-1 receptor agonists .
An alternative approach for precision medicine in type 2 diabetes
Precision diabetes in monogenic diabetes has been an easy early win but, in contrast, its implementation in type 2 diabetes will be considerably more difficult. Monogenic diabetes has the advantage that there are discrete subgroups that are easily defined by molecular genetics. Frequently, the knowledge of the biology that results from the aetiological gene being identified has helped to define likely treatment response. These early successes have coloured our approach to precision diabetes, favouring genomics-based approaches that search for a single biomarker or a genetic variant with very large effect on treatment response. However, type 2 diabetes is a polygenic condition in which environment, as well as genetic predisposition, play a big role. In this case, an approach concentrating on newer technologies may not be optimum and certainly examining simple clinical criteria like BMI, sex and age should be carried out before rushing to molecular technologies. Finally, one thing that we have learnt from monogenic diabetes, particularly MODY, is that even when there is a clear case, both clinically and economically, for a precision diabetes approach, implementation may be difficult.
We would like to thank our many colleagues within the Exeter Monogenic Team, both past and present, and also our numerous collaborators in the UK and internationally. We accept that many of the ideas expressed in this article have come from collective discussion over many years. Some parts of this article were developed from A. T. Hattersley’s EASD/Novo Nordisk Foundation Diabetes Prize for Excellence 2016.
This work is supported by the MASTERMIND Consortium sponsored by the Medical Research Council (MRC; MR-K005707-1) and by a Wellcome Trust Senior Investigator award given to ATH (and S. Ellard, University of Exeter Medical School, Exeter, UK [WT098395/Z/12/Z]). The work is also supported by the National Institute for Health Research (NIHR) Clinical Research Facility.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
Both authors were responsible for drafting the article and revising it critically for important intellectual content and approved the version to be published.
- 1.National Research Council (US) Committee on A Framework for Developing a New Taxonomy of Disease (2011) Toward precision medicine: building a knowledge network for biomedical research and a new taxonomy of disease. National Academies Press, Washington, DC, USAGoogle Scholar
- 41.Pihoker C, Gilliam LK, Ellard S et al (2013) Prevalence, characteristics and clinical diagnosis of maturity onset diabetes of the young due to mutations in HNF1A, HNF4A, and glucokinase: results from the SEARCH for Diabetes in Youth. J Clin Endocrinol Metab 98:4055–4062CrossRefPubMedPubMedCentralGoogle Scholar
- 45.Besser REJ, Shepherd MH, McDonald TJ et al (2011) Urinary C-peptide creatinine ratio is a practical outpatient tool for identifying hepatocyte nuclear factor 1-/hepatocyte nuclear factor 4- maturity-onset diabetes of the young from long-duration type 1 diabetes. Diabetes Care 34:286–291CrossRefPubMedPubMedCentralGoogle Scholar
- 46.Flannick J, Beer NL, Bick AG et al (2013) Assessing the phenotypic effects in the general population of rare variants in genes for a dominant Mendelian form of diabetes. Nat Publ Group 45:1380–1385Google Scholar
- 54.Dawed AY, Zhou K, Pearson ER (2016) Pharmacogenetics in type 2 diabetes: influence on response to oral hypoglycemic agents. Pharm Pers Med 9:17–29Google Scholar
- 57.Shields BM, Longergan M, Dennis J et al (2015) Patient characteristics are associated with treatment response to second line glucose lowering therapy: a MASTERMIND study abstracts of 51st EASD annual meeting. Diabetologia 58(Suppl 1):S405Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.