BACKGROUND Sepsis remains a major clinical challenge for which successful treatment requires greater precision in identifying patients at increased risk of adverse outcomes requiring different therapeutic approaches. Predicting clinical outcomes and immunological endotyping of septic patients generally relies on using blood protein or mRNA biomarkers, or static cell phenotyping. Here, we sought to determine whether functional immune responsiveness would yield improved precision.METHODS An ex vivo whole-blood enzyme-linked immunosorbent spot (ELISpot) assay for cellular production of interferon γ (IFN-γ) was evaluated in 107 septic and 68 nonseptic patients from 5 academic health centers using blood samples collected on days 1, 4, and 7 following ICU admission.RESULTS Compared with 46 healthy participants, unstimulated and stimulated whole-blood IFN-γ expression was either increased or unchanged, respectively, in septic and nonseptic ICU patients. However, in septic patients who did not survive 180 days, stimulated whole-blood IFN-γ expression was significantly reduced on ICU days 1, 4, and 7 (all P < 0.05), due to both significant reductions in total number of IFN-γ–producing cells and amount of IFN-γ produced per cell (all P < 0.05). Importantly, IFN-γ total expression on days 1 and 4 after admission could discriminate 180-day mortality better than absolute lymphocyte count (ALC), IL-6, and procalcitonin. Septic patients with low IFN-γ expression were older and had lower ALCs and higher soluble PD-L1 and IL-10 concentrations, consistent with an immunosuppressed endotype.CONCLUSIONS A whole-blood IFN-γ ELISpot assay can both identify septic patients at increased risk of late mortality and identify immunosuppressed septic patients.TRIAL REGISTRY N/A.FUNDING This prospective, observational, multicenter clinical study was directly supported by National Institute of General Medical Sciences grant R01 GM-139046, including a supplement (R01 GM-139046-03S1) from 2022 to 2024.
Evan L. Barrios, Monty B. Mazer, Patrick W. McGonagill, Christian B. Bergmann, Michael D. Goodman, Robert W. Gould, Mahil Rao, Valerie E. Polcz, Ruth J. Davis, Drew E. Del Toro, Marvin L.S. Dirain, Alexandra Dram, Lucas O. Hale, Mohammad Heidarian, Caleb Y. Kim, Tamara A. Kucaba, Jennifer P. Lanz, Ashley E. McCray, Sandra Meszaros, Sydney Miles, Candace R. Nelson, Ivanna L. Rocha, Elvia E. Silva, Ricardo F. Ungaro, Andrew H. Walton, Julie Xu, Leilani Zeumer-Spataro, Anne M. Drewry, Muxuan Liang, Letitia E. Bible, Tyler J. Loftus, Isaiah R. Turnbull, Philip A. Efron, Kenneth E. Remy, Scott C. Brakenridge, Vladimir P. Badovinac, Thomas S. Griffith, Lyle L. Moldawer, Richard S. Hotchkiss, Charles C. Caldwell
Usage data is cumulative from March 2025 through March 2026.
| Usage | JCI | PMC |
|---|---|---|
| Text version | 1,918 | 277 |
| 200 | 41 | |
| Figure | 325 | 0 |
| Table | 225 | 0 |
| Supplemental data | 186 | 5 |
| Citation downloads | 102 | 0 |
| Totals | 2,956 | 323 |
| Total Views | 3,279 | |
Usage information is collected from two different sources: this site (JCI) and Pubmed Central (PMC). JCI information (compiled daily) shows human readership based on methods we employ to screen out robotic usage. PMC information (aggregated monthly) is also similarly screened of robotic usage.
Various methods are used to distinguish robotic usage. For example, Google automatically scans articles to add to its search index and identifies itself as robotic; other services might not clearly identify themselves as robotic, or they are new or unknown as robotic. Because this activity can be misinterpreted as human readership, data may be re-processed periodically to reflect an improved understanding of robotic activity. Because of these factors, readers should consider usage information illustrative but subject to change.