Functional immune profiling of hyper- and hypo-inflammatory subphenotypes of critical illness: a secondary analysis

Abstract

Introduction: Recent studies of adult sepsis patients demonstrate the existence of two subphenotypes that differ in risk of mortality: a hyper-inflammatory subphenotype with a high risk of mortality, and a hypo-inflammatory or "not hyper-inflamed" subphenotype with a relatively lower risk of mortality. We recently investigated the association of organ dysfunction with ex vivo immune profiling in sixty (60) critically ill adult patients with sepsis. In this secondary analysis we measured cytokine biomarkers with an automated, microfluidic immunoassay device (Ella™) and sought to investigate the functional immune profiles of patients in the hyper/hypo-inflammatory subphenotype groups.

Methods: Subjects were consecutively identified adults, older than 18 years, and enrolled within 48 hours of sepsis onset. Whole blood cytokine analysis was performed in all patients. Additionally, ex vivo cytokine production was measured following 4h of stimulation. Cytokine concentrations were measured using the Ella™ automated immunoassay system.

Results: Subjects were divided into hypo-inflammatory (42 patients) and hyper-inflammatory (18 patients) subtypes using a previously validated parsimonious model based on concentrations of IL-6, TNFR1 and bicarbonate. The hyper- and hypo-inflammatory clusters demonstrated a near four-fold difference in 30-day mortality (44.4% vs 11.9%, p=0.0046). Following 4h of ex vivo stimulation with LPS, TNF production was lower in the hyper-inflammatory group as compared with the hypo-inflammatory group (p=0.0159). Ex vivo phorbol 12-myristate 13-acetate (PMA)-stimulated IFN-γ production (4h) by whole blood did not differ between groups.

Conclusions: These data further validate the use of IL-6, TNFR1 and bicarbonate to discern inflammatory sub-groups of patients with critical illness. They also confirm the observation that the presence of the hyper-inflammatory subphenotype is often accompanied by a compensatory anti-inflammatory response syndrome. Future investigations should focus on prospective validation of this panel for prognostic enrichment of clinical research studies.

Keywords: biomarker; critical illness; cytokines; humans; phenotype; sepsis.

Figure 1

Defining inflammatory subphenotypes by IL-6, TNFR1 and bicarbonate. Sepsis patients were divided into hyper-inflammatory (red) and hypo-inflammatory (blue) groups using IL-6, TNFR1 and bicarbonate levels based on a previously validated model (48). The association of other cytokines with the inflammatory groups is shown, with cytokines listed in order of increasing significant differences between the groups (A). Bicarbonate, the non-cytokine variable in the model, by definition, was significantly different (***p<0.001) between the groups (B). Mortality in the hyper-inflammatory group was significantly higher than the hypo-inflammatory group (44.4% vs 11.9%, **p<0.01) (C). Comparisons between the groups were performed using t-tests (A, B) and chi-squared tests (C). *p<0.05.

Figure 2

Comparison of functional immune profiling as determined by ex vivo cytokine production between inflammatory subphenotypes. Total TNF production (4h) was significantly (*p=0.016) lower in the hyper-inflammatory cluster (A), but not when corrected for absolute monocyte count (C). Ex vivo IFN-γ production did not differ between the inflammatory clusters in total (B), or after correction for absolute lymphocyte count (D). Significance was determined using non-parametric (Wilcoxon signed-rank) testing.