Plasma Metabolomics of Dietary Intake of Protein-Rich Foods and Kidney Disease Progression in Children

Abstract

Objective:

There is inconsistent evidence on the efficacy of a low-protein diet for CKD patients and recommending a low-protein diet for pediatric patients is controversial. There is also a lack of objective biomarkers of dietary intake. The purpose of this study was to identify plasma metabolites associated with dietary intake of protein and to assess whether protein-related metabolites are associated with CKD progression.

Methods:

Non-targeted metabolomics was conducted in plasma samples from 484 Chronic Kidney Disease in Children (CKiD) participants. Multivariable linear regression estimated the cross-sectional association between 949 known, non-drug metabolites and dietary intake of total protein, animal protein, plant protein, chicken, dairy, nuts and beans, red and processed meat, fish, and eggs, adjusting for demographic, clinical, and dietary covariates. Cox proportional hazards models assessed the prospective association between protein-related metabolites and CKD progression defined as the initiation of kidney replacement therapy or 50% eGFR reduction, adjusting for demographic and clinical covariates.

Results:

127 (26%) children experienced CKD progression during 5 years of follow-up. Sixty metabolites were significantly associated with dietary protein intake. Among the 60 metabolites, 10 metabolites were significantly associated with CKD progression (animal protein: n=1, dairy: n=7, red and processed meat: n=2, nuts and beans: n=1), including one amino acid, one cofactor and vitamin, four lipids, two nucleotides, one peptide, and one xenobiotic. 1-(1-enyl-palmitoyl)-2-oleoyl-glycerophosphoethanolamine (GPE, P-16:0/18:1) was positively associated with dietary intake of red and processed meat, and a doubling of its abundance was associated with 88% higher risk of CKD progression. 3-ureidopropionate was inversely associated with dietary intake of red and processed meat, and a doubling of its abundance was associated with 48% lower risk of CKD progression.

Conclusion:

Untargeted plasma metabolomic profiling revealed metabolites associated with dietary intake of protein and CKD progression in a pediatric population.

INTRODUCTION

Nutrition is an important part of chronic kidney disease (CKD) management. A low protein diet is recommended for people with CKD to delay its progression to end-stage kidney disease (ESKD) and the onset of kidney replacement therapy [1]. However, the evidence on the efficacy of a low protein diet for CKD patients is inconsistent [2–6]. Additionally, there is concern that restricting dietary protein intake in pediatric patients will adversely impact growth, unless protein restriction can be implemented with maintenance of adequate energy intake [7].

Dietary intake is typically self-reported in clinical and research settings. However, systematic error introduced by self-reported dietary assessment attenuate measures of association and reduce statistical power [8–10]. Biomarkers can provide a more objective assessment of dietary intake. In order to measure total protein intake, 24-hour urea nitrogen is the gold standard, but it can overestimate lower dietary protein intake levels and underestimate higher protein intake levels [11]. Additionally, it is tedious for participants to collect urine over 24 hours, especially for children [12]. The accuracy of measuring protein intake could be improved by combining conventional instruments and biomarkers. There is a need to find novel blood biomarkers of total protein, as well as specific protein sources, that can more objectively and easily assess dietary protein intake.

Non-targeted metabolomics is useful due to its high-throughput, high-resolution phenotyping characteristics [12]. It has the potential to reveal novel metabolites of dietary intake and metabolites of functional significance in kidney disease, which can be targeted by nutritional interventions.

The objective of this study was to identify plasma metabolites associated with dietary intake of protein among children with CKD and to determine whether protein-related metabolites are associated with CKD progression. We hypothesized that metabolites associated with higher dietary intake of red and processed meat would be associated with a higher risk of CKD progression, while metabolites associated with a higher dietary intake of plant protein, fish, nuts and beans would be associated with a lower risk of CKD progression.

METHODS

Study Population

The Chronic Kidney Disease in Children (CKiD) study is an ongoing, multicenter, prospective cohort study of children with CKD [13, 14]. Participants included in the present investigation were enrolled in CKiD between April 2005 and March 2015 at 54 participating medical centers in the United States and Canada. Children were enrolled if they were between the ages of 6 months and 16 years and had an estimated glomerular filtration rate (eGFR) of 30–90 mL/min per 1.73 m². Written informed consent was obtained from parents or legal guardians, along with assent, when appropriate, from the enrolled children. The CKiD study was approved by the institutional review board of each participating institution.

Out of 960 participants in the CKiD study (from cohort 1 and cohort 2), 673 children had plasma samples at the 6-month follow-up visit for metabolomic profiling. Of these, participants were excluded from the analysis if the CKD progression outcome was not available, covariates were missing, more than 10 food items on the food frequency questionnaire were missing, total energy intake was <500 or >3500 kcal for children <6 years old and <500 or >5000 kcal for older children, and percent estimated energy requirement was not in the range of the 2.5^th and the 97.5^th percentile. A total of 484 participants were available for our analyses (Figure 1).

Figure 1.

Flow diagram of study participants in the CKiD study

Number of participants excluded due to missing baseline covariates: race (n=1), body mass index z-score (n=6), hypertension (n=15), anemia (n=17), serum albumin (n=8), protein-tocreatinine ratio (n=20).

Abbreviations: CKD, chronic kidney disease; CKiD, Chronic Kidney Disease in Children study; EER, estimated energy requirement; FFQ, food frequency questionnaire

Covariate Assessment

Covariates were obtained from the first annual CKiD visit, with a median time since enrollment of 1.2 years, or, when missing, at the 6-month follow-up visit after enrollment. The first annual visit was chosen as the anchoring point for the analysis because the majority of participant data were collected at annual visits and it was the closest annual visit after metabolite measurement. Covariate information included demographic characteristics (age, sex, race), glomerular or non-glomerular etiology, urine protein-to-creatinine ratio, body mass index (BMI) Z-score, anemia, hypertension, eGFR, serum albumin, total energy intake, and dietary intake of fruits, whole grains, and refined grains. Hypertension was defined as systolic or diastolic blood pressure ≥95th percentile for age, sex, and height, or any self-reported antihypertensive medication use. eGFR was calculated using the age- and sex-dependent CKiD under 25 equation [15].

Dietary Assessment

The Child Harvard Service food frequency questionnaire was used to assess dietary intake [16]. This tool was adapted from the self-administered food frequency questionnaire developed and evaluated by Willett et al. [17], redesigned to assess usual eating habits in the prior 28 days, and was validated in Native American and Caucasian children and pregnant women [18, 19]. Three similar age-specific versions (2–5, 6–13 and 14–18 years old) were adopted in the CKiD study, assessing dietary intake of a total of 86 or 87 food items (depending upon age group). The food frequency questionnaire was completed by interview with a study coordinator or self-administered by children or their guardian(s) and collected six months after CKiD study enrollment.

Dietary intake of protein, animal protein, plant protein, and six specific protein-rich food groups were estimated from reported food consumption using the Nutrition Data System for Research (NDSR, version 2013): 1) dairy (milk, cheese, yogurt), 2) nuts and beans (beans, peanut butter), 3) red and processed meat (hot dogs, sausage, hamburger, cold cuts, pork or ham, roast beef or steak, liver or organ, bacon), 4) fish (canned tuna, fried fish, other fish), 5) chicken (chicken nuggets, other chicken), and 6) eggs.

Metabolomic Profiling

Metabolite profiling was performed on plasma samples collected six months after enrollment in the CKiD study. These plasma samples were stored at −80°C until processed and were assayed with an untargeted ultra-high-performance liquid chromatography tandem mass spectrometry–based metabolomics quantification protocol (Metabolon, Inc., Durham, North Carolina) [20–22]. For 35 replicates, 74% of metabolites had a correlation coefficient ≥0.8 and 63% of metabolites had a coefficient of variation <20%. A total of 1,518 metabolites were detected. Samples were excluded if >50% metabolite values were missing. Metabolites were excluded if >80% of the values were missing across all samples (n=45). For the remaining metabolites, missing values were imputed with the minimum value for that metabolite within the sample, and metabolite values were normalized to the run-day medians to correct for variations across runs or instruments. After logarithmic (base 2) transformation, we removed metabolites with low variability (variance <0.01) (n=16) or outliers (greater than 5 standard deviations above or below the mean) (n=21). A total of 1,363 metabolites were available after these exclusions, and 949 known, non-drug compounds were retained for the analysis.

CKD Progression

CKD progression was a composite outcome defined as either initiation of kidney replacement therapy (dialysis or transplantation) or 50% eGFR decline [13]. Time to event was defined as years from the first annual visit after the time of metabolite measurement to CKD progression, death, or the end of the study (March 1, 2018), whichever came first.

Statistical Analysis

Multivariable linear regression was used to estimate the cross-sectional association between dietary intake of protein (exposure) and plasma metabolites (outcome), adjusting for age, sex, race, BMI Z-score, glomerular or non-glomerular diagnosis, eGFR, total energy intake, and dietary intake of fruit, whole grains, and refined grains. In addition, models for each of the 6 protein-rich food groups (dairy, nuts and beans, red and processed meat, fish, chicken, eggs) were adjusted for the other 5 food groups. We adjusted for confounders to improve specificity of each exposure-outcome relationship.

Metabolites that were significantly associated with dietary intake of protein in the cross-sectional analysis were analyzed prospectively with CKD progression. Cox proportional hazards regression models were used to assess the prospective association between metabolites (exposure) and CKD progression (outcome) adjusted for age, sex, race, BMI Z-score, glomerular or non-glomerular diagnosis, eGFR, log₂-transformed urine protein-to-creatinine ratio, anemia, hypertension, and serum albumin. To account for multiple comparisons, we used a false discovery rate <0.05 as the significance threshold using Benjamini and Hochberg approach [23].

RESULTS

Baseline Characteristics

In the overall study sample (n=484), mean age was 12 years, 40% were female, 12% were black, and mean eGFR was 54 mL/min per 1.73 m² (Table 1). A third of the participants had anemia (32%) and a glomerular diagnosis (31%). Half of participants had hypertension. Baseline characteristics were similar for the participants included in our analysis relative to the entire CKiD study population. None of the baseline characteristics were significantly different between participants who had plasma samples available from the 6-month follow-up visit and those who did not.

Table 1.

Baseline characteristics for the overall study population and the original CKiD study participants^*

Characteristics	Overall Study Population (n=484)	Participants with Plasma Samples (n=673)	Original CkiD Study (n=960)
Age, years	12 ± 4	12 ± 4	11 ± 5
Female, n (%)	192 (40)	263 (39)	359 (37)
Black, n (%)	60 (12)	102 (15)	153 (17)
Glomerular diagnosis, n (%)	149 (31)	206 (31)	254 (26)
Anemia, n (%)	155 (32)	224 (35)	283 (31)
Hypertension, n (%)	242 (50)	342 (54)	465 (52)
Body mass index Z-score	0.39 ± 1.14	0.39 ± 1.17	0.39 ± 1.19
eGFR, mL/min per 1.73 m2†	54 ± 21	52 ± 21	52 ± 22
Serum albumin, g/dL	4.35 ± 0.43	4.33 ± 0.43	4.35 ± 0.42
Urine protein-to-creatinine ratio	0.89 ± 1.88	0.93 ± 1.87	0.90 ± 1.74

^*Values for continuous variables are presented as mean ± standard deviation. Values for categorical variables are presented as n (%).

^†eGFR was measured using CKiD under 25 equation (Pierce et al., 2021). Abbreviations: CKiD, Chronic Kidney Disease in Children study; eGFR, estimated glomerular filtration rate.

Sixty unique metabolites were significantly associated with dietary protein intake (total protein: n=2; animal protein: n=2; plant protein: n=1; chicken: n=3; dairy: n=21; nuts and beans: n=13; red and processed meat: n=20; fish: n=2) (Figure 2, Supplemental Table 1). No metabolites were significantly associated with eggs. The most common category of protein-related metabolites was lipids (n=29, 48%), followed by amino acids (n=14, 23%).

Figure 2.

Volcano plot of false discovery rate p-values and β coefficients for the association between plasma metabolites and dietary intake of protein

The horizontal line represents the false discovery rate (FDR) threshold (0.05) for statistical significance, and the vertical line is set at a β coefficient value of 0. Metabolites to the right of the vertical line indicate that they were positively associated with dietary protein intake, and metabolites to the left of the vertical line indicate that they were inversely associated with dietary protein intake.

Abbreviation: GPC, glycerophosphocholine.

Metabolites Related to Dietary Protein-Rich Foods

Among sixty protein-related metabolites, the top five significant metabolites were tryptophan betaine, 3-carboxy-4-methyl-5-propyl-2-furanpropanoate (CMPF), hydroxyl-CMPF, sphingomyelin (d18:1/14:0, d16:1/16:0)*, and 1-lignoceroyl-GPC (24:0) (Figure 2, Supplemental Table 1).

Metabolites related to total protein, animal protein, plant protein, chicken, and fish were all positively associated with these five types of protein (Supplemental Table 1). The majority of dairy-related metabolites (17 out of 21) were positively associated with dairy. The majority of red and processed meat-related metabolites (14 out of 21) were positively associated with red and processed meat. All 13 metabolites, except for one, were positively associated with nuts and beans. Four metabolites were associated with two types of dietary intake of protein (Supplemental Table 1). Argininate was positively associated with both total protein and animal protein. Sphingomyelin (d18:1/14:0, d16:1/16:0) was positively associated with both animal protein and dairy. S-methylcysteine sulfoxide and 4-oxo-retinoic acid were positively associated with nuts and beans and inversely associated with red and processed meat.

Metabolites Related to CKD Progression

Over a median follow-up of 5 years, 127 (26%) children experienced CKD progression. Among the 60 protein-associated metabolites, 10 metabolites were significantly associated with CKD progression (animal protein: n=1, dairy: n=7, red and processed meat: n=2, nuts and beans: n=1) (Figure 3, Table 2). The majority of these metabolites were lipids (n=4, 44%), followed by nucleotides (n=2), amino acid (n=1), cofactors and vitamins (n=1), peptide (n=1), and xenobiotic (n=1).

Figure 3.

Volcano plot of false discovery rate p-values and hazard ratios for the association between dietary protein-related metabolites and chronic kidney disease (CKD) progression

The horizontal line represents the false discovery rate (FDR) threshold (0.05) for statistical significance. The vertical line is set at a hazard ratio of 1.

Abbreviations: FDR, false discovery rate; GPE, glycerophosphoethanolamine; HR, hazard ratio.

Table 2.

Plasma Metabolites Cross-sectionally Associated with Dietary Intake of Protein-Rich foods and Prospectively Associated with CKD Progression in the CKiD Study

			Cross-sectional analysis with dietary intake of protein†		Prospective analysis with CKD progression‡
Compound	Superpathway	Subpathway	β (SE)	FDR p-value	HR (95% CI)	FDR p-value
Metabolites Significantly Associated with Dietary Intake of Animal Protein (n=1)
sphingomyelin (d18:1/14:0, d16:1/16:0)*	Lipid	Sphingomyelins	0.0039 (0.0009)	0.008	2.49 (1.46, 4.26)	0.02
Metabolites Significantly Associated with Dietary Intake of Dairy (n=7)
carotene diol (2)	Cofactors and Vitamins	Vitamin A Metabolism	−0.0271 (0.0064)	0.003	1.50 (1.11, 2.02)	0.048
myristoyl dihydrosphingomyelin (d18:0/14:0)*	Lipid	Dihydrosphingomyelins	0.0236 (0.0050)	0.0006	2.10 (1.32, 3.36)	0.02
sphingomyelin (d18:1/14:0, d16:1/16:0)*	Lipid	Sphingomyelins	0.0235 (0.0039)	2.55 × 10−6	2.49 (1.46, 4.26)	0.02
sphingomyelin (d17:1/16:0, d18:1/15:0, d16:1/17:0)*	Lipid	Sphingomyelins	0.0205 (0.0038)	3.73 × 10−5	2.45 (1.40, 4.29)	0.02
phenylacetylglycine	Peptide	Acetylated Peptides	0.0742 (0.0164)	0.0012	1.22 (1.06, 1.40)	0.04
urate	Nucleotide	Purine Metabolism, (Hypo)Xanthine/Inosine containing	−0.0104 (0.0031)	0.0396	3.25 (1.44, 7.31)	0.04
homostachydrine*	Xenobiotics	Food Component/Plant	0.0334 (0.0097)	0.0374	1.55 (1.24, 1.93)	0.006
Metabolites Significantly Associated with Dietary Intake of Red and Processed Meat (n=2)
1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1)*	Lipid	Plasmalogen	0.0130 (0.0034)	0.0135	1.88 (1.19, 2.99)	0.0470
3-ureidopropionate	Nucleotide	Pyrimidine Metabolism, Uracil containing	−0.0124 (0.0037)	0.0427	0.52 (0.35, 0.77)	0.02
Metabolites Significantly Associated with Dietary Intake of Nuts and Beans (n=1)
tryptophan betaine	Amino Acid	Tryptophan Metabolism	0.1770 (0.0211)	5.58 × 10−13	1.24 (1.10, 1.39)	0.008

^†For the cross-sectional analysis of dietary intake of protein and metabolites, models were adjusted for age, sex, race, body mass index Z-score, glomerular or non-glomerular diagnosis, estimated glomerular filtration rate (calculated using the CKiD under 25 equation), total energy intake, and dietary intake of fruit, whole grains, and refined grains. The model for each of the six protein-rich food was further adjusted for the other five food groups.

^‡For the prospective analysis of metabolites and CKD progression, models were adjusted for age, sex, race, body mass index Z-score, glomerular or non-glomerular diagnosis, estimated glomerular filtration rate (calculated using the CKiD under 25 equation), log₂-transformed protein-to-creatinine ratio, anemia, hypertension, and serum albumin.

^*Asterisk indicates metabolites not officially confirmed (tier 2 identification).

Abbreviations: CKD, chronic kidney disease; CKiD, Chronic Kidney Disease in Children study; CI, confidence interval; FDR, false discovery rate; GPE, glycerophosphoethanolamine; HR, hazard ratio; SE, standard error.

Sphingomyelin (d18:1/14:0, d16:1/16:0) was positively associated with animal protein and dairy and positively associated with CKD progression (Table 2). For the remaining 6 dairy-related metabolites, 2 metabolites were inversely associated with dairy (carotene diol, urate), 4 metabolites were positively associated with dairy [myristoyl dihydrosphingomyelin (d18:0/14:0), sphingomyelin (d17:1/16:0, d18:1/15:0, d16:1/17:0), phenylacetylglycine, homostachydrine], and all metabolites were positively associated with CKD progression. 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1) was positively associated with red and processed meat and CKD progression, while 3-ureidopropionate was inversely associated with red and processed meat and CKD progression. 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1) and 3-ureidopropionate were associated with an 88% higher (HR, 1.88; 95% CI, 1.19–2.99) and 48% lower (HR, 0.52; 95% CI, 0.35–0.77) risk of CKD progression per doubling of metabolite abundance, respectively. Tryptophan betaine was positively associated with nuts and beans and CKD progression.

DISCUSSION

In this study of 484 children with CKD, using untargeted metabolomic profiling, we revealed a total of 60 unique metabolites associated with dietary intake of protein after adjusting for demographic characteristics, clinical factors, eGFR, and dietary factors. We also found that 10 of these unique dietary protein-related metabolites were associated with CKD progression over 5 years of follow-up. Consistent with our hypothesis, sphingomyelin (d18:1/14:0, d16:1/16:0) was positively associated with animal protein and CKD progression, 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1), a plasmalogen, was positively associated with red and processed meat and CKD progression, and 3-ureidopropionate, a uracil-containing compound, was inversely associated with red and processed meat and CKD progression.

Many of our findings are consistent with prior research on known biomarkers of dietary protein intake. Tryptophan betaine is a biomarker of plant protein intake, especially of nuts and beans [24–26]. Homostachydrine is associated with dietary intake of dairy products (milk) [27]. Plasmalogens are associated with red and processed meat consumption [28]. 3-carboxy-4-methyl-5-propyl-2-furanpropanoate (CMPF) is an established biomarker of fish intake [27, 29, 30], and several metabolites of the histidine metabolic pathway were indicators of animal-source protein as demonstrated in previous literature [24, 31]. For example, 3-methylhistidine is a robust biomarker of chicken, red meat, and eggs.

Urate was inversely associated with dairy intake and positively associated with CKD progression based on our analysis. Uric acid is related to renal inflammation, oxidative stress, endothelial dysfunction, and vascular smooth muscle cell proliferation in animals and human studies [32–34]. Higher urate acid levels and increases in urate acid load were associated with more severe CKD progression among children in the CKiD study [35]. A cohort study identified higher serum uric acid level as an independent risk factor for ESKD [36]. In a pooled analysis of the Atherosclerosis Risk in Communities study and the Cardiovascular Health Study, elevated serum uric acid level was associated with higher risk of incident kidney disease [37]. Taken together, urate is a strong risk factor of nephropathy.

Two sphingomyelins and one dihydrosphingomyelin were positively associated with animal protein, dairy, and CKD progression in our study [sphingomyelin (d18:1/14:0, d16:1/16:0), sphingomyelin (d17:1/16:0, d18:1/15:0, d16:1/17:0), myristoyl dihydrosphingomyelin (d18:0/14:0)]. In a prospective study of people with type 1 diabetes, serum level of sphingomyelins was associated with eGFR decline and progression to ESKD [38]. Among individuals with type 1 diabetes, sphingomyelins were strongly associated with macroalbuminuria relative to other serum lipids [39]. Intracellular accumulation of sphingolipids, including sphingomyelins, in glomerular cells is present in glomerular disease [40], and was associated with glomerular proliferation and hypertrophy. Together, these established mechanisms suggest that sphingomyelins may be early indicators of risk of CKD progression.

We found that a plasmalogen, 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1), was positively associated with dietary intake of red and processed meat and positively associated with CKD progression. This finding corroborated a prior study in which plasmalogens were associated with focal segmental glomerular sclerosis among pediatric patients with CKD [41]. Plasmalogens are biomarkers of oxidative stress among people with kidney failure, and higher level of oxidative stress was associated with a higher prevalence of cardiovascular disease [42, 43]. Since cardiovascular disease and CKD are closely related in terms of their pathology and risk factors, plasmalogens may be a marker of CKD progression.

Tryptophan metabolites, such as kynurenine and quinolinic acid, have been positively associated with kidney disease in previous research [44]. Patients with CKD are unable to excrete tryptophan metabolites due to impaired kidney function, resulting in the accumulation of uremic toxins, which can lead to CKD progression [45]. Tryptophan betaine was identified in our analysis as a biomarker of CKD progression. However, tryptophan betaine was positively associated with dietary intake of nuts and beans, which is consistent with prior studies documenting tryptophan betaine as a marker of legumes and plant protein [24–26]. There may be other, non-dietary, contributors to circulating levels of tryptophan metabolites.

3-ureidopropionate, which is involved in pyrimidine metabolism, was inversely associated with red and processed meat and inversely associated with CKD progression in our study. It was previously reported as a potential biomarker of sepsis-associated acute kidney injury among pediatric patients [46]. Carotene diol, a carotenoid involved in vitamin A metabolism, was inversely associated with dairy and positively associated with CKD progression in our study. In contrast, an inverse association between serum carotenoid level and kidney outcomes was previously reported [47, 48]. Phenylacetylglycine, which was related to dairy and CKD progression in our study, was associated with progressive kidney injury in rats [49]. More research is needed to better understand the role of 3-ureidopropionate, carotene diol, and phenylacetylglycine in kidney disease.

Whereas metabolites associated with total protein in our study were not associated with CKD progression, metabolites that were positively associated with animal protein and red and processed meat were positively associated with CKD progression [sphingomyelin (d18:1/14:0, d16:1/16:0), 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1)] and the metabolite that was inversely associated with red and processed meat was inversely associated with CKD progression (3-ureidopropionate). Current guidelines from Kidney Disease Outcomes Quality Initiative (KDOQI) note that there is insufficient evidence to recommend dietary intake of a specific type of protein [1]. Our study suggests that higher intake of animal protein, especially red and processed meat, is related to metabolites that are associated with higher risk of CKD progression. Future research is needed to evaluate the impact of consuming specific sources of protein on CKD progression in children and adults.

Previous metabolomic studies of CKD progression in adults conducted in the Chronic Renal Insufficiency Cohort (CRIC) study identified different biomarkers, but the biomarkers in previous studies were not examined for their relationship with dietary intake of protein [50–52]. Urate was similarly associated with CKD progression in the present study and the CRIC study. To our knowledge, few studies have examined dietary protein-related metabolites in association with CKD progression among adults or children.

The strengths of our study include the use of data and plasma from participants in the CKiD study, which has sufficient sample size, extensive follow-up for the observation of CKD progression, detailed phenotypic data enabling us to adjust for multiple covariates in regression models, and extensive dietary data to investigate protein sources (animal vs. plant) as well as multiple protein-rich foods. We used untargeted plasma metabolomic profiling to maximize the opportunity for discovery of novel biomarkers and accounted for multiple comparisons to reduce false positive findings. Less is known about biomarkers of diet and CKD progression in children relative to adults, so our results contribute important insights to pediatric nephrology.

However, our study has several limitations. Dietary data were self-reported, which is subject to misclassification. Usual intake of protein was assessed at a single point in time, but dietary intake may have changed over time as CKD progresses. Additionally, although we adjusted for many covariates, there may be residual confounding given the observational study design. We were unable to differentiate between low-fat dairy and high-fat dairy products based on the food items that were assessed on the food frequency questionnaire, which may be warranted in future research. Replication of our study findings in independent pediatric study populations is necessary.

In conclusion, untargeted plasma metabolomic profiling identified metabolites associated with dietary protein and CKD progression within a pediatric population. Our findings corroborate previously identified biomarkers of protein intake and CKD progression and provides evidence for novel dietary markers. If replicated, these metabolites may be used as biomarkers to target for preventing CKD progression via dietary interventions.

PRACTICAL APPLICATION.

There are unique metabolic properties of specific sources of protein, and distinct biomarkers for dietary sources of protein. Metabolites representative of animal sources of protein and red and processed meat in particular may be detrimental for kidney health in children.

ABBREVIATIONS

BMI

body mass index

CKiD

Chronic Kidney Disease in Children

CMPF

3-carboxy-4-methyl-5-propyl-2-furanpropanoate

CRIC

Chronic Renal Insufficiency Cohort

eGFR

estimated glomerular filtration rate

ESKD

end-stage kidney disease

GPE

glycerophosphoethanolamine

KDOQI

Kidney Disease Outcomes Quality Initiative