Endothelial Cell Phenotypes Demonstrate Different Metabolic Patterns and Predict Mortality in Trauma Patients

Abstract

In trauma patients, shock-induced endotheliopathy (SHINE) is associated with a poor prognosis. We have previously identified four metabolic phenotypes in a small cohort of trauma patients (N = 20) and displayed the intracellular metabolic profile of the endothelial cell by integrating quantified plasma metabolomic profiles into a genome-scale metabolic model (iEC-GEM). A retrospective observational study of 99 trauma patients admitted to a Level 1 Trauma Center. Mass spectrometry was conducted on admission samples of plasma metabolites. Quantified metabolites were analyzed by computational network analysis of the iEC-GEM. Four plasma metabolic phenotypes (A–D) were identified, of which phenotype D was associated with an increased injury severity score (p < 0.001); 90% (91.6%) of the patients who died within 72 h possessed this phenotype. The inferred EC metabolic patterns were found to be different between phenotype A and D. Phenotype D was unable to maintain adequate redox homeostasis. We confirm that trauma patients presented four metabolic phenotypes at admission. Phenotype D was associated with increased mortality. Different EC metabolic patterns were identified between phenotypes A and D, and the inability to maintain adequate redox balance may be linked to the high mortality.

Integrating Genome-Scale Metabolic Models with Patient Plasma Metabolome to Study Endothelial Metabolism In Situ

“…In this sense our computational analysis revealed notable differences in the activity states of metabolic tasks, reminiscent of those described in Henriksen et al 2023 [17] (see Figure 4c). For instance, an imbalance in ATP generation between glycolytic and oxidative phosphorylation pathways, as observed in the metabo-groups outlined by Henriksen et al 2023, was also evident in our analysis. Furthermore, alterations in other metabolic tasks were identified.…”

“…The individual group clustering analysis revealed notable differences in the activity states of metabolic tasks, reminiscent of those described in Henriksen et al 2023 [19]. For instance, an imbalance in ATP generation between glycolytic and oxidative phosphorylation pathways, as observed in the metabogroups outlined by Henriksen et al 2023, was also evident in our analysis. Other relevant tasks that were identified were the synthesis of Cholesterol, Arginine, methiylglyoxal, alanine, Hydroxymethylglutaryl-CoA, mevalonate, glucuronate, cysteine, also synthesis of anthranilate, N-formylanthranilate, quinolinate from tryptophan and the degradation of methionine and homocysteine (Figure 4c).…”

“…To mitigate this issue, the presented pipeline incorporates the ADSB algorithm, aimed at drastically reducing computational time [20]. The performance of ADSB was assessed through a comparative analysis with the gpSampler algorithm, a well-established and widely used method implemented in Matlab [17]. Both, ADBS and gpSampler algorithms were applied on two GEMs: the E. coli GEM EcoliCore [35] and the EC GEM iEC3006 [19].…”

“…The initial analysis failed to cluster patients into the four metabo-groups as described in Henriksen et al 2020 [17] (see Figure 4a). To gain deeper insights, we conducted the same clustering analysis on the same GEMs but now separated by metabo-groups and standardizing by row (See Figure 4b).…”

“…As a proof of concept, the pipeline outlined in the previous section was applied to a cohort of 95 trauma patients, for whom blood plasma samples were obtained to measure the metabolome [17]. To establish a reference point, metabolomic data from a control group of 20 healthy individuals was utilized to calculate the relative metabolomic data employed in characterizing the reconstructions of patient-specific GEMs.…”