Editorial: sepsis: studying the immune system to highlight biomarkers for diagnosis, prognosis and personalized treatments

Sepsis, defined by The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3), is a life-threatening organ dysfunction caused by a dysregulated host response to infection emphasizing the pivotal role-playing by the innate and adaptive immune response in the development of the clinical syndrome (1). Due to its complex pathogenesis that involves networks of multiple systems, a marked heterogeneity is observed from patient to patient suffering from sepsis, both in terms of organ dysfunction distribution and its severity (2, 3). Despite advances in organ support and antimicrobial therapy up to 25% of patients still succumb to sepsis. But the situation is even more dramatic in septic shock, with a hospital mortality rate of around 60% (4–6). Therefore, there is a high priority to improve the prevention, recognition, diagnosis and management of sepsis. Unfortunately, there is no single diagnostic available test that establishes the diagnosis of sepsis or septic shock (7, 8). Clinical criteria, that distinguish sepsis from localized microbial infection is a failing host response with multiple organ dysfunction and potentially septic shock (9). Despite the cited consensus Sepsis-3 (1), new criteria will be needed to predict life-threatening sepsis (8). There is heterogeneity in patient’s responses that may be related to multiple endophenotypes of sepsis that may have implications for personalized treatments (10, 11). Currently, emerging research evaluating omics technologies and bioinformatics methods may improve the care of sepsis patients. Using existing datasets of genetic expression of septic patients, artificial intelligence systems (AI) are trained to recognize disease progression and clinical outcomes (12, 13). For instance, in 2019 Seymour et al. published a paper describing the application of machine learning algorithms to readily available clinical data. Based on the analysis of 29-sepsis-related variables four novel sepsis phenotypes (α, β, γ and δ) with different demographics, laboratory values and patterns of organ dysfunction could be individualized. Compared to other phenotypes, the δ-phenotype is characterized by greater rates of acute renal, hepatic and endothelial dysfunction and mortality rates (14, 15). The pathophysiology of sepsis has recently been reviewed (3, 16, 17). A major research effort to identify biomarkers in patients with sepsis or septic shock predict mortality has been made. 250 biomarkers have been identified and evaluated including complement system, cytokines, chemokines, cell membrane receptors, soluble receptors, metabolites, damage-associated molecular patterns, non-coding RNAs, miRNAs and more, but no single biomarker or combinations accurately differentiated between and sepsis-like syndrome. Thus, prediction of patient outcomes in sepsis is still driven by clinical signs. It is expected that integrated biomarker-guided algorithms together with the application of AI using machine learning methods may hold promise to improve both the diagnosis and prognosis of sepsis, as well as, to optimize more personalized patient care at the bedside (6, 11, 15, 18). Moreover, extensive clinical and experimental research has compellingly demonstrated that sepsis can induce a state of immunosuppression, leading to heightened vulnerability to secondary, predominantly opportunistic infections. The mechanisms underpinning sepsis-induced immunosuppression encompass immune cell apoptosis, the proliferation of regulatory T (Treg) cells, and impaired microbial clearance by macrophages (19). In this context, our recent work has yielded fresh insights into the purinergic signaling pathway in severe experimental sepsis in mice. Our findings unveiled a notable expansion of t