Clinical-Genomic Analysis of 1261 Patients with Ehlers–Danlos Syndrome Outlines an Articulo-Autonomic Gene Network (Entome)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thanks for your very interesting submission. The hypothesis is that the diversity of genetic basis of EDS underpins the variety of clinical presentations. In addition (convenience sampling), the EDS patients are compared with developmental disability - the basis for this comparison is not very clear.

A few notes:

In general, I believe this submission is too long and insufficiently focused. 

It's unclear if the EDS patients recruited by the 3 different methods differ from each other, or if there's any accuracy problems in subtyping. If only 967 EDS patients had genetic profiling, perhaps it's best to exclude EDS patients without genetic profiling; it would make analysis much easier.

 The analysis starts by comparing frequencies of findings between males and females - which is not at all mentioned in the methods. The authors should explain why the males were excluded from analysis - if here's no difference, why not keep the males?

The analysis meanders, and not predicted by the methods section. 

The x-axis for Figure 1 should be in ordinal numbers, or 0.5 increments .

The next few sections describe the frequencies in EDS patients. This is not really in keeping with the hypothesis. These differences in symptom/sign frequencies could be a stand-alone submission, and in my opinion, distract from this submission.

Figure 2 is very confusing. There's no description in the methods about correction of values based on 21-40 y.o females. The rationale should be explained a priori.

Similarly, the V*DU distinctions should be described in the methods. I'm unclear how these were made.

Overall, I think shortening the manuscript and / or breaking into multiple submissions would be best.

Author Response

REVIEWER 1

 

	Yes	Can be improved	Must be improved	Not applicable
Does the introduction provide sufficient background and include all relevant references?	( )	(x)	( )	( )
Are all the cited references relevant to the research?	(x)	( )	( )	( )
Is the research design appropriate?	( )	( )	(x)	( )
Are the methods adequately described?	( )	( )	(x)	( )
Are the results clearly presented?	( )	( )	(x)	( )
Are the conclusions supported by the results?	( )	( )	(x)	( )

Comments and Suggestions for Authors

• The hypothesis is that the diversity of genetic basis of EDS underpins the variety of clinical presentations. In addition (convenience sampling), the EDS patients are compared with developmental disability - the basis for this comparison is not very clear.—clarified in results and methods, the mostly different genes variant in patients with developmental disability taken to show contribution rather than coincidental association with the findings of EDS patients.
• In general, I believe this submission is too long and insufficiently focused. Detailed description of findings (former Table 1) and their statistical analysis (former S1 Table) eliminated, focus now is solely on the gene changes in EDS patients
• It's unclear if the EDS patients recruited by the 3 different methods differ from each other, or if there's any accuracy problems in subtyping. If only 967 EDS patients had genetic profiling, perhaps it's best to exclude EDS patients without genetic profiling; it would make analysis much easier. Description of different EDS patient groups now condensed, discussing only how their minor differences were accounted for in gene group comparisons
• The analysis starts by comparing frequencies of findings between males and females - which is not at all mentioned in the methods. The authors should explain why the males were excluded from analysis - if here's no difference, why not keep the males? Males are not eliminated, their frequencies corrected to those for females when composite gene groups are compared
• The analysis meanders, and not predicted by the methods section. Now focused mostly on gene changes in EDS patients, the findings condensed to a brief S1 Table A and discussed regarding the need for their complete ascertainment to qualify the changes in 317 genes as relevant to EDS
• The x-axis for Figure 1 should be in ordinal numbers, or 0.5 increments . Figure 1 eliminated
• The next few sections describe the frequencies in EDS patients. This is not really in keeping with the hypothesis. These differences in symptom/sign frequencies could be a stand-alone submission, and in my opinion, distract from this submission. Suggestion followed by eliminating Figure 1, S1 Table and including only discussion necessary to gene relevance and clinical-DNA correlations.
• Figure 2 is very confusing. There's no description in the methods about correction of values based on 21-40 y.o females. The rationale should be explained a priori. Now Table 2 and the discussions clarified
• Similarly, the V*DU distinctions should be described in the methods. I'm unclear how these were made. These discussions also focused by describing the qualification protocol at the beginning of Results
• Overall, I think shortening the manuscript and / or breaking into multiple submissions would be best. Suggestion followed.

Reviewer 2 Report

Comments and Suggestions for Authors

The article is very well structured, in my opinion a little long. It's a very clinical article and I don't know if it fits the requirements of the journal.

There are figures in which the letter is very small (Fig.1,2,3)

In the results, check the numbering of the different sections, I think that from 3 it goes to 8.

Author Response

Reviewer 2

 

	Yes	Can be improved	Must be improved	Not applicable
Does the introduction provide sufficient background and include all relevant references?	(x)	( )	( )	( )
Are all the cited references relevant to the research?	(x)	( )	( )	( )
Is the research design appropriate?	(x)	( )	( )	( )
Are the methods adequately described?	(x)	( )	( )	( )
Are the results clearly presented?	(x)	( )	( )	( )
Are the conclusions supported by the results?	(x)	( )	( )	( )

Comments and Suggestions for Authors

• The article is very well structured, in my opinion a little long. It's a very clinical article and I don't know if it fits the requirements of the journal. Shortened article and focused on genomic results that we hope makes it relevant to CIMB
• There are figures in which the letter is very small (Fig.1,2,3) Figure 1 eliminated, Figure 2 a Table, and Figure 3 able to be expanded by readers for better viewing
• In the results, check the numbering of the different sections, I think that from 3 it goes to 8. Thank you!--Corrected

Reviewer 3 Report

Comments and Suggestions for Authors

This is a very serious effort that obviously represents many years of clinical observation, extensive genomic testing and thorough study. For a large population of clinicians and scientists, this article will be of both use and interest.

However, at this point, it is too long and too complicated to make much sense. I think you have combined at least two articles in one. The first would be your clinical observations about EDS symptoms compared to other neuro-skeletal-dermatological syndromes. Your classification could create clarity where you observe there is currently a degree of confusion--especially when the clinical symptoms are aligned with the genomic information about what genes are in play and how they are expressed. The second article is clearly your analysis of the genomic data your lab has organized for you.

I see that you have invested significantly in this study since I saw no outside funding source and the exome testing could not have been cheap. It deserves more focus. 

Also, you have stuck the most important tables in the Supplementary Information. You constantly refer to Table S1 and S4. So often that they need to be in the primary text. Splitting the article into two would help you include the tables in the text.

You have also organized your data and done your analysis in Excel. I think you would be better served to have moved the analysis of the data to an established statistical program such as Stata, GraphPad Prism, SPSS or a statistical language such as R or Python. Excel is so prone to errors that I limit its use in my practice to preparing table for importing to R. (I am a biostatistician/virologist). Doing so would also create opportunities for more in-depth statistical analysis, one of which I will refer to below. Including an author who is a statistical specialist will also help the sophistication of your statistical results mirror the very sophisticated research design.

MATERIALS & METHODS: line 74: when you refer to your prior report or paper, it would be a good idea to give the general idea of what it covers and how it prepares the ground for this paper.

lines 122 -3: you should specify which online resource you used. In some cases, they have known biases and we should have to wait for the references to see which resources they are.

RESULTS: line 142: I think you meant to say "profile is" rather than "profile that is".

lines 146 - 7: how do joint hypermobility and and skin elasticity (1) confer advantages and (2) likely explain abundance of females? Not clear to me.

Figure 1A: As in many cases, your "figures" are really tabular and should be labeled "Table".

Figure 1B: needs a new version. I can't follow the x-axis--the values are too close together. I don't understand the y-axis--the label doesn't seem to speak to numeric values. If I understand it correctly, it is a scatterplot, but you have the lines connected suggesting a trend of y varying according to the values of the x-axis, but I don't see that suggested in either the Caption or the text.

line 245: why do different referral sources matter with regard to profiles of female patients? It seems that it's more a commentary on the skills and specialities of the medical professionals involved and if so why?

line 275: why does "date seen" (3rd data column) suggest awareness of dysautonomia? It could be, I guess, but you need to give us an explanation for this observation.

line 295 - 6: you speak of many subjective observational variables, some of which are overlapping. This cries out for the application of a dimension reduction strategy such as Principal Components or Factor Analysis to find the most orthogonal (distinct) dimensions to measure the genomic differences.

line 319: measurement imprecision should be covered by variance within the sample or, if it based on some consistent error or misapprehension by the physician, you should be speaking about it.

line 374: I think you meant EDS patient groups, not DS.

Section 3.7: EDS genomics needing clinical qualifications of DNA variants. Are you proposing then a new subtyping system for EDS based on genomic variants rather than clinical symptoms? Possibly a good idea. What you're suggesting here is you have to test the association of genomic variants to sets of symptoms. You speak a lot of this, but we need a clearer sub typing system specified.

lines 476 - 484: The details of this paragraph are interesting and useful, but I don't see the point of including it. Why are you speaking to this issue?

Table 1: bring this way forward so we readers have a chance to appreciate who your panel of patients is. Page 12 is way late.

lines 539 - 554: you are in essence suggesting a statistical power analysis, relating sample size with the level of difference in levels of variables and other factors. You should simply do it. Then the question of how large your sample should be given subgroup sizes, etc., will be clear.

Gene Network Specification: it seems this is the guts of the paper, where you get to after all your prior analysis. It needs to be rewritten to first show the network (Figure 6) and then break it down. This way, the reader can always refer back to the diagram as a reference. 

line 1057: diagnostic delays average 14 years? what could this mean? Does this mean that patients go to a doctor for 14 years before they say EDS? How counterintuitive is that? This needs more exposition.

In conclusion, you've done an enormous amount of work that deserves to be published. I think your article will have a very positive impact on your field. But, it needs serious tightening and restructuring to make the points clearer and perhaps division into two or more articles.

Comments on the Quality of English Language

Your English is grammatically fine. My only concern is the over-reliance on technical medical words that only scientists and clinicians in your sub-specialty will understand. In many cases, simple English words would have sufficed. A careful rereading and simplification of the sentences will do well to make the article clearer and more interesting.

Let me now take the article in the order of the text.

ABSTRACT: Your language is so summary that I struggled to understand it. I think if you try to make fewer points in the abstract, you can abstract the article more clearly and attractively (so as to make the reader want to read the rest).

INTRODUCTION: You dive directly into the details of sub-types of EDS, many of which are overlapping. I think one or two introductory paragraphs giving an overview of EDS will enable readers to join you in the deep dive linking symptoms with genomic information. 

Line 66: I'm not sure relation to COVID-19 is really relevant to your article or your overall point. It strikes me as noise. Long COVID clearly has joint-related symptoms, but your interest (deep in the Results section) should be to differentiate these symptoms from those related to EDS.

 

Author Response

Reviewer 3

 

	·         Yes	·         Can be improved	·         Must be improved	·         Not applicable
·         Does the introduction provide sufficient background and include all relevant references?	·         (x)	·         ( )	·         ( )	·         ( )
·         Are all the cited references relevant to the research?	·         (x)	·         ( )	·         ( )	·         ( )
·         Is the research design appropriate?	·         ( )	·         (x)	·         ( )	·         ( )
·         Are the methods adequately described?	·         ( )	·         ( )	·         (x)	·         ( )
·         Are the results clearly presented?	·         ( )	·         ( )	·         (x)	·         ( )
·         Are the conclusions supported by the results?	·         (x)	·         ( )	·         ( )	·         ( )

Comments and Suggestions for Authors  (responses in red)

• This is a very serious effort that obviously represents many years of clinical observation, extensive genomic testing and thorough study. For a large population of clinicians and scientists, this article will be of both use and interest.---this is a remarkably thorough review which we should all try to emulate and we have tried to address all of the very appropriate criticisms. The clinical author (GNW) has spent the $145 on the academic version of the recommended Stata software and will use it for factor analysis of the clinical findings in a separate article as recommended
• However, at this point, it is too long and too complicated to make much sense. I think you have combined at least two articles in one. The first would be your clinical observations about EDS symptoms compared to other neuro-skeletal-dermatological syndromes. Your classification could create clarity where you observe there is currently a degree of confusion--especially when the clinical symptoms are aligned with the genomic information about what genes are in play and how they are expressed. The second article is clearly your analysis of the genomic data your lab has organized for you. We have condensed discussions to those relevant to the DNA findings that are now the subject of this shortened and refocused article
• I see that you have invested significantly in this study since I saw no outside funding source and the exome testing could not have been cheap. It deserves more focus. Now focused on the molecular data, obtained by conventional medical testing funded by patient insurance after ensuring coverage amounts.
• Also, you have stuck the most important tables in the Supplementary Information. You constantly refer to Table S1 and S4. So often that they need to be in the primary text. Splitting the article into two would help you include the tables in the text. Have split the article and eliminated Table S1 but the many details in S2-S4, Sheet 5 would occupy a huge amount of article text and be distracting to the main data and discussion points.
• You have also organized your data and done your analysis in Excel. I think you would be better served to have moved the analysis of the data to an established statistical program such as Stata, GraphPad Prism, SPSS or a statistical language such as R or Python. Excel is so prone to errors that I limit its use in my practice to preparing table for importing to R. (I am a biostatistician/virologist). Doing so would also create opportunities for more in-depth statistical analysis, one of which I will refer to below. Including an author who is a statistical specialist will also help the sophistication of your statistical results mirror the very sophisticated research design. I (GNW) purchased Stata but eliminated the analysis of finding variability that it might be more useful for. I am beginning the considerable learning curve, so have checked some of the simple SD by Excel and t test calculations of MedCalc using actual formulae and find them reasonably accurate.
• MATERIALS & METHODS: line 74: when you refer to your prior report or paper, it would be a good idea to give the general idea of what it covers and how it prepares the ground for this paper. We did so.
• lines 122 -3: you should specify which online resource you used. In some cases, they have known biases and we should have to wait for the references to see which resources they are. referenced the MedCalc website
• RESULTS: line 142: I think you meant to say "profile is" rather than "profile that is". Yes, corrected
• lines 146 - 7: how do joint hypermobility and and skin elasticity (1) confer advantages and (2) likely explain abundance of females? Not clear to me. eliminated
• Figure 1A: As in many cases, your "figures" are really tabular and should be labeled "Table". Agree and changed for Tables 1-5
• Figure 1B: needs a new version. I can't follow the x-axis--the values are too close together. I don't understand the y-axis--the label doesn't seem to speak to numeric values. If I understand it correctly, it is a scatterplot, but you have the lines connected suggesting a trend of y varying according to the values of the x-axis, but I don't see that suggested in either the Caption or the text. Eliminated
• line 245: why do different referral sources matter with regard to profiles of female patients? It seems that it's more a commentary on the skills and specialities of the medical professionals involved and if so why? Eliminated
• line 275: why does "date seen" (3rd data column) suggest awareness of dysautonomia? It could be, I guess, but you need to give us an explanation for this observation. Eliminated
• line 295 - 6: you speak of many subjective observational variables, some of which are overlapping. This cries out for the application of a dimension reduction strategy such as Principal Components or Factor Analysis to find the most orthogonal (distinct) dimensions to measure the genomic differences. Will attempt this in what will be a separate paper as recommended.
• line 319: measurement imprecision should be covered by variance within the sample or, if it based on some consistent error or misapprehension by the physician, you should be speaking about it. Eliminated
• line 374: I think you meant EDS patient groups, not DS. Yes, corrected
• Section 3.7: EDS genomics needing clinical qualifications of DNA variants. Are you proposing then a new subtyping system for EDS based on genomic variants rather than clinical symptoms? Possibly a good idea. What you're suggesting here is you have to test the association of genomic variants to sets of symptoms. You speak a lot of this, but we need a clearer sub typing system specified. Are advocating the opposite by saying most EDS patients have a similar clinical profile as a consequence of multiple genes operating in network fashion—only rare gene changes like some of those in COL3 will act as nodes in the network and produce significantly different (for management) finding profiles like that of the vascular EDS type (reference 7). Hope this view is clarified in the refocused article.
• lines 476 - 484: The details of this paragraph are interesting and useful, but I don't see the point of including it. Why are you speaking to this issue? Rephrased and focused
• Table 1: bring this way forward so we readers have a chance to appreciate who your panel of patients is. Page 12 is way late. Suggestion followed—Table 1 leads off completely rearranged Results
• lines 539 - 554: you are in essence suggesting a statistical power analysis, relating sample size with the level of difference in levels of variables and other factors. You should simply do it. Then the question of how large your sample should be given subgroup sizes, etc., will be clear. Will attempt in separate article
• Gene Network Specification: it seems this is the guts of the paper, where you get to after all your prior analysis. It needs to be rewritten to first show the network (Figure 6) and then break it down. This way, the reader can always refer back to the diagram as a reference. We feel that the list of genes and mechanisms would be incomprehensible without the prior data and that the (now Figure 2) provides a nice summary of the main conclusion.
• line 1057: diagnostic delays average 14 years? what could this mean? Does this mean that patients go to a doctor for 14 years before they say EDS? How counterintuitive is that? This needs more exposition. That is exactly what it means for most patients, but that subject is mostly eliminated
• In conclusion, you've done an enormous amount of work that deserves to be published. I think your article will have a very positive impact on your field. But, it needs serious tightening and restructuring to make the points clearer and perhaps division into two or more articles. Have definitely restructured and greatly condensed the clinical data into a brief S1 Table A.

Comments on the Quality of English Language

• Your English is grammatically fine. My only concern is the over-reliance on technical medical words that only scientists and clinicians in your sub-specialty will understand. In many cases, simple English words would have sufficed. A careful rereading and simplification of the sentences will do well to make the article clearer and more interesting.
• Let me now take the article in the order of the text.
• ABSTRACT: Your language is so summary that I struggled to understand it. I think if you try to make fewer points in the abstract, you can abstract the article more clearly and attractively (so as to make the reader want to read the rest). Rewritten
• INTRODUCTION: You dive directly into the details of sub-types of EDS, many of which are overlapping. I think one or two introductory paragraphs giving an overview of EDS will enable readers to join you in the deep dive linking symptoms with genomic information. Completely rewritten
• Line 66: I'm not sure relation to COVID-19 is really relevant to your article or your overall point. It strikes me as noise. Long COVID clearly has joint-related symptoms, but your interest (deep in the Results section) should be to differentiate these symptoms from those related to EDS. Refer only to the prior article (now ref 14) regarding similar Methods and similar S2 Table that is now reorganized

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

This paper is considerably clearer than in the first submission. It is still VERY long and should be shorted some. As I have previously said, this paper combines a number of themes that could be expressed in separate papers.

However, it is a major improvement.

Comments on the Quality of English Language

Much better. Sentence structure could still be simplified more.

Author Response

In response to reviewer 3's request to shorten the manuscript further, we have eliminated section 3.10 that examined many of the genes that were variant in only a few EDS patients, deleting 5 manuscript pages to go from 28 to 23. This leaves section 3.9 that covers genes variant in 6 or more EDS patients, these these patient groups sufficiently large (most with 15-49 patients) to compare their EDS-dysautonomia finding profiles. These are the groups outlined in the main Tables 2-3 of the paper and justify the conclusion of gene networks underlying EDS.