Next Article in Journal
Possible Missing Sources of Atmospheric Glyoxal Part I: Phospholipid Oxidation from Marine Algae
Previous Article in Journal
Non-Targeted Metabolomics of White Rhinoceros Colostrum and Its Changes During Early Lactation by 1H Nuclear Magnetic Resonance Spectroscopy
Previous Article in Special Issue
Causal Metabolomic and Lipidomic Analysis of Circulating Plasma Metabolites in Autism: A Comprehensive Mendelian Randomization Study with Independent Cohort Validation
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Metabolites
Volume 14
Issue 11
10.3390/metabo14110638
metabolites-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Analysis of the Urine Volatilome of COVID-19 Patients and the Possible Metabolic Alterations Produced by the Disease

by
Jennifer Narro-Serrano
1,
Maruan Shalabi-Benavent
2,
José María Álamo-Marzo
2,
Álvaro Maximiliam Seijo-García
3 and
Frutos Carlos Marhuenda-Egea
3,*
1
Department of Physical Chemistry, University of Alicante, 03690 Alicante, Spain
2
Hospital Marina Baixa, 03570 Alicante, Spain
3
Department of Biochemistry and Molecular Biology and Soil Science and Agricultural Chemistry, University of Alicante, 03690 Alicante, Spain
*
Author to whom correspondence should be addressed.
Metabolites 2024, 14(11), 638; https://doi.org/10.3390/metabo14110638
Submission received: 13 September 2024 / Revised: 14 November 2024 / Accepted: 18 November 2024 / Published: 19 November 2024
(This article belongs to the Special Issue Advanced Metabolomics and Lipidomics Approaches in Studying Human Diseases)
Browse Figures
Versions Notes

Abstract

:
Alterations in metabolism caused by SARS-CoV-2 infection have been highlighted in various investigations and have been used to search for biomarkers in different biological matrices. However, the selected biomarkers vary greatly across studies. Our objective is to provide a robust selection of biomarkers, including results from different sample treatments in the analysis of volatile organic compounds (VOCs) present in urine samples from patients with COVID-19. Between September 2021 and May 2022, urine samples were collected from 35 hospitalized COVID-19 patients and 32 healthy controls. The samples were analyzed by headspace (HS) solid phase microextraction (SPME) coupled to gas chromatography–mass spectrometry (GC-MS). Analyses were conducted on untreated urine samples and on samples that underwent specific pretreatments: lyophilization and treatment with sulfuric acid. Partial Least Squares Linear Discriminant Analysis (PLS-LDA) and Subwindow Permutation Analysis (SPA) models were established to distinguish patterns between COVID-19 patients and healthy controls. The results identify compounds that are present in different proportions in urine samples from COVID-19 patients compared to those from healthy individuals. Analysis of urine samples using HS-SPME-GC-MS reveals differences between COVID-19 patients and healthy individuals. These differences are more pronounced when methods that enhance VOC formation are used. However, these pretreatments can cause reactions between sample components, creating additional products or removing compounds, so biomarker selection could be altered. Therefore, using a combination of methods may be more informative when evaluating metabolic alterations caused by viral infections and would allow for a better selection of biomarkers.

1. Introduction

The COVID-19 pandemic emerged in 2019 and, since then, more than 775 million cases have been confirmed worldwide, according to the World Health Organization [1]. It has been observed that around a third of patients may present some type of pathology two years after passing the acute phase of the disease [2]. Therefore, although the state of emergency was suspended on 5 May 2023, the disease continues to affect a large portion of the population. The set of symptoms that remain after the disease has been referred to as “long-COVID” or “post-COVID-19” [3]. Several of these symptoms are directly related to metabolism, but the underlying mechanisms have not yet been elucidated.
Among the strategies used to study metabolism, metabolomic studies stand out due to the large amount of information they provide. These studies can be conducted on different biological matrices, such as urine, blood, feces or breath. Urine is particularly useful because it is an easily accessible biofluid and has the advantage of being a concentrated biological sample.
Metabolomic studies have already been used in the analysis of COVID-19 patients using nuclear magnetic resonance (NMR) [4,5,6,7,8], high-performance liquid chromatography coupled with mass spectrometry (HPLC-MS) [9,10,11,12], or gas chromatography coupled with mass spectrometry (GC-MS) [13,14,15,16,17,18].
Various studies aim to distinguish COVID-19 patients from healthy individuals based on the analysis of volatile organic compounds (VOCs) [13,15,16,19,20,21,22,23,24]. However, the selected biomarkers vary greatly across investigations. To reduce errors in the subsequent chemometric analysis of VOCs, different techniques are often used to increase the signal of the peaks of interest and reduce interference from particles released by the fiber used in the GC-MS equipment. These techniques include adding a strong acid to promote protonation and the volatility of the compounds or lyophilizing the sample before analysis [25,26].
Our objective is to provide a more robust selection of biomarkers in the analysis of VOCs present in urine samples from COVID-19 patients, including results from different sample pretreatments. This is not a comparison of pretreatment methods but rather an integrated analysis of the results. To this end, VOCs present in urine samples from COVID-19 patients will be analyzed, and biomarkers that could aid in identifying the infection and assessing the possible metabolic alterations caused by the SARS-CoV-2 virus in the body will be sought. The samples will be analyzed by GC-MS using three different preparations: untreated samples, lyophilized samples, and samples treated with H2SO4.

2. Materials and Methods

2.1. Study Participants

Patient recruitment and sampling procedures were conducted in accordance with the Declaration of Helsinki, as well as applicable local regulatory requirements and laws, and after receiving approval from the Ethics Committee of the Hospital Universitario de San Juan (Alicante, Spain). Written informed consent was obtained from each participant before inclusion in this study. Urine samples from COVID-19 patients were obtained from the Hospital de la Marina Baixa (Vila Joiosa, Alicante, Spain). The first set of COVID-19 patient samples (n = 20) was collected between September and November 2021, when the Delta variant was predominant in Spain. The second set of samples (n = 15) was collected between February and May 2022, when the Omicron variant was dominant. Healthy control samples were obtained from the University of Alicante personnel and outpatients from the Hospital de la Marina Baixa with another chronic and common illness (Vila Joiosa, Alicante, Spain) (Table 1).

2.2. Sample Preparation and Gas Chromatography–Mass Spectrometry (GC–MS) Analysis

Human urine samples (first pass, morning, with 10–14 h of fasting and at least 4 to 6 h since the last urination) were collected from volunteers in 120 mL sterile urine specimen cups. Upon receipt (typically within 1 h of collection), all samples were stored at −20 °C. Urine samples were stored for two weeks while all testing was performed. Prior to analysis, the samples were thawed at room temperature for 30 min and centrifuged at 12,000 rpm for 5 min. Afterward, samples were subjected to three different sample treatments, i.e., no treatment, lyophilization, and H2SO4 addition. Samples without treatment were analyzed directly [25]. The required amount of NaCl (0.270 g) was added to 1 mL of urine in a 10 mL glass vial with a micro-stirring bar (PTFE bar, 5mm diameter, 15mm length). For the lyophilized samples, 1 mL of urine was lyophilized before being placed in glass vials [26]. For the samples containing H2SO4, 0.2 mL of 2.5 M H2SO4 was added to 1 mL of urine in a 10 mL glass vial [25].
The glass vials were sealed with aluminum crimp caps equipped with a needle-pierceable polytetrafluoroethylene/silicone septa. The solid phase microextraction (SPME) fiber used was divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) 50/30 µm, StableFlex, 1 cm long, mounted on an SPME automatic holder assembly (Supelco, Bellefonte, PA, USA). The sample vial was then placed in a thermostatic water bath at 45 °C and stirred at 500 rpm for 30 min. After this equilibration, the SPME needle was exposed to the VOCs in the headspace for 30 min. Once the extraction was complete, the SPME needle was removed, and the fiber was immediately desorbed into the GC-MS injection port at 250 °C for 10 min. To avoid carryover, an additional step was added to our GC method. After analyzing each sample, the fiber was introduced into the oven with a helium flow (20 mL/min) for 2 min at 250 °C. Additionally, at the beginning and end of each sample processing batch, a blank analysis (empty, capped vials) was performed.
Volatile analysis was carried out using an Agilent 6890N GC coupled to a 5973N MS (Agilent Technologies, Palo Alto, CA, USA) operating in electron impact ionization mode (EI 70 eV). The ion source and GC-MS transfer line temperatures were set to 230 °C and 280 °C, respectively. A DB-624 column (30 m length × 0.25 mm internal diameter × 1.4 µm film thickness, Agilent Technologies, Palo Alto, CA, USA) was used. The oven temperature was programmed as follows: 50 °C initial temperature with a hold for 3 min, then from 50 to 250 °C at 5 °C min−1, and finally hold for 10 min at 250 °C; the total run time was 53 min. Helium was used as the carrier gas (1 mL/min). Mass spectrum data were acquired in scan mode over a mass range of 30–350 mass-to-charge ratio (m/z), using MassHunter GCMS Acquisition B.07.06 (Agilent Technologies, Inc., 5301 Stevens Creek Boulevard. Santa Clara, CA, USA). Peak identification was based on a comparison of the mass spectrum data with spectra in the NIST/Wiley mass spectral library [27] using Enhanced ChemStation—MSD ChemStation F.01.03., Level 2 identification, according to the Metabolomics Standards Initiative (MSI) [28].

2.3. Statistical Analysis

GC-MS spectra were filtered for signals corresponding to the fiber and medications, such as the anesthetic Propofol, which was present in a great number of the samples from patients suffering from COVID-19 and hospitalized in intensive care units.
To better identify the VOCs, the raw data were exported to MATLAB R2024a [29] for mathematical processing. The maximum and minimum of the signals were marked, and an integration of these was performed, using, as a reference, the two minimum points on either side of peak’s maximum. Each segment was then reviewed to evaluate which signals were not artifacts due to baseline alterations in the chromatogram, nor were signals that appear only in urine samples from COVID-19 patients, which might indicate they are drugs used in treating the disease, such as propofol. We searched for signals present in both healthy individuals and COVID-19 patients that were sufficiently intense to enable correct classification.
Data analysis of the GC-MS spectra was performed using the PLS-LDA algorithm. PLS-LDA is a supervised method that groups data according to a mathematical model. This algorithm allows for the determination of whether data are correctly grouped and which properties (peaks in our GC-MS data) are important for accurate classification (COVID-19 patients or healthy controls). The statistical parameters used to determine the accuracy of the model were the R2Y, R2X, and AUC. Data were subjected to Pareto scaling prior to PLS-LDA; three components were selected [30,31]. The optimal number of latent variables of PLS-LDA was chosen by a five-fold cross validation [32].
In the GC-MS spectra, the peaks corresponding to VOCs were identified and integrated. The integrated data were classified with the SPA algorithm [32]. This algorithm ranks the variables based on the p-value. The PLS-LDA algorithm analyzes the ranking of each sample using a model created by excluding the sample in question from the complete set. This analysis is represented by tpScore [33]. We have confirmed that the model enables accurate classification across all samples and preparation methods for GC-MS.

3. Results

3.1. Samples Without Any Treatment

First, we analyzed the urine samples directly [25], without any alteration except for adding a small amount of sodium chloride to adjust the ionic strength of the urine and promote the formation of VOCs. The PLS-LDA analysis was used to assess VOC features capable of classifying the urine samples in the group of COVID-19 patients or the group of healthy controls. The signals obtained were not of high intensity, and the spectra had to be processed to eliminate the signals generated by the fiber itself, which absorbs the VOCs during the analysis. However, they enabled us to achieve a good separation between the two groups analyzed, COVID-19 patients and healthy individuals (Figure 1). As shown on the score plots (Figure 1A), there is a good separation of the two groups. The pseudospectral representation of the tpLoadings indicates signals of the GC-MS spectrum which are most important in the separation observed in the PLS-LDA model (Figure 1B). The peaks of the negative tpLoadings correspond to compounds present at higher concentrations in COVID-19 patients, while the positive tpLoadings correspond to compounds found at higher concentrations in healthy individuals. These volatile compounds may appear or disappear in urine as a result of the metabolic alterations caused by COVID-19 [5,8].
GC-MS spectra allow for effective classification, but identifying which peaks correspond to the different VOCs was difficult. To improve identification, once the peaks were identified, we integrate the region where each peak is located. Before applying any mathematical classification model, we eliminate the integrals associated with peaks that appear due to the degradation of the fiber used to absorb volatile compounds in the GC-MS technique. The fiber is made of a silicon compound (polydimethylsiloxane), which is released during the analysis itself.
To identify the VOC peaks that differ between the groups of COVID-19 patients and healthy controls, we used the SPA (Subwindow Permutation Analysis) algorithm [32]. This algorithm identifies and classifies the values of the peak integrals that are most important for distinguishing between the groups. SPA is based on the COSS (Conditional Synergistic Score) value for variable selection (Figure 2). COSS is defined as -log10(p-value); thus, a p-value of 0.05 corresponds to a COSS value of 1.3. Several variables have much higher COSS values, indicating p-values much lower than 0.05 (Figure 2).
Figure 3 shows the peaks contributing the most to the group separation, and their boxplot peak area. Among the peaks selected as most important following SPA classification, more intense peaks were found in both patients with COVID-19 and healthy controls, which would indicate alterations in metabolism [5,8].
Table 2 presents the 15 signals with the highest COSS values (lowest p-value), the compounds identified according to the Wiley library [27], the confidence percentage of the identification, and whether the compound was found at higher concentrations in COVID-19 or healthy controls group. These compounds are often present in urine samples and are associated with secondary metabolism or environmental contaminants [34,35,36,37,38]. In some samples of COVID-19 patients, signals associated with medications administered during hospital treatment were found, such as the anesthetic Propofol. Peaks associated with medication were eliminated from the statistical analysis.
Table 2. VOCs obtained from the analysis of untreated urine samples from COVID-19 patients and healthy controls. They are classified based on the SPA algorithm and their importance in classifying urine samples. The first four compounds in the table correspond to those shown in Figure 3.
Table 2. VOCs obtained from the analysis of untreated urine samples from COVID-19 patients and healthy controls. They are classified based on the SPA algorithm and their importance in classifying urine samples. The first four compounds in the table correspond to those shown in Figure 3.
Time (min)m/zCompoundConfidence (%)Higher in
26.07441642′-Hydroxy-4′,5′-dimethylacetophenona87COVID-19
34.2336206Phenol, 2,4-bis(1,1-dimethylethyl)97COVID-19
29.751208Decahydro-4,4,8,9,10-pentamethylnaphthalene49Control
13.2882106p-Xylene97Control
29.9481523-(But-3-enyl)-cyclohexanone53COVID-19
26.5193135Benzothiazole94COVID-19
9.520492Toluene95Control
25.8843721-(1-Propen-1-yl)-2-(2-thiopent-3-yl) disulfide72COVID-19
41.0242362,4-Diphenyl-4-methyl-2(E)-pentene95Control
24.4244154l-Menthone98Control
30.43162525-Octadecene, (E)-49COVID-19
32.0588170Methacrylic acid, tetradecyl ester74COVID-19
17.4058126Dimethyl trisulfide95COVID-19
31.43912001-Dodecanol, 2-methyl-, (S)-80COVID-19
30.28711382-(2,2-Dimethylvinyl)thiophene83COVID-19
The variables selected by the SPA algorithm can be evaluated as potential biomarkers. Since the signal intensities were low, we set out to evaluate other strategies in order to increase both the signal generated by the volatile compounds and the number of volatile compounds generated. Sample lyophilization and sulfuric acid addition were further performed as two distinct sample treatments [25,26], and following the same data analysis procedure applied for samples without treatment.

3.2. Lyophilized Samples

Urine samples were freeze-dried, as detailed in the Materials and Methods section. This method of sample pretreatment increases both the number of volatile compounds generated and the intensity of the signals, minimizing the impact of signals produced by the fiber on which the VOCs are adsorbed during the GC-MS analysis [26]. A good separation was obtained after PLS-LDA (Figure 4).
Although found to be less intense than in the analysis of untreated urine samples, signals corresponding to the fiber were also eliminated at this stage. We obtained a good selection of peaks, as illustrated in Figure 4B. Peak integrals were further obtained to improve the identification of urine samples from COVID-19 patients compared to those from healthy individuals. The data analysis included the SPA algorithm, which, with the COSS value, allows the selection of the most important variables (Figure 5) [32].
The representation of the selected regions and the corresponding peaks allows for a better visualization of the different selected volatile compounds of interest with a boxplot peak area (Figure 6).
Table 3 presents the 15 signals with the highest COSS values for the lyophilized urine samples, the corresponding identified compounds according to the Wiley library [27], the confidence percentage of the identification, and whether the compound is found at a higher concentration in the COVID-19 group or the healthy control group.
Table 3. VOCs obtained from the analysis of lyophilized urine samples from COVID-19 patients and healthy controls. They are classified based on the SPA algorithm and their importance in classifying urine samples. The first four compounds in the table correspond to those shown in Figure 6.
Table 3. VOCs obtained from the analysis of lyophilized urine samples from COVID-19 patients and healthy controls. They are classified based on the SPA algorithm and their importance in classifying urine samples. The first four compounds in the table correspond to those shown in Figure 6.
Time (min)m/zCompoundConfidence (%)Higher in
12.6913902,3-Butanediol80Control
28.8461113Caprolactam96COVID-19
13.58198Cyclopentanone, 3-methyl-96Control
28.5382148Benzaldehyde, 2,4,5-trimethyl-93COVID-19
14.934578Dimethyl Sulfoxide (DMSO)96Control
22.8465422H-Pyran-2-one, tetrahydro-90Control
25.16271Oxalic acid, 2-ethylhexyl hexyl ester64COVID-19
5.1405119Methane-d, trichloro-95COVID-19
16.9914170Decane93Control
15.7596962-Cyclopenten-1-one, 2-methyl-94Control
29.1009158Formamide, N,N-dibutyl-97COVID-19
12.851902,3-Butanediol90Control
26.4242135Benzothiazole93COVID-19
19.8201134Benzene, 1,2,3,4-tetramethyl-80COVID-19
15.1208108Pyrazine, 2,6-dimethyl-91Control

3.3. H2SO4 Samples

The urine samples were treated with sulfuric acid, as detailed in the Materials and Methods section [25] with the aim of increasing both the number of volatile compounds generated and the intensity of the signals while minimizing the impact of the signals generated by the fiber itself, on which the volatile compounds are adsorbed during GC-MS analysis [25]. A good separation was obtained after PLS-LDA modelling (Figure 7).
As before, signals from the fiber were eliminated before building the model with PLS-LDA [33]. We obtained a good selection of peaks, as shown in Figure 7B, with a good R2Y value (0.77).
To improve the selection and identification of variables associated with VOCs, we applied the SPA algorithm, which allows us to select the most important variables through the COSS value (Figure 8) [32].
This allows us to select a series of variables that correspond to the integral values (boxplot) of the peaks in the GC-MS spectrum of the urine samples (Figure 9).
Table 4 presents the 15 signals with the highest COSS values for urine samples treated with sulfuric acid, the corresponding identified compounds according to the Wiley library [27], the confidence percentage of the identification, and whether the compound is found at a higher concentration in the COVID-19 patient group or the healthy control group.
Table 4. VOCs obtained from the analysis of H2SO4-treated urine samples from COVID-19 patients and healthy controls. They are classified based on the SPA algorithm and their importance in classifying urine samples. The first four compounds in the table correspond to those shown in Figure 9.
Table 4. VOCs obtained from the analysis of H2SO4-treated urine samples from COVID-19 patients and healthy controls. They are classified based on the SPA algorithm and their importance in classifying urine samples. The first four compounds in the table correspond to those shown in Figure 9.
Time (min)m/zCompoundConfidence (%)Higher in
8.949894Disulfide, dimethyl (DMDS)97COVID-19
29.2831971,1,5-Trimethyl-1,2-dihydronaphthalene97Control
17.3637126Dimethyl trisulfide (DMTS)97COVID-19
22.2721124Phenol, 2-methoxy- (guaiacol)94Control
5.121284Methane-d, trichloro-94COVID-19
17.7782105Benzene, 1,2,4-trimethyl-90COVID-19
13.8925714-Heptanone91COVID-19
33.0889204Spiro[5.5]undeca-1,8-diene, 1,5,5,9-tetramethyl-, (R)-98COVID-19
29.1006224Cetene97COVID-19
25.49631382-Methoxy-5-methylphenol80Control
14.4474Butanoic acid, 3-methyl-81COVID-19
9.497391Toluene95COVID-19
29.00551591H-Inden-1-one, 2,3-dihydro-3,4,7-trimethyl-91COVID-19
20.9414170trans-Linalool oxide (furanoid)90Control
18.534768D-Limonene98COVID-19

4. Discussion

In 2010, it was proposed to use dogs for the detection of certain urological tumors and, during the 2020 pandemic, dogs were used to detect individuals infected with the COVID-19 virus. This suggested that the volatile compounds present in urine could be used as possible biomarkers of COVID-19 [19,39]. Changes in the composition of urine could be due to the metabolic alterations caused by SARS-CoV-2 infection [5,8]. Electronic noses could also be used to detect these volatile compounds in urine, but first, it is necessary to identify the biomarkers associated with the disease, as has been done in the detection of certain types of tumors [24,37,40]. Electronic noses also need to identify scent profiles to be effective, which would significantly limit their use in the case of COVID-19. In the present study, we have found a number of volatile compounds present in urine that could be considered biomarkers of COVID-19, using PLS-LDA models fit for discriminating individuals according to their status. The samples presented with an error rate of 0% and a sensitivity and specificity of 100% in all analyzed cases.
We performed three different treatments for the urine samples. Initially, a certain amount of NaCl was added to increase ionic strength and promote the formation of volatile compounds [25]. With the same objective, we freeze-dried the samples for the second treatment [26]. The third sample treatment consisted of sulfuric acid addition, as protonation of the molecules favors the formation of volatile compounds by decreasing the solubility of the protonated compounds [25]. In the first treatment, although the signal intensity was not very high, several compounds were identified (ID 2 according to the MSI), such as those listed in Table 2. The more intense signals in the urine of COVID-19 patients corresponded to 2′-hydroxy-4′,5′-dimethylacetophenone, 2,4-bis(1,1-dimethylethyl) phenol, and benzothiazole, which are aromatic compounds. In plants, compounds such as 2,4-bis(1,1-dimethylethyl) phenol appear in response to infections and are induced by arachidonic acid, playing an antioxidant role [41], but its role in human metabolism has not been identified. In healthy individuals, the signals corresponding to decahydro-4,4,8,9,10-pentamethylnaphthalene, toluene, and p-xylene were more intense. Some of these compounds, such as toluene, seem to have an exogenous origin, while others result from secondary metabolism [42,43,44], making it difficult to identify the metabolic pathways or precursors that generate them. We can assume that the differences between illness (COVID-19) and health generate distinct profiles in the spectra of volatile compounds, highlighting the metabolic alterations caused by COVID-19 [34,35,45,46].
For lyophilized samples, the spectra provided a greater number of volatile compounds and also allowed a good classification of the urine from COVID-19 patients using the PLS-LDA algorithm (Figure 4). When integrating the regions of the peaks, the SPA algorithm classifies these regions based on the p-value (Figure 5), allowing us to identify a series of potential biomarkers that are elevated in COVID-19 patients, such as caprolactam and benzaldehyde, 2,4,5-trimethyl, a biomarker in bladder cancer [47] or anxiety in mice [48]. In healthy individuals, we identified 2,3-butanediol; cyclopentanone, 3-methyl; and dimethyl sulfoxide (DMSO), or 2H-pyran-2-one, tetrahydro, as the most abundant compounds. 2,3-butanediol is a product of microbiota metabolism that contributes to reducing plasma cholesterol [49], and cyclopentanone has a protective effect against issues associated with diabetic nephropathy [50]. Other compounds, such as DMSO or 2H-pyran-2-one, tetrahydro, are products that do not form naturally and appear due to exposure [51].
With H2SO4-treated urine samples, another series of volatile compounds were found (Table 4). Some are present in higher proportions in samples from COVID-19 patients, such as disulfide, dimethyl (DMDS); dimethyl trisulfide (DMTS); benzene, 1,2,4-trimethyl-; and 4-heptanone. In healthy individuals, we predominantly have 1, 1, 5-trimethyl-1, 2-dihydronaphthalene; trans-linalool oxide (furanoid); or phenol, 2-methoxy- (guaiacol). As noted, many of these compounds may have an external origin, such as diet [34] or exposure to certain environments. For example, 1,2-dihydronaphthalene is related to exposure to naphthalene, although it can also be a product of the degradation of lutein or beta-carotene [42]. The fact that many volatile compounds have a dietary origin makes interpreting them as biomarkers for SARS-CoV-2 infection very challenging. Additionally, it is important to consider that the bacterial microflora also plays a significant role in generating these volatile compounds present in urine, such as DMDS [52,53,54]. The microbiota also appears to be altered during COVID-19. The enormous complexity of human metabolism, along with the microbiota [54], has been highlighted in studies analyzing volatile compounds in the urine of cancer patients [55,56].
As shown in Table 2, Table 3 and Table 4, the different pretreatments influence biomarker selection, favoring certain chemical families over others. In the analysis of untreated samples, hydrocarbons, ketones, organosulfur compounds, esters, and phenols are prominent. Without any additional treatment, this group presents a range of compounds that are inherently volatile and thermostable, making them suitable for GC-MS analysis. In the case of lyophilized samples, the primary compounds detected are alcohols, carboxylic acids, amines, lactams, and some halogenated compounds. The removal of water during lyophilization enhances the volatility of alcohols and carboxylic acids, improving their detectability in GC-MS [26]. Amines and nitrogen-containing compounds also benefit from lyophilization, as the process increases their concentration without altering their structure [26]. In contrast, samples treated with H2SO4 show the presence of phenols, organosulfur compounds, monoterpenes, ketones, and aromatic derivatives. The addition of sulfuric acid can protonate functional groups, particularly in phenols and amines, making some compounds more reactive [25]. This treatment can decompose unstable compounds, generating volatile derivatives or fragments that are detectable in GC-MS. Sulfuric acid may also induce dehydration in certain compounds, facilitating their volatilization [25].
Thus, while using a single sample treatment may be useful for creating a patient segregation model—allowing differentiation based on measurable changes in metabolites—a combination of different treatments is more valuable for a deeper study of disease-related metabolic alterations.

5. Conclusions

The analysis of urine samples using HS-SPME-GC–MS reveals differences between samples from COVID-19 patients and healthy individuals. These differences are clearer when methods enhancing the formation of volatile compounds, such as sample lyophilization or sulfuric acid addition, are used.
We have identified a number of compounds whose concentration could serve as biomarkers, but further studies for biomarker confirmation using reference standards are required, as well as biomarker confirmation in an independent cohort, in order to ascertain their use in COVID-19 infections.
It is also important to note that a single sample preparation method for GC-MS may be sufficient to classify patients as healthy or diseased but is very limited for studying metabolic alterations. Therefore, using a combination of methods may be more informative when evaluating metabolic alterations caused by viral infections and would allow for a better selection of biomarkers.

Author Contributions

F.C.M.-E. designed the experiment, performed the experiment, analyzed the data, wrote the manuscript, and generated the figures. J.N.-S. performed the experiment, analyzed the data, and wrote the manuscript. Á.M.S.-G. performed the experiment and reviewed the manuscript. M.S.-B. and J.M.Á.-M. collected clinical samples and reviewed the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the University of Alicante (POSTCOVID-10-09) and with the support of the Generalitat Valenciana (ACIF/2021/298).

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of the Hospital Universitario de San Juan (Alicante, Spain) (protocol code 21/014 Tut., 30 June 2021).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The data presented in this study are openly available in Marhuenda, Frutos (2024), “Urine GC-MS COVID-19”, Mendeley Data, V2, https://doi.org/10.17632/j3szbwjnnd.2.

Acknowledgments

The authors would like to thank P. Blasco and P. Candela (SSTTI of the University of Alicante) for technical assistance with headspace (HS) solid phase microextraction (SPME) coupled to gas chromatography–mass spectrometry (GC-MS). They also thank all of the staff members at Hospital Marina Baixa who contributed to sample collection and handling.

Conflicts of Interest

The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

References

  1. World Health Organization. WHO COVID-19 Dashboard. Available online: https://data.who.int/dashboards/covid19/cases?n=c (accessed on 24 September 2024).
  2. Fernandez-De-Las-Peñas, C.; Notarte, K.I.; Macasaet, R.; Velasco, J.V.; Catahay, J.A.; Ver, A.T.; Chung, W.; Valera-Calero, J.A.; Navarro-Santana, M. Persistence of post-COVID symptoms in the general population two years after SARS-CoV-2 infection: A systematic review and meta-analysis. J. Infect. 2024, 88, 77–88. [Google Scholar] [CrossRef] [PubMed]
  3. Soriano, J.B.; Murthy, S.; Marshall, J.C.; Relan, P.; Diaz, J.V. A clinical case definition of post-COVID-19 condition by a Delphi consensus. Lancet Infect. Dis. 2022, 22, e102–e107. [Google Scholar] [CrossRef] [PubMed]
  4. Ansone, L.; Rovite, V.; Brīvība, M.; Jagare, L.; Pelcmane, L.; Borisova, D.; Thews, A.; Leiminger, R.; Kloviņš, J. Longitudinal NMR-Based Metabolomics Study Reveals How Hospitalized COVID-19 Patients Recover: Evidence of Dyslipidemia and Energy Metabolism Dysregulation. Int. J. Mol. Sci. 2024, 25, 1523. [Google Scholar] [CrossRef] [PubMed]
  5. Marhuenda-Egea, F.C.; Narro-Serrano, J.; Shalabi-Benavent, M.J.; Álamo-Marzo, J.M.; Amador-Prous, C.; Algado-Rabasa, J.T.; Garijo-Saiz, A.M.; Marco-Escoto, M. A metabolic readout of the urine metabolome of COVID-19 patients. Metabolomics 2023, 19, 7. [Google Scholar] [CrossRef] [PubMed]
  6. Ghini, V.; Maggi, L.; Mazzoni, A.; Spinicci, M.; Zammarchi, L.; Bartoloni, A.; Annunziato, F.; Turano, P. Serum NMR Profiling Reveals Differential Alterations in the Lipoproteome Induced by Pfizer-BioNTech Vaccine in COVID-19 Recovered Subjects and Naïve Subjects. Front. Mol. Biosci. 2022, 9, 839809. [Google Scholar] [CrossRef]
  7. Holmes, E.; Wist, J.; Masuda, R.; Lodge, S.; Nitschke, P.; Kimhofer, T.; Loo, R.L.; Begum, S.; Boughton, B.; Yang, R.; et al. Incomplete Systemic Recovery and Metabolic Phenoreversion in Post-Acute-Phase Nonhospitalized COVID-19 Patients: Implications for Assessment of Post-Acute COVID-19 Syndrome. J. Proteome Res. 2021, 20, 3315–3329. [Google Scholar] [CrossRef]
  8. Bruzzone, C.; Bizkarguenaga, M.; Gil-Redondo, R.; Diercks, T.; Arana, E.; de Vicuña, A.G.; Seco, M.; Bosch, A.; Palazón, A.; Juan, I.S.; et al. SARS-CoV-2 Infection Dysregulates the Metabolomic and Lipidomic Profiles of Serum. iScience 2020, 23, 101645. [Google Scholar] [CrossRef]
  9. Frankevich, N.; Tokareva, A.; Chagovets, V.; Starodubtseva, N.; Dolgushina, N.; Shmakov, R.; Sukhikh, G.; Frankevich, V. COVID-19 Infection during Pregnancy: Disruptions in Lipid Metabolism and Implications for Newborn Health. Int. J. Mol. Sci. 2023, 24, 13787. [Google Scholar] [CrossRef]
  10. Lomova, N.; Dolgushina, N.; Tokareva, A.; Chagovets, V.; Starodubtseva, N.; Kulikov, I.; Sukhikh, G.; Frankevich, V. Past COVID-19: The Impact on IVF Outcomes Based on Follicular Fluid Lipid Profile. Int. J. Mol. Sci. 2022, 24, 10. [Google Scholar] [CrossRef]
  11. Lomova, N.; Chagovets, V.; Dolgopolova, E.; Novoselova, A.; Petrova, U.; Shmakov, R.; Frankevich, V. Altered amino acid profiles of the “mother–fetus” system in COVID-19. Bull. Russ. State Med Univ. 2022, 3, 51–60. [Google Scholar] [CrossRef]
  12. Lionetto, L.; Ulivieri, M.; Capi, M.; De Bernardini, D.; Fazio, F.; Petrucca, A.; Pomes, L.M.; De Luca, O.; Gentile, G.; Casolla, B.; et al. Increased kynurenine-to-tryptophan ratio in the serum of patients infected with SARS-CoV-2: An observational cohort study. Biochim. Biophys. Acta (BBA) Mol. Basis Dis. 2020, 1867, 166042. [Google Scholar] [CrossRef]
  13. Chuachaina, S.; Thaveesangsakulthai, I.; Sinsukudomchai, P.; Somboon, P.; Traipattanakul, J.; Torvorapanit, P.; Chatdarong, K.; Kulsing, C.; Nhujak, T. Identification of Volatile Markers in Sweat for COVID-19 Screening by Gas Chromatography-Mass Spectrometry. ChemistrySelect 2024, 9, e202304388. [Google Scholar] [CrossRef]
  14. Chen, X.; Gu, X.; Yang, J.; Jiang, Z.; Deng, J. Gas Chromatography–Mass Spectrometry Technology: Application in the Study of Inflammatory Mechanism in COVID-19 Patients. Chromatographia 2023, 86, 175–183. [Google Scholar] [CrossRef] [PubMed]
  15. Di Gilio, A.; Palmisani, J.; Picciariello, A.; Zambonin, C.; Aresta, A.; De Vietro, N.; Franchini, S.A.; Ventrella, G.; Nisi, M.R.; Licen, S.; et al. Identification of a characteristic VOCs pattern in the exhaled breath of post-COVID subjects: Are metabolic alterations induced by the infection still detectable? J. Breath Res. 2023, 17, 047101. [Google Scholar] [CrossRef] [PubMed]
  16. Woollam, M.; Angarita-Rivera, P.; Siegel, A.P.; Kalra, V.; Kapoor, R.; Agarwal, M. Exhaled VOCs can discriminate subjects with COVID-19 from healthy controls. J. Breath Res. 2022, 16, 036002. [Google Scholar] [CrossRef] [PubMed]
  17. Lv, L.; Jiang, H.; Chen, Y.; Gu, S.; Xia, J.; Zhang, H.; Lu, Y.; Yan, R.; Li, L. The faecal metabolome in COVID-19 patients is altered and associated with clinical features and gut microbes. Anal. Chim. Acta 2021, 1152, 338267. [Google Scholar] [CrossRef] [PubMed]
  18. Shi, D.; Yan, R.; Lv, L.; Jiang, H.; Lu, Y.; Sheng, J.; Xie, J.; Wu, W.; Xia, J.; Xu, K.; et al. The serum metabolome of COVID-19 patients is distinctive and predictive. Metabolism 2021, 118, 154739. [Google Scholar] [CrossRef]
  19. Angeletti, S.; Travaglino, F.; Spoto, S.; Pascarella, M.C.; Mansi, G.; De Cesaris, M.; Sartea, S.; Giovanetti, M.; Fogolari, M.; Plescia, D.; et al. COVID-19 sniffer dog experimental training: Which protocol and which implications for reliable sidentification? J. Med. Virol. 2021, 93, 5924–5930. [Google Scholar] [CrossRef]
  20. David, P.; Shoenfeld, Y. The Smell in COVID-19 Infection: Diagnostic Opportunities. Isr. Med. Assoc. J. 2020, 22, 401–403. [Google Scholar]
  21. Dickey, T.; Junqueira, H. Toward the use of medical scent detection dogs for COVID-19 screening. J. Am. Osteopat. Assoc. 2021, 121, 141–148. [Google Scholar] [CrossRef]
  22. Giovannini, G.; Haick, H.; Garoli, D. Detecting COVID-19 from Breath: A Game Changer for a Big Challenge. ACS Sensors 2021, 6, 1408–1417. [Google Scholar] [CrossRef] [PubMed]
  23. Ibrahim, W.; Cordell, R.L.; Wilde, M.J.; Richardson, M.; Carr, L.; Dasi, A.S.D.; Hargadon, B.; Free, R.C.; Monks, P.S.; Brightling, C.E.; et al. Diagnosis of COVID-19 by exhaled breath analysis using gas chromatography-mass spectrometry. ERJ Open Res. 2021, 7, 00139–2021. [Google Scholar] [CrossRef] [PubMed]
  24. Snitz, K.; Andelman-Gur, M.; Pinchover, L.; Weissgross, R.; Weissbrod, A.; Mishor, E.; Zoller, R.; Linetsky, V.; Medhanie, A.; Shushan, S.; et al. Proof of concept for real-time detection of SARS-CoV-2 infection with an electronic nose. PLoS ONE 2021, 16, e0252121. [Google Scholar] [CrossRef] [PubMed]
  25. Aggarwal, P.; Baker, J.; Boyd, M.T.; Coyle, S.; Probert, C.; Chapman, E.A. Optimisation of Urine Sample Preparation for Headspace-Solid Phase Microextraction Gas Chromatography-Mass Spectrometry: Altering Sample pH, Sulphuric Acid Concentration and Phase Ratio. Metabolites 2020, 10, 482. [Google Scholar] [CrossRef] [PubMed]
  26. Aggio, R.B.M.; Mayor, A.; Coyle, S.; Reade, S.; Khalid, T.; Ratcliffe, N.M.; Probert, C.S.J. Freeze-drying: An alternative method for the analysis of volatile organic compounds in the headspace of urine samples using solid phase micro-extraction coupled to gas chromatography—Mass spectrometry. Chem. Cent. J. 2016, 10, 9. [Google Scholar] [CrossRef]
  27. Wiley Science Solutions. Wiley Registry/NIST Mass Spectral Library 2023; John Wiley & Sons Ltd.: Hoboken, NJ, USA, 2023. [Google Scholar]
  28. Sumner, L.W.; Amberg, A.; Barrett, D.; Beale, M.H.; Beger, R.; Daykin, C.A.; Fan, T.W.-M.; Fiehn, O.; Goodacre, R.; Griffin, J.L.; et al. Proposed minimum reporting standards for chemical analysis. Metabolomics 2007, 3, 211–221. [Google Scholar] [CrossRef]
  29. The MathWorks Inc. MATLAB, version: 24.2 (R2024a); The MathWorks Inc.: Natick, MA, USA, 2022. [Google Scholar]
  30. Li, H.; Liang, Y.; Xu, Q.; Cao, D. Key wavelengths screening using competitive adaptive reweighted sampling method for multivariate calibration. Anal. Chim. Acta 2009, 648, 77–84. [Google Scholar] [CrossRef]
  31. Tang, L.; Peng, S.; Bi, Y.; Shan, P.; Hu, X. A New Method Combining LDA and PLS for Dimension Reduction. PLoS ONE 2014, 9, e96944. [Google Scholar] [CrossRef]
  32. Li, H.-D.; Zeng, M.-M.; Tan, B.-B.; Liang, Y.-Z.; Xu, Q.-S.; Cao, D.-S. Recipe for revealing informative metabolites based on model population analysis. Metabolomics 2010, 6, 353–361. [Google Scholar] [CrossRef]
  33. Li, H.-D.; Xu, Q.-S.; Liang, Y.-Z. libPLS: An integrated library for partial least squares regression and linear discriminant analysis. Chemom. Intell. Lab. Syst. 2018, 176, 34–43. [Google Scholar] [CrossRef]
  34. Amann, A.; Costello, B.d.L.; Miekisch, W.; Schubert, J.; Buszewski, B.; Pleil, J.; Ratcliffe, N.; Risby, T. The human volatilome: Volatile organic compounds (VOCs) in exhaled breath, skin emanations, urine, feces and saliva. J. Breath Res. 2014, 8, 034001. [Google Scholar] [CrossRef] [PubMed]
  35. de Lacy Costello, B.; Amann, A.; Al-Kateb, H.; Flynn, C.; Filipiak, W.; Khalid, T.; Osborne, D.; Ratcliffe, N.M. A review of the volatiles from the healthy human body. J. Breath Res. 2014, 8, 014001. [Google Scholar] [CrossRef] [PubMed]
  36. Daulton, E.; Wicaksono, A.N.; Tiele, A.; Kocher, H.M.; Debernardi, S.; Crnogorac-Jurcevic, T.; Covington, J.A. Volatile organic compounds (VOCs) for the non-invasive detection of pancreatic cancer from urine. Talanta 2021, 221, 121604. [Google Scholar] [CrossRef] [PubMed]
  37. Drabińska, N.; Flynn, C.; Ratcliffe, N.; Belluomo, I.; Myridakis, A.; Gould, O.; Fois, M.; Smart, A.; Devine, T.; Costello, B.P.J.d.L. A literature survey of all volatiles from healthy human breath and bodily fluids: The human volatilome. J. Breath Res. 2021, 15, 034001. [Google Scholar] [CrossRef] [PubMed]
  38. Sharma, A.; Kumar, R.; Varadwaj, P. Smelling the Disease: Diagnostic Potential of Breath Analysis. Mol. Diagn. Ther. 2023, 27, 321–347. [Google Scholar] [CrossRef]
  39. Grandjean, D.; Sarkis, R.; Lecoq-Julien, C.; Benard, A.; Roger, V.; Levesque, E.; Bernes-Luciani, E.; Maestracci, B.; Morvan, P.; Gully, E.; et al. Can the detection dog alert on COVID-19 positive persons by sniffing axillary sweat samples? A proof-of-concept study. PLoS ONE 2020, 15, e0243122. [Google Scholar] [CrossRef]
  40. Lamote, K.; Brinkman, P.; Vandermeersch, L.; Vynck, M.; Sterk, P.J.; Van Langenhove, H.; Thas, O.; Van Cleemput, J.; Nackaerts, K.; van Meerbeeck, J.P. Breath analysis by gas chromatography-mass spectrometry and electronic nose to screen for pleural mesothelioma: A cross-sectional case-control study. Oncotarget 2017, 8, 91593–91602. [Google Scholar] [CrossRef]
  41. Teresa, R.-C.M.; Rosaura, V.-G.; Elda, C.-M.; Ernesto, G.-P. The avocado defense compound phenol-2,4-bis (1,1-dimethylethyl) is induced by arachidonic acid and acts via the inhibition of hydrogen peroxide production by pathogens. Physiol. Mol. Plant Pathol. 2014, 87, 32–41. [Google Scholar] [CrossRef]
  42. Marais, J. 1, 1,6-Trimethyl-1,2-dihydronaphthalene (TDN): A Possible Degradation Product of Lutein and beta-Carotene. S. Afr. J. Enol. Vitic. 1992, 13, 52–55. [Google Scholar] [CrossRef]
  43. Ramos, G.; Antón, A.P.; Sánchez, M.d.N.; Pavón, J.L.P.; Cordero, B.M. Urinary volatile fingerprint based on mass spectrometry for the discrimination of patients with lung cancer and controls. Talanta 2017, 174, 158–164. [Google Scholar] [CrossRef]
  44. Wu, R.; Waidyanatha, S.; Henderson, A.P.; Serdar, B.; Zheng, Y.; Rappaport, S.M. Determination of dihydroxynaphthalenes in human urine by gas chromatography–mass spectrometry. J. Chromatogr. B 2005, 826, 206–213. [Google Scholar] [CrossRef] [PubMed]
  45. Agapiou, A.; Amann, A.; Mochalski, P.; Statheropoulos, M.; Thomas, C. Trace detection of endogenous human volatile organic compounds for search, rescue and emergency applications. TrAC Trends Anal. Chem. 2015, 66, 158–175. [Google Scholar] [CrossRef]
  46. Broza, Y.Y.; Mochalski, P.; Ruzsanyi, V.; Amann, A.; Haick, H. Hybrid Volatolomics and Disease Detection. Angew. Chem. Int. Ed. 2015, 54, 11036–11048. [Google Scholar] [CrossRef] [PubMed]
  47. Lett, L.; George, M.; Slater, R.; Costello, B.D.L.; Ratcliffe, N.; García-Fiñana, M.; Lazarowicz, H.; Probert, C. Investigation of urinary volatile organic compounds as novel diagnostic and surveillance biomarkers of bladder cancer. Br. J. Cancer 2022, 127, 329–336. [Google Scholar] [CrossRef] [PubMed]
  48. Fujita, A.; Ihara, K.; Kawai, H.; Obuchi, S.; Watanabe, Y.; Hirano, H.; Fujiwara, Y.; Takeda, Y.; Tanaka, M.; Kato, K. A novel set of volatile urinary biomarkers for late-life major depressive and anxiety disorders upon the progression of frailty: A pilot study. Discov. Ment. Health 2022, 2, 20. [Google Scholar] [CrossRef]
  49. Veeravalli, S.; Scott, F.H.; Varshavi, D.; Pullen, F.S.; Veselkov, K.; Phillips, I.R.; Everett, J.R.; Shephard, E.A. Treatment of wild-type mice with 2,3-butanediol, a urinary biomarker of Fmo5−/− mice, decreases plasma cholesterol and epididymal fat deposition. Front. Physiol. 2022, 13, 859681. [Google Scholar] [CrossRef]
  50. Tang, C.; Wang, M.; Liu, J.; Zhang, C.; Li, L.; Wu, Y.; Chu, Y.; Wu, D.; Liu, H.; Yuan, X. A Cyclopentanone Compound Attenuates the Over-Accumulation of Extracellular Matrix and Fibrosis in Diabetic Nephropathy via Downregulating the TGF-β/p38MAPK Axis. Biomedicines 2022, 10, 3270. [Google Scholar] [CrossRef]
  51. Barupal, D.K.; Fiehn, O. Generating the Blood Exposome Database Using a Comprehensive Text Mining and Database Fusion Approach. Environ. Health Perspect. 2019, 127, 97008. [Google Scholar] [CrossRef]
  52. Meldau, D.G.; Meldau, S.; Hoang, L.H.; Underberg, S.; Wunsche, H.; Baldwin, I.T. Dimethyl disulfide produced by the naturally associated bacterium bacillus sp b55 promotes nicotiana attenuata growth by enhancing sulfur nutrition. Plant Cell 2013, 25, 2731–2747. [Google Scholar] [CrossRef]
  53. Thorn, R.M.S.; Greenman, J. Microbial volatile compounds in health and disease conditions. J. Breath Res. 2012, 6, 024001. [Google Scholar] [CrossRef]
  54. Zhao, J.; Gao, J.; Jin, X.; You, J.; Feng, K.; Ye, J.; Chen, J.; Zhang, S. Superior dimethyl disulfide degradation in a microbial fuel cell: Extracellular electron transfer and hybrid metabolism pathways. Environ. Pollut. 2022, 315, 120469. [Google Scholar] [CrossRef] [PubMed]
  55. Taunk, K.; Porto-Figueira, P.; Pereira, J.A.M.; Taware, R.; da Costa, N.L.; Barbosa, R.; Rapole, S.; Câmara, J.S. Urinary Volatomic Expression Pattern: Paving the Way for Identification of Potential Candidate Biosignatures for Lung Cancer. Metabolites 2022, 12, 36. [Google Scholar] [CrossRef] [PubMed]
  56. Liu, Q.; Li, S.; Li, Y.; Yu, L.; Zhao, Y.; Wu, Z.; Fan, Y.; Li, X.; Wang, Y.; Zhang, X.; et al. Identification of urinary volatile organic compounds as a potential non-invasive biomarker for esophageal cancer. Sci. Rep. 2023, 13, 18587. [Google Scholar] [CrossRef] [PubMed]
Metabolites 14 00638 g001
Figure 1. (A) Score plots of the three components of the PLS-LDA model for GC-MS spectra of urine samples from COVID-19 patients (red diamonds) and healthy controls (blue circles); (B) tpLoadings pseudospectrum from PLS-LDA model for GC-MS spectra of urine samples from COVID-19 patients and healthy controls. The intensity of the peaks (positive or negative) in the pseudospectrum is determined by the most significant spectral shift regions in the PLS-LDA model. The model performances were the following: R2X = 0.35, R2Y = 0.86, and AUC = 1. (*) Selected peaks appearing in Table 2.
Figure 1. (A) Score plots of the three components of the PLS-LDA model for GC-MS spectra of urine samples from COVID-19 patients (red diamonds) and healthy controls (blue circles); (B) tpLoadings pseudospectrum from PLS-LDA model for GC-MS spectra of urine samples from COVID-19 patients and healthy controls. The intensity of the peaks (positive or negative) in the pseudospectrum is determined by the most significant spectral shift regions in the PLS-LDA model. The model performances were the following: R2X = 0.35, R2Y = 0.86, and AUC = 1. (*) Selected peaks appearing in Table 2.
Metabolites 14 00638 g001
Metabolites 14 00638 g002
Figure 2. COSS values of the variables selected by the SPA algorithm as most important for determining the separation between groups of untreated urine samples from COVID-19 patients and healthy controls.
Figure 2. COSS values of the variables selected by the SPA algorithm as most important for determining the separation between groups of untreated urine samples from COVID-19 patients and healthy controls.
Metabolites 14 00638 g002
Metabolites 14 00638 g003
Figure 3. Boxplots of the integrals of the GC-MS spectrum regions and the peaks selected by the SPA algorithm as most important for determining the separation between COVID-19 patients and healthy controls.
Figure 3. Boxplots of the integrals of the GC-MS spectrum regions and the peaks selected by the SPA algorithm as most important for determining the separation between COVID-19 patients and healthy controls.
Metabolites 14 00638 g003
Metabolites 14 00638 g004
Figure 4. (A) Score plots of the three components of the PLS-LDA model for the GC-MS spectra of lyophilized urine samples from COVID-19 patients (red diamonds) and healthy controls (blue circles); (B) tpLoadings pseudospectrum from the PLS-LDA model for the GC-MS spectra of urine samples from COVID-19 patients and healthy controls. The intensity of the peaks (positive or negative) in the pseudospectrum is determined by the most significant spectral shift regions in the PLS-LDA model. The model performances were the following: R2X = 0.22, R2Y = 0.87, and AUC =1. (*) Selected peaks appearing in Table 3.
Figure 4. (A) Score plots of the three components of the PLS-LDA model for the GC-MS spectra of lyophilized urine samples from COVID-19 patients (red diamonds) and healthy controls (blue circles); (B) tpLoadings pseudospectrum from the PLS-LDA model for the GC-MS spectra of urine samples from COVID-19 patients and healthy controls. The intensity of the peaks (positive or negative) in the pseudospectrum is determined by the most significant spectral shift regions in the PLS-LDA model. The model performances were the following: R2X = 0.22, R2Y = 0.87, and AUC =1. (*) Selected peaks appearing in Table 3.
Metabolites 14 00638 g004
Metabolites 14 00638 g005
Figure 5. COSS values of the variables selected by the SPA algorithm as most important for determining the separation between groups of lyophilized urine samples from COVID-19 patients and healthy controls.
Figure 5. COSS values of the variables selected by the SPA algorithm as most important for determining the separation between groups of lyophilized urine samples from COVID-19 patients and healthy controls.
Metabolites 14 00638 g005
Metabolites 14 00638 g006
Figure 6. Boxplots of the integrals of the GC-MS spectrum regions and the peaks selected by the SPA algorithm as most important for determining the separation between COVID-19 patients and healthy controls.
Figure 6. Boxplots of the integrals of the GC-MS spectrum regions and the peaks selected by the SPA algorithm as most important for determining the separation between COVID-19 patients and healthy controls.
Metabolites 14 00638 g006
Metabolites 14 00638 g007
Figure 7. (A) Score plots of the three components of the PLS-LDA model for GC-MS spectra of H2SO4-treated urine samples from COVID-19 patients (red diamonds) and healthy controls (blue circles); (B) tpLoadings pseudospectrum from the PLS-LDA model for GC-MS spectra of urine samples from COVID-19 patients and healthy controls. The intensity of the peaks (positive or negative) in the pseudospectrum is determined by the most significant spectral shift regions in the PLS-LDA model. The model performances were the following: R2X = 0.25, R2Y = 0.77, and AUC =1. (*) Selected peaks appearing in Table 4.
Figure 7. (A) Score plots of the three components of the PLS-LDA model for GC-MS spectra of H2SO4-treated urine samples from COVID-19 patients (red diamonds) and healthy controls (blue circles); (B) tpLoadings pseudospectrum from the PLS-LDA model for GC-MS spectra of urine samples from COVID-19 patients and healthy controls. The intensity of the peaks (positive or negative) in the pseudospectrum is determined by the most significant spectral shift regions in the PLS-LDA model. The model performances were the following: R2X = 0.25, R2Y = 0.77, and AUC =1. (*) Selected peaks appearing in Table 4.
Metabolites 14 00638 g007
Metabolites 14 00638 g008
Figure 8. COSS values of the variables selected by the SPA algorithm as most important for determining the separation between groups of H2SO4-treated urine samples from COVID-19 patients and healthy controls.
Figure 8. COSS values of the variables selected by the SPA algorithm as most important for determining the separation between groups of H2SO4-treated urine samples from COVID-19 patients and healthy controls.
Metabolites 14 00638 g008
Metabolites 14 00638 g009
Figure 9. Boxplots of the integrals of the GC-MS spectrum regions and the peaks selected by the SPA algorithm as most important for determining the separation between COVID-19 patients and healthy controls.
Figure 9. Boxplots of the integrals of the GC-MS spectrum regions and the peaks selected by the SPA algorithm as most important for determining the separation between COVID-19 patients and healthy controls.
Metabolites 14 00638 g009
Table 1. Population and clinical data of the people whose urine have been analyzed.
Table 1. Population and clinical data of the people whose urine have been analyzed.
Healthy Controls (n = 32)COVID-19 Patients (n = 35)
Age [Median (IQR)]52.5 (19.1)59 (20.0)
Sex, distribution
    Male [n (%)]9 (28.1)11 (31.4)
    Female [n (%)]23 (71.9)24 (68.6)
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Narro-Serrano, J.; Shalabi-Benavent, M.; Álamo-Marzo, J.M.; Seijo-García, Á.M.; Marhuenda-Egea, F.C. Analysis of the Urine Volatilome of COVID-19 Patients and the Possible Metabolic Alterations Produced by the Disease. Metabolites 2024, 14, 638. https://doi.org/10.3390/metabo14110638

AMA Style

Narro-Serrano J, Shalabi-Benavent M, Álamo-Marzo JM, Seijo-García ÁM, Marhuenda-Egea FC. Analysis of the Urine Volatilome of COVID-19 Patients and the Possible Metabolic Alterations Produced by the Disease. Metabolites. 2024; 14(11):638. https://doi.org/10.3390/metabo14110638

Chicago/Turabian Style

Narro-Serrano, Jennifer, Maruan Shalabi-Benavent, José María Álamo-Marzo, Álvaro Maximiliam Seijo-García, and Frutos Carlos Marhuenda-Egea. 2024. "Analysis of the Urine Volatilome of COVID-19 Patients and the Possible Metabolic Alterations Produced by the Disease" Metabolites 14, no. 11: 638. https://doi.org/10.3390/metabo14110638

APA Style

Narro-Serrano, J., Shalabi-Benavent, M., Álamo-Marzo, J. M., Seijo-García, Á. M., & Marhuenda-Egea, F. C. (2024). Analysis of the Urine Volatilome of COVID-19 Patients and the Possible Metabolic Alterations Produced by the Disease. Metabolites, 14(11), 638. https://doi.org/10.3390/metabo14110638

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop