Strain Versus 2D Shear-Wave Elastography Parameters—Which Score Better in Predicting Thyroid Cancer?

Round 1

Reviewer 1 Report

At present, there are many literatures comparing the diagnostic performance of SE and SWE, and there are also literatures comparing the application of Se and swe in thyroid nodules. What are the unique advantages of SE and SWE?We do not think that this article has much clinical significance.

Author Response

Dear reviewer,

Thank you for your time on reviewing our paper. We did some changes, please see attached the revised version of the manuscript. 

Elastography is a much studied topic in the field of thyroid nodules and especially in improving the US diagnostic of thyroid cancer. It is more and more used in everyday practice although clear cut-offs have not been established yet. We do believe our work has clinical relevance and contributes significantly to the current knowledge in the field, given that we compare two elastography techniques employed on high-end equipment. There are more studies comparing elasto techniques, most of them in different populations and different machines. Some technical details have also been added. Please let us know if you have additional comments so that we can make further improvements.

Kind regards,

Andreea Borlea

Reviewer 2 Report

The manuscript "Strain versus 2D-shear-wave elastography parameters – which 2 score better in predicting thyroid cancer?" compare the diagnosis ability of two kinds of Ultrasound elastography, i.e. shear wave based USE and strain based USE of thyroid cancer. The paper has clinical interests but I would like the authors to address the questions below before the consideration of publication.

1) to Understand why there are clinical diagnosis ability differences, the specifications of devices should be analysed. However in the current manuscript, the information is not enough. For example, the resolution of all US systems? The frequency of the shear wave and type? continuous or transient? The stress/force applied in the strain based USE ,etc? Do you follow the same process to all patients? What is the error induced by the operation? Are there repeating scan in the same area? A lot of experimental details are missing in the current version.

2) The potential for each USE are not fully analysed. Authors mentioned in the sample there are different size of nodules and different malignancy of the cancer. What is the potential for different USE to diagnosis different malignancy of the tumour? Now in the paper the analysis has been done only based on YES/No with cancer and I do not think it is innovative/significant enough. As USE has contribution in the diagnosis of cancer is commonly agreed now, how about their special contribution in, e.g. help with small lesions, distinguish the early and late cancers and more? 

3) what is the contribution and diagnosis accuracy of conventional US?

4) There is no results come from the comparison between histology and USE, as mentioned in question 2, can USE provide more information? e.g. localisation of cancerous area and indicate the malignancy?

5) Other general comments on the figures, i.e. low quality, lack of unit (for example figure 7 c and d, etc.

Author Response

Dear Reviewer,

Thank you for your time and for your kind suggestions to improve our paper. We did find your comments truly helpful and relevant. We hope that the with adjustments that we made, the manuscript meets your requirements. Please find attached the modified version of the manuscript and our punctual answers below:

1)  The clinical differences are not given by general characteristics of the US machines, both transducers provide very high resolution images, which depends on the frequency of the transducer that is used - that is similar for the ones used in this study (5-18 MHz). The clinical differences refer more to the elastography techniques. More details on the elasto techniques have been added to the manuscript in material and methods section- thank you for this suggestion. We did follow the same process to all patients, in terms of image acquisition and number of measurements as described in material and methods. There may be errors induced by applying pressure but these are minimized given that an experimented operator performed the measurements. Unfortunately there are no quality parameters for the SWE evaluation using the Aixplorer machine, the only way is to avoid compression artifacts (this was included in the manuscript). For the SE evaluation, we do have a quality assessment, a graph on the display of the machine that shows the degree of compression. - added to the material and methods.

2) Although it would have been interesting to compare the differences between subtypes of thyroid cancer, our group included mostly PTCs which is the most prevalent type of thyroid cancer, so a difference could not be made between PTCs and FTC or MTC in this study. We continue our research and hope to be able to find elastography differences after a longer period of time.

This is discussed in the manuscript - discussions.

Regarding the size of the nodule, the number of nodules <1 cm included in this study was too small (8 nodules). so the few microcarcinomas that were diagnosed were discussed separately as individual cases with their differences in SE and SWE. This could make the research aim of a future study

Our main aim was not to only to reassess that elastography is useful in thyroid evaluation, but to compare two high-end techniques for both SE and 2D-SWE in the same population in order to demonstrate if either has a general advantage. it is thought until now that SE has certain advantage in superficial tissues such as breast and thyroid, but performant SWE, such as the one provided by the Supersonic machine seem to have similar performance with excellent diagnostic quality in predicting thyroid cancer.  

3) The accuracy of conventional US is presented separately (according to EU TIRADS) and combined with the elasto assessment in table 7. It was initially referd to as "2B" in our paper, but it is now referred to as "B-mode"  - suggested by another reviewer. 

4) Thank you for this remark. We added this to the discussions. As thyroid cancers are monoclonal proliferations, finding an area with higher risk of malignancy inside the nodule is not our goal. The goal is to find the lesions with increased stiffness which are entirely with high risk, although it is true that the measurements are made from the region with the highest stiffness (excluding artifacts).

5) units have been added to the figures, thank you for the comment

We hope we did answer your concerns. Please let us know if you still consider we should add technical details or if we should do other adjustments to the current version of the manuscript.

Thank you for your support.

With kind regrads,

Andreea Borlea

Reviewer 3 Report

The authors report a head to head comparison of two elasticity imaging methods for the evaluation of patients with thyroid nodules in order to  diagnose the  malignant from the  benign nodules. It is a very  relevant topic, as thyroid nodules are common. They included 437patients in only 7 months, but only  the  90 patients who finally had surgical resection of their nodule were included in the  analysis. It is not stated clearly, but I suppose the  analysis is performed retrospectively. I have the  following comments:

1. Introduction: The introduction should end up with a clear formulation of the primary and possibly secondary aims. This is lacking, as the statement that "a head to head comparison" is not performed is not an aim. The aim of differentiating malignant from non-malignant thyroid lesions with histology as a reference method must be stated clearly.

2. In the  introduction page 2, line 59-61, you write that "In shear-wave elas- tography (SWE), a dynamic stress is applied by the device, generating perpendicular (point-SWE and 2D-SWE) or parallel (1D-SWE) shear waves." Is this information given to  explain a difference between 2 D SWE and SWE? 1D SWE is not a generally used term. The  methods for numbers and directions which shear waves are measured in the designated area is mostly  at the  discretion of the producers of the equipment, and as what you call 1D SWE is not used in this study, I would suggest to leave this sentence out, as it is more confusing than explanatory. If included, it should belong to the  M&M section, and not the  introduction. 

3. Introduction, page 2, L62: "quantitative appraisal" I do not understand this. If you mean quantitative measurement, please use that instead

4. Introduction, line 68: Reference to WFUMB is spelled erroneously as "WFSUMB" for reference 18. 

5. Material and Method section: 437 patients were included, 215 had biopsies taken, and 90 patients had surgery and had a post surgical pathology report (whole specimen). Why  could you  not use the  pathology report made on the  basis of biopsies, is not that the purpose of taking biopsies in the first place? It would also be clarifying if you made a flow chart of the exclusion process, and this could help  save some text as the M&M section is relatively long, and could be shortened.

6. P. 3, L99-100: "blood was drawn", better use "blood samples were taken", why do you mention that TSH was measured, but not included in the final analyses?

7. Page 3,L129-134: TIRADS description. You made a 3- level categorization of the thyroid nodules based on B-mode US findings, low risk, intermediate risk and high risk. You refer to the  TIRADS, but this represents a 5-level score, and the 3- level quantification you use should be described according to  TIRDAS levels, do  you e.g. put TIRADS 1 and 2 as low risk?, 4 and 5 as high risk? Please rewrite and clarify this. 

8. You use the mean and SD of 5 measurements in the analysis. Previous reports have reported median values and IQR/Median as the measurement of variation. This is because all elastography methods ofte provide one or two outliers that are typically in the high range, rather than in the low, and 5 measurements is i little few datapoints to determine if the data is normally distributed. I would based on this stick to reporting median values for  elastography. If the median values are differing  much  from the mean values, it means that the data are probably not normally distributed after all (even if Kruskalis Willis test finds it normally distributed).

9. P4, L169-170: The  description of the levels of the Asteria score is very limited and each level should have description particularly since it is a visual scoring method. 

10. P. 4, l 170: Do not use "elastic" when you mean soft. Remember that steel is also an elastic material as defined by physics. 

11. Perhaps a sketch of the thyroid anatomy and how you define the  longitudinal section etc would be clarifying for readers who normally  do not examine the thyroid? This is also important since you  have taller-than-wide as a worrisome feature.

12. p6. L233: "pathology report". You should use pathological evaluation of the  surgical specimen, as to me a "pathological report" can also be made on basis of a biopsy or fine needle biopsy.

13. P 6, line 234: PTC is defined as a footnote in table 2, but I did not see it defined at it's first use un the main text.

14. p7, L249: "suspicion features" Please rephrase: Suspect features, worrisome features, you could again refer to TIRADS, as these features are described there. 

15. Table 4, 6 and figure 7 summarizes the most central data that I assume are the main aim of this paper. For Asteria strain score, which is a categorical scoring, and not a longitudinal scale, the  box plot presentation looks more odd than informative, this could be excluded. It is very interesting to  look at the  similarity  between  box plot a (strain ratio) and SWE ratio (also referred to  earlier in the paper as Q-box ratio), they  look very  similar although the  scales are of course different. This is an interesting  finding that could be commented in the discussion as they represent the same index with two different methods. 

16: Please avoid to use 2B when you mean B-mode US, use gray scale or B-mode as these are generally understood terms. Usually the  term E, Elasticity is used with the  unit kPa, and not Elasticity Index, where does the index come from Is it a Supersonic term?

17. Discussion: p14, L 388: You discuss lack of uniformity. In SWE the  amount of pressure is important, in Strain elastography, both pressure, pulsatile pressure, pulsatile pressure frequency (or blood pressure) and the position of the reference area represents "degrees of freedom" that tent do individualize the examination more than the SWE. This could be discussed, and you come back to this in the  discussion (L431-440). 

18. P15, L 395-396, it is conventional to mention sensitivity before specificity when comparison is done, please be consequent on this throughout the paper.

19: You  end up with a cut off between malignant and benign lesions of 30.5 kPa according to the Youden index of the  ROC curve, In the  discussion you find three references with similar results (50,51,52). Perhaps a cut-off rounded to 30kPa would be a useful in clinical practice, if you want to  provide an advice?

19B: Did you apply the  cut-offs scores of SWE and strain ratio to the remaining number of patients in your series who did not undergo surgery (n=347)? If not, why?

20 Figures: 10  figures. 6 are images of scans. They are illustrative, but may  be reduced in numbers, perhaps the  false negative could be omitted. (fig 10). ROC curves: Fig 3, 5 is fine. Figure 8 has too many ROC-curves some ar related from former figures, and it is hard to read, perhaps remove B-mode, B-mode SWE and B-mode SE to a separate panel or just omit the figure. 

21. Tables: 8 tables: Short and easy  to read. Table 7 is a statistical comparison of AUROC AUC where all parameters are compared to  B-mode ROC-AUC, but this is irrelevant. The most relevant ROC-AUC comparison is between EI and Strain ratio, and this could be given in the  text. 

22. References: 55 references. Most seem relevant, but it is a lot for an original paper, 7 references include one of the co-authors, and thus represent self-references. Although this may be difficult to avoid, as some of the authors have contributed with a high number of publications in this field.

Author Response

Dear Reviewer,

Thank you for your time and your generous suggestions

Please find attached the revised version of the manuscript and below our poiny-by-point answers. We do hope that our adjustments to the manuscript do meet your requirements. If not, please let us now how we should further improve it.

1) thank you for the suggestions, we rephrased and tried to make the aims simple and clear.

2) thank you for the remark, we changed 1D with transient elastography, which is usually the term that is used. we found 1D in some references, but it might generate confusion (https://www.eurekaselect.com/images/graphical-abstract/cdt/22/3/big-007.jpg)

We did not delete the phrase, please let us know if we should do so

3) "appraisal" was replaced with "measurement"

4) WFUMB was corrected

5) thank you for your suggestion. fine-needle aspiration (FNA) was initially performed in order to detect high-risk nodules, not biopsy. The FNA(cytology) results are not comparable to biopsies (histology), this is why we included  only the cases with pathology report (post-surgery cases). 

6) thank you for the suggestion. - we deleted this sentence (p3, l 105).

7) we added the explananion in the text (l135-138)

8) thank you for this suggestion. we also believe a reliability index such as the IQR/M should be used in thyroid elastography, as it is in liver for example. However, most studies if not all, do not use this parameter for thyroid. maybe the variability is less important, given its superficial location, but this should be studied. However, we chose to perform 5 measurements because previous studies have proven that no significant differences are found between taking 10 or 5 measurements for thyroid tissue elasto (ex. DOI: 10.4329/wjr.v4.i4.174)

9) we defined all 4 scores

10) thank you for the remark, we changed the term with "soft"

11) we do not consider that a sketch of thyroid anatomy is necessary given that the article should reach readers practicing or learning thyroid elastography, who should first perform well conventional thyroid US. please let us know if you think otherwise. we did add in the text how the transducer is held for the longitudinal view. (l 190)

12) we did this modification so to not generate confusion

13) we defined PTC (line251)

14) we replaced wit "suspect" (l266)

15) thank you for this truly helpful comment. we erased the box plot for the asteria score, which was not relevant. 

16) we did replace 2B in the text (except for figures but it is explained under the figure). As for EI, it is not necessarily specific for the Aixplorer but we did find it in several papers and found it to be more appropriate than just elasticity or elastic modulus. Please let us know if you think otherwise.

17) Thank you for the comment, we did discuss the impact of applying pressure and how we should avoid it, also added some bits in M&M

however, by lack of uniformity we meant that there are no clear guidelines regarding how to perform elasto in the thyroid, how many measurements, how big the ROI, regions to avoid, etc; this is why we discussed lack of uniformity

18) thank you, we changed the order in Se and Sp

19) thank you very much, we added this comment in discussions (l441)

19B we did not apply retrospectively the cut-off to the patients without surgery in the study (for the study it is not useful as we don't have a gold-standard to compare to), at least not to the ones that did not present for another evaluation after, but actually we do use this cut-off currently in our daily practice

20) as we could not make comparisons between types of thyroid cancers and also the discordances were only few between the 2 elasto techniques, we would like to keep the figures as case discussions. Please tell us if you think they should be removed from the manuscript

21) you are right, the most relevant comparison is between SR and mean EI but we think it is important to underline that all parameters are actually useful and have excellent dg quality and compare them to the b-mode alone

22) our team has publications in the field of US and elastography, please let us know if you think any of them is used inappropriately

thank you again for your time and effort in reviewing our paper and improving how we deliver our findings.

we are waiting patiently for your response,

with kind regards,

Andreea Borlea

Round 2

Reviewer 1 Report

There are a few queries：

1.Whether “3ntramodular” on the 131 lines, 133 lines, and 145 lines in the text is the wrong term? Please check carefully and confirm that this is correct.

2.The area under the receiver operating characteristic (AUROC) should be AUC (Area Under Curve).

3. The note in the figure 3 “Max SWE, “Mean SWE” and “SWE ratio” does not agree with the caption.

Author Response

Dear reviewer,

Thank you for your comments. We hope we answered to all your requests. Please see the revised version of the manuscript attached and the point-by-point answers to your review:

1) Thank you for the comment. it was a typing error, we changed to "intranodular"

2) We changed AUROC to AUC in the text and tables

3) We explained the terms in the figure caption. thank you for this suggestion.

Kind regards,

the authors

Reviewer 2 Report

Authors answered the comments and I am happy to recommend the publication. 

Author Response

Thank you for your response and thank you again for your kind suggestions.

The authors