Editorial 

Caring for the Critically Ill Patient

Precision Therapy for Sepsis: The End of the Beginning?

Derek C. Angus

JAMA Published Online: December 8, 2025

doi: 10.1001/jama.2025.24099

Sepsis is a macabre dance between host and pathogen.¹ With development of germ theory by Lister, Semmelweis, Pasteur, and Koch, pathogen eradication was the focus. Only in recent decades, with observations by Roger Bone² and others that patients succumbed to sepsis despite successful eradication of the pathogen, did attention swing to the host. It is now well accepted that a hallmark of sepsis is the systemic host immune response to infection that is directly injurious to vital organs. Unfortunately, the response is also extremely complex and highly variable in a way that is not easily appreciated from routine clinical examination or laboratory tests. This clinically elusive heterogeneity in the underlying pathogenesis of sepsis has bedeviled attempts to develop immune modulating therapies, with a litany of failed trials. The great hope is that success will lie in a strategy of precision therapy for sepsis, where a particular host response pattern can be identified swiftly and treated with more targeted interventions. Although ideas abound on how to measure and treat specific host-response patterns, the hope has remained largely theoretical, with little demonstration that it is practical and effective in clinical practice.

In this issue of JAMA, Giamarellos-Bourboulis and colleagues³ address this issue directly with their report on the results of ImmunoSep, a multicenter trial of precision therapy for sepsis. ImmunoSep was a randomized clinical trial of patients with sepsis cared for in the intensive care units of 33 hospitals in 6 European countries. Patients were eligible if they met Sepsis 3⁴ criteria secondary to pneumonia or primary bacteremia. With informed consent, blood was drawn and sent by expedited dispatch to national laboratories to perform 2 assays (serum ferritin and human leukocyte antigen [HLA]–DR expression on circulating monocytes) to characterize the host response as macrophage-activation-like syndrome (high ferritin and normal HLA-DR), sepsis-induced immune paralysis (normal ferritin and low HLA-DR), or unclassified (normal ferritin and HLA-DR expression). Those with an unclassified response were not randomized. The remainder were randomized 1:1 to receive precision therapy or placebo. Those in the precision therapy group received a 15-day course of either intravenous anakinra if they had macrophage-activation-like syndrome or subcutaneous recombinant interferon if they had immune paralysis. To ensure blinding, all patients in both arms received intravenous and subcutaneous injections, with control patients receiving double placebo and patients in the precision therapy arm receiving the placebo form of the nonindicated intervention in addition to their indicated active agent.

All study procedures up to the first randomized dose had to occur within 72 hours of meeting sepsis criteria. The primary outcome was the change in vital organ dysfunction from baseline, measured as the difference between the mean Sequential Organ Dysfunction Assessment (SOFA) score at enrollment and that averaged over days 2 to 9. The authors powered their study to find a more favorable change in the mean SOFA score of at least 1.4 points, a magnitude shown previously to correlate with subsequent survival. The authors also specified several secondary and sensitivity analyses.

Over 30 months, the research team screened 672 patients, of whom 355 (53%) had an unclassified host response and 281 (42%) met criteria with either macrophage-activation-like syndrome (7%, n = 48) or immune paralysis (34%, n = 228) and were randomized to intervention or placebo. Removing 5 individuals who withdrew consent, 276 were available for analysis. Within this group, intervention and placebo delivery and follow-up was excellent. A greater proportion of those in the intervention group had at least a 1.4-point improvement in their SOFA score compared with the control group (35.1% [46 of 131] vs 17.9% [26 of 145], P = .002). Secondary analyses suggested that therapy was effective in both host immune signatures regarding the primary outcome. In a subset of patients with sequential biomarkers, host immune signatures appeared to normalize more rapidly in the intervention group. Secondary analyses of organ dysfunction broadly supported the primary findings, but there was no significant improvement in mortality. Serious adverse events were infrequent and similar between groups. It appeared therefore that the precision therapy approach was safe, feasible, at least in this setting, and beneficial regarding resolution of organ dysfunction.

There are however several caveats. First, the authors’ strategy to identify different host immune signatures involved several somewhat arbitrary decisions. Although both macrophage-activation–like syndrome and immune paralysis are well described, they may not be mutually exclusive and there are alternative approaches to classify them. There are also other potential immune response signatures. All of these signatures can evolve over days or hours, and none are part of current parlance in clinical practice.⁵ Although ferritin is routinely available in clinical laboratories, HLA-DR expression is not. Classifying biomarkers dichotomously may be overly simplistic, and any chosen cut point may not carry the same implications across different laboratories. The 2 signatures together accounted for less than half of all screened cases, leaving unaddressed the role of precision therapy for the majority of patients.

The choice of interventions is also somewhat arbitrary. There are alternative doses and durations, dosing could have been titrated to individual responses, and many other immune-modulating agents could have been chosen. In particular, immune paralysis is a very broad concept for which a plethora of immune-stimulating agents could have been deployed.⁵

Finally, there is high face validity in using a composite measure of organ dysfunction as an intermediate outcome in critical illness. However, the analytic approach taken by the authors was somewhat unusual, summarizing the mean score across days 2 through 9, and expressing success as an improvement of 1.4 or more. It is reassuring that the additional analyses were supportive of the primary findings. However, as noted by the authors, it would be nice to see this approach tested in a larger trial with a primary outcome that was more patient-centric.

That all said, it is remarkable to be at a point where a multicenter precision therapy trial for sepsis has been fully executed and available for scrutiny. That the trial ran with such fidelity across multiple sites and countries is a crucially important testament to the feasibility for future efforts. That the approach, despite testing only a limited set of biomarker and therapeutic strategies, met the primary outcome is also highly promising. ImmunoSep has not found the cure for sepsis, and, indeed, there are a number of issues over which many may quibble. However, to paraphrase Winston Churchill: this study represents neither the end of the war on sepsis nor the beginning of the end. But it is the end of the beginning.⁶ The battle is joined: trials of precision therapy for sepsis are now a reality. We look forward to many more.