Urine Metabolites as Biomarkers and Metabolism Mechanism Studies of Alcohol-Associated Liver Disease

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Manuscript by He et al describes potential ALD biomarkers in the urine of patients. However, lack of validation cohort and a weak correlation between suggested biomarkers and disease severity dampen the enthusiasm for this study.

Comments:

1. Authors previously examine bile acids, caffeine metabolism and keratin 18 in the urine of the same patients. How do findings in this paper correlate with other changes in urine metabolites and proteins, will the combination of these measurements serve as a better predictor, or these are redundant changes?
2. Authors showed that using the abundance of 3-hydroxysebacic acid and androsterone glucuronide could differentiate patients from HC with an accuracy of 0.97. Authors need to show that this can be reproduced in a validation cohort.
3. Authors demonstrate the change in appr 200 metabolites between healthy control and AH patients; however, it is not clear whether these changes are specific to AH. Non-alcohol disease control should be included in the study.
4. The presented data suggest that urine metabolites could differentiate between ALD and severe AH, however, it is not clear whether these metabolites had better predictive value than the MELD score for example for patients’ survival, decompensation etc.
5. It is hard to determine the distribution of individual patients’ measurements from bar graph. Violin plots would be more appropriate for data presentation.
6. Histidine and arginine metabolism were the most affected in ALD patients. Authors speculate that these changes represent defect in liver metabolic pathways. Authors should confirm that corresponding changes exist in AH patients’ livers for example by assessing liver transcriptomic or metabolomic changes in AH compared to ALD control (several published dataset could be utilized).

Author Response

Manuscript by He et al describes potential ALD biomarkers in the urine of patients. However, lack of validation cohort and a weak correlation between suggested biomarkers and disease severity dampen the enthusiasm for this study.

Answer: The reviewer is right, it will be more confident if these discoveries were confirmed by a validation cohort and we plan to confirm this result in a larger cohort. However, collecting human samples takes time and it is extremely difficult for the control group. We added a weak correlation of the biomarkers to the limitations in our revised manuscript (Page 9, Line 352 - 354). The changed pathways discovered in the current study are very important. Several publications showed the changes in urea synthesis capacity in patients with liver disease, which is related to the arginine biosynthesis pathway. However, they focused on the urea circle and the changes of individual metabolites in this pathway, and the impact of this pathway on disease development was seldom discussed. In addition, no literature reports show the changes in the histidine pathway in ALD. Therefore, this is the first report focusing on the entire arginine biosynthesis pathway changes and histidine metabolism. As the top changed pathways, we believe they play an important role in ALD.

Comments:

1. Authors previously examine bile acids, caffeine metabolism and keratin 18 in the urine of the same patients. How do findings in this paper correlate with other changes in urine metabolites and proteins, will the combination of these measurements serve as a better predictor, or these are redundant changes?

Answer: We agree with the reviewer, i.e., combining different metabolites will increase the accuracy of disease stage differentiation. As in our previous publication, combining seven significantly changed bile acids could differentiate severe-AH and non-severe ALD with a perfect score in the ROC analysis. Combining potential biomarker molecules detected in different methods requires the same sample analyzed using multiple methods, which consumes more biological samples and extends analysis time. Overall, it is a great idea to combine different potential biomarkers (BAs, Arg, His, and Caffeine metabolites) for disease diagnosis if a comprehensive method is developed to quantify all these metabolites in one injection.   

1. Authors showed that using the abundance of 3-hydroxysebacic acid and androsterone glucuronide could differentiate patients from HC with an accuracy of 0.97. Authors need to show that this can be reproduced in a validation cohort.

Answer: The reviewer is right; it will be more confident if these discoveries could be confirmed by a validation cohort. We plan to confirm this result in a larger cohort. However, collecting human samples takes time and it is extremely difficult for the control group. We added limitations in a weak correlation of the biomarkers in our revised manuscript (Page 9, Line 352 - 354).

1. Authors demonstrate the change in appr 200 metabolites between healthy control and AH patients; however, it is not clear whether these changes are specific to AH. Non-alcohol disease control should be included in the study.

Answer:  For 194 significantly changed metabolites, many of them were also changed in NAFLD, such as glutamic acid, carnitine, metabolites in the arginine metabolism pathway, histidine, and amino acids. We didn’t find reports on some of them in NAFLD, but different extraction and detection methods in publications may detect different metabolites. Therefore, we are not sure whether these not reported in NAFLD are specific to AH. We added a limitation of our study in the revised manuscript (Page 10, Line 355 - 357). 

1. The presented data suggest that urine metabolites could differentiate between ALD and severe AH, however, it is not clear whether these metabolites had better predictive value than the MELD score for example for patients’ survival, decompensation etc.

Answer: We did find several metabolites may have better predictive value than the MELD score and added them to our revised manuscript (Page 5, Line 202 – 219 and Page 6, Fig. 2).

1. It is hard to determine the distribution of individual patients’ measurements from bar graph. Violin plots would be more appropriate for data presentation.

Answer: Thanks for the reviewer’s suggestion. We changed the bar charts to violin plots in our revised manuscript (Page 7, Fig. 3 and Page 8, Fig. 4).  

6. Histidine and arginine metabolism were the most affected in ALD patients. Authors speculate that these changes represent defect in liver metabolic pathways. Authors should confirm that corresponding changes exist in AH patients’ livers for example by assessing liver transcriptomic or metabolomic changes in AH compared to ALD control (several published dataset could be utilized).

Answer: We added some results from AH patients’ liver samples (Page 6, Line 235 - 240 and Page 8, Line 287 - 289). It is a good idea to explain our results, thanks for the reviewer’s suggestion.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript by He et al. provides details regarding the analysis of urine Metabolites as Biomarkers and Metabolism Mechanism Studies of Alcohol Associated Liver Disease

Some of the research results are of interest as they suggest the possibility of using urine polar metabolites as non-invasive biomarkers of alcohol-associated liver disease (ALD) for early-stage diagnosis and severity assessment.

There are, however, some questions that the authors should address and some suggestions which need to be considered.

• In Methods and Patients section is described that all study participants had a complete history, physical examination, and laboratory evaluation. In table S1 age and sex are available. Authors have investigated whether age affect results.
  1. The variable sex should be also taken into account. Female patients seem to be underrepresented in comparison to male patients, mainly in severe AH. Are severe AH female patients scarce? Does severe AH predominantly affect men? Authors should consider that results could be different between men and women due to differences in metabolism.
  2. Are there other interesting data such as genetic background due to ethnicity, which can modulate or affect liver alcohol metabolism, available? Did the inclusion criteria pay attention to other factors, such as race, diet, etc.?
  3. As a laboratory panel was undertaken, were any correlations between urinary biomarkers and serum parameters analyzed?

 

• In some paragraphs authors claim that single samples were used for the analysis but according to other paragraphs of the manuscript group based pooled samples were used. This information should be clarified.

 

• Authors claim that the decreased levels of urea and NO in patients suggests that urea synthesis and the conversion of arginine to ornithine might be inhibited in the livers of patients with AH. Regarding arginine biosynthesis pathway authors suggest that metabolites in this pathway are not good candidates to stage ALD diagnosis. Do authors think that these metabolites may have any interest as biomarkers in some other situations? What other studies could be undertaken to explain these changes? Was ornithine measured?
• Regarding nitric oxide, authors point out that additional studies need to be conducted to determine the change of NO levels. Which additional studies would authors suggest to study this pathway?

 

• Authors discuss several limitations of the study, such as the explanation of the mechanism(s) of ALD development due to the fact that the current study did not include metabolomics study in patient liver tissue samples. Have authors considered measuring any of these metabolites in serum? Other measurements could be done with the available samples such as oxidative stress biomarkers such as malondialdehyde (MDA) levels, catalase (CAT), superoxide dismutase (SOD-1) activities, etc. to support discussion and conclusions.

Minor changes

• In line 29 healthy controls (HC)
• In Fig. S1 caption, healthy controls (HC) should be added and “could differentiate non-severe ALD from severe AH” instead of non-severe AH
• The arginine biosynthesis pathway and histidine metabolism pathway figures 2A and 3A could be improved

Author Response

The manuscript by He et al. provides details regarding the analysis of urine Metabolites as Biomarkers and Metabolism Mechanism Studies of Alcohol Associated Liver Disease

Some of the research results are of interest as they suggest the possibility of using urine polar metabolites as non-invasive biomarkers of alcohol-associated liver disease (ALD) for early-stage diagnosis and severity assessment.

There are, however, some questions that the authors should address and some suggestions which need to be considered.

• In Methods and Patients section is described that all study participants had a complete history, physical examination, and laboratory evaluation. In table S1 age and sex are available. Authors have investigated whether age affect results.
1. The variable sex should be also taken into account. Female patients seem to be underrepresented in comparison to male patients, mainly in severe AH. Are severe AH female patients scarce? Does severe AH predominantly affect men? Authors should consider that results could be different between men and women due to differences in metabolism.

Answer: Gender should be taken into account, however, samples from females are scarce and the analysis may not be objective, therefore, we did not analyze gender influence in this study.

1. Are there other interesting data such as genetic background due to ethnicity, which can modulate or affect liver alcohol metabolism, available? Did the inclusion criteria pay attention to other factors, such as race, diet, etc.?

Answer: Although these samples were collected without consideration of genetic background, gender or ethnicity, the overall sample size remains small. Given the diversity among patients, certain subgroups may have very limited representation. Categorizing samples based on these variables could result in insufficient group sizes, potentially compromising the objectivity and reliability of the analysis. Therefore, we did not evaluate the influence of these parameters in this study.

1. As a laboratory panel was undertaken, were any correlations between urinary biomarkers and serum parameters analyzed?

Answer: We analyzed urinary bile acids with ALT, AST and the ratio of AST/ALT that was detected in patient serum samples. The correlation was not as good as the MELD score. Therefore, we did not analyze the correlation of urinary metabolites with serum parameters in this study.

 

• In some paragraphs authors claim that single samples were used for the analysis but according to other paragraphs of the manuscript group based pooled samples were used. This information should be clarified.

Answer: Each biological sample was analyzed in Full MS mode for metabolite quantification. Group-based pooled samples were analyzed in MS/MS mode for metabolite identification. We tried to make this more clear in the revised manuscript. 

• Authors claim that the decreased levels of urea and NO in patients suggests that urea synthesis and the conversion of arginine to ornithine might be inhibited in the livers of patients with AH. Regarding arginine biosynthesis pathway authors suggest that metabolites in this pathway are not good candidates to stage ALD diagnosis. Do authors think that these metabolites may have any interest as biomarkers in some other situations? What other studies could be undertaken to explain these changes? Was ornithine measured?

Answer: Several publications have reported the changes in the urea cycle in ALD. Urea and NO produced by the arginine biosynthesis pathway in liver cells may not be consistent with that in urine since they may be consumed or participate in signaling pathways. To confirm the changes in urea and NO levels that are caused by alcohol consumption, it is better to detect their levels in human liver samples, as well as the enzyme activity and expression that is responsible for this reaction. We did polar profiling in patient urine samples, unfortunately, ornithine was not detected. The possible reason is a low concentration of ornithine in urine samples, or it is the matrix effect that reduced the chance of ornithine detection.

• Regarding nitric oxide, authors point out that additional studies need to be conducted to determine the change of NO levels. Which additional studies would authors suggest to study this pathway?

          Answer: NO produced in liver cells may not be consistent with that in urine. To confirm the changes in NO levels caused by alcohol consumption, it is better to detect the NO level in human liver samples, as well as the enzyme activity and expression that is responsible for this reaction.  

• Authors discuss several limitations of the study, such as the explanation of the mechanism(s) of ALD development due to the fact that the current study did not include metabolomics study in patient liver tissue samples. Have authors considered measuring any of these metabolites in serum? Other measurements could be done with the available samples such as oxidative stress biomarkers such as malondialdehyde (MDA) levels, catalase (CAT), superoxide dismutase (SOD-1) activities, etc. to support discussion and conclusions.

Answer: We did a polar profiling in serum samples collected from these patients. Pathway analysis showed that the arginine biosynthesis pathway was the top changed pathways, and the histidine pathway was also changed in these patients, the results from serum which will be published elsewhere.

Minor changes

• In line 29 healthy controls (HC)

Answer: We added (HC) in Line 29.

• In Fig. S1 caption, healthy controls (HC) should be added and “could differentiate non-severe ALD from severe AH” instead of non-severe AH

Answer: Non-severe ALD includes AUD and moderate AH patients, not only AH patients. Therefore, we think non-severe ALD is more accurate. In addition, we included the HC group in Fig. S1.

• The arginine biosynthesis pathway and histidine metabolism pathway figures 2A and 3A could be improved.

Answer: We improved the figures.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Authors  responded somewhat to all my comments.

Author Response

Authors responded somewhat to all my comments.

Thanks.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript by He et al. provides details regarding the analysis of urine Metabolites as Biomarkers and Metabolism Mechanism Studies of Alcohol Associated Liver Disease

Some of the research results are of interest as they suggest the possibility of using urine polar metabolites as non-invasive biomarkers of alcohol-associated liver disease (ALD) for early-stage diagnosis and severity assessment.

After revision some questions and suggestions have been addressed and considered but there are still some important limitations of the study as discussed by authors.

 

The variable sex should also be taken into account. Female patients seem to be underrepresented in comparison to male patients, mainly in severe AH. According to the authors samples from females are scarce and they did not analyze gender influence in this study. Findings should be representative of both sexes. Sex as a variable should be added to the discussion section.

 

Other interesting data or criteria such as genetic background due to ethnicity, diet, etc. can modulate or affect liver alcohol metabolism. According to authors, samples were collected without consideration of genetic background, gender or ethnicity, as the overall sample size is small. Therefore, this should be added to the discussion section.

 

A laboratory panel was undertaken, but according to the authors correlations between urinary biomarkers and serum parameters were not analyzed in this study due to the fact that they considered MELD score was a better parameter. This could be added to the discussion section.

 

Regarding urea and NO levels authors claim that to confirm that changes are caused by alcohol consumption, it is better to detect their levels in human liver samples, as well as the enzyme activity and expression that is responsible for this reaction.

 

Authors discuss several limitations of the study, such as the explanation of the mechanism(s) of ALD development due to the fact that the current study did not include metabolomics study in patient liver tissue samples. Authors have measured metabolites in serum but results will be published elsewhere. The results from these measurements could support discussion and conclusions.

After revision, information about samples has been added, minor changes were added and figures have been improved

Author Response

The manuscript by He et al. provides details regarding the analysis of urine Metabolites as Biomarkers and Metabolism Mechanism Studies of Alcohol Associated Liver Disease. Some of the research results are of interest as they suggest the possibility of using urine polar metabolites as non-invasive biomarkers of alcohol-associated liver disease (ALD) for early-stage diagnosis and severity assessment. After revision some questions and suggestions have been addressed and considered but there are still some important limitations of the study as discussed by the authors.

The variable sex should also be taken into account. Female patients seem to be underrepresented in comparison to male patients, mainly in severe AH. According to the authors samples from females are scarce and they did not analyze gender influence in this study. Findings should be representative of both sexes. Sex as a variable should be added to the discussion section.

 Answer: Thanks for the reviewer’s suggestion, we added this limitation in our revised manuscript (Page 10, Line 359 – 361).

Other interesting data or criteria such as genetic background due to ethnicity, diet, etc. can modulate or affect liver alcohol metabolism. According to authors, samples were collected without consideration of genetic background, gender or ethnicity, as the overall sample size is small. Therefore, this should be added to the discussion section.

 Answer: Thanks for the reviewer’s suggestion. We added this as another limitation in our revised manuscript (Page 10, Line 359 – 363).

A laboratory panel was undertaken, but according to the authors correlations between urinary biomarkers and serum parameters were not analyzed in this study due to the fact that they considered MELD score was a better parameter. This could be added to the discussion section.

 Answer: Thanks for the reviewer’s suggestion. We actually discussed this in our previous publication. We now added this discussion to our revised manuscript (Page 9, Line 312 – 314).

Regarding urea and NO levels authors claim that to confirm that changes are caused by alcohol consumption, it is better to detect their levels in human liver samples, as well as the enzyme activity and expression that is responsible for this reaction.

 Answer: The reviewer is correct. Liver tissue is ideal to confirm the urea and NO changes in the arginine biosynthesis pathway. However, it is hard to collect liver samples since biopsy is an invasive method that could cause several complications. Samples in this project were collected from patients with different stages of ALD, and several patients had passed away. Therefore, the confirmation study will done in newly collected urine samples and possible liver samples from liver biopsy.  

Authors discuss several limitations of the study, such as the explanation of the mechanism(s) of ALD development due to the fact that the current study did not include metabolomics study in patient liver tissue samples. Authors have measured metabolites in serum but results will be published elsewhere. The results from these measurements could support discussion and conclusions.

Answer: The reviewer is right.

After revision, information about samples has been added, minor changes were added and figures have been improved

Answer: Thanks, and we appreciate the reviewer’s constructive comments.