The scent of COVID-19: viral (semi-)volatiles as fast diagnostic biomarkers?

'What would Hippocrates do?', one may ask in these desperate times in search for answers to the coronavirus disease (COVID-19) outbreak. Considering the lack of modern technology in the early days of medicine, he would probably rely on his senses and instruct physicians to smell the patient's breath for a diagnosis. Although physically smelling the breath of a potential contaminated patient is the last thing we would want to do right now due to the risk of transmission it entails, smelling with a modern twist, for instance through electronic noses or similar technology, might have some use in today's battle against the COVID-19 pandemic.

Coronaviruses (CoVs) are enveloped, positive single-stranded RNA viruses (+ssRNA) that classify into four genera: α-, β-, γ-, and δ-CoVs [1]. The severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), which currently causes COVID-19, is a pleiomorphic β-CoV of 60–140 nm in diameter [1, 2]. SARS-CoV-2 is transmitted by respiratory droplets that are released during coughing and sneezing and infection is characterised by a median incubation time of 5 days [3]. The early symptoms of COVID-19 are non-specific and mostly harmless, including coughing, sneezing, fever, and dyspnoea [4, 5]. However, SARS-CoV-2 infection can also lead to an uncontrolled systemic inflammatory reaction—a cytokine storm [6]. This can cause acute respiratory distress syndrome and multiple organ failure, and lead to the decease of the patient [1]. Currently, treatment of severely infected patients is only supportive and symptomatic, where oxygen therapy or mechanical ventilation is the main choice of intervention. The main focus lies within the preventive measures to flatten the incidence curve by isolation of patients, social distancing, hygienic measures and local lockdowns [5].

Patients experiencing symptoms are urged to contact their physicians and referred for SARS-CoV-2 diagnostic testing when COVID-19 is suspected. This causes a rush of anxious symptomatic patients to the hospitals, whether or not infected with SARS-CoV-2, putting a burden on the health care system. A key aspect is to get a good differential diagnosis with other infectious respiratory disorders (adenovirus, influenza, respiratory syncytial virus). Currently, COVID-19 is diagnosed by collecting specimens from either the upper (naso- and oropharyngeal swabs) or lower respiratory tract (induced sputum, endotracheal aspirate, bronchoalveolar lavage) using swabs. The genetic material extracted from these samples is then amplified by RT-PCR, searching for the SARS-CoV-2 genetic code. If positive, repeating the test for confirmation is recommended. Furthermore, in patients with COVID-19, this testing should be repeated to evaluate viral clearance, all taking time. Furthermore, concern has been raised about the false negative rate of RT-PCR and the sample techniques [7]. As current sampling techniques heavily rely on the skills of the person taking the sample and timing, it will introduce a variability within test results. A systematic review concluded that up to 29% of patients could have an initial false negative RT-PCR result [8]. Although this review has several limitations, such as a risk of bias and high heterogeneity, at least two other studies have reported similar results. One study showed an 11%, 27% and 40% false negative rate in respectively sputum, nasal and throat samples during the first 7 days after onset of the symptoms [9]. Another study reported a true positivity rate of 67% using RT-PCR in patients with acute respiratory symptoms and a chest CT indicative of COVID-19, while an antibody seroconversion of 93% was observed [10].

As the use of swabs is perceived as very unpleasant by the patient, a quick, non-invasive alternative would be beneficial. This is where breathomics enters the scene. Exhaled breath consists of a gaseous phase and a liquid phase. The gaseous phase, encloses N₂, CO₂ and volatile organic compounds (VOCs) present in picomolar concentrations. VOCs can arise from cellular in vivo metabolic activity and can also be induced by pathological processes, although one should be cautious not to mistake these VOCs for exogenously originating VOCs linked to drugs, diet, or environment that can go into equilibrium with the body. The liquid phase, on the other hand, encloses both exhaled breath condensate (EBC) and aerosols (EBA), containing a wide range of non-volatile molecules, such as cytokines, chemokines, hydrogen peroxide, ammonia, adenosine, leukotrienes, isoprostanes, nitrogen oxides, peptides, DNA and RNA (figure 1) [11, 12]. Analysis of EBC has shown to be a more recent, non-invasive technique allowing the detection of biomarkers, originating mainly from the lower respiratory tract [13]. It is collected during tidal breathing by cooling and condensation of the exhaled breath. The detection of inflammatory markers, reflecting chronic airway diseases, such as cystic fibrosis (CF), asthma and chronic obstructive pulmonary disease (COPD) have already been described in literature [13, 14]. Above of this, the identification of metabolomic, proteomic and genomic fingerprints of the exhaled breath for the early detection of both respiratory and systematic diseases, have recently gained a lot of interest. In addition to water, EBC also contains exhaled droplets, containing entrapped semi-volatile and non-volatile compounds, such as proteins, metabolites, smaller polar compounds, cellular fractions, fatty acids, cytokines, bacteria and viruses. These droplets or aerosols originate both from the lower airways due to surfactant disruption and from the upper airways by turbulence [12, 13, 15]. These droplets are released during exhalation, coughing and sneezing and can also be sampled and analysed separately. Sampling of EBA is possible by trapping the aerosols in a filter or sampling a worn facial mask, although in accordance to VOCs, caution is advised to control for exogenous exposures. The initial concept of analysing EBA arose from investigating how dogs were able to track humans long after they were gone and VOCs would have likely disappeared [16]. The importance of non-volatile compounds was further enhanced by the claim that trained dogs are able to diagnose cancer based on the smell of simple surgical masks worn by the patient [16, 17]. As VOCs are not expected to be captured by such a mask and should definitely not survive shipping and handling, the discriminative power is most likely linked to residual absorbed low volatile organics or dried aerosols. Identifying these chemicals to discriminate between samples could lead to the development of instruments capable of detecting the same compounds [12]. Although no studies based on EBA analysis of COVID-19 patients are ongoing yet, a proof of concept study using sweat samples of COVID-19 positive patients showed promising results, with diagnostic success rates ranging from 84% to 100% [18].

Zoom In Zoom Out Reset image size

Furthermore, the biochemistry of infection is a complex multistep process that induces the formation of VOCs (figure 1). First, SARS-CoV-2 binds the ACE2 receptor on the lower respiratory tract cells by its Spike(S)-glycoprotein [1, 19], inducing endocytosis. This fusion is accompanied by a decrease of pH in the endosome, resulting in specific changes in protein synthesis and VOC generation. Since cellular infection simultaneously affects many signal transduction and protein expression pathways, a large downstream VOC production effect can be expected. This viral entry differs from those of influenza or rhinovirus that use respectively sialic acids and Toll-like receptor 3 suggesting SARS-CoV-2-specific VOCs to be formed [20, 21]. Subsequently, the viral genome is released in the cytoplasm and translates into two polyproteins and structural proteins [1]. These form a replication-transcription complex wherein the viral genome begins to replicate, forming accessory and structural proteins. This involves multiple small molecules as cofactors, reactants, and (side)products which can cross the cell membrane and possibly be detected in the exhaled breath. Using the endoplasmic reticulum and Golgi apparatus, newly formed genomic RNA, nucleocapsid proteins and envelope glycoproteins assemble and form viral particle buds that fuse with the plasma membrane to release the virus in the airways.

Second, SARS-CoV-2 is taken up by dendritic cells that present their antigens, stimulating humoral and cellular immunity, and forming virus-specific B- and T-cells [1]. In particular, B-cells will likely produce unique and distinct VOCs upon infection as a result of the specific virus–cell interactions considering that cell lines with various HLA gene expression profiles exhibit unique VOCs as an immunologic fingerprint based on exposure to a specific antigen [21, 22].

Third, the immune response is vital for the control and resolution of SARS-CoV-2 infection, but it can possibly lead to a cytokine storm. The viral RNAs serve as pathogen-associated molecular patterns, and are detected by endosomal pattern recognition receptors. These trigger downstream cascade molecules that lead to the activation of NF-κB and the production of IFN-α and pro-inflammatory cytokines, inducing a cytokine storm in the body [6]. This leads to lung injury, and may be associated with the critical condition of COVID-19 patients, causing destruction of the cellular structure due to oxidative stress and liberating several additional VOCs that can change according to the severity of damage [11, 23].

VOCs could, hence, allow to assess pulmonary infection with COVID-19 and probably predict disease outcome, stressing their use as diagnostic and potential prognostic biomarkers. Differential diagnosis in early stages with influenza or rhinovirus will thus be key. And although infection will induce common VOCs, unique VOC fingerprints have been found in response to three different subtypes of influenza viruses, respiratory syncytial virus, and rhinovirus [20, 21] (table 1).

Taking the alternative infection pathway of SARS-CoV-2 into account, specific VOCs associated with COVID-19 severity are also to be expected. Establishing a correlation between the identified VOCs and the biological pathways related with viral infection, could lead to a better understanding of COVID-19 and provide novel treatment targets. A recent review by Oliver Gould et al gives a comprehensive overview of the different techniques that can be used for the detection of viral infection through breath analysis [26]. Mass spectrometry-based techniques are most used for the analysis of VOCs, however they often require extensive data manipulations, elaborate analyses and complex modelling. In contrast, handheld breathalyzers can establish a fast diagnosis by detecting a specific combination of VOCs linked to a certain disease, without analyzing single VOCs in detail [27]. We would therefore recommend to combine the best of both and use the handheld breathalyzer as a rapid, non-invasive, easy to use device in the clinic, while also collecting samples for further analysis of the VOC pattern and underlying pathophysiology and to aid in finetuning the breathalyzer.

In addition to VOCs, both EBC and EBA could play an important role in the diagnosis of COVID-19. Given the fact that viral particles are transmitted through respiratory droplets, EBC and EBA should be considered for diagnostics [28]. Both viral RNA and DNA of viruses, such as of rhinoviruses and influenza, have already been detected in this non-volatile fraction of the exhaled breath [29]. Furthermore Patrucco et al demonstrated a sensitivity and specificity of 66% and 100% respectively for the detection of non-herpes viruses as compared to bronchoalveolar lavage analysis [30]. Although the possibility of differential diagnosis should further be investigated, we believe that a distinction can be made through the difference in the genetic makeup of the viruses. Also, the aforementioned cytokines, which are released due to infection, have been detected in a reproducible manner in both EBC and EBA [12, 31]. As the levels of cytokines are related to the severity of the symptoms, it might have prognostic capacities. A recent study investigating cytokines as biomarkers of severity and prognosis for respiratory syncytial virus infection (hRSV) has linked IL-8, interferon-alpha (IFN-α), IL-6, thymic stromal lymphopoietin (TSLP), IL-3 and IL-33 to the severity of a hRSV infection [32].

In conclusion, both VOC as EBC and EBA analysis offer a new research field for non-invasive testing. The current diagnostic tests do not have a perfect specificity and sensitivity, leaving room for improvement. As sensitivity and specificity are extremely important for the development of a diagnostic test, linking VOCs to pathophysiology or the combination of both VOCs and EBC/EBA information might increase the clinical implementation. Another advantage is the access to an unlimited sample and the use of inert materials that are easy to clean or disposable. Above this, recommendations of how to minimize the risk of human-mediated transmission when performing breath research, have recently been published [33]. Since SARS-CoV-2 is sensitive to ultraviolet light and heat, the sampling material can be effectively decontaminated by cleaning with lipid solvents containing ether, ethanol, or chlorine-containing disinfectant, or be autoclaved, limiting cross-contamination of patients and medical staff. Above all, the method allows us to adhere to social distancing when 'smelling out' COVID-19. We therefore fully encourage researchers to explore volatiles and non-volatiles in exhaled breath for the non-invasive and broad screening for COVID-19. Although establishing both a high sensitivity and specificity will be challenging, we do believe that the advantages of breath analysis outweigh its challenges. Hippocrates' ancient medicine might just be the one that skyrockets the breakthroughs we need today.

The authors acknowledge "Kom op tegen kanker" (Stand up to Cancer), the Flemish cancer society.