Evolving concepts in sepsis: we are making progress!

Sepsis, a word of Greek origin initially used by Hippocrates to refer to putrefaction, has become the term used to describe a serious infection, i.e., an infection complicated by organ dysfunction. The word sepsis is more appropriate than the term 'septicemia' which has been widely used but implies the presence of microorganisms in the blood whereas blood cultures are positive in scarcely 50% of patients with sepsis.Sepsis is a global disease, responsible for some 20% of total annual deaths and designated a worldwide health priority by the World Health Organization (World Health Organization 2020). It is difficult to estimate the full individual and societal burden of sepsis, especially as there are limited data available from low and middle income countries, yet these populations are likely to be disproportionately affected given the general poorer hygiene and resource availability for sepsis prevention and treatment in these areas (Schultz et al. 2017). Moreover, in addition to its high mortality rates, sepsis is also responsible for considerable short and long-term morbidity with associated high economic impact in terms of costs of hospitalization and treatment, of long-term care if needed, and of lost workforce productivity.The impact of sepsis on the emotional, psychological, and social well-being of affected individuals and their families is also substantial. Recent initiative such as the Global Sepsis Alliance (https://globalsepsisalliance.org), the International Sepsis Forum (https://sepsisforum.org), and World Sepsis Day (https://www.worldsepsisday.org) have raised awareness of sepsis, but many challenges remain to address and reduce the huge burden of this condition worldwide.The current management of sepsis relies on hemodynamic stabilization and infection control.Hemodynamic stabilization requires the administration of intravenous fluids, vasopressor agents (primarily norepinephrine and sometimes vasopressin), and inotropic agents (primarily dobutamine) when required. Infection control requires adequate antibiotic therapy and source control. Nevertheless, these measures are not, and will never be, fully effective. It is now recognized that not even the most effective antibiotic therapy can control all cases of sepsis.Being able to modulate the sepsis response is a tantalizing prospect but is currently limited to corticosteroid administration in severe cases and is a subject of ongoing debate and controversy (Bode et al. 2023).The critical care community assumed for far too long that sepsis was a homogeneous, primarily hyperinflammatory host response to infection. This influenced our approach to developing potential therapies, focusing on agents with anti-inflammatory or immunosuppressive effects. We now consider that the basic underlying mechanism is better described as a "dysregulation" of the host response (Singer et al. 2016). Understanding the different facets of this dysregulation will help in the development of more specific, targeted sepsis therapeutics. These recent concepts and the possible systems immunology approaches that can be used to improve our understanding of the complexities of the sepsis response and thus move towards more precision-based treatments are beautifully presented by Hancock and colleagues in their lead article (Hancock et al. 2025). Important aspects to remember include, first, that the underlying immune alterations of sepsis are highly complex and need to be better characterized in each individual patient, which is now becoming possible, especially with the assistance of artificial intelligence-based models. And, second, these alterations can change rapidly over time, implying the need for regular, repeated assessments of the host response.We can thus identify a path towards real progress in this field. After many years of negative trials trying to identify sepsis drugs that would be effective for all patients with sepsis, we have come to recognize that this was an oversimplistic illusion. Patients with sepsis are so different in terms of demographics, comorbidities, genetics, causative microorganisms, stage of disease at presentation, prior treatments, and degree of host response, among other factors, that identifying a single agent that would work for all was never going to be realistic. It has become obvious that more specific interventions are needed. Better characterization of individual patients will help determine which therapy is most likely to be of benefit in which patient. Hancock et al. (Hancock et al. 2025) excellently describe some of the tools available to achieve this, focusing on endotypes, in which patients are characterized according to underlying pathophysiological mechanisms, such as degree or type of immune response. A recent roundtable conference held in Brussels proposed focusing also on patient subphenotypes or treatable traits, which characterize patients more according to specific clinical features or outcomes rather than biological mechanisms (Gordon et al. 2024). The development of theranostics, combining diagnostic approaches (using biomarkers, endotypes, phenotypes, etc, to characterize patients) with appropriate therapeutic choices, has been used to guide the selection of the most relevant medication for individual patients in clinical trials in sepsis. For example, Vincent et al. selected only patients with sepsis-associated coagulopathy for inclusion in a randomized trial of thrombomodulin versus placebo (Vincent et al. 2019) and Francois et al. assessed response to nangibotide, which modulates triggering receptor expressed on myeloid cells (TREM)-1, according to concentrations of soluble TREM-1-a known sepsis biomarker (Francois et al. 2023). It has even been suggested that we may no longer need the word "sepsis" to describe a patient's condition, and this could be replaced using methods that more precisely evaluate and define the immune status.Nevertheless, several hurdles remain. One is that a patient's characteristics may change rapidly over time, and the trend is not predictable or identical in each individual. The specific moment of onset of sepsis is also generally not known with precision; sepsis may have developed before the admission to the intensive care unit (ICU) or even before admission to the hospital. Second, different types of response may coexist: some cells may be in a hyperinflammatory state at the same time as others are immunosuppressed (van Vught et al. 2017). Third, we usually assess the host response in the blood, but any alterations may be different in the tissues. To overcome some of these challenges, new studies need to investigate not only mortality outcomes but also other patient-relevant benefits, including limiting the development of organ failure and facilitating an uncomplicated recovery with shorter ICU and hospital stays. Even if a therapeutic strategy is not demonstrated to increase survival, effects on other outcomes can be clinically meaningful. Trial designs other than the traditional randomized controlled trial may also help in identifying and assessing new interventions (Gordon et al. 2024). For example, adaptive clinical trial designs, in which multiple trial arms are initially included and those showing promise are continued while others are rapidly discontinued. With the improved patient characterization methods highlighted by Hancock et al. (Hancock et al. 2025), clinical trials could also focus not so much on the presence of an infection, which is sometimes difficult to establish definitely (Maraolo et al. 2025), but on a particular pattern, characterized by a specific marker, endotype, phenotype and so on. So-called 'basket trials' are now used in oncology to test whether a new drug can be effective in patients who have a certain abnormality regardless of the type of cancer. Likewise, critically ill patients could be enrolled in a trial when they have a particular profile, regardless of documented presence of infection.As new therapies become available assisted by these novel approaches, a reasonable management option, based on current knowledge, may be to initially use an intervention that could reduce the inflammatory response when present, and then to immunostimulate the host in the later phase of immunosuppression. Limitations to this approach are that the proinflammatory response may be quite short (van Amstel et al. 2024), the two phases may be present simultaneously in some patients, and the immunosuppressive phase may not contribute markedly to mortality (van Vught et al. 2016). The better characterization of patients now becoming possible, as Hancock et al. (Hancock et al. 2025) discuss, will facilitate appropriate treatment choices for individual patients.In conclusion, better characterization of the host response over time in individual patients with sepsis will help advance research in this field, allowing potential therapies to be trialed in more precisely defined populations who are most likely to benefit. This in turn will enable the host response to be controlled more precisely and thus more effectively. Global collaboration of multiple stakeholders-research scientists, clinicians, industry, healthcare managers, politicians, and governments-is needed to help overcome the remaining challenges and obstacles, including the high associated costs, and drive the incorporation of precision medicine into clinical practice to help improve sepsis outcomes.