Longitudinal changes in uric acid concentration and their relationship with chronic kidney disease progression in children and adolescents

Abstract

Background

Elevated serum uric acid concentration is a risk factor for CKD progression. Its change over time and association with CKD etiology and concomitant changes in estimated glomerular filtration rate (eGFR) in children and adolescents is unknown.

Methods

Longitudinal study of 153 children/adolescents with glomerular [G] and 540 with non-glomerular [NG] etiology from the CKD in Children study. Baseline serum uric acid, change in uric acid and eGFR over time, CKD etiology, and comorbidities were monitored. Adjusted linear mixed-effects regression models quantified the relationship between within-person changes in uric acid and concurrent within-person changes in eGFR.

Results

Participants with stable uric acid over follow-up had CKD progression which became worse for increased baseline uric acid (average annual percentage changes in eGFR were −1.4%, −7.7%, and −14.7% in those with G CKD with baseline uric acid <5.5 mg/dL, 5.5–7.5 mg/dL, and >7.5 mg/dL, respectively; these changes were −1.4%, −4.1%, and −8.6% in NG CKD). Each 1 mg/dL increase in uric acid over follow-up was independently associated with significant concomitant eGFR decreases of −5.7% (95%CI −8.4 to −3.0%) (G) and −5.1% (95%CI −6.3 to −4.0%) (NG) for those with baseline uric acid <5.5 mg/dL and −4.3% (95%CI −6.8% to −1.6%) (G) and −3.3% (95%CI −4.1% to −2.6%) (NG) with baseline uric acid between 5.5 and 7.5 mg/dL.

Conclusion

Higher uric acid levels and increases in uric acid over time are risk factors for more severe progression of CKD in children and adolescents.

Graphical Abstract

INTRODUCTION

Hyperuricemia is associated with the development and progression of chronic kidney disease (CKD) in adults [1–14] and in children [15]. Whereas several clinical studies have shown that reduction of uric acid levels may slow progression of CKD in adults [16–20], in a large randomized controlled trial in adults with proteinuric stage 3–4 kidney disease, those assigned to a urate lowering treatment with allopurinol showed a similar annual decline in estimated glomerular filtration rate (eGFR) as those receiving a placebo [21]. Moreover, in a study of patients with long-standing type 1 diabetes mellitus and mild-to-moderate CKD [22] urate lowering treatment with allopurinol failed to influence the decline in measured GFR. These adult populations had either long-standing diabetes or proteinuric kidney disease, and so the chronicity of kidney disease could have been irreversible.

It seemed appropriate to examine the effects of hyperuricemia in children in the NIH funded Chronic Kidney disease in Children (CKiD) study, without the confounding effects of other factors such as atherosclerosis, diabetes mellitus, cardiac impairment, or multiple drug prescriptions. In a previous analysis of participants in the CKiD study [15] we showed that initial serum uric acid levels above 7.5 mg/dL were associated with a 38% shorter time to a >30% decline in eGFR or kidney replacement therapy independent of blood pressure, GFR, or proteinuria. In children and adolescents with CKD there is a dearth of data on uric acid levels as a function of age, sex, and CKD etiology as well as within-person changes in uric acid levels over time. The purpose of this longitudinal study was to examine how changes in uric acid levels in children and adolescents with CKD from the CKiD study were associated with CKD etiology, sex, age, and baseline level of uric acid and quantify how within-person changes in uric acid correlate with contemporaneous within-person changes in eGFR.

METHODS

Study population

CKiD is a multicenter, prospective cohort study of children with mild-to-moderate CKD across North America. The study design and conduct were approved by an observational study-monitoring board appointed by the National Institute of Diabetes and Digestive and Kidney Diseases, and by the institutional review boards of each participating center. Participant and parent/legal guardian provided informed consent and assent. The demographic and clinical characteristics of the cohort have been published elsewhere [23]. Beginning in January 2005, study participants were seen at annual follow-up visits and provided data on kidney, cardiovascular, neurocognitive, and growth parameters. Since June 2008 serum uric acid has been part of the laboratory panel obtained at each annual visit. Uric acid was measured using the Roche Cobas enzymatic colorimetric assay (Roche Diagnostics GmbH, D-68305 Mannheim) wherein uricase cleaves uric acid to form allantoin and hydrogen peroxide in the presence of peroxidase resulting in an oxidation of 4-aminophenazone; the red color intensity of the quinone-diimine formed is directly proportional to the uric acid concentration.

Our source population consisted of the 1095 CKiD participants enrolled as of November 2020; 734 had at least two visits with serum uric acid measured, eGFR [24] determined, and covariate data (sex, race [white, non-white], age, time-varying clinical blood pressure (BP) stage [25] [normal, elevated, stage 1 or stage 2 hypertension], urine protein–creatinine ratio [<0.5, 0.5–2.0, >2.0], body mass index (BMI), and diuretic use) at each visit with uric acid measured. Because CKD progression and characteristics related to CKD progression are different in children with glomerular (G) CKD and non-glomerular (NG) CKD [26–28], all analyses presented were done separately by CKD etiology. Also, since hemolytic uremic syndrome (HUS) differs from other non-HUS glomerular diagnoses with respect to CKD progression [29–32], 41 children with HUS were excluded from those with glomerular CKD in all analyses yielding a final study population of 153 participants with non-HUS G CKD (hereafter referred to as G CKD) and 540 participants with NG CKD.

Statistical analyses

Linear mixed models were used to estimate the average yearly change in uric acid level for each additional year of age by G/NG CKD etiology, sex, and baseline uric acid level while accounting for repeated uric acid measurements contributed by the same participant over time. Specifically, for a group of participants defined by their G/NG CKD etiology and sex we fit a model of the form to the data:

$${uricacid}_{ij}={\alpha }_0+{\alpha }_1\cdot I\left({5.5\le uricacid}_{i0}\le 7.5\right)+{\alpha }_2\cdot I\left({uricacid}_{i0}>7.5\right)+{\alpha }_n\cdot z_n+{\beta }_0\cdot {(age}_{ij}-12)+{\beta }_1\cdot {(age}_{ij}-12)\cdot I({5.5\le uricacid}_{i0}\le 7.5)+{\beta }_2\cdot {(age}_{ij}-12)\cdot I({uricacid}_{i0}>7.5)+{\phi }_n\cdot w_n+a_i+b_i\cdot ({age}_{ij}-12)+{\epsilon }_{ij}$$

where uric acid_ij is the uric acid of participant i at time j; α₀ is the average uric acid at 12 years of age for a participant with baseline uric acid <5.5 mg/dL and reference values of all covariates; (α₀ + α₁) and (α₀ + α₂) are the average uric acid levels at 12 years of age for a participant with baseline uric acid between 5.5 and 7.5 mg/dL and >7.5 mg/dL, respectively and reference values of all covariates; z_n is the time-fixed covariate for race; α_n are the parameter estimates showing the between-person differences in uric acid by race; age_ij is the age of participant i at time j; β₀, (β₀ + β₁), and (β₀ + β₂) are the average changes in uric acid for each one-year increment in age for a participant with baseline uric acid <5.5 mg/dL, between 5.5 and 7.5 mg/dL, and >7.5 mg/dL, respectively; w_n are the time-varying covariates of BP stage, urine protein-creatinine ratio, BMI, and diuretic use for participant i at time j; φ_n are parameter estimates for the time-varying covariates; a_i is the random departure of the i^th individual’s baseline uric acid from the overall average baseline uric acid; b_i is the random departure of the i^th individual’s slope from the overall slope; and ε_ij is a N(0, σ²) random variable representing the residual error for the j^th uric acid measurement of the i^th individual. Four separate models were fit to the data; one model for each category of G/NG CKD etiology and sex.

Finally, using linear mixed models we quantified the association between a 1.0 mg/dL increase in uric acid and a contemporaneous change in eGFR by G/NG CKD etiology and baseline uric acid as previously performed in analyses of the association of changes in eGFR with changes in dyslipidemia [33]. Time on study from the first visit where uric acid was measured was used as the time scale in this set of analyses. Specifically, for a group of participants defined by their G/NG CKD etiology and baseline uric acid level (<5.5 mg/dL, 5.5–7.5 mg/dL, or >7.5 mg/dL) we fit a model of the form to the data:

$$\log \left({eGFR}_{ij}\right)={\alpha }_0+{\alpha }_n\cdot z_n+\rho \left({uricacid}_{ij}-{uricacid}_{i0}\right)+\beta \cdot t_{ij}+{\phi }_n\cdot w_n+a_i+b_i\cdot t_{ij}+{\epsilon }_{ij}$$

where log(eGFR_ij) is the log transformed eGFR for participant i at time j; exp(α₀) is the geometric mean baseline eGFR; z_n are the time-fixed covariates (sex, race, and age at the first visit with uric acid available); α_n are parameter estimates showing the between-person differences in log(eGFR) based on the z_n covariates at the baseline visit; (uricacid_ij − uricacid_i0) is the difference in uric acid values for participant i between times j and first visit with uric acid available; 100% ⋅ (exp(ρ) − 1) is the average percentage change in eGFR for a concurrent one-unit increase in uric acid; 100% ⋅ (exp(β) − 1) is the yearly percentage change in eGFR for participants who do not have a change in uric acid; t_ij is the time j following the first visit with uric acid available for participant i; w_n are the time-varying covariates of BP stage, urine protein-creatinine ratio, BMI, and diuretic use for participant i at time j; φ_n are parameter estimates for the time-varying covariates; a_i is the random departure of the i^th individual’s baseline eGFR from the overall geometric mean baseline eGFR; b_i is the random departure of the i^th individual’s slope from the overall slope; and ε_ij is the residual error ~ N(0, σ²). Six separate models were fit to the data; one model for each category of G/NG CKD etiology and baseline uric acid level (<5.5 mg/dL, 5.5–7.5 mg/dL, >7.5 mg/dL). SAS version 9.4 was used to conduct all statistical analyses.

RESULTS

Six hundred and ninety-three CKiD participants, 153 (22%) with G CKD and 540 (78%) with NG CKD, were included in our analyses. Table 1 describes demographic and clinical characteristics at the first visit with uric acid available stratified by CKD etiology and baseline uric acid level. Higher levels of uric acid were cross-sectionally associated with older age and lower eGFR. Among children and adolescents with G CKD, those with uric acid ≥5.5 mg/dL had greater proteinuria. In NG participants, uric acid >7.5 mg/dL was associated with higher levels of proteinuria and BMI; higher levels of uric acid were also associated with increased antihypertensive medication use.

Table 1.

Demographic and clinical characteristics^a at the first visit with uric acid measured stratified by baseline uric acid levels among 153 non-HUS glomerular and 540 non-glomerular Chronic Kidney Disease in Children study participants.

	Glomerularb			Non-Glomerularc
Characteristic	<5.5 mg/dL (n=38)	5.5–7.5 mg/dL (n=77)	>7.5 mg/dL (n=38)	<5.5 mg/dL (n=156)	5.5–7.5 mg/dL (n=260)	>7.5 mg/dL (n=124)
Age, years	13 [11, 16]	15 [12, 17]	16 [14, 17]	6 [5, 11]	11 [6, 14]	14 [11, 16]
Male sex	53%	52%	58%	69%	65%	71%
White Race	53%	51%	42%	67%	68%	74%
eGFRd, ml/min|1.73 m2	80 [69, 93]	65 [50, 81]	49 [40, 64]	64 [49, 75]	47 [36, 60]	42 [34, 51]
Proteinuria, mg/mg
<0.5	74%	39%	39%	71%	73%	53%
0.5 to 2.0	18%	39%	37%	20%	20%	40%
>2.0	8%	22%	24%	9%	7%	7%
BMI, kg/m2	22 [18, 28]	23 [20, 27]	24 [19, 30]	17 [15, 19]	18 [16, 21]	21 [18, 26]
Blood pressuree
Normal BP	74%	61%	55%	63%	63%	65%
Elevated BP	13%	17%	19%	10%	12%	15%
Stage 1 or Stage 2 Hypertensionf	13%	22%	26%	27%	25%	20%
ACEi/ARB use	79%	79%	89%	25%	45%	54%
Diuretic use	8%	10%	16%	4%	4%	6%
Urate lowering therapy useg	32%	27%	18%	5%	3%	13%

^amedian [inter-quartile range] is reported for continuous characteristics and percentage is reported for categorical characteristic

^bfocal segmental glomerulosclerosis (34%), systemic immunological (including SLE) (18%), chronic glomerulonephritis (10%), IgA Nephropathy (Berger’s) (9%), or other (29%)

^caplastic/hypoplastic/dysplastic kidneys (25%), obstructive uropathy (24%), reflux nephropathy (17%), or other (34%)

^destimated glomerular filtration rate (eGFR) calculated from averaging equations provided in Kidney Int 2021; 99:948–956.

^eBlood pressure (BP) categories determined by American Academy of Pediatrics (AAP) 2017 updated BP guidelines.

^fHypertension defined as either Stage 1 or Stage 2 based on the AAP 2017 guidelines.

^gReported use of losartan and/or allopurinol

Figure 1 shows the distribution of uric acid over all person-visits by CKD etiology, sex, and two-year age increments. The 153 G participants contributed 719 person-visits (296 female (Figure 1; panel A) and 423 male (Figure 1; panel B)) over follow-up. Whereas the median level of uric acid remained relatively constant with age among females, the males showed an increasing trend of median uric acid level with age; the percentage of person-visits with uric acid above 7.5 mg/dL also increased with age in males with more than 40% in those ≥14 years of age exceeding that value. The 540 NG participants contributed 2927 person-visits (991 female (Figure 1; panel C) and 1936 male (Figure 1; panel D)) over study follow-up. Both females and males showed increasing median levels of uric acid as well as increasing proportion of uric acid concentrations >7.5 mg/dL with age. However, males had higher levels of uric acid such that more than half of person-visits had a serum uric acid above 7.5 mg/dL in those ≥14 years of age compared to 37% of females.

Figure 1.

Boxplots of uric acid by age grouped into two-year increments across all person-visits with uric acid data available. Panel A: 296 person-visits from 71 females with non-HUS glomerular diagnosis. Panel B: 423 person-visits from 82 males with non-HUS glomerular diagnosis. Panel C: 991 person-visits from 177 females with non-glomerular diagnosis. Panel D: 1936 person-visits from 363 males with non-glomerular diagnosis. The number of person-visits in each age group are shown on the bottom of each panel.

Each participant’s uric acid level by attained age over follow-up is shown by CKD etiology, sex and baseline uric acid level in Figure 2. Uric acid level increased significantly with older age in NG participants and in G boys with baseline uric acid ≤7.5 mg/dL. Table 2 shows the estimated annual changes in uric acid per 1 year in age from the multivariable linear mixed-effects models, stratified by CKD etiology, sex, and baseline uric acid level. In females with G CKD, uric acid levels were constant over time. In males with G CKD, uric acid increased significantly with older age for those with baseline uric acid <5.5 mg/dL (each additional year of age was associated with an average increase in uric acid of 0.26 mg/dL (95% confidence interval (95%CI): 0.16 to 0.37)) and for those with baseline uric acid between 5.5 and 7.5 mg/dL (each additional year of age was associated with an average increase in uric acid of 0.15 mg/dL (95%CI 0.07 to 0.24)). In both male and female children and adolescents with NG CKD, older age was associated with significant increases in uric acid levels for baseline uric acid values ≤7.5 mg/dL.

Figure 2.

Individual plots of uric acid by attained age of follow-up of participants. Panel A: 296 person-visits from 71 females with non-HUS glomerular diagnosis. Panel B: 423 person-visits from 82 males with non-HUS glomerular diagnosis. Panel C: 991 person-visits from 177 females with non-glomerular diagnosis. Panel D: 1936 person-visits from 363 males with non-glomerular diagnosis. Dashed lines correspond to the estimated regression lines obtained from the linear mixed models. The unadjusted estimated average changes in uric acid for each baseline uric acid level are also shown (from the first regression model shown in the statistical methods section, ${\widehat{\beta }}_0$ is the unadjusted estimated average change in uric acid per year of age in those with baseline uric acid <5.5 mg/dL, ${\widehat{\beta }}_0+{\widehat{\beta }}_1$is the estimated average change in uric acid per year of age in those with a baseline uric acid between 5.5 and 7.5 mg/dL, and ${\widehat{\beta }}_0+{\widehat{\beta }}_2$ is the estimated average change in uric acid per year of age in those with a baseline uric acid > 7.5 mg/dL). A star denotes an average change in uric acid per year of age that is significantly different from 0.

Table 2.

Multivariable^a results from linear mixed models of uric acid on attained age over follow-up by CKD etiology, sex, and baseline uric acid category.

	Average uric acid change per 1 year of age
	Estimate (95% Confidence Interval)	p-value
Females with Glomerular CKD (n=71; 296 person-visits)
Baseline uric acid <5.5 mg/dL	0.07 (−0.01, 0.16)	0.087
Baseline uric acid 5.5–7.5 mg/dL	−0.03 (−0.08, 0.03)	0.339
Baseline uric acid >7.5 mg/dL	−0.001 (−0.10, 0.10)	0.978
Males with Glomerular CKD (n=82; 423 person-visits)
Baseline uric acid <5.5 mg/dL	0.26 (0.16, 0.37)	<0.001
Baseline uric acid 5.5–7.5 mg/dL	0.15 (0.07, 0.24)	<0.001
Baseline uric acid >7.5 mg/dL	−0.01 (−0.14, 0.12)	0.891
Females with non-glomerular CKD (n=177; 991 person-visits)
Baseline uric acid <5.5 mg/dL	0.08 (0.04, 0.12)	<0.001
Baseline uric acid 5.5–7.5 mg/dL	0.04 (0.01, 0.07)	0.007
Baseline uric acid >7.5 mg/dL	0.05 (−0.0002, 0.10)	0.051
Males with non-glomerular CKD (n=363; 1936 person-visits)
Baseline uric acid <5.5 mg/dL	0.14 (0.10, 0.18)	<0.001
Baseline uric acid 5.5–7.5 mg/dL	0.11 (0.08, 0.15)	<0.001
Baseline uric acid >7.5 mg/dL	0.01 (−0.04, 0.06)	0.683

^aAdjusted for race and current values of blood pressure stage, urine protein–creatinine ratio, BMI, and diuretic use.

Table 3 presents the results of the multivariable, linear mixed-effects analyses of the association of time and change in uric acid on the annual percentage change in eGFR, stratified by CKD etiology and baseline uric acid level. The parameter estimate for “eGFR percent change per year” shows the average rate of change per year in eGFR among study participants with no change in uric acid over follow-up (the value of $100\%\cdot (\exp (\widehat{\beta })-1)$ from the second model described in the statistical methods). Regardless of CKD etiology, children with stable uric acid over follow-up had annual percentage decreases in eGFR which became worse as baseline uric acid level increased. Specifically, children with baseline uric acid >7.5 mg/dL had annual changes in eGFR of −14.7% (95%CI −18.9 to −10.2%) and −8.6% (95%CI −10.3 to −6.8%) for participants with G and NG CKD, respectively, compared to children with baseline uric acid <5.5 mg/dL whose eGFR changed on average −1.4% (95%CI −3.6 to +0.8%) per year for G and −1.4% (95%CI −2.3 to −0.5%) for NG CKD.

Table 3.

Multivariable ^a results from linear mixed models of eGFR (mL/min|1.73 m²) on both time and change in uric acid by baseline uric acid level among 153 non-HUS glomerular and 540 non-glomerular CKiD participants.

	Baseline uric acid <5.5 mg/dL		Baseline uric acid 5.5–7.5 mg/dL		Baseline uric acid >7.5 mg/dL
	Estimate (95% CI)	p-value	Estimate (95% CI)	p-value	Estimate (95% CI)	p-value
Glomerular CKD	n= 38; 200 person-visits		n= 77; 361 person-visits		n= 38; 158 person-visits
eGFR % change per year b	−1.4% (−3.6%, 0.8%)	0.212	−7.7% (−10.1%, −5.3%)	<0.001	−14.7% (−18.9%, −10.2%)	<0.001
eGFR % change from baseline per 1 mg/dL increase in uric acid	−5.7% (−8.4%, −3.0%)	<0.001	−4.3% (−6.8%, −1.6%)	0.002	−1.6% (−4.0%, 0.9%)	0.211
Non-glomerular CKD	n= 156; 820 person-visits		n= 260; 1469 person-visits		n= 124; 638 person-visits
eGFR % change per year b	−1.4% (−2.3%, −0.5%)	0.002	−4.1% (−4.9%, −3.2%)	<0.001	−8.6% (−10.3%, −6.8%)	<0.001
eGFR % change from baseline per 1 mg/dL increase in uric acid	−5.1% (−6.3%, −4.0%)	<0.001	−3.3% (−4.1%, −2.6%)	<0.001	−0.8% (−1.8%, 0.2%)	0.110

Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate

^aAdjusted for sex, race, age at first uric acid measurement, and current values of blood pressure stage, urine protein–creatinine ratio, BMI, and diuretic use.

^bExpected percent change in eGFR per year among children with no concomitant change in uric acid in the same time period.

Parameter estimates for “eGFR percent change from baseline per 1 mg/dL increase in uric acid” (the value of $100\%\cdot (\exp \left(\widehat{\rho })-1\right)$ from the second model described in the statistical methods) were also statistically significant for both CKD etiology groups for initial uric acid levels ≤7.5 mg/dL, which indicate that increases in uric acid over time were independently associated with significant concomitant decreases in eGFR. Furthermore, as baseline uric acid levels increased, the decreases in eGFR were attenuated indicating that an increase in uric acid over follow-up was most deleterious (in terms of a concomitant decrease in eGFR) for those with lower initial uric acid levels. Specifically, for those with G CKD, each 1 mg/dL increase in uric acid over follow-up was independently associated with concomitant eGFR decreases of −5.7% (95%CI −8.4 to −3.0%), −4.3% (95%CI −6.8 to −1.6%), and −1.6% (95%CI −4.0%, 0.9%) for those with baseline uric acid <5.5 mg/dL, between 5.5–7.5 mg/dL, and >7.5 mg/dL, respectively. For those with NG CKD, each 1 mg/dL increase in uric acid was associated with concomitant eGFR decreases of −5.1% (95%CI −6.3 to −4.0%), −3.3% (95%CI −4.1 to −2.6%), and −0.8% (95%CI −1.8 to 0.2%) for those with baseline uric acid <5.5 mg/dL, between 5.5–7.5 mg/dL, and >7.5 mg/dL, respectively.

The percentage changes in eGFR shown in Table 3 for a given CKD etiology/baseline uric acid group are additive (in the log scale) so that for example, a participant with glomerular CKD and a baseline uric acid level of <5.5 mg/dL with an increase of 1.0 mg/dL in uric acid in the first x years of follow-up would have an expected percentage change in eGFR over the first x years of follow-up ≈ −1.4 x% − 5.7% (e.g., −7.1% decrease in eGFR over one-year, −8.5% decrease in eGFR over two years).

DISCUSSION

Previously, we reported on hyperuricemia and the progression of CKD in children and adolescents in the CKiD study using the first measurement of uric acid as the exposure of interest [15]. We found that the progression to kidney replacement therapy or a >30% decrease in GFR was faster in those with baseline uric acid levels above 5.5 mg/dL. Specifically, those with baseline uric acid of 5.5–7.5 mg/dL and those with uric acid above 7.5 mg/dL progressed 17% and 38%, respectively, faster than those with baseline uric acid levels below 5.5 mg/dL. These findings were adjusted for sex, race, CKD etiology and baseline values of age, BP, eGFR, urine protein–creatinine ratio, and BMI. Our current work corroborates and extends our earlier findings by showing that increased uric acid levels at baseline were an independent risk factor for more severe progression of CKD in children and adolescents and for initial uric acid levels ≥5.5 mg/dL compared to those with NG CKD, the progression was more severe in those with G CKD. More importantly, our current work shows that within-participant increases in uric acid over time were associated with significant concomitant within-participant decreases in eGFR (Table 3) in children and adolescents with an initial uric acid <7.5 mg/dL after adjusting for current values of BP stage, urine protein–creatinine ratio, BMI, and diuretic use.

A traditional analytical approach would simply quantify the relationship between baseline uric acid and longitudinal changes in eGFR after adjustment for baseline factors without (1) addressing the effects that changes in uric acid have on concurrent eGFR changes during follow-up, or (2) adjusting for current values of potential confounding factors. Thus, traditional analyses do not account for changes in the exposure or covariates over follow-up which are anticipated in clinical practice. The analytical approach that was illustrated by Saland et al. [33] and that we used herein allows for the quantification of the relationship between concurrent changes in uric acid and eGFR over time; our data show that the contribution of increases in uric acid to eGFR decline is more pronounced at lower initial values of uric acid (Table 3). This relationship is important as we showed uric acid generally increased over time in our study population (Figure 2 and Table 2), and one would expect such a change in uric acid in a group of children with CKD.

It is likely that serum uric acid levels are not constant throughout the progression of CKD. The spaghetti plots in Figure 2 show the changes in uric acid for each participant stratified by CKD etiology, sex, and initial uric acid and the results shown in Table 2 are from corresponding adjusted models. It is evident that uric acid generally rises with age in males with G diagnosis and baseline uric acid ≤7.5 mg/dL and in both males and females with NG diagnoses with baseline uric acid ≤7.5 mg/dL. Increases in serum uric acid levels have substantial negative effects on eGFR, with the largest effects seen at lower baseline levels of urate (Table 3). Even though our study was not designed to test for the heterogeneity of effects on eGFR change over time by the use of urate lowering treatment (as only 7 (18%) of 38 participants with glomerular CKD and initial uric acid >7.5 mg/dL and 16 (13%) of 124 participants with non-glomerular CKD and initial uric acid >7.5 mg/dL reported using losartan (a urate lowering antihypertensive angiotensin II receptor antagonist) or allopurinol at baseline), in an unadjusted descriptive analysis we found that a 1 mg/dL increase in uric acid was associated with a qualitative difference in % change in eGFR in those not taking urate lowering therapy compared to those taking urate lowering therapy at baseline (−3.1% vs. +0.2% in glomerular participants and −1.5% vs. +2.9% in non-glomerular participants).

Since the uric acid-dependent changes in eGFR are more pronounced at lower levels of initial uric acid concentration, it would seem reasonable to consider early treatment of hyperuricemia to prevent accelerated progression of CKD. Treatment once the uric acid concentration is above 7.5, would appear to be too late to reverse or slow the progression of CKD. While some small studies and reviews have suggested this in adults [34, 35], a number of larger randomized studies [21, 22] and some Cochrane reviews [36] have not confirmed any salutary effects of uric acid lowering on progression of CKD. Perhaps the lowering of uric acid occurred too late in the progression of CKD. This is a reasonable hypothesis since mitigation of short-term exposure to hyperuricemia in 5/6 nephrectomized rats reduced kidney injury [37].

Uric acid levels may be lower in younger children/adolescents because of a maturational increase in the URAT1 protein, which mediates uric acid reabsorption in the proximal tubule. Indeed, the expression of URAT1 is higher in adults and children than infants [38] and a maturational increase is also described in developing rodents [39], which would result in a decrease in fractional excretion of uric acid with age [40–42]. These physiologic changes would result in a maturational increase in serum uric acid concentration, as shown by others previously [42–45] and confirmed for the NG CKiD participants with repetitive annual uric acid measurements (see Figures 1 and 2).

A review of the negative clinical studies suggests it is possible that participants with long-standing type 1 diabetes and mild CKD, as in the PERL trial, may not be as responsive as others with CKD without chronic diabetes, and that longer time trials are necessary. Similarly for the non-diabetic CKD trial (CKDfix) [22], the CKD might have been too long standing to be affected by short-term uric acid lowering. It is also possible that allopurinol treatment might have resulted in the accumulation of nephrotoxic lipid peroxides which could decrease the GFR directly [46], thereby obscuring any salubrious effects of uric acid lowering. Agents that enhance the excretion of uric acid such as verinurad, an inhibitor of URAT1 [47], may be more useful in a randomized control trial. Moreover, studies in children with shorter duration of CKD, no diabetes mellitus, and fewer other co-morbidities, such as in our CKiD population, may show a reversible effect of early uric acid lowering on CKD progression as seen in experimental animals. The CKD may be more reversible in the younger patient, compared to treatment of long-standing kidney disease in adults, but a randomized control study is necessary to establish this.

Limitations of this study include a loss of participants from follow-up because of progression of kidney disease to kidney replacement therapy, which may bias the results towards the null. Also, our study did not have an adequate number of children and adolescents who reported use of urate lowering treatment (e.g. losartan) to formally test whether urate lowering therapy use modified the effect of a 1.0 mg/dL increase in uric acid on the percentage change in eGFR in the group of children and adolescents with initial uric acid >7.5 mg/dL. Future studies designed to directly assess whether urate lowering therapy modifies the association between uric acid changes with concurrent changes in eGFR in those with advanced CKD are warranted. Finally, as with all observational studies, our findings are subject to possible unmeasured confounding.

In all, we show that an increase in serum uric acid is a risk factor for CKD progression in children and adolescents with an initial uric acid <7.5 mg/dL. Both higher baseline uric acid level and increases in uric acid concentration are independent risk factors for more severe progression of CKD in our CKiD population. Safe, early lowering of elevated uric acid levels may indeed slow kidney progression early in the course of pediatric CKD.