Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper covers important clinical issue: the possible impact of the ultrafiltration rate (UFR) on ESA resistance for older people getting regular hemodialysis. It is a relevant subject. Interest for clinicians is growing too because dialysis patients are more older. The research has some biological reasonability and brings some interesting outcomes, but there are various problems in methods and interpretations that need changes before this can be accepted.

Major comments

1. Because this is retrospective, cross-sectional study, it cannot really say what causes what with certainty. Sometimes the authors indicate there might be a direct effect of high UFR on increased ERI; however what they really have is only proof of some connections. The Results and Conclusions parts should more clearly highlight that the findings are just about connections, not cause and effect.
2. Some things that could interfere with results are not dealt well enough. To cover more, more full multivariate tests should be added. This means they should also look at things like how well-fed or frail the patients are, other heart-related illnesses, how much weight changes between dialysis sessions, drops in blood pressure during dialysis and check more things that can show inflammation (since only IL-6 and TNF-α are measured).
3. A big UFR might be just sign of more frail patients, not something that directly makes people resistant to ESA treatment.
4. There needs to be more detail for the definition of a UFR cut-off. The use of 10 mL/h/kg is okay, but there is no testing for if bigger doses have an bigger effects, and methods like linear regression or spline should be checked.
5. Their statement about the hepcidin findings, meaning it falls because ESA is increased, is interesting but only a guess. They need to be more careful with this, as the hepcidin can be moved by many things and there was no before and after dialysis checking.

Minor comments

Several areas in the Discussion are a bit repetitive.

Terms should be used in the same manner everywhere (pick ESRD or ESKD and stay with it).

Add flow diagram for the patients.

Make the talk on confounding factors more clear.

Analysis with UFR as a continuous variable should be made.

Author Response

We thank the reviewer for the comments and suggestions, to which we reply point by point:

The paper covers important clinical issue: the possible impact of the ultrafiltration rate (UFR) on ESA resistance for older people getting regular hemodialysis. It is a relevant subject. Interest for clinicians is growing too because dialysis patients are more older. The research has some biological reasonability and brings some interesting outcomes, but there are various problems in methods and interpretations that need changes before this can be accepted.

Major comments

1. Because this is retrospective, cross-sectional study, it cannot really say what causes what with certainty. Sometimes the authors indicate there might be a direct effect of high UFR on increased ERI; however what they really have is only proof of some connections. The Results and Conclusions parts should more clearly highlight that the findings are just about connections, not cause and effect.

R: We agree with this observation. In the first version of the manuscript, we addressed this issue in the limitations of the study. But we agree that we can turn the text clearer and therefore we revised Discussion and Conclusions sections to reinforce connections, not causality.

2. Some things that could interfere with results are not dealt well enough. To cover more, more full multivariate tests should be added. This means they should also look at things like how well-fed or frail the patients are, other heart-related illnesses, how much weight changes between dialysis sessions, drops in blood pressure during dialysis and check more things that can show inflammation (since only IL-6 and TNF-α are measured).

R: We appreciate this comment. We have perception that these studies are complex and involve multiple variables in complex patients. We did not collect drops in blood pressure and we have not sufficient data to characterize frailty patients. But we measured several parameters that indicate nutrition status (“well-fed”), namely phosphorus, albumin, urea and creatinine. These are standard parameters that are followed by nutritionists on a regular basis. In addition, IL-6 and TNF- were not the only inflammatory parameters that we measured. Albumin and transferrin are negative acute-phase reactants and ferritin is a positive acute-phase reactant. In our database we have other inflammatory variables but are not described as being related to the erythropoiesis response, and we decided not to place them, not to enlarge the manuscript. Finally, the comment about weight change is fully addressed by our variable UF. The UF value that is decided for each hemodialysis session is calculated according to the initial weight and to the target (final) weight, which is defined as dry weight. In this new version we introduced the number of hypertensive patients.

3. A big UFR might be just sign of more frail patients, not something that directly makes people resistant to ESA treatment.

R: We appreciate this comment. We are aware that frailty could influence these results, but, in our opinion, not in that direction. The most common pattern in frail patients is lower intake, leading to less interdialytic weight gain and lower ultrafiltration requirements. And as referred before, there were no statistically significant differences in the levels of albumin, urea, phosphorus and creatinine between the studied groups. Although these biomarkers cannot reliably estimate frailty on their own in hemodialysis patients, they can provide useful clues. We are also aware that BMI was lower in the UFR-H, but this parameter has important limitations when characterizing frailty, especially in older adults or medically complex patients. In addition, after adjusting for BMI, the differences obtained for ERI remained statistically significant. Considering this, we have decided to maintain our original version.

4. There needs to be more detail for the definition of a UFR cut-off. The use of 10 mL/h/kg is okay, but there is no testing for if bigger doses have an bigger effects, and methods like linear regression or spline should be checked.

R: We decided to use a cut-off value (10 mL/kg/h) as this is frequently used in good practice guidelines and our sample size would not advise creating more subgroups. But we followed your suggestion and further evaluated UFR as a continuous variable. In the revised version, we performed multiple regression analysis. To assess the independent relationship of various factors with ERI, multiple regression analysis was conducted following the log transformation of the non-normally distributed variables, through stepwise selection, applying an entry threshold of p < 0.05. We found that UFR/W was an independent predictor of ERI, together with gender, TSAT and circulating levels of IL-6. This data is now provided on a new table and discussed.

5. Their statement about the hepcidin findings, meaning it falls because ESA is increased, is interesting but only a guess. They need to be more careful with this, as the hepcidin can be moved by many things and there was no before and after dialysis checking.

R: We agree with this observation. In the first version, we dedicated a paragraph in the Discussion to the modifiers of hepcidin. In this new version we emphasized that this is just a guess and also that we performed no excretion studies.

Minor comments

Several areas in the Discussion are a bit repetitive.

R: We revised this.

Terms should be used in the same manner everywhere (pick ESRD or ESKD and stay with it).

R: We uniformized the term as end-stage kidney disease (ESKD).

Add flow diagram for the patients.

R: The flow diagram was added.

Make the talk on confounding factors more clear.

R: We revised the text.

Analysis with UFR as a continuous variable should be made.

R: As previously stated, we performed this analysis.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript by Belo et al investigated the association between UFR and EPO response among elderly HD patients in a retrospective observational study. This is an interesting study and provides further evidence how UFR and anemia may be linked in a certain dialysis population. The manuscript is well written and focused but some more details could be helpful to better understand the manuscript.

Here are some specific comments which may help to improve the manuscript:

1) Methods/Results: Information about dialysis treatments should be added, e.g. blood / dialysate flow rates, treatment modality, anticoagulation

2) Line 155/216: “UFR groups were matched for age.” The wording is confusing as the study was not based on a matched-pair design.

3) Results: It is explained how the cut-off of 10ml/h/kg has been selected. However, by investigating two groups reduces the information as compared to a continuous variable. The authors could consider adding correlations / multivariate regression analyses using UFR as continuous variable.

4) Could (residual) confounding be a reason for the missing association between inflammation and UFR?

5) Figure 1: All results of the figure should be shortly mentioned in the text, not just Figure 1A and D.

6) Figure 1: The font size could be increased and the figure is little bit blurred.

Author Response

We thank the reviewer for the comments and suggestions, to which we reply point by point:

The manuscript by Belo et al investigated the association between UFR and EPO response among elderly HD patients in a retrospective observational study. This is an interesting study and provides further evidence how UFR and anemia may be linked in a certain dialysis population. The manuscript is well written and focused but some more details could be helpful to better understand the manuscript.

Here are some specific comments which may help to improve the manuscript:

1) Methods/Results: Information about dialysis treatments should be added, e.g. blood / dialysate flow rates, treatment modality, anticoagulation

R: We provided this data immediately after the initially reported information (first version) about treatment frequency and class dialyzers, in the Results section.

2) Line 155/216: “UFR groups were matched for age.” The wording is confusing as the study was not based on a matched-pair design.

R: We corrected the wording.

3) Results: It is explained how the cut-off of 10ml/h/kg has been selected. However, by investigating two groups reduces the information as compared to a continuous variable. The authors could consider adding correlations / multivariate regression analyses using UFR as continuous variable.

R: We agree with this comment, which was also raised by the other reviewer. In this new version, we enlarged statistical analysis using UFR as a continuous variable and performing multiple regression analysis.

4) Could (residual) confounding be a reason for the missing association between inflammation and UFR?

R: We think that we adjusted for all the potential confounding variables that we analyzed. Nevertheless, in the former version we addressed this missing association in the Discussion section, describing possible reasons.

5) Figure 1: All results of the figure should be shortly mentioned in the text, not just Figure 1A and D.

R: As suggested, we mentioned the results of the figure in the text.

6) Figure 1: The font size could be increased and the figure is little bit blurred.

R: As suggested, we improved Figure 1.