Effect of high-normal compared with low-normal arterial pH on protein balances in automated peritoneal dialysis patients^,

Abstract

Background: Although the protein catabolic effects of metabolic acidosis are well established, it is unclear whether the entire reference range of arterial pH (7.37–7.44) is equivalent for protein balance.

Objective: We undertook this study to test the hypothesis that in patients undergoing automated peritoneal dialysis, an arterial pH of 7.43–7.45, as compared with a pH of 7.36–7.38, is associated with more-positive nitrogen balances.

Design: Eight stable subjects (5 men) aged 43.1 ± 15.3 y completed a randomized, crossover nitrogen balance study for ≥42 d. Arterial pH was varied by changing the daily doses of sodium citrate/citric acid and ammonium chloride.

Results: The subjects attained mean (±SD) arterial pH values of 7.37 ± 0.01 and 7.44 ± 0.02 during the low-normal and high-normal pH phases, respectively. The higher arterial pH was associated with higher net nitrogen balances (3.22 ± 1.37 compared with 2.29 ± 2.18 g/d; P = 0.06), lower serum urea nitrogen (54.1 ± 13.7 compared with 64.4 ± 20.2 mg/dL; P = 0.01), higher fasting leucine flux (P = 0.02), and increased fasting total-body protein synthesis (P = 0.01) and degradation (P = 0.02). In 7 of 8 study subjects, nitrogen balances were more positive at the higher arterial pH (P = 0.004). There were no significant changes in anthropometric measurements, other biochemical measurements, and the mRNA content of selected proteins in skeletal muscle.

Conclusion: This study suggests that in most stable automated peritoneal dialysis patients, a mean arterial pH of 7.44, as compared with 7.37, is associated with more-positive nitrogen balances. This trial was registered at clinical trials.gov as NCT00586131.

INTRODUCTION

Metabolic acidosis is frequently present in patients with chronic kidney disease (CKD); correction of metabolic acidosis is one of the beneficial effects of maintenance dialysis therapy. A large body of evidence supports the thesis that metabolic acidosis contributes to the protein wasting commonly associated with CKD—both by its catabolic and its antianabolic effects (1). The arterial pH for optimal protein balance, however, has not been well defined. Most clinical laboratories define normal arterial pH to be between 7.37 and 7.44, and this entire range is implied in clinical practice as equivalent for protein balance. However, in cultured BC3H1 myocytes, an increase in the pH of the culture medium from 7.40 to 7.50 results in a sharp increase in protein synthesis (2). Consistent with these observations, in a randomized controlled clinical trial of outpatients undergoing peritoneal dialysis, subjects randomly assigned to peritoneal dialysis solutions with a higher lactate concentration (40 mmol/L) had a higher arterial pH (7.44 compared with 7.40) at the end of 1-y of follow-up and had a greater increase in body weight than did those treated with peritoneal dialysis solutions with lower lactate concentrations (35 mmol/L) (3). In contrast, in all but one kinetic study undertaken to examine the effect of correction of metabolic acidosis on leucine and/or phenylalanine kinetics in CKD subjects, the arterial pH in the control group was in the acidemic range (4–8). Thus, it remains unclear whether increasing the arterial pH within the reference “normal” range may improve the protein wasting of CKD. It needs to be remembered that the so-called reference normal range of arterial pH derives not from detailed and sophisticated investigations of optimality of physiologic and metabolic function, but from clinical criteria of no evidence of overt ongoing disease processes.

We undertook this study to test the hypothesis that in maintenance dialysis subjects treated with automated peritoneal dialysis, an arterial pH of 7.43–7.45 is associated with significantly greater positive nitrogen balances than is an arterial pH of 7.36–7.38.

SUBJECTS AND METHODS

Study design

The study was undertaken as an open-label, crossover, randomized controlled trial. Whether the subjects were assigned to a lower or higher arterial pH during the first half of the study was determined by using a computer-generated randomization list drawn up by a statistician. The allocation sequence was concealed from the investigators responsible for enrolling the participants and implementing the study protocol. The study was approved by the Institutional Review Board at the Los Angeles Biomedical Research Center. An independent monitor, as well as the local General Clinical Research Center Advisory Committee, served as the Data Safety and Monitoring Board for the trial, and 6-monthly safety reviews were performed.

Subjects

Subjects with end-stage renal disease between the ages of 21 and 65 y of age, undergoing treatment with automated peritoneal dialysis for ≥1 mo, were eligible to participate in the study. Other inclusion criteria were 1) hemoglobin > 11.0 g/dL, 2) stable erythropoietin dose for ≥3 mo, 3) serum albumin > 3.3 g/dL (bromocresol green method), 4) normalized protein equivalent of nitrogen appearance ≥0.8 g · kg⁻¹ · d⁻¹, 5) total weekly Kt/V urea ≥1.7, and 6) no evidence of primary or secondary myopathy. The exclusion criteria were as follows: 1) symptomatic, severe cardiac, liver, lung, or musculoskeletal disease or previous amputation; 2) diabetes mellitus requiring insulin therapy; 3) subjects treated with l-carnitine or anabolic hormones during the preceding 6 mo; 4) active vasculitis or cancer; 5) treatment with glucocorticoids in the preceding 6 mo; 6) severe chronic infection or acute or chronic inflammatory or catabolic illnesses; 7) alcohol or drug abuse; and 8) unable to give informed consent or follow study protocol.

Interventions

The sequence of interventions for the study is summarized in Figure 1. To minimize the time required for equilibration, the total energy and protein intakes of each subject during the study were designed to be the same as his or her intake before joining the study, as determined by dietary diaries and interviews and 24-h collections of dialysate and urine. The dietary protein and energy intakes were kept constant during the entire period of the metabolic protocol. Total energy intake was derived from a combination of food intake and d-glucose uptake from the peritoneal dialysate. The usual dietary energy intake was determined by using the information collected from a 3-d food diary, a dietary interview, and measurement of resting energy expenditure. The energy provided by the absorption of glucose from the peritoneal dialysate was estimated by using the Adequest software (Baxter Health Care, Deerfield, IL) (9). The peritoneal dialysis prescription, including the tonicity of the dialysate, was kept constant during the entire period of hospitalization to maintain a constant energy intake from the glucose absorption from the peritoneal dialysis solution. Subjects also maintained their usual dietary protein intake, which was estimated by using the 3-d food diary and the measurement of the protein equivalent of nitrogen appearance in a 24-h collection of dialysate and urine. The diets, provided as a 2-d diet cycle, with identical estimated nutrient content, were prepared and fed on alternate days during the entire period of hospitalization. All subjects were treated with at least twelve 800-mg tablets of sevelamer hydrochloride as phosphate binder; the dose for each subject was kept constant during the entire period of the study.

FIGURE 1.

Summary of the key procedures and outcome measures for the outpatient and inpatient phases of the trial. GCRC, General Clinical Research Center.

Arterial blood gas analysis was performed on admission (day 0), to determine the baseline pH. Ammonium chloride capsules were used to lower and citric acid/sodium citrate (bicitra) was used to raise the arterial pH, as needed, during each of the 2 phases. To maintain the nitrogen intake constant, subjects who were randomly assigned to the low-normal pH phase first and required ammonium chloride were given an equimolar amount of ammonium acetate during the high-normal pH phase. Those who were randomly assigned to the high-normal pH phase first and, based on the baseline pH, were anticipated to require ammonium chloride during the low-normal phase, were given ammonium acetate initially. Arterialized venous blood was drawn every 2–4 d to titrate the dose of the medications to achieve the desired pH. The success of titration was monitored by arterial blood pH measurements every 10 d. The achieved arterial pH was defined as the one measured on the last day of the equilibrated period for nitrogen balances.

Outcome measures

The primary outcome measure was the net, daily nitrogen balance during the last 5 d (equilibrated) of each treatment arm. [¹³C]Leucine kinetics, anthropometric measures, and expression of mRNA in muscle biopsy specimens were secondary outcome measures.

Nitrogen balance was computed as the difference between nitrogen intake (diet and medications, including ammonium chloride and ammonium acetate) and output (dialysate, urine, and feces), after adjustment for changes in the body's urea nitrogen content. Three sets of each 2-d diet cycle for each subject and all medications taken by each subject during the study were separately homogenized to measure the daily nitrogen intake. The dialysate effluent, collected over ice, for each 24-h period was pooled, and after measurement of the total volume, an aliquot was taken to measure the nitrogen content. All urine and stool collected were refrigerated and each was pooled, generally over 5-d periods; the pools of fecal collection were homogenized for measurement of nitrogen content. Nitrogen was measured by the macro-Kjeldahl technique with a semiautomated technique (10). The net nitrogen balance was adjusted for changes in the body urea nitrogen content, calculated from the changes in serum urea nitrogen and body weight as previously described (11).

For each subject, anthropometric assessments were made on the day of admission and on the last day of each of the 2 final equilibrated periods of the higher and lower arterial pH levels. They included the measurement of skinfold thicknesses at 5 sites (suprailiac, triceps, biceps, subscapular, and midcalf) and midarm, and midcalf circumference. Skinfold thicknesses were measured by using the Holtain Skinfold Caliper (Holtain Ltd, Cronwell, United Kingdom). Midarm and midcalf muscle areas were calculated as described by Heymsfield et al (12).

[¹³C]Leucine studies were performed after a 10-h overnight fast as follows. Subjects underwent their usual nighttime peritoneal dialysis cycling and then emptied all the intraperitoneal dialysate 2 h before the start of the study. After an initial priming dose of [¹³C]leucine (0.55 mg/kg, 98% purity) and [¹³C]NaHCO₃ (0.11 mg/kg), a constant infusion of [¹³C]leucine (0.37 mg · kg⁻¹ · h⁻¹) was given for 10 h. Subjects fasted from 2000 the night before until they had completed the first 4 h of the infusion; over the next 6 h, they were fed their usual breakfast and lunch in small isonitrogenous and isocaloric portions at 30-min intervals. Indirect calorimetry was performed at hourly intervals throughout the [¹³C]leucine turnover study by using a Vmax Spectra series model V29n indirect calorimeter (Sensor Medics Corporation/VIASYS health care; Yorba Linda, CA). Serum and breath samples were collected at 30-min intervals during the last 2 h of the fasting (180–240 min) and feeding (480–600 min) phases of the study. ¹³CO₂ Enrichment of expired air and [¹³C]ketoisocaproate enrichment of plasma were measured at Metabolic Solutions laboratory (Nashua, NH). Leucine flux and oxidation and protein synthesis and degradation were calculated for the fasting and fed states, for each treatment arm, for each subject by using methods defined earlier (13).

Muscle biopsy samples were collected from the right vastus lateralis muscle at the end of the 4 h of the fasting phase of the leucine turnover study, as previously described (14). The tissue was processed to measure the mRNA for the following proteins (14): insulin-like growth factor (IGF) splice variants IGF-IEa and IGF-IEc, IGF-I receptor, IGF-II, IGF-II receptor, IGF binding protein-2, myostatin, and MAFBx (muscle-specific ubiquitin ligase atrogin-1).

Statistical analyses

A sample size of 8 subjects was chosen a priori based on anticipated differences in nitrogen balances during the equilibrated period for each goal arterial pH. The sample size provided us with an estimated 80% power to detect differences of 1.16 times the SD of change and 90% power to detect differences of 1.34 times the SD of change.

The results are expressed as means ± SDs or medians with interquartile ranges, as appropriate. The significance of differences was tested by using paired t tests or Wilcoxon's signed-rank test, as appropriate. All analyses were done by using SPSS version 16 (SPSS Inc, Chicago, IL). A P value < 0.05 was considered significant.

RESULTS

Subject flow and characteristics

The flow of subjects into the study protocol is summarized in Figure 2. Because the study design required paired comparisons to determine the metabolic effect of different levels of arterial pH, only the data from the subjects who completed the equilibrated phase at both levels of arterial pH were used for analysis. Of the 5 subjects who dropped out, 3 did so before starting the inpatient phase of the study, and 2 dropped out after starting the inpatient phase. The latter 2 subjects dropped out within the first week of the study, long before reaching the first equilibrated period. The total duration of the phase with a lower arterial pH goal was 21.6 ± 2.6 d, and the duration of the phase with the higher arterial pH goal was for 21.6 ± 3.2 d. The last 5 d of each arterial phase was considered to be the equilibrated phase of the study, and the primary outcome measure, nitrogen balance, reflects the average for this period (Figure 1). The achieved arterial pH, anthropometric, and biochemical measurements reported below refer to the measurements made at the end of this equilibrated period.

FIGURE 2.

Chart summarizing the flow of patients throughout the clinical trial.

The key characteristics of the study cohort and the peritoneal dialysis prescription are summarized in Table 1. None of the 3 diabetic subjects were receiving insulin therapy. The dietary intakes of the study subjects are summarized in Table 2. There was no significant effect of the sequence of intervention on the results; therefore, the results are grouped by the achieved arterial pH.

TABLE 1.

Demographic, clinical, and peritoneal dialysis therapy characteristics of the 8 study subjects

Characteristic	Value
Age (y)	43.1 ± 15.31
Sex (M/F)	5/3
Race-ethnicity (n)
Latino	6
Asian	1
Black	1
History of diabetes mellitus (n)2	3
Peritoneal dialysis vintage (mo)	19.0 (58.4)3
4-h Dialysate/plasma creatinine ratio	0.73 ± 0.11
Weekly total Kt/V urea	2.30 ± 0.40
Weekly peritoneal Kt/V urea	2.15 ± 0.44
Nighttime cycling duration (h)	9.6 ± 0.8
Nighttime cycles (n)	4.8 ± 1.0
Daytime exchanges (n)	1.4 ± 0.7
Dialysate volume, nighttime (L)	11.4 ± 1.6
Dialysate volume, daytime (L)	3.4 ± 1.9
Dialysate volume, daily (L)	14.8 ± 2.6

¹Mean ± SD (all such values).

²Not requiring insulin.

³Median (interquartile range).

TABLE 2.

Summary of energy and protein intakes of the 8 study subjects during the inpatient nitrogen balance study¹

	Value
Dietary energy intake (kcal/d)	1687 ± 324
Energy intake from peritoneal dialysate (kcal/d)	498 ± 127
Total energy intake (kcal/d)	2220 ± 345
Dietary protein intake (g/d)2	74 ± 17
Dietary energy intake (kcal · kg−1 · d−1)	22.8 ± 5.2
Energy intake from peritoneal dialysate (kcal · kg−1 · d−1)	6.8 ± 2.2
Total energy intake (kcal · kg−1 · d−1)	30.0 ± 6.3
Dietary protein intake (g · kg−1 · d−1)2	0.98 ± 0.13

¹All values are means ± SDs.

²Does not include nitrogen provided from medications, including ammonium salts.

Achieved arterial pH

Of the 8 study subjects, 3 were randomly assigned to a goal arterial pH of 7.36–7.38 during the first half of the study; the remaining 5 were randomly assigned to this goal during the second half. The mean achieved arterial pH during this phase was 7.37 ± 0.01. To achieve this goal pH, 4 subjects required treatment with ammonium chloride (final dose: 40.3 ± 21.5 mEq/d), and 4 required treatment with sodium citrate/citric acid (final dose: 20.5 ± 12.9 mEq/d).

The mean achieved arterial pH during the high-normal pH phase was 7.44 ± 0.02. To increase the arterial pH, 7 of the 8 study subjects required treatment with sodium citrate/citric acid (final dose: 78.7 ± 30.9 mEq/d). In one subject, the arterial pH was spontaneously in the target range and, thus, did not require the administration of sodium citrate/citric acid.

Nitrogen balances

The total nitrogen intakes of the subjects were nearly identical during both phases of the study. The higher average arterial pH of 7.44, as compared with 7.37, was associated with a significant decrease in dialysate nitrogen output but a slight increase in urinary nitrogen losses. After adjustment for changes in the body urea nitrogen content, the higher arterial pH was associated with a trend toward a more positive nitrogen balance (P = 0.06; Table 3).

TABLE 3.

Summary of nitrogen balances for the 8 study subjects

	Low-normal pH	High-normal pH	P value1
Final arterial pH	7.37 ± 0.012	7.44 ± 0.02	<0.001
Intake (g N/d)
Diet	12.57 ± 1.88	12.54 ± 1.79	0.73
Medications3	2.31 ± 0.41	2.32 ± 0.42	0.32
Total	14.88 ± 1.89	14.86 ± 1.80	0.82
Output (g N/d)
Dialysate	9.21 ± 3.30	7.99 ± 2.73	0.04
Urinary4	0.04 (0.21)5	0.12 (1.00)	0.02
Fecal	3.01 ± 0.67	2.66 ± 0.87	0.26
Total	12.64 ± 3.15	11.35 ± 1.75	0.06
Adjustment in nitrogen retention for changes in serum urea nitrogen and body weight (g N/d)	0.07 (0.54)	0.11 (0.63)	0.26
Net nitrogen balance (g/d)	2.29 ± 2.18	3.22 ± 1.37	0.06

¹Significance of difference between means tested by paired t test and of medians by Wilcoxon's signed-rank test.

²Mean ± SD (all such values).

³Includes ammonium chloride and/or ammonium acetate in the 4 patients who received them.

⁴Six of the 8 subjects had measurable urine outputs.

⁵Median; interquartile range in parentheses (all such values).

In all but one study subject, the higher arterial pH was associated with more positive nitrogen balances (Figure 3). After the data for this outlier subject were excluded, a higher arterial pH was associated with a significantly higher net positive nitrogen balances (P = 0.004; Table 4).

FIGURE 3.

Nitrogen (N) balances during the low-normal and high-normal arterial pH phases (P) for each of the individual patients enrolled in the clinical trial. In 7 of the 8 subjects, the net nitrogen balance was higher during the high-normal arterial pH phase of the study than during the low-normal arterial pH phase.

TABLE 4.

Summary of nitrogen balances for 7 study subjects, after exclusion of the outlier

	Low-normal pH	High-normal pH	P value1
Final arterial pH	7.37 ± 0.012	7.44 ± 0.01	<0.001
Intake (g N/d)
Diet	12.62 ± 2.03	12.58 ± 1.93	0.74
Medications3	2.34 ± 0.43	2.35 ± 0.44	0.30
Total	14.96 ± 2.03	14.93 ± 1.93	0.83
Output (g N/d)
Dialysate	9.67 ± 3.27	8.17 ± 2.89	0.02
Urinary4	0.03 (0.24)5	0.03 (0.52)	0.03
Fecal	3.13 ± 0.64	2.80 ± 0.83	0.37
Total	13.26 ± 2.81	11.61 ± 1.70	0.02
Adjustment in nitrogen retention for changes in serum urea nitrogen and body weight (g N/d)	0.12 (0.55)	0.18 (0.67)	0.40
Net nitrogen balance (g/d)	1.69 ± 1.48	2.97 ± 1.25	0.004

¹Significance of difference between means tested by paired t test and between medians by Wilcoxon's signed-rank test.

²Mean ± SD (all such values).

³Includes ammonium chloride and/or ammonium acetate in the 4 patients who received them.

⁴Six of the 8 subjects had measurable urine outputs.

⁵Median; interquartile range in parentheses (all such values).

Leucine turnover studies

Throughout the leucine turnover studies, the subjects had no intraperitoneal dialysate. During the fasting state, leucine flux, protein synthesis, and protein degradation were significantly greater at the higher pH level (Table 5). However, leucine oxidation and protein balance were not different between the 2 pH levels. During the postprandial state, there was no difference in any of the kinetic parameters between the 2 pH levels (Table 5). As expected, at both pH levels, there was a marked increase in protein balance in the postprandial state.

TABLE 5.

Summary of [¹³C]leucine turnover studies in the 8 study subjects¹

	Low-normal pH	High-normal pH	P value2
	μmol · kg−1 · h−1	μmol · kg−1 · h−1
Final arterial pH	7.37 ± 0.01	7.44 ± 0.02	<0.001
Fasting
Leucine flux	77.7 ± 10.4	84.0 ± 13.8	0.02
Leucine oxidation	10.0 ± 2.6	10.8 ± 4.7	0.54
Protein synthesis	67.7 ± 9.0	73.3 ± 12.0	0.01
Protein degradation	77.7 ± 10.4	84.0 ± 13.8	0.02
Protein balance	−10.0 ± 2.6	−10.8 ± 4.7	0.54
Postprandial
Leucine flux	90.2 ± 17.4	95.9 ± 13.7	0.17
Leucine oxidation	15.9 ± 3.5	17.2 ± 6.2	0.63
Protein synthesis	73.6 ± 16.4	78.7 ± 13.0	0.23
Protein degradation	29.0 ± 24.9	35.2 ± 15.2	0.22
Protein balance	44.6 ± 11.7	43.5 ± 7.3	0.77

¹All values are means ± SDs.

²Significance of difference between means tested by paired t test.

Arterial blood gas measurements were not performed on the day of the [¹³C]leucine kinetic studies; however, arterialized venous blood pH was measured in 6 of the 8 subjects in 9 of the 16 turnover studies. Six of these 9 were paired measurements in 3 subjects. The achieved arterialized pH values were 7.36 ± 0.03 and 7.43 ± 0.03 during the 2 phases of the study.

Biochemical and anthropometric measures

The biochemical measurements at baseline and at the end of each of the 2 arterial phases are summarized in Table 6. The higher arterial pH was associated with a significantly greater urine volume (median urine volume in 6 of 8 subjects who had a urine output: 129.8 ± 534.3 compared with 26.1 ± 209.7 mL/d; P = 0.03). The higher arterial pH was associated with significantly lower serum urea nitrogen concentrations; however, there was no significant change in serum creatinine concentrations. There was no significant change in any of the other biochemical measurements. The anthropometric data are summarized in Table 7. There was no significant effect of change in arterial pH on any of the measurements performed.

TABLE 6.

Summary of urine and serum biochemical measures at baseline and at the end of the 2 study periods¹

	Baseline	Low-normal pH	High-normal pH	P value2
Final arterial pH	—	7.37 ± 0.013	7.44 ± 0.02	<0.001
Urine volume (mL/d)4	—	16.9 (84.4)5	41.1 (335.3)	0.03
Urinary urea nitrogen (mg/d)4	—	15.3 (129.5)	78.2 (668.3)	0.25
Urinary urea clearance (mL/min)4	—	0.02 (0.14)	0.09 (1.02)	0.03
Urinary creatinine clearance (mL/min)4	—	0.08 (0.51)	0.16 (1.4)	0.35
Urine protein excretion (mg/d)4	—	47 (143)	148 (457)	0.12
Venous bicarbonate (mEq/L)	23.0 ± 2.6	18.9 ± 2.2	24.6 ± 2.6	0.002
Arterial pCO2 (mm Hg)	37.5 ± 3.6	35.9 ± 3.0	41.8 ± 3.7	0.001
Serum urea nitrogen (mg/dL)	55.0 ± 17.1	64.4 ± 20.2	54.1 ± 13.7	0.01
Serum creatinine (mg/dL)	13.7 ± 4.6	13.6 ± 4.1	12.9 ± 3.5	0.18
Serum potassium (mEq/L)	4.2 ± 0.8	4.3 ± 0.7	3.9 ± 0.8	0.28
Serum phosphorus (mg/dL)	6.3 ± 1.8	5.7 ± 1.2	5.5 ± 1.1	0.48
Serum parathyroid hormone (pg/mL)6	337 (273)	224 (242)	318 (252)	0.21
Serum albumin (g/dL)6	3.5 ± 0.6	3.1 ± 0.4	3.1 ± 0.4	0.12
Serum prealbumin (g/dL)	44.7 ± 10.6	40.7 ± 9.5	39.9 ± 7.5	0.72
Serum transferrin (mg/dL)	233.5 ± 57.8	206.6 ± 35.2	217.0 ± 73.8	0.57
Serum insulin-like growth factor I (μg/L)	192 (143)	189 (173)	159 (191)	0.89
Serum C-reactive protein (mg/L)	3.2 ± 2.7	2.1 ± 1.4	1.8 ± 1.2	0.37
Serum tumor necrosis factor (pg/mL)	5.8 ± 1.7	6.4 ± 1.8	5.5 ± 1.3	0.09
Serum interleukin-6 (pg/mL)	3.8 ± 2.1	3.3 ± 1.8	2.6 ± 1.3	0.09

¹pCO₂, partial pressure of carbon dioxide.

²Significance of difference (end of low-normal and high-normal pH phases) between means tested by paired t test and between medians by Wilcoxon's signed-rank test.

³Mean ± SD (all such values).

⁴Values are the means for the last 5 d of each phase; 6 of the 8 study subjects had measurable urine outputs.

⁵Median; interquartile range in parentheses (all such values).

⁶Measured by using bromocresol purple during the inpatient phase of the study.

TABLE 7.

Summary of anthropometric measures for the 8 study patients at baseline and at the end of the 2 study periods

	Baseline	Low-normal pH	High-normal pH	P value1
Final arterial pH	—	7.37 ± 0.012	7.44 ± 0.02	<0.001
Body weight (kg)	74.9 ± 10.3	75.2 ± 11.2	76.7 ± 10.0	0.08
Suprailiac skinfold thickness (mm)	18.3 ± 8.0	18.2 ± 7.8	18.0 ± 7.4	0.78
Triceps skinfold thickness (mm)	22.7 ± 7.2	21.6 ± 6.8	22.0 ± 7.2	0.63
Biceps skinfold thickness (mm)	8.0 (3.8)3	7.8 (1.4)	8.2 (3.1)	0.26
Subscapular skinfold thickness (mm)4	19.6 ± 7.3	20.9 ± 6.5	20.8 ± 7.1	0.96
Midcalf skinfold thickness (mm)	9.2 (11.9)	8.9 (13.3)	8.3 (11.9)	0.50
Midarm circumference (cm)	33.5 ± 3.0	33.1 ± 3.1	33.0 ± 2.9	0.75
Midarm muscle area (cm2)	47.1 ± 8.9	46.9 ± 10.6	45.8 ± 10.2	0.54
Midcalf circumference (cm)4	35.2 ± 3.2	35.3 ± 2.8	35.2 ± 3.6	0.97
Midcalf muscle area (cm2)4	70.3 ± 16.0	70.1 ± 15.4	68.2 ± 20.1	0.51

¹Significance of difference (end of low-normal and high-normal pH phases) between means tested by paired t test and between medians by Wilcoxon's signed-rank test.

²Mean ± SD (all such values).

³Median; interquartile range in parentheses (all such values).

⁴Mean of values for 7 study subjects; the skin of one subject each was too tight in the subscapular and midcalf areas to allow for measurement of skinfold thicknesses in that area.

Muscle biopsy

Muscle biopsy samples were collected on the day of the leucine turnover study. The mRNA levels of various proteins in the skeletal muscle biopsy are summarized in Table 8; there was no significant effect of change in arterial pH on the mRNA content of any of the proteins studied. Even though the mean mRNA for myostatin at the high-normal arterial pH was 63% lower than that at the low-normal pH, the trend was not statistically significant (P = 0.07).

TABLE 8.

Summary of mRNA content in the quadriceps femoralis muscle of the 8 study subjects measured at the end of each study period¹

	Low-normal pH	High-normal pH	P value2
Final arterial pH	7.37 ± 0.013	7.44 ± 0.02	<0.001
Insulin-like growth factor IEa	20.5 (20.6)4	23.9 (23.3)	0.67
Insulin-like growth factor IEc	0.17 (0.24)	0.08 (0.15)	0.33
Insulin-like growth factor I receptor	10.8 (9.6)	13.0 (9.4)	0.67
Insulin-like growth factor II	256.7 (392.2)	192.6 (156.3)	0.67
Insulin-like growth factor II receptor	77.9 (79.0)	58.3 (43.7)	0.16
Insulin-like growth factor binding protein-2	66.8 (50.8)	67.2 (49.0)	0.89
Myostatin	127.2 (271.2)	47.1 (185.4)	0.07
MAFBx	133.1 (94.2)	85.5 (88.0)	0.40

¹MAFBx, muscle-specific ubiquitin ligase atrogin-1.

²Significance of difference between means tested by paired t test and between medians by Wilcoxon's signed-rank test.

³Mean ± SD (all such values).

⁴Median; interquartile range in parentheses (all such values).

Adverse events

The first study subject developed acute bacterial peritonitis during the second part of the study (goal arterial pH: 7.43–7.45). All study procedures were suspended for 7 d and resumed after all evidence of intraperitoneal and systemic inflammation had subsided. None of the other subjects experienced any adverse events.

DISCUSSION

The adverse metabolic consequences of metabolic acidosis and its contribution to protein wasting in maintenance dialysis subjects have been well documented. This study, however, questions the premise that the entire range of the reference arterial pH (7.37–7.44) is equivalent for protein balance. In an otherwise healthy cohort of subjects with end-stage renal disease undergoing automated peritoneal dialysis, a mean arterial pH of 7.44 was associated with higher net-positive nitrogen balances in 7 of 8 study subjects. This suggests that a more aggressive improvement of systemic acid-base status in the direction of the upper range of the clinically adjudged normal range of arterial pH, and perhaps beyond that upper range, may be beneficial for the amelioration and/or prevention of protein wasting in maintenance dialysis patients.

The study environment was very carefully regulated, and all attempts were made to keep everything related to the subject's medical treatment, nutritional intake, and living environment, but not the arterial pH, constant during both phases of the study. The subjects were fed an isonitrogenous diet during both phases of the study. The strength and dose formulation of all medications, (other than the interventions) were kept constant during the entire study. This was particularly essential for sevelamer hydrochloride (used as a phosphate binder), which has 125 mg N per 800-mg tablet. Furthermore, the higher nitrogen intakes with ammonium chloride in 4 of the 8 study subjects during the low-normal phase of the study were compensated for by an equivalent intake of ammonium acetate during the high-normal phase of the study. The daily energy intakes were also kept constant, not only from the diets but also from the peritoneal dialysate; the prescription (tonicity, volume, and dwell time) was identical for each day for the entire study. During the high arterial pH phase, the subjects had a higher urine volume and slightly but significantly greater urinary nitrogen losses. This was offset by a much larger decrease in dialysate nitrogen output. This, in conjunction with constant nitrogen intakes, the high-normal arterial pH was associated with higher net positive daily nitrogen balances in all but one study subject. Neither the magnitude nor the direction of change was affected by the sequence of intervention. The thesis that these greater net positive nitrogen balances reflect greater anabolism at high-normal arterial pH values is supported by significantly lower serum urea nitrogen concentrations during this phase of the study. Even though the urinary urea clearances increased >4-fold during the high-normal arterial pH phase of the study (from 0.02 to 0.09 mL/min), the clearances were still too low to account for the substantially lower serum urea nitrogen concentrations at the higher pH level.

There was only one study subject in whom this trend of higher net positive nitrogen balances during the high-normal arterial pH was not observed. This subject, a healthy young male, had substantially positive nitrogen balances during the entire 6-wk course of the study, but body weight remained stable. Despite the rigorous methodology, nitrogen balance studies do not measure several sources of daily nitrogen loss. These include, but are not limited to, respiratory losses, integumentary growth and shedding, perspiration, and ejaculate. There are no published data on the magnitude of respiratory losses in dialysis patients. With higher circulating concentrations of nitrogenous substances, it is likely that these respiratory losses may be substantially higher than in the general population. Thus, it is possible that this subject had substantially higher respiratory losses, which may have varied differently during the 2 intervention phases of the study. However, we did not measure respiratory losses, and this possibility remains speculative.

It needs to be noted that the fecal nitrogen losses were significantly higher in our study subjects than has previously been reported. This was accounted for by the use of sevelamer hydrochloride—the drug binds phosphorus in the gastrointestinal tract, is not absorbed, and thus, appears in the stool. Because each study subject took ≥12 tablets of the drug daily, it accounted for ≥1.5 g fecal nitrogen/d.

Despite the higher net positive nitrogen balances, we were unable to demonstrate higher net protein balances with the leucine turnover studies at the end of the high-normal arterial pH phase. There are potentially 2 reasons to explain this discrepancy. First, even though the subjects did not eat any food for >10 h before the start of the turnover study, peritoneal dialysis was continued until ≈2 h before. Thus, subjects continued to receive a carbohydrate load, possibly with increased circulating serum insulin concentrations, until shortly before the turnover study commenced. Second, the maintenance of arterial pH during the 2 equilibrated phases of the study was achieved, in part, by maintaining the acid intake (diet and phosphate binder) and dialytic removal constant. Peritoneal dialysis provides slow, but continuous, absorption of lactate, which is then metabolized to bicarbonate. However, peritoneal dialysis therapy was interrupted for the entire course of the turnover study. It follows, then, that the subjects may not have had the same arterial pH during the turnover study, as during the equilibrated phase for nitrogen balances. Other investigators have shown increased catabolism and decreased anabolism by using [¹³C]leucine kinetics, at lower pH levels, compared with normal pH (reviewed in reference 1). However, unlike this study, the lower arterial pH in those studies was in the acidemic range, and the turnover studies were performed more acutely. In our study, the response to changes in arterial pH was measured after 20 d, and there may have been some metabolic adjustments over that period of time, accounting for the apparently discrepant results.

We were unable to identify the pathophysiologic mechanisms that may have led to the improvement in nitrogen balances with the high-normal arterial pH. Two important trends were, however, identified. First, serum tumor necrosis factor-α and interleukin-6 tended to be lower with higher arterial pH. This is consistent with a previous observation by Pickering et al (15), in which treatment of 8 stable continuous ambulatory peritoneal dialysis patients with dialysis solutions with a higher lactate concentration was associated with a higher serum bicarbonate concentration and a significantly lower serum tumor necrosis factor α concentration. Second, at the end of the high-normal arterial pH phase, the muscle expression of myostatin was almost one-third of what was seen at the end of low-normal arterial pH phase. Myostatin is an antianabolic protein in the muscle and is down-regulated with aerobic exercise training in maintenance dialysis patients (14). It is possible that down-regulation of myostatin may be, in part, responsible for the anabolic effects of high-normal arterial pH.

Both the Nutrition and the Bone Metabolism and Disease Working Groups of the National Kidney Foundations' Kidney Disease Outcome Quality Initiative recommend a serum bicarbonate concentration ≥22 mEq/L for maintenance dialysis subjects (16, 17). The mean serum bicarbonate concentration in our cohort during the high-normal arterial pH phase was 25 mEq/L, significantly higher than 22 mEq/L. In a randomized controlled trial, bicarbonate supplementation for peritoneal dialysis patients with venous bicarbonate concentrations ≤25 mmol/L was associated with a higher subjective global nutritional assessment score (indicating a lesser degree of protein-energy wasting), an increased normalized protein equivalent of nitrogen appearance, and shorter hospitalizations (18). These data suggest that, in some subjects, higher concentrations (≥25 mEq/L) may have greater metabolic benefits than concentrations of 22 mEq/L. However, we did not directly compare the 2 levels of serum bicarbonate (22 compared with 25 mEq/L), and this remains speculative.

We chose to study subjects with end-stage renal disease because metabolic acidosis is a frequent accompaniment of this disease state. Even though most patients with end-stage renal disease are treated with maintenance hemodialysis, the intermittent nature of therapy (performed thrice weekly for 3 to 4 h at a time) does not allow for steady metabolic control. On the other hand, peritoneal dialysis is performed daily, often continuously, and this makes it feasible to achieve a steady metabolic state. This formed the rationale for the selection of the study population. Our findings are consistent with those previously reported in postmenopausal women in whom neutralization of endogenous acid production with potassium bicarbonate was associated with a reduction in urinary nitrogen losses (19). This suggests that the implications of our findings extend beyond the population that we studied and raises a fundamental question as to what range of arterial pH constitutes a pathogenetically important degree of metabolic acidemia (20). The studies of the effect of pH on protein synthesis in cultured myocytes also leave open the possibility that an arterial pH even higher than 7.44 may lead to more positive nitrogen balances. The advantages of a higher arterial pH may extend beyond nitrogen balances—it may allow patients to liberalize their intake of potassium-rich foods, and hence higher intakes of antioxidants, and phytochemicals present in such foods. The liberalization of intakes may also improve a patient's quality of life and muscle function. However, these potential benefits would need to be balanced against the risk of respiratory suppression with higher arterial pH. However, we did not study these effects, and thus, they remains speculative.

This study had several limitations. First, only 8 subjects were studied. Second, we compared only 2 levels of arterial pH and did not study intermediate pH levels. It is possible that it may not be necessary to increase the arterial pH to 7.44 in all patients to reduce protein wasting. Third, we cannot exclude the possibility that the changes noted herein were the direct result of the interventions (ammonium chloride and sodium citrate/citric acid) and not secondary to changes in arterial pH. Finally, chronic ingestion of ammonium chloride and sodium citrate/citric acid may lead to changes in total body acid content, which may not be completely reflected in either the arterial pH or serum bicarbonate values (21).

To conclude, this is one of the few studies that used arterial pH as a measure of a subject's acid-base status. Our study showed that a high-normal arterial pH is associated with greater net positive nitrogen balances than is a low-normal arterial pH in most, otherwise stable patients with end-stage renal disease undergoing automated peritoneal dialysis. This suggests that the entire “normal” range of arterial pH may not be equivalent, and a more aggressive correction of metabolic acidosis to an arterial pH of 7.44 may be better at ameliorating and/or preventing protein-energy wasting in maintenance dialysis patients.