Distinct MAIT cell phenotypes associated with sepsis clinical outcome in emergency department patients

Abstract

Objectives: Rapid diagnosis and intervention are critical for sepsis patient outcomes. However, diagnosis is challenging because of a heterogenic patient group as well as sometimes vague symptoms when the patient presents at the emergency department. Mucosal-associated invariant T (MAIT) cells are rapid responders to infection, but their role and characteristics in the early course of sepsis remain unknown. Here, we evaluate the early MAIT cell characteristics in the blood of patients triggering a clinical sepsis alert system at the emergency department.

Methods: Peripheral blood mononuclear cells were isolated from freshly drawn blood and immediately stained. MAIT cell phenotyping analyses were conducted using multiparameter flow cytometry. All analyses were completed prior to the stratification of patients into sepsis or non-sepsis groups. Soluble factors in plasma were measured using a multiplex assay.

Results: Unsupervised high-dimensional phenotyping identified distinct MAIT cell activation profiles in sepsis and non-sepsis groups. Among sepsis patients, hierarchical clustering of MAIT cell phenotypes separated clinical endotypes into three groups with different infection focus, severity and aetiology. A prominent characteristic of sepsis severity was high expression of CD69 on MAIT cells, which was associated with organ dysfunction, lymphopenia and poor outcome. Plasma levels of IL-12, IL-15, TNF, IFNγ and CXCL10 correlated with the magnitude of MAIT cell activation in sepsis patients.

Conclusions: These clinical endotype-specific MAIT cell phenotypes presenting already in the emergency department are of interest for early patient identification and prognostication in sepsis.

Keywords: MAIT cells; immunophenotyping; sepsis; sepsis endotypes.

Figure 1

Study design and patient characteristics. (a) Schematic overview of study design, inclusion and exclusion criteria. (b) Patient characteristics and number of patients within the patient subgroups. (c, d) The distribution of (c) patients infected with indicated pathogens and (d) patients with indicated infection focuses within the sepsis and no sepsis groups.

Figure 2

Decline of MAIT cell frequencies in sepsis patients compared to healthy donors. (a) UMAP plots of total live CD3⁺ cells in peripheral blood of healthy donors (HD) and sepsis patients below 65 years of age and expression of indicated markers. (b) Total CD3⁺ cells overlaid with immune cell subsets identified by manual gating. (c) Total CD3⁺ cells of healthy donors (blue) or sepsis patients below 65 years of age (green). Red circles indicate MAIT cells. (d) Frequencies of MAIT cells and Vα7.2⁺CD4⁺CD161⁻ cells among total CD3⁺ cells and frequencies of non‐MAIT CD4⁺, CD8⁺, and double negative (DN) T cells among non‐MAIT CD3⁺ cells of healthy donors and sepsis patients below 65 years of age, identified by manual gating. (e) Change in MAIT cell frequency among CD3⁺ cells over time in sepsis and no sepsis patients. Each line represents one patient. (f) Frequencies of MAIT cells among total CD3⁺ cells and frequencies of non‐MAIT CD4⁺, CD8⁺ and DN T cells among non‐MAIT CD3⁺ cells in sepsis patients with and without positive blood cultures. (g) MAIT cell frequency among CD3⁺ cells in patients with or without viral infection. Data in (d), (f) and (g) are presented as median ± IQR and each dot represents an individual patient. Statistical analysis was performed using the nonparametric Mann–Whitney test. Statistically significant differences are indicated by **P < 0.01, *P < 0.05. In case of borderline significance, i.e. < 0.1, the actual P‐value is shown.

Figure 3

MAIT cells are activated early in sepsis. (a) Frequency of indicated markers on MAIT cells identified by manual gating in healthy donors and geriatric controls combined (Ctrl), no sepsis and sepsis patients. Data are presented as median ± IQR, and each dot represents an individual patient. Statistical analysis was performed using the nonparametric Kruskal‐Wallis test followed by Dunn's multiple comparisons test. ***P < 0.001, **P < 0.01, *P < 0.05. (b) UMAP of all MAIT cells from all patients included in the automated analysis, showing representative expression of phenotypic markers used for clustering. (c) UMAP of all MAIT cells from patients used for the analysis, split according to their sepsis status. (d) Percentage of 16 PhenoGraph clusters within total cells in sepsis and no sepsis patient groups. (e) Expression of phenotypic markers across 16 detected PhenoGraph clusters (column z‐score of median expression values). (f) Changes over time in frequencies of MAIT cells expressing indicated markers in non‐sepsis and sepsis patients. Each line represents one patient. Only statistically significant differences are indicated by ***P < 0.001, *P < 0.05.

Figure 4

MAIT cell activation in sepsis patients correlates with inflammatory cytokine levels and lymphopenia (a, b). Concentration of (a) IL‐12, IL‐18 and IL‐15 and (b) TNF, IFNγ, IL‐17A and granzyme B in plasma of healthy donors and geriatric controls combined (Ctrl), non‐sepsis and sepsis patients. Data are presented as median ± IQR, and each dot represents an individual patient. Statistical analysis is performed using the nonparametric Kruskal‐Wallis test, followed by Dunn's multiple comparisons test. Statistically significant differences are indicated by ***P < 0.001, **P < 0.01, *P < 0.05. In case of borderline significant, i.e. < 0.1, the actual P‐value is shown. (c) A heatmap displaying pairwise Spearman correlations between the frequency of surface markers on MAIT cells, soluble factors in plasma, and clinical parameters in sepsis patients. Colour indicates the strength of the correlation. P and rho values are shown in Supplementary table 4. (d) Spearman correlations between CD69 expression on MAIT cells and IL‐12 in plasma and CD69 expression and IL‐15 in plasma in sepsis. (e) Spearman correlations between CD69 expression on MAIT cells and IFNγ (log₁₀‐transformed) in plasma and CD69 expression and TNF (log₁₀‐transformed) in plasma in sepsis. (f) Spearman correlations between CD69 expression on MAIT cells and CXCL10 (log₁₀₋transformed) in plasma and CD38 expression and CXCL10 (log₁₀‐transformed) in plasma in sepsis. (g) Spearman correlations between CD25 and CD38 expression on MAIT cells and suPAR in plasma and CD4⁻CD8⁻ MAIT cells and suPAR in plasma in sepsis. (h) Spearman correlations between IL‐10 (log₁₀‐transformed) and suPAR and TNF (log₁₀‐transformed) and suPAR in plasma in sepsis. (i) Spearman correlations between CD69 expression on MAIT cells and lymphocyte counts. (j) Spearman correlations between IL‐15, IL‐12, or CXCL10 (log₁₀‐transformed) in plasma and lymphocyte counts. The Spearman correlation coefficient (rho) and the associated calculated P‐value (p) are indicated on graphs. ***P < 0.001, **P < 0.01, *P < 0.05.

Figure 5

MAIT cell activation profiles associated with clinical parameters and outcomes in sepsis patients. (a) Concentration of CXCL10 and IL‐15 in plasma, (b) MAIT cell frequencies among total CD3⁺ cells and (c) frequencies of CD69⁺, CD25⁺ and HLA‐DR⁺ MAIT cells in sepsis patients with moderate (2, 3) or high (≥ 4) Δ‐SOFA score. (d) UMAP of all MAIT cells from patients with or without sepsis, split according to their Δ‐SOFA score. (e) Percentage of 16 PhenoGraph clusters within total cells in Δ‐SOFA^high, Δ‐SOFA^mod and Δ‐SOFA^low patient groups. (f) Expression of indicated markers in selected PhenoGraph clusters that are over‐ or under‐represented in Δ‐SOFA^high patient group. (g) Concentrations of IL‐10, IL‐8 and TNF in sepsis patients with or without septic shock. (h, i) Frequencies of CD38⁺, CD69⁺ and PD‐1⁺ MAIT cells in sepsis patients (h) with or without septic shock and (i) patient treated in ICU or not. (j) Frequencies of LAG‐3⁺ MAIT cells in sepsis patients who developed secondary infections within 60 days. (k) Concentration of IL‐6 in plasma of sepsis patients who developed secondary infections within 60 days. (l) Frequencies of CD38⁺, CD69⁺ and PD‐1⁺ MAIT cells in sepsis patients who were alive, A, or died, D, within 28 days after the onset of sepsis. (m) Expression of indicated markers in selected PhenoGraph clusters that are over‐ or under‐represented in sepsis patients who died within 28 days after the onset of sepsis. (n) Concentration of IL‐1β in plasma of sepsis patients who were alive, A, or died, D, within 28 days after the onset of sepsis. (a–c, g– l, n) Data are presented as median ± IQR, each dot representing an individual patient. Statistical analysis was performed using the nonparametric Mann–Whitney test. Statistically significant differences are indicated by **P < 0.01, *P < 0.05. In case of borderline significance, i.e. < 0.1, the actual P‐value is shown.

Figure 6

The MAIT cell phenotype in sepsis is associated with clinical endotypes. (a) Receiver operating characteristic (ROC) analyses for high versus moderate Δ‐SOFA and for lymphopenia. The ROC plots show results, including the AUC values, of MAIT CD69 expression and the clinical markers white blood cells (WBC), procalcitonin (PCT) and lactate. (b) Hierarchical clustering of PhenoGraph clusters and clinical categorical parameters. The heatmap was calculated as column z‐score of cluster percentages within each clinical categorical parameter group. Putative MAIT cell endotypes are indicated as group 1 (orange), group 2 (purple) and group 3 (green).