Introduction: Thrombocytopenia-associated multiple organ failure (TAMOF) encompasses a broad spectrum of pathologies, and it is unknown if immune dysregulation drives coagulopathy for a subset of TAMOF. We hypothesized that high-dimensional proteomic data could identify a protein signature that differentiates TAMOF from other causes of multiple organ failure (MOF). Methods: Heparinized plasma was obtained from 85 patients with and without sepsis within 48 hours from the onset of ≥2 organ system failures defined by Proulx criteria. Targeted proteomics was performed using proximity extension assay (Olink, Sweden). We compared patients with platelet counts < 100,000/mm3 (TAMOF, n=46) to patients without thrombocytopenia (MOF, n=39). Penalized regression (LASSO) with 10-fold cross-validation was used to identify a minimal set of proteins that differentiate TAMOF from MOF. To define biologic subgroups, we used XGBoost to derive high-dimensional signatures of TAMOF, then performed hierarchical clustering to identify proteomic MOF endotypes. Results: Lasso identified 13 proteins that differentiate TAMOF from MOF with positive predictors including IL-15, a key cytokine in NK cell and CD8+ T cell proliferation, and negative predictors including PRELP, an inhibitor of complement activation. Supervised clustering revealed 4 subgroups: TAMOF-predominant (cluster 1: 27 TAMOF, 3 MOF), MOF-predominant (cluster 2: 1 TAMOF, 24 MOF), mixed hypoinflammatory (cluster 3: 7 TAMOF, 5 MOF), and mixed hyperinflammatory (cluster 4: 7 TAMOF, 7 MOF). A heatmap of hallmark immune pathways across clusters showed upregulation of IL6-JAK-STAT3, IL2-STAT5, TNFα, and IFNγ signaling in both clusters 1 and 4 relative to clusters 2 and 3. Conclusions: High-dimensional proteomic data biologically distinguishes TAMOF from MOF. Proteomic data, agnostic to thrombocytopenia, accurately predicted TAMOF patients. TAMOF is related to, but distinct from, hyperinflammatory sepsis and has parsimonious features. IL-15 has been implicated in endothelial dysfunction and thrombogenesis in other disease states, thus its role in TAMOF will be further investigated. Additionally, the impact of IL-15 on complement-mediated coagulopathy will also be studied.
Ham et al. (Sun,) studied this question.