Body Volatilome Study Strategy for COVID-19 Biomarker Identification Considering Exogenous Parameters

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript provid a useful approach for quick detection of possible COVID-19 patient.

I have some concerns.

Abstract

25 compounds highlighted, 13 associated with cosmetics and 7 linked to air pollution excluded, 4 compounds retained.

Looking at Figure 1, there was also 1 compound which was not mentioned in the abstract. Was it benzoic acid, 2,4-dichloro- or the underexpressed compound B5? It is not clear. Please specify.

 

The introduction provides detailed background information and relevant references, giving sufficient foundational support for this study and demonstrating its significance.

 

Line 85

gas chromatography-ion mobility spectrometry [24]

Please add its abbreviation in a ‘()’ followed.

 

Line 94; line 150

two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (TD-GC×GC/ToFMS)

The abbreviation should be corrected to be: (GC×GC/TOF MS).

 

Line 113: taken minimum 2 hours after meals.

 

In previous references, the levels of biomarker compounds were not significantly influenced by factors such as use of deodorants or diet, while the identified biomarkers varied significantly. Compounds p-cymene and linalool are natural flavors probably deriving from the food. In this context, did this study consider the influence of food intake (meals)?

 

The sampling process is well designed (patients, healthy individuals, sampling position, two sides of the body, blanks) and stated in specific.

 

Line 131

This was made possible thanks to preliminary results, which indicated that the VOCs emanating from both sides of the body were analogous.

Please provide a reference for these preliminary results.

 

Line 157

dimensions 50 cm × 0.18 mm × 0.18 µm (Agilent) was utilized in the secondary oven.

I am not sure, but please check the length of the 2D column, 50 cm or 50 m?

 

The TD-GC×GC/ToFMS conditions are described in full detail.

 

Equations 1 and 2: FC1, 2 values

When using this fold change index, did the authors use statistics, such as t test or ANOVA, for comparison of the means? Were the sampling carried out repeatedly, in triplicate for example, so as to perform comparisons?

 

Equation 1: The FC1 value

There are two equations, covid+/- and covid-/+, both defined as FC1, which is a little confused. With such a definition, when looking at a FC value, one cannot tell if it represents overexpress or underexpress. Please clarify.

Table 1

RT 1D RT 2D

Please give the full name of these terms with units in the footnote of the table.

Please supplement the FC1 values for these compounds in the Table, meanwhile specify if it is covid+/- or covid-/+.

 

Table 2

Please also supplement the FC1 values, meanwhile specify covid+/- or covid-/+, for these compounds in the Table.

 

Line 269

benzoic acid, 2,4-dichloro-

Which group in Figure 1 did this compound belong to, the 27 compounds or 25 compounds? Please specify.

 

Figure 1

Looking at the abstract: 25 compounds highlighted, 13 associated with cosmetics and 7 linked to air pollution excluded, 1 compound underexpressed, 4 compounds retained. It is somewhat not clear. Which group did benzoic acid, 2,4-dichloro- (line 269) belong to?

Author Response

Response to Reviewer 1 Comments
Summary
Thank you very much for taking the time to review this manuscript. Please find the detailed responses below, the corresponding revision can also be found in red in the manuscript. We thank you for your comments, which have helped us to improve the quality of our manuscript. Sentences in bold in our answers correspond to modification brought to the manuscript.
Point-by-point response to Comments and Suggestions for Authors
The manuscript provides a useful approach for quick detection of possible COVID-19 patient. I have some concerns. Comments 1: Abstract 25 compounds highlighted, 13 associated with cosmetics and 7 linked to air pollution excluded, 4 compounds retained. Looking at Figure 1, there was also 1 compound which was not mentioned in the abstract. Was it benzoic acid, 2,4-dichloro- or the underexpressed compound B5? It is not clear. Please specify. Response 1: Thank you for pointing this out. We agree with this comment as this point lacks clarity. The compound that seemed to be missing is nonanoic acid as this compound was initially among the retained candidate biomarkers but ended up being excluded when comparing the peak areas between body volatilome samples and hospital air samples. To ensure a better understanding of this point we updated the number in the abstract including nonanoic acid among the air pollution molecules (even if this exclusion does not result from the same type of comparison) “13 associated with cosmetic products were excluded, and 8 linked to air pollution” (page 1, line 21). Also, when nonanoic acid peak’s areas are compared between individuals and hospital air samples we specified that given this observation it was set aside along with the 7 molecules previously highlighted as linked to environment “To avoid this type of undesirable influence, nonanoic acid was set aside, along with the 7 other molecules associated with a parasitic effect from the ambient environment.” (page 10, line 347-349). The benzoic acid, 2,4-dichloro- was just used as example in section 3.1.2. to show that both urban and hospital environments can have an impact on the samples, but this compound was not highlighted by our biomarkers research methodology. To clarify that, when benzoic acid, 2,4-dichloro- is mentioned we specified that it was observable but not highlighted by our biomarkers research methodology to not confound it with the other 25 compounds “Although the difference in intensity regarding this compound was visually observable on the chromatograms, the compound had not been highlighted by the biomarkers research methodology.”  (page 8, line 297-299). Comments 2: The introduction provides detailed background information and relevant references, giving sufficient foundational support for this study and demonstrating its significance. Line 85 gas chromatography-ion mobility spectrometry [24] Please add its abbreviation in a ‘()’ followed. Response 2: The abbreviation (GC-IMS) has been added (page 2, line 91) Comments 3: Line 94; line 150 two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (TD-GC×GC/ToFMS) The abbreviation should be corrected to be: (GC×GC/TOF MS). Response 3: The abbreviation (TD-GC×GC/ToFMS) has been corrected to be (TD-GC×GC/TOF MS) everywhere in the manuscript. Comments 4: Line 113: taken minimum 2 hours after meals. In previous references, the levels of biomarker compounds were not significantly influenced by factors such as use of deodorants or diet, while the identified biomarkers varied significantly. Compounds p-cymene and linalool are natural flavors probably deriving from the food. In this context, did this study consider the influence of food intake (meals)? Response 4: In our study the influence of food intake was not considered. We made the hypothesis that it didn’t influence the results by performing the sampling outside mealtimes but also because the diet of the sampled individuals was different, even among the sick individuals which were sampled in different hospitals. Nevertheless, we think it would be interesting to study this parameter in future studies. Here, our main point was to show that for some molecules it is sometimes difficult to state with certainty that they are real biomarkers. For example, p-cymene and linalool are indeed natural flavors that can be found in food but also in fragrances, cosmetics, detergents (used a lot in hospitals) or used as material processing agent (surfactant, emulsifier…). We consider that some of those sources could have a higher impact than food intake. Moreover, previous references stated indeed that the levels of biomarker compounds were not significantly influenced by factors such as deodorants or diet but we didn’t clearly know how these parameters were studied (except for the food intake using orange juice for p-cymene). We were therefore suspicious of these compounds, especially as a previous study we had carried out had shown a marked difference in the volatilome of an individual sampled while using cosmetics and after having stopped using them (certain molecules associated with cosmetics can be detected for 3 days even after their use had been discontinued). This is for these reasons that we looked for p-cymene and nonanal in our data but unfortunately, they were found at non-significant levels compared to sampling system blank or hospital air.     The sampling process is well designed (patients, healthy individuals, sampling position, two sides of the body, blanks) and stated in specific.  Comments 5: Line 131 This was made possible thanks to preliminary results, which indicated that the VOCs emanating from both sides of the body were analogous. Please provide a reference for these preliminary results. Response 5: These results were obtained by our team and are not published for the moment but will be included in my thesis manuscript (which has been submitted), so we have added its reference. However, some details on how these results were obtained have been added “These results were obtained by our team after comparing left/right hands, armpits and forearm for 19 individuals [27]” (page 3, line 145-146). Your comment makes us think that it might be interesting to publish these results.  Comments 6: Line 157 dimensions 50 cm × 0.18 mm × 0.18 µm (Agilent) was utilized in the secondary oven. I am not sure, but please check the length of the 2D column, 50 cm or 50 m? Response 6: Thank you for your vigilance, the length of the 2D column is indeed 50 cm as the separation in the second dimension needs to be performed in 3 seconds. The TD-GC×GC/ToFMS conditions are described in full detail. Comments 7: Equations 1 and 2: FC1, 2 values When using this fold change index, did the authors use statistics, such as t test or ANOVA, for comparison of the means? Were the sampling carried out repeatedly, in triplicate for example, so as to perform comparisons? Response 7: Fold change index was used on the peak areas of compounds which were already selected as significant after using Fisher’s test (used in ANOVA). Thus, at this level, no additional statistical tests were performed. Unfortunately, sampling was not carried out repeatedly due to hospital constraints. However, sick individuals were sampled from armpits, hands and groin and here these sampling areas have been combined in order to search for biomarkers that would be common, so it is possible to consider them as a kind of replicates. This precision has been added in section 2.4 “Data obtained from different sampling body areas (hands, armpits, groin) have been combined to search for biomarkers that would be common.” (page 4, line 184-186).   Comments 8: Equation 1: The FC1 value There are two equations, covid+/- and covid-/+, both defined as FC1, which is a little confused. With such a definition, when looking at a FC value, one cannot tell if it represents overexpress or underexpress. Please clarify. Response 8: We agree with this comment. To clarify the difference between fold changes related to overexpressed and underexpresseed compounds we differenciated the formulae and named the fold changes FC1 (overexpressed) and FC1’ (underexpressed) “the remaining hits were evaluated on a fold-change basis (FC1 and FC1’), using the formula (1) and (2)” (page 4, line 200-201). Moreover, to stay consistent, the Figure 1 has been updated to show that the criteria related to these fold changes was “FC1 or FC1’ > 10” (page 6). Comments 9: Table 1 RT 1D RT 2D Please give the full name of these terms with units in the footnote of the table. Response 9: Full name and units of these terms have been added in the footnote of the Table 1 “Retention time first dimension (RT 1D), retention time second dimension (RT 2D) both in seconds.” (page 7, line 274). Comments 10: Please supplement the FC1 values for these compounds in the Table, meanwhile specify if it is covid+/- or covid-/+. Response 10: As all compounds in Table 1 are overexpressed in Covid(-) group, the fold change values were added using FC1’ (page 7). Comments 11: Table 2 Please also supplement the FC1 values, meanwhile specify covid+/- or covid-/+, for these compounds in the Table. Response 11: Same as above, as all compounds in Table 2 are overexpressed in Covid(-) group, the fold change values were added using FC1’ (page 7). To stay consistent, FC1 and FC1’ values have also been added to Table 3 (page 10) Comments 12: Line 269 benzoic acid, 2,4-dichloro- Which group in Figure 1 did this compound belong to, the 27 compounds or 25 compounds? Please specify. Figure 1 Looking at the abstract: 25 compounds highlighted, 13 associated with cosmetics and 7 linked to air pollution excluded, 1 compound underexpressed, 4 compounds retained. It is somewhat not clear. Which group did benzoic acid, 2,4-dichloro- (line 269) belong to? Response 11: I think that the Response 1 answers also those points.

 

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript focuses on the study of the volatilome in COVID-19 patients and healthy individuals. The article is well-written and argued, and the research appears to provide sufficient evidence to support the data presented. However, I have a few considerations that I believe should be addressed.

Although the authors state that the work is focused on the identification of markers specific to COVID-19-positive patients, it seems to me that the study is more focused on identifying compounds that may appear at different levels in COVID-19 patients due to other factors, such as hospital environmental exposure or lack of exposure to urban environments or cosmetic. Therefore, rather than directly identifying COVID-19 markers, the article appears to serve more as a guide outlining key considerations for future studies on COVID-19 markers, highlighting which compounds should be excluded as potential disease markers due to their origin in the conditions experienced by hospitalized COVID-19 patients.

I believe these points are important to clarify in the manuscript to better align its claims with its findings.

 

Therefore, I have some specific comments:

• The title of the manuscript should be revised to better reflect the findings mentioned above.
• Please use the term “volatilome” consistently throughout the manuscript, as it is the more accurate terminology.
• In lines 84-85, reference 24 requires further explanation. Does this article employ a targeted or non-targeted analytical approach?
• Can a single marker or a small set of markers be sufficient to diagnose the disease? Non-targeted approaches tend to be more effective in addressing these challenges in "omics" studies.
• Line 214: The manuscript states that “130 samples were analyzed from 53 individuals (40 from the COVID(+) group and 13 from the COVID(-) group).” Were the samples taken from the same individuals at different times, or were duplicate samples collected? Please clarify this point.
• Regarding cosmetic and environmental influence, how many individuals were exposed to these conditions, and how many of them were positive for the markers related with these conditions?
• Lines 293-298: Please provide additional details on the origin of these compounds.
• Lines 375-377: In my opinion, this section contains the most important conclusion of the manuscript—establishing a protocol for conducting studies of this nature.

Comments on the “2.1. Clinical study” section:

• The number of samples, especially for healthy individuals, seems insufficient to draw strong conclusions. Recruiting more healthy individuals, who are generally easier to find, would strengthen the study.
• If environmental exposure plays such a critical role, the manuscript or supplementary material should include information regarding the environments the individuals were exposed to, including the use of cosmetics and other relevant factors.
• It appears that only hospitalized COVID-19 patients were studied. Were COVID-19-positive patients who did not require hospitalization also considered? Has the study accounted for different severities of the disease?

Author Response

Response to Reviewer 2 Comments
Summary
Thank you very much for taking the time to review this manuscript. Please find the detailed responses below, the corresponding revision can be found in red in the manuscript. We thank you for your comments, which have helped us to improve the quality of our manuscript. Sentences in bold in our answers correspond to modification brought to the manuscript.
Point-by-point response to Comments and Suggestions for Authors
The manuscript focuses on the study of the volatilome in COVID-19 patients and healthy individuals. The article is well-written and argued, and the research appears to provide sufficient evidence to support the data presented. However, I have a few considerations that I believe should be addressed. Comments 1: Although the authors state that the work is focused on the identification of markers specific to COVID-19-positive patients, it seems to me that the study is more focused on identifying compounds that may appear at different levels in COVID-19 patients due to other factors, such as hospital environmental exposure or lack of exposure to urban environments or cosmetic. Therefore, rather than directly identifying COVID-19 markers, the article appears to serve more as a guide outlining key considerations for future studies on COVID-19 markers, highlighting which compounds should be excluded as potential disease markers due to their origin in the conditions experienced by hospitalized COVID-19 patients. I believe these points are important to clarify in the manuscript to better align its claims with its findings. Response 1: We thank you for your comment which shows a great level of understanding of our intentions which are indeed beyond biomarker identification and more focused on showing how careful we need to be to suggest relevant biomarkers, highlighting compounds that could have been mistaken for biomarkers but are in fact linked to exogenous parameters. We will take your comments into careful consideration to make this point even clearer. Therefore, I have some specific comments: Comments 2: ·         The title of the manuscript should be revised to better reflect the findings mentioned above. Response 2: We have revised the title of the manuscript replacing “study” by “study strategy” to imply that we used a multi-step process and adding “considering exogenous parameters” at the end of the tittle to mention that the work is not solely focused on finding “statistically significant” biomarkers but more on finding relevant biomarker considering the context of the study and the complexity of the matrix, i.e. its exogenous influences. Thus, the full title is now “Body Volatilome Study Strategy for Covid-19 Biomarker Identification Considering Exogenous Parameters”. Comments 3: ·         Please use the term “volatilome” consistently throughout the manuscript, as it is the more accurate terminology. Response 3: Thank you for this comment. The term “volatolome” has been replaced by “volatilome” in the manuscript. Comments 4: In lines 84-85, reference 24 requires further explanation. Does this article employ a targeted or non-targeted analytical approach? Response 4: More details have been added regarding the reference 24 “In this study, 55 features were highlighted thanks to a non-targeted method as significantly different over time, but only 4 specific ones were investigated in detail, including methanol. On the basis of these features, differences were noted regarding the volatilome depending on the type of vaccine (Pfizer vs Moderna).” (page 2, line 91-95). In this article, non-targeted approach was initially used highlighting 55 features but then only 4 of them were investigated in detail. Comments 5: ·         Can a single marker or a small set of markers be sufficient to diagnose the disease? Non-targeted approaches tend to be more effective in addressing these challenges in "omics" studies. Response 5: Thank you for pointing this out. This is one of the big questions surrounding the field, and I don't think anyone has a definitive answer at the moment. In our study, we presented what our method allowed us to obtain and above all that the quantity of biomarkers is not necessarily a guarantee of quality. Further studies would be needed to confirm the diagnostic power of this biomarker. Moreover, our biomarker has been identified thanks to samples taken for individuals with Covid-19 but further studies would also be needed to know whether it is a biomarker more generally associated with lung infections, for example, or if it is pathology-specific. This precision has been added to the conclusion of the manuscript “This biomarker has been identified in samples taken from individuals with Covid-19, but further studies would be needed to determine whether it could be more generally associated with lung infections, for example, or if it is pathology-specific (page 11, line 414-416). Finally, we agree that a non-targeted approach is more efficient to address this type of challenge, this is why we used in a first step the software ChromaToF Tile which allowed to screen our whole data set before conducting more detailed data treatment, manually feasible. Comments 6: ·         Line 214: The manuscript states that “130 samples were analyzed from 53 individuals (40 from the COVID(+) group and 13 from the COVID(-) group).” Were the samples taken from the same individuals at different times, or were duplicate samples collected? Please clarify this point. Response 6: Unfortunately, sampling was not carried out repeatedly due to hospital constraints. However, sick individuals were sampled at the same time from armpits, hands and groin and here these sampling areas have been combined for the data processing step in order to search for biomarkers that would be common (justifying the obtained number of samples). This precision has been added in section 2.4 “Data obtained from different sampling body areas (hands, armpits, groin) have been combined to search for biomarkers that would be common.” (page 4, line 184-186).  Comments 7: ·         Regarding cosmetic and environmental influence, how many individuals were exposed to these conditions, and how many of them were positive for the markers related with these conditions? Response 7: Exposure to cosmetics and the environment has not been deliberately set up. The identification of molecules linked to these factors is more the result of not excluding these criteria initially (no restriction given on cosmetics and no possibility of sampling all individuals in the same environment). It was by looking at the molecules highlighted by our research methodology (and thanks to related studies) that we became aware that these factors had an impact, and that it was important to consider them a posteriori to avoid providing biased results. Comments 8: ·         Lines 293-298: Please provide additional details on the origin of these compounds. Response 8: Concerning the compounds found in hospital air samples, according to the cited studies, their main origin was linked to cleaning/disinfectant products and other type of chemical products. We added that precision in the manuscript “According to those studies, most of the compounds found in hospital air were related to cleaning/disinfectant products and other type of chemical products used for different health activities.” (page 9, line 313-315). Comments 9: ·         Lines 375-377: In my opinion, this section contains the most important conclusion of the manuscript—establishing a protocol for conducting studies of this nature. Response 9: Thank you for this comment. Indeed, we put those lines in the conclusion to summarize our main idea. Comments on the “2.1. Clinical study” section: Comments 10: ·         The number of samples, especially for healthy individuals, seems insufficient to draw strong conclusions. Recruiting more healthy individuals, who are generally easier to find, would strengthen the study. Response 10: We agree with this comment, however as this study was initially designed to be an “exploratory clinical study” it was not possible to add more individuals. However, we took that point into consideration for the design of another study currently conducted where the sets are more balanced and extended. Comments 11: ·         If environmental exposure plays such a critical role, the manuscript or supplementary material should include information regarding the environments the individuals were exposed to, including the use of cosmetics and other relevant factors. Response 11: As this kind of exposure was not “controlled” but “incurred”, all that information is not available. For example, for cosmetics (as said in section 2.1) no specific restriction were given which means that each individual was free to use any product he wanted. After observing the critical impact it had, we conducted another “small scale” study to see how many days a cosmetic product can last on the skin (regarding the volatilome point of view) and this is why we stated in the conclusion that in new protocols this kind of product should be prohibited for a certain time before the sampling. Comments 12: ·         It appears that only hospitalized COVID-19 patients were studied. Were COVID-19-positive patients who did not require hospitalization also considered? Has the study accounted for different severities of the disease? Response 12: In this study only hospitalized COVID-19 patients were studied because they suffered from a high severity of the disease (respiratory difficulties for example). Not to add too many parameters to the study, we decided to restrict ourselves to this population.

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The authors present a study looking at differences between the body volatolome in human subjects with and without Covid-19. I think this is an interesting research area, but I have some comments.

Abstract - why have the authors selected disproportionate numbers of Covid-19 patients and controls. Are healthy people an appropriate control for this study? Surely there should be some detail here about the tests used to diagnose Covid-19 infection?

Also is it true that since the 1950s this has been an established method of detecting disease?

Is it enough to say patients were sampled using PowerSorb? maybe mention the medium i.e. body VOCs etc. "were sampled using PowerSorb® as sorbent phase,"

What method specifically was used here " Biomarker research methodology compared Covid(+) and Covid(-) chromatograms"

Check maths at end of abstract as 25 compoiunds with 20 excluded seems to be 5, but 4 compounds is mentioned.

Also is 2-ethyl hexanoic acid concentration used to differentiate the groups? this is not completely clear.

Maybe have a sentence at the end of the abstract summing up what future studies need to be undertaken and how this work could be useful?

 

Introduction - I think the beginning of the introduction could be condensed and better referenced. For example although I agree with the purpose of research not sure it is necessary to state this. Also some of the background information seems unnecessary.

THe authors give an extensive overview of one groups work on Covid-19 but I think this is maybe too narrowly focused. There are other groups that have undertaken studies and so it would be good to overview this work. Also it would be good to overview the work to differentiate between Covid-19 and other viral infections rather than healthy controls.

The justification of the authors study is quite weak if they are claiming that these are the only studies undertaken by a single group and that different markers were identified etc.

Methods

This needs to be more specific "40 patients were sampled and included in the Covid(+) group 105 as they had either a positive PCR or antigenic test, respectively carried out at most 96h 106 and 48h before."

If you are trying toaccount for environmental VOCS and exclude do you not think sampling patients/controls in the same environment is preferable?

" The Covid(-) individuals were sampled at the ESPCI (Paris, France) or in a partner canine training center involved in Covid-19 detection (French K9 Academy, Parthenay, France), not in the hospital due to logistical constraints."

Has it been establisged that it is odourless? " the area of skin to be sampled was cleaned with an odorless cleansing wipe (Ront) allowing to remove potential cosmetic residues."

Were blanks run? "Before sampling, PowerSorb® were individually cleaned using the tube conditioner TC-20 (Markes International, Bridgend, UK) for 4 hours at 240°C with a N2 flow at 50 mL/min."

Needs more detail "This was made possible thanks to preliminary results, which indicated that the VOCs emanating from both sides of the body were analogous."

Does this give a good measure of the interfering VOCs? "To do so, 2 conditioned PowerSorb® were put in the PP/AB box 136 which was then wrapped in bandage strip and placed in an oven at 40°C for 1 hour to 137 simulate body heat."

I think the data processing section needs more clarity - I don't think the current description is easy to follow. Maybe requires a figure to show the process?

What about the other 3 compounds were these not identified? or didn't match standards?

"Over the 5 compounds kept as potential volatile biomarker of Covid-19, two attribu- 203 tions were confirmed using standards: nonanoic acid and 2-ethyl hexanoic acid (Sigma- 204 Aldrich)."

Is it rig rather than ring? "using the calibration solution loading ring (CSLR)"

Results

Figure 1 legend needs to be better to explain the terms used.

 

Why do you report Hexanoic acid 2 ethyl in the abstract when other compounds give a better AUROC?

What if you combine 2 or more VOCs does this give a better classifier?

These statements do seem to call into question the whole study if you can't guarantee sample validity, yet you are comparing them? "This was  particularly true for p-cymene, for which the significance could probably be due to a different level of expertise of the individuals performing the sampling: Covid(-) samples were collected by experienced individuals accustomed to the sampling system, whereas in hospital, the nursing staff was less experienced and hospital constraints resulted in less time available."

conclusions - this could be rewritten "But it was only through a critical eye that compounds might coming from exogenous sources could be excluded."

This isn't that clearly/scientifically written? " However, if this kind of restriction will now be added to future clinical study launched by our team, the effect of surrounding air cannot be avoided in the same way."

Maybe need to discuss the limitations and mitigating these in future studies in the discussion in addition to the conclusions.

 

Author Response

Response to Reviewer 3 Comments
Summary
Thank you very much for taking the time to review this manuscript. Please find the detailed responses below, the corresponding revision can be found in red in the manuscript. We thank you for your comments, which have helped us to improve the quality of our manuscript. Sentences in bold in our answers correspond to modification brought to the manuscript.
Point-by-point response to Comments and Suggestions for Authors

The authors present a study looking at differences between the body volatolome in human subjects with and without Covid-19. I think this is an interesting research area, but I have some comments.

Comments 1:

Abstract - why have the authors selected disproportionate numbers of Covid-19 patients and controls. Are healthy people an appropriate control for this study? Surely there should be some detail here about the tests used to diagnose Covid-19 infection?

Response 1: As this study was initially designed to be an “exploratory clinical study” the number of patients and control were restricted. However, we took that into account for a new clinical study currently undertaken (on another disease) where our groups are bigger and more balanced. To explore this kind of problematic, healthy people were the simplest control to have as we are looking for biomarkers which could be used as pre-diagnosis tool, but then if biomarker specificity needs to be evaluated it could be possible to compare for example COVID-19 patients with flu patients. As the tests used to diagnose COVID-19 infection where the most commonly used (PCR and antigenic) we preferred mentioning this in the Materials and Methods section.

Comments 2:

Also is it true that since the 1950s this has been an established method of detecting disease?

Response 2: The first references citing the use of body volatilome as a diagnostic tool in the literature date back to this period, and it is since then that studies have multiplied. Nevertheless, this method is not yet “established” because it is still not commonly used. However, we confirm that according to the articles we read it has been recognized as a powerful tool since the 1950s.

 

Comments 3:

Is it enough to say patients were sampled using PowerSorb? maybe mention the medium i.e. body VOCs etc. "were sampled using PowerSorb® as sorbent phase,"

Response 3: Thank you for your comment. Indeed, this sentence lacked of precision so we added the details you mentioned “The VOCs constituting the body volatilome of 40 Covid-19 patients and 13 healthy subjects were sampled using PowerSorb® as sorbent phase” (page 1, line 15).

Comments 4:

What method specifically was used here " Biomarker research methodology compared Covid(+) and Covid(-) chromatograms"

Response 4: We decide to use “biomarker research methodology” to describe our method and its different steps illustrated in the Figure 1. We specified in the same sentence that this methodology combined statistical significance measure and peak area changes. However, to add more precision we decided to describe our methodology as non-targeted “Non-targeted biomarker research methodology compared Covid(+) and Covid(-) chromatograms” (page 1, line 18-19).

Comments 5:

Check maths at end of abstract as 25 compounds with 20 excluded seems to be 5, but 4 compounds is mentioned.

Response 5: Thank you for this comment which highlights a lack of clarity on this point. 13 associated with cosmetics (based on their attributions), 7 linked to air pollution (based on their attribution and observation prevalence among individuals’ groups) and thus 5 initially retained as candidate biomarkers. However, thanks to the added comparison between the areas of these 5 peaks in individuals and hospital air samples, nonanoic acid was found at the same level. Therefore, this compound was excluded as it was also linked to environment air. To get mathematics right we decided to update the number of compounds related to air pollution, going from 7 to 8 to include the nonanoic acid “13 associated with cosmetic products were excluded, and 8 linked to air pollution” (page 1, line 20-21).

Comments 6:

Also is 2-ethyl hexanoic acid concentration used to differentiate the groups? this is not completely clear.

Response 6: 2-ethyl hexanoic acid concentration was not used to differentiate the groups, this information was added because we realized when looking at the literature that the VOCs of the body volatolome were seldom quantified and we wanted to add this information. However, as it seemed to hinder understanding, we preferred to remove it from the abstract. The sentence is now “Among them, hexanoic acid, 2-ethyl- was confirmed with standard and led to an area under curve of 92%.” (page 1, line 23-24).

Comments 7:

Maybe have a sentence at the end of the abstract summing up what future studies need to be undertaken and how this work could be useful?

Response 7: We agree with this comment; thus, we added the suggested sentence at the end of the abstract “More in-depth studies would be needed to investigate the specificity of the biomarker in relation to Covid-19, but the strategy of the present study showed how to avoid obtaining data biased by exogenous factors.” (page 1, line 24-26). However, we hope it will fit the requirements of the journal as the abstract is now over 200 words long (225 to be exact).

Comments 8:

Introduction - I think the beginning of the introduction could be condensed and better referenced. For example although I agree with the purpose of research not sure it is necessary to state this. Also some of the background information seems unnecessary.

THe authors give an extensive overview of one groups work on Covid-19 but I think this is maybe too narrowly focused. There are other groups that have undertaken studies and so it would be good to overview this work. Also it would be good to overview the work to differentiate between Covid-19 and other viral infections rather than healthy controls.

The justification of the authors study is quite weak if they are claiming that these are the only studies undertaken by a single group and that different markers were identified etc.

Response 8: The choice was made to limit our consideration to articles studying COVID-19 thanks to body volatilome, meaning not including breath test for example. A search was again carried out when reading reviewers’comments, but we were unable to find any additional articles corresponding to our criteria and therefore coming from other research groups than the one mentioned. In our opinion, it is important to limit the study to the body volatilome in order to compare matrices that are genuinely comparable. Nevertheless, if you have any references to suggest, we would be delighted to add them to this manuscript.

Moreover, the justification of our study is not solely linked to the fact that different markers were identified by one research team. Here we studied COVID-19 but, to us, the impact that the exogenous compounds can have over the matrix is the same for any disease as long as body volatilome is studied. Thus, our purpose was rather to highlight that it is important to be vigilant when identifying biomarkers from the body volatilome, as the matrix remains relatively unknown.

Comments 9:

Methods

This needs to be more specific "40 patients were sampled and included in the Covid(+) group 105 as they had either a positive PCR or antigenic test, respectively carried out at most 96h 106 and 48h before."

Response 9: Slight changes has been made to this sentence and some precisions have been added to precise why the PCR or antigenic test needed to be done at most 96h or 48h before inclusion. “40 patients were sampled and included in the Covid(+) group after confirming their health status with a positive PCR or antigenic test, respectively carried out at most 96h or 48h before the inclusion. This allowed to include patients who were at the maximum of their symptoms.” (page 3, line 114-117).

Comments 10:

If you are trying to account for environmental VOCS and exclude do you not think sampling patients/controls in the same environment is preferable? " The Covid(-) individuals were sampled at the ESPCI (Paris, France) or in a partner canine training center involved in Covid-19 detection (French K9 Academy, Parthenay, France), not in the hospital due to logistical constraints."

Response 10: We totally agree with this comment, but at the time of the study it was not possible because of the pandemic and the constraints of the hospital environment. Thanks to these results, for a current study, we have ensured that the sampling environments are identical (or at least similar), while ambient air samples are taken for each individual sampled.

Comments 11:

Has it been establisged that it is odourless? " the area of skin to be sampled was cleaned with an odorless cleansing wipe (Ront) allowing to remove potential cosmetic residues."

Response 11: Yes, we studied that point by simply rubbing the wipe on a PowerSorb® and then we analyzed the PowerSorb by TD-GC×GC/TOF MS (with the same method used for real samples) to be sure that it remained clean.

Comments 12:

Were blanks run? "Before sampling, PowerSorb® were individually cleaned using the tube conditioner TC-20 (Markes International, Bridgend, UK) for 4 hours at 240°C with a N2 flow at 50 mL/min."

Response 12: Yes, for each batch of 20 conditioned PowerSorb® 2 PowerSorb® were kept to be analyzed as blanks. We added that precision in the manuscript “For each batch of 20 conditioned PowerSorb®, 2 PowerSorb® were kept to be analyzed by TD-GC×GC/TOF MS as blanks.” (page 3, line 136-137).

Comments 13:

Needs more detail "This was made possible thanks to preliminary results, which indicated that the VOCs emanating from both sides of the body were analogous."

Response 13: These results were obtained by our team and are not published for the moment but will be included in my thesis manuscript (which has been submitted), so we have added its reference. However, some details on how these results were obtained have been added “These results were obtained by our team after comparing left/right hands, armpits and forearm for 19 individuals [27].” (page 3, line 145-146).

Comments 14:

Does this give a good measure of the interfering VOCs? "To do so, 2 conditioned PowerSorb® were put in the PP/AB box 136 which was then wrapped in bandage strip and placed in an oven at 40°C for 1 hour to 137 simulate body heat."

Response 14: We ended up using this method after comparing with others (like thermodesorbing individually each material of the sampling system to know its VOCs emissions, for example) but this one was the closest one to the real sampling situation where we expected the VOCs emissions of the bandage strip to be the more problematic. By wrapping the box in the bandage strip, we got the worst-case scenario of a not properly fastened sampling device. Thus, it was a good way to measure the interfering VOCs to then exclude them by using the FC₂.

Comments 15:

I think the data processing section needs more clarity - I don't think the current description is easy to follow. Maybe requires a figure to show the process?

Response 15: The data processing has already been described visually using the Figure 1.

Comments 15:

What about the other 3 compounds were these not identified? or didn't match standards?

"Over the 5 compounds kept as potential volatile biomarker of Covid-19, two attribu- 203 tions were confirmed using standards: nonanoic acid and 2-ethyl hexanoic acid (Sigma- 204 Aldrich)."

Response 15: The other 3 compounds had an attribution given by the software but standards of those molecules were not commercially available so we were not able to confirm it.

Comments 16:

Is it rig rather than ring? "using the calibration solution loading ring (CSLR)"

Response 16: Thank you, it was indeed a mistake. We corrected it in the manuscript “calibration solution loading rig” (page 5, line 225).

Comments 17:

Results

Figure 1 legend needs to be better to explain the terms used.

Response 17: To improve the quality of the explanations, in line with comment 15, the legend has been enriched with information and details “ChromaToF Tile was used as non-targeted step to obtain a list of statistically significant hits, the fold changes FC1 and FC1’ allowed to keep only compounds 10 times over- or underexpressed for Covid-19 patients and FC2 allowed to eliminate interfering compounds coming from the sampling device.” (page 6, line 250-253).

Comments 18:

Why do you report Hexanoic acid 2 ethyl in the abstract when other compounds give a better AUROC?

What if you combine 2 or more VOCs does this give a better classifier?

Response 18: We made the choice to only report 2-ethyl hexanoic acid in the abstract as this was the only compound which had a confirmed attribution. As there was a lack of information for the other compounds (despite their better AUROC) we preferred to highlight only the information which has the highest level of confidence.

For the combination of different VOCs we tried to use a multivariate exploratory ROC analysis using PLS-DA (with the software MetaboAnalyst) as classification method. The best classifier was obtained using the combination of the 4 molecules and gave an AUC of 98.5%. It is very interesting but we were afraid that by adding a new statistical method to an article where data processing is already tedious, the article would lose some of its clarity.

Comments 19:

These statements do seem to call into question the whole study if you can't guarantee sample validity, yet you are comparing them? "This was particularly true for p-cymene, for which the significance could probably be due to a different level of expertise of the individuals performing the sampling: Covid(-) samples were collected by experienced individuals accustomed to the sampling system, whereas in hospital, the nursing staff was less experienced and hospital constraints resulted in less time available."

Response 19: We understand your point of view as we asked ourselves the same type of question and this is why the implementation of the FC₂ is important: indeed, we saw that the different levels of expertise of the individuals using the sampling system could lead to higher level of contamination coming from the VOCs of the sampling device. However, as these emissions were evaluated previously, we were able to eliminate compounds that would seem significant but yet not originating from the individuals. For p-cymene we still looked for it as it was an identified Covid-19 biomarker in the literature and we ended up finding it at same levels in Covid(+) samples as in the sampling system blanks. Still, we mentioned it, as we wanted to point out that sometimes it can seem “easy” to get a statistically significant biomarker (to discriminate between healthy and sick groups) while it is not linked at all with the disease. For all the other molecules highlighted by our biomarkers research methodology we avoided this risk by using FC₂ which basically eliminates compounds that are not at least twice higher in the samples compared with the sampling system blanks.

Finally, after conducting this study we focused our effort on developing a new sampling system which emits far fewer VOCs to further reduce this type of constraint.

Comments 20:

conclusions - this could be rewritten "But it was only through a critical eye that compounds might coming from exogenous sources could be excluded."

Response 20: The sentence has been rewritten. “However, despite the use of these high-performance tools, it was necessary to remain vigilant and critical with regard to the results in order to eliminate compounds coming from exogenous sources.” (page 11, line 392-394)

Comments 21:

This isn't that clearly/scientifically written? " However, if this kind of restriction will now be added to future clinical study launched by our team, the effect of surrounding air cannot be avoided in the same way."

Response 21: Thank you, we agree with your point of view. The sentence has been rewritten “Thus, a restriction on the use of cosmetics, minimum 24h before the sampling, will now be added for future studies. However, regarding the effect of the surrounding air, the same kind of restriction cannot be applied and collecting air samples seemed to be the only way to avoid bias related to that parameter.” (page 11, line 397-400)

Comments 22:

Maybe need to discuss the limitations and mitigating these in future studies in the discussion in addition to the conclusions.

Response 22: Some limitations mentioned in the conclusion have been added to the different discussion parts: regarding cosmetics “Nevertheless, to avoid this type of influence in future studies, a restriction on the use of cosmetic products will be added before the sampling.” (page 6, line 268-270) and regarding the environment “To avoid this type of influence, future studies should ideally perform the sampling of the different groups in the same environment. However, if this is not possible, a systematic ambient air sampling must be carried out simultaneously with volatilome sampling.” (page 9, line 318-321). And the limitations regarding the specificity of the biomarkers were already written in the discussion “More advanced studies would be needed to determine whether it is a specific marker of Covid-19 or, more generally, a biomarker of a viral/respiratory infection. “(page 10, line 356-358).

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has been improved following most of my suggestions, and I now recommend it for publication in its current form

Author Response

Comment 1: The manuscript has been improved following most of my suggestions, and I now recommend it for publication in its current form

Response 1: We would like to thank you for your comments, which have helped us to improve the quality of the article, and for your recommendation on its current version.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors addressed some of my comments. Regarding the introduction and overviewing more studies. I think that there are some dog and electronic nose studies that look at body odour which may be worth including?

I personally feel there is so much work on breath and Covid-19 detection that it may be worth the authors mentioning some of that work also. Only because it gives reassurance on the concept and there are distinct mechanisms where VOCs may present on the breath and in body odour etc.  So maybe add one of the systematic review/meta-analysis references?

Guest C, Dewhirst SY, Lindsay SW, Allen DJ, Aziz S, Baerenbold O, Bradley J, Chabildas U, Chen-Hussey V, Clifford S, Cottis L, Dennehy J, Foley E, Gezan SA, Gibson T, Greaves CK, Kleinschmidt I, Lambert S, Last A, Morant S, Parker JEA, Pickett J, Quilty BJ, Rooney A, Shah M, Somerville M, Squires C, Walker M, Logan JG; COVID Dogs Research Team. Using trained dogs and organic semi-conducting sensors to identify asymptomatic and mild SARS-CoV-2 infections: an observational study. J Travel Med. 2022 May 31;29(3):taac043. doi: 10.1093/jtm/taac043. PMID: 35325195; PMCID: PMC9047163.

Devillier P, Gallet C, Salvator H, Lecoq-Julien C, Naline E, Roisse D, Levert C, Breton E, Galtat A, Decourtray S, Prevel L, Grassin-Delyle S, Grandjean D. Biomedical detection dogs for the identification of SARS-CoV-2 infections from axillary sweat and breath samples*. J Breath Res. 2022 Apr 28;16(3). doi: 10.1088/1752-7163/ac5d8c. PMID: 35287115.

Snitz K, Andelman-Gur M, Pinchover L, Weissgross R, Weissbrod A, Mishor E, Zoller R, Linetsky V, Medhanie A, Shushan S, Jaffe E, Sobel N. Proof of concept for real-time detection of SARS CoV-2 infection with an electronic nose. PLoS One. 2021 Jun 2;16(6):e0252121. doi: 10.1371/journal.pone.0252121. PMID: 34077435; PMCID: PMC8172018.

Subali AD, Wiyono L, Yusuf M, Zaky MFA. The potential of volatile organic compounds-based breath analysis for COVID-19 screening: a systematic review & meta-analysis. Diagn Microbiol Infect Dis. 2022 Feb;102(2):115589. doi: 10.1016/j.diagmicrobio.2021.115589. Epub 2021 Oct 30. PMID: 34879323; PMCID: PMC8556067.

I understand what the authors are saying about the difficulty of obtaining balanced groups for their study, however, when they then identify VOCS that may be linked to Covid-19 it potentially becomes more problematic, as having disproportionate groups makes it more likely that differences will be observed. Also it increases the chances of overfgitting in the statistical models etc. 

My point regarding healthy controls was that it may be better to use asymptomatic patients who had been in contact with Covid-19 patients (family members etc.). Also a group who are symptomatic but not Covid-19 + would have been good. 

Author Response

Response to Reviewer 3 Comments
1. Summary
Thank you for taking the time to clarify your comments in this second round of review. Please find the detailed responses below, the corresponding revision can be found in red in the manuscript. Sentences in bold in our answers correspond to modification brought to the manuscript. We thank you for your comments, which have helped us to improve the quality of our manuscript.
3. Point-by-point response to Comments and Suggestions for Authors The authors addressed some of my comments. Comments 1: Regarding the introduction and overviewing more studies. I think that there are some dog and electronic nose studies that look at body odour which may be worth including? I personally feel there is so much work on breath and Covid-19 detection that it may be worth the authors mentioning some of that work also. Only because it gives reassurance on the concept and there are distinct mechanisms where VOCs may present on the breath and in body odour etc.  So maybe add one of the systematic review/meta-analysis references? Guest C, Dewhirst SY, Lindsay SW, Allen DJ, Aziz S, Baerenbold O, Bradley J, Chabildas U, Chen-Hussey V, Clifford S, Cottis L, Dennehy J, Foley E, Gezan SA, Gibson T, Greaves CK, Kleinschmidt I, Lambert S, Last A, Morant S, Parker JEA, Pickett J, Quilty BJ, Rooney A, Shah M, Somerville M, Squires C, Walker M, Logan JG; COVID Dogs Research Team. Using trained dogs and organic semi-conducting sensors to identify asymptomatic and mild SARS-CoV-2 infections: an observational study. J Travel Med. 2022 May 31;29(3):taac043. doi: 10.1093/jtm/taac043. PMID: 35325195; PMCID: PMC9047163. Devillier P, Gallet C, Salvator H, Lecoq-Julien C, Naline E, Roisse D, Levert C, Breton E, Galtat A, Decourtray S, Prevel L, Grassin-Delyle S, Grandjean D. Biomedical detection dogs for the identification of SARS-CoV-2 infections from axillary sweat and breath samples*. J Breath Res. 2022 Apr 28;16(3). doi: 10.1088/1752-7163/ac5d8c. PMID: 35287115. Snitz K, Andelman-Gur M, Pinchover L, Weissgross R, Weissbrod A, Mishor E, Zoller R, Linetsky V, Medhanie A, Shushan S, Jaffe E, Sobel N. Proof of concept for real-time detection of SARS CoV-2 infection with an electronic nose. PLoS One. 2021 Jun 2;16(6):e0252121. doi: 10.1371/journal.pone.0252121. PMID: 34077435; PMCID: PMC8172018. Subali AD, Wiyono L, Yusuf M, Zaky MFA. The potential of volatile organic compounds-based breath analysis for COVID-19 screening: a systematic review & meta-analysis. Diagn Microbiol Infect Dis. 2022 Feb;102(2):115589. doi: 10.1016/j.diagmicrobio.2021.115589. Epub 2021 Oct 30. PMID: 34879323; PMCID: PMC8556067. Response 1: Thank you for the given information. We absolutely agree with you as we already mentioned in the introduction that some work exist relating sensor-based methods, dog detection and use of exhaled air. To reinforce this point, we have slightly modified the text and added all the references suggested “However, alternative approaches were also developed, including sensor-based methods [15, 16] and the analysis of volatile organic compounds (VOCs) in exhaled air, which has been studied a lot more than body volatilome [17, 18]. The earliest studies using body volatilome to detect SARS-CoV-2 primarily utilized dogs' powerful sense of smell [19–22]. (page 2, line 62-67)”. Nevertheless, we have chosen not to detail the results of these studies so as not to confuse the reader with those obtained for the body volatilome. Comments 2: I understand what the authors are saying about the difficulty of obtaining balanced groups for their study, however, when they then identify VOCS that may be linked to Covid-19 it potentially becomes more problematic, as having disproportionate groups makes it more likely that differences will be observed. Also it increases the chances of overfgitting in the statistical models etc.  My point regarding healthy controls was that it may be better to use asymptomatic patients who had been in contact with Covid-19 patients (family members etc.). Also a group who are symptomatic but not Covid-19 + would have been good.  Response 2: Thank you for you comment. Although we can no longer adjust these points by including new individuals, we consider your comments relevant and have added this information in the conclusion to qualify the results and provide avenues for improvement for future studies. Regarding the impact of unbalanced groups on statistical model: “Finally, balanced groups should be encouraged in order to reinforce the reliability of statistical tests.” (page 11, line 426-427). Regarding the healthy controls: “One way of doing this would be to use as a control group asymptomatic patient who have been in contact with Covid-19 patients, and/or a group with symptoms but not tested positive for Covid-19.” (page 11, line 418-420).