Sodium Bicarbonate Treatment and Vascular Function in CKD: A Randomized, Double-Blind, Placebo-Controlled Trial

Abstract

Significance Statement

Lower serum bicarbonate levels, even within the normal range, are strongly linked to risks of cardiovascular disease in CKD, possibly by modifying vascular function. In this randomized, controlled trial, treatment with sodium bicarbonate (NaHCO₃) did not improve vascular endothelial function or reduce arterial stiffness in participants with CKD stage 3b–4 with normal serum bicarbonate levels. In addition, NaHCO₃ treatment did not reduce left ventricular mass index. NaHCO₃ did increase plasma bicarbonate levels and urinary citrate excretion and reduce urinary ammonium excretion, indicating that the intervention was indeed effective. NaHCO₃ therapy was safe with no significant changes in BP, weight, or edema. These results do not support the use of NaHCO₃ for vascular dysfunction in participants with CKD.

Background

Lower serum bicarbonate levels, even within the normal range, are strongly linked to risks of cardiovascular disease in CKD, possibly by modifying vascular function. Prospective interventional trials with sodium bicarbonate (NaHCO₃) are lacking.

Methods

We conducted a randomized, double-blind, placebo-controlled trial examining the effect of NaHCO₃ on vascular function in 109 patients with CKD stage 3b–4 (eGFR 15–44 ml/min per 1.73 m²) with normal serum bicarbonate levels (22–27 mEq/L). Participants were randomized 1:1 to NaHCO₃ or placebo at a dose of 0.5 mEq/lean body weight-kg per day for 12 months. The coprimary end points were change in brachial artery flow-mediated dilation (FMD) and change in aortic pulse wave velocity over 12 months.

Results

Ninety patients completed this study. After 12 months, plasma bicarbonate levels increased significantly in the NaHCO₃ group compared with placebo (mean [SD] difference between groups 1.35±2.1, P = 0.003). NaHCO₃ treatment did not result in a significant improvement in aortic pulse wave velocity from baseline. NaHCO₃ did result in a significant increase in flow-mediated dilation after 1 month; however, this effect disappeared at 6 and 12 months. NaHCO₃ resulted in a significant increase in 24-hour urine citrate and pH and a significant decrease in 24-hour urine ammonia. There was no significant change in left ventricular mass index, ejection fraction, or eGFR with NaHCO₃. NaHCO₃ treatment was safe and well-tolerated with no significant changes in BP, antihypertensive medication, weight, plasma calcium, or potassium levels.

Conclusion

Our results do not support the use of NaHCO₃ for vascular dysfunction in participants with CKD and normal serum bicarbonate levels.

Introduction

CKD is associated with significant cardiovascular morbidity and mortality. Arterial dysfunction begins early in the course of kidney disease and is a key factor responsible for the development of left ventricular hypertrophy in this population.^1,2 Left ventricular hypertrophy is a strong predictor of cardiovascular mortality in CKD.³ Acid retention is a common complication of CKD because the diseased kidney is unable to excrete the daily dietary acid load.⁴ The subsequent acid retention results in unfavorable effects leading to adverse clinical outcomes. Lower serum bicarbonate levels, even within the normal laboratory range, are strongly linked to risks of hypertension,^5–7 cardiovascular disease (CVD),^8–11 CKD progression,^9,12–14 and death.^15,16 Experimental data suggest that acid retention may result in CVD through activation of the complement system, renin angiotensin aldosterone system, and inflammation.

Small interventional trials have shown that treatment with alkali therapy slows the progression of kidney disease, even in patients with normal serum bicarbonate levels.^13,17,18 Although observational data have shown an association between acid retention and CVD, prospective interventional data are lacking. In a small pilot study, sodium bicarbonate (NaHCO₃) significantly improved vascular endothelial function in patients with CKD stage 3–4.¹⁹ Hence, treatment with alkali therapy may represent an inexpensive and novel therapeutic paradigm in CKD. However, no randomized trials have been performed examining the effect of NaHCO₃ therapy on vascular endothelial function or arterial stiffness in patients with CKD.

We conducted a prospective, randomized, double-blind trial to examine the effect of 12 months of NaHCO₃ therapy versus placebo on vascular endothelial function and arterial stiffness in patients with CKD stage 3b–4 (eGFR 15–44 ml/min per 1.73 m²) with normal serum bicarbonate levels (22–27 mEq/L). The primary aim was to determine the effect of NaHCO₃ on brachial artery flow-mediated dilation (vascular endothelial function) and aortic pulse wave velocity (aPWV) (arterial stiffness). In addition, we examined the effect of NaHCO₃ on the left ventricular mass (LVM) measured by cardiac magnetic resonance imaging (MRI). We tested the hypothesis that treatment with NaHCO₃ would improve vascular endothelial function and reduce arterial stiffness compared with placebo.

Methods

Study Population

Participants were recruited from the CKD clinics at the University of Colorado (UCH) between January 2017 and May 2021. Eligibility criteria included age between 18 and 80 years, CKD stage 3b–4 (eGFR 15–44 ml/min per 1.73 m²), serum bicarbonate level 22–27 mEq/L on two separate measurements at least 1 day apart, BP <140/90 mm Hg before randomization, body mass index <40 kg/m², and ability to provide informed consent. Estimated GFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation. All participants were on a stable antihypertensive regimen for a least 1 month before randomization. Women could not be pregnant, breastfeeding, or unwilling to use birth control. Exclusion criteria included use of chronic daily oral alkali within the past 3 months, serum potassium <3.3 or ≥5.5 mEq/L, New York Heart Association Class 3 or 4 heart failure symptoms, known ejection fraction (EF) <30% or hospital admission for heart failure within the past 3 months, chronic use of supplemental oxygen, or significant comorbid conditions that would lead the investigator to conclude that life expectancy was <1 year. The study protocol was approved by the Colorado Multiple Institutional Review Board. All participants provided written informed consent before study entry. All authors declare adherence to the Declaration of Helsinki. This study was registered at ClinicalTrials.gov (NCT02915601).

Study Design

Participants who met eligibility criteria and provided informed consent were randomized using block randomization by age, sex, and CKD stage (3b or 4) to receive NaHCO₃ or placebo. A 1-month run-in period occurred after screening if participants did not meet the BP goal of <140/90 mm Hg. During the run-in phase, an up-titration of a participant's current antihypertensive medications occurred to reach the target of <140/90 mm Hg. The rationale for the run-in phase was to achieve a stable antihypertensive regimen so that changes in antihypertensive medications throughout this study were minimal. The study medication was prepared by Belmar Pharmacy in Lakewood, Colorado. Commercial NaHCO₃ tablets were used, which contained 7.7 mEq bicarbonate, and placebo tablets were manufactured to be identical in size, shape, and color to the commercial NaHCO₃ tablets. Study investigators, coordinators, participants, and data analyzers were blinded to treatment allocation.

Study Drug Dosing

Participants received 0.5 mEq/kg-lean body weight (LBW) per day of study medication for the entire 12 months. Participants took half of the daily dose in the morning and the other half of the daily dose in the evening. The number of capsules was rounded to the nearest whole number. To reduce pill burden and increase compliance, the maximum number of pills per day was six. A dose of 0.5 mEq/kg-LBW per day has been shown to be safe and effective for the treatment of metabolic acidosis in patients with CKD.^17,20 Previous studies using this dose of NaHCO₃ have not reported significant changes in edema, body weight, or BP.^17,18,20

Study End Points

The primary end points were change in brachial artery flow-mediated dilation (FMD) and change in aPWV over 12 months. Prespecified secondary end points included change in LVM and change in 24-hour urine ammonium and citrate.

Brachial Artery FMD

Vascular measurements were performed in the morning after a 12-hour fast (water allowed), 12 hours after abstaining from tobacco and caffeine, and 24 hours after abstaining from alcohol and vigorous exercise. All measurements were performed by trained ultrasonographers at the Kidney Disease Research Center at the UCH Anschutz Medical Campus. Measurements were performed at baseline, 1 month, 6 months, and 12 months. FMD was determined using high-resolution ultrasonography (Toshiba Xario 200) as described originally by Celermajer et al.²¹ and, more recently, by our group.^20,22 Electrocardiogram-gated end-diastolic ultrasound images were acquired during baseline and FMD conditions. For FMD, reactive hyperemia was produced by inflating a pediatric BP cuff around the forearm to 250 mm Hg for 5 minutes, followed by rapid deflation. Brachial artery endothelium independent dilation was determined by measuring brachial artery dilation for 10 minutes after administration of sublingual nitroglycerin (0.4 mg).^22,23 A commercially available software package (Vascular Analysis Tools 6.0 Medical Imaging Applications, LLC, Iowa City, IA) was used to concurrently acquire electrocardiogram-gated brachial artery diameters. FMD was expressed as percentage change from baseline diameter. The FMD images were analyzed by an independent experienced reader who was blinded to treatment assignment.

Aortic Stiffness

aPWV was performed at baseline, 6 months, and 12 months. All measurements were obtained after fasting and abstaining from tobacco, caffeine, alcohol, and exercise, as described earlier. aPWV was measured as described by Seals et al.^24,25 Transcutaneous custom tonometers (Noninvasive hemodynamics Workstation, Cardiovascular Engineering Inc., Norwood, MA) were positioned at the carotid, radial, aorta, and femoral artery with simultaneous electrocardiogram gating of the R-wave to measure the time delay between the foot of the carotid and femoral arterial pressure waves. aPWV was calculated as the distance between the arterial sites (m) divided by the arterial pressure wave transit time (s) at each site.

Cardiac MRI

LVM, EF, and cardiac output were measured using noncontrast cardiac MRI at baseline and 12 months. Cardiac MRI was performed on a 3.0-T imaging system with cardiac sequence package using an electrocardiograph-gated, breath-hold, two-dimensional, cardiac steady-state free precession imaging for assessing left ventricular volumes and function. All measurements were obtained at the UCH Brain Imaging Center, a research-dedicated facility with a 3-T whole-body MRI scanner. Postprocessing to evaluate cardiac function was performed on Syngo using the Siemens Healthineers workstation using an MR cardiac functional analysis package. LVM was indexed to body surface area to calculate the left ventricular mass index (LVMI). All MRI scans were interpreted by an independent radiologist blinded to treatment allocation.

24-Hour Urine Collection

A 24-hour urine collection was performed at baseline, 6 months, and 12 months. Participants were given the collection kits with instructions before each visit. Participants returned their 24-hour urine collection on the date of their visit, and the sample was sent to Litholink for measurement of urinary sodium, ammonia, citrate, and pH. Twenty-four hour urine samples were included in the analysis if the duration of collection was 20–28 hours, and all voids were collected.

Laboratory Measurements

Fasting blood samples were collected at baseline, 6 months, and 12 months. Plasma basic metabolic panels, phosphorus, and albumin were measured at the UCH Clinical Laboratory using standard techniques. Venous blood gases were collected at baseline and 12 months and delivered to the UCH Clinical Laboratory within 1 hour. Random urine albumin and urine creatinine were measured from morning voids at baseline, 6 months, and 12 months.

Patient Safety

Participants had monthly safety visits. Visits were in person at months 1, 3, 6, 9, and 12 and otherwise were performed through telephone. Weight, vital signs, plasma bicarbonate, potassium, and calcium were obtained at each in-person visit. Adverse events were assessed monthly. Pill counts for compliance occurred at months 3, 6, 9, and 12. Set dose reduction and stopping criteria were predefined in the protocol for plasma bicarbonate and potassium levels and for worsening BP or edema (see Supplemental Methods for details on dose reduction and stopping criteria). Rescue therapy with open-label NaHCO₃ was initiated if the serum bicarbonate level was <20 mEq/L on two consecutive measurements at least 1 week apart. Open-label NaHCO₃ was given until the serum bicarbonate level was at least 20 mEq/L. Participants remained on study medication even if open-label NaHCO₃ was prescribed.

Statistical Analyses

The mean and standard deviation (SD) were calculated for continuous variables, and frequency and proportion were calculated for categorical variables. For continuous variables with right-skewed distribution, median and interquartile range (IQR) were calculated, and its natural log value was calculated for further inference analysis. Adverse events were tabulated. Tables and figures were used to present the results. Repeated measures analysis with a mixed-effect model was performed to examine the effect of NaHCO₃ treatment on vascular function and urinary and blood parameters, including data from all time points in one model. All participants with at least one measure of an outcome variable were included in the analysis. The model included the randomization group, months of study, their interaction, and adjustment for covariates, if appropriate. The interaction effect is the key result to assess the treatment effect on an outcome variable. On the basis of the mixed model, the estimation and test of a change within a group or the difference between groups in a change were performed if appropriate. For reference purpose and as an exploratory analysis, the pre-post change from baseline to months 1, 6, and 12 was calculated, and change within a group was tested by a paired t-test and change between groups by a two-group t-test. Prespecified subgroup analyses were also performed. To examine the treatment effect on cardiac parameters, including EF, stroke volume, cardiac output, and LVMI, analysis with the two-group t-test was performed to compare the pre-post change between groups. Only those with data of both pre- and post-treatment measures were included. The two-sided significance level of 0.05 was used in making a conclusion. All data analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC).

Power

Effect sizes were estimated for the primary end points of FMD and aPWV on the basis of published data of a similar population.^26–28 For FMD, assuming a SD of change of 0.8%, 54 patients in the placebo group and 54 patients in the bicarbonate group achieved 99% power to detect a 1.5% difference among the means using the two-sample t-test with a two-sided 0.025 significance level (i.e., 5% significance level after Bonferroni correction for two outcomes) for intent-to-treat analysis and 99% power to detect the same effect size in per-protocol analysis (allowing for 15% attrition). For aPWV, assuming a common SD within a group of 1.5 m/s,^27,28 54 patients in each group will provide 88% power to detect a 1.0 m/s difference among the means using the two sample t-test with a two-sided 0.025 significance level (i.e., 5% significance level after Bonferroni correction for two outcomes) for intent-to-treat analysis and achieve 80% power to detect the same effect size in per-protocol analysis.

Results

Figure 1 shows the consort flow diagram of this study. One hundred nine participants were enrolled in this study, with 54 randomly assigned to placebo and 55 randomly assigned to NaHCO₃. Nineteen participants did not complete this study, 11 in the NaHCO₃ group and eight in the placebo group. Reasons for discontinuation are given in Supplemental Table 2. Compliance with study medication (defined as the percentage of number of pills taken per number of pills provided) was 96.9% in the placebo group and 90.1% in the NaHCO₃ group at 12 months. Baseline characteristics of participants by study group are presented in Table 1. The mean age and eGFR of participants were 62±12 years and 35.9±9.8 ml/min per 1.73 m², respectively. Fifty percent of participants were women; 78.9% of participants were White; and 22% of participants were Hispanic. Plasma bicarbonate levels increased significantly in the treatment group compared with placebo at 12 months (NaHCO₃: mean [SD] 1.11±1.9 mEq/L; placebo: −0.24±2.3 mEq/L; P = 0.003). The median (IQR) and range of bicarbonate levels by treatment group are presented in Supplemental Table 3.

Figure 1.

Patient enrollment, randomization, and completion (CONSORT) diagram.

Table 1.

Baseline characteristics of study participants

Characteristic	Placebo (n=54)	Sodium Bicarbonate (n=55)
Age (yr)	62.6±10.7	60.8±12.5
Female, N (%)	27 (50)	28 (50.9)
Race, N (%)
White	44 (81.5)	42 (76.4)
Black	5 (9.3)	8 (14.5)
Ethnicity, N (%)
Hispanic	13 (24.1)	11 (20.0)
Diabetes, N (%)	21 (38.9)	29 (52.7)
Hypertension, N (%)	51 (94.4)	51 (92.7)
Etiology of kidney disease, N (%)
Diabetes	20 (37.0)	28 (50.9)
Hypertension	9 (16.7)	12 (21.8)
PKD	2 (3.7)	4 (7.3)
Glomerulonephritis	7 (12.9)	7 (12.7)
Drugs/Toxins	3 (5.6)	5 (9.1)
CHF, N (%)	2 (3.7)	6 (10.9)
Smoking status, N (%)
Current	6 (11.1)	2 (3.6)
Former	17 (31.5)	21 (38.2)
Use of ACEi/ARB, N (%)	38 (70.4)	34 (61.8)
Use of diuretic, N (%)	27 (50.0)	28 (50.9)
Body mass index (kg/m2)	30.3±6.0	29.5±5.6
Lean body weight (kg)	54.1±12.0	53.5±11.1
Systolic BP (mm Hg)	120±15.2	121±14.3
Bicarbonate (mEq/L)	23.5±2.3	23.3±2.2
GFR (ml/min per 1.73 m2)	33.9±10.2	35.8±9.4
Urine ACR (mg/g) median (IQR)	86.4 (17.1, 384.4)	33.2 (15.7, 365.6)
Study drug dose at baseline (tablets)	3.5±0.8	3.7±0.7

All values are mean±SD unless otherwise specified. PKD, polycycstic kidney disease; CHF, congestive heart failure; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ACR, albumin-to-creatinine ratio; IQR, interquartile range.

Effect of NaHCO₃ on Vascular Function

At baseline, participants in both groups had vascular endothelial dysfunction because the mean FMD was <4% in each group.²⁹ FMD was similar between the two groups (Figure 2). After 1 month of treatment with NaHCO₃, FMD increased significantly from baseline (3.99±4.8 versus 6.39±7.3 [mean %Δ±SD], P = 0.003). However, there was no statistical difference from placebo in change in FMD at 1 month (P = 0.35, Figure 2). There was no significant change from baseline in either group and no significant difference in FMD between groups at 6 or 12 months (Figure 2). The results were similar when presented as absolute change (Table 2). When examining prespecified subgroups, participants with CKD stage 4 had significantly decreased FMD with NaHCO₃ as compared with placebo at 12 months (Figure 3A). There were no differences among subgroups at month 1 or 6.

Figure 2.

Change in FMD and aPWV with sodium bicarbonate and placebo. aPWV, aortic pulse wave velocity; FMD, flow-mediated dilation.

Table 2.

Changes in vascular function, hemodynamic factors, and weight from baseline to 12 months

Parameter	Placebo			Sodium Bicarbonate			Between Group P Value
	Baseline	12 mo	P Value	Baseline	12 mo	P Value
FMD (%)	3.78±4.3	4.43±3.9	0.32	3.99±4.8	4.25±3.51	0.69	0.68
FMD (mm)	0.13±0.2	0.15±0.1	0.31	0.14±0.2	0.15±0.1	0.75	0.62
Brachial artery dilation to NTG (%)	13.2±8.7	15.8±8.0	0.25	17.0±8.5	14.3±9.1	0.31	0.13
aPWV (cm/s)	994.8±373.5	983.4±347.7	0.60	998.3±447.5	1067.2±457.9	0.26	0.24
crPWV (cm/s)	931.5±151.6	949.7±151.3	0.49	954.3±177.8	987.9±215.8	0.17	0.62
SBP (mm Hg)	119.8±16.0	122.1±16.4	0.33	122.0±14.4	125.6±12.3	0.07	0.55
DBP (mm Hg)	70.9±11.0	70.9±9.8	0.92	70.8±10.6	72.0±11.5	0.93	0.99
Weight (kg)	83.6±17.6	85.2±22.6	0.12	84.4±17.7	85.4±18.0	0.33	0.69

All values are mean (SD) and were calculated using all available data. P values were calculated from mixed-effect models. FMD, brachial artery flow-mediated dilation; NTG, nitroglycerin; aPWV, aortic pulse wave velocity; crPWV, carotid radial pulse wave velocity; SBP, systolic BP; DBP, diastolic BP.

Figure 3.

Changes in FMD and a PWV with sodium bicarbonate and placebo across subgroups. (A) Forest plot showing FMD change in subgroup analyses. (B) Forest plot showing aPWV change in subgroup analyses. aPWV, aortic pulse wave velocity; FMD, flow-mediated dilation.

There was no significant change in aPWV from baseline at 6 or 12 months with NaHCO₃ treatment (Figure 2). There was also no change in carotid-radial pulse wave velocity (a measure of peripheral arterial resistance) from baseline with NaHCO₃ treatment (Table 2). There were no significant differences in aPWV among subgroups (Figure 3B). In addition, in the mixed model, change in FMD and aPWV over time did not depend on baseline bicarbonate levels (FMD P = 0.83; aPWV P = 0.86). Thus, there was no beneficial effect of NaHCO₃ on vascular function even in those with lower bicarbonate levels at baseline.

Effect of NaHCO₃ on Cardiac Parameters

At baseline, EF, stroke volume, cardiac output, and LVMI were similar between the groups (Table 3). After 12 months of treatment with NaHCO₃, there were no significant changes from baseline in EF (median [IQR] change placebo: −1.9 [−5.3, 2.1]; NaHCO₃: −0.7 [−6.7, 3.0]; P = 0.41), stroke volume (median [IQR] change placebo: −1.3 [−9.0, 10.3]; NaHCO₃: −3.6 [−13.7, 5.7]; P = 0.13), cardiac output (median [IQR] change placebo: −0.2 [−0.5, 0.8]; NaHCO₃: −0.2 [−1.1, 0.4]; P = 0.21), or LVMI (median [IQR] change placebo: −1.6 [−9.7, 6.5]; NaHCO₃: 2.7 [−4.1, 6.6]; P = 0.51).

Table 3.

Change from baseline in cardiac parameters according to the treatment group

Parameter	Placebo	Sodium Bicarbonate	P Value
Ejection fraction, %
Baseline	67.1 (60.3, 72.2)	68.1 (55.9, 73.0)
End of study	65.3 (59.1, 69.4)	65.0 (57.8, 73.2)
Change from baseline	−1.9 (−5.3, 2.1)	−0.7 (−6.7, 3.0)	0.41
Stroke volume, ml
Baseline	71.3 (59.4, 82.3)	67.9 (60.5, 78.9)
End of study	72.2 (58.6, 82.4)	68.9 (54.9, 87.4)
Change from baseline	−1.3 (−9.0, 10.3)	−3.6 (−13.7, 5.7)	0.13
Cardiac output, L/min
Baseline	4.7 (4.0, 5.9)	4.7 (4.1, 5.8)
End of study	4.7 (4.2, 5.6)	4.5 (3.4, 5.7)
Change from baseline	−0.2 (−0.5, 0.8)	−0.2 (−1.1, 0.4)	0.21
LVMI, g/m2
Baseline	59.4 (51.4, 76.1)	57.4 (51.5, 67.9)
End of study	67.6 (51.6, 78.3)	57.2 (53.1, 66.8)
Change from baseline	−1.6 (−9.7, 6.5)	2.7 (−4.1, 6.6)	0.51

The median and IQR at baseline and at the end of the study were calculated using all available data. The median (IQR) of the pre-post change were calculated for only those with data both before and after intervention. P values were calculated by the two-group t test in comparison of the pre-post change. LVMI, left ventricular mass index; IQR, interquartile range.

Effect of NaHCO₃ on Urinary and Blood Parameters

After 12 months, there was no change in serum potassium, calcium, phosphate, albumin, or eGFR with NaHCO₃ treatment (Table 4). After 12 months, NaHCO₃ significantly increased 24-hour urinary citrate and urinary pH from baseline, whereas there was no change in the placebo group (Table 4). NaHCO₃ treatment also significantly reduced 24-hour urinary ammonia at 12 months (Table 4). There was no change in 24-hour urine sodium from baseline in either group. Albuminuria increased significantly from baseline to 12 months in the placebo group (mean [SD] percentage change from baseline 9.86%±31.0%; P = 0.04), whereas there was no change in the NaHCO₃ group (mean [SD] percentage change from baseline 4.0%±26.3%; P = 0.60). However, there was no significant difference between the groups in urine albumin to creatinine ratio (ACR) after 12 months (P = 0.31).

Table 4.

Changes in blood and urine parameters from baseline to 12 months

Parameter	Placebo				Sodium Bicarbonate				Between Group P Value
	Baseline	6 mo	12 mo	P Value	Baseline	6 mo	12 mo	P Value
Bicarbonate (mEq/L)	23.4±2.2	22.9±2.3	23.2±2.8	0.51	23.6±2.1	24.3±2.5	24.7±2.5	0.001	0.003
Potassium (mg/dl)	4.1±0.4	4.2±0.5	4.2±0.5	0.46	4.0±0.5	4.1±0.3	4.0±0.4	0.86	0.52
Calcium (mg/dl)	9.4±0.4	9.4±0.4	9.4±0.5	0.34	9.4±0.4	9.3±0.4	9.4±0.4	0.79	0.39
Phosphorus (mg/dl)	3.4±0.5	3.5±0.6	3.4±0.5	0.98	3.5±0.5	3.5±0.6	3.4±0.5	0.95	0.98
eGFR (ml/min per 1.73 m2)	35.4±9.5	34.5±9.2	33.6±8.6	0.08	37.4±8.8	36.2±9.9	36.0±10.7	0.17	0.80
Albumin (g/dl)	4.0±0.4		4.1±0.3	0.06	4.0±0.4		4.1±0.3	0.48	0.07
Venous pH	7.37±0.04		7.37±0.04	0.83	7.38±0.05		7.39±0.04	0.18	0.41
Venous pCO2	40.5±4.6		39.6±5.6	0.25	40.1±4.9		41.3±5.5	0.17	0.07
Urine Parameters
24-h urine citrate (mg/d)	195.2 (80.8, 331.1)	199.1 (70.5, 341.9)	221.4 (124.7, 288.6)	0.94	133.3 (46.2, 383.4)	324.4 (233.6, 546.1)	337.5 (234.5, 566.2)	<0.0001	<0.0001
24-h urine ammonia (mmol/d)	20.2±12.8	22.1±14.4	23.0±21.6	0.17	23.0±14.7	21.3±11.7	15.4±12.3	0.0004	0.0005
24-h urine sodium (mEq/d)	151.7±64.2	147.7±57.8	167.6±57.3	0.37	158.2±103.4	154.2±55.2	178.0±80.4	0.21	0.79
24-h urine pH	5.8±0.5	5.8±0.6	5.9±0.7	0.46	5.7±0.5	6.4±0.7	6.4±0.6	<0.001	0.0001

The mean and standard deviation at baseline and at 12 months of study were calculated using all available data. P values were calculated from mixed-effect models. All values are mean (SD) or median (IQR). IQR, interquartile range,

Adverse Events

Adverse events are presented in Table 5. There were no differences in serious adverse events between the groups. More participants in the NaHCO₃ group (n=13) compared with the placebo group (n=1) had an elevated bicarbonate level of >28 mEq/L during this study. Per protocol, the study drug was reduced by 50%, and participants returned for a follow-up one week later. Only 1 participant continued to have a high bicarbonate level after dose reduction and the study drug was discontinued. There was no change in systolic BP, diastolic BP, or weight with the NaHCO₃ treatment (Table 2). Six patients were started on open-label NaHCO₃ therapy because of a plasma bicarbonate level of <20 mEq/L, five of which were in the placebo group. Open-label NaHCO₃ was continued until the plasma bicarbonate level was ≥20 mEq/L. Participants remained on the study drug while on open-label bicarbonate. There was no significant difference in gastrointestinal symptoms between the two groups (Supplemental Table 1).

Table 5.

Adverse events

Adverse Event	Placebo N=54	Sodium Bicarbonate N=55
Serious adverse events
No. of hospitalizations, N (%)	6 (11.1)	6 (10.9)
Reasons for hospitalizations
Infections	1 (1.9)	1 (1.8)
Volume overload	1 (1.9)	0
MI/stroke	1 (1.9)	1 (1.8)
Acute on chronic renal failure	1 (1.9)	0
Surgery	2 (3.7)	0
Ileus	0	1 (1.8)
High/low glucose levels	0	2 (3.6)
Loss of consciousness	0	1 (1.8)
No. of deaths, N (%)	1 (1.9)	0
Deaths related to the study	0	0
Chronic dialysis initiation, N (%)	1 (1.9)	1 (1.8)
Adverse events
New edema, N (%)	1 (2.2)	1 (2.3)
Increase in diuretic therapy, N (%)	7 (12.9)	8 (14.5)
Increase in antihypertensive therapy, N (%)	12 (22.2)	11 (20.0)
High serum bicarbonate level (>28 mEq/L), N (%)	1 (1.9)	13 (23.6)a
Low serum bicarbonate level (<20 mEq/L), N (%)	8 (14.8)	4 (7.3)

^aP value <0.05.

Discussion

In this randomized, double-blind, placebo-controlled trial, we found that 12 months of NaHCO₃ therapy did not improve vascular endothelial function or reduce arterial stiffness in patients with CKD stage 3b–4 with normal serum bicarbonate levels. Furthermore, treatment with NaHCO₃ did not reduce LVMI. NaHCO₃ did significantly increase urinary citrate and pH and reduce urinary ammonia. To our knowledge, this is the first interventional trial examining the effect of NaHCO₃ versus placebo on surrogate cardiovascular outcomes in patients with CKD.

Acid retention is common in patients with CKD. As kidney function declines, the kidneys progressively lose the ability to synthesize ammonia and excrete hydrogen ions.^4,30 Low serum bicarbonate levels are more common in patients with decreasing kidney function, but over 80% of patients with CKD have normal serum bicarbonate levels.^4,30–32 Studies in patients with CKD have found a positive acid balance despite serum bicarbonate levels within the normal laboratory range.^31,32 This acid retention results in unfavorable effects that can lead to adverse outcomes. Several observational studies have found that lower serum bicarbonate levels, even those in the normal laboratory range, are associated with adverse cardiovascular outcomes.^6–11 Small interventional trials have shown that alkali therapy reduces kidney disease progression even in patients with normal serum bicarbonate levels.^13,17,18 However, these studies were small and not placebo-controlled. We did not find any improvement in vascular function or a reduction in arterial stiffness after NaHCO₃ treatment compared with placebo. In addition, we did not observe any significant change in eGFR or reduction in ACR with NaHCO₃ treatment. Previous studies regarding ACR have been conflicting. Some have shown a reduction in ACR³³ with NaHCO₃ treatment, whereas others found an increase in urine ACR with NaHCO₃ treatment.³⁴ Nonetheless, our results do not support NaHCO₃ treatment for CKD patients with serum bicarbonate levels in the normal laboratory range to improve vascular function or proteinuria.

It is possible that if we would have included participants with low serum bicarbonate levels (<22 mEq/L), our results may have been different. Although patients with CKD may be in a positive acid balance, even with serum bicarbonate levels in the normal range, it may only be those with levels <22 mEq/L that may benefit from treatment. In our previous pilot study of patients with CKD stage 3b–4 with serum bicarbonate levels of 16–22 mEq/L, we did find a significant improvement in FMD with NaHCO₃ treatment compared with control.¹⁹ However, this study was short in duration (only 6 weeks) and was not blinded. In our study, the pattern of how vascular function changed over time did not depend on baseline bicarbonate levels. In addition, a recent randomized placebo-controlled study in adults with CKD and bicarbonate levels <22 mEq/L did not find an improvement in kidney function with 2 years of NaHCO₃ treatment compared with placebo.³⁵ Thus, it is possible that NaHCO₃ treatment does not improve outcomes in patients with CKD even in those with low bicarbonate levels. An interesting finding in our current trial is that FMD did improve significantly with NaHCO₃ after 4 weeks, suggesting that there may be a short-term effect, similar to what we found in our pilot study. However, this effect disappeared at 6 and 12 months. The reason for the short-term improvement in FMD is unclear but should be explored in future studies examining alkali therapy.

Observational data suggest that high serum bicarbonate levels may be associated with worsening heart failure and arterial stiffness.^9,10 The mechanisms behind the increased risk of adverse outcomes with higher serum bicarbonate levels is unknown, but it is hypothesized that it may worsen vascular stiffness and calcification.³⁶ We did not observe any significant increase in arterial stiffness or LVMI or any decrease in EF with NaHCO₃. Thus, 1 year of treatment with NaHCO₃ did not seem to worsen vascular stiffness or heart failure in participants with CKD. However, when examining the primary outcome, FMD, across subgroups, participants with CKD stage 4 had a significant decrease in FMD compared with placebo, without a concomitant increase in arterial stiffness compared with placebo. The reason for this decrease in vascular function is unclear because there were no significant changes in BP, edema, or weight gain in these participants compared with placebo. The subgroup of participants with CKD stage 4 was small (n=11 in the NaHCO₃ group and n=14 in placebo), and thus, it is possible that these results were spurious.

Participants receiving NaHCO₃ did have a significant increase in plasma bicarbonate levels, a significant decrease in urinary ammonium excretion, and a significant increase in urinary citrate excretion compared with placebo, indicating that participants took the study medication. The use of NaHCO₃ was safe. There is a potential concern with long-term supplementation with NaHCO₃ and volume overload in participants with CKD, but in our trial, NaHCO₃ did not result in worsening BP, edema, or weight gain. There was no change in LVMI or EF. Although there was an increase in urine sodium excretion in the NaHCO₃ group, it did not reach statistical significance. We did not collect dietary data on participants, so it is possible that participants in the NaHCO₃ group had lower dietary sodium intake. There was no difference in adverse events between the two groups, including no difference in hospitalizations for fluid overload. More patients on NaHCO₃ treatment had plasma bicarbonate levels >28 mEq/L, but only 1 participant had to discontinue the drug because of bicarbonate levels remaining >28 mEq/L after a 50% reduction in the dose. There was no difference in plasma potassium levels between groups, and there were no significant hypokalemia events (plasma potassium<3 mEq/L). NaHCO₃ treatment did not result in worsening gastrointestinal symptoms compared with placebo. Compliance with study medication was over 90% in both groups.

Our study does have limitations, including that study duration was only 1 year; thus, we were unable to determine the effect of NaHCO₃ on hard clinical outcomes. In addition, it is possible that the dose of NaHCO₃ that we used (0.5 mEq/kg-LBW per day) was too low to result in changes in vascular function. Systolic BP is known to be an important determinant of vascular function and arterial stiffness and could be a theoretical confounder. However, all participants had to have BP controlled before entering this study, and we included a run-in phase to achieve this to minimize this risk. Finally, we did not collect dietary information from the participants and diet can affect bicarbonate levels. Our study also has several strengths, including that it was a double-blinded, randomized, placebo-controlled trial that was powered to detect a clinically significant difference in FMD and aPWV. We obtained 24-hour urine collections and venous blood gases to determine the overall acid-base status of participants.

In conclusion, 12 months of treatment with NaHCO₃ did not improve vascular endothelial function or reduce arterial stiffness in patients with CKD stage 3b–4 with normal serum bicarbonate levels. Our results do not support the use of NaHCO₃ for vascular dysfunction in participants with CKD and normal serum bicarbonate levels.

Supplementary Material

jasn-34-1433-s001.pdf ^{(33.7KB, pdf)}

jasn-34-1433-s002.doc ^{(218.5KB, doc)}

Disclosures

J. Kendrick has received advisory fees from AstraZeneca. J. Kendrick has received grant funding from AstraZeneca, Bayer Healthcare, Fresenius Renal Therapies, NHBLI, NIA, NIH NIDDK, and the Juvenile Diabetes Research Foundation. K.L. Nowak has received consulting fees from Otsuka and honoraria from the National Kidney Foundation. K.L. Nowak has received grant funding from the NIH and Polycystic Kidney Disease (PKD) Foundation. M. Chonchol has received grant funding from NIH. All remaining authors have nothing to disclose.

Funding

Funding for this study was provided by the National Institutes of Health/National Heart, Lung, and Blood Institute R01 HL132868.

Author Contributions

Conceptualization: Jessica Kendrick.

Data curation: Michel Chonchol, Heather Farmer-Bailey, Jessica Kendrick, Kerrie Moreau, Kristen L. Nowak, Nayana Patel, Cortney Steele, Wei Wang.

Formal analysis: Jessica Kendrick, Kristen L. Nowak, Nayana Patel, Wei Wang, Zhiying You.

Funding acquisition: Jessica Kendrick.

Investigation: Jessica Kendrick.

Methodology: Emily Andrews, Michel Chonchol, Heather Farmer-Bailey, Jessica Kendrick, Kerrie Moreau, Nayana Patel, Zhiying You.

Project administration: Emily Andrews, Jessica Kendrick.

Resources: Michel Chonchol, Jessica Kendrick.

Software: Zhiying You.

Supervision: Jessica Kendrick, Kristen L. Nowak.

Writing – original draft: Jessica Kendrick.

Writing – review & editing: Emily Andrews, Michel Chonchol, Heather Farmer-Bailey, Jessica Kendrick, Kerrie Moreau, Kristen L. Nowak, Nayana Patel, Cortney Steele, Wei Wang, Zhiying You.

Data Sharing Statement

Data, which have been stripped of all personal identification and information and coded with a number, will be made available to qualified individuals within the scientific community who apply for data use.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/E443 and http://links.lww.com/JSN/E444.

Supplemental Methods.

Supplemental Table 1. Gastrointestinal side effects by treatment group.

Supplemental Table 2. Reasons for study discontinuation.

Supplemental Table 3. Changes in plasma bicarbonate levels by treatment group.