Global 
Deutschland Translational Science Corner
Nat Med. 2026;32(2):633-644
Interpretable inflammation landscape of circulating immune cells
Inflammation is a biological phenomenon beneficial for homeostasis, but it is unfavorable if dysregulated. Although major progress has been made in characterizing inflammation in specific diseases, a global, holistic understanding is still elusive. This is particularly intriguing, considering its function for human health and the potential for modern medicine if fully deciphered. In this study, the authors leveraged advances in single-cell transcriptomics to delineate inflammatory processes of circulating immune cells during infection, immune-mediated inflammatory diseases and cancer. The authors’ single-cell atlas of more than 6.5 million peripheral blood mononuclear cells from 1047 patients (56% female, 43% male) and 19 diseases allowed us to learn a comprehensive model of inflammation in circulating immune cells. The atlas expands current knowledge of the biology of inflammation of immune-mediated diseases, acute and chronic inflammatory diseases, infections and solid tumors and lays the foundation to develop a disease classification framework using unsupervised as well as explainable machine learning. Beyond a disease-centered analysis, the authors charted altered activity of inflammatory molecules in peripheral blood cells, depicting discriminative inflammation-related genes to further understand mechanisms of inflammation. The authors present a rich resource for the community and lay the groundwork for learning a classifier for inflammatory diseases, presenting cells in circulation as living biomarkers.
DOI: 10.1038/s41591-025-04126-3
Dr. Lena Sophie Mayer
Specialist Internal Medicine, University Medical Center Freiburg, Department of Internal Medicine II, Hugstetter Str. 55, 79106 Freiburg, Germany
Systemic inflammatory programs in circulating immune cells: Relevance for biomarker development and precision medicine
Inflammatory responses protect the body from pathogenic microbes, play a central role in tissue repair, and are key for cancer control. The immune system must maintain a delicate balance; when this balance is disrupted, a protective immune response can become pathological. Because cellular and molecular inflammatory mediators are involved in nearly every disease and can be modulated pharmacologically, a comprehensive, cross-disease understanding of inflammatory processes is essential.
In this study, the authors generated a large single-cell atlas of inflammatory processes by analyzing the transcriptomes of peripheral blood mononuclear cells (PBMCs). In total, 6.5 million cells from 1047 patients across 19 diseases—including autoimmune, infectious, and malignant conditions—were profiled. This exceptionally large dataset, encompassing individuals of both sexes and all age groups across diverse diseases, provides substantial biological variability and enables systematic cross-disease comparisons.
The analysis identified conserved inflammatory programs, particularly interferon-associated pathways, representing shared immunological mechanisms across diseases. At the same time, disease-specific transcriptional signatures were found to complement these shared inflammatory programs, enabling differentiation between individual diseases. Importantly, the identified signatures are cell-type-specific, indicating that global analyses must always be interpreted in the context of the relevant immune cell populations. Overall, this study provides a valuable framework for biomarker discovery and advances the conceptual basis of precision medicine.
Methodologically, the authors employ state-of-the-art techniques. While the individual methods—single-cell RNA sequencing (scRNA-seq), scVI, scANVI, and SHAP—are each well established, the novelty of this work lies in their combination: 1) large-scale single-cell datasets across multiple diseases and studies, 2) probabilistic models that summarize the key biological features of cells, and 3) interpretability through systematic application of SHAP at the level of cells, diseases, and genes, enabling the identification of genes driving disease classification. Biological signals and batch effects are considered separately, making potential confounders visible.
Despite these strengths, the study has several limitations. First, sampling was restricted to individuals of European descent; expanding the dataset to diverse populations will be necessary to capture global immunological variability. Second, the dataset is limited to PBMCs, whereas critical disease mechanisms occur in tissues and may be tissue-specific, limiting the generalizability of the findings. Third, integrating datasets from multiple studies risks introducing batch effects that may not be fully eliminated, even after applying advanced correction methods. Finally, functional validation of the identified transcriptional pathways is lacking; the analyses primarily reveal correlations and cannot conclusively establish whether the identified signatures are key causal factors in disease pathogenesis. Experimental studies, ideally prospective and longitudinal if appropriate, are required to assess the pathogenic relevance of individual signatures. Additional follow-up studies are also needed to demonstrate the clinical utility of this work, for instance in improving decision-making.
In conclusion, this study highlights the potential of modern machine-learning techniques for the systematic analysis of complex immunological processes at single-cell resolution. Analyses of circulating cells enable minimally invasive monitoring of disease activity, potentially capturing pathological changes in the blood before they manifest in tissues. The identification of disease-specific signatures offers the potential to define novel biomarkers, which could in turn enhance diagnostic precision and support the development of targeted therapies.