High Tg/HDL-Cholesterol Ratio Highlights a Higher Risk of Metabolic Syndrome in Children and Adolescents with Severe Obesity

Round 1

Reviewer 1 Report

This is an interesting study comparing several indices to detect
metabolic syndrome. One strength of the manuscript is the large   sample size and comparison of many indices. Overall, the manuscript   is well written and the results are interesting and should be published.

I have some problems to understand the analysis and ask the authors if this can be clarified.

1) line 158: A quantile regression is used to model body composition index as a function of age. I cannot find the results of this analysis.

2) I assume that the significance test in table 1 compare boys and
      girls? Please clarify this in the caption.

3) In table 2 the significance tests compare MetS and no MetS?

4) The ROC analysis does not reveal information about the best
   cut-off values and the corresponding sensitivity and specificity. Can you provide this information?

5) I don't understand the analysis in supp table S1: What is shown
     in the three tables? What contrast method is used? Looks like a
      Dunnet contrast was applied? Alpha error adjustment?

6) How was data in table 5 analysed? Logistic Regression? What pseudo R^2 was used?

7) Which method was used in figure 3? A linear regression with    response TG/HDL-C ratio? How were the age related cut-off values
computed? Do you tested for interaction age X gender?

     

Author Response

Comments and Suggestions for Authors

This is an interesting study comparing several indices to detect
metabolic syndrome. One strength of the manuscript is the large   sample size and comparison of many indices. Overall, the manuscript   is well written and the results are interesting and should be published.
I have some problems to understand the analysis and ask the authors if this can be clarified.
Thank you for taking the time to review our article carefully. Your feedback was very helpful in improving the overall quality of the manuscript. Here enclosed please find a point by point response to your queries below.

Q1. line 158: A quantile regression is used to model body composition index as a function of age. I cannot find the results of this analysis.

A1. The percentile of the ordered values corresponding to the optimal cutoff value identified with the Youden index, was used as regression quantile. The age-predictive values resulting from the regression are represented in the cutoff graph for the Mets. The statistics for regressions are reported in the following boxes (for boys and girls).

 

 

Boys TG/HDL

Summary Statistics
Variable	Q1	Median	Q3	Mean	Standard Deviation	MAD
Age	12.6600	14.7050	16.1600	14.4527	2.1107	2.5278
TG_HDL	0.7173	1.0310	1.4346	1.1597	0.6380	0.5226

 

Quantile and Objective Function
Quantile	0.6573705
Objective Function	115.2791
Predicted Value at Mean	1.2404

 

Parameter Estimates
Parameter	DF	Estimate
Intercept	1	0.5334
Age	1	0.0489

 

Test Results
Test	Test Statistic	DF	Chi-Square	Pr > ChiSq
Wald	8.2419	1	8.24	0.0041
Likelihood Ratio	7.9382	1	7.94	0.0048

 

Girls TG/HDL

Summary Statistics
Variable	Q1	Median	Q3	Mean	Standard Deviation	MAD
Age	12.6600	14.7050	16.1600	14.4527	2.1107	2.5278
TG_HDL	0.7173	1.0310	1.4346	1.1597	0.6380	0.5226

 

Quantile and Objective Function
Quantile	0.5932504
Objective Function	116.5757
Predicted Value at Mean	1.1526

 

Parameter Estimates
Parameter	DF	Estimate
Intercept	1	0.3424
Age	1	0.0561

 

Test Results
Test	Test Statistic	DF	Chi-Square	Pr > ChiSq
Wald	13.4575	1	13.46	0.0002
Likelihood Ratio	10.3072	1	10.31	0.0013

 

 

Boys VAI

Summary Statistics
Variable	Q1	Median	Q3	Mean	Standard Deviation	MAD
Age	12.6600	14.7050	16.1600	14.4527	2.1107	2.5278
VAI	0.9408	1.3793	1.9300	1.5751	0.8884	0.7139

 

Quantile and Objective Function
Quantile	0.669323
Objective Function	158.6746
Predicted Value at Mean	1.7137

 

Parameter Estimates
Parameter	DF	Estimate
Intercept	1	0.5325
Age	1	0.0817

 

Test Results
Test	Test Statistic	DF	Chi-Square	Pr > ChiSq
Wald	10.8939	1	10.89	0.0010
Likelihood Ratio	10.7044	1	10.70	0.0011

 

Girls VAI

Summary Statistics
Variable	Q1	Median	Q3	Mean	Standard Deviation	MAD
Age	12.6600	14.7050	16.1600	14.4527	2.1107	2.5278
VAI	0.9408	1.3793	1.9300	1.5751	0.8884	0.7139

 

Quantile and Objective Function
Quantile	0.589698
Objective Function	160.4864
Predicted Value at Mean	1.5268

 

Parameter Estimates
Parameter	DF	Estimate
Intercept	1	0.3086
Age	1	0.0843

 

Test Results
Test	Test Statistic	DF	Chi-Square	Pr > ChiSq
Wald	15.3390	1	15.34	<.0001
Likelihood Ratio	14.6530	1	14.65	0.0001

 

 

 

BOYS CMI

Summary Statistics
Variable	Q1	Median	Q3	Mean	Standard Deviation	MAD
Age	12.6600	14.7050	16.1600	14.4527	2.1107	2.5278
CMI	0.5005	0.7493	1.0436	0.8416	0.4802	0.3938

 

Quantile and Objective Function
Quantile	0.6533865
Objective Function	87.4906
Predicted Value at Mean	0.8879

 

Parameter Estimates
Parameter	DF	Estimate
Intercept	1	0.3406
Age	1	0.0379

 

Test Results
Test	Test Statistic	DF	Chi-Square	Pr > ChiSq
Wald	7.6504	1	7.65	0.0057
Likelihood Ratio	11.0503	1	11.05	0.0009

 

GIRLS CMI

Summary Statistics
Variable	Q1	Median	Q3	Mean	Standard Deviation	MAD
Age	12.6600	14.7050	16.1600	14.4527	2.1107	2.5278
CMI	0.5005	0.7493	1.0436	0.8416	0.4802	0.3938

 

Quantile and Objective Function
Quantile	0.7282416
Objective Function	83.5634
Predicted Value at Mean	1.0082

 

Parameter Estimates
Parameter	DF	Estimate
Intercept	1	0.2922
Age	1	0.0495

 

Test Results
Test	Test Statistic	DF	Chi-Square	Pr > ChiSq
Wald	10.8086	1	10.81	0.0010
Likelihood Ratio	8.0041	1	8.00	0.0047

 

 

Q2. I assume that the significance test in table 1 compare boys and girls? Please clarify this in the caption.

A2. Your interpretation was correct. We have added a sentence to clarify this point.

Q3. In table 2 the significance tests compare MetS and no MetS?

A3. Your interpretation was correct. We have added a sentence to clarify this point.

Q4. The ROC analysis does not reveal information about the best cut-off values and the corresponding sensitivity and specificity. Can you provide this information?

A4. Cut offs identified with ROC analysis were the following:

 

	Boys	Girls
VAI	1.71	1,89
TG/HDL	1.26	0.99
CMI	0.89	0.87

Sensitivity and other indexes are reported in table 3.

 

Q5. I don't understand the analysis in supp table S1: What is shown in the three tables? What contrast method is used? Looks like a Dunnet contrast was applied? Alpha error adjustment?

A5. The analysis compared the different ROC models (in terms of AUC) and produced a contrast matrix of differences between each ROC curve and the reference curve (VAI, TG/HDL, CMI). Differences were tested with chi-square test and the 95% Wald confidence limits of the differences was showed.

 

Q6. How was data in table 5 analyzed? Logistic Regression? What pseudo R^2 was used?

A6. Data were analyzed by logistic regression. We apologize for the typing error: R square was actually used. In the revised table the corrected values have been reported.

 

	Boys		Girls
	OR (95% CI)	R square	OR (95% CI)	R square
Large WC (IDF)	n.c.*	n.c.*	267.1 (0.1 to infinite)	0.006
High Glucose (IDF)	1.7 (0.5-6.1)	0.002	1.8 (0.6-5.2)	0.003
Low HDL cholesterol (IDF)	17.1 (9.8-29.9)&	0.321	22.9 (12.4-42.2)&	0.341
High Triglycerides (IDF)	121.2 (39.6-370.6)&	0.351	167.3 (50.3-556.9)&	0.374
High Blood Pressure (IDF)	0.9 (0.6-1.2)	0.141	1.3 (1-1.7)	0.042

Table 5. Odds ratio and 95% confidence interval for cardiometabolic risk factors across TG/HDL adjusted for age.

 

Q7. Which method was used in figure 3? A linear regression with response TG/HDL-C ratio? How were the age related cut-off values computed? Do you tested for interaction age X gender?

A7. The percentile of the ordered values corresponding to the optimal cutoff value identified with the Youden index, was used as regression quantile. The age-predictive values resulting from the regression have been represented in the cutoff graph for the Mets. Age rounded to first decimal place and separated analysis by gender.

 

 

Reviewer 2 Report

 The manuscript “High Tg/HDL-cholesterol ratio highlights a higher risk of metabolic syndrome in children and adolescents with severe obesity” is well written. The authors used extensive clinical and paraclincal data that are well documented. I think that the research question is relevant to the field, yet it can be better addressed in the methodology section. The conclusions are consistent with the evidence and arguments presented and I consider the references appropriate.

Suggestions to the authors:

The manuscript needs a few language corrections before being published (revision by a native English speaker is strongly advised).

 “..offers the same accuracy of other more sophisticated”- I suggest “…as other”

 “For this reason, the need to have available tools allowing early identification of patients with high degree of obesity at risk for MetS has become mandatory.” I think the phrase needs to be rephrased

I would suggest “unnecessary investigations”

WHR- please use the abbreviations consistently

ABSI- define the terms at the first use in text

“The data were first scrutinized for outliers, using a cutoff of 4.5 SDS”.- please be more specific to which variable this cut-off refers to

“In Table 4 the correlation between TG/HDL-C ratio with the”…. I suggest “The correlation between… and…”

“In Table 5 the odds ratio between TG/HDL-C ratio with the different cardiometabolic  risk factors are reported. In both sexes, a strong association with high triglycerides and low HDL-C was found, but not with glucose and blood pressure”.- the phase should de reformulated and the data better explained in text; the findings seem redundant

“Figure 5 Trend of TG/HDL cutoff for MetS, according to age and sex”- it is not very clear to which cut off are you referring to; what was the value of this cutoff? is there a significant difference between genders?

I suggest a more clear and comprehensive expression of your results in text

As both TG and HDL-C are included in the definition of MetS, the association between the two, or their ratio, and the presence of MetS is more than obvious from the beginning, and the demonstration may seem redundant; maybe the association between TG/HDL and cardiovascular risk should have been of more value

“Conversely to what observed in a previous paper by our group, showing that BMI  was similar to other indexes in detecting MetS in children and adolescents (15), in the  present study BMI resulted less accurate in identifying MetS than the TG/HDL-C ratio.”- as I already mentioned before both TG and HDL are already included in the definition of MetS

 

Author Response

The manuscript “High Tg/HDL-cholesterol ratio highlights a higher risk of metabolic syndrome in children and adolescents with severe obesity” is well written. The authors used extensive clinical and paraclincal data that are well documented. I think that the research question is relevant to the field, yet it can be better addressed in the methodology section. The conclusions are consistent with the evidence and arguments presented and I consider the references appropriate.

Thank you for taking the time to review our article carefully. Your feedback was very helpful in improving the overall quality of the manuscript. Here enclosed please find a point by point response to your queries below.

 

Q1. The manuscript needs a few language corrections before being published (revision by a native English speaker is strongly advised).

 “..offers the same accuracy of other more sophisticated”- I suggest “…as other”

 “For this reason, the need to have available tools allowing early identification of patients with high degree of obesity at risk for MetS has become mandatory.” I think the phrase needs to be rephrased

I would suggest “unnecessary investigations”

A1. The paper has been revised by an English teacher.

 

Q2. WHR- please use the abbreviations consistently

ABSI- define the terms at the first use in text

A2. The requested corrections have been done.

 

Q3. “The data were first scrutinized for outliers, using a cutoff of 4.5 SDS”.- please be more specific to which variable this cut-off refers to

A3. The cut-offs were referred to VAI, TG/HDL, CMI. We have clarified this point in the text

 

Q4. “In Table 4 the correlation between TG/HDL-C ratio with the”…. I suggest “The correlation between… and…”

A4. Corrected as suggested

 

Q5. “In Table 5 the odds ratio between TG/HDL-C ratio with the different cardiometabolic  risk factors are reported. In both sexes, a strong association with high triglycerides and low HDL-C was found, but not with glucose and blood pressure”.- the phrase should be reformulated and the data better explained in text; the findings seem redundant

 A5. The sentence has been reformulated

 

Q6. “Figure 5 Trend of TG/HDL cutoff for MetS, according to age and sex”- it is not very clear to which cut off are you referring to; what was the value of this cutoff? is there a significant difference between genders?

A6. The trend in Figure 3 (not fig 5!) shows the cut-off values at various ages calculated with a quantile regression starting from the percentile of the cut-off identified with the ROC curve. The mean difference of the cut-off values between sexes, calculated at all ages, was always significant (p<0.05). We have added a sentence to clarify this point

 

Q7. I suggest a more clear and comprehensive expression of your results in text.

As both TG and HDL-C are included in the definition of MetS, the association between the two, or their ratio, and the presence of MetS is more than obvious from the beginning, and the demonstration may seem redundant; maybe the association between TG/HDL and cardiovascular risk should have been of more value

A7. The reviewer is correct, in fact in table 4 and 5 the correlations and the OR between TG/HDL and the cardiovascular risk factors are reported.

 

Q8. “Conversely to what observed in a previous paper by our group, showing that BMI  was similar to other indexes in detecting MetS in children and adolescents (15), in the  present study BMI resulted less accurate in identifying MetS than the TG/HDL-C ratio.”- as I already mentioned before both TG and HDL are already included in the definition of MetS

A8. It is correct that TG and HDL-C are included in the definition of MetS, as for other indices. However, the aim of this study was to find out the simplest and more accurate index, among all indices, able to detect MetS in a particular population of extremely obese patients