Genetic Risk Score Distinguishes Between Diabetes Types

Miriam E Tucker

November 23, 2015

An investigational genetic risk score for type 1 diabetes can help in determining between types of diabetes in young adults and will help predict whether they will require insulin treatment within 3 years, a new study suggests.

The findings were published online November 17 in Diabetes Care by Richard A Oram, MD, of the University of Exeter Medical School, United Kingdom, and colleagues.

The increase in obesity has made it often difficult to distinguish between type 1 and type 2 diabetes, particularly in young adults aged between 20 and 45.

Misdiagnosis in that age range is common: About 10% are initially thought to have type 2 and then fail on multiple oral treatments because they really have type 1. At the same time, about 10% are inappropriately put straight on insulin when they do not need it because they actually have type 2 diabetes.

And monogenic diabetes — usually maturity-onset diabetes of youth (MODY) — is commonly misdiagnosed as both types.

"The wrong diagnosis leads to the wrong treatment," principal investigator Andrew T Hattersley, MD, professor of molecular medicine at Exeter, told Medscape Medical News.

Currently available diagnostic tests have limitations. Antibody tests are not always specific for type 1 diabetes, and islet-cell autoantibody levels are typically lower in adults than in children and may diminish over time.

And C-peptide levels may not identify type 1 diabetes if the person experiences a brief resurgence of beta-cell function, a so-called "honeymoon" period.

"A big advantage is that the genetic test is unaltered over time.…At present I think this testing will be very helpful when other testing or clinical criteria are opposing or equivocal," Dr Hattersley said. The new test is available now, at a cost of around $75, he noted.

Type 1 Genetic Risk Score: Highly Sensitive, Specific

Using published studies of genetic variants, the investigators derived genetic risk scores for both type 1 and type 2 diabetes. For the type 1 score, single nucleotide polymorphisms (SNPs) from both the HLA- and non-HLA regions were combined for a total of 30 associated risk variants. For the type 2 score, they selected a set of 69 SNPs.

They tested the discriminatory ability of the scores in 1938 people who had received a clinical diagnosis of type 1 diabetes at age less than 17 years and were treated with insulin from the time of diagnosis and in 1914 who had been diagnosed with type 2 diabetes between the ages of 25 and 75 and were glutamic acid decarboxylase (GAD)–autoantibody negative. All were from the UK Wellcome Trust Case Control Consortium.

The type 1 diabetes genetic risk score was highly discriminatory, with an area under the curve (AUC) of 0.88. The mean type 1 diabetes genetic risk score was 0.279 in type 1 diabetes patients vs 0.229 in type 2 diabetes patients (P < .0001).

The type 2 diabetes score was much less discriminatory, with an AUC of 0.64 that did not add significantly to the discriminatory power of the type 1 diabetes score (combined AUC 0.89).

Subsequent analyses focused on the type 1 diabetes genetic risk score.

A type 1 diabetes score greater than 0.280 (50th percentile of type 1 diabetes genetic risk score in the Wellcome Trust type 1 diabetes cohort) was indicative of type 1 diabetes with 95% specificity and 50% sensitivity, while a score of less than 0.234 (fifth percentile of type 1 diabetes score in the Wellcome type 1 diabetes cohort) identified type 2 diabetes with 95% specificity and 53% sensitivity.

Identifying Severe Insulin Deficiency

Next, the investigators assessed the diagnostic accuracy of the type 1 diabetes genetic risk score in identifying severe insulin deficiency (requiring insulin within 3 years of diagnosis and low C-peptide) in 223 individuals with diabetes between 20 and 40 years of age. Of those, 21% were severely insulin-deficient.

The type 1 diabetes genetic risk score was highly discriminatory in this group, with AUC 0.87. Here, cutoff scores as defined previously were similarly sensitive and specific for severe insulin deficiency: a type 1 diabetes score greater than 0.280 had 92% specificity and 54% sensitivity for severe insulin deficiency, with a positive predictive value of 63% and a negative predictive value of 88%.

A type 1 diabetes score less than 0.234 had 96% specificity and 56% sensitivity for the absence of severe insulin deficiency with a positive predictive value of 98% and a negative predictive value of 37%.

When the type 1 diabetes genetic risk score was combined with the other predictors of severe insulin deficiency — islet-autoantibody status and age at diagnosis — the discriminatory power was increased to an AUC of 0.96.

This is similar to the Exeter group's previous work integrating biochemical results and clinical features to predict MODY, the authors note.

A subset of just nine single nucleotide polymorphisms discriminated accurately between type 1 and type 2 diabetes, with an AUC of 0.873 compared with 0.8880 for all 30.

Availability and Affordability of New Genetic Risk Score

Dr Hattersley said that the Exeter lab offers a diagnostic service for the type 1 diabetes genetic risk score that costs about £50 (approximately US $75), which is available now and simply requires a blood sample. They're currently setting up a service to provide gold-standard diagnostic service to patients.

This price is expected to fall considerably if there is a large volume of testing done at a single time, he said.

Ultimately, the goal is "to work on the commercialization of this test so it can be performed very cheaply all over the world."

Regarding the near-100% accuracy of the combination of genetic score with antibodies and clinical features, Dr Hattersley noted, "You could argue that given you have diabetes for life, that to spend $100 to get the right diagnosis from the beginning is money well spent."

Dr Oram was supported by a Diabetes UK Clinical Training Fellowship during this project and is currently supported by an Alberta Health Services Transplant Fellowship. This work was supported by a Wellcome Trust Senior Investigator award to Dr Hattersley, which funded genotyping costs. Dr Hattersley is also supported by a National Institute for Health Research (NIHR) Senior Investigator award. Additional support came from the University of Exeter and the NIHR Exeter Clinical Research Facility. Disclosures for the coauthors are listed in the paper.

Diabetes Care. Published online November 17, 2015. Abstract


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.