Metabolic Dysfunction-Associated Fatty Liver Disease (MAFLD) Is Associated with Cervical Stromal Involvement in Endometrial Cancer Patients: A Cross-Sectional Study in South China

Round 1

Reviewer 1 Report

I thank you for asking me to review this manuscript that aimed to evaluate the relationship between MAFLD and cervical stromal invasion of endometrial cancer. Although the manuscript sought to investigate a very important topic, however, I have some major concerns about the methods, statistical analysis and the results presented. 

In sum, the authors were not focused in their result presentation and it appears that the objective(s) were not clearly delineated in the results. Specific comments are as stated below.

Abstract: Line 26: CSI should be defined at first use. Same for other abbreviations.

Abstract: Line 24: Authors stated "The hepatic steatosis index (HSI) was assessed for the best cutoff using the ROC method" What cut-off are you referring to? What is the outcome? In the methods section of the main manuscript, the methodology for the ROC and the cut-off were not sufficiently described for the sake of reproducibility. 

Line 26: Please state the 95% confidence interval of the prevalence of MAFLD in the abstract and in the main manuscript. 

Line 26: delete "positively"

Line 31: " highly prevalent" . A 25% prevalence may not be referred to as "highly prevalent". Authors may wish to revise. You may state "About one-quarter of our cohort had MAFLD"

The Introduction is essentially good. However, authors should improve by reviewing literature on the association between MAFLD and CSI, including the interrelationship of obesity and HSI.

Line 66-67: Authors stated "However, there has not yet been a study reporting MAFLD in the context of the prognosis or clinical outcomes of EC" It appears that this statement is far-fetched with regards to the objective of the present study. The authors did not evaluate nor report on prognosis of the EC.  Thus, this should be revised

Line 67 - 68: "Therefore, the aim of this study was to investigate the associations and clinical consequences of MAFLD on the risk of CSI in EC". Authors should delete "clinical consequences" which appears to be far- fetched from present study

Materials and methods: Fairly well-written. But authors should address the following:

Authors should explicitly state that this study was a " retrospective analytic cross-sectional study"

Line 84: Insert " (Figure 1)" after "Patients" 

Line 85: delete " (Figure 1)"

Authors stated "We obtained each participant's informed consent prior to collecting the medical examination data for our study" . This appears technically difficult to do since the study was a retrospective study. The requirement for participant's consent is not usually extended to retrospective studies. Authors should motivate and justify how it is possible to obtain consent from patients that had been previously managed and discharged home before the commencement of the study. 

Line 93: Please ensure that "FIGO" has been defined at first use.

Statistical analysis: Lines 115 - 123.  The description of the statistical analysis as reported in the results section is deficient. There are several results that were stated in the statistical analysis that were not described here. All statistical analyses underlying the results should be described here. Authors also conducted some sub-analysis/intraclass comparisons. All must be stated here. (Though I am not too sure if the study is adequately powered for such intraclass/sub-analyses)

Some specifics are : 

Please state the country of origin, city and the company producing SPSS and R software.

Did you check for normality before reporting continuous variables as medians? Please show the interquartile range alongside the medians. Age is usually normally distributed which should be reported as mean and standard deviation. Please confirm that normality test was conducted for age and other continuous variables.

Since all the continuous variables were expressed as median, then it may not be appropriate to utilize Student's ttest which is a parametric test.

Authors stated : "Student’s t tests or Mann‒Whitney U tests were used to analyze the continuous variables." What was the categorical variable that was used for this bivariate analysis?. This analysis (Student's ttest) is conducted as the comparison of mean/median of a continuous variable across a binary categorical variable.

"Categorical variables were analyzed by Chi-square tests or Fisher’s exact tests". What is/are the explanatory variables and what is/are the outcomes?

How was the cut-off obtained? was it through Younden's index or other methods? was the ROC a post logistic regression estimation? What is the continuous variable? what is the outcome? Please the relevant ROC curves should be shown even if it will appear as supplementary materials. ROC estimates such as the area under the curve should also be shown. Even if it is in the supplementary material.

"Adjusted analyses were performed using multiple logistic regression to investigate binary associations"

The description of the regression modelling should be explicit. Was there a univariable regression modelling? What cut-off of P-value was utilized for selecting variables for multivariable regression modelling? What type of regression modelling was done? backward elimination, stepwise forward etc. What is the outcome? What are the explanatory variables/confounding variables?

Table 1. Title: Revise to "Comparison of the demographic and clinical characteristics among Endometrial cancer patients with and without MAFLD".

Lines 133 - 134: "Among prognostic factors, including FIGO stage, histological grade, CSI, MI, LVSI and LNM, the difference was statistically significant only for CSI (P 133 = 0.024, Table 1)." 

Please authors should compare the proportions between CSI and MAFLD in the Table. 

The Chi-square association between CSI and MAFLD has been shown in Table 1. Why are you repeating the association again in Table 3?

Although the study was a cross sectional study and causality may not be easily ascertained, however from the objective/hypothesis of this study, I suspect the outcome was CSI while MAFLD was the primary explanatory variable.

Conventionally, we usually compared other characteristics among the groups of the outcome. Thus, Table 1 which compared characteristics among MAFLD positive or negative is conventionally not required. You may send Table 1 to the appendix/supplementary Table. If not, it will not be clear what was your explanatory variable or outcome for this study.

Line 145: change "in" to " among". Delete "total"

Line 145:  "The optimal cutoff value of the hepatic steatosis index (HSI) in total EC patients was 34.37  according to ROC analysis"  This statement is not clear. Is it that you want to use the optimal HSI level to diagnose EC? But all your participants were EC? So what is the outcome? What is optimum cut-off referring to as it may not be EC.? Please produce the ROC curves and show them in the main text or as supplementary materials. The optimal cut-off is a different sentence/paragraph that may come after concluding everything about the associations in Tables 2 and 3.

Line 146: "There was no significant difference between the two groups in HSI,.."  It is a bit difficult to conduct optimal cut-off when there is no statistically significant relationship. What is the value of the area under the ROC curve of this relationship? The ROC curves should be shown. Authors may reconsider this and possibly delete this analysis and result from the research/study. 

Line 143: Delete "Moreover"

Line 152 -154 : Authors stated "After adjusting for age, FIGO stage and grade, MAFLD was significantly associated 152 with cervical stromal involvement in the total population (odds ratios (OR) 2.204; 95% 153 confidence intervals (CI) 1.221-3.978; P = 0.009; Figure 2)."

This expression should be stated after Table 3 and close to Figure 2. 

From Table ER expression and Ki were statistically associated with CSI". Why did you not adjust for these variables in the multivariable model? By convention, variables with P-value < 0.2 at univariable analysis are included in the multivariable model building.

Figure 2. delete "Padj" and replace with "P-value"

Figure 2.  please change "This The Multivariate Logistic Regression Analysis for CSI in Women with EC"

to 

"Multivariable logistic regression of the association between MAFLD and CSI among endometrial cancer patients"

Figure 2: Please make "MAFLD No Vs Yes" as the first variable in the figure since this is supposed to be the primary explanatory variable in the model

Figure 2: You have included many variables in this model. But authors stated that only "age, FIGO stage and grade" were corrected for in the model. Please state the other variables in the model. 

Authors should check for collinearity of the variables in the model and report their results

Line 173: Authors stated : "Thus, we performed a subgroup analysis by dividing the subjects into type I EC and type II EC groups." This sub-group analysis should be described in the statistical analysis plans. Furthermore, such sub-analysis should be justified by reviewing the relevant literature in the introduction.

Line 175: Change "cervical stromal invasion" to "CSI". Please do the same throughout the manuscript. When you define an abbreviation, you are expected to continue with the abbreviation throughout the manuscript.

Line 179 - 180 "However, no significant relationship was observed between MAFLD and cervical stromal involvement in the type II EC group (P = 0.332, Table 4)"

You may add in the limitation of the study that being a retrospective study, sample size was not calculated. Thus, sample size of 112 may not be adequately powered for this sub-analysis and you run the risk of Type 2 error of stating that there was no association whereas there was a true association. Small/inadequate sample size can cause type 2 error.  

The authors stated :"The optimal cutoff values of HSI 180 in type I and type II patients were 34.37 and 31.77, respectively. However, we did not find any statistically significant difference between HSI and the risk of cervical stromal involvement in the type I EC group (P = 0.079, Table 4) or type II EC group (P = 0.103, Table 183 4)." As stated earlier, it may be difficult to find the optimum cut-off when there is no statistically significant association. Authors either delete this ROC/optimum cut-off or describe in detail in the statistical analysis on how it was obtained and also show the ROC diagrams together with the AUCs

Lines 184 - 185: "Then, we analyzed 5 factors, including MAFLD, grade, PR expression, Ki67 index and HSI, in a stepwise manner" This statement should be in the statistical analysis section of the methodology, and you need to describe how you utilized only five factors out all the factors available. If not, the authors run the risk of "data dredging"

Table 4. Title. Change "Correlation" to "Association". Change "in" to "among"

Table 4. Please change "P1" and "P2" to "P-value"

Figure 3a and 3b. MAFLD should be the first variable being the primary explanatory variable of the model. Please explain in detail how these models are built in the methods/statistical analysis section

3.4 . Change "Correlation" to "Relationship" 

I do not see the need for this section 3.4 in this paper. It makes the paper not to be focused. This is obviously not part of the objective of the study. My recommendation is that it should be deleted. Or a strong motivation should be stated in the methodology/statistical analysis section. Indeed, it is not easy to vouch for the power of this sub-analysis. Furthermore, why are you focusing on non-obese women only?

Conclusion

line 217: change "demonstrate" to "demonstrated"

Line 218: delete "positive"

Line 219 - 222: " We also show that HSI was associated with the presence of CSI in nonobese patients with Type I endometrial cancer" . This may be deleted if previous recommendation is followed.

Line 247: Change "stromal invasion of the cervix" to "CSI". 

Line 249 - 250 : "We aim to identify if NAFLD could 249 act as a driving force in cancer development. The correlation between MAFLD and CSI is not clear" This sentence is out of place here. You may delete it or recast.

Line 272-275: " Another interesting finding was that the use of the Hepatic Steatosis Index may be helpful in nonobese type I EC patients and can identify EC patients at high risk for hepatic disturbances" This statement/conclusion may be far-fetched

Line 278: change "cervical stromal invasion" to "CSI" . Do this throughout the manuscript

Line 281: Change "most" to "only"

Line 294: "In addition, it is recommended that HSI be included in the prognostic assessment of EC patients, especially in nonobese women with type I EC." This sentence/conclusion may not be correct. Please confirm

Once again, I thank you for asking me to review the manuscript

Author Response

We deeply thank the reviewers’ and editor for their work and hope that the corrections will meet with your approval. Please see the attachment.

Author Response File: Author Response.doc

Reviewer 2 Report

The authors wrote in the introduction that there has not yet been a study reporting metabolic dysfunction-associated liver disease (MAFLD) in the context of the prognosis or clinical outcomes of endometrial cancer. Then the authors investigated the associations and clinical consequences of MAFLD on the risk of cervical stromal invasion in endometrial cancer.

The problem with this article is that the authors did not assess the prognosis or clinical outcomes of endometrial cancer patients with cervical stromal invasion. Cervical stromal invasion is usually associated with a poor prognosis, but node-negative patients with cervical stromal invasion may have a favorable prognosis. Survival data are necessary.

Obesity or overweight is defined as BMI ≥23 kg/m2, based on the Asia-Pacific criteria in this article. This sentence must be referenced. I do not think "Asia-Pacific criteria" is appropriate, because pacific island nations are known to be the heaviest countries, in contrast to Asian countries. In addition, obesity rates in western and eastern Asia appear different.

Materials and Methods

The authors evaluated EC patients who underwent TAH and BSO with or without lymphadenectomy. However, the authors excluded patients who did not undergo TAH and BSO with or without surgery from this study population (exclusion criterion (2)).

Are cervical lesions and uterine leiomyoma malignant tumors?

How do you calculate homeostasis model assessment and insulin resistance (HOMA-IR)?

The authors need to explain the meaning of HSI values.

Results

The study included 792 EC patients, consisting of 205 patients with MAFLD and 592 patients with non-MAFLD (Line 126). 792 = 205 + 592 ?

In contrast, 824 EC patients were included in this study (Abstract). Which is the correct number?

Is the age difference between the MAFLD group and the non-MAFLD group (54 yr vs 53 yr) clinically meaningful?

The definitions of type I and type II EC are necessary.

Lynch syndrome-associated endometrial cancers often involve the lower uterine segment, which often invade cervical stroma. How many patients in this study were suspected to have Lynch syndrome based on family history?

Author Response

We deeply thank the reviewers’ and editor for their work and hope that the corrections will meet with your approval. Please see the attachment.

Author Response File: Author Response.doc

Reviewer 3 Report

I read with great interest the Manuscript titled "Metabolic dysfunction-associated fatty liver disease (MAFLD) is associated with cervical stromal involvement in endometrial cancer patients: a cross-sectional study in South China" which falls within the aim of the Journal.

In my honest opinion, the topic is interesting enough to attract the readers’ attention. Methodology is accurate and conclusions are supported by the data analysis. Nevertheless, authors should clarify some point and improve the discussion citing relevant and novel key articles about the topic.

-The whole text should be corrected by a native English speaker in order to make the work clearer and more readable.

-The introduction should be extended and completed. I find interesting a reference to the efforts made for the prevention and early diagnosis of gynecological cancers (see PMID: 36141217).

 

- Discussions can be expanded and improved by citing relevant articles (I suggest authors to read and insert in references the following article PMID: 35742340; 36553988).

Considered all this points, I think it could be of interest for the readers and, in my opinion, it deserves the priority to be published after minor revisions.

 

Author Response

We deeply thank the reviewers’ and editor for their work and hope that the corrections will meet with your approval. Please see the attachment.

Author Response File: Author Response.doc

Round 2

Reviewer 1 Report

Authors have substantially improved the manuscript. However, authors should consider the following points before acceptance/publication

1.The ethical statement should be reworked in past tense. Authors should just state that ethical approval was obtained and state the ethics certificate number. Authors can then state that being a retrospective study, informed consent was waived by the ethics committee.

2. Since the associations were not significant, it is difficult to believe that a cut-off point can be obtained. Moreover, the Area under the ROC curves showed very poor values  of around 50% with some even crossing the diagonal line. This patently showed poor diagnostic/predictive value and to now utilise same for calculation of cut-off appears far-fetched. Authors should review statistical books and be sure what they did with such poor ROC values is not flawed. Such statistical references should be cited. 

3. Authors produced some supplementary figures of Kaplan Meir plots. Are these part of the objectives of this present study? If yes, please authors should describe such statistical analysis in the statistical analysis section of the methods. Authors should also write a report of the Kaplan Meir plots in the result section and explicitly refer to the supplementary material on Kaplan Meir plots

4. An English language editor should thoroughly the manuscript 

I suggest these issues should be adequately addressed before accepting the manuscript for publication. 

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Reviewer 2 Report

In this revised version, some problems remain. The data accuracy in this manuscript must be improved.

 

What is the true number of patients in this study?

Line 171. The authors write, 'The study included 797 EC participants who had abdominal ultrasound and biochemical data, among whom 205 (25.7%) had MAFLD and 592 (74.3%) had non-MAFLD (95% confidence interval [Cl]=20.5-28.5).'

However, the total number of patients is 824 (= 696 + 128) in Tables 1 and 2.

In Table 2, the number of non MAFLD patients is 592 and that of MAFLD patients is 127. Thus, the total number is 719.

All of the data must be re-examined. Based on the correct numbers, Tables and Figures need to be recreated.

 

Similarly, what is the definition of overweight or obesity?

Line 116. The authors write, 'The non-obesity population was defined as BMI <25 kg/m2.'

Line 129. overweight or obesity (BMI ≥23 kg/m2)

Line 247 nonobese patients (BMI ≤25)

 

Line 98-101 must be rewritten, as uterine leiomyoma is not a malignant tumor and cervical lesions also may not be a malignant tumor when they do not have stromal invasion.

 

Line 130. What are metabolic abnormalities? What is the difference between metabolic abnormalities and metabolic dysregulation?

The authors should comment on the importance of CSI based on survival analysis.

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Round 3

Reviewer 2 Report

Abstract.

  The authors write, "Kaplan-Meier survival curves were used for survival analysis." However, survival analysis was not reported in the Results. If this study evaluates MAFLD and clinical (not pathological) outcomes, the authors need to mention a survival difference between patients with MAFLD and patients without MAFLD.

  Spell out HSI, when it is first mentioned.

Introduction

  Line 45. What are "the majority of factors associated with EC"? Please give examples.

Materials and Methods

  Line 99. What are "those"? Please explain in the text.

  Line 114. This sentence must be referenced.

Results

  Line 189. The authors state that "The association between MAFLD and CSI in EC patients is presented in Tables 1-2." However, Table 1 indicates the association between CSI and patient demographics and clinical characteristics, and Table 2 indicates the association between CSI and clinical biochemical and imaging indicators.

  In Table 1, tumors were classified into G1 and G2-G3. In contrast, Type 1 EC was defined as grade 1-2 endometrioid carcinomas. The authors need to explain why different classifications were used in this manuscript.

 The authors need to indicate how to classify ER/PgR into low and high.

 In Table 2, the authors use "Ultrasound, Fatty liver." However, in Materials and Methods section, the authors use "ultrasonography-diagnosed hepatic steatosis (Line 127)." What is the difference?

 Figure 2. The title is wrong.

Discussion

  Line 289. I do not understand what this sentence means.

  Line 336. What is "a prognostic factor for the development of EC"?

Conclusions

  The authors state, "clinicians may be able to identify individuals with a high risk of EC early by closely monitoring for metabolic abnormalities." However, the results of this study do not support this statement. The results only indicate the association between CSI and MAFLD.

As another reviewer commented, the whole text should be corrected by a native English speaker in order to make the work clearer and more readable. By doing this, the number of revisions may be reduced.

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Round 4

Reviewer 2 Report

Most points are adequately corrected in this version.