DocTIS’ results and the origin of a universal diagnostic tool for inflammatory diseases

The scientific article “Interpretable Inflammation Landscape of Circulating Immune Cells”, recently published in bioRxiv, an open-access preprint repository for the biological sciences, leverages data from the DocTIS project.

In this investigation, a comprehensive single-cell atlas was crafted, encompassing over two million peripheral blood mononuclear cells (PBMCs) obtained from 356 patients afflicted by 18 distinct inflammatory diseases and healthy controls. A significant portion of the sequencing data (79%) was either internally generated or provided by collaborators. Notably, the DocTIS initiative played a pivotal role, providing the data for 184 patients (+50%) of six immune-mediated disorders including: Rheumatoid Arthritis (RA), Psoriatic Arthritis (PSA), Crohn’s Disease (CD), Ulcerative Colitis (UC), Psoriasis (PS), and Systemic Lupus Erythematosus (SLE), as well as healthy subjects.

Furthermore, many of the article’s authors are participants in DocTIS and represent the project’s partner organizations: the Centro Nacional de Análisis Genómico (CNAG), Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Vall d’Hebron Research Institute, IMIDomics, Cardiff University, University of Verona, and Charité-Universitätsmedizin Berlin.

The study provides compelling evidence for the potential of using circulating immune cells as a liquid biopsy for Immune mediated inflammatory diseases (IMIDs) diagnostics, when combined with the use of single-cell technologies and advanced machine learning models. These models represent a cornerstone for the development of classifiers for inflammatory diseases and, ultimately, as potential precision diagnostic tools for clinical applications.

Summary of the article:

Inflammation, a biological response with both physiological and pathological implications, has been the subject of extensive research. While controlled inflammation aids in restoring homeostasis, dysregulated inflammatory responses can lead to adverse outcomes. Recent advancements have allowed for the characterization of acute and chronic inflammation in specific diseases, primarily through the application of single-cell sequencing. This technique has been instrumental in identifying alterations in cell type composition and disease-driving mechanisms, thereby highlighting potential therapeutic targets.

Despite these advancements, a comprehensive map detailing immune cell states across various diseases and effectively charting immune plasticity in inflammatory diseases remains elusive. To address this gap, researchers have integrated single-cell transcriptomic data of immune circulating cells across a multitude of diseases, analysing a large number of cells to extract a full spectrum of features representing inflammatory processes. This approach has facilitated the generation of an interpretable model of inflammation in circulating immune cells.

Leveraging advances in single-cell genomics, researchers have been able to delineate the full spectrum of immune cell activation underlying inflammatory processes during infection, immune-mediated inflammatory diseases, and cancer. The use of sequencing technologies has enabled unbiased immuno-phenotyping of single cells, providing comprehensive cellular landscapes without the need for prior knowledge.

The researchers employed a three-phase analysis strategy, beginning with the evaluation of well-known inflammation-related patterns, followed by the unsupervised discovery of interpretable biomarkers associated to cell types and diseases, and culminating in patient classification. This approach has paved the way for the development of precision medicine diagnostic tools for patients experiencing severe acute or chronic inflammation.

In conclusion, the researchers have generated a comprehensive landscape of inflammation in circulating immune cells from acute and chronic inflammatory diseases using single-cell genomics. By employing advanced machine learning pipelines, they have developed interpretable models for biomarker identification. Furthermore, they have classified patients based on generative models that have learned the full inflammatory feature space across cell types and diseases, thereby paving the way for a universal diagnostic tool for inflammatory diseases.

Access the full article here: https://www.biorxiv.org/content/10.1101/2023.11.28.568839v1