Potassium Bicarbonate Supplementation Lowers Bone Turnover and Calcium Excretion in Older Men and Women: A Randomized Dose-Finding Trial

Abstract

The acid load accompanying modern diets may have adverse effects on bone and muscle metabolism. Treatment with alkaline salts of potassium can neutralize the acid load, but the optimal amount of alkali is not established. Our objective was to determine the effectiveness of two doses of potassium bicarbonate (KHCO₃) compared with placebo on biochemical markers of bone turnover, and calcium and nitrogen (N) excretion. In this double-blind, randomized, placebo-controlled study, 244 men and women age 50 years and older were randomized to placebo or 1 mmol/kg or 1.5 mmol/kg of KHCO₃ daily for 3 months; 233 completed the study. The primary outcomes were changes in 24-hour urinary N-telopeptide (NTX) and N; changes in these measures were compared across the treatment groups. Exploratory outcomes included 24-hour urinary calcium excretion, serum amino-terminal propeptide of type I procollagen (P1NP), and muscle strength and function assessments. The median administered doses in the low-dose and high-dose groups were 81 mmol/day and 122 mmol/day, respectively. When compared with placebo, urinary NTX declined significantly in the low-dose group (p =0.012, after adjustment for baseline NTX, gender, and change in urine creatinine) and serum P1NP declined significantly in the low-dose group (p =0.004, adjusted for baseline P1NP and gender). Urinary calcium declined significantly in both KHCO₃ groups versus placebo (p < 0.001, adjusted for baseline urinary calcium, gender, and changes in urine creatinine and calcium intake). There was no significant effect of either dose of KHCO₃ on urinary N excretion or on the physical strength and function measures. KHCO₃ has favorable effects on bone turnover and calcium excretion and the lower dose appears to be the more effective dose. Long-term trials to assess the effect of alkali on bone mass and fracture risk are needed.

Introduction

Acid-producing diets have been implicated as potential contributors to bone and muscle loss in older adults. These diets are characterized by low content of fruits and vegetables in relation to their content of cereal grains and protein. According to the 2010 Dietary Guidelines,⁽¹⁾ among U.-S. adults, fruit and vegetable intake is one-half of the amount recommended, grain intake is twice the amount recommended, and protein intake is close to the recommended amount. It is therefore likely that much of the population is consuming an acid-producing diet. In our previous study, in fact, 96% of the participants consumed acid-producing diets at entry, based on their 24-hour urinary net acid excretion (NAE) levels.⁽²⁾ It is important to know the impact of acid-producing diets on the declines in bone and muscle mass and function that occur in older adults.

We recently reported that treatment with the alkaline salt, potassium bicarbonate (KHCO₃), 67.5 mmol per day, for 3 months, lowered urinary calcium excretion and urinary N-telopeptide of collagen type I (NTX), a biochemical marker of bone resorption, in healthy older men and women.⁽²⁾ After 3 months of treatment, the subjects with the lowest NAE levels had the lowest NTX levels.⁽²⁾ The lowest urinary NTX concentrations were seen in the subjects who were excreting alkali, raising the possibility that a higher dose of KHCO₃ may further lower bone resorption. In support of a higher dose of alkali for bone, Moseley and colleagues⁽³⁾ observed that calcium balance progressively improved in adults treated with placebo, 60 mmol per day, and 90 mmol per day of the alkaline salt, potassium citrate.

In our recent trial, KHCO₃ treatment also reduced nitrogen (N) excretion, an indicator of muscle wasting, and increased lower extremity muscle power in the women but not the men.⁽⁴⁾ The men were over 20% heavier than women. We postulated that the men may require a higher dose of KHCO₃ for muscle benefits and that KHCO₃ dosing should be weight-based. ⁽⁴⁾

This study was conducted to determine the dose of KHCO₃ that is optimal for short-term indicators of bone and muscle preservation in men and women. The primary aim of the trial was to describe and compare changes in urinary NTX and N excretion across KHCO₃ doses and to describe the safety and tolerability of the supplements. Exploratory aims were to determine the effects of supplementation on serum amino-terminal propeptide of type I procollagen (P1NP), urinary calcium excretion, insulin-like growth factor 1 (IGF-1), and insulin-like growth factor binding protein 3 (IGFBP-3) concentrations and on physical strength and function measures, including grip strength, the timed stair climb, and the Short Physical Performance Battery (SPPB). Finally, we describe the diet characteristics of the subjects in relation to their NAE levels at baseline.

Subjects and Methods

Subjects

Healthy ambulatory men and women age 60 years and older, weighing ≥45 kg or ≤113.5 kg, and with an estimated glomerular filtration rate (GFR) of at least 50 mL/min/1.73 m²,⁽⁵⁾ were recruited through direct mailings and advertisements in the community. The women were menopausal for at least 1 year. Exclusion criteria included: serum potassium > 5.3 mEq/L, serum bicarbonate >33 mmol/L, fasting sugar >130 mg/dL, >2 units of alcohol/day, gastroesophageal reflux disease, active malignancy, adrenal insufficiency, hyperparathyroidism, untreated thyroid disease, significant immune disorder, heart disease, salt-restricted diets, kidney stones in the last 5 years, fasting spot urine calcium/creatinine >0.38 mmol/mmol after 1 week off of calcium supplements, serum calcium outside the range of 8.3 to 10.2 mg/dL, current use of diuretics, angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, or nonsteroidal anti-inflammatory drugs >3 times/week, over-the-counter antacids, potassium supplements, and salt substitutes; bisphosphonates in the last 2 years, other osteoporosis medications in the last 6 months, or glucocorticoids for >10 days in the last 3 months.

We prescreened 1955 subjects by telephone and screened 356 of those subjects. Of these, 90 were ineligible, 22 were potentially eligible but not enrolled, and 244 (126 men and 118 women) were enrolled (224 whites, 18 blacks, and 2 Asians); 233 subjects completed the study (see Supporting Fig. 1). With a planned sample size of 231 subjects, we had an estimated 80% power at the two-sided 0.05 level of significance to detect differences between any two groups in changes in NTX/Cr of 5.0 nmol/mmol and N/Cr of 6.5 mmol/mol. The protocol was approved by the Tufts Medical Center-Tufts University Institutional Review Board, and written informed consent was obtained from each subject. All subject visits took place at the U.S. Department of Agriculture (USDA) Human Nutrition Research Center on Aging (HNRCA) at Tufts University. The study is registered at Clinicaltrials.gov (http://clinicaltrials.gov/show/NCT1475214).

Fig. 1.

Adjusted mean 84-day changes in urinary calcium, NTX, and NAE and serum P1NP. For adjustments, see Table 3. ^*Differs from placebo at p < 0.001. +Differs from placebo at p =0.036. #Differs from placebo at p =0.011.

Study design and supplements

In this double blind, placebo-controlled, parallel group trial, subjects were randomized to 84 days of treatment with: placebo, 1.0 mmol/kg per day of KHCO₃, or 1.5 mmol/kg per day of KHCO₃. Within each treatment arm, subjects were dosed according to their weight category: A =45 to 68.9 kg; B =69 to 90.9 kg; and C =91 to 113.5 kg. The random allocation sequence was based on the program at http://www.randomization.com and randomization was stratified by weight category (A, B, or C) in blocks of three. The randomization and treatment assignments were performed by SSH; other team members and all subjects were blinded. BD-H enrolled the subjects. The weight categories were chosen based on the weight of subjects who participated in our previous KHCO₃ trial and the range we would likely recruit naturally for this study.⁽²⁾ The number of capsules was rounded to approximate the desired dose for the midpoint of each weight group. Subjects came to the HNRCA to sign the consent and receive instructions for record-assisted 24-hour diet recalls. After the record-assisted 24-hour recall, they returned for the baseline visit (day 1) bringing a 24-hour urine collection. On that visit they had medical history and physical activity questionnaires, physical performance measures, a dual-energy X-ray absorptiometry (DXA) total body scan, and a blood draw. They were given their study capsules and compliance calendars. They returned for serum potassium safety checks on days 10, 13, 16, 19, 22, and 50; GFR was also estimated on days 22 and 50. At the end of the study, the 24-hour recall (day 83) and the 24-hour urine collection (day 83) were repeated and subjects returned fasting on day 84 for a final blood draw, questionnaires, and physical function measures.

The KHCO₃ capsules contained 13.5 mmol of KHCO₃ each and the matching placebos contained microcrystalline cellulose. Capsules were preloaded into containers that held a week’s supply, with each day’s supply divided equally across three cells. Subjects in weight group A had 6 capsules per day, subjects in B had 9, and subjects in C had 12. However, the proportion of KHCO₃ and placebo capsules in each cell varied across the three treatment arms. Subjects were instructed to take the capsules after breakfast, lunch, and dinner each day with a full glass of water. To allow subjects to adapt to the KHCO₃ (to reduce risk of gastrointestinal intolerance), they gradually increased the number of capsules taken daily until day 10 when all subjects were taking their full assigned dose. Subjects who could not tolerate their assigned dose were encouraged to remain in the study and to take the number of capsules that they could tolerate. The KHCO₃ capsules and placebo capsules were purchased from Life Enhancement Products, Inc (Petaluma, CA, USA). Independent analysis by Covance (Princeton, NJ, USA) indicated that they contained 13.6 g (101% of the stated content) of KHCO₃.

Diet assessment

Record-assisted 24-hour dietary recalls were collected at the beginning and end of the study under the supervision of a research dietitian, on days coinciding with the 24-hour urine collections. Subjects were instructed to record everything they ate and drank and the dietary supplements taken. Prior to the second set of 24-hour recalls, the instructions were repeated by phone the day before the subject started their food log. They were provided with two-dimensional measuring aids to assist with estimating portion sizes. Subjects reported what they consumed to a diet technician on the study visits. The dietary interview training was adapted from the 24-hour recall protocol described at http://www.ncc.umn.edu/index.html Dietary intake data were coded and analyzed using Nutrition Data System for Research software version 2011, Nutrition Coordinating Center, University of Minnesota. This version defines food group servings according to the Dietary Guidelines for Americans, 2005.

Subjects agreed to consume their usual diets, maintain a stable exercise pattern, and not to attempt to gain or lose weight during the study. They also agreed to discontinue their own calcium supplements during the study. They were allowed to continue their vitamin D supplements in amounts up to 4000 IU/day but agreed not to change their pattern of vitamin D supplement use during the study. During the study, all subjects were provided with 600 mg of elemental calcium per day as calcium triphosphate, a pH-neutral calcium supplement (Posture D; US Rhodia, Cranbury, NJ, USA). This supplement also contained 500 IU of vitamin D₃.

Physical strength and function and lean mass assessments

Height was measured with a stadiometer and weight with a digital scale. Leisure, household, and occupational activity was estimated with use of the Physical Activity Scale for the Elderly (PASE) questionnaire.⁽⁶⁾ Handgrip strength of both dominant and nondominant hands was determined using a hand-held dynamometer (Model TKK5401; Takei). The highest of three consecutive readings was recorded with each hand. For the stair climbing, subjects were asked to climb a standard set of stairs as quickly and safely as possible. A stopwatch was used to record the time to the nearest 0.01 s when the subject’s first foot landed on the 10th step (up) and on the starting level (down). They could use the handrail if necessary and any use was recorded. Stair climbing is a good indicator of knee extensor strength and of functional capacity in the elderly and is significantly correlated with lower extremity strength and power.⁽⁷⁾ This test has excellent reliability (interclass coefficient =0.96). ⁽⁸⁾ For the SPPB, subjects were asked to perform a balance assessment (open, semi-tandem, and tandem stance), a timed 4-m walk, and a chair rise test (timed 5 rises). Each of these three tests has a maximum score of 4 points (total score 12). The SPPB captures domains of lower extremity strength and endurance and balance; it is highly predictive of subsequent disability.⁽⁹⁾ Lean (nonfat, nonbone) tissue mass was measured at baseline on a GE Lunar Prodigy DXA Scanner (Madison, WI, USA) using acquisition software version 6.1 and analysis version 12.2, with a precision of 0.77%.⁽¹⁰⁾ Multiple thickness phantoms were scanned weekly to monitor instrument stability.

Adherence monitoring

We monitored adherence to study capsules with a phone call on day 5 and by pill counts (subjects were instructed to return all pill holders) on days 10, 22, 50, and 84. Subjects also indicated the number of capsules they had taken after each meal on compliance calendars, which they returned on days 22, 50, and the final visit.

Biochemical measurements

Blood was drawn between 7:00 a.m. and 9:30 a.m. after the subjects had fasted for 12 hours. All samples from individual subjects except the safety bloods were batched for analyses. Serum 25-hydroxyvitamin D [25(OH)D] was measured by liquid chromatography–tandem mass spectrometry (LC-MS/MS) on a Waters Acquity UPLC with TQD triple quadrupole mass spectrometer with a coefficient of variation (CV) of 6%. Separation was on a C18 UPLC column. NIST 25OHD standards were run to calibrate the assay. Serum IGF-1, and IGFBP-3 were measured by chemiluminescent immunoradiometric assay on an automated immunoassay system (IMMULITE 1000; Diagnostic Product Corp., Los Angeles, CA, USA), with CVs of 3 to 9%. Serum potassium, bicarbonate, and urinary creatinine and potassium were measured on an automated clinical chemistry analyzer (Olympus AU400; Olympus America Inc., Melville, NY, USA) with CVs of 3.0% to 6.0%. Serum P1NP was measured by competitive radioimmunoassay with RIA kits from Orion Diagnostica Uni P1NP (Espoo, Finland) with intraassay and interassay CVs of 5.0% and 8.1%, respectively. Urinary calcium was measured by direct-current plasma emission spectroscopy (Beckman SpectraSpan VI Direct Current Plasma Emission Spectrophotometer; Beckman Instruments, Fullerton, CA, USA) with a CV of 3% to 5%. Urinary NTX was measured by enzyme-linked immunosorbent assay (Wampole, Princeton, NJ, USA) with a CV of 5.6% to 7.7%. Urinary N was measured with a model FP-2000 nitrogen/protein determinator (LECO, St. Joseph, MI, USA) with intraassay and interassay CVs of 6.5% and 8.6%.⁽⁴⁾ Urinary pH was determined on the Accumet Excel pH meter (Fisher Scientific, Pittsburgh, PA, USA). NAE (= titratable acid +NH₄⁺ HCO₃⁻) was measured in 24-hour urine collections by a modification of the Jorgensen titration method,⁽¹¹⁾ as described by Chan,⁽¹²⁾ with precision in our laboratory of 10.1%.⁽²⁾

Statistical analysis

In accordance with intention-to-treat principles, analyses were conducted on all 233 subjects who completed the final visit, whether or not they were compliant with treatment. There were several missing laboratory and physical function values. For analyses in which one or more value was missing, the resulting sample size is indicated in parentheses in the relevant table or figure. Data were examined graphically to rule out the presence of outliers and to evaluate the linearity of bivariate associations. One N excretion value was set aside because it was implausibly high (2685 mmol/day compared on the final visit compared with a baseline value of 614 mmol; the next highest value among all subjects was 2012). We did not stratify by gender because we found no significant interactions of gender with KHCO₃ on the endpoints, but we did adjust for gender in all analyses.

Pearson correlation coefficients were calculated to describe selected bivariate associations. Proportions were compared with the chi-square test. Subject characteristics and other means were compared across groups by analysis of variance for unadjusted values (overall differences and/or linear trends) and by analysis of covariance (ANCOVA) with least-squares means for adjusted values. A Bonferroni adjustment was applied to pairwise comparisons to account for multiple testing. Covariates that were adjusted for in all analyses of changes in outcome variables included gender and the baseline value of the outcome variable. In addition, changes in urine values were adjusted for concurrent changes in creatinine excretion to account for any differences in completeness of the baseline and final urine collections. Also, changes in outcome variables with direct links to diet (urine calcium and N) were adjusted for differences in dietary intakes of the relevant nutrient during the baseline and final urine collections. Finally, changes in measures of physical strength and function were adjusted for concordance of the tester(s) who made the measurements. The same tester was used for baseline and final measurements for 186 of the 233 subjects studied. Potential interactions were investigated by adding cross-product terms to ANCOVA models. Preliminary analyses indicated that there was no interaction of treatment group with gender in the analyses of any of the outcomes, so men and women were analyzed together. Values of p <0.05 were considered to indicate statistical significance. Analyses were conducted with SPSS, version 22 (IBM Corp., Armonk, NY, USA).

Results

Subject visits took place from February 2012 through December 2014. Dietary and selected other characteristics of the subjects are shown by baseline NAE level in Table 1. Only 1 subject had an NAE below −5.0 mmol and her value was −5.89 mmol/day. Subjects with lower NAE reported consuming more fruits and vegetables than those with higher NAE (eg, 8.1 ± 5.4 servings/day in the lowest compared with 5.5 in the highest NAE category), p =0.032. Grain consumption also increased with NAE from 5.5 servings/day in the lowest category to 7.2 in the highest. Protein intake and dairy food intake did not differ significantly across the NAE categories.

Table 1.

Baseline Characteristics of 231 Subjects by Baseline Net Acid Excretion

	Net acid excretion (mmol)
	−5 to 5	5 to 15	15 to 50	≥50
n	83	67	56	25
% Female	65.1	53.7	26.8	32.0
Age (years)	66.4 ± 4.9	67.6 ± 5.5	67.5 ± 6.3	65.2 ± 5.2
GFR	76.3 ± 12.3	75.8 ± 14.0	76.2 ± 12.8	75.7 ± 10.2
Body mass index (kg/m2)	25.7 ± 4.2	25.8 ± 4.3	25.4 ± 3.1	26.0 ± 4.0
kcal/day	1968 ± 665	2010 ± 681	2168 ± 731	2009 ± 596
F+V (servings/d)a	8.1 ± 5.4	6.4 ± 3.3	7.8 ± 5.9	5.5 ± 3.1
Grains, servings/db	5.5 ± 3.6	6.2 ± 3.3	6.6 ± 4.4	7.2 ± 3.9
F+V/grain servings ratioc	1.47	1.07	1.18	0.78
Meat, poultry, and fish (servings/day)	6.0 ± 4.0	7.2 ± 4.3	6.8 ± 5.4	6.0 ± 4.2
Dairy (servings/day)	2.0 ± 1.7	2.2 ± 2.0	2.4 ± 2.3	1.9 ± 1.4

Values are mean ± SD or %. One subject was not included due to NAE < −5 (−5.89 nmol) and another subject had missing diet data.

GFR =glomerular filtration rate; F+V =fruits and vegetables.

^aOverall differences among groups, p =0.032.

^bLinear trend, p =0.023.

^cCalculated as mean daily fruits and vegetable servings divided by grain servings.

The mean dose of KHCO₃ administered was 74 ± 19 mmol per day (median, 81 mmol per day) in the low-dose group and 111 ± 24 mmol per day (median, 122 mmol per day) in the high-dose group. The clinical characteristics of the subjects are shown by treatment group in Table 2. There were no statistically significant group differences in any of these characteristics.

Table 2.

Baseline Characteristics of 233 Completers

	Placebo	Low-dose	High-dose
n	79	79	75
Age (years)	67.0 ± 6.2	67.2 ± 5.3	66.4 ±4.9
% Female	46.8	49.4	49.3
Height (cm)	168.3 ± 9.2	170.2 ± 9.5	168.2 ±9.0
Weight (kg)	72.7 ± 13.6	74.2 ± 13.9	73.8 ±13.4
Lean mass (kg)	47.7 ± 9.9 (73)	47.9 ± 11.2 (74)	47.0 ±9.8 (70)
Supplemental vitamin D (IU)a	448 ± 1056	304 ± 760	376 ±756
Calcium intake (mg/day)	1013 ± 464	980 ± 470	1028 ±579
Protein intake (g/day)	86.9 ± 35.5	83.7 ± 37.4	87.7 ±35.8
Nitrogen intake (g/day)	14.2 ± 5.8	13.6 ±6.0	14.3 ±5.8
Serum 25(OH)D (nmol/L)	72.6 ± 22.4	68.5 ±22.9	68.5 ±24.5
PASE score	149 ± 69	143 ±66	141 ±75
GFR (mL/min/1.73 m2)	75 ± 13	77 ±13	76 ±12

Values are mean ± SD or %. When values were missing, the sample size is indicated in parentheses.

25(OH)D =25-hydroxyvitamin D; PASE =Physical Activity Scale for the Elderly; GFR =glomerular filtration rate.

^aVitamin D intake from personal supplements. (All subjects received an additional 500 IU of vitamin D₃/day as part of the study.)

Adherence with the study capsules averaged 92.2% in the placebo group, 91.3% in the low-dose KHCO₃ group, and 87.4% in the high-dose KHCO₃ group. The study completion rate was 97.5% in the placebo group, 94.0% in the low-dose KHCO₃ group, and 94.9% in the high-dose KHCO₃ group. Two subjects in the placebo group, 5 in the low-dose group, and 4 in the high-dose group withdrew from the study for personal reasons or loss of interest. Seventeen subjects stopped taking study pills, 7 because of hyperkalemia confirmed on a second measurement to rule out hemolysis (1 subject in the placebo group, 1 in the low-dose group, and in 5 in the high-dose group), 5 because of a decline in GFR (3 in the low-dose group and 2 in the high-dose group), and 5 because of gastrointestinal complaints (2 in the placebo group, 1 in the low-dose group, and 2 in the high-dose group) (Supporting Fig. 1). These 17 subjects remained in the study.

At baseline, urinary NAE was significantly correlated with urinary NTX (r =0.668, p < 0.001), but not with urinary calcium (r =−0.091, p =0.116) or serum P1NP (r =0.001, p =0.994). Baseline levels of urinary calcium, NTX, and NAE excretion and serum P1NP and their 84-day adjusted changes are shown by treatment group in Table 3 and the adjusted changes are illustrated in Fig. 1. There were no significant differences in any of the measures at baseline.

Table 3.

Baseline and 84-Day Changes in Laboratory Values by Treatment Group

	Placebo	Low-dose	High-dose	pa
n	79	79	75
Serum
P1NP (nmol/L)
Baseline	1.33 ±0.06	1.31 ±0.05	1.36 ±0.05 (74)	0.796
Changeb	−0.03 ±0.03	−0.14 ±0.03 (78)c	−0.10 ±0.03 (74)	0.013
Urine
Calcium (mmol/day)
Baseline	3.08 ±0.19	2.85 ±0.16 (78)	3.38 ±0.25	0.189
Changed	0.45 ±0.14	−0.31 ±0.14 (78)e	−0.52 ±0.14e	<0.001
NTX (nmol/day)
Baseline	241 ±17	240 ±17	230 ±20	0.903
Changef	−14 ±11	−53 ±11g	−43 ±11	0.035
NAE (mmol/day)
Baseline	16.7 ±2.5	19.8 ±3.1 (78)	19.8 ±3.7	0.719
Changef	−2.1 ±2.6	−19.4 ±2.6 (78)e	−26.4 ±2.6e	<0.001
pH
Baseline	6.12 ±0.05	6.19 ±0.06	6.10 ±0.05	0.473
Changeb,h	−0.03 ±0.06	0.81 ±0.06	1.02 ±0.06	<0.001

Values are mean ±SE. When values were missing, sample size is indicated in parentheses.

^aValues of p were calculated using analysis of covariance.

^bAdjusted for baseline value and gender.

^cDiffers from placebo, p =0.011.

^dAdjusted for baseline value, gender, and changes in calcium intake and urine creatinine.

^eDiffers from placebo, p < 0.001.

^fAdjusted for baseline value, gender, and change in urine creatinine.

^gDiffers from placebo, p =0.036.

^hEach group differs from both others, p < 0.023.

Mean NAE declined significantly in the low-dose and high-dose KHCO₃ groups compared with placebo (p < 0.001, Table 3). Urinary calcium and urinary NTX also declined in the low-dose and high-dose KHCO₃ groups compared with placebo (p ≤ 0.035, Table 3). Serum P1NP declined in the low-dose group relative to placebo (p =0.004, Table 3). There was no interaction of treatment group with baseline NAE whether the latter was included as a continuous (p > 0.59) or categorical (p > 0.42) variable.

To estimate the levels of NAE associated with the lowest levels of urinary calcium and NTX and serum P1NP, we display levels of these measures in relation to category of NAE in all subjects after 84 days in the study (Fig. 2). Urinary calcium levels were sequentially lower across declining categories of NAE to the NAE range of −10 to −5 mmol/day. Urinary NTX was lowest at NAE levels in the range of −5.0 to 5.0 mmol/day. At entry into the study, 1 subject was slightly below this range, 83 subjects had levels in the range, and 149 subjects had NAE levels above the range. The pattern for serum P1NP was similar to that of urinary calcium.

Fig. 2.

Mean urinary calcium, NTX, and P1NP by category of NAE at the end of the study.

On the day prior to the baseline visit, N intake and 24-hour urinary N were significantly correlated (r =0.53, p < 0.001, n =233). Baseline levels of N excretion, IGF-1, IGF-BP3, and of physical strength and function and their 3-month adjusted mean changes are shown in Table 4. There was no effect of treatment on N excretion in the group as a whole (p =0.884) or in men (p =0.316) or women (p =0.243), after the same adjustments (excluding gender). Similarly there were no significant treatment group differences in IGF-1 or IGF-BP3 or any of the muscle strength and performance measures in this study (Table 4).

Table 4.

Baseline and 84-Day Changes in Biochemical and Physical Measures Related to Muscle

	Placebo	Low-dose	High-dose	p
n	79	79	75
Serum
IGF-1 (nmol/L)
Baseline	19.5 ±0.8	17.7 ±0.7	18.0 ±0.8 (74)	0.189
Changea	0.49 ±0.37	0.26 ±0.37	−0.03 ±0.38 (74)	0.618
IGFBP-3 (mg/L)
Baseline	4.81 ±0.12	4.55 ±0.10	4.55 ±0.12 (74)	0.179
Changea	0.01 ±0.06	0.09 ±0.06	0.05 ±0.06 (74)	0.600
Urine
Nitrogen excretion (mmol/day)b
Baseline	864 ±33	769 ±32	834 ±39	0.143
Change	−21.0 ±18.7	−29.4 ±18.6	−16.4 ±19.1 (74)	0.884
Strength and function
Grip strength, dominant (kg)c
Baseline	32.1 ±1.0	33.6 ±1.1	31.6 ±1.0	0.372
Change	−0.59 ± 0.37	−0.06 ±0.35	0.02 ±0.36 (74)	0.366
Grip strength, nondominant (kg)c
Baseline	30.3 ±0.9	31.2 ±1.1	29.6 ±1.0	0.538
Change	−0.42 ±0.35 (78)	0.10 ±0.33	0.05 ±0.34	0.431
SPPB scorec
Baseline	11.2 ±0.1	11.0 ±0.1	11.1 ±0.1	0.435
Change	0.03 ±0.10	−0.09 ±0.09	−0.01 ±0.09	0.618
Stair climb, ascent time (s)c
Baseline	3.81 ±0.12	4.08 ±0.20 (78)	4.09 ±0.14	0.344
Change	−0.06 ±0.08	−0.07 ±0.08 (78)	−0.00 ±0.08	0.776
Stair climb, descent time (s)c
Baseline	3.30 ±0.12	3.58 ±0.20 (78)	3.74 ±0.17	0.178
Change	−0.06 ±0.08	0.02 ±0.08 (78)	−0.06 ±0.08	0.715

Values are mean ±SE. When values are missing, the sample size is indicated in parentheses.

^aChanges adjusted for baseline value and gender.

^bChanges adjusted for baseline value, gender, change in nitrogen intake, and change in urine creatinine excretion.

^cChanges adjusted for baseline value, gender, and tester concordance.

Discussion

The results of this study, including effects of KHCO₃ on calcium excretion and markers of bone turnover, suggest that correcting mild metabolic acidosis is beneficial to bone. Compared with placebo, urine calcium excretion was reduced by 169% on the lower dose and by 213% on the higher dose of KHCO₃. This finding is consistent with a recent report of decline in calcium excretion with increasing alkali dosage in 52 postmenopausal women.⁽³⁾ Moseley and colleagues⁽³⁾ documented that calcium absorption was not affected by alkali supplementation and reasonably concluded that the reduction in calcium excretion likely resulted from reduced calcium losses from bone. In other studies, alkali supplementation has repeatedly reduced calcium excretion,^(3,13,14) although this has not been an entirely consistent finding.⁽¹⁵⁾

Supplementation with KHCO₃ significantly reduced circulating levels of biochemical markers of bone turnover. The maximal reduction of both markers was observed with the lower dose of KHCO₃, indicating that 1 mmol/kg per day of alkali is sufficient to optimize short-term indicators of bone health. Treatment with the lower-dose of KHCO₃ reduced the bone resorption marker urinary NTX, by 18.7%, and the formation marker serum P1NP by 10.7%. The observed decline in serum P1NP likely reflects the coupling of bone formation to bone resorption. It is not clear why the higher dose did not significantly reduce NTX and P1NP, although the direction of change in both markers was as predicted. Our NTX result is consistent with the work of Moseley and colleagues,⁽³⁾ who found maximal reduction in the bone resorption marker, serum CTX, at their lower 60 mmol/day dose of alkali; in contrast, in that study, neither the 60-mmol nor the 90-mmol dose significantly altered the bone formation marker, bone-specific alkaline phosphatase. The only other multiple dose study was that of Macdonald and colleagues,⁽¹⁵⁾ who tested 18 mmol/day and 50 mmol/day doses of potassium citrate and did not find any effect of either dose on serum CTX or P1NP. In observational studies^(16,17) and single-dose supplementation trials,^{(2,13,18–20)} alkali lowered biochemical markers of bone resorption, whereas bone formation markers either declined⁽¹⁹⁾ or did not change significantly.^(2,13,18,20) In contrast, Jehle and colleagues⁽²¹⁾ reported that supplementation with 60 mmol/day of potassium citrate significantly increased serum P1NP levels by 13.8% and decreased urinary NTX by about 5% over a 2-year study period. To our knowledge, this is the only report of an increase in a bone formation marker with alkali supplementation.

We did not identify any impact of supplementation with either dose of KHCO₃ on N excretion in the group as a whole or in men and women separately. This is in agreement with our earlier finding that supplementation with 67.5 mmol/day of KHCO₃ did not alter N excretion in men, but is at odds with our previous observation that supplementation lowered N excretion by 7% to 8% in women.⁽⁴⁾ In an intervention study in women on controlled high protein diets, treatment with alkali decreased nitrogen excretion⁽²²⁾ and in a small study in 24 women, alkali reduced N excretion by 12%, but this decline was not statistically significant.⁽¹⁹⁾ These mixed findings raise questions about the impact of acid-base balance on N excretion and muscle mass that can only be answered in subsequent studies. Supplementation had no significant effect on handgrip strength, as seen previously,⁽⁴⁾ or on the physical function measures, stair climbing and the SPPB, perhaps because of ceiling effects. The mean baseline SPPB score, for instance, was >11 on a 12-point scale. We had previously observed improved double leg press power and endurance in women treated with 67.5 mmol/day of KHCO₃.⁽⁴⁾ Long-term effects of supplementation of physical performance measures are unknown.

One of the objectives of this study was to identify the optimal range of NAE for bone and describe the diet composition of subjects in that range. We examined the association of NAE with bone-related measures after 84 days of treatment to try to identify a potential NAE target for optimal bone health. From inspection of Fig. 2, a reasonable estimate of the NAE range for maximal suppression of bone resorption, as indicated by the lowest urinary NTX, is −5.0 to 5.0 mmol/day. Reductions in bone turnover are of clinical significance because of their associations with reduced bone loss and fracture rates.^(23,24) The other two measures, urinary calcium and serum P1NP, had minimal values at NAE levels slightly below this range, and the significance of this for bone health is uncertain.

In this study, at baseline, 36% of the subjects were in the desirable −5.0 to 5.0 mmol/day range of NAE and 64% were above the range. Examination of the subjects’ diets at entry revealed that intakes of protein and dairy foods were not linked to baseline NAE. Protein is acid-producing, but, in this study did not vary across categories of NAE, perhaps because of the balance of animal and plant protein sources consumed by our participants. Dairy foods, with the exception of hard cheeses, are generally metabolized to neutral compounds⁽²⁵⁾ and were not associated with the NAE level. As expected,⁽²⁵⁾ subjects in the desired NAE range had higher fruit and vegetable intake and lower grain intake than those above the range. Specifically, they consumed an average of 8.1 servings per day of fruits and vegetables and 5.5 servings of grains giving them a high daily mean fruit+vegetable/grain servings ratio of 1.47 (8.1 ÷ 5.5). The diet composition of these subjects is within the 2010 Dietary Guidelines, which are, for a 2000-calorie diet, 2.5 cups of vegetables, 2 cups of fruits, and 6 oz of grains.⁽¹⁾ This translates to about 9 servings of fruits and vegetables and 6 servings/day of grains (ratio 1.5). These subjects’ diets were similar to the Dietary Approaches to Stop Hypertension (DASH) diet, which contained 9.6 servings of fruits and vegetables and 7.5 servings of grains per day.⁽²⁶⁾ Consumption of the DASH diet not only lowered blood pressure,⁽²⁶⁾ it lowered bone resorption markers to a similar degree as our lower dose of KHCO₃.⁽²⁷⁾ Specifically, serum CTX levels were decreased by 16% to 18% on that diet, and serum osteocalcin levels, an indicator of bone formation, decreased by 8% to 11%.⁽²⁷⁾ The low acid load of the DASH diet likely contributed to the reduction in bone turnover in that study. From our study, we would expect that subjects with daily ratios of fruit+vegetable/grain servings <1.2 would be most likely to benefit from KHCO₃ or potassium citrate supplementation. These subjects would also be expected to benefit from changing their diets to increase that ratio.

It remains to be shown whether maintaining a near neutral NAE over the long term would reduce bone loss and lower risk of fracture in older adults. It is reasonable to expect that it would, based on the work of Jehle and colleagues,⁽²¹⁾ who found that older men and women treated with 60 mmol/day of potassium citrate for 2 years had reduced rates of bone loss from the spine and femoral neck. In that study, the mean NAE level in the supplemented group was in the target range identified in this study. In contrast, however, Macdonald and colleagues⁽¹⁵⁾ did not identify an effect of supplementation with potassium citrate on rates of bone loss (or bone resorption); the achieved NAE level in that study is unknown.

There were no safety concerns with the lower dose of KHCO₃. The higher dose did cause a few more episodes of hyperkalemia. The safety of long-term ingestion of alkaline salts of potassium has not been rigorously evaluated because intervention trials have been of no more than 2 years’ duration.

In conclusion, supplementation with the lower dose of KHCO₃, 1 mmol/kg/day or a median dose of 81 mmol/day, significantly improved intermediary indicators of bone health (lowered calcium excretion, NTX excretion, and serum P1NP levels) but had no significant effect on N excretion or measures of muscle strength or function in healthy older men and women. This dose was safe and well tolerated. The higher KHCO₃ dose of 1.5 mmol/kg/day produced no greater decline in urinary NTX but did cause more episodes of hyperkalemia. The lowest levels of urinary NTX during treatment were seen at a NAE range of −5.0 to 5.0 mmol per day, suggesting that this may be a reasonable target range for bone health. The subset of individuals in this range at entry into the study, comprising 36% of our study population, had diets resembling the DASH diet with respect to fruit and vegetable grain intake. Their diets were also compatible with the 2010 Dietary Guidelines. Further work is needed to determine whether achieving and sustaining the nearly neutral NAE range of −5.0 to 5.0 mmol/day would reduce risk of osteoporosis.