Uncovering Shared and Divergent Pathogenic Pathways in SLE, MS, RA, and T1D
General Audience Summary
Autoimmune diseases occur when the body’s immune system mistakenly attacks healthy cells. Although conditions like multiple sclerosis (MS), rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and type 1 diabetes (T1D) affect very different parts of the body, they share a common underlying problem: a breakdown in the immune system’s ability to distinguish “self” from “non self.” Because many autoimmune diseases run in families and share similar genetic risk factors, researchers believe they may arise from overlapping biological mechanisms. Understanding what these diseases have in common may reveal powerful new ways to prevent, diagnose, and treat them.
This study has three main objectives. The first is to identify new treatment opportunities by combining genetic findings with other biological data (such as proteins in the blood). Large genetic studies have already identified many DNA regions linked to RA, SLE, and T1D. By examining how these genetic markers influence protein levels, we can pinpoint proteins that may play a direct role in causing disease. Some of these proteins might already be targeted by existing drugs, creating opportunities to repurpose medications originally designed for other conditions. This approach has already shown success in MS, where several promising targets and drug candidates were identified.
The second objective is to develop computer based prediction models that can identify who is at risk of developing autoimmune diseases long before symptoms appear. Many autoimmune conditions have a long “silent” phase, where the immune system is beginning to malfunction but obvious symptoms have not yet developed. Sweden’s nationwide health registries offer a unique resource for this kind of research. These registries contain decades of medical information, prescription records, family links, and demographic data for nearly the entire population. Using advanced machine learning methods, the research team aims to build prediction tools capable of recognizing early warning patterns. These models can then be enhanced using genetic information and lifestyle data to improve accuracy and help identify individuals who may benefit from early monitoring or preventive interventions.
The third objective is to better understand which genetic factors are shared across autoimmune diseases and which are unique to specific conditions. Many genetic risk markers—especially in the HLA region—affect multiple autoimmune diseases, while others influence only one. Even shared genetic variants can have opposite effects depending on the disease. By separating HLA based risk from other genetic influences and studying them across several autoimmune diseases, we aim to map out the biological pathways that underlie both common and distinct disease features. This will also help identify subgroups of patients with similar genetic profiles, which could explain why symptoms and treatment responses vary so much from person to person.
The anticipated outcomes of this study are wide reaching. We expect to identify new biological targets for autoimmune treatments, including existing medications that can be repurposed. We also aim to develop early warning tools that can predict disease risk with high accuracy, giving clinicians the ability to monitor or intervene earlier. Finally, the study will generate detailed genetic maps and identify patient subgroups, paving the way for more personalized and effective treatment strategies.
Overall, this project aims to improve the understanding of autoimmune diseases, accelerate treatment discovery, and support earlier diagnosis—ultimately helping reduce long term disability and improving quality of life for those affected.
Scientific Abstract
Autoimmune diseases share some risk factors while others differ. There is familial co-aggregation between several autoimmune diseases for example between multiple sclerosis (MS), and systemic lupus erythematosus (SLE), and to a lesser extent with rheumatoid arthritis (RA), and type 1 diabetes (T1D). Genetically autoimmune diseases cluster in different groups, with diseases where B-cells are thought to have a more dominant role forming one clusters and those where T-cells are more important forming another cluster. MS falls in-between these two clusters. It is thus likely that some of the processes leading to autoimmune diseases are shared between the diseases while others are unique or shared with other autoimmune diseases. Understanding these processes and which can be targeted with drugs can lead to novel treatment some of which may be shared between several diseases. These diseases have prodromal phase when alterations in immune reactions have started but have not yet lead to overt autoimmunity. If individuals at risk of developing autoimmune diseases can be identified earlier changes in lifestyle or preventative treatments may reduce the burden of autoimmune disease.
We therefore have the following aims
1. Drug repurposing using multiomics integration based on genetic associations for T1D, SLE and RA, similar to what we have already done for MS
2. Build predictive models using deep learning methods such as XGboost and DeepHit for MS, RA, SLE and T1D using national healthcare registry linkage data enhancing them with other risk factors when such data is available.
3. Construct high-fidelity genetic risk maps for each disease that separate HLA and non-HLA contributions using state of the art methods as well as novel machine learning methods.
4. Identify genetic loci and functional elements that drive shared versus divergent risk across diseases.
5. Characterize patient-level heterogeneity that will inform prediction and mechanistic follow-up.
These goals are aimed to distinguish common and disease-specific mechanisms in autoimmunity though comparative, mechanism-oriented genetics using state of the art methods and exploring the use of artificial intelligence for several of the aims (aims 2, 3 and 5). We will use both publicly available data (aim 1), health administrative data (aims 2 and 5) as well as local genetic and lifestyle (aims 2, 3, 4 and 5).
Sweden is one of few countries where individual data on life-long healthcare and drug usage can be followed in national registries and linked to other available data. We have already used this for studies of MS and will now extend it to the other diseases. We will utilize large cohorts of MS, RA and SLE patients which have already been well characterized by us. Hence no new collection of samples or profiling of them will be included in this study, although the possibility of following up findings with additional lab work exists.
Expected outputs include- Novel protein drug targets for autoimmune diseases- Potential existing drugs for re-purposing for autoimmune diseases- Potential predictive models for autoimmune diseases that in future can be employed to diagnose patients earlier enabling earlier treatment/prevention by change of lifestyle and hence protection from tissue damage and disease burden.- Disease-specific and cross-disease genetc models, ranked lists of shared versus unique functional genetic loci and regulatory elements.- Sex-stratified mechanistic hypotheses, and genetic discriminators suitable for inclusion in the disease prediction models. Collectively, these deliverables will prioritize credible genetic targets for drug repurposing and discovery across autoimmune diseases and will integrate directly with the project’s other modules on drug-target identification and predictive modelling. In this project researchers from several departments at Karolinska Insitutet are joining their expertise and unique recourses from different autoimmune diseases.