Urine citrate excretion identifies changes in acid retention as eGFR declines in patients with chronic kidney disease

Abstract

Previous studies have shown that acid (H⁺) retention in patients with chronic kidney disease (CKD) but without metabolic acidosis increases as the estimated glomerular filtration rate (eGFR) decreases over time. The present study examined whether changes in urine excretion of the pH-sensitive metabolite citrate predicted changes in H⁺ retention over time in similar patients with CKD that were followed for 10 yr. We randomized 120 CKD2 nondiabetic, hypertension-associated nephropathy patients with plasma total CO₂ of >24 mM to receive 0.5 meq·kg body wt⁻¹·day⁻¹ NaHCO₃ (${\text{HCO}}_3^{-}$; n = 40), 0.5 meq·kg body wt⁻¹·day⁻¹ NaCl (NaCl; n = 40), or usual care (UC; n = 40). We assessed eGFR (CKD-EPI) and H⁺ retention by comparing the observed with expected plasma total CO₂ increase 2 h after an oral NaHCO₃ bolus (0.5 meq/kg body wt). Although 10 yr versus baseline eGFR was lower for each group, 10-yr eGFR was higher (P < 0.01) in ${\text{HCO}}_3^{-}$ (59.6 ± 4.8 ml·min⁻¹·1.73 m⁻²) than NaCl and UC (52.1 ± 5.9 and 52.3 ± 4.1 ml·min⁻¹·1.73 m⁻², respectively) groups. Less eGFR preservation was associated with higher 10-yr versus baseline H⁺ retention in the NaCl group (26.5 ± 13.1 vs. 18.2 ± 15.3 mmol, P < 0.01) and UC group (24.8 ± 11.3 vs. 17.7 ± 10.9 mmol, P < 0.01) and with lower 10-yr versus baseline 8-h urine citrate excretion (U_citrateV) for the NaCl group (162 ± 47 vs. 196 ± 52 mg, respectively, P < 0.01) and UC group (153 ± 41 vs. 186 ± 42 mg, respectively, P < 0.01). Conversely, better eGFR preservation in the ${\text{HCO}}_3^{-}$ group was associated with no differences in 10-yr versus baseline H⁺ retention (14.2 ±13.5 vs. 16.1 ± 15.1 mmol, P = 1.00) or U_citrateV (212 ± 45 vs. 203 ± 49 mg, respectively, P = 0.74). An overall generalized linear model for repeated measures showed that U_citrateV predicted H⁺ retention (P < 0.01). Less eGFR preservation in patients with CKD2 without metabolic acidosis was associated with increased H⁺ retention that was predicted by decreased U_citrateV.

INTRODUCTION

The kidneys are responsible primarily for excreting “fixed” acid (as opposed to “volatile” acid derived from CO₂ excreted by the lungs) from acid (H⁺)-producing diets typical of developed societies (38). Individuals without known kidney disease given dietary acid H⁺ excrete it and eventually achieve steady-state urine H⁺ excretion sufficient to avoid progressive metabolic acidosis (27, 32) but cumulatively excrete less H⁺ than ingested, consistent with H⁺ retention (28, 29). Indeed, patients with reduced estimated glomerular filtration rate (eGFR) who eat the high-H⁺ diets typical of developed societies can have H⁺ retention (19, 20, 41, 47) with the magnitude inversely related to eGFR (19), and this is not necessarily reflected by plasma acid-base parameters characteristic of metabolic acidosis. High-H⁺ diets increase indexes of kidney injury in patients with reduced eGFR but no metabolic acidosis (16) and are associated with increased CKD incidence (4).

Overall, chronic kidney disease (CKD) incidence in the United States has not increased in recent analyses, yet the incidence of prevalent CKD progressing to more advanced stages (40), including to end-stage kidney disease (ESKD) (34), has nevertheless increased, as has its associated death rate (7). The higher ESKD incidence in at-risk United States population groups is due more to faster progression of prevalent CKD to ESKD than to greater CKD incidence (22), emphasizing the importance of identifying modifiable factors for CKD progression to ESKD (34). Because NaHCO₃ slows the eGFR decline in patients with reduced eGFR who are eating high-H⁺ diets but have no metabolic acidosis (33), the underlying H⁺ retention in patients with reduced eGFR might contribute to the subsequent eGFR decline that such patients often experience (3, 33). Epidemiological studies showing that high-H⁺ diets increase serum anion gap (6) and are associated with an increased risk for CKD progression to ESKD (5, 50) support this hypothesis. Therefore, dietary H⁺ reduction that reduces underlying H⁺ retention might limit the progression of prevalent CKD to ESKD.

Dietary H⁺ reduction treatment of underlying H⁺ retention might include Na⁺-based alkali and/or base-producing diets along with their associated challenges. Na⁺ accompanying alkali salts can be problematic for the Na⁺-sensitive comorbidities that often accompany CKD (8), requiring additional or increased doses of natriuretic agents (12). Base-producing diets reduce the underlying H⁺ retention (20), but only a minority of patients can sustain adherence (15). The challenges described for these two available interventions support identification of underlying H⁺ retention in patients with CKD without metabolic acidosis before these largely asymptomatic patients are subjected to these treatments.

Current methods for identifying H⁺ retention in patients with CKD are invasive and cumbersome, and so they are not useful in clinical settings. Recent studies have supported that reduced urine citrate excretion (U_citrateV) can identify patients with CKD with reduced eGFR without metabolic acidosis, yet they have H⁺ retention (20). The present study tested the hypothesis that changes in U_citrateV identify changes in H⁺ retention as eGFR changes over time (19). If this hypothesis tests true, U_citrateV might be a clinically useful biomarker for tracking the risk for nephropathy progression in patients with CKD, thereby revealing their candidacy for potentially kidney-protective dietary H⁺ reduction therapy.

MATERIALS AND METHODS

This longitudinal study examined the effects of oral NaHCO₃ (${\text{HCO}}_3^{-}$group; 0.5 meq·kg body wt⁻¹·day⁻¹) or oral NaCl (NaCl group; 0.5 meq·kg body wt⁻¹·day⁻¹) compared with usual care (UC group), as recommended by guidelines at study initiation (9, 30), on the cystatin C-calculated (CKD-EPI) eGFR (23) over 10 yr as the primary analysis. Secondary analyses included the longitudinal effect(s) of these interventions on H⁺ retention (see below) and the association of changes in H⁺ retention with changes in eGFR. Post hoc, we measured 8-h U_citrateV and assessed the association of changes in U_citrateV with changes in H⁺ retention. Interim analysis at 5 yr showed better eGFR preservation in the ${\text{HCO}}_3^{-}$ group (33), and we advised all patients of this outcome and gave them the opportunity to receive oral NaHCO₃ as described. Thirty-four patients elected to receive NaHCO₃ from years 6−10 of the protocol in substitution for the regimen to which they had been randomized in blinded fashion (UC, NaCl, or NaHCO₃), of whom 14 patients were in the UC group, 9 patients were in the NaCl group, and 11 patients were in the ${\text{HCO}}_3^{-}$ group. Each unblinded patient received NaHCO₃ from years 6−10 but remained in their originally assigned group in intention-to-treat fashion as per the protocol. Nevertheless, the analysis reported censored those 34 participants who received NaHCO₃ in an unblinded fashion.

Figure 1 shows that we identified, through clinic screening (33), 491 macroalbuminuric, hypertensive, nondiabetic CKD2 [eGFR: 60–89 ml·min⁻¹·1.73 m⁻² by the Modification of Diet in Renal Disease Study formula (25)] patients without clinical evidence of glomerulonephritis or evidence of a systemic disease associated with glomerulonephritis, from whom 120 triplicates were matched for age (±2 yr), sex, ethnicity (black vs. Hispanic or white), eGFR (±4 ml·min⁻¹·1.73 m⁻²), and urine albumin (±40 mg/g creatinine). We measured plasma cystatin C and subsequently calculated eGFR with this parameter (23), and eGFR is so reported because earlier reports have supported that cystatin C compared with creatinine estimated eGFR more accurately at the comparatively higher eGFR levels of our recruited patients (39). We chose macroalbuminuric patients because they are at higher risk of eGFR progression despite recommended kidney-protective interventions than those with lower levels of albuminuria (42). This selection enhanced our ability to examine the effect of changes of H⁺ retention on the course of eGFR. Other inclusion criteria were 1) nonmalignant hypertension; 2) at least two primary care physician visits in the preceding year, showing compliance with clinic visits; and 3) age ≥ 18 yr and ability to give consent. Exclusion criteria were 1) primary kidney disease or findings consistent thereof, such as at least 3 red blood cells/high-powered field or urine cellular casts; 2) history of diabetes or fasting blood glucose ≥ 110 mg/dl; 3) history of malignancies, chronic infections, pregnancy, or clinical evidence of cardiovascular disease; 4) peripheral edema or diagnoses associated with edema such as heart/liver failure or nephrotic syndrome; and 5) inability to tolerate angiotensin-converting enzyme inhibition because extant guidelines at the start of the study recommended this therapy for patients with macroalbuminuria (9). None of the participants had a kidney biopsy to exclude other CKD causes. Secondary causes of hypertension such as renal artery stenosis and hyperaldosteronism were excluded clinically. We did not do kidney Doppler or plasma aldosterone-to-renin ratio studies. We recruited most participants from those referred to our academic nephrology clinic for management of hypertension and/or evaluation of reduced kidney function. Participant systolic blood pressure (SBP) was reduced pharmacologically toward a goal of <130 mmHg, as recommended by extant guidelines for those with albuminuria (9). We measured 8-h urine net acid excretion (NAE) from urine titratable acidity (TA), ammonium (NH₄⁺), and HCO₃ ([NH₄⁺] + [TA] – [${\text{HCO}}_3^{-}$]) (47).

Fig. 1.

Outline of protocol to longitudinally assess the association between changes in acid (H⁺) retention and changes in estimated glomerular filtration rate (eGFR) among patients with chronic kidney disease stage 2 (CKD2; eGFR: 60–89 ml·min⁻¹·1.73 m⁻²) undergoing usual care or treatment with oral NaCl or oral NaHCO₃ [meq·kg body wt (bw)⁻¹·day⁻¹].

We assessed H⁺ retention as “unaccounted ${\text{HCO}}_3^{-}$” at baseline and at 10 yr by comparing the difference between the expected and observed change in plasma total CO₂ in response to retained ${\text{HCO}}_3^{-}$ (dose minus excretion) 2 h after an oral bolus of 0.5 meq·kg⁻¹·body wt⁻¹ NaHCO₃ with the following equation: H⁺ retention = [(retained ${\text{HCO}}_3^{-}$/0.5 × body weight) – observed increase in plasma ${\text{HCO}}_3^{-}$] × 0.5 body weightt, assuming 50% body weight space distribution for ${\text{HCO}}_3^{-}$ (2).

We assessed steady-state H⁺ retention by measuring 2-h NAE and venous plasma total CO₂ in the three CKD groups after an oral 0.5 meq/kg lean body wt bolus of NaHCO₃. We told participants to take nothing by mouth after midnight, except for medications before reporting, after which they voided to empty, had venous blood drawn for pH, Pco₂, and total CO₂, and had venous access established. They received the NaHCO₃ bolus at 8 AM and had urine collected for NAE and venous blood drawn for acid-base parameters at 10 AM. They received 8 oz of chilled distilled H₂O, but no other intake, every 2 h to promote urine output. Our assessment of H⁺ retention assumed that higher H⁺ retention would manifest by greater H⁺ titration of the administered ${\text{HCO}}_3^{-}$, as reflected by a smaller observed compared with expected plasma total CO₂ increase (assuming 50% ${\text{HCO}}_3^{-}$ space of distribution) (2) and/or by less urine ${\text{HCO}}_3^{-}$ excretion. This method also assumes similar buffering capacity among CKD stages, an assumption supported by no differences in levels of the major contributors to whole blood buffer base capacity reported in previous studies (47). Because measured parameters did not permit comparison of intracellular buffering capacity, we assumed that this parameter was similar among CKD groups, as done in previous studies (19, 20, 47).

Because patients were recruited before 2006, our study is not registered with clinicaltrials.gov. Our local Institutional Review Board approved the study protocols.

Analytic methods.

We measured plasma and urine creatinine and urine albumin using the Sigma Diagnostics Creatinine Kit (procedure no. 555, Sigma Diagnostics) (36). We measured plasma cystatin C with a particle-enhanced immunonephelometric assay (N Latex cystatin C, Dade Behring, Somerville, NJ) and a nephelometer (BNII, Dade Bering) (14). The IRMA SL Series 2000 blood analysis system (Diametrics, Edison, NJ) measured venous blood pH and Pco₂, and [${\text{HCO}}_3^{-}$] was calculated from these two parameters. Urine and plasma total CO₂ were measured using ultrafluoremetry (43). Urine TA was measured by correction to the ambient blood pH by NaOH addition and NH₄⁺ by the formalin titrametric (to ambient blood pH) method (10). We measured potential renal acid load (PRAL) using a 3-day dairy assessment of the type and amount of foods ingested and scoring them as to their H⁺ or base content as previously described (37) and as previously performed (19, 20, 47). This calculation does not include an estimate of organic acid excretion that combines with PRAL to estimate urine NAE (20); further details regarding this measurement have been reported in an earlier study from our laboratory (20).

Statistical methods.

Characteristics of the sample are described using descriptive statistics. We used frequencies and percentages to describe categorical variables and used means and SDs to describe continuous variables. Group comparisons were made using one-way ANOVA (or a Kruskal-Wallis test when continuous data were not distributed approximately normally). We used a generalized linear model for repeated-measures data to evaluate change in H⁺ retention, as measured by unaccounted ${\text{HCO}}_3^{-}$, from baseline to 10 yr. Tukey-Kramer adjustments were applied for post hoc comparisons between groups and between time points. We used similar models to evaluate changes from baseline for cystatin C-calculated eGFR and plasma total CO₂. We used simple linear regression to assess the effect of the net change in H⁺ retention on the net change in eGFR between baseline and 10 yr, with a statistical significance level set to 0.05. Followed by the intent-to-treat analyses, we carried out sensitivity analyses after excluding the 34 participants who were unblinded and/or changed their randomization status from years 6−10. Similar to the intent-to-treat analyses, the sensitivity analyses used generalized linear models for repeated measures along with Tukey-Kramer adjustments for post hoc comparisons between groups and between time points.

RESULTS

Figure 1 shows the protocol for the 120 patients with CKD2 randomized to the UC (n = 40), NaCl (0.5 meq·kg body wt⁻¹·day⁻¹, n = 40), and ${\text{HCO}}_3^{-}$ (0.5 meq·kg body wt⁻¹·day⁻¹, n = 40) groups and followed with the indicated measurements at baseline and at 10 yr. Table 1 shows that there were no difference in sex, age, body weight (used to calculate H⁺ retention), or body mass index among groups. Table 2 shows there were no differences in SBP among groups at baseline and 10 yr; all three groups had lower than baseline SBP at 10 yr. Table 2 also shows there were no difference in baseline 8-h urine Na⁺ excretion (UNaV) and urine K⁺ excretion (UKV). Ten-year values for UNaV and UKV were higher than baseline in the NaCl and ${\text{HCO}}_3^{-}$ groups but not in the UC group, showing that the two intervention groups ingested the provided NaCl and NaHCO₃. Ten-year UKV was higher than the respective baseline in the ${\text{HCO}}_3^{-}$ group but not in the NaCl or UC group.

Table 1.

General participant characteristics

	Usual Care Group	NaCl Group	${\text{HCO}}_3^{-}$ Group	P Value
Men/women, %	47.5/52.5	47.5/52.5	47.5/52.5	1.0
Ethnicity
Black, %	62.5	62.5	62.5
Hispanic, %	25.0	22.5	17.5	0.97
White, %	12.5	15.0	20.0
Age, yr	51.3 (8.5)	51.5 (8.3)	51.2 (8.2)	0.98
Body weight, kg	84.1 (4.5)	83.8 (5.2)	83.2 (5.4)	0.75
Body mass index	28.8 (2.3)	28.1 (2.2)	28.6 (2.2)	0.81

Values are means (SD) where appropriate; n = 40 participants/group. Patients with chronic kidney disease stage 2 were divided into the following three groups: usual care or treatment with oral NaCl or NaHCO₃ (${\text{HCO}}_3^{-}$).

Table 2.

Baseline and followup values for systolic blood pressure and 8-h urine excretion of Na⁺ and K⁺

	Usual Care Group	NaCl Group	${\text{HCO}}_3^{-}$ Group	P Value
Systolic blood pressure, mmHg
Year 0	153.7 (13.3)	156.1 (15.7)	155.3 (13.7)	0.75*
Year 10	134.1 (8.9)	133.0 (7.6)	135.6 (7.1)	0.25*
P value†	<0.01	<0.01	<0.01
Urine Na+ excretion, mmol/8 h
Year 0	78.4 (5.8)	77.3 (6.0)	75.9 (6.1)	0.82*
Year 10	75.8 (6.0)	87.6 (5.9)	84.3 (6.2)	<0.01*
P value†	0.25	<0.01	<0.01
Urine K+ excretion, mmol/8 h
Year 0	36.8 (5.1)	34.6 (5.3)	35.5 (5.0)	0.85*
Year 10	35.3 (5.2)	36.0 (5.2)	37.8 (5.1)	0.23*
P value†	0.63	0.19	0.039

Values are means (SD); n = 40 participants/group. Patients with chronic kidney disease stage 2 were divided into the following three groups: usual care or treatment with oral NaCl or NaHCO₃ (${\text{HCO}}_3^{-}$). P values at the end of the rows compare values in that row.

^*Kruskal-Wallis test for comparing three groups. All other P values were from one-way ANOVA.

^†P values below year 0 and year 10 values compare values for those years.

Table 3 shows there were no difference in baseline plasma levels of cystatin C or creatinine but higher than baseline values at 10 yr for each parameter in all three groups. Nevertheless, both plasma levels of cystatin C and creatinine were lower in the ${\text{HCO}}_3^{-}$ group than in the remaining groups at 10 yr (P < 0.01). Figure 2 shows that cystatin C-calculated eGFR was not different among groups at baseline but was lower at 10 yr (P < 0.01) than at baseline for the UC (52.3 ± 4.1 vs. 73.9 ± 6.4 ml·min⁻¹·1.73 m⁻²), NaCl (52.1 ± 5.9 vs. 73.7 ± 6.6 ml·min⁻¹·1.73 m⁻²), and ${\text{HCO}}_3^{-}$ (59.6 ± 4.8 vs. 72.9 ± 6.3 ml·min⁻¹·73 m⁻²) groups. Ten-year eGFR was higher in the ${\text{HCO}}_3^{-}$ group than in the NaCl and UC groups (P < 0.01) but was not different between the NaCl and UC groups (P = 0.99).

Table 3.

Baseline and followup values for plasma cystatin C and creatinine

	Usual Care Group	NaCl Group	${\text{HCO}}_3^{-}$ Group	P Value
Plasma cystatin C, mg/dl
Year 0	1.04 (0.09)	1.04 (0.09)	1.05 (0.10)	0.90*
Year 10	1.33 (0.11)	1.34 (0.11)	1.24 (0.10)	<0.02*
P value†	<0.01	<0.01	<0.01
Plasma creatinine, mg/dl
Year 0	0.95 (0.08)	0.96 (0.08)	0.97 (0.07)	0.95*
Year 10	1.23 (0.08)	1.24 (0.08)	1.14 (0.07)	<0.01*
P value†	<0.01	<0.01	<0.01

Values are means (SD); n = 40 participants/group. Patients with chronic kidney disease stage 2 were divided into the following three groups: usual care or treatment with oral NaCl or NaHCO₃ (${\text{HCO}}_3^{-}$). P values at the end of the rows compare values in that row.

^*Kruskal-Wallis test for comparing three groups. All other P values were from one-way ANOVA.

^†P values below year 0 and year 10 values compare values for those years.

Fig. 2.

Dot plots showing cystatin C-calculated estimated glomerular filtration rate (eGFR) for patients with chronic kidney disease stage 2 (CKD2) undergoing usual care (UC group) or additionally treated with oral NaCl (NaCl group; 0.5 meq·kg body wt⁻¹·day⁻¹) or NaHCO₃ (${\text{HCO}}_3^{-}$ group; 0.5 meq·kg body wt⁻¹·day⁻¹) at baseline (light gray circles) and at 10 yr (dark gray circles). Black circles show mean ± SD values for the respective groups. *P < 0.05, year 10 vs. the respective baseline value within that group; +P < 0.05 vs. the respective UC group.

Table 4 shows that there were no differences among groups in PRAL at any time point. Table 5 shows 8-h urine NAE and its components at the three time points. Baseline 8-h NAE, 8-h ${\text{NH}}_4^{+}$ excretion, 8-h TA, and 8-h urine ${\text{HCO}}_3^{-}$ excretion were not different among groups. Five- and ten-year values for 8-h NAE, 8-h NH₄⁺ excretion, and 8-h TA were not different among groups. In contrast, as shown in Table 5, there were higher values of 8-h urine ${\text{HCO}}_3^{-}$ excretion for ${\text{HCO}}_3^{-}$ patients at 5 and 10 yr. Compared with baseline, 5- and 10-yr values for 8-h NH₄⁺ excretion were lower and those for 8-h TA were higher in ${\text{HCO}}_3^{-}$ patients but not in NaCl or UC patients (Table 5). Five- and ten-year 8-h urine ${\text{HCO}}_3^{-}$ excretion were higher than baseline in ${\text{HCO}}_3^{-}$ patients but not in NaCl or UC patients. There was also higher 8-h urine ${\text{HCO}}_3^{-}$ excretion in years 5 and 10 compared with baseline in ${\text{HCO}}_3^{-}$ patients but not in NaCl or UC patients (Table 5). Although baseline plasma total CO₂ was not different among groups at baseline, 5-yr plasma total CO₂ was lower than baseline in the NaCl and UC groups and higher than baseline in the ${\text{HCO}}_3^{-}$ group (Table 6). Ten-year plasma total CO₂ was lower still than the 5-yr value in NaCl and UC groups but was not different in the ${\text{HCO}}_3^{-}$ group compared with plasma total CO₂ at 5 yr, and 10-yr plasma total CO₂ for the ${\text{HCO}}_3^{-}$ group was not different from its respective baseline.

Table 4.

Changes in potential renal acid load during followup

	Usual Care Group	NaCl Group	${\text{HCO}}_3^{-}$ Group	P Value
Year 0	65.7 (17.3)	64.2 (11.7)	63.4 (12.0)	0.78*
Year 5	64.1 (15.3)	62.9 (11.3)	66.4 (8.3)	0.67*
Year 10	61.8 (11.8)	63.4 (12.3)	64.8 (8.3)	0.59*
Difference: year 5 vs. year 0	−1.6 (13.4)	−1.3 (16.7)	3.0 (14.1)	0.67
P value†	0.79	0.64	0.66
Difference: year 10 vs. year 0	−3.9 (13.2)	−0.8 (15.7)	1.4 (13.0)	0.43
P value†	0.26	0.81	0.85
Difference: year 10 vs. year 5	−2.3 (12.5)	0.5 (15.8)	−1.6 (14.1)	0.74
P value†	0.45	0.83	0.75

Values are means (SD) of potential renal acid load (in mmol/day); n = 40 participants/group. Patients with chronic kidney disease stage 2 were divided into the following three groups: usual care or treatment with oral NaCl or NaHCO₃ (${\text{HCO}}_3^{-}$). Mean differences were determined as follows: mean difference = average (SD) of the year 10 value minus baseline, year 5 value minus baseline, and year 10 value minus year 5 value, respectively, for each patient.

^*Kruskal-Wallis test for comparing three groups. All other P values are from one-way ANOVA.

^†Signed-rank test for assessing the within-group difference between baseline and year 5, baseline and year 10, and year 5 and year 10.

Table 5.

Changes in 8-h urine NAE and its components during followup

	8-h Net Acid Excretion				8-h Urine ${\text{NH}}_4^{+}$ Excretion†				8-h Urine Titratable Acid Execretion				8-h Urine ${\text{HCO}}_3^{-}$ Excretion
	Usual caregroup	NaCl group	${\text{HCO}}_3^{-}$ group	P value	Usual caregroup	NaCl group	${\text{HCO}}_3^{-}$ grop	P value	Usual caregroup	NaCl group	${\text{HCO}}_3^{-}$ grop	P value	Usual caregroup	NaCl group	${\text{HCO}}_3^{-}$ group	P value
Year 0	24.8 (5.1)	24.5 (4.3)	25.3 (4.2)	0.37*	14.1 (3.1)	14.0 (2.9)	14.6 (2.9)	0.92*	10.9 (2.5)	10.5 (2.3)	10.8 (2.2)	0.94*	0.2 (0.1)	0.1 (0.1)	0.1 (0.1)	0.77*
Year 5	23.9 (5.3)	24.1 (4.2)	25.4 (3.5)	0.29*	14.3 (3.3)	14.5 (2.8)	13.5 (2.5)	0.07*	9.7 (2.3)	9.6 (2.4)	12.4 (2.5)	0.06*	0.1 (0.2)	0.1 (0.2)	0.6 (0.2)	0.02*
Year 10	23.7 (5.4)	24.0 (4.0)	25.4 (3.2)	0.17*	14.7 (3.2)	14.9 (2.9)	13.3 (2.4)	0.06*	9.2 (2.4)	9.2 (2.4)	12.5 (2.7)	0.06*	0.1 (0.2)	0.1 (0.1)	0.4 (0.2)	0.04*
Difference: year 5 vs. year 0	−0.9 (13.4)	−0.4 (16.7)	0.1 (14.1)	0.06	0.2 (2.6)	0.5 (3.0)	−1.1 (3.2)		−1.2 (2.9)	−0.9 (3.2)	1.6 (4.2)	0.03	−0.1 (1.2)	0.0 (1.1)	0.5 (0.3)	0.02
P value†	0.34	0.51	0.73		0.64	0.61	0.03		0.26	0.35	0.05		0.34	0.89	<0.01
Difference: year 10 vs. year 0	−1.1 (12.5)	−0.5 (15.8)	0.1 (14.1)	0.08	0.6 (3.7)	0.9 (2.9)	−1.3 (3.3)	0.02	−1.7 (3.1)	−1.3 (2.9)	1.7 (4.0)	0.03	−0.1 (1.1)	0.0 (1.2)	0.3 (0.2)	0.03
P value†	0.54	0.71	0.73		0.42	0.5	0.02		0.19	0.23	0.04		0.45	0.92	<0.01
Difference: year 10 vs. year 5	−0.2 (13.2)	−0.1 (15.7)	0.0 (13.0)	0.41	0.4 (3.5)	0.4 (2.5)	−0.2 (2.5)	0.21	−0.5 (2.9)	−0.4 (2.2)	0.1 (2.6)	0.08	0.0 (1.2)	0.0 (1.0)	−0.2 (0.3)	0.18
P value†	0.61	0.72	0.79		0.33	0.71	0.59		0.39	0.47	0.66		0.85	0.87	0.18

Values are means (SD); n = 40 participants/group. Patients with chronic kidney disease stage 2 were divided into the following three groups: usual care or treatment with oral NaCl or NaHCO₃ (${\text{HCO}}_3^{-}$). Mean differences were calculated as follows: mean difference = average (SD) of the year 10 value minus baseline, year 5 value minus baseline, and year 10 value minus year 5 value for each patient.

^*Kruskal-Wallis test for comparing three groups. All other P values were from one-way ANOVA.

^†Signed-rank test for assessing the within-group difference between baseline and year 5, baseline and year 10, and year 5 and year 10.

Table 6.

Changes in plasma total CO₂ during followup

	Usual Care Group	NaCl Group	${\text{HCO}}_3^{-}$ Group	P Value
Year 0	26.1 (0.8)	25.8 (0.5)	25.9 (0.6)	0.42*
Year 5	26.0 (0.7)	25.7 (0.5)	26.2 (0.7)	<0.01
Year 10	25.2 (0.4)	25.1 (0.5)	26.1 (0.7)	<0.01
Difference: year 5 vs.  year 0	−0.1 (0.2)	−0.1 (0.1)	0.3 (0.4)	<0.01
P value†	<0.01	<0.01	<0.01
Difference: year 10 vs.  year 0	−0.9 (0.7)	−0.7 (0.3)	0.2 (0.4)	<0.01
P value†	<0.01	<0.01	<0.01
Difference: year 10 vs.  year 5	−0.8 (0.7)	−0.6 (0.2)	−0.1 (0.2)	<0.01
P value†	<0.01	<0.01	0.20

Values are means (SD) of plasma total CO₂ (in mM); n = 40 participants/group. Patients with chronic kidney disease stage 2 were divided into the following three groups: usual care or treatment with oral NaCl or NaHCO₃ (${\text{HCO}}_3^{-}$). Mean differences were calculated as follows = average (SD) of the year 10 value minus baseline, year 5 value minus baseline, and year 10 value minus year 5 value for each patient.

^*Kruskal-Wallis test for comparing three groups. All other P values were from one-way ANOVA.

^†Signed-rank test for assessing the within-group difference between baseline and year 5, baseline and year 10, and year 5 and year 10.

Table 7 shows venous plasma electrolytes and acid-base parameters, with the reported ${\text{HCO}}_3^{-}$ calculated from measured pH and Pco₂. There were no differences in plasma values for Na⁺, K⁺, or Cl⁻ among groups at baseline or at 10 yr, and 10 yr compared with baseline values for plasma Na⁺ or Cl⁻ was not different. Plasma K⁺ was lower at 10 yr compared with baseline in ${\text{HCO}}_3^{-}$ patients but was not different in NaCl or UC patients. Baseline plasma ${\text{HCO}}_3^{-}$, pH, Pco₂, and plasma anion gap were not different among groups. Ten-year ${\text{HCO}}_3^{-}$, pH, and Pco₂ were all lower than baseline for NaCl and UC patients but were not different for ${\text{HCO}}_3^{-}$ patients. Ten-year compared with baseline plasma anion gap was greater in NaCl and UC patients but was not different for ${\text{HCO}}_3^{-}$ patients.

Table 7.

Baseline and followup plasma electrolytes and acid-base parameters

	Na+, meq/l	K+, meq/l	Cl−, meq/l	${\text{HCO}}_3^{-}$, meq/l	pH	Pco2, mmHg	Anion Gap, meq/l
Usual care group
Year 0	139.3 (1.3)	4.1 (0.3)	103.1 (1.4)	24.9 (0.6)	7.405 (0.019)	40.8 (0.2)	11.3 (0.1)
Year 10	139.7 (1.4)	4.2 (0.3)	103.6 (1.4)	23.9 (0.6)	7.393 (0.020)	40.3 (0.1)	12.2 (0.1)
P value*	0.32	0.79	0.21	<0.01	<0.01	0.04	<0.01
NaCl group
Year 0	139.1 (1.3)	4.1 (0.2)	103.3 (1.4)	24.6 (0.5)	7.403 (0.022)	40.5 (0.1)	11.2 (0.1)
Year 10	139.7 (1.3)	4.1 (0.2)	103.9 (1.4)	23.7 (0.5)	7.391 (0.021)	40.1 (0.1)	12.1 (0.1)
P value*	0.25	0.96	0.22	<0.01	<0.01	0.03	<0.01
${\text{HCO}}_3^{-}$ group
Year 0	139.2 (1.2)	4.1 (0.2)	103.2 (1.4)	24.7 (0.6)	7.404 (0.020)	40.6 (0.1)	11.3 (0.1)
Year 10	139.5 (1.2)	4.0 (0.2)	103.5 (1.4)	24.8 (0.7)	7.405 (0.020)	40.7 (0.1)	11.2 (0.1)
P value*	0.95	0.047	0.78	0.62	<0.01	0.46	0.54

Values are means (SD); n = 40 participants/group. Patients with chronic kidney disease stage 2 were divided into the following three groups: usual care or treatment with oral NaCl or NaHCO₃ (${\text{HCO}}_3^{-}$).

^*P values were from a signed-rank test for assessing the within-group difference between baseline and year 10.

Figure 3A shows that baseline H⁺ retention was not different among groups, and the 10-year compared with baseline value was not different for the ${\text{HCO}}_3^{-}$ group (14.2 ± 13.5 vs. 16.1 ± 15.1 mmol, respectively, P = 1.00). In contrast, 10-yr compared with baseline H⁺ retention was higher in the NaCl group (26.5 ± 13.1 vs. 18.2 ± 15.3 mmol, respectively, P < 0.01) and UC group (24.8 ± 11.3 vs. 17.7 ± 10.9 mmol, P < 0.01). Overall, there was no association between H⁺ retention and eGFR across time for the three treatment groups (P = 0.13).

Fig. 3.

Dot plots showing acid (H⁺) retention in patients with chronic kidney disease stage 2 (CKD2) undergoing usual care (UC group) or treatment with oral NaCl (NaCl group; 0.5 meq·kg body wt⁻¹·day⁻¹) or NaHCO₃ (${\text{HCO}}_3^{-}$ group; 0.5 meq·kg body wt⁻¹·day⁻¹) at baseline (light gray circles) and at 10 yr (dark gray circles). Black circles show mean ± SD values for the respective groups. *P < 0.05, year 10 vs. the respective baseline value within that group; +P < 0.05 vs. the respective UC group.

Figure 3B also shows that baseline U_citrateV was not different among groups and that the 10-yr compared with baseline value, respectively, was not different for the ${\text{HCO}}_3^{-}$ group (212 ± 45 vs. 203 ± 49 mg, respectively, P = 0.74). In contrast, 10-yr compared with baseline U_citrateV was lower in the NaCl group (162 ± 47 vs. 196 ± 52 mg, respectively, P < 0.01) and UC group (153 ± 41 vs.186 ± 42 mg, respectively, P < 0.01). Although U_citrateV at 10 yr was not different between NaCl and UC group (P = 1.00), 10-yr U_citrateV in the ${\text{HCO}}_3^{-}$ group was higher than both NaCl (P < 0.01) and UC (P < 0.01) groups. A generalized linear model for repeated measures, adjusted for time, showed that U_citrateV predicted H⁺ retention overall when the data for all three groups were combined (P < 0.01; visual depiction shown in Fig. 4). When this model was applied to the individual groups, U_citrateV did not predict H⁺ retention for the UC group (P = 0.06) but did so for the NaCl (P < 0.01) and ${\text{HCO}}_3^{-}$ (P < 0.01) groups.

Fig. 4.

Generalized linear model for repeated measures, adjusted for time, for H⁺ retention versus 8-h urine citrate excretion (U_citrateV) at baseline (left) and at 10 yr (right) combining the data for usual care (UC), NaCl (0.5 meq·kg body wt⁻¹·day⁻¹), and NaHCO₃ (${\text{HCO}}_3^{-}$; 0.5 meq·kg body wt⁻¹·day⁻¹) groups.

DISCUSSION

This longitudinal analysis shows that better eGFR preservation with chronic oral NaHCO₃ than oral NaCl or UC in patients with CKD2 with reduced eGFR and no metabolic acidosis was associated with no further increase in H⁺ retention, whereas H⁺ retention increased with NaCl and UC. In addition, unchanged H⁺ retention in the ${\text{HCO}}_3^{-}$ group was associated with no change in U_citrateV, but U_citrateV decreased with increased H⁺ retention in the NaCl and UC groups. Furthermore, decreased U_citrateV predicted increased H⁺ retention over time. Although increased H⁺ retention in the NaCl and UC groups was associated with significant decreases in plasma total CO₂ and plasma anion gap, the quantitative decreases were small, and 10-yr plasma total CO₂ and anion gap remained within normal ranges of clinical laboratories. The data support that unlike plasma total CO₂ or plasma anion gap, the change in U_citrateV (decrease) is quantitatively sufficient for clinicians to identify increasing H⁺ retention as eGFR declines in patients with CKD and reduced eGFR but no metabolic acidosis (19), thereby designating them as candidates for dietary H⁺ reduction to mitigate against further eGFR decline.

The data showing that measurable increases in H⁺ retention in patients with CKD with reduced eGFR but no metabolic acidosis was associated with only minimal decreases in plasma total CO₂ and minimal increases in anion gap challenge clinicians as to how to identify patients with CKD who are potential candidates for dietary H⁺ reduction therapy. Although high-H⁺ diets more likely reduce plasma total CO₂ in patients with CKD with decreased rather than normal eGFR (1), marked changes in dietary H⁺ yield only modest changes in plasma acid-base parameters within normal ranges for clinical laboratories (26). This robust protection of plasma against acid-base challenges shows that clinicians cannot rely upon changes in plasma total CO₂ to assess underlying H⁺ retention or to identify if patients are eating a high-H⁺ diet. The method used to identify H⁺ retention in the present study and previous studies (19, 20, 47) is invasive and cumbersome, and so is without clinical utility. Recent studies have supported that reduced urine excretion of the pH-sensitive metabolite citrate might be a clinically useful, noninvasive way by which to identify underlying H⁺ retention in patients with CKD without metabolic acidosis (20). The present study supports that decreases in U_citrateV predict increases in H⁺ retention in patients with CKD without metabolic acidosis followed over time and might aid clinical decisions as to whether to institute or increase dietary H⁺ reduction therapy for kidney protection.

There is growing evidence that an acid milieu contributes to CKD progression and that its correction with dietary H⁺ reduction is kidney protective. Treatment of CKD-related metabolic acidosis in patients with plasma total CO₂ < 22 mM, as recommended by current guidelines (24), slows the decline rate of creatinine clearance (11) and eGFR (35) in individuals with very low GFR. Correction of metabolic acidosis that is less severe than that for which current guidelines recommend treatment (i.e., plasma total CO₂ = 22–24 mM) and also slows eGFR decline (18). In addition, alkali treatment of individuals with reduced eGFR without metabolic acidosis but eating the high-H⁺ diets of developed societies (38) slows eGFR decline (33). Furthermore, an epidemiological study has shown that high-H⁺ diets are associated with increased CKD incidence in a general population (4). Together, these studies support that the full spectrum of “H⁺ stress” that ranges from high dietary H⁺ with initially normal plasma total CO₂ in individuals with normal eGFR to those with very low eGFR and metabolic acidosis by plasma acid-base parameters is harmful to kidneys and that its amelioration with dietary H⁺ reduction is kidney protective. Kidney-protective interventions such as dietary H⁺ reduction will likely provide greater benefit to the large cadre of patients with CKD with with better preserved eGFR than in those with less preserved eGFR, like the initial patients with CKD2 of the present study, to reduce the increasing incidence of progression of prevalent CKD to its more advanced stages (40).

Modest dietary H⁺ reduction in patients with CKD with reduced eGFR but no metabolic acidosis incompletely corrected underlying H⁺ retention (47) and slowed but did not stop eGFR decline (33). A cross-sectional study has shown that lower eGFR was associated with higher H⁺ retention and showed longitudinally that as eGFR decreased, H⁺ retention increased and was associated with a faster rate of eGFR decline (19). The present study shows that this same modest dietary H⁺ reduction regimen prevented a further increase in H⁺ retention as eGFR declined but did not decrease it. Together, these studies support examining whether more aggressive dietary H⁺ reduction in patients with CKD with initially lower eGFR, and thus possibly greater initial H⁺ retention, more completely reduces H⁺ retention and thereby provides greater kidney protection in patients at increased risk for CKD progression. Dietary H⁺ reduction accomplished by adding base-producing food components decreased H⁺ retention in patients with CKD without metabolic acidosis (20) and improved CKD-related metabolic acidosis (17, 18). These data support a combination of a low-H⁺ diet composed of base-producing food components and Na⁺-based alkali as one strategy by which to enhance dietary H⁺ reduction. Adding base-producing food components might spare the amount of Na⁺-based alkali needed to achieve the acid-base goal given the untoward effects of Na⁺-based alkali (8, 12, 31).

Animal and some patient studies have provide insights as to the mechanisms by which H⁺ retention might promote CKD progression. Animals with reduced GFR without metabolic acidosis eating a H⁺-producing diet have H⁺ retention by microdialysis (45, 46, 48, 49) and high kidney levels of endothelin, angiotensin II, and aldosterone (46, 48, 49). These agents increase nephron acidification in the setting of overall reduced nephron mass (46, 48) but also mediate kidney injury and nephropathy progression (46, 49, 50). Some patients with CKD and reduced eGFR but no metabolic acidosis who eat the high-H⁺ diets of developed societies have H⁺ retention associated with high urine levels of endothelin and aldosterone that decrease in response to dietary H⁺ reduction (47). These data support that increased kidney levels of these three agents mediate the short-term benefit of increased per nephron acidification to maintain overall kidney H⁺ excretion in a setting of reduced nephron mass and high dietary H⁺. On the other hand, these agents appear to mediate long-term progressive nephropathy when dietary H⁺ intake is high.

Present methods for identifying H⁺ retention in patients with CKD (19, 20, 47) are cumbersome and invasive and thus are not clinically practical. A recent study has shown that decreased urine excretion of the pH-sensitive metabolite citrate that is commonly and easily measured in clinical laboratories noninvasively identifies H⁺ retention in patients with CKD with reduced eGFR but no metabolic acidosis (20). The present longitudinal analysis shows that reductions in U_citrateV noninvasively identify increases in H⁺ retention that occur in response to decreasing eGFR over time in patients with CKD but no metabolic acidosis. These data support leveraging the pathophysiological insight that H⁺ retention reduces U_citrateV to identify patients with CKD without metabolic acidosis who during followup develop worsening H⁺ retention that is not manifested by noticeable reductions in plasma total CO₂ or increases in plasma anion gap and might benefit from dietary H⁺ reduction for kidney protection. Furthermore, decreasing U_citrateV in patients receiving dietary H⁺ reduction for kidney protection might indicate a need for more aggressive dietary H⁺ reduction.

Emerging data support an earlier stage in the spectrum of H⁺ accumulation that is insufficient to manifest with changes in plasma acid-base parameters characteristic of metabolic acidosis yet is sufficient to increase urine levels of substances associated with progressive eGFR decline like endothelin and aldosterone (47) and increase urine biomarkers of kidney injury (16, 33). Retaining citrate through proximal tubule reabsorption rather than its urine excretion helps in the setting of H⁺ accumulation because its metabolism consumes H⁺ and yields ${\text{HCO}}_3^{-}$ (21). Small-scale studies have shown that dietary H⁺ reduction with chronic oral NaHCO₃ reduces urine excretion of endothelin, aldosterone, and biomarkers of kidney injury (16, 33). Nevertheless, current guidelines (24) recommend alkali therapy only for patients with CKD at the metabolic acidosis stage of the H⁺ stress spectrum, so patients at this earlier stage of H⁺ retention would not be candidates for such therapy. Larger-scale studies will better determine the utility of dietary H⁺ reduction for kidney protection in patients with CKD with H⁺ retention but without metabolic acidosis. Using U_citrateV rather than the present invasive method to identify and track H⁺ retention in response to interventions would greatly facilitate conduct of such studies.

A spectrum of H⁺ stress, which includes H⁺ retention without metabolic acidosis, questions where the accumulated H⁺ resides before it attains levels that manifest with changes in plasma acid-base parameters characteristic of metabolic acidosis. A dietary H⁺ increment sufficient to increase urine H⁺ excretion but not to decrease plasma total CO₂ in animals with intact nephron mass increased H⁺ addition to microdialysate perfused against the kidney cortical interstitium (44), consistent with increased interstitial fluid H⁺ content. The H⁺ entering the microdialysate might also have derived from plasma buffers and/or intracellular stores. Other studies showing that this dietary protocol increased H⁺ content in skeletal muscle have suggested that H⁺ retention is a systemic phenomenon (45). These data support the interstitial fluid compartment as at least one locus of retained H⁺ in these settings. Other investigators have reported a direct relationship between extracellular fluid volume and interstitial fluid volume and pressure and that each of the latter were higher patients with reduced GFR (13), suggesting that the interstitial fluid compartment reflects systemic status of other kidney-regulated phenomenon. Together, these data reveal that interstitial fluid can reflect systemic Na⁺ and H⁺ status before their respective manifestation in plasma, including pathological excess of each observed with decreased GFR.

Study limitations include our indirect method of assessing H⁺ retention. In addition, the study population was limited to nondiabetic patients with CKD and contained few white patients, so future studies with larger, more diverse sample sizes will determine whether these findings apply to those with CKD to other etiologies and apply equally across ethnic groups.

In summary, this study shows that better eGFR preservation in NaHCO₃-treated patients with CKD2 without metabolic acidosis was associated with no increase in their baseline level of H⁺ retention, but H⁺ retention increased in patients that received eqimolar NaCl or placebo that had less eGFR preservation. In addition, decreasing U_citrateV identified those patients with increasing H⁺ retention during followup. These data support using longitudinal reductions in U_citrateV to identify increasing H⁺ retention in patients with CKD despite not having metabolic acidosis but who might nevertheless benefit from dietary H⁺ reduction or who require more aggressive dietary H⁺ reduction.

GRANTS

This work was supported by funds from the Larry and Jane Woirhaye Memorial Endowment in Renal Research the Texas Tech University Health Sciences Center, by the Statistics Department of Texas A&M University, and by the Academic Operations Division of Baylor Scott and White Health. This work was also supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant R21-DK-11344.

DISCLOSURES

D. E. Wesson has a portion of his salary paid through his employer to serve as a consultant for Tricida (San Francisco, CA). None of the other authors have conflicts to disclose.

AUTHOR CONTRIBUTIONS

N.E.M. and D.E.W. conceived and designed research; N.G. and J.S. performed experiments; J.S., L.N.S., and A.M. analyzed data; N.G. and D.E.W. interpreted results of experiments; L.N.S. prepared figures; D.E.W. drafted manuscript; N.G., N.E.M., and D.E.W. edited and revised manuscript; N.G., A.M., N.E.M., and D.E.W. approved final version of manuscript.