Review Reports

Round 1

Reviewer 1 Report

The study is well presented with all tables and figures. It impresses by it's large sample size and long perspective design. It tries to answer what is the role of diet as a modifable factor, it's  possible impact on kidney function. I found your work valuable, though found at  moments quite difficult to understand how the food groups were  defined, some cultural differences could be seen in the products used, but it's input is quite huge in further understanding of different protein, carbohydrates role in kidney function.   It also gives a broad social picture of the population described.

Author Response

Please see the attachment.

Reviewer 1:

The study is well presented with all tables and figures. It impresses by its large sample size and long perspective design. It tries to answer what is the role of diet as a modifiable factor, it’s possible impact on kidney function. I found your work valuable, though found at moments quite difficult to understand how the food groups were defined, some cultural differences could be seen in the products used, but its input is quite huge in further understanding of different protein, carbohydrates role in kidney function. It also gives a broad social picture of the population described.

Response: We thank the reviewer for his/her positive remarks about our study. The point that it is difficult to understand how the food groups were defined is well taken. We would like to take the opportunity to clarify this. We used FFQ information to assess the habitual intake of 110 food items during the last month in our study population of the Northern Netherlands. An existing validated Dutch FFQ formed the basis for the FFQ used in the Lifelines study. In the present study, food items were manually classified into 50 food groups based on similarities in food and nutrient composition. While we agree that this approach is somewhat subjective, several recent studies used the same approach and based our food groups on these studies [1-2]. We clarified the methodology in the revised manuscript and added these publications as references (see Methods 2.2, page 2).

 

References:

1. Slagter SN, Corpeleijn E, Van Der Klauw MM, Sijtsma A, Swart-Busscher LG, Perenboom CW, De Vries JH, Feskens EJ, Wolffenbuttel BH, Kromhout D, van Vliet-Ostaptchouk JV. Dietary patterns and physical activity in the metabolically (un) healthy obese: the Dutch lifelines cohort study. Nutrition journal. 2018 Dec;17(1):18.
2. Dekker LH, Rijnks RH, Strijker D, Navis GJ. A spatial analysis of dietary patterns in a large representative population in the north of The Netherlands–the Lifelines cohort study. international journal of behavioral nutrition and physical activity. 2017 Dec 1;14(1):166.

 

 

Reviewer 2 Report

Qingqing Cai, et al. demonstrated an interesting relationship between dietary pattern and eGFR decline in the large cohort using the RRR method. This study is important to reveal an effective diet pattern preventing eGFR decline. However, there are several questions that need to be answered.

1. The eGFR decline rate depends on baseline eGFR, therefore (1) it is better to include the baseline eGFR at first visit as a cofounding factor in the multiple regression analysis. (2) The number of population in each group of eGFR according to the CKD stages (≥90, 60-89, 45-59, 30-44, 15-29, 15>) should be shown in the manuscript, and investigation in each population may show us an important suggestion.
2. As mentioned in the Discussion section, the authors could not include the albuminuria in the definition of CKD due to unavailable data at the second visit. Then, was the data of albuminuria or proteinuria available at the first visit? If so, analysis in the population with or without albuminuria/ proteinuria at the first visit may be interesting.
3. Considering the difference between men and women, (1) the authors described in the manuscript, with an increasing eGFR-DP score, the proportion of protein consumption was higher, carbohydrate consumption was lower and fat consumption was higher in women while the opposite was true for all three parameters in men. On the other hand, higher eGFR-DP was associated with lower eGFR decline in both men and women. Why? The contributor of eGFR decline may not be these nutrients such as protein, fat or carbohydrate, but other factors? (2) Is it necessary to consider a different pattern of nutrition guidance in men and women for prevention of CKD or its progression? Please show the authors’ speculation in the discussion section in more detail.

Author Response

Reviewer 2:

Point 1. The eGFR decline rate depends on baseline eGFR, therefore (1) it is better to include the baseline eGFR at first visit as a cofounding factor in the multiple regression analysis. (2) The number of populations in each group of eGFR according to the CKD stages (≥90, 60-89, 45-59, 30-44, 15-29, 15>) should be shown in the manuscript, and investigation in each population may show us an important suggestion.

Response: We kindly thank the reviewer for these comments and suggestions.

The first point brought up by the reviewer has been intensively discussed among the authors on several occasions. Since we constructed the RRR model to find dietary patterns based on the baseline eGFR, we wondered whether it would be appropriate to consider baseline eGFR a confounder (and additionally adjust for it). To accommodate the Reviewer, we have now decided to additionally adjust for baseline eGFR in our multiple regression analyses (model 5 in Tables 3 and 4). This did not materially change the results related to the primary outcome: the eGFR-DP score remained significantly associated with a ≥20% eGFR decline both in women and men (Table 3). Additional adjustment for baseline eGFR did influence the association between the eGFR-based dietary pattern and the secondary outcome, incident CKD, which was no longer statistically significant in women (adjusted hazard ratio 0.93 [95% CI 0.85-1.01]) and men (adjusted hazard ratio 0.96 [95% CI 0.85-1.09]). This may be explained by the fact that baseline eGFR is a very strong determinant of the risk to develop CKD, and there would be limited power for other variables to be additionally associated with this outcome, independent of eGFR. In line, we observed that the association between the Mediterranean Diet Score and incident CKD was also blunted after adjusting for baseline eGFR (Table S7). We made changes to Tables 3, 4, S6 and S7, Figure 1 (splines for incident CKD outcome were not illustrative, so we decided to remove them), and in the text to the Abstract, Results and Discussion (page 11) sections.

 

The second suggestion is related to the numbers of patients according to eGFR stages. It is important to note, and apparently, we have made this insufficiently clear in our manuscript, that all individuals with a baseline eGFR below 60 mL/min/1.73 m² were excluded from the analyses. Therefore, a division as proposed by the reviewer is not possible. We have clarified this exclusion criterion in section 2.1 on page 2 of the revised manuscript.

 

Point 2. As mentioned in the Discussion section, the authors could not include the albuminuria in the definition of CKD due to unavailable data at the second visit. Then, was the data of albuminuria or proteinuria available at the first visit? If so, analysis in the population with or without albuminuria/ proteinuria at the first visit may be interesting.

Responses: Thank you for this suggestion. In the Lifelines cohort, about 39.6% participants have the information of urinary albumin excretion at baseline. We tested the association between eGFR-DP score and outcomes according to baseline micro-albuminuria status (≥30 mg/24h or <30 mg/24h) stratified by sex (Please see Tables X and Y below). In participants without micro-albuminuria, the results were in line with those of the full cohort. However, in the participants with micro-albuminuria, the number of events were too low, resulting in insufficient statistical power to test this hypothesis. We therefore think that these analyses provide insufficient information to warrant inclusion in the manuscript, although we are prepared to include them as supplementary tables at the discretion of the reviewer and editorial team.

 

Point 3. Considering the difference between men and women, (1) the authors described in the manuscript, with an increasing eGFR-DP score, the proportion of protein consumption was higher, carbohydrate consumption was lower and fat consumption was higher in women while the opposite was true for all three parameters in men. On the other hand, higher eGFR-DP was associated with lower eGFR decline in both men and women. Why? The contributor of eGFR decline may not be these nutrients such as protein, fat or carbohydrate, but other factors? (2) Is it necessary to consider a different pattern of nutrition guidance in men and women for prevention of CKD or its progression? Please show the authors’ speculation in the discussion section in more detail.

Response: Thank you for your comments. Regarding the first comment, as you mentioned, we observed small differences in macronutrient consumption over the eGFR-DP quartiles between men and women. The differences in macronutrient consumption might be caused by differences in dietary patterns among men and women (See Table 1; for example, high egg intake is relatively strongly determining the eGFR-DP in women compared with men). Previous studies on diet and chronic diseases often focused on macronutrients or micronutrients. However, food-based dietary patterns take into account complex interactions between food items. As people do not eat isolated nutrients, dietary patterns may be more informative because it addresses the effect of the diet as a whole and it provides a broader picture of food and nutrient intake. Although we found the same associations between dietary pattern score and eGFR decline in men and women, the dietary pattern was more complex than could be discerned based on single macronutrients.

Thank you for your second comments. It is well-established that dietary behaviors between men and women are different, highlighting the need to examine men and women separately in healthy population groups [1]. We derived sex-specific dietary patterns that associated with eGFR decline. Although the derived dietary patterns partly overlapped for men and women (high consumption of cheese and low consumption of white meat) but there are also differences, suggesting that it could be useful to consider different advices for men and women to prevent CKD progression. We further discussed it in the revised manuscript (Discussion section, page 11). 

 

Reference:

[1]. Devlin UM, McNulty BA, Nugent AP, Gibney MJ (2012) The use of cluster analysis to derive dietary patterns: methodological con- siderations, reproducibility, validity and the effect of energy mis- reporting. Proc Nutr Soc 71(4):599–609. https ://doi.org/10.1017/ S002

Table X. Risk of ≥20% eGFR decline according to baseline eGFR-based dietary pattern score in the participants with or without albuminuria at baseline.

Women		Continuous dietary pattern score
		With albuminuria		Without albuminuria
		OR (95% CI)	P	OR (95% CI)	P
	Cases/population	49/521		1116/17844
	eGFR decline ≥20%	9.4%		6.3%
	Model 1	1.04 (0.95-1.14)	0.374	0.98 (0.96-0.99)	0.008
	Model 2	1.05 (0.96-1.15)	0.315	0.98 (0.96-0.99)	0.009
	Model 3	1.02 (0.93-1.13)	0.649	0.98 (0.96-0.99)	0.009
	Model 4	1.02 (0.93-1.13)	0.666	0.98 (0.96-0.99)	0.009
	Model 5	1.03 (0.93-1.13)	0.602	0.97 (0.95-0.99)	0.002
Men		Continuous dietary pattern score
		With albuminuria		Without albuminuria
		OR (95% CI)	P	OR (95% CI)	P
	Cases/population	60/513		607/12125
	eGFR decline ≥20%	11.7%		5.0%
	Model 1	0.96 (0.89-1.04)	0.352	0.97 (0.94-0.99)	0.010
	Model 2	0.96 (0.89-1.04)	0.360	0.97 (0.94-0.99)	0.007
	Model 3	0.97 (0.89-1.06)	0.531	0.97 (0.94-0.99)	0.014
	Model 4	0.97 (0.89-1.06)	0.543	0.97 (0.94-0.99)	0.015
	Model 5	0.98 (0.89-1.07)	0.603	0.97 (0.94-0.99)	0.012

Model 1. Adjusted for age and BSA

Model 2. Model1 plus BMI, waist circumference, cholesterol, triglycerides, diabetes, hypertension and cardiovascular disease

Model 3. Model 2 plus physical activity, smoker, and total energy intake

Model 4. Model 3 plus education and income

Model 5. Model 4 plus baseline eGFR

Albuminuria was defined as urinary albumin excretion >30 mg/24h

 

 

Table Y. Risk of CKD incidence according to baseline eGFR-based dietary pattern score in the subgroup of participants with or without albuminuria at baseline.

Women		Continuous dietary pattern score
		With albuminuria		Without albuminuria
		OR (95% CI)	P	OR (95% CI)	P
	Cases/population	31/521		371/17844
	CKD incidence	6.0%		2.1%
	Model 1	0.97 (0.89-1.09)	0.650	0.98 (0.94-1.01)	0.119
	Model 2	0.97 (0.86-1.09)	0.574	0.97 (0.94-1.01)	0.106
	Model 3	0.92 (0.81-1.05)	0.237	0.97 (0.94-1.01)	0.072
	Model 4	0.93 (0.81-1.06)	0.288	0.97 (0.94-1.01)	0.079
	Model 5	0.90 (0.76-1.07)	0.227	1.01 (0.97-1.05)	0.663
Men		Continuous dietary pattern score
		With albuminuria		Without albuminuria
		OR (95% CI)	P	OR (95% CI)	P
	Cases/population	28/513		181/12125
	CKD incidence	5.5%		1.5%
	Model 1	0.94 (0.83-1.06)	0.347	0.91 (0.87-0.96)	<0.001
	Model 2	0.95 (0.84-1.07)	0.404	0.91 (0.86-0.96)	<0.001
	Model 3	0.97 (0.85-1.10)	0.648	0.93 (0.88-0.98)	0.009
	Model 4	0.96 (0.84-1.10)	0.569	0.93 (0.88-0.98)	0.010
	Model 5	1.01 (0.86-1.20)	0.870	1.01 (0.94-1.06)	0.989

Model 1. Adjusted for age and BSA

Model 2. Model1 plus BMI, waist circumference, cholesterol, triglycerides, diabetes, hypertension and cardiovascular disease

Model 3. Model 2 plus physical activity, smoker, and total energy intake

Model 4. Model 3 plus education and income

Model 5. Model 4 plus baseline eGFR

Albuminuria was defined as urinary albumin excretion >30 mg/24h