Background
The current Coronavirus disease 2019 (COVID-19) pandemic, caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) [
1], strained critical care resources in many countries, and the management of lung injury in these patients posed a tremendous challenge for clinicians [
2]. Several patients with COVID-19 developed severe acute respiratory distress syndrome (ARDS) with mortality rates around 30% [
3] in the first pandemic wave. ARDS is characterized by lung inflammation and hyperpermeability pulmonary edema. Currently, the diagnosis of ARDS is based on the presence of clinical, physiological, and radiological criteria [
4‐
6]. Unlike other clinical conditions, to date, there are no specific molecular markers that help in the prognosis of this condition. Advancement in the understanding of the pathogenesis of ARDS is necessary for designing innovative and effective therapeutic approaches.
Molecular approaches are needed to understand the mechanisms of ARDS induced by COVID-19. Of particular interest is the use of metabolomics for the characterization of this condition. The metabolome reflects early and specific alterations in the pathophysiological state of biological systems. In this context, Magnetic Resonance Spectroscopy (MRS) emerges as a highly potential tool for studying metabolic disorders in respiratory diseases [
7]. Several studies have proved the potential of MRS-based metabolomics to monitor patients with ARDS induced by respiratory infections [
8‐
10]. A previous study compared the specific metabolic fingerprint of ARDS patients with either influenza A pneumonia (IAP) caused by the H1N1-2009 influenza virus or pneumonia caused by
Streptococcus pneumoniae [
11]. Here, we used a similar approach for the characterization of the metabolic fingerprint of COVID-19-induced ARDS. We compared COVID-19 and IAP patients to identify the metabolic reprogramming involved in these two conditions. The identification of metabolic pathways involved in ARDS caused by the H1N1-2009 influenza virus or by SARS-CoV-2 will improve our understanding of the pathogenesis of COVID-19. Finally, as a proof of concept of the diagnostic potential of these metabolic biomarkers, we developed a predictive model to identify the etiological pathogens responsible for ARDS.
Discussion
The description of the metabolic alterations induced by SARS-CoV-2 infection in ICU patients is fundamental for a better understanding of the pathobiology of the disease. In the present study, we compared the metabolomic profile of ARDS due either to IAP or to COVID-19 by MRS using untargeted multivariate statistical analysis and metabolic pathway analysis. We found that the activation of many metabolic pathways was different between ARDS patients with COVID-19 or IAP. Furthermore, the serum metabolite profile of patients with ARDS discriminates the specific virus infection (H1N1-2009 influenza pneumonia versus SARS-CoV-2 pneumonia). PLS-DA model provided a classification accuracy of 100%. These findings are helpful for the understanding of the pathogenesis of severe COVID-19. Specifically, the metabolomic profile of ARDS in these patients suggests alterations in energy pathways, inflammatory response, and oxidative stress.
Previous studies that have analyzed the metabolism of patients with COVID-19 [
34‐
39] were designed to compare the metabolic profile of COVID-19 patients with healthy controls or to evaluate the metabolic differences between patients with a positive or negative outcome. Thus they cannot discern between the metabolic dysregulation due to SARS-CoV-2 infection or due to ARDS development. To the best of our knowledge, this is the first study designed to compare the metabolic profile of ICU patients with similar severity of ARDS due to COVID-19 or to other viral respiratory infections, e.g. IAP.
We have found specific metabolic differences between ARDS patients induced by COVID-19 or IAP. Most of these metabolic alterations have been previously reported as biomarkers of ARDS or ARDS severity [
8‐
10]. For example, a similar serum metabolic profile including proline, glutamate, phenylalanine, and valine was reported as a sensitive biomarker of ARDS severity (mild, moderate, and severe) [
40]. However, we have to consider that, unlike previous studies, in which metabolic changes induced by the viral infection itself or by the occurrence of ARDS cannot be distinguished, IAP and COVID-19 patients in the present study all met the criteria for the diagnosis of ARDS, and disease severity was similar. Thus differences in the metabolic profile herein reported are better explained by virus-specific pathogenetic mechanisms rather than by the occurrence of ARDS or by disease severity.
The ability of patients to normalize energy metabolism has been reported as one of the critical factors determining the COVID-19 progression [
39]. Compared with IAP patients, COVID-19 patients showed up-regulation of energy-generating pathways, i.e. glycolysis, fatty acid degradation, CoA biosynthesis, glycerolipids, and glycerophospholipids metabolism. The increase of lactate-to-glucose ratio found in COVID-19 patients is a biomarker of the up-regulation in the glycolysis pathway [
41]. In the same context, dysregulation of the choline metabolism and elevated levels of free poly-unsaturated fatty acids are biomarkers of the energy deficiency reported in COVID-19 patients [
42]. However, other authors [
34,
43] have associated the dysregulation of lipid metabolism with a higher atherosclerotic risk in COVID-19 patients. In the same line, COVID-19 patients showed a higher phenylalanine-tyrosine ratio that has been associated with an adverse outcome [
39] and may indicate a higher cardiovascular risk [
44] in COVID-19 patients [
45]. Other supplementary energy-generating pathways were also up-regulated. The excess of ketone bodies such as acetone suggests that they are used as an alternative energy source. Ketosis could be explained in the context of acute illness and lack of adequate caloric intake.
Alteration in amino acidic metabolism has been reported as one of the key features of ARDS development [
8,
9], and it has also been found significantly up-regulated in COVID-19 patients [
34,
39,
46,
47]. However, when we compared the metabolic profile of patients with ARDS due to COVID-19 or to IAP, we found that the amino acidic metabolism was decreased in COVID-19 patients. The serum concentrations of branched-chain amino acids (BCAAs), including isoleucine and valine, were decreased in COVID-19 compared with IAP patients. As elevated circulating BCAAs may promote oxidative stress [
48], the lower levels of BCAAs in patients with COVID-19 may result in less intense inflammatory response as compared to patients with influenza A [
49]. Downregulated BCAAs may also be considered a potential marker of the infection and its further involvement in the dysregulation of pantothenate and CoA biosynthesis [
47], as confirmed by the enrichment analysis. Pantothenic acid (vitamin B5) is required for coenzyme A formation and is also essential for α-ketoglutarate and pyruvate dehydrogenase complexes as well as fatty acid oxidation, compromising the mitochondrial energy metabolism [
50]. The increase in 2-hydroxybutyric acid, a readout of hepatic glutathione synthesis and marker of oxidative stress [
50], and essential amino acids such as proline [
51] confirmed more marked inflammatory and oxidative stress responses [
52] in IAP than in COVID-19 patients. A previous study also identified dysregulation of propanoate metabolism as a novel pathway in the progression of COVID-19 [
46], suggesting potential roles played by gut microbiota in the immune response [
53]. Finally, the methyl-guanidine-to-creatinine ratio is an index of hydroxyl radical formation in the lung, and it was identified previously as a specific metabolic pattern of IAP [
11].
Several limitations of the study should be acknowledged. First, the metabolic concentrations reported are relative to the total metabolic concentration, and baseline clinical differences among groups should be taken into consideration when interpreting the results. We did not perform absolute quantification because of limitations in sample manipulation for HR-MAS NMR analysis. External validation is required before the application of the specific metabolic fingerprint in clinical practice. Second, despite the overall similarity in disease severity (as measured by the SAPS II score, the SOFA score, and the mortality), patients differed in some characteristics, such as age and the requirement of noradrenaline, that could have an impact on the metabolic profile. Also, oxygenation impairment differed in the two groups, although the difference did not reach statistical significance. Third, some aspects of patients management could differ in the cohorts, as they span some years apart. Specifically, the way patients were mechanically ventilated could have had an impact on the metabolic profile. After the publication of the ARMA trial in 2000 [
54], different studies reported changes in the way mechanical ventilation was used, i.e., a lower tidal volume and slightly higher PEEP levels [
55,
56]. However, other studies have failed to show significant changes after 2010 [
57,
58]. Thus, it is unlikely that the metabolic changes reported in the present study are due to different mechanical ventilation strategies in the two groups. Fourth, ARDS was diagnosed according to the AECC, followed when the first cohort was recruited.
Acknowledgements
This research was supported by grants: (i) PID2019-10656RJ-I00 from the Spanish Ministry of Science and Innovation; (ii) S2017/BMD-3727-EXOHEP-CM from Comunidad de Madrid and Fondos FEDER, Madrid, Spain; (iii) B2017/BMD3875 from the Comunidad de Madrid, Madrid, Spain; (iv) SAF2017-84494-C2-1-R from the Spanish Ministry of Economy, Industry, and Competitiveness (MEIC-AEI); (v) Fondo Sectorial de Salud (155219), Agencia Nacional de Investigación e Innovación (vi) KK-2019/bmG19 from the Gobierno Vasco, Dpto. Industria, Innovación, Comercio y Turismo, under the ELKARTEK program and (vii) the European Union’s Horizon 2020 Research and Innovation Program, under the Marie Skłodowska-Curie grant agreement no. 823854 (INNOVA4TB). JRC received grants from the BBVA Foundation (Ayudas a Equipos de Investigación Científica de Biomedicina 2018) and from La Caixa Foundation (Health Research Call 2020 / HR20-00075). CIC biomaGUNE is supported by the Maria de Maeztu Units of Excellence Programme from the Spanish State Research Agency (Grant No. MDM-2017-0720). The funders were not involved at any stage, from study design to submission of the manuscript for publication. NMR studies were performed at the CAI BIOIMAC (Centro de Asistencia a la Investigación Bioimagen Complutense) node of the ICTS (Infraestructura Científicas y Técnicas Singulares) REDiB. The authors would also like to thank the nursing staff of the ICU and the medical team for their collaboration in managing these patients under unusually stressful conditions.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.