The Influence of Peripheral Blood Eosinophil Counts in Asthma Comorbidities in Adults: A Real Life Study

Round 1

Reviewer 1 Report

This manuscript by Esteban-Gorgojo et al. deals with the important and incompletely understood relationship between eosinophils and features of asthma. An understanding of this relationship is critically important for the future design and appropriate prescription of asthma therapies. This retrospective study has a large pool of subjects and examines associations between elevated peripheral eosinophil counts and asthma features, comorbidities, and pulmonary function using several different eosinophil count thresholds.

However, several issues of concern remain in the manuscript. My specific comments are as follows:

1. Data that are discussed in much of the Results section do not appear in any of the tables. For example, none of the associations using thresholds other than 300 eosinophils per microliter appear in the tables. If these data form a key part of the paper, as they currently do, they must be included.
2. The title does not accurately convey the findings of the manuscript: it asserts the comorbidities determine eosinophil counts when causation (in either direction) was not examined and cannot be examined using this dataset. Similar conclusions drawn in the manuscript, e.g., “Atopy is a determinant of elevated eosinophilia in this study” (line 233), should be re-worded to indicate an association or correlation instead.
3. Similarly, lack of evidence for a particular association is not equivalent to evidence of a lack of association. The statement that “we can conclude from this study that the risk of rhinitis only increases in non-asthmatic patients” (line 236-237) should be re-worded.
4. It appears that logistic regressions were not performed to adjust for patient demographics or comorbidities. If these regressions were performed, the results should be shown. Given that elevated eosinophil counts are more common in subjects who are younger or male, some of the other associations with eosinophil levels may instead be related to gender or age. It is also not clear that current treatments that the subjects were taking were considered, which may similarly skew results if they differ between those with and without elevated eosinophil counts.
5. The eosinophil count means and standard deviations in the tables are suspiciously round numbers. The meaning of the numbers in parentheses for other characteristics, such as FeNO and total IgE, are also unclear. Please check that these numbers are accurate and explained in the table legends.
6. OR and RR values should be greater than 1 when referring to a particular feature being favored/more prevalent in the eosinophilic group of subjects.
7. The concept that elevated peripheral eosinophil counts are associated with particular asthma features, especially severe, exacerbation-prone asthma, is not novel, and eosinophil counts are regularly considered in the asthma endotyping. The manuscript should be clearer about the novelty of and contributions made by these findings.
8. There are a number of grammatical and typographical errors in the manuscript that should be corrected.

Author Response

Thank you for your suggestions, we appreciate your comments. We will now proceed to respond to your suggestions.

• Data that are discussed in much of the Results section do not appear in any of the tables. For example, none of the associations using thresholds other than 300 eosinophils per microliter appear in the tables. If these data form a key part of the paper, as they currently do, they must be included.

Thank you for your contribution, we agree with you that there are results that are not shown in the tables, but the maximum number of tables to be included in the journal did not allow us to include them all, so we included the tables that we thought were most relevant. If we were allowed to add more tables, we would be happy to provide more tables with the data you suggest.

• The title does not accurately convey the findings of the manuscript: it asserts the comorbidities determine eosinophil counts when causation (in either direction) was not examined and cannot be examined using this dataset. Similar conclusions drawn in the manuscript, e.g., “Atopy is a determinant of elevated eosinophilia in this study” (line 233), should be re-worded to indicate an association or correlation instead.

Thank you for your suggestions, we have modified the title according to your comments. We have also modified line 233 (191 in our manuscript version) according to your suggestions.

• Similarly, lack of evidence for a particular association is not equivalent to evidence of a lack of association. The statement that “we can conclude from this study that the risk of rhinitis only increases in non-asthmatic patients” (line 236-237) should be re-worded.

Dear reviewer, we believe that in this sentence we are not talking about causality, we are talking about risk, we have measured odds ratio and relative risk, so we believe that our sentence is correct.

• It appears that logistic regressions were not performed to adjust for patient demographics or comorbidities. If these regressions were performed, the results should be shown. Given that elevated eosinophil counts are more common in subjects who are younger or male, some of the other associations with eosinophil levels may instead be related to gender or age. It is also not clear that current treatments that the subjects were taking were considered, which may similarly skew results if they differ between those with and without elevated eosinophil counts.

Thank you for your comment. As the statistical method used to analyze the continuous variables has been ANCOVA, the therapeutic step has been introduced as a covariate. This bias has been eliminated with this statistical method. We introduce a sentence in methods section to clarify this ("Discrete variables were analyzed using Chi-square test and continuous variables were analyzed using ANCOVA test. Medication steps according to GINA were considered as a covariate in this analysis ").

You are right that sex and age were not entered as covariates. 

We did not correct the data for sex and age because no statistically significant differences were observed in the preliminary analysis of the data. The stratification required for this analysis would have reduced the sample size and therefore the power of the statistical tests used. This reasoning was included in the text in lines 215-217 of our version ("Despite stratified analysis by sex and age, no significant differences were found, so the analysis of eosinophilia should be performed equally regardless of sex and age stratum")

• The eosinophil count means and standard deviations in the tables are suspiciously round numbers. The meaning of the numbers in parentheses for other characteristics, such as FeNO and total IgE, are also unclear. Please check that these numbers are accurate and explained in the table legends.

These number are accurate. It is a real-life study and the laboratory that performs the hemograms at our center provides the data in this manner. FENO and IgE does not have a normal distribution in this study. Therefore the data in parentheses is the standard deviation, as in the case of eosinophilia. The standard deviation has been measured with SPSS, therefore its accuracy should be the same as all the data in this study. We have detected that the journal has sent to reviewers a preliminary version with small changes, in the version that should have reached them the legends in the tables are included.

• OR and RR values should be greater than 1 when referring to a particular feature being favored/more prevalent in the eosinophilic group of subjects.

We totally agree with your statement anyway when the odds and relative risk are less than one this means that it is a protective factor, decreasing the risk. When 1 is included in the confidence interval, the risk is not statistically significant. Consequently, both risks, greater and lower than 1, were only included when their confidence interval did not include 1 and were therefore, statistically significant.

• The concept that elevated peripheral eosinophil counts are associated with particular asthma features, especially severe, exacerbation-prone asthma, is not novel, and eosinophil counts are regularly considered in the asthma endotyping. The manuscript should be clearer about the novelty of and contributions made by these findings.

In our study we have not discussed asthma exacerbations. The novelty of our study is that we describe the characteristics of asthma in relation to its eosinophils. Independently of phenotyping.

• There are a number of grammatical and typographical errors in the manuscript that should be corrected.

We agree with you. We believe that the version that has reached you is a preliminary version that was not the final version sent to the journal. The final version sent to the editor has already been revised by a native English speaker, therefore a substantial grammatical and typographical modification was included.

Author Response File: Author Response.pdf

Reviewer 2 Report

The clinical studies conducted by the authors have demonstrated rhinitis only appears in non-asthmatic patients with elevated blood eosinophil levels, while atopy is elevated in parallel to eosinophilia regardless of whether the patients are asthmatic or not. A decrease in lung function is also observed the higher the blood eosinophil levels, but total IgE is only elevated in patients with high peripheral blood eosinophil levels and asthma. This is very informative to readers in understanding the relationships among asthma, rhinitis, and eosinophilia.

Some comments:

1. Apart from eosinophil counts, has the authors also count basophils?
2. For the patients with more than 300 eosinophils, is there any correlation between eosinophils counts and disease severity?
3. Does any basic science support their clinical findings? The authors may add some content in their discussion.

Author Response

Dear reviewer, thanks for yor comments, we kindly appreciate your contribution.

We answer to your comments below:

1. Apart from eosinophil counts, has the authors also count basophils? Thanks for yoy enriching comment, but unfortunately basophils counts were not gathered for the database
2. For the patients with more than 300 eosinophils, is there any correlation between eosinophils counts and disease severity? Disease severity was not evaluated because only the first visit was considered for the study. Medication needed to achieve control and therefore severity are not available in the database.
3. Does any basic science support their clinical findings? The authors may add some content in their discussion. Thanks for the observation, this is an observational study to define phenotypic characteristics of asthma, so basic science was not applied.

Author Response File: Author Response.pdf