Beyond Autoantibodies: How to Optimize Type 1 Diabetes Screening?

To better meet the expectations of our readers, we asked our subscribers to choose the topic they were most interested in regarding immunology. This article answers one of the most popular questions. You can also find our other two articles from our special coverage of the IDS, available for free and also translated into English: “Therapies to Alter the Course of Type 1 Diabetes and Achieve a Cure: Exploring Diverse Pathways” and “Type 1 Diabetes Cure: What to Expect from Immunology Specialists?”.

 

During the Bruges congress, Belgian researcher and endocrinologist Chantal Mathieu shared her excitement about the current era in diabetes research and the potentialities it seems to open up: “It’s very exciting, and it’s a whole new world opening up to us. As endocrinologists and pediatric endocrinologists, we need to accept the concept that we are now able to diagnose T1D not only at the time of hyperglycemia, but also when two or more autoantibodies are present.” This new era of prediction, that is before the onset of clinical symptoms, is able to anticipate treatment and care, resulting in unavoidable and growing questions regarding the screening.

 

The current president of the EASD (European Association for the Study of Diabetes) is clear about the role of screening, stating that, thanks to it, it is now possible to “really change the course of the disease through education and follow-up of the screened individuals, and prevent them from developing diabetic ketoacidosis at stage 3, that is, at the clinical diagnosis of T1D”.

 

This explains the global launch of screening initiatives to detect the presence of these antibodies in the blood of first-degree relatives of individuals with T1D, as well as in children and adolescents of the general population. Emphasizing their importance, Chantal Mathieu specifies, “The time has come to organize our healthcare systems so that screening and follow-up of screened individuals can take place.”

However, the choice of individuals to be screened and the methods used vary globally due to the highly heterogeneous nature of the disease, which makes it difficult to predict. Despite this, several sessions at the congress contributed to providing insights on improving screening by addressing the concept of genetic risk scores. A genetic risk score is defined as a tool that evaluates the predisposition to the disease by analyzing key genetic variants to estimate the risk of its development.

 

For Professor Richard Oram of the University of Exeter in the UK, the genetic risk score is important in T1D because “it helps identify the risk of developing the disease and can thus be applied to screening”. Professor Oram further specifies that “genetic risk works quite well in individuals who test positive for autoantibodies,” demonstrating the value of this indicator. This has been shown in various studies conducted in India and Africa, where the identification of typical T1D cases was done effectively, though with limitations for atypical cases.

 

In general, polygenic scores are effective at differentiating T1D patients from other populations with great precision. The genetic risk score, which aggregates multiple key genetic data points, allows classification by risk level: low, medium, or high. However, the British researcher acknowledges, “Capturing the genetic risk of T1D is not enough.” One of the main limitations lies in genetic variations between populations: scores must, therefore, be adjusted to account for ancestral and geographical factors.

 

Professor Oram remains optimistic about the potential improvement of T1D prediction models. According to him: “Integrating these scores into large genetic databases, such as newborn sequencing programs, will provide a more accurate prediction when combined with other biomarkers, paving the way for optimized, personalized care.”

 

Meanwhile, Italian researcher in translational and precision medicine at the University of Siena, Erika Pedace, shared her work on the role of microRNAs, small molecules that regulate gene expression. Her goal was to determine if they could be relevant indicators of progression to stage 3 of T1D, i.e., its clinical diagnosis. For Erika Pedace, “Autoantibodies alone are good predictors, but they do not provide precise information about when the disease will appear. Combined microRNAs could fill this gap.”

 

Erika Pedace’s Ph.D. project led her to compare the presence of these microRNAs in samples from individuals who were not affected by T1D. These individuals, who were autoantibody-positive and had a family history of the disease, were divided into two groups: those who progressed to the clinical diagnosis and those who did not. A total of 19 microRNAs were identified as expressed differently between the two groups, with four being validated as significantly more expressed in individuals who were eventually diagnosed.

 

This contributes to the ongoing progress in refining and improving screening in order to predict, with greater precision, who will be diagnosed and when. Furthermore, complementary tools based on Artificial Intelligence (AI) and mathematics are contributing to this improvement effort.

Professor Fabian Theis, Head of the Health Informatics Center and Director of the Institute for Computational Biology at the Helmholtz Center in Munich, spoke about the potential of AI assisting research. He emphasized the importance of relying on machines both for their large-scale data analysis capabilities and their ability to perform automated learning. “Generative AI creates models that allow us to extract possibilities from complex data.”

 

Taking the example of the study of the individuality of cells through an “omic” approach—that is blending chemistry, biology, and data science tools and technologies—he wants to shed light on diabetes and its prediction. Thanks to AI’s analysis, life trajectories of cells, their evolution, and their interactions are better understood. Through the development of virtual models, AI can accelerate research by simulating biological responses and thus predicting the progression of the disease. According to Theis, “In ten years, an AI assistant could suggest a researcher which experiments to perform next: a real co-pilot for research.”

Lauric Ferrat, a researcher from the University of Geneva, suggests a combination involving AI and mathematics for more precise and economically viable screening strategies. For the mathematician, the most effective alliance for disease development prediction is to rely on the genetic risk and the presence of autoantibodies. “The first remains stable throughout life, while the latter is specific and dynamic. Together, they form a magical duo.” When integrated into mathematical models, these two variables help halve the number of tests required while ensuring a high level of detection for affected individuals.

 

Similarly, mathematical models should minimize the costs of screening while maximizing its benefits. What is sought is “optimization, which means doing more with less”. To illustrate his point, Lauric Ferrat uses the example of a pediatric test conducted at key ages, such as 2 and 6 years old, which are particularly relevant for establishing a diagnosis. He also suggests focusing efforts on children with high genetic risk scores to significantly reduce the cost of screening. His microsimulations, which help determine the best screening strategies, have shown that a targeted approach can reduce by 40%, severe complications, such as diabetic ketoacidosis, while avoiding universal screening of the entire population.

 

While the possibilities for T1D screening seem wider than ever, there are still many questions to address in order to enable its effective generalization. Reducing the rate of diabetic ketoacidosis at diagnosis, considering anxiety and psychological burden, addressing social inequalities, and ensuring economic viability for healthcare systems are all issues that shape the path toward a logical and virtuous approach, yet still far from actual prevention and a cure.

 

This article is part of a three-part series on the Immunology of Diabetes Society Congress. To ensure free access to this content, the series is funded by IDS. Glucose toujours retains full editorial independence.

 

Translation reviewed by Anna Jones