Cytokine Dynamics and Herpesvirus Interactions in Pediatric Liver and Kidney Transplant Recipients: The Distinct Behavior of HCMV, HHV6, HHV7 and EBV

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript is well written.

Major comments

Within section 2.3. Statistical Analysis the authors use substitution method to maintain statistical rigor. While this reviewer understands that sample cytokine concentrations that lie outside the sensitivities of the assay are problematic for sample analysis, as any substitution using a constant value when reporting limits will distort estimates of the standard deviation, and therefore all subsequent (parametric) hypothesis tests. 

This may be further compounded if one sample set has >50% “negative” sample as seen for IL-(2), 4, 5, 6 and 10 and . Correlation coefficients, regression slopes, and their p-values should be considered particularly suspect when values are substituted for nondetects, especially if the statistics are found to be significant. Substitution methods should be considered at best ‘‘semi-quantitative’’, to be used only when approximate answers are required. Some indication that the authors recognize this inherent error should be stated clearly in the results. 

The authors may also consider revising  “raw data” values  in figure 1C to show a more detailed scatter plot to allow the reader to quickly grasp the number of negative value samples.

ROUT test does not tell you the number of outliers, rather it is the % success rate of detecting an outlier in a given sample set. Prism uses 1% a stringent means to tell the user their chance of detecting an outlier in multiple sets. The authors may be under the assumption that the data plots scatter is Gaussian curve. As an exercise perhaps presented in the appendix that would better  assure the reader that data was correctly analyzed the use of nonlinear regression, Cronbach’s alpha test or a ranking plot as a straightforward way to measure whether the raw data that included outliers is reliable.

DNAemia and virus reactivation should be used with caution. Increased plasma DNA as a reflection of EBV-virus production is still controversial and may be due to cell fragmentation of EBV harboring cells and not free virus.

 

Minor comments

At first use of RR define as “relative risk” in the text.

Author Response

The manuscript is well written.

Response: We sincerely appreciate the positive evaluation of our work and the insightful comments that have significantly enhanced our study.

Major comments

1. Within section 2.3. Statistical Analysis the authors use substitution method to maintain statistical rigor. While this reviewer understands that sample cytokine concentrations that lie outside the sensitivities of the assay are problematic for sample analysis, as any substitution using a constant value when reporting limits will distort estimates of the standard deviation, and therefore all subsequent (parametric) hypothesis tests.  This may be further compounded if one sample set has >50% “negative” sample as seen for IL-(2), 4, 5, 6 and 10 and . Correlation coefficients, regression slopes, and their p-values should be considered particularly suspect when values are substituted for nondetects, especially if the statistics are found to be significant. Substitution methods should be considered at best ‘‘semi-quantitative’’, to be used only when approximate answers are required. Some indication that the authors recognize this inherent error should be stated clearly in the results.

Response: In addressing the reviewer's concerns regarding the classification of positive samples below the limit of detection (LOD), we carefully considered the methodological implications of various substitution strategies as outlined in the literature. Specifically, we explored a recognized substitution method where samples falling below the LOD are either assigned a value of 0 (indicating a negative result) or the LOD itself (Croghan C EP. Methods of Dealing with Values Below the Limit of Detection using SAS. ResearchGate, 2003). https://ntrl.ntis.gov/NTRL/dashboard/searchResults.xhtml?searchQuery=PB2004-100886; (Cohen, A. C. Simplified Estimators for the Normal Distribution When Samples Are Singly Censored or Truncated, 1959. Technometrics, 1(3), 217–237. https://doi.org/10.2307/1266442).

We tested both approaches without finding significant differences in the outcomes, leading us to adopt the LOD substitution as the preferred method. This decision was also supported by the manufacturer guidelines of the cytokine 17-plex immunoassay, which advise that samples identified with concentrations below the LOD are not negative but fall within a range where the accuracy of the concentration measurements is compromised due to the method's detection power being below 95% confidence. Our choice to classify these samples using the LOD substitution method is rooted in an understanding that, particularly in the context of immunosuppressed patients, the presence of low or undetectable cytokine concentrations does not equate to an absence of biological significance. Indeed, our analysis, distinguishes between truly negative samples and those with concentrations below the LOD, thereby ensuring a more nuanced interpretation of the data. This approach is consistent with widely accepted practices across environmental studies, medical sciences, and analytical chemistry, providing a robust methodological foundation for our study.

Additionally, we explored the use of Maximum Likelihood Estimation (MLE) as an alternative analytical technique. However, MLE significantly altered the values of samples above the LOD (true positive samples), thereby skewing our data interpretation. Given these considerations, we concluded that the LOD substitution method offered the most accurate and meaningful approach for classifying samples with concentrations below the LOD, aligning with established scientific principles and ensuring the integrity of our data analysis.

Acknowledging the reviewer´s observation, we have added the following paragraphs to the 2.4 Statistical Analysis section: “Using substitution methods for values below the LOD can potentially distort estimates and statistical tests, particularly for cytokines with an elevated proportion of samples under the LOD. While we believe our approach is sound and follows established practices, we acknowledge this potential limitation.”

 

2. The authors may also consider revising “raw data” values  in figure 1C to show a more detailed scatter plot to allow the reader to quickly grasp the number of negative value samples.

Response. Because some cytokines had several negative samples, it is difficult to visualize them in any type of graph. However, we include graph 1b, which specifies the proportion of positive and negative samples for each cytokine. In addition, Table 3 shows the exact number of negative and positive samples for each cytokine. But we understand the reviewer's point, so we include the number of negative samples in graph 1c as well.

 

3. ROUT test does not tell you the number of outliers, rather it is the % success rate of detecting an outlier in a given sample set. Prism uses 1% a stringent means to tell the user their chance of detecting an outlier in multiple sets. The authors may be under the assumption that the data plots scatter is Gaussian curve. As an exercise perhaps presented in the appendix that would better  assure the reader that data was correctly analyzed the use of nonlinear regression, Cronbach’s alpha test or a ranking plot as a straightforward way to measure whether the raw data that included outliers is reliable.

Response: Thank you very much for the suggestion, in appendix B we add two scatter dot plots (Figure B1), the first is the distribution of the raw data, the second is the distribution of the clean data after the ROUT test.

 

4. DNAemia and virus reactivation should be used with caution. Increased plasma DNA as a reflection of EBV-virus production is still controversial and may be due to cell fragmentation of EBV harboring cells and not free virus.

Response: We were aware of the potential contamination from fragmented cells. Therefore, before the extraction of plasma DNA, a DNase test was performed to digest free-floating DNA. This ensures that the readings obtained were derived from protected DNA, reducing the likelihood of contamination from broken cells. We appreciate the observation and have added a small paragraph in section 2.3 of the Methods detailing this process: “Before the extraction of plasma DNA, a DNase test was performed to avoid quantifying viral DNA from broken cells.”

 

Minor comments

At first use of RR define as “relative risk” in the text.

Response: Thank you for your comment, we have already made the correction on line 163.

Reviewer 2 Report

Comments and Suggestions for Authors

The study is very exhaustive. Its compression may be complex for readers who are not experts on the subject. I would recommend to simplify the content by selecting the most relevant data. 

Specific comments:

Material and methods:

The authors describe the methodology used to determine plasma levels of cytokines and the rejection criteria. However, they do not describe the viral detection methodology or the cut-off points used to differentiate between latent infection or reactivation.

Line 81. ¨…..comprising 19 pediatric patiens”  However in Table 1: “13 renal and 7 liver”. Clarify how many patients have been included.

Results:

Lines 181 to 186: This paragraph is complex and difficult to understand.

 

 

Author Response

The study is very exhaustive. Its compression may be complex for readers who are not experts on the subject. I would recommend simplifying the content by selecting the most relevant data. 

Response: We are deeply thankful for the positive appraisal of our work and the thoughtful comments that have been pivotal in enriching our study.

The interaction between the various herpesvirus life cycles and the immune system in the context of pharmacological immunosuppression due to solid organ transplantation is extremely complex. Our work provides only a small snapshot of this complexity, which we understand is reflected in the paper. However, we would like to note that we have already selected the most relevant data from our full study, omitting some pieces of information to report a coherent narrative in the simplest possible way.

 

Specific comments:

Material and methods:

1.The authors describe the methodology used to determine plasma levels of cytokines and the rejection criteria. However, they do not describe the viral detection methodology, or the cut-off points used to differentiate between latent infection or reactivation.

Response: We did not describe the in-house multiplex qPCR used for viral detection in this manuscript because it was previously detailed in our earlier publication: Sanchez-Ponce, Y. et al. Simultaneous Detection of Beta and Gamma Human Herpesviruses by Multiplex qPCR Reveals Simple Infection and Coinfection Episodes Increasing Risk for Graft Rejection in Solid Organ Transplantation. Viruses, 2018. 10(12). doi: 10.3390/v10120730. PMID 30572622. This study is cited as reference number 11 in our current manuscript. To address this omission and enhance the understanding of our study, we have now added a section on viral detection in clinical samples to the Methodology section (lines 121-132), which reads as follows:

2.3. Viral Detection in Clinical Samples

Viral detection was performed using an in-house multiplex qPCR that simultaneously detects and quantifies beta and gamma human herpesviruses, as previously reported [11]. In this earlier study, blood samples were fractionated into cellular components and plasma. Since herpesviruses exhibit a bipartite life cycle oscillating between latent and lytic phases, viral loads found in cells were considered more indicative of latency, while viral loads in plasma were considered more indicative of an active lytic cycle.

Below is the complete description of the methodology used for viral detection as described in our previous article:

DNA was purified from 2–4 × 106 leukocytes using QIAamp DNA Mini Kit (Qiagen, Hilden, NRW, Germany) according to manufacturer’s instructions. Quantification was done in a nanodrop 1000 spectrophotometer (Thermo Fisher Scientific). The purity and integrity of DNA were evaluated through optical density (260/280 ratio), 1.5% agarose gel electrophoresis and amplification of the β-actin endogenous gene. Viral detection was conducted using 100 ng of DNA and then expressed as copy number per µg of DNA (Conversion factor = 10). We used 200 µL (1/5 of 1 mL) of plasma for DNA purification using QIAsymphony virus/bacterium mini kit (Qiagen, Hilden, NRW, Germany) in QIAsymphony SP equipment according to the manufacturer’s instructions. DNA was eluted in 60 µL of elution buffer of the kit. Virus detection was conducted in 5 µL (1/12 of the eluate) and then expressed as copy number per ml of plasma (conversion factor = 60 (12 × 5)). A total of 495 paired samples (leukocytes-plasma) were tested by triplicate with the two-tube multiplex qPCRs. The number of viral copies per sample was calculated from the standard curves and multiplied by the corresponding conversion factor to be reported throughout the text as viral copies per μg of DNA (from leukocytes) or per mL of plasma.

Regarding the cut-off points, we did not use specific cut-off points to differentiate between latent infection or reactivation. We hope this clarification, along with the new methodology section and sub-section 3.6 of the results (lines 355-361), makes it clearer. As explained, "In our previous analysis [11], we separated blood samples into cell and plasma fractions, reasoning that EBV detection in the cell fraction would be indicative of latent infection, while detection in the plasma fraction would be indicative of lytic infection and, consequently, of viral reactivation."

 

2. Line 81. ¨…..comprising 19 pediatric patients” However in Table 1: “13 renal and 7 liver”. Clarify how many patients have been included.

Response: Thank you very much for pointing out this discrepancy. It was a typographical error. Samples from 20 patients were analyzed, comprising 13 renal and 7 liver transplant recipients. The correct number has been updated in line 81 of the Materials and Methods section.

 

3. 3. Results: Lines 181 to 186: This paragraph is complex and difficult to understand.

Response. We appreciate the comment and have reworded the paragraph to enhance clarity (now lines 209-218). This paragraph now reads as follows:

“We compared cytokine concentrations in samples with single viral DNAemia, multiple viral DNAemia, and without DNAemia. Significant increases in the levels of TNF-α, IFN-γ, IL-17, IL-12, IL-8, IL-7, IL-2, IL-1β, IL-4, and IL-10 were observed in samples with viral detection, whether single or multiple, compared with those without viral detection. Generally, greater increases in cytokine levels were seen in samples positive for more than one herpesvirus, except for IL-7, IL-17, and IFN-γ, which were more elevated in samples with single viral DNAemia. IL-4 and IL-10 were only detected in samples with multiple DNAemia and were never detected in virus-negative or single-virus detection cases (Figure 2a). Cytokines MIP-1β, IL-13, IL-6, IL-5, G-CSF, and GM-CSF did not show any differences between sample groups and are therefore not presented.”

Reviewer 3 Report

Comments and Suggestions for Authors

In this manuscript, Sánchez-Ponce and others analyzed viral DNA and cytokine levels in the plasma specimens of solid transplant pediatric patients. Research was carefully planned and carried out. Figures look good and convincing. I only suggest some minor revisions.

 

Abstract, line 24-26, “However, in co-detection scenarios with β-herpesviruses, EBV transitioned to a lytic state, contributing to heightened cytokinemia and graft rejection”. I agree with the authors that CMV/HHV6/7 might cause EBV reactivation that led to high cytokine levels. In the meantime, I also speculate that there is a possibility that higher expression of cytokines, due to unknown reasons, came first, and the higher cytokines caused reactivation of multiple herpesviruses.

 

Typos all over the manuscript even in the figures and tables: INFg must read IFNg.

 

Typos “ADNemia” (page 13 line 324 and 338) must be DNAemia.

 

The authors use the word “Simple” as a counterpart of “Multiple”, all over the manuscript. However, I believe “Single” is a more accurate word here as opposed to “Multiple”.

 

 

Comments on the Quality of English Language

Proofreading in English is prerequisite.

 

Author Response

1. In this manuscript, Sánchez-Ponce and others analyzed viral DNA and cytokine levels in the plasma specimens of solid transplant pediatric patients. Research was carefully planned and carried out. Figures look good and convincing. I only suggest some minor revisions.

Response: We are deeply thankful for the reviewer's positive appraisal of our work and the thoughtful and detailed comments that have significantly enhanced our study.

 

2. Abstract, line 24-26, “However, in co-detection scenarios with β-herpesviruses, EBV transitioned to a lytic state, contributing to heightened cytokinemia and graft rejection”. I agree with the authors that CMV/HHV6/7 might cause EBV reactivation that led to high cytokine levels. In the meantime, I also speculate that there is a possibility that higher expression of cytokines, due to unknown reasons, came first, and the higher cytokines caused reactivation of multiple herpesviruses.

Response: Thank you for this observation. In agreement with the reviewer, it is crucial to emphasize that our study is inherently associative. We have added this clarification to the abstract, which now states that this is an associative study. In addition, the sentence in question has been revised to: “However, in co-detection scenarios with β-herpesviruses, EBV transitioned to a lytic state, also associating with heightened cytokinemia and graft rejection.” This change eliminates the implication of causality and emphasizes the associative nature of our observations.

Additionally, we include a study limitation paragraph in the discussion section to further emphasize the observational nature of our study (lines 526-549). This paragraph now begins with: “An important limitation of this study is that it is an associative study, and in this complex interplay of viruses, cytokines, and graft rejection, we cannot conclude which one comes first, triggering the others.” Our analyses aim to shed light on the associations between viral dynamics and clinical outcomes, providing a foundational understanding that could inform future studies designed to interrogate causal relationships more directly.

 

3. Typos all over the manuscript even in the figures and tables: INFg must read IFNg.

Response: Thanks for the comment, we have already corrected all typos, in the manuscript (lines 172, 189, 191 y 197), in the table 3 and figures (1, 2, 3 and 4). 

 

4. Typos “ADNemia” (page 13 line 324 and 338) must be DNAemia.

Response: Thanks for the comment. We have already corrected all typos, in the manuscript (lines 336 and 350).

 

5. The authors use the word “Simple” as a counterpart of “Multiple”, all over the manuscript. However, I believe “Single” is a more accurate word here as opposed to “Multiple”.

Response: Thanks for the insightful suggestion, we agree with the reviewer and therefore we modified the word simple to single in the whole paper.

 

6. Comments on the Quality of English Language. Proofreading in English is prerequisite.

Response: We have carefully read the manuscript correcting all typos and grammatical errors found.