Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The scientific article under consideration is devoted to one of the current topics of modern abdominal radiology. namely the possibility of transitioning from multiparametric MRI (mpMRI) to biparametric MRI (bpMRI) without the use of contrast. It is known that the use of gadolinium-based contrast agents has contraindications (e.g., renal failure) and carries the risk of metal accumulation in the body. Concentration of contrast agents during MRI balances the increase in sensitivity or image quality with the potential for harm to the patient. Confirming that the non-contrast protocol achieves an accuracy of up to 88.6% (sensitivity) is crucial for patients with limitations. The Authors emphasize the critical role of a near-subjective criterion in the form of radiologist competence (agreement 73.9% for an expert versus 56.5% for a novice), which warns clinics against hastily abandoning contrast without proper staff training.

1. To what extent can it be justified to use the conclusion of the initial MRI report with contrast as a reference standard, rather than the results of the pathomorphological examination after resection or cystectomy?
2. General question: Given the high rate of overstaging after resection, what specific biparametric protocols can help the clinician differentiate post-manipulation changes from true invasion without perfusion data?

 

Author Response

 

1. The scientific article under consideration is devoted to one of the current topics of modern abdominal radiology. namely the possibility of transitioning from multiparametric MRI (mpMRI) to biparametric MRI (bpMRI) without the use of contrast. It is known that the use of gadolinium-based contrast agents has contraindications (e.g., renal failure) and carries the risk of metal accumulation in the body. Concentration of contrast agents during MRI balances the increase in sensitivity or image quality with the potential for harm to the patient. Confirming that the non-contrast protocol achieves an accuracy of up to 88.6% (sensitivity) is crucial for patients with limitations. The Authors emphasize the critical role of a near-subjective criterion in the form of radiologist competence (agreement 73.9% for an expert versus 56.5% for a novice), which warns clinics against hastily abandoning contrast without proper staff training.

We thank the Reviewer for the positive and thoughtful assessment of our manuscript. We appreciate the recognition of the clinical relevance of evaluating a contrast-free bpMRI approach in bladder cancer imaging, particularly for patients with contraindications to gadolinium-based contrast agents. We also thank the Reviewer for highlighting the importance of radiologist experience in bpMRI interpretation, which is one of the key practical messages of our study. We agree that our findings support a cautious and selective adoption of bpMRI, especially when reader expertise and clinical context are carefully considered.

2. To what extent can it be justified to use the conclusion of the initial MRI report with contrast as a reference standard, rather than the results of the pathomorphological examination after resection or cystectomy?

We thank the Reviewer for this important comment. We agree that histopathology is the gold standard for determining muscle invasion. In our study, the reference standard based on the original contrast-enhanced mpMRI report was used specifically for the reproducibility objective, namely to assess whether bpMRI could reproduce routine mpMRI-based VI-RADS categorization in clinical practice. Histopathology, on the other hand, represents the biological reference for the binary endpoint of muscle-invasive versus non-muscle-invasive disease. We have revised the manuscript to clarify this distinction and to better explain that imaging-based agreement analyses and pathological correlation address two related but different study objectives.

1. General question: Given the high rate of overstaging after resection, what specific biparametric protocols can help the clinician differentiate post-manipulation changes from true invasion without perfusion data?

 

We thank the Reviewer for this important practical question. We agree that the post-resection setting is the most challenging scenario for a contrast-free approach. At present, no dedicated biparametric protocol has been fully validated to replace perfusion imaging in post-TURBT assessment. However, available evidence suggests that overstaging can be reduced by combining strict acquisition quality and a DWI-dominant reading strategy. In particular, MRI should ideally be performed before TURBT or at least 2 weeks after TURBT/intravesical treatment, with adequate bladder distension and antiperistaltic preparation. In a biparametric setting, careful evaluation of high-b-value DWI together with ADC maps is essential, because postoperative edema may produce T2 shine-through whereas fibrosis and benign post-manipulation changes usually do not show marked true diffusion restriction. We have revised the Discussion to clarify that these measures may improve contrast-free assessment, but that equivocal and post-TURBT cases should still preferentially undergo standard contrast-enhanced mpMRI rather than bpMRI-only evaluation.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors aimed to test relying on a single center experience the differences in bpMRI and mpMRI to evaluate VI-RADS. The topic is novel and warrant a consideration.

The authors included 65 patients (69 lesions) underwent bladder multiparametric MRI. Two blinded radiologists assigned VI-RADS scores using only T2-weighted and diffusion-weighted imaging (biparametric MRI), while the reference was the mpMRI VI-RADS. 

They found briefly than the AUC for VI-RADS ≥4 was 0.87 (0.78–0.95) for the expert and 0.81 (0.69–0.91) for the non-expert. Sensitivity at a biparametric threshold of VI-RADS ≥4 was 88.6% for both readers; specificity was 85.3% vs 73.5%. 

Despite the interest in the field, several observations should be pointed out.

First how the confirmation bias was managed (if was managed). The flow used in PI-RADS by PRIME was randomization, then assessing bpMRI and then mpMRI. Here, the flow was all the way around. First mpMRI, then bpMRI, with the huge selection bias of readers, as well as the absence of randomization. All these aspects should be discussed properly due to their inherently importance. 

Second, the setting of MRI should be better defined. Was it a post-resection setting, or patients were naïve? All these aspects may lead to an artificial VI-RADS evaluation. Furthermore, homogeneity is required. We don't know if mpMRI may be usefull in some clinical scenario rather than others. It's the moment to prove it. Nonetheless, the main concern of VI-RADS was still not addressed. The unanswered question indeed is in which clinical setting VI-RADS should be used. This question was totally ignored by the current authors. 

 

Author Response

 

1. The authors aimed to test relying on a single center experience the differences in bpMRI and mpMRI to evaluate VI-RADS. The topic is novel and warrant a consideration.
   The authors included 65 patients (69 lesions) underwent bladder multiparametric MRI. Two blinded radiologists assigned VI-RADS scores using only T2-weighted and diffusion-weighted imaging (biparametric MRI), while the reference was the mpMRI VI-RADS. They found briefly than the AUC for VI-RADS ≥4 was 0.87 (0.78–0.95) for the expert and 0.81 (0.69–0.91) for the non-expert. Sensitivity at a biparametric threshold of VI-RADS ≥4 was 88.6% for both readers; specificity was 85.3% vs 73.5%. Despite the interest in the field, several observations should be pointed out.

 

We thank the Reviewer for the positive appraisal of our work and for recognizing the novelty and clinical relevance of the topic. We also appreciate the concise summary of our study design and main findings. We have carefully considered the following observations and addressed each point in detail below.

2. First how the confirmation bias was managed (if was managed). The flow used in PI-RADS by PRIME was randomization, then assessing bpMRI and then mpMRI. Here, the flow was all the way around. First mpMRI, then bpMRI, with the huge selection bias of readers, as well as the absence of randomization. All these aspects should be discussed properly due to their inherently importance. 

We thank the Reviewer for this important methodological comment. We agree that potential confirmation bias and reading-order effects should be explicitly addressed, particularly when comparing our retrospective design with prospective randomized paradigms such as those used in prostate MRI studies. In our study, however, several measures were adopted to reduce this risk. First, the two bpMRI readers were not the radiologist who had originally reported the full mpMRI examinations used as the reference standard. Second, bpMRI datasets were reviewed in randomized and anonymized fashion, and both readers were blinded to clinical information, histopathology, and the original mpMRI assessments. Therefore, the bpMRI interpretation was independent from the original routine mpMRI reporting process.

At the same time, we agree that our study did not use a fully prospective paired-reading design in which the same cases are interpreted in a randomized sequence under both bpMRI and mpMRI conditions. This should be considered a limitation of the study and differs from more controlled multireader paradigms. We have revised the manuscript to clarify how confirmation bias was mitigated and to explicitly discuss the residual limitations related to the retrospective single-center design. Finally, the inclusion of one expert and one non-expert reader was intentional and consistent with the study aim of evaluating the impact of radiologist experience, rather than representing an unintended source of selection bias.

3. Second, the setting of MRI should be better defined. Was it a post-resection setting, or patients were naïve? All these aspects may lead to an artificial VI-RADS evaluation. Furthermore, homogeneity is required. We don't know if mpMRI may be usefull in some clinical scenario rather than others. It's the moment to prove it. Nonetheless, the main concern of VI-RADS was still not addressed. The unanswered question indeed is in which clinical setting VI-RADS should be used. This question was totally ignored by the current authors.

We thank the Reviewer for this important comment. We agree that the clinical setting of MRI acquisition must be more clearly defined, because VI-RADS performance is influenced by whether imaging is performed before or after transurethral resection and by the broader clinical context. In our cohort, most patients underwent MRI in the pre-treatment setting, whereas a minority had undergone TURBT before MRI; we have now clarified this point more explicitly in the Methods and Results. We also agree that post-resection MRI may lead to more challenging and potentially artificial VI-RADS interpretation because postoperative inflammation, edema, fibrosis, and blood products can mimic residual tumor invasion. This is consistent with the original VI-RADS concept and with more recent reviews indicating that VI-RADS is primarily intended for local staging before TURBT, whereas post-TURBT assessment is a more problematic and less standardized scenario.

Regarding the broader question of in which clinical setting VI-RADS should be used, we agree that this deserves clearer discussion. Our study was not designed to define all possible indications for VI-RADS, but rather to compare bpMRI and mpMRI within a real-world cohort. We have therefore revised the Discussion to better position our findings within the current clinical framework: VI-RADS is most established for preoperative local staging and risk stratification of suspected or known bladder cancer before TURBT; its use in post-TURBT or other post-treatment settings is less robust and should be interpreted with caution; and newer applications, such as response assessment with nacVI-RADS after systemic therapy, should be considered distinct extensions rather than interchangeable with the original primary-staging scenario.

Reviewer 3 Report

Comments and Suggestions for Authors

I have reviewed the manuscript by Urraro et al  on bpMRI vs mpMRI for VI-RADS, with focus on reader experience in a single-center cohort. The topic is clinically practical (contrast contraindication, workflow time), and the manuscript is generally readable. However, the current study design uses mpMRI VI-RADS from the original report as the “reference standard” (not histopathology), and this creates a circular benchmark that limits the strength of the conclusions. Also, several reporting gaps (MRI parameters, lesion-level correlation, subgroup precision) reduce interpretability.

Major comments

-Authors  compare bpMRI-derived VI-RADS against the “routine mpMRI report VI-RADS” as ground truth. This is not diagnostic accuracy; it is agreement with a prior opinion (and possibly from the same institutional reading culture). Please be explicit in the title/aim/conclusions that this is reproducibility vs mpMRI reporting, not “accuracy” for muscle invasion.

Direct author query: Do you have histopathology (TURBT/cystectomy) to report sensitivity/specificity for MIBC using pathology as reference, at least in a subset?

-Authors conclude conclude bpMRI “may approximate” mpMRI in selected low-risk cases, but your cohort has a high fraction of VI-RADS 4–5 (≈50%). This is not “low-risk enriched”. Please define a priori what “selected” means (VI-RADS 2–3 only? non-postTURBT? lesion size? no hydronephrosis?). Right now it reads post hoc. What exact clinical rule do you propose for selecting patients for bpMRI-only pathway?

-Authors state non-independence when multiple lesions per patient and say analyses are “primarily descriptive”, but you still report ROC/AUC, kappa, CI bootstraps, etc. This is not purely descriptive. If multiple lesions exist in same patient, your CIs may be optimistic.

Direct author query: How many patients had >1 lesion? Can you re-run key metrics with patient-level approach (worst lesion per patient) as sensitivity analysis?

-In methods state that technical parameters “should be reported” (b-values, slice thickness, etc.). This must be fully specified in the main text or Supplementary, otherwise reproducibility is weak. Please provide b-values, DWI sequence type, slice thickness, in-plane resolution, DCE temporal resolution, and contrast dose

-only 9 post-TURBT cases, yet you emphasize that discrepancies increase there. The conclusion is plausible, but statistical uncertainty is large. Present exact counts of overstage/understage in this subgroup and avoid strong wording.

Direct author query: Provide a mini-table for post-TURBT: reference VI-RADS vs bpMRI VI-RADS for each reader (even anonymized case-level counts).

-high sensitivity is reported for both readers at ≥4, but specificity drops for non-expert due to overstaging. Discuss clinical consequences: overstaging may push to aggressive workup or change surgical planning. Quantify expected false positives per 100 patients in your sample.

-There are visible typos and formatting artifacts: “Priot TURBT”, mixed decimal separators (0,74 vs 0.74), and “Figure .” without number. Also some fig captions include confusing wording (Fig 1 caption mentions DCE in the middle with grammar breaks).

A one page flow: cohort selection + number of lesions + number of post-TURBT + distribution of reference VI-RADS (simple bar).

A stratified plot of disagreement rate by scenario: low-risk (2–3) vs high-risk (4–5), and TURBT vs non-TURBT could improve

Minor

Standardize terms: bpMRI vs mpMRI, VI-RADS capitalization, “non-expert” definition (years of practice? exposure to VI-RADS training)..

Table 1 is extremely minimal; add comorbidity relevant to contrast contraindication (eGFR categories) if available, since this is part of rationale.

Clarify whether antiperistaltic agent was uniform for all, and which agent (hyoscine, gluca

Author Response

 

1. I have reviewed the manuscript by Urraro et al  on bpMRI vs mpMRI for VI-RADS, with focus on reader experience in a single-center cohort. The topic is clinically practical (contrast contraindication, workflow time), and the manuscript is generally readable. However, the current study design uses mpMRI VI-RADS from the original report as the “reference standard” (not histopathology), and this creates a circular benchmark that limits the strength of the conclusions. Also, several reporting gaps (MRI parameters, lesion-level correlation, subgroup precision) reduce interpretability.

We thank the Reviewer for the careful evaluation of our manuscript and for recognizing the practical clinical relevance of the topic. We also appreciate the Reviewer’s constructive comments regarding the use of the original mpMRI VI-RADS report as the reference standard and the need for more complete reporting of technical and analytical details. We have carefully addressed these issues throughout the revised manuscript by clarifying the study aims and reference standards, expanding the description of MRI acquisition and image interpretation, and better discussing the limitations related to lesion-level analysis and subgroup precision. Each point is addressed in detail below.

2. Authors  compare bpMRI-derived VI-RADS against the “routine mpMRI report VI-RADS” as ground truth. This is not diagnostic accuracy; it is agreement with a prior opinion (and possibly from the same institutional reading culture). Please be explicit in the title/aim/conclusions that this is reproducibility vs mpMRI reporting, not “accuracy” for muscle invasion.Direct author query: Do you have histopathology (TURBT/cystectomy) to report sensitivity/specificity

We thank the Reviewer for this important methodological comment. We agree that comparing bpMRI-derived VI-RADS with the original routine mpMRI VI-RADS report primarily evaluates reproducibility of imaging-based risk stratification rather than absolute diagnostic accuracy for muscle invasion. Accordingly, we have revised the manuscript to make this distinction more explicit in the title, study aim, Methods, Discussion, and Conclusions. In particular, we now clarify that the main objective was to assess whether a contrast-free bpMRI protocol could reproduce routine mpMRI-based VI-RADS categorization used in clinical practice, and that this agreement-based analysis should not be interpreted as a pathology-based validation of muscle invasion staging.

Regarding histopathology, we agree that pathology is the biological gold standard for differentiating muscle-invasive from non-muscle-invasive bladder cancer. However, the primary design of the present study was focused on reproducibility versus routine mpMRI reporting rather than on a dedicated pathology-based diagnostic accuracy analysis. We have therefore revised the wording throughout the manuscript to avoid overstating the term “accuracy” where the analysis is based on agreement with mpMRI reporting rather than histopathological confirmation. This limitation is now discussed more explicitly.

3. Authors conclude conclude bpMRI “may approximate” mpMRI in selected low-risk cases, but your cohort has a high fraction of VI-RADS 4–5 (≈50%). This is not “low-risk enriched”. Please define a priori what “selected” means (VI-RADS 2–3 only? non-postTURBT? lesion size? no hydronephrosis?). Right now it reads post hoc. What exact clinical rule do you propose for selecting patients for bpMRI-only pathway?

We thank the Reviewer for this important observation. We agree that the expression “selected low-risk cases” was too vague and could be interpreted as a post hoc statement, particularly because our cohort was enriched in intermediate- and high-risk lesions. We have therefore revised the manuscript to define more explicitly what we mean by a potentially suitable bpMRI-only scenario.

Based on our data, we do not propose bpMRI as a general alternative to mpMRI, nor do we claim to have derived a formally validated selection algorithm. Rather, our findings support a cautious and restricted use of bpMRI only in treatment-naïve, pre-TURBT patients with adequate image quality and clearly low-risk, non-equivocal imaging features on T2WI and DWI/ADC. In practical terms, equivocal lesions, suspected higher-risk lesions, post-TURBT examinations, and technically suboptimal studies should still undergo standard contrast-enhanced mpMRI. We have revised the Abstract, Discussion, and Conclusions accordingly to make this point more explicit and less post hoc.

4. Authors state non-independence when multiple lesions per patient and say analyses are “primarily descriptive”, but you still report ROC/AUC, kappa, CI bootstraps, etc. This is not purely descriptive. If multiple lesions exist in same patient, your CIs may be optimistic. Direct author query: How many patients had >1 lesion? Can you re-run key metrics with patient-level approach (worst lesion per patient) as sensitivity analysis?

We thank the Reviewer for this important methodological observation. We agree that the analyses are not purely descriptive, since we reported agreement statistics, ROC analysis, and bootstrap-based confidence intervals. We have therefore revised the manuscript to describe these analyses more appropriately as exploratory rather than purely descriptive.

Regarding lesion-level non-independence, 61 patients had a single lesion and 4 patients had two lesions, for a total of 69 lesions in 65 patients. Thus, only a small minority of patients contributed more than one lesion. Although each lesion was evaluated separately and lesion staging was assigned independently, we agree that some within-patient correlation cannot be completely excluded because lesions from the same patient share the same examination context. Given the very limited degree of clustering in our cohort, we believe that its impact on the main results is likely modest, but confidence intervals may still be somewhat optimistic. We have clarified this point more explicitly in the revised manuscript and now describe the analyses as exploratory.

We acknowledge that a patient-level sensitivity analysis based on the worst lesion per patient would be of interest. However, the present study was designed primarily as a lesion-based exploratory comparison of bpMRI and mpMRI VI-RADS assessment, and this additional analysis was not included in the current version. We have added this point as a methodological consideration and as a possible direction for future studies.

5. In methods state that technical parameters “should be reported” (b-values, slice thickness, etc.). This must be fully specified in the main text or Supplementary, otherwise reproducibility is weak. Please provide b-values, DWI sequence type, slice thickness, in-plane resolution, DCE temporal resolution, and contrast dose

We thank the Reviewer for this important comment. We agree that MRI acquisition details must be reported clearly to ensure reproducibility. In our institution, the bladder MRI protocol was performed in accordance with the technical recommendations of the VI-RADS acquisition framework. We have therefore revised the Methods section to explicitly state that the protocol was VI-RADS-compliant and to report the key recommended technical elements, including field strength, T2-weighted slice thickness, DWI sequence orientation and b-values, and DCE acquisition and contrast administration parameters.

6. only 9 post-TURBT cases, yet you emphasize that discrepancies increase there. The conclusion is plausible, but statistical uncertainty is large. Present exact counts of overstage/understage in this subgroup and avoid strong wording. Direct author query: Provide a mini-table for post-TURBT: reference VI-RADS vs bpMRI VI-RADS for each reader (even anonymized case-level counts).

We thank the Reviewer for this important comment. We agree that the post-TURBT subgroup is small and that the corresponding findings should be interpreted cautiously. In the revised manuscript, we have therefore added the exact numbers of concordant cases, overstaging, and understaging in the post-TURBT subgroup for each reader, and we have softened the wording throughout the Results, Discussion, and Conclusions to reflect the descriptive nature and statistical uncertainty of this subgroup analysis. In the revised manuscript, we now report the exact distribution of concordant, overstaged, and understaged cases in the post-TURBT subgroup for each reader, and we provide a summary mini-table to improve transparency. We also softened the wording throughout the manuscript to reflect the descriptive nature and limited precision of this subgroup analysis.

7. high sensitivity is reported for both readers at ≥4, but specificity drops for non-expert due to overstaging. Discuss clinical consequences: overstaging may push to aggressive workup or change surgical planning. Quantify expected false positives per 100 patients in your sample.

We thank the Reviewer for this important comment. We agree that the reduced specificity observed for the non-expert reader has relevant clinical implications, because bpMRI overstaging may increase the likelihood of unnecessary aggressive workup, stronger suspicion of muscle-invasive disease, and potentially more invasive treatment planning discussions. In the revised manuscript, we have expanded the Discussion to better address these consequences. In addition, using the prevalence of mpMRI-negative cases in our cohort and the observed specificities, we estimated the expected number of false-positive classifications per 100 similar cases. This corresponded to approximately 7 false positives per 100 cases for the expert reader and 13 false positives per 100 cases for the non-expert reader, highlighting the practical impact of reduced specificity in less experienced readers.

8. There are visible typos and formatting artifacts: “Priot TURBT”, mixed decimal separators (0,74 vs 0.74), and “Figure .” without number. Also some fig captions include confusing wording (Fig 1 caption mentions DCE in the middle with grammar breaks).

We thank the Reviewer for this careful editorial observation. We agree that the manuscript contained several typographical errors and formatting inconsistencies, including spelling mistakes, mixed decimal separators, incomplete figure numbering, and some unclear wording in figure captions. In the revised version, we have carefully proofread the manuscript, corrected typographical and formatting artifacts, standardized decimal notation throughout the text and tables, corrected figure numbering, and revised the figure captions for clarity and grammatical consistency.

9. A one page flow: cohort selection + number of lesions + number of post-TURBT + distribution of reference VI-RADS (simple bar).

We thank the Reviewer for this helpful suggestion. In the revised manuscript, we have added a one-page overview figure summarizing the study cohort and the distribution of reference mpMRI VI-RADS scores. The figure includes the number of included patients, the total number of lesions, the number of pre-TURBT and post-TURBT cases, and a simple bar chart showing the distribution of reference VI-RADS categories. This addition was intended to improve readability and provide an immediate overview of the study population and case-mix.

10. A stratified plot of disagreement rate by scenario: low-risk (2–3) vs high-risk (4–5), and TURBT vs non-TURBT could improve

We thank the Reviewer for this helpful suggestion. We agree that a stratified presentation of disagreement rates improves interpretability. In the revised manuscript, we have added a dedicated figure showing disagreement rates stratified by reference risk category (VI-RADS 2–3 vs 4–5) and by clinical setting (post-TURBT vs non-TURBT), separately for the expert and non-expert readers. This additional visualization helps clarify that disagreement rates were higher in high-risk categories for both readers and that the effect of the post-TURBT setting was more evident for the non-expert reader.

11. Standardize terms: bpMRI vs mpMRI, VI-RADS capitalization, “non-expert” definition (years of practice? exposure to VI-RADS training)..

We thank the Reviewer for this helpful comment. In the revised manuscript, terminology has been standardized throughout, including consistent use of bpMRI and mpMRI, as well as uniform capitalization of VI-RADS. We have also clarified the definition of reader experience in the Methods section: the expert reader had more than 15 years of radiology experience and dedicated expertise in genitourinary MRI, whereas the non-expert reader had 5 years of radiology experience and no dedicated subspecialty training in genitourinary imaging.

12. Table 1 is extremely minimal; add comorbidity relevant to contrast contraindication (eGFR categories) if available, since this is part of rationale.

We respectfully clarify that contrast-related comorbidities were not part of the study design and are therefore not relevant baseline variables for Table 1 in the present analysis. The rationale of the study was not to evaluate a selected cohort of patients unable to receive contrast medium, nor to compare outcomes according to renal function or other contraindications to gadolinium administration. Rather, the aim was to assess whether a contrast-free abbreviated bpMRI reading could reproduce routine mpMRI-based VI-RADS assessment when all patients had undergone the same standard multiparametric MRI protocol. In this setting, the clinical motivation for exploring a non-contrast pathway is part of the broader background, but it was not an inclusion criterion and did not define the study population. Accordingly, variables such as eGFR categories would not materially contribute to interpretation of the present results.

13. Clarify whether antiperistaltic agent was uniform for all, and which agent

We thank the Reviewer for this comment. In our routine clinical practice, an antiperistaltic agent is not routinely administered for bladder MRI. Therefore, its use was not standardized across the cohort and no specific antiperistaltic drug formed part of the acquisition protocol. In our experience, satisfactory image quality for bladder MRI can be consistently achieved without mandatory administration of antispasmodic medication. We have revised the Methods section accordingly to avoid implying that antiperistaltic preparation was uniformly performed in all patients.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

No further comments

Reviewer 3 Report

Comments and Suggestions for Authors

The authors responded to the letters and made substantial changes to their manuscript. I consider it suitable for publication.