Assessment of soluble PD-L1 in septic shock in relation to immunosuppressive phenotypes

Abstract

Background: Septic shock triggers a complex immune response characterized by the coexistence of hyperinflammation and immunosuppression, the latter being a major driver of ICU-acquired infections and increased mortality. Currently, the most established biomarkers for assessing sepsis-induced immunosuppression rely on flow cytometry-a technique not universally available in clinical practice. In contrast, soluble biomarkers are, in principle, easier to measure. Although assays for soluble PD-L1 (sPD-L1) are not yet standardized, sPD-L1 concentrations may represent a pragmatic alternative, given the putative role of PD-1/PD-L1 signaling in immunosuppressive pathways during sepsis. In this study, we investigated sPD-L1 in relation to established cellular markers of immunosuppression in a cohort of 161 patients with septic shock. sPD-L1 levels were measured using the ELLA microfluidic platform during the first week of ICU admission. We assessed their association with clinical outcomes and explored the relationship between sPD-L1 and immunosuppressive profiles defined by low monocytic HLA-DR expression (mHLA-DR) and absolute lymphocyte count.

Results: Upon admission, patients exhibited elevated sPD-L1 levels compared to healthy controls (medians: 179 vs. 54 pg/mL, p < 0.001). No correlation was observed between sPD-L1 levels and severity scores (SOFA, SAPS II). Elevated sPD-L1 was independently and significantly associated with increased mortality at both 28 and 90 days. Longitudinal analysis using K-means clustering revealed that the cluster with consistently highest sPD-L1 levels was associated with unfavorable outcomes. Overall, and at any single time point, sPD-L1 concentrations did not correlate with mHLA-DR expression or lymphopenia. However, the combined presence of high sPD-L1 and low mHLA-DR levels at the end of the first week identified a subgroup of patients with particularly poor clinical outcomes.

Conclusions: These findings highlight the potential of sPD-L1 as a clinically relevant biomarker in the context of sepsis immunopathology. Further studies are warranted to elucidate its role in the mechanisms underlying sepsis-induced immunosuppression. Such insights could support the integration of sPD-L1 into multimodal biomarker panels for immune monitoring and risk stratification in patients with septic shock.

Keywords: Biomarkers; Hyperinflammation; Immunosuppression; Septic shock.

Fig. 1

Temporal Dynamics of Soluble PD-L1 in Septic Shock: Analysis in the Whole Cohort and Stratification by Clinical Outcome. (A) sPD-L1 concentration in septic shock patients compared to healthy volunteers (HV) and its evolution at D1-2, D3-4, D5-8 and 6 months. (B) sPD-L1 concentrations comparison between 28-day survivors (n = 122) and non-survivors (n = 39) at D1-2, D3-4, D5-8. (C) sPD-L1 concentrations comparison between 90-day survivors (n = 110) and non-survivors (n = 45) at D1-2, D3-4, D5-8. (D) sPD-L1 concentrations comparison between patients with (n = 45) and without ICU-acquired infection (n = 116) at D1-2, D3-4, D5-8. sPD-L1 values were censored once infections occurred. Comparison of each with Mann-Whitney tests = ns: non significant; * p-value < 0,05; ** p-value < 0.01; *** p-value < 0.001.

Fig. 2

sPD-L1 association with mortality and cluster analysis. (A) ROC curves to predict 28- and 90-day mortality based on D3-4 or D5-8 sPD-L1 concentrations. (B) Table of multivariate analysis to predict 28- and 90-day mortality. Only the parameters significant in univariate analysis were included. (C) Kaplan Meier of 28-day mortality, stratification on D3-4 median sPD-L1, with high sPD-L1 and low sPD-L1. (D) Kaplan Meier of 28-day mortality, stratification on D5-8 median sPD-L1, with high sPD-L1 and low sPD-L1. (E) Kaplan Meier of 90-day mortality, stratification on D3-4 median sPD-L1, with high sPD-L1 and low sPD-L1. (F) Kaplan–Meier of 90-day mortality, stratification on D5-8 median sPD-L1, with high sPD-L1 and low sPD-L1. The log rank test was used to test the difference between the curves. (G) Median sPD-L1 trajectories according to the clusters obtained with K-means approach, cluster 1 (n = 127) and cluster 2 (n = 34). (H) Box plot presentation of sPD-L1 concentrations according to both clusters at different time-points. (I) Kaplan–Meier curves depicting 90-day mortality according to cluster assignment.

Fig. 3

sPD-L1 and mHLA-DR combined can predict adverse outcomes in septic shock patients. (A) sPD-L1 concentrations comparison between immunosuppressed (n = 36) and non immunosuppressed patients (n = 106). Comparison of each populations at different time-points with Mann-Whitney test = * p-value < 0.05 ; ** p-value < 0.01. Immunosuppressed status at D5-8 was defined when patients presented with both mHLA-DR < 8,000 AB/C (20) and ALC < 0.9 G/L (22). (B) Alluvial plot to compare two groups of patients with high or low sPD-L1 stratified on D5-8 median and two groups of patients with high or low mHLA-DR stratified on D5-8 median. (C) Percentage of 28-day mortality according to different categories of patients: severe patients with low mHLA-DR and high sPD-L1; patients with either low mHLA-DR or high sPD-L1 and non-severe patients with high mHLA-DR and low sPD-L1. (D) Kaplan–Meier analysis for 28-day mortality, between severe patients with low mHLA-DR and high sPD-L1 (n = 49) and the rest of the cohort (n = 109). (E) Kaplan Meier analysis for 90-day mortality, between severe patients with low mHLA-DR and high sPD-L1 (n = 49) and the rest of the cohort (n = 109). (F) Kaplan–Meier of ICU-acquired infections (% cumulative infection free), between severe patients with low mHLA-DR and high sPD-L1 (n = 45) and the rest of the cohort (n = 103). The log rank test was used to test the difference between the curves.