Circulating pyruvate is a potent prognostic marker for critical COVID-19 outcomes

Abstract

Background: Coronavirus-19 (COVID-19) disease is driven by an unchecked immune response to the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus which alters host mitochondrial-associated mechanisms. Compromised mitochondrial health results in abnormal reprogramming of glucose metabolism, which can disrupt extracellular signalling. We hypothesized that examining mitochondrial energy-related signalling metabolites implicated in host immune response to SARS-CoV-2 infection would provide potential biomarkers for predicting the risk of severe COVID-19 illness.

Methods: We used a semi-targeted serum metabolomics approach in 273 patients with different severity grades of COVID-19 recruited at the acute phase of the infection to determine the relative abundance of tricarboxylic acid (Krebs) cycle-related metabolites with known extracellular signaling properties (pyruvate, lactate, succinate and α-ketoglutarate). Abundance levels of energy-related metabolites were evaluated in a validation cohort (n=398) using quantitative fluorimetric assays.

Results: Increased levels of four energy-related metabolites (pyruvate, lactate, a-ketoglutarate and succinate) were found in critically ill COVID-19 patients using semi-targeted and targeted approaches (p<0.05). The combined strategy proposed herein enabled us to establish that circulating pyruvate levels (p<0.001) together with body mass index (p=0.025), C-reactive protein (p=0.039), D-Dimer (p<0.001) and creatinine (p=0.043) levels, are independent predictors of critical COVID-19. Furthermore, classification and regression tree (CART) analysis provided a cut-off value of pyruvate in serum (24.54 µM; p<0.001) as an early criterion to accurately classify patients with critical outcomes.

Conclusion: Our findings support the link between COVID-19 pathogenesis and immunometabolic dysregulation, and show that fluorometric quantification of circulating pyruvate is a cost-effective clinical decision support tool to improve patient stratification and prognosis prediction.

Keywords: COVID-19; energy-related metabolites; fluorometric quantification; pyruvate; semi-targeted metebolomics.

Figure 1

Study design and patient cohort distribution. Cohort 1 comprised 273 patients consecutively recruited during the first COVID-19 wave (March-May 2020) from the outpatient clinics of the participating hospitals. Cohort 2 included 398 patients with COVID-19 recruited from March 2020 to December 2020 at the HUJ23. Patients were stratified by disease severity. Serum samples were used for semi-targeted metabolomic analysis (cohort 1) and quantitative fluorimetric assays (cohort 2).

Figure 2

Semi-targeted metabolomics study of patients in cohort 1. (A) Relative abundance of succinate, α-ketoglutarate, lactate and pyruvate acid in patients grouped by disease severity (mild, severe and critical). Statistical significance between different groups was estimated using the Kruskal-Wallis test. Bars represent median values ± SEM. (B) Correlation matrix between relevant parameters previously related to COVID-19 severity and metabolites measured in patients of cohort 1. The color of the squares corresponds to the absolute value of the Spearman correlation coefficient, with orange or blue color indicating negative or positive correlation, respectively. A blank square indicates a lack of correlation between variables. The results were considered significant at *P<0.05; **P<0.01; ***P<0.001.

Figure 3

Energy-related metabolites in the validation cohort. (A) Serum levels (µM) of succinate, lactate, pyruvate and α-ketoglutarate determined by fluorimetric assay in patients with mild (n=65), severe (n=218) and critical (n=115) COVID-19 disease. Statistical significance was estimated using the Kruskal-Wallis test and Dunn’s multiple comparisons test. (B) Receiver operating characteristic curves predicting COVID-19 severity included the parameters of the regression model (Table 3). (C) Regression tree (CART) analysis including all the previously selected variables as predictors of COVID-19 severity. The results were considered significant at *P<0.05; **P<0.01; ***P<0.001.