Medical and behavioural interventions in individuals at high risk of type 2 diabetes are effective in delaying or preventing the onset of the disease, but a substantial proportion of people with prediabetes are missed by current clinical screening and diagnostic techniques. Often individuals with isolated impaired glucose tolerance (iIGT), a common subtype of prediabetes that affects about one in forty Europeans, can only be identified through oral glucose tolerance testing. Oral glucose tolerance testing is a time-consuming procedure requiring repeated blood draws, and is therefore not routinely performed as part of type 2 diabetes clinical screening strategies.
The scientists around Prof. Claudia Langenberg, head of the division for Computational Medicine at the Berlin Institute of Health at Charité (BIH), who led the study, used a proteomic assay to measure the relative levels of nearly 5,000 proteins in blood plasma samples from more than 11,000 participants in the Fenland study, each of whom underwent an oral glucose tolerance test. The assay employs a library of chemically modified single-stranded DNA molecules called aptamers that bind specific proteins, and then DNA microarrays that bind the aptamers to determine the relative amount of each protein present in a sample. The authors created a machine learning algorithm that was able to extract a core set of few proteins out of the thousands measured that were most informative to identify participants with too high glucose levels even before the oral glucose tolerance test was performed.
The scientists identified a signature of only three proteins that when combined with standard screening techniques for impaired glucose tolerance improved identification of individuals with iIGT in the Fenland study cohort, and subsequently confirmed this finding in the independent Whitehall II study. Their results also indicate that fasting before the blood sample is taken does not significantly change the reliability of the three protein signature for identifying people with impaired glucose tolerance, which would greatly increase the application of the test in clinical practice.
Julia Carrasco Zanini, a PhD student at the MRC Epidemiology Unit and first author on the paper said: “The Fenland Study is unique for its size in combining genetic data and blood sampling with objective measurements of a range of clinical characteristics that includes oral glucose tolerance testing. By combining this resource with broad-capture proteomics technology we were able to identify protein signatures that substantially improved detection of impaired glucose tolerance.”
The authors suggest that by replacing the two-step screening strategy recommended by current guidelines with a three-step screening strategy that incorporates testing for the three-protein signature, the number of individuals who need to undergo oral glucose tolerance testing to identify an iIGT case could potentially be halved. However, they note that some individuals with iIGT would still be missed, an important consideration for clinical implementation.
Senior author Professor Claudia Langenberg, who has been appointed recently as Director of Queen Mary University of London’s Precision Healthcare University Research Institute, said: “Our strategy has the potential to address an important unmet clinical need: the identification of a meaningful proportion of people at high risk of developing type 2 diabetes who currently remain undetected. Early identification would enable preventive lifestyle and behavioural interventions to improve the health of affected individuals and alleviate the burden to health-care systems caused by their delayed diagnosis.
We would now like to evaluate the three-protein signature in other populations and ethnic groups, and ultimately to test the three step strategy for identifying prediabetes in randomised screening trials.”
The research was funded primarily by the Medical Research Council and Wellcome. Proteomic measurements were supported and governed by a collaboration agreement between the University of Cambridge and SomaLogic.