Quantity of SARS-CoV-2 RNA copies exhaled per minute during natural breathing over the course of COVID-19 infection

Reviewer #1 (Public Review):

Summary:

This manuscript introduces an exciting way to measure SARS-CoV-2 aerosolized shedding using a disposable exhaled breath condensate collection device (EBCD). The paper draws the conclusion that the contagious shedding of the virus via aerosol route persists at a high level 8 days after symptoms.

Strengths:

The methodology is potentially of high importance and the paper is clearly written. The study design is clever. If aerosolized viral load kinetics truly differed from those of nasal swabs, then this would be a very important finding.

Weaknesses:

The study conclusions are not entirely supported by the data for several reasons:

(1) Most data points in the study are relatively late during infection when viral loads from other compartments (nasal and oral swabs) are typically much lower than peak viral loads which often occur in the pre-symptomatic or early symptomatic phase of infection. Moreover, the generation time for SARS-CoV-2 has been estimated to be 3-4 days on average meaning that most infections occur before or very early during symptoms. Therefore, the available epidemiologic data does not support 12 days of infection (day 8 symptoms) as important for most transmissions. Therefore, many of the measurement timepoints in this study may not be relevant for transmission.

(2) Fig 1A would be more powerful as a correlation plot between viral load from nasal samples (x-axis) and aerosol (y-axis). One would expect at least a rough correlation (as has been seen between viral loads in oral and nasal samples) and deviations from this correlation would provide crucial information about how and when aerosol shedding is discordant from nasal samples (ie early vs late time points, low versus high viral loads< etc...). It is too strong to state correspondence is 100% when viral load is only measured in one compartment and nasal swabs are reduced to the oversimplified "positive or negative".

(3) Results are reported in RNA copies which is fine but particle-forming units (pfu, or quantitative culture) are likely a more accurate surrogate of infectivity. It is quite possible that all of these samples would have been negative for pfu given that the ratio of RNA: pfu is often >1000 (though also dynamic over time during infection). This could be another indicator that most samples in the study were collected too late during infection to represent contagious time points.

(4) Individual kinetic curves should be shown for participants with more than three time points to demonstrate whether there are clear kinetic trends within individuals that would help further validate this approach. The inclusion of single samples from individuals is less informative.

(5) The S-shaped model in 2A is somewhat misleading as it is fit to means but there is tremendous variability within the data. Therefore the 8-day threshold should be listed clearly as a mean but not a rule for all individuals. The statement that viral RNA copies do not decrease until 8 days from symptom onset is unlikely to be true for all infected people and can't be made based on the available data in this study given that many people contributed only one datapoint.

(6) The incubation period for SARS-CoV-2 is highly variable. Therefore duration of symptoms is a rather poor correlate of the duration of infection. This further diminishes the interpretive value of positive samples from individuals who were only sampled once.

Reviewer #2 (Public Review):

Summary:

In this manuscript, Lane and colleagues measured the abundance of SARS-CoV-2 on breath in 60 outpatients after the development of COVID-19 symptoms using a novel breath collection apparatus. They found that, overall, viral abundance remains high for approximately eight days following the development of symptoms, after which viral abundance on breath drops to a low level that may persist for approximately 20 days or more. They did not identify significant differences in viral shedding on breath by vaccination status or viral variant. They also noted substantial variation in the degree and duration of shedding across individuals.

Strengths:

The primary strengths of this study are (1) the focus on breath, rather than the more traditional nasal/oropharyngeal swabs, and (2) the fact that the data were collected at multiple time points for each infection. This allows the authors to characterize not only mean viral abundance across individuals but also how that abundance changes over time, allowing for a better understanding of the potential duration of infectiousness of SARS-CoV-2.

Weaknesses:

The sample size is moderate (60) and focuses only on outpatients. While these are minor weaknesses (as the authors note, the majority of SARS-CoV-2 transmission likely occurs among those with symptoms below the threshold of hospitalization), it would nevertheless be useful to have a fuller understanding of variation in viral shedding across clinical groups. Furthermore, the study lacks information on viral shedding prior to the development of symptoms, which may be a critical period for transmission. Since the samples were collected at home by study participants using a novel apparatus, it is difficult to assess the degree to which actual variation in viral abundance, user variability, and/or measurement variation is inherent to the apparatus.

Author Response

Reviewer #1 (Public Review):

Summary:

This manuscript introduces an exciting way to measure SARS-CoV-2 aerosolized shedding using a disposable exhaled breath condensate collection device (EBCD). The paper draws the conclusion that the contagious shedding of the virus via aerosol route persists at a high level 8 days after symptoms.

Strengths:

The methodology is potentially of high importance and the paper is clearly written. The study design is clever. If aerosolized viral load kinetics truly differed from those of nasal swabs, then this would be a very important finding.

Thank you for your encouraging remarks. We agree that a comparison between aerosolized viral load and nasal swabs would strengthen our findings, and we have collected new specimens which will enable this comparison: In each session we collected both nasal swabs and exhaled breath samples, and we are in the process of analyzing these data. These data will be included in our revised manuscript.

Weaknesses:

The study conclusions are not entirely supported by the data for several reasons:

(1) Most data points in the study are relatively late during infection when viral loads from other compartments (nasal and oral swabs) are typically much lower than peak viral loads which often occur in the pre-symptomatic or early symptomatic phase of infection. Moreover, the generation time for SARS-CoV-2 has been estimated to be 3-4 days on average meaning that most infections occur before or very early during symptoms. Therefore, the available epidemiologic data does not support 12 days of infection (day 8 symptoms) as important for most transmissions. Therefore, many of the measurement timepoints in this study may not be relevant for transmission.

Thank you for your comment. Notably, our new data set includes a small number of specimens that were collected prior to the start of symptoms, and so we may be able to partially address this concern with those data. That said, we agree that a limitation of our study is that we were unable to collect specimens prior to symptom onset, and that this pre-symptomatic period represents a fruitful area for future work. However, significant questions do remain open regarding transmission dynamics of SARS-CoV-2, including the extent of transmission after symptom onset, and therefore, despite this limitation of our data, we feel that our method may contribute to further understanding of those dynamics. However, we will include a more prominent discussion of this limitation in the revised manuscript.

(2) Fig 1A would be more powerful as a correlation plot between viral load from nasal samples (x-axis) and aerosol (y-axis). One would expect at least a rough correlation (as has been seen between viral loads in oral and nasal samples) and deviations from this correlation would provide crucial information about how and when aerosol shedding is discordant from nasal samples (ie early vs late time points, low versus high viral loads< etc...). It is too strong to state correspondence is 100% when viral load is only measured in one compartment and nasal swabs are reduced to the oversimplified "positive or negative".

Thank you for this suggestion, we agree that the figure would be more powerful as a correlation plot between viral load from nasal samples and aerosol. Unfortunately, at the time these samples were collected, the ER at Northwestern Hospital was diagnosing SARS-CoV-2 patients using the Abbott ID NOW rapid diagnostic platform, which, despite being a PCR-based system, does not provide quantitative information about viral loads, and instead provides a binary positive/negative result. Since we were looking for a direct comparison between the clinical diagnostic test and our test, we considered the binary aspect of our data (detected/undetected), and found 100% correspondence, meaning that when the clinical test detected SARS-CoV-2, our test did too. We have collected additional data which includes quantitative PCR values from nasal swabs collected at the same time as breath samples and we will include these data in the format you suggest, once analyzed, in our revised manuscript.

(3) Results are reported in RNA copies which is fine but particle-forming units (pfu, or quantitative culture) are likely a more accurate surrogate of infectivity. It is quite possible that all of these samples would have been negative for pfu given that the ratio of RNA: pfu is often >1000 (though also dynamic over time during infection). This could be another indicator that most samples in the study were collected too late during infection to represent contagious time points.

We agree that culturing exhaled breath samples would be an important addition to our understanding of the transmission dynamics of SARS-CoV-2 and we consider this to be an important next step for our method. Because we did not perform culturing of our breath samples in this study, we avoided making claims about infectivity of our samples in this manuscript, and instead speculate about the future utility of our method in understanding transmission dynamics, once an appropriate surrogate of infectivity is performed. We will make sure this is clearer in the revised manuscript. That said, other groups have successfully cultured breath samples with corresponding CT values in a range that are well within the range we found in our study, and sufficient for transmission (for example, Alsved et al, 2023, CT range ~33-38). These studies support the idea that a significant portion of the viral RNA measured in our samples may come from viable virus. Therefore, quantifying the ratio of viable to nonviable virus in our samples is an important next step. We appreciate this comment, and we will add a clearer discussion of this point to the revised manuscript.

(4) Individual kinetic curves should be shown for participants with more than three time points to demonstrate whether there are clear kinetic trends within individuals that would help further validate this approach. The inclusion of single samples from individuals is less informative.

We will add individual kinetic curves to the revised manuscript.

(5) The S-shaped model in 2A is somewhat misleading as it is fit to means but there is tremendous variability within the data. Therefore the 8-day threshold should be listed clearly as a mean but not a rule for all individuals. The statement that viral RNA copies do not decrease until 8 days from symptom onset is unlikely to be true for all infected people and can't be made based on the available data in this study given that many people contributed only one datapoint.

We will clarify the language in the manuscript and make limitations of the 8-day interpretation clearer.

(6) The incubation period for SARS-CoV-2 is highly variable. Therefore duration of symptoms is a rather poor correlate of the duration of infection. This further diminishes the interpretive value of positive samples from individuals who were only sampled once.

We will add a discussion of this point to the revised manuscript.

Reviewer #2 (Public Review):

Summary:

In this manuscript, Lane and colleagues measured the abundance of SARS-CoV-2 on breath in 60 outpatients after the development of COVID-19 symptoms using a novel breath collection apparatus. They found that, overall, viral abundance remains high for approximately eight days following the development of symptoms, after which viral abundance on breath drops to a low level that may persist for approximately 20 days or more. They did not identify significant differences in viral shedding on breath by vaccination status or viral variant. They also noted substantial variation in the degree and duration of shedding across individuals.

Strengths:

The primary strengths of this study are (1) the focus on breath, rather than the more traditional nasal/oropharyngeal swabs, and (2) the fact that the data were collected at multiple time points for each infection. This allows the authors to characterize not only mean viral abundance across individuals but also how that abundance changes over time, allowing for a better understanding of the potential duration of infectiousness of SARS-CoV-2.

Weaknesses:

The sample size is moderate (60) and focuses only on outpatients. While these are minor weaknesses (as the authors note, the majority of SARS-CoV-2 transmission likely occurs among those with symptoms below the threshold of hospitalization), it would nevertheless be useful to have a fuller understanding of variation in viral shedding across clinical groups.

We agree this would be very interesting and feel our method, which is straightforward to perform in clinical settings, lends itself to future studies across clinical groups. We have added discussion of this to the discussion section of the manuscript.

Furthermore, the study lacks information on viral shedding prior to the development of symptoms, which may be a critical period for transmission. Since the samples were collected at home by study participants using a novel apparatus, it is difficult to assess the degree to which actual variation in viral abundance, user variability, and/or measurement variation is inherent to the apparatus.

This is a great point, which we will discuss in our revised manuscript.