A Metabolomics approach for the diagnosis Of SecondAry InfeCtions in COVID-19 (MOSAIC): a study protocol

The novel coronavirus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) continues to present a major healthcare burden globally, with over 170,000,000 confirmed cases and 3,500,000 deaths worldwide [1]. Coronavirus Disease 2019 (COVID-19) has a spectrum of severity, and while the vast majority of cases result in a minor self-limiting illness, approximately 5% of patients will become critically unwell with severe respiratory failure which can progress to sepsis and multiple organ failure [2, 3]. Despite recent advances in treatment strategies, the mortality rate for COVID-19 patients admitted to the intensive care unit (ICU) in Scotland remains high at around 35% [4, 5].

The incidence of secondary infections in hospitalised patients with COVID-19 appears relatively low with rates between 6 and 15% [6‐8]. However the rates are significantly higher in critically ill patients and carry a mortality rate of around 50% [8‐10]. The clinical features of COVID-19 such as pyrexia, cough and dyspnoea are non-specific and are also observed in bacterial pneumonia [11]. It can therefore be difficult to make a diagnosis of secondary infection by clinical means [12]. Biomarkers such as lactate, C-reactive protein (CRP) and procalcitonin have a well-established role in identifying septic patients who are at risk of further deterioration, but as of yet a specific biomarker to detect the presence of a secondary infection remains elusive [13‐15]. As such, liberal use of broad spectrum antibiotics has been observed in critically ill COVID-19 patients [2]. The World Health Organisation (WHO) and the Surviving Sepsis Campaign both recommend initiating empirical antibiotics for all severe cases of COVID-19 [12, 16], whereas the National Institutes of Health (NIH) and the National Institute for Health and Care Excellence (NICE) suggest only starting antibiotics when there is a clear clinical suspicion of a secondary infection [17, 18]. Early initiation of antibiotic therapy has been shown to reduce mortality in bacterial sepsis [19], however unnecessary use of broad spectrum antibiotics increases the risk of side effects and promotes antimicrobial resistance [20]. Prompt diagnosis of a secondary infection and identification of the causative pathogen is therefore important in optimal management. Microbiological cultures can guide antimicrobial therapy, but these are time-consuming, and have a low sensitivity [21]. Alternative diagnostic strategies to improve sensitivity and provide rapid specificity would therefore be valuable.

All infections have the potential to cause major disruption to physiological processes, potentially leading to a degree of metabolic dysfunction [22, 23]. This can lead to alteration of the normal serum metabolome (all low molecular weight metabolites less than 1 kDa circulating in the bloodstream at a given point in time) [24]. The physiological response to an infection can result in depletion of certain important nutrients, while simultaneously causing accumulation of other toxic by-products [23]. As such, variation in the composition of the metabolome can be indicative of pathological processes occurring further upstream [24].

Metabolomics is a discipline which is gaining much traction as a potential diagnostic tool. Metabolomic analysis using techniques such as liquid chromatography–mass spectrometry (LC–MS) can be used to provide a metabolic profile of a patient. This acts as a ‘snapshot’ of the patient’s metabolome, providing a description of the metabolic state of a patient as a result of both genetic contributions and environmental factors [24].

Metabolomics has been investigated for aiding diagnosis of a wide variety of diseases, however in recent years particular focus has been given to its use in identifying biomarkers for infections and sepsis, with several candidate metabolites showing promise. Derangement of fatty acid metabolism has been observed during a septic response, with increased levels of acylcarnitines in septic patients compared with non-septic controls, as well as lower levels of lysophosphatidylcholines in septic shock patients who responded poorly to initial therapy [25‐28]. Animal and human studies have shown that changes in levels of tricarboxylic acid cycle intermediates can occur in response to infection. Derangement of concentrations of citrate, malate and succinate have been observed, particularly in response to gram positive infections including Staph. aureus and Strep. Pneumoniae [29‐31]. Increases in protein catabolism and amino acid degradation have also been reported during a septic inflammatory response. Multiple studies have shown depletion of the essential amino acid tryptophan leading to a detectable accumulation of the toxic metabolite kynurenine [25, 32, 33]. The biogenic amine trimethylamine-N-oxide (TMAO) has also shown promise as a biomarker of infection. Gut microflora metabolise quaternary ammonium compounds such as betaine and choline to trimethylamine (TMA), which is then converted to TMAO in the liver. TMAO levels are thus indirectly dependant on a functioning gut microbiome [34, 35].

The wide range of possibilities in terms of site of infection, organism type and physiological response means that the resulting metabolic changes will vary from patient to patient. As a result, a single biomarker is unlikely to be sufficient to diagnose a secondary infection. Consideration of groups, or panels of metabolites may be more sensitive for diagnosing secondary infections than individual biomarkers [36]. Moreover, different types of infections may produce a unique signature metabolic profile, thus potentially making identification of the specific organism possible [37].

In critically ill patients with COVID-19, examination of metabolic profiles may permit characterisation of novel biomarkers. These could permit detection of specific underlying infective organisms more rapidly than traditional culture methods. This would allow for earlier initiation of targeted antibiotic therapy, potentially improving outcomes in critically ill patients with COVID-19.

The primary aim of this study is to ascertain the diagnostic capability of metabolomics for identifying secondary infections in critically ill patients with COVID-19 compared with routinely collected markers of infection.

Secondary objectives will compare profiles between the following subgroups:

This study aims to determine the potential diagnostic capability of metabolomics in diagnosing secondary infections in critically ill patients with COVID-19. We hypothesise that metabolomic profiling of patients with COVID-19 may permit identification of those with secondary infections. As such it is hoped that metabolomic profiling may provide additional perspectives on the underlying pathogen, thus aiding targeted antimicrobial therapy.

A multi-centre prospective diagnostic observational study.

Inclusion criteria:

Patients over the age of 18 with a confirmed positive SARS-CoV-2 test in the previous 7 days who are admitted for at least level 2 care will be eligible for inclusion.

Exclusion criteria:

Patients will be excluded if they cannot provide informed consent directly or by their nearest relative if they are incapacitated. Pregnant patients will be excluded, as will those who are deemed to be end of life.

Taking into consideration the number of intensive care admissions during the first wave of the COVID-19 pandemic, we aim to recruit 75–100 critically ill COVID-19 patients and 50 healthy volunteers within a period of 1 year.

The first patient was recruited in November 2020, and recruitment will continue until December 2021, or until a sufficient number of patients have been recruited to provide 50 healthy control samples and 300 critically ill COVID-19 samples.

Primary endpoint

Secondary endpoints

Patients eligible for recruitment will be identified and approached by the local research team upon their admission to the critical care department. Eligibility will be assessed against the inclusion and exclusion criteria by a doctor with an up-to-date Good Clinical Practice certificate.

Informed consent will be obtained for all patients recruited to the study. Where possible, consent will be sought directly from the patient. Patients will be asked to sign two copies of the consent form: one which they will retain, and a second to be filed in their medical notes. The consent forms will be counter-signed by the researcher gaining consent. Where patients are incapacitated, their nearest relative will be approached for consent. This will be discussed either in person or by telephone. Patients who regain capacity after recruitment will be approached to discuss consent for ongoing participation in the study.

Recruited patients will have serial blood samples collected at three time points: Day 0, day 3 and day 10. In addition, a blood sample will be collected if participants show evidence of a secondary infection. This includes growth of organisms from microbial cultures, or clinical deterioration which results in a septic screen being sent by the clinical team.

Blood samples will be centrifuged within 1 h of sample collection. Serum will be aliquoted and frozen at – 80 °C for storage. Metabolites will be extracted in chloroform:methanol:water, 1:3:1, v/v/v solution. Untargeted metabolomic analysis will be undertaken using LC–MS in both positive and negative ion modes on a Thermo Orbitrap system interfaced to a Dionex UltiMate 3000 Rapid Separation LC platform with a zwitterionic polymeric hydrophilic interaction chromatography (ZIC-pHILIC) column and ammonium carbonate in water/acetonitrile gradient. The raw LC–MS data sets will be processed by in-house bioinformaticians through an established pipeline, which uses XCMS and MZMatch [38‐40]. Core metabolite identifications will be validated by comparison of the chromatographic retention times and the mass to charge values (m/z) values against a panel of authentic standards.

Relevant patient data will be extracted from clinical notes. Data will be collected by the local research team using standardised data collection forms. Completed forms will be transcribed to a password protected electronic spreadsheet kept on a secure server. Paper forms will then be stored securely in the site file within a locked office at the participating site. All data will be anonymised at site and a unique study number will be allocated to each patient.

The following clinical data will be collected:

Common contaminants of blood cultures such as coagulase-negative staphylococci, diphtheroids, Bacillus species or Propionibacterium species will not be considered positive microbiological cultures unless they are cultured from serial samples and treatment is initiated.

Statistical analysis of the untargeted metabolomic data sets will be guided by in-house bioinformaticians and will consist of univariate- and multivariate analyses.

Patient confidentiality will be maintained throughout the study as per the Data Protection Act. Recruited patients will be assigned a unique study number. All data collection sheets, study reports and communication regarding the study will identify the patients by this study number and initials only.

This is an observational study, and so adverse events are unlikely to occur. Any adverse events or serious adverse events will be recorded on case report forms, and an annual summary will be given to the ethics committee and research and development office.

A variation of the approved protocol will be considered a protocol deviation. Any protocol deviations will be recorded on case report forms and reported to the sponsor.

Authorisation for all publications will be sought from the study chief investigator and will be reviewed by the sponsor prior to publication.