Machine learning identified distinct serum lipidomic signatures in hospitalized COVID-19-positive and COVID-19-negative patients

  1. Helena Castañé
  2. Simona Iftimie
  3. Gerard Baiges-Gaya
  4. Elisabet Rodríguez-Tomàs
  5. Andrea Jiménez-Franco
  6. Ana Felisa López-Azcona
  7. Pedro Garrido
  8. Antoni Castro
  9. Jordi Camps
  10. Jorge Joven

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Background

Lipids are involved in the interaction between viral infection and the host metabolic and immunological response. Several studies comparing the lipidome of COVID-19-positive hospitalized patients vs . healthy subjects have already been reported. It is largely unknown, however, whether these differences are specific to this disease. The present study compared the lipidomic signature of hospitalized COVID-19-positive patients with that of healthy subjects, and with COVID-19-negative patients hospitalized for other infectious/inflammatory diseases. Potential COVID-19 biomarkers were identified.

Methods

We analyzed the lipidomic signature of 126 COVID-19-positive patients, 45 COVID-19-negative patients hospitalized with other infectious/inflammatory diseases and 50 healthy volunteers. Results were interpreted by machine learning.

Results

We identified acylcarnitines, lysophosphatidylethanolamines, arachidonic acid and oxylipins as the most altered species in COVID-19-positive patients compared to healthy volunteers. However, we found similar alterations in COVID-19-negative patients. By contrast, we identified lysophosphatidylcholine 22:6-sn2, phosphatidylcholine 36:1 and secondary bile acids as the parameters that had the greatest capacity to discriminate between COVID-19-positive and COVID-19-negative patients.

Conclusion

This study shows that COVID-19 infection shares many lipid alterations with other infectious/inflammatory diseases, but differentiating them from the healthy population. Also, we identified some lipid species the alterations of which distinguish COVID-19-positive from Covid-19-negative patients. Our results highlight the value of integrating lipidomics with machine learning algorithms to explore the pathophysiology of COVID-19 and, consequently, improve clinical decision making.

Article activity feed

  1. ScreenIT

    SciScore for 10.1101/2021.12.14.21267764: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    EthicsIACUC: This study was approved by the Comitè d’Ètica i Investigació en Medicaments (Institutional Review Committee) of the Institut d’Investigació Sanitària Pere Virgili (Resolution CEIM 040/2018, modified on April 16, 2020). 2.2.
    Sex as a biological variablenot detected.
    RandomizationTo evaluate the diagnostic accuracy of different combinations of lipids, we constructed a Monte Carlo cross validation model that combined from 5 to 100 random variables, and subsequently calculated the area under the curve of the Receiver Operating Characteristics (ROC) curves, and confusion matrices [18].
    Blindingnot detected.
    Power Analysisnot detected.

    Table 2: Resources

    Software and Algorithms
    SentencesResources
    Lipids were then matched with the Metlin database (Scripps Research Institute, La Jolla, CA,) and quantified with calibration curves generated with internal standards. 2.3.
    Metlin
    suggested: (METLIN, RRID:SCR_010500)
    Statistical analyses: Statistical assessments were performed with the R program (RStudio version 4.0.5).
    RStudio
    suggested: (RStudio, RRID:SCR_000432)
    The MetaboAnalystR package was used to generate scores and loading plots and included False Discovery Rates (FDR), Volcano plots, Principal Component Analysis (PCA), Partial Least Square Discriminant Analysis (PLS-DA), and hierarchically clustered heatmaps [17].
    MetaboAnalystR
    suggested: (MetaboAnalystR, RRID:SCR_016723)

    Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

    Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.

    Results from TrialIdentifier: No clinical trial numbers were referenced.

    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

    Results from JetFighter: We did not find any issues relating to colormaps.

    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    Results from scite Reference Check: We found no unreliable references.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  2. Version published to 10.1101/2021.12.14.21267764 on medRxiv