Periodic Reporting for period 1 - PredRegNetworks (Listening to Silence: The First Blood Test for Risk of Individuals at Prophylactic Time Windows to Develop Type 2 Diabetes Mellitus)
Reporting period: 2024-04-01 to 2025-09-30
Over 350 million individuals worldwide currently live with pre-Type 2 diabetes (prediabetes), defined as impaired glucose tolerance (IGT) or impaired fasting glucose (IFG). This number is projected to approach 548 million by 2045. Nearly half of them (48.1%) are under the age of 50, and one third (28.3%) are in the age group of 20-39 years, thus likely to spend many years at high risk of developing severe outcomes. In the face of the currently available treatment protocols, conversion rate of prediabetes to full Type 2 Diabetes (T2D) is substantial: 5-10% of prediabetics progress each year, with as many as 70% developing T2D during their lifetime3. The consequences are harmful, including blindness, kidney failure, cardiovascular disease, stroke, and limb amputations. More than 3 million people die annually from the disease and its complications. The global economic burden exceeds €1 trillion per year, ranking among the greatest health-related challenges worldwide.
Early treatment of the people at risk with available medications, combined with lifestyle modifications, can reduce progression risk and improve survival by 40-70%. These interventions are most effective when implemented before the onset of metabolic deterioration, manifested as impaired glucose metabolism (IGT, IGF). However, by the time prediabetes can be identified through currently available tests, both insulin resistance and β-cell damage are already present, resulting in severe insulin sensitivity and secretion problems, with limited treatment abilities. This inability to detect those at risk early enough is a direct driver of an enormous burden on individuals, communities, and governments (Figure 1).
Our long-term goal is to develop an effective, simple, blood-based test of the risk of young individuals with normal glucose levels to develop prediabetes during their life. Based on the results of a former ERC-funded research program, we developed a new class of epigenetic biomarkers that capture the earliest shifts in gene regulation, and therefore, may provide far greater predictive power than existing approaches. In this Proof-of-Concept research we have validated the prediction power of this novel biomarkers in identifying the people at risk. This new ability to identify the people at risk, and provide them with effective treatments in time, may dramatically improve the way we are dealing with these harmful diseases.
Early treatment of the people at risk with available medications, combined with lifestyle modifications, can reduce progression risk and improve survival by 40-70%. These interventions are most effective when implemented before the onset of metabolic deterioration, manifested as impaired glucose metabolism (IGT, IGF). However, by the time prediabetes can be identified through currently available tests, both insulin resistance and β-cell damage are already present, resulting in severe insulin sensitivity and secretion problems, with limited treatment abilities. This inability to detect those at risk early enough is a direct driver of an enormous burden on individuals, communities, and governments (Figure 1).
Our long-term goal is to develop an effective, simple, blood-based test of the risk of young individuals with normal glucose levels to develop prediabetes during their life. Based on the results of a former ERC-funded research program, we developed a new class of epigenetic biomarkers that capture the earliest shifts in gene regulation, and therefore, may provide far greater predictive power than existing approaches. In this Proof-of-Concept research we have validated the prediction power of this novel biomarkers in identifying the people at risk. This new ability to identify the people at risk, and provide them with effective treatments in time, may dramatically improve the way we are dealing with these harmful diseases.
Relying on measuring methylation levels at a few dozen sites, carefully selected from a list of more than three thousand newly discovered prediabetes-associated biomarkers, we developed a new, effective test of the risk of young individuals with normal glucose levels to develop prediabetes conditions. Unlike polygenic risk scores or symptom-based tools, this assay captures early regulatory changes that arise from the interaction between genetic and environmental factors and affect critical disease genes. These changes cause distinct, biologically critical alterations of key transcription regulators, which ultimately promote disease development. These unique biomarkers capture the earliest shifts in gene regulation, and therefore, they provide far greater predictive power than existing approaches.
Using a well-characterized prospective cohort, we analyzed whole-blood samples collected at ages 31-34 from individuals with normal glucose metabolism (fasting blood glucose <100 mg/dL; A1C<5.7). Some of the tested individuals developed prediabetes (fasting blood glucose 100-125 mg/dL; A1C 5.7-6.4) at ages 41-46 (average of 13 years from initial sample), while the controls remained normoglycemic. Prediabetes-related biomarkers were examined in these baseline samples, and prediction models were developed using different subsets of the novel biomarkers, wherein each biomarker is assigned a defined weight, which contributes to the final test value. Validation of the tests was performed in an independent set of individuals (progressors to prediabetes and matched controls) of the same age group, not included in the samples used for the test development.
Receiver Operating Characteristic (ROC) analysis demonstrated superior predictive power across multiple prediction models based on subsets of the newly discovered biomarkers, with all models achieving an AUC above 0.95. The most efficient model, using only 30 methylation sites, achieved an AUC of 0.97 (95% confidence interval: 0.962-0.988) with accuracy of 0.97 and precision of 0.94. Comparative performance analysis showed that the new test consistently outperformed existing prediction models and scrambled-data controls, achieving markedly higher accuracy and separation between progressors to prediabetes and non-progressors.
Using a well-characterized prospective cohort, we analyzed whole-blood samples collected at ages 31-34 from individuals with normal glucose metabolism (fasting blood glucose <100 mg/dL; A1C<5.7). Some of the tested individuals developed prediabetes (fasting blood glucose 100-125 mg/dL; A1C 5.7-6.4) at ages 41-46 (average of 13 years from initial sample), while the controls remained normoglycemic. Prediabetes-related biomarkers were examined in these baseline samples, and prediction models were developed using different subsets of the novel biomarkers, wherein each biomarker is assigned a defined weight, which contributes to the final test value. Validation of the tests was performed in an independent set of individuals (progressors to prediabetes and matched controls) of the same age group, not included in the samples used for the test development.
Receiver Operating Characteristic (ROC) analysis demonstrated superior predictive power across multiple prediction models based on subsets of the newly discovered biomarkers, with all models achieving an AUC above 0.95. The most efficient model, using only 30 methylation sites, achieved an AUC of 0.97 (95% confidence interval: 0.962-0.988) with accuracy of 0.97 and precision of 0.94. Comparative performance analysis showed that the new test consistently outperformed existing prediction models and scrambled-data controls, achieving markedly higher accuracy and separation between progressors to prediabetes and non-progressors.
We have developed and validated the first effective test to detect the risk of developing prediabetic states among individuals with normal glucose metabolism. We have demonstrated that our test provides higher performance than existing tests. The new test use biomarkers that can only be measured using our technology and cannot be detected with conventional assays including Genome Sequencing, Nanopore sequencing, circulating cell-free DNA sequencing (cfDNAseq), or standard microarrays. As a result, they are absent from existing databases and have not been used in previous attempts to develop predictive models. Current undergoing development steps include optimization and demonstration in particular population groups.