Associations of serum MCP-1, GDF-15, PCSK9, and NLRP3 with fecal Bifidobacterium and Lactobacillus abundance and 28-day mortality in sepsis: a prospective single-center cohort study

Sepsis is a leading cause of intensive care unit death, and conventional severity scores capture only part of the biology determining outcome. Intestinal dysbiosis, particularly reduced abundance of short-chain fatty acid-producing commensals, has been associated with systemic inflammation in sepsis, yet the circulating mediators linking microbial changes to deterioration remain poorly defined. We examined whether serum monocyte chemoattractant protein-1 (MCP-1), growth differentiation factor-15 (GDF-15), proprotein convertase subtilisin/kexin type 9 (PCSK9), and nucleotide-binding oligomerization domain-like receptor protein 3 (NLRP3) are associated with lower abundance of the putatively beneficial commensal genera Bifidobacterium and Lactobacillus and with 28-day mortality. In this prospective single-center cohort, 215 adults meeting Sepsis-3 criteria were enrolled and stratified by septic shock ( n = 52) and 28-day vital status (61 deaths). Serum biomarkers were quantified by enzyme-linked immunosorbent assay, and fecal Escherichia coli , Bifidobacterium , and Lactobacillus were measured by absolute real-time polymerase chain reaction. Continuous variables were assessed for distributional properties and compared by the Welch t test or the Mann to Whitney U test as appropriate, and categorical variables by the Pearson chi-square test or Fisher exact test. Associations were evaluated with Spearman correlation, multivariable logistic regression adjusted for age, Sequential Organ Failure Assessment (SOFA) score, and lactate, and receiver operating characteristic analysis with DeLong comparison, calibration assessment, and 1,000-sample bootstrap internal validation in which the full model-fitting process was repeated in each resample. All four biomarkers were higher and Bifidobacterium and Lactobacillus were lower in patients with shock and in non-survivors (all P < 0.05). Each biomarker correlated inversely with Bifidobacterium and Lactobacillus (Spearman rho − 0.31 to − 0.46, all P < 0.001; associations persisted in stratified and partial-correlation analyses). After adjustment, MCP-1, GDF-15, PCSK9, and NLRP3 were each independently associated with 28-day mortality, whereas higher Bifidobacterium and Lactobacillus abundance was associated with lower 28-day mortality. The four-biomarker panel discriminated 28-day mortality (area under the curve [AUC] 0.917) better than any single marker (DeLong P < 0.001), with an apparent 86.9% sensitivity and 81.8% specificity at an internally derived cut-off ; adding clinical variables raised the AUC to 0.954 (bootstrap-corrected 0.943; optimism-corrected calibration slope 0.86; Brier score 0.078). Discrimination was preserved across antibiotic-exposure and sampling-time subgroups. Serum MCP-1, GDF-15, PCSK9, and NLRP3 are associated with lower abundance of Bifidobacterium and Lactobacillus and with 28-day mortality; in combination with routine clinical variables, they provide internally validated risk prediction with modest optimism, supporting an integrated inflammatory-microbial approach to sepsis risk stratification that requires external validation in multicentre, ethnically diverse cohorts before any clinical implementation. Clinical trial number: not applicable.