Rapid and sensitive detection of SARS-CoV-2 infection using quantitative peptide enrichment LC-MS analysis

  1. Andreas Hober
  2. Khue Hua Tran-Minh
  3. Dominic Foley
  4. Thomas McDonald
  5. Johannes PC Vissers
  6. Rebecca Pattison
  7. Samantha Ferries
  8. Sigurd Hermansson
  9. Ingvar Betner
  10. Mathias Uhlén
  11. Morteza Razavi
  12. Richard Yip
  13. Matthew E Pope
  14. Terry W Pearson
  15. Leigh N Andersson
  16. Amy Bartlett
  17. Lisa Calton
  18. Jessica J Alm
  19. Lars Engstrand
  20. Fredrik Edfors

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This study describes the addition of peptide immunoprecipitation by means of SISCAPA technology to the Sars-Cov2 mass spectrometry-based diagnostics toolbox. While MS-based viral detection has been described by several other groups, the presented method enriches SARS-CoV-2 peptides from clinical samples, thus enhancing the sensitivity and addressing one of the major limitations of this new method for viral diagnostics. The work is straightforward and of potentially great importance to the field. It shows beyond any doubt that mass spectrometry can become a clinically applied diagnostic instrument to detect (viral) infection.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Reliable, robust, large-scale molecular testing for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is essential for monitoring the ongoing coronavirus disease 2019 (COVID-19) pandemic. We have developed a scalable analytical approach to detect viral proteins based on peptide immuno-affinity enrichment combined with liquid chromatography-mass spectrometry (LC-MS). This is a multiplexed strategy, based on targeted proteomics analysis and read-out by LC-MS, capable of precisely quantifying and confirming the presence of SARS-CoV-2 in phosphate-buffered saline (PBS) swab media from combined throat/nasopharynx/saliva samples. The results reveal that the levels of SARS-CoV-2 measured by LC-MS correlate well with their correspondingreal-time polymerase chain reaction (RT-PCR) read-out (r = 0.79). The analytical workflow shows similar turnaround times as regular RT-PCR instrumentation with a quantitative read-out of viral proteins corresponding to cycle thresholds (Ct) equivalents ranging from 21 to 34. Using RT-PCR as a reference, we demonstrate that the LC-MS-based method has 100% negative percent agreement (estimated specificity) and 95% positive percent agreement (estimated sensitivity) when analyzing clinical samples collected from asymptomatic individuals with a Ct within the limit of detection of the mass spectrometer (Ct ≤ 30). These results suggest that a scalable analytical method based on LC-MS has a place in future pandemic preparedness centers to complement current virus detection technologies.

Article activity feed

  1. Version published to 10.7554/elife.70843 on eLife
  2. Version published to 10.1101/2021.06.02.21258097v2 on medRxiv
  3. eLife

    Author Response:

    Reviewer #1 (Public Review):

    Summary: In " Rapid and Sensitive Detection of SARS-CoV-2 Infection Using Quantitative Peptide Enrichment 1 LC-MS/MS Analysis" Hober, A. et al. describe the addition of peptide immunoprecipitation by means of SISCAPA technology to the Sars-Cov2 mass spectrometry-based diagnostics toolbox. The work shows in a straightforward way that this is a huge improvement and of great importance to the field. It shows beyond any doubt that mass spectrometry can become a clinically applied diagnostic instrument to detect (viral) infection.

    Overall remark: The main concern is the reported number of 83% sensitivity. This is not because the number is too low, but because the number is misleading. In line with "CLSI EP 12-A2 User Protocol for Evaluation of Qualitative Test Performance guidance" a summary of the sample analysis results are shown in a 2x2 contingency table. Unfortunately, I oppose to this representation of the results at this stage for three reasons: (i) reporting a percentage shouldn't be done on less than 100 samples because of the weight of a few misannotated samples on these numbers, be it in the qPCR or the MS results; (ii) because both assays are imperfect, it is impossible to assess the ground truth for calling patients and thus assess sensitivity and specificity; (iii) the authors still only target a single peptide, which is not conventional in MS-based assays that targets proteins.

    We have changed to PPA and NPA in the new version of the manuscript. We have also included 264 RT-PCR negative samples collected in the same study. We agree that protein quantification should not be done using only one single peptide. We have updated the manuscript to clarify that we do not perform protein quantification, but rather peptide quantification.

    Rather than the proposed confusion matrix, which assumes that the ground truth is known to call it e.g. "false negatives", the authors could refer to it as an agreement matrix and not be tempted to calculate threshold values like sensitivity, which have too much of an impact on the clinical readership that is used to seeing this value in a more controlled context. This is in line with the recent Lancet manuscript from Fitzpatrick, M. et al (2021), proposing percent positive agreement (PPA) and percent negative agreement (PNA) instead (Fitzpatrick et al., 2021).

    We have decided to keep the confusion matrix but we are referring to it as PPA and NPA and rephrased sensitivity to “estimated sensitivity” based on PPA.

    More specifically, as we and others have shown, qPCR Ct values rarely agree in two (consecutive) analyses, even within accredited settings (personal communication NHS). Above Ct30, patients regularly turned negative in our hands (https://doi.org/10.1021/jacsau.1c00048), even with an assay that had proven detectability of 1 plasmid at Ct40. Furthermore, we suspect that freeze-thaw cycles further inflate this uncertainty, two of which the current samples were subjected to. Undetected mRNA would then classify these patient samples as "false positives" if they did yield signal in the LCMS results. By chance, this did not happen in this manuscript, yet this could very well be the reason for the highest signal reported in Figure 3 as a green dot at log2 MRM response of -6 (see minor remarks).

    The authors already distinguished the patients in a High Pool of Ct <30, a Low Pool 30{less than or equal to}Ct<33 and the negative samples (Ct>40). It is clear from the gap (no 34<Ct<39) that finding patients between Ct33 and Ct39 is challenging. Indeed, qPCR has its own "diagnostic grey zone" of LOQ negative and LOQ positive that rarely is being referenced. Thus, a "sensitivity" of 95% for patients <Ct30, despite the low number of samples and considering the uncertainties in qPCR (just above or below Ct30) at least limits the comparison to samples that are positive beyond any doubt. But again, we would be thresholding against a trembling metric, in turn making the claim from the authors dangerous that "the estimated LLOQ is 3 amol/μL approximates to Ct {less than or equal to}30". Rather, the Ct30 threshold should be set for a different reason, if one is chosen at all.

    What is needed is good thresholding for clinical diagnostics, as is done in qPCR. In the public hospital in Belgium that provided us with patient samples, the positive threshold is set to Ct33 on the first measurement and practitioners use higher Ct values only in the context of physical symptoms of the disease to come to a final conclusion. For MS, we now need to measure >1000 samples in order to decide what log2 MRM response for a given set of peptides corresponds to an LOQ positive from - say - Ct27 to Ct30 and an LOQ Negative from Ct31 to Ct33. In other words, the linearity of the correlation between qPCR and MS illustrates the intrinsic value of MS; the point up until which we can provide clinically relevant information remains to be determined on large patient cohorts. In turn, these large patient cohorts can allow to sort (clinically) validated patients according to signal intensity and set a log2 threshold at which e.g. 2% or 5% negatives are expected, in line with False Discovery calculations for target decoy strategies. At this stage however, it might be most straightforward to conclude with percent positive agreement (PPA) and percent negative agreement (PNA), as is recommended for laminar flow tests validated on <100 samples.

    Finally, realizing the importance of this pivotal moment in the implementation of MS in the clinic, I find it somewhat tricky to only focus on one peptide. In fact, the authors perform the qPCR on two genes (three genes being even more common) because of the drop-outs that can occur. I feel like the use of peptide IP with MRM for detecting pathogens has not yet matured enough to rely solely on one peptide. Still, I understand that asking for a second peptide would mean repeating all the measurements, so that is most probably not realistic. Yet, I do consider this to be yet another reason not to report % sensitivity and specificity in the current manuscript and the potential to gain robustness with more peptides should clearly be emphasized at every stage of the manuscript.

    We agree that the method would be much improved by adding another peptide to the repertoire. The method was developed using the most sensitive antibody-peptide pair and the most promising pair was used in the downstream process. We have highlighted the limitations of using only one peptide and emphasized that this is a proof-of-principle study.

    In conclusion, because patient batches in the thousands are currently unavailable to MS-oriented diagnostic labs and because of all the reasons mentioned above, we cannot report the numbers of sensitivity and specificity in this manuscript, as they are misleading and do not quantify what they are intended to do.

    Fitzpatrick, M. C. et al. (2021) 'Buyer beware: inflated claims of sensitivity for rapid COVID-19 tests', The Lancet. Lancet Publishing Group, pp. 24-25. doi: 10.1016/S0140-6736(20)32635-0.

    We agree and have changed to PPA and NPA for this reason.

    Major remarks: P3L250: "on-column amount of 60 amol." Because of the enrichment procedure, could the authors specify what initial conditions they spiked into the dilution series prior to enrichment. This would allow recalculation and avoid confusion about the correctness of the 60 amol on column claim (which in our hands is still detectable).

    We made changes to this in the updated version of the manuscript.

    P8L181: "50 μL elution buffer (0.5 % 180 formic acid, 0.03% CHAPS, 1X PBS) and incubated for 5 min at room temperature." This minor sentence is placed under major remarks, because in our understanding the elution buffer needs to be acidic and adding PBS will reduce acidity. If this is a typo, please correct. If this is not, could the authors try and use H2O instead and see if their results improve?

    The access to the raw data was denied.

    The raw data is accessible through the provided Panorama link and can be accessed under the tab “Raw Data”. The entry in ProteomeXchange, however, is only a reserved data set identifier for now, but the data will be made available through this link after the review process.

    Reviewer #2 (Public Review):

    MS-based proteomics is currently discussed as a method for detection of viruses from clinical samples. Several studies have already shown the potential of this method on the example of the detection of SARS-CoV-2 from respiratory specimens. However, one of the major drawbacks still remains the low sensitivity of MS-based virus detection compared to real-time PCR, which is the gold-standard method. In their manuscript Hober and colleagues apply specific antibody-based enrichment of SARS-CoV-2 peptides from upper airway samples to concentrate the analyte prior to analysis by targeted MS (MRM). The authors determined the dynamic range of the method for four different SARS-CoV-2 NCAP peptides using a calibration curve. On the example of the SARS-CoV-2 NCAP peptide AYNVTQAFGR a correlation between the MS result and the cT value is shown. Furthermore, using stable isotope labelled (SIL) peptides as internal reference, a quantitative MS measurement was achieved. The presented approach is able to distinguish real-time PCR SARS-CoV-2 positive samples from negative samples in the used set of 88 samples from asymptomatic patients. Combined with a specificity of 100 % and sensitivities of up to 94.7 % for samples with cT values {less than or equal to} 30 the authors conclude that the method could be an alternative to real-time PCR.

    Strengths of the manuscript:

    I think the applied technique (SISCAPA) is highly interesting in the context of virus proteomics. This is because virus proteins are often underrepresented in relation to the host proteins, especially during early time points of infection, hampering their detection. Recently, the application of SISCAPA for SARS-CoV-2 diagnostics has been suggested in the discussion of a manuscript from Van Puyvelde and colleagues. The manuscript from Hober and colleagues presents the first study demonstrating that this technique can be applied to enrich, detect and quantify SARS-CoV-2 peptides from upper airway samples. The manuscript is clearly arranged, the data is sound and supports the main conclusions.

    Weaknesses of the manuscript:

    I think the manuscript in some points underestimates the PCR and vice versa overemphasizes the proteomics approach. For example, I don't agree that real-time PCR generally suffers from technical problems, degraded probes or non-specific amplification. Vice versa I think the LC-MS/MS approach is not inherently absolute specific and does not outperform PCR in terms of specificity. Further, LC-MS/MS does not eliminate the problem of false positives, which could be introduced during sample preparation or by inter-run contaminations. Although in real-time PCR no internal standards analogous to isotopically labelled peptides are used there are internal controls used to assure the quality of the extraction and the PCR reaction itself. The method presented by Hober and colleagues is clearly beneficial for the field of proteomics-based virus detection, but I suggest a more balanced discussion also including also the potential drawbacks of the method.

    Another point I like to raise is that the authors conclude at the end of the results section that patient samples were collected at an infectious stage.

    We have made changes to the manuscript accordingly and removed the claim that the samples were collected in an infectious stage since this cannot be confirmed. The patients did not show any symptoms when sampled, which has been highlighted in the new version.

    However, an assessment of the infectivity cannot be drawn from the presented data. The analysis of real-time PCR results in the manuscript is based on cT values. But to draw the conclusion, that the analysed samples contained infectious virus particles, the number of viral genome equivalents has to be determined, which in turn can be correlated to infectivity.

    We have removed this section since we cannot make any claim on infectious virus-particles.

    The detection of viral proteins itself does not proof that samples were collected at an infectious stage and there is currently no correlate of the amount of NCAP protein and infectivity. Since viral proteins are likely more stable than viral RNA, they could even be detectable for a more prolonged time in patient samples.

    Reviewer #3 (Public Review):

    Major comments

    P2, l245, Figure 2: It is not completely clear to me what is represented in panels A and B. Is this the pure SIL peptide of the endogenous peptide in a complex matrix? This may make a large difference for the determination of the LLOQ. Panel B shows a calibration curve and as these are curves for which the signal is detected based on known input amounts of sample, I assume that the input is pure SIL peptide here?

    In panel A, what does '3 amol/ul' in the middle chromatogram exactly mean? Is this the endogenous peptide that was calculated to be present at 3 amol/ul based on a known concentration of spiked-in SIL peptide?

    P4, l276: The authors need to explain the details of data imputation. It is unclear which data were imputed and how this was done. In Figure 3 the grey data points represent "not detected" or "inconclusively identified" samples by LC-MS, while some of the data points seem to have a higher 'response' values than others. Please explain.

    In Figure 3, how is 'response' defined? I don't understand the following sentence (p4, l277): "… for the LC-MS results the lowest response divided by three was used, mimicking….". Which variable does the data point size reflect? There seem to be clear differences in ball sizes. Please explain. For clarity, it would be advisable to keep the y-axes for panels A and B identical. Also, how could RT-PCR values be not obtained, apparently leading to missing Ct values (p5, l278)?

    Assuming that all collected samples from individuals in the test group in this study are visualized in Figure 3, the majority was tested positive for SARS-CoV-2. This is very different from the percentages oberserved in regular testing facilities. How was the study group composed? Were these individuals who were already admitted to the hospital?

    We have specified that the sampels were selected based on RT-PCR result and have included more negative samples in the new version of the mansucript. We have also speciied how individuals were enrolled into the study.

    It would be interesting to include more negatively tested individuals to see the distribution of 'MRM response' values in this group, since some of the negatively tested individuals (green data points) show higher than expected MRM response values if no viral protein is present at all. Related to this, I do not understand how a specificity score of 100 % (p5, l292) was obtained while some green data points (negative by RT-PCR) have higher associated MRM response values than some of the blue (positive by RT-PCR) samples. Can the authors explain this?

    The negative samples that show a stronger MRM response do not have the required qualifying ions, thereby failing the QC parameter of the assay. This has been clarified in the new version of the manuscript.

    I find the text from p6, l298 ("However…") onward more suited for the Discussion section, since this is about the interpretation of the results presented here and the use of the described methodology in diagnostics; no results are shown in this part.

  4. eLife

    Evaluation Summary:

    This study describes the addition of peptide immunoprecipitation by means of SISCAPA technology to the Sars-Cov2 mass spectrometry-based diagnostics toolbox. While MS-based viral detection has been described by several other groups, the presented method enriches SARS-CoV-2 peptides from clinical samples, thus enhancing the sensitivity and addressing one of the major limitations of this new method for viral diagnostics. The work is straightforward and of potentially great importance to the field. It shows beyond any doubt that mass spectrometry can become a clinically applied diagnostic instrument to detect (viral) infection.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

  5. eLife

    Reviewer #1 (Public Review):

    Summary:
    In " Rapid and Sensitive Detection of SARS-CoV-2 Infection Using Quantitative Peptide Enrichment 1 LC-MS/MS Analysis" Hober, A. et al. describe the addition of peptide immunoprecipitation by means of SISCAPA technology to the Sars-Cov2 mass spectrometry-based diagnostics toolbox. The work shows in a straightforward way that this is a huge improvement and of great importance to the field. It shows beyond any doubt that mass spectrometry can become a clinically applied diagnostic instrument to detect (viral) infection.

    Overall remark:
    The main concern is the reported number of 83% sensitivity. This is not because the number is too low, but because the number is misleading. In line with "CLSI EP 12-A2 User Protocol for Evaluation of Qualitative Test Performance guidance" a summary of the sample analysis results are shown in a 2x2 contingency table. Unfortunately, I oppose to this representation of the results at this stage for three reasons: (i) reporting a percentage shouldn't be done on less than 100 samples because of the weight of a few misannotated samples on these numbers, be it in the qPCR or the MS results; (ii) because both assays are imperfect, it is impossible to assess the ground truth for calling patients and thus assess sensitivity and specificity; (iii) the authors still only target a single peptide, which is not conventional in MS-based assays that targets proteins.

    Rather than the proposed confusion matrix, which assumes that the ground truth is known to call it e.g. "false negatives", the authors could refer to it as an agreement matrix and not be tempted to calculate threshold values like sensitivity, which have too much of an impact on the clinical readership that is used to seeing this value in a more controlled context. This is in line with the recent Lancet manuscript from Fitzpatrick, M. et al (2021), proposing percent positive agreement (PPA) and percent negative agreement (PNA) instead (Fitzpatrick et al., 2021).

    More specifically, as we and others have shown, qPCR Ct values rarely agree in two (consecutive) analyses, even within accredited settings (personal communication NHS). Above Ct30, patients regularly turned negative in our hands (https://doi.org/10.1021/jacsau.1c00048), even with an assay that had proven detectability of 1 plasmid at Ct40. Furthermore, we suspect that freeze-thaw cycles further inflate this uncertainty, two of which the current samples were subjected to. Undetected mRNA would then classify these patient samples as "false positives" if they did yield signal in the LCMS results. By chance, this did not happen in this manuscript, yet this could very well be the reason for the highest signal reported in Figure 3 as a green dot at log2 MRM response of -6 (see minor remarks).

    The authors already distinguished the patients in a High Pool of Ct <30, a Low Pool 30{less than or equal to}Ct<33 and the negative samples (Ct>40). It is clear from the gap (no 34<Ct<39) that finding patients between Ct33 and Ct39 is challenging. Indeed, qPCR has its own "diagnostic grey zone" of LOQ negative and LOQ positive that rarely is being referenced. Thus, a "sensitivity" of 95% for patients <Ct30, despite the low number of samples and considering the uncertainties in qPCR (just above or below Ct30) at least limits the comparison to samples that are positive beyond any doubt. But again, we would be thresholding against a trembling metric, in turn making the claim from the authors dangerous that "the estimated LLOQ is 3 amol/μL approximates to Ct {less than or equal to}30". Rather, the Ct30 threshold should be set for a different reason, if one is chosen at all.

    What is needed is good thresholding for clinical diagnostics, as is done in qPCR. In the public hospital in Belgium that provided us with patient samples, the positive threshold is set to Ct33 on the first measurement and practitioners use higher Ct values only in the context of physical symptoms of the disease to come to a final conclusion. For MS, we now need to measure >1000 samples in order to decide what log2 MRM response for a given set of peptides corresponds to an LOQ positive from - say - Ct27 to Ct30 and an LOQ Negative from Ct31 to Ct33. In other words, the linearity of the correlation between qPCR and MS illustrates the intrinsic value of MS; the point up until which we can provide clinically relevant information remains to be determined on large patient cohorts. In turn, these large patient cohorts can allow to sort (clinically) validated patients according to signal intensity and set a log2 threshold at which e.g. 2% or 5% negatives are expected, in line with False Discovery calculations for target decoy strategies. At this stage however, it might be most straightforward to conclude with percent positive agreement (PPA) and percent negative agreement (PNA), as is recommended for laminar flow tests validated on <100 samples.

    Finally, realizing the importance of this pivotal moment in the implementation of MS in the clinic, I find it somewhat tricky to only focus on one peptide. In fact, the authors perform the qPCR on two genes (three genes being even more common) because of the drop-outs that can occur. I feel like the use of peptide IP with MRM for detecting pathogens has not yet matured enough to rely solely on one peptide. Still, I understand that asking for a second peptide would mean repeating all the measurements, so that is most probably not realistic. Yet, I do consider this to be yet another reason not to report % sensitivity and specificity in the current manuscript and the potential to gain robustness with more peptides should clearly be emphasized at every stage of the manuscript.

    In conclusion, because patient batches in the thousands are currently unavailable to MS-oriented diagnostic labs and because of all the reasons mentioned above, we cannot report the numbers of sensitivity and specificity in this manuscript, as they are misleading and do not quantify what they are intended to do.

    Fitzpatrick, M. C. et al. (2021) 'Buyer beware: inflated claims of sensitivity for rapid COVID-19 tests', The Lancet. Lancet Publishing Group, pp. 24-25. doi: 10.1016/S0140-6736(20)32635-0.

    Major remarks:
    P3L250: "on-column amount of 60 amol." Because of the enrichment procedure, could the authors specify what initial conditions they spiked into the dilution series prior to enrichment. This would allow recalculation and avoid confusion about the correctness of the 60 amol on column claim (which in our hands is still detectable).

    P8L181: "50 μL elution buffer (0.5 % 180 formic acid, 0.03% CHAPS, 1X PBS) and incubated for 5 min at room temperature." This minor sentence is placed under major remarks, because in our understanding the elution buffer needs to be acidic and adding PBS will reduce acidity. If this is a typo, please correct. If this is not, could the authors try and use H2O instead and see if their results improve?

    The access to the raw data was denied.

  6. eLife

    Reviewer #2 (Public Review):

    MS-based proteomics is currently discussed as a method for detection of viruses from clinical samples. Several studies have already shown the potential of this method on the example of the detection of SARS-CoV-2 from respiratory specimens. However, one of the major drawbacks still remains the low sensitivity of MS-based virus detection compared to real-time PCR, which is the gold-standard method. In their manuscript Hober and colleagues apply specific antibody-based enrichment of SARS-CoV-2 peptides from upper airway samples to concentrate the analyte prior to analysis by targeted MS (MRM). The authors determined the dynamic range of the method for four different SARS-CoV-2 NCAP peptides using a calibration curve. On the example of the SARS-CoV-2 NCAP peptide AYNVTQAFGR a correlation between the MS result and the cT value is shown. Furthermore, using stable isotope labelled (SIL) peptides as internal reference, a quantitative MS measurement was achieved. The presented approach is able to distinguish real-time PCR SARS-CoV-2 positive samples from negative samples in the used set of 88 samples from asymptomatic patients. Combined with a specificity of 100 % and sensitivities of up to 94.7 % for samples with cT values {less than or equal to} 30 the authors conclude that the method could be an alternative to real-time PCR.

    Strengths of the manuscript:

    I think the applied technique (SISCAPA) is highly interesting in the context of virus proteomics. This is because virus proteins are often underrepresented in relation to the host proteins, especially during early time points of infection, hampering their detection. Recently, the application of SISCAPA for SARS-CoV-2 diagnostics has been suggested in the discussion of a manuscript from Van Puyvelde and colleagues. The manuscript from Hober and colleagues presents the first study demonstrating that this technique can be applied to enrich, detect and quantify SARS-CoV-2 peptides from upper airway samples. The manuscript is clearly arranged, the data is sound and supports the main conclusions.

    Weaknesses of the manuscript:

    I think the manuscript in some points underestimates the PCR and vice versa overemphasizes the proteomics approach. For example, I don't agree that real-time PCR generally suffers from technical problems, degraded probes or non-specific amplification. Vice versa I think the LC-MS/MS approach is not inherently absolute specific and does not outperform PCR in terms of specificity. Further, LC-MS/MS does not eliminate the problem of false positives, which could be introduced during sample preparation or by inter-run contaminations. Although in real-time PCR no internal standards analogous to isotopically labelled peptides are used there are internal controls used to assure the quality of the extraction and the PCR reaction itself. The method presented by Hober and colleagues is clearly beneficial for the field of proteomics-based virus detection, but I suggest a more balanced discussion also including also the potential drawbacks of the method.
    Another point I like to raise is that the authors conclude at the end of the results section that patient samples were collected at an infectious stage. However, an assessment of the infectivity cannot be drawn from the presented data. The analysis of real-time PCR results in the manuscript is based on cT values. But to draw the conclusion, that the analysed samples contained infectious virus particles, the number of viral genome equivalents has to be determined, which in turn can be correlated to infectivity. The detection of viral proteins itself does not proof that samples were collected at an infectious stage and there is currently no correlate of the amount of NCAP protein and infectivity. Since viral proteins are likely more stable than viral RNA, they could even be detectable for a more prolonged time in patient samples.

  7. eLife

    Reviewer #3 (Public Review):

    Major comments

    P2, l245, Figure 2: It is not completely clear to me what is represented in panels A and B. Is this the pure SIL peptide of the endogenous peptide in a complex matrix? This may make a large difference for the determination of the LLOQ. Panel B shows a calibration curve and as these are curves for which the signal is detected based on known input amounts of sample, I assume that the input is pure SIL peptide here?

    In panel A, what does '3 amol/ul' in the middle chromatogram exactly mean? Is this the endogenous peptide that was calculated to be present at 3 amol/ul based on a known concentration of spiked-in SIL peptide?

    P4, l276: The authors need to explain the details of data imputation. It is unclear which data were imputed and how this was done. In Figure 3 the grey data points represent "not detected" or "inconclusively identified" samples by LC-MS, while some of the data points seem to have a higher 'response' values than others. Please explain.

    In Figure 3, how is 'response' defined? I don't understand the following sentence (p4, l277): "... for the LC-MS results the lowest response divided by three was used, mimicking....". Which variable does the data point size reflect? There seem to be clear differences in ball sizes. Please explain. For clarity, it would be advisable to keep the y-axes for panels A and B identical. Also, how could RT-PCR values be not obtained, apparently leading to missing Ct values (p5, l278)?

    Assuming that all collected samples from individuals in the test group in this study are visualized in Figure 3, the majority was tested positive for SARS-CoV-2. This is very different from the percentages oberserved in regular testing facilities. How was the study group composed? Were these individuals who were already admitted to the hospital? It would be interesting to include more negatively tested individuals to see the distribution of 'MRM response' values in this group, since some of the negatively tested individuals (green data points) show higher than expected MRM response values if no viral protein is present at all. Related to this, I do not understand how a specificity score of 100 % (p5, l292) was obtained while some green data points (negative by RT-PCR) have higher associated MRM response values than some of the blue (positive by RT-PCR) samples. Can the authors explain this?

    I find the text from p6, l298 ("However...") onward more suited for the Discussion section, since this is about the interpretation of the results presented here and the use of the described methodology in diagnostics; no results are shown in this part.

  8. ScreenIT

    SciScore for 10.1101/2021.06.02.21258097: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    EthicsIRB: Sample collection: The study was performed in accordance with the declaration of Helsinki and the study protocol (“Jämförande studier av Covid-19 smitta och antikroppssvar i olika grupper i samhället”) was approved by the Ethical Review Board of Linköping, Sweden (Regionala etikprövningsnämnden, Linköping, DNR - 2020-06395).
    Sex as a biological variablenot detected.
    Randomizationnot detected.
    Blindingnot detected.
    Power Analysisnot detected.

    Table 2: Resources

    No key resources detected.

    Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

    Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.

    Results from TrialIdentifier: No clinical trial numbers were referenced.

    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

    Results from JetFighter: We did not find any issues relating to colormaps.

    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    Results from scite Reference Check: We found no unreliable references.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  9. Version published to 10.1101/2021.06.02.21258097v1 on medRxiv