Introduction

Sarcopenia is characterized by a loss of muscle mass and strength, a decreased quality of life (QoL) with age, morbidities, and immobility and is associated with protein-energy wasting in chronic kidney disease (CKD) and end-stage kidney disease patients.1 The prevalence of sarcopenia is increased with declining kidney function.1, 2 CKD is associated with chronic low-grade inflammation leading to progressive weight loss, muscle weakness, and the loss of the ability to exercise. Inflammatory cytokines, oxidative stress, and the inactivity-mediated destruction of protein homeostasis result in the catabolic destruction of structural and functional proteins, skeletal muscle wasting, and a decrease in exercise capacity.3, 4 Mitochondrial dysfunction of skeletal muscles in CKD is also considered a cause of the loss of muscle mass and exercise capacity.5-7

Uraemic toxins enter target cells via specific transporters, such as organic anion transporters (OATs).8-10 IS also enters various cells via OATs (OAT1 and OAT3), and OATs are also expressed in muscles.11-13 Protein-bound uraemic toxins, including IS and p-cresyl sulfate, exert their toxicity via the activation of cellular NAD(P)H oxidase, which results in the overproduction of reactive oxygen species (ROS) and inflammatory cytokines.14, 15 IS accumulation in muscle cells and subsequent superoxide production and the up-regulation of inflammatory cytokines such as tumour necrosis factor (TNF)-α, interleukin (IL)-6, and transforming growth factor-β induce muscle wasting through myostatin and atrogin-1.13 And these changes are associated with mitochondrial dysfunction, which is mediated by metabolic alterations, that is, the up-regulation of antioxidative responses and down-regulation of energy-generation pathways.16

A significant inverse association between plasma IS levels and skeletal muscle mass has been found in CKD patients.16 In addition, IS down-regulates Klotho expression through ROS-associated nuclear factor-kB activation in the kidney.17, 18 Klotho levels and skeletal muscle physiology are closely related.19, 20

AST-120 reduces the accumulation of IS in organs, including skeletal muscles21; increases Klotho expression; and inhibits cell senescence in uraemic mouse and rat.22 AST-120 in in vivo and in vitro experiments also prevented the CKD-induced physical inactivity mainly by maintaining mitochondrial function, suppressing atrogin-1/myostatin expression, and recovering Akt phosphorylation in skeletal muscles.23

However, the effect of AST-120 on sarcopenia in CKD patients has never been studied. Therefore, we aimed to determine the effects of AST-120 on sarcopenia and sarcopenia-associated factors in CKD patients.

Materials and methods Study design

This was a 48 week, randomized controlled, parallel group, open-label, multicentre trial. This study was registered in clinicaltrials.gov on 27 December, 2018 [RolE of AST-120 in sarCOpenia preVEntion in pRe-dialYsis chronic kidney disease patients (RECOVERY): NCT03788252]. The participants were randomly assigned in a 1:1 ratio to the control (CON) and AST-120 (Renamezin®, REN) groups. Measurements were taken at baseline and every 24 weeks for 48 weeks; the measurements included vital sign recordings, blood and urine laboratory examination findings, body composition including bioelectric impedance study findings, physical performances, and questionnaire responses. This study was approved by the institutional review boards of the participating hospitals. We conducted this study in compliance with the principles of the Declaration of Helsinki. All participants provided informed consents.

Study outcomes

The primary outcome was physical performance (gait speed difference ≥0.1 m/s between the two groups) in CKD patients. We chose the gait speed as the primary outcome because of following reasons24: (1) it is hard to reverse muscle mass due to old age and advanced kidney dysfunction of participants; (2) gait speed serves as a core indicator of health and function in ageing and diseases; (3) gait speed is a quick and reliable measure of functional capacity with high interrator and test–retest reliability. The secondary outcomes included muscle mass and strength, physical activity level, levels of inflammatory and muscle-related markers (IS, TNF-α, IL-6, and myostatin), health-related QoL, and renal function [serum creatinine (SCr) and estimated glomerular filtration rate (eGFR)].

Recruitment and population

A total of 150 CKD patients were recruited from six general hospitals in Korea.

Patients were eligible for inclusion in this study only if the patient (1) had pre-dialysis CKD and was aged 20 years or older; (2) had stable renal function with an SCr level between 2.0 and 5.0 mg/dL or MDRD (or CKD-EPI) eGFR between 15 and 60 mL/min/1.73 m2 for 3 months; (3) had a serum albumin level higher than 3.0 g/dL; (4) was naïve to AST-120 during the 4 weeks before screening; (5) did not have significant changes in medical treatment within the 4 weeks before screening; (6) could ambulate without any help from caregivers except for using auxiliary devices; and (7) could thoroughly understand the protocol and sign the informed consent form.

Patients were excluded for any of the following reasons: (1) they had passage disorders in the gastrointestinal tract and uncontrolled constipation; (2) they were kidney transplantation recipients or were expected to undergo dialysis or kidney transplantation within 3 months after enrolment; (3) they were taking immune-suppressive agents; (4) they had active ulcers or oesophageal varices; (5) they had an uncontrolled blood pressure of ≥180/110 mmHg; (6) they had acute or subacute cardiovascular diseases within the last 3 months; (7) they had active infections or uncontrolled inflammatory diseases; (8) they had abnormal aspartate- and alanine- aminotransferase >2.5 times the upper normal limit; (9) they had an uncontrolled blood sugar level (fasting glucose >250 mg/dL or HbA1c > 10.0%); (10) they had a progressive malignancy; (11) they were pregnant, lactating, or planning to be pregnant during the study period; (12) they had severe retinopathies such as proliferative diabetic retinopathy or vitreal haemorrhage; (13) they had orthopaedic disorders that can be aggravated by physical activity; (14) they underwent leg amputation but did not wear a prosthetic legs or they showed claudication; (15) they were participating in other clinical studies; and (16) they were considered ineligible for the study by the investigators.

Randomization and interventions

The patients who were enrolled were randomly assigned at a ratio of 1:1 to the CON and REN groups. We performed the block randomization (6 blocks of size 25 for 1:1 allocation). An independent statistician used SAS (version 9.4) to generate random numbers, and only an independent physician had the pregenerated codes. All the participants received standard care, including angiotensin-converting enzyme inhibitors and/or angiotensin receptor blockers and lipid modifiers. The participants in the REN group were instructed to self-administer the drug orally in three divided doses (Renamezin® 21 capsules a day) for a total of 6 g a day. Designated researchers counted the residual capsules of AST-120 at every visit to assess compliance.

Sample size

The sample size was estimated such that a mean between-treatment group difference in gait speed of ≥0.1 m/s could be detected. We assumed a standard deviation (SD) of gait speed of 0.2 m/s. Assuming a two-tailed hypothesis, an alpha value of 0.05 and a desired power of 80%, 50 participants were found to be needed per group to complete the study. Allowing for a 34% dropout rate, we calculated the required sample size to be 75 patients per treatment group.

Measurements Six-metre gait speed test

The participants walked at their usual walking speed after the examiner instructed to do so. The participants could use walking aids such as walking sticks, canes, or walkers. We adopted both static and dynamic start methods. For the static start method, the participants stood on a 6 m start line and started walking after examiner instructed to do so. For the dynamic start method, the participants started walking 2 m prior to the measurement point (acceleration zone) and ended 2 m after the measurement point (deceleration zone). The participants repeated each type of walking speed test two times, and all data including the mean were analysed.

Hand grip strength

Hand grip strength (HGS) was assessed by using digital hand dynamometers (T.K.K.5401, Takei Scientific Instruments Co. Ltd., Niigata, Japan) in both hands. The participants measured HGS in two positions: sitting and standing. The participants stood upright with the shoulders facing forward without rotation, the elbows extended, and the wrists in neutral flexion for the standing position measurements. The participants were seated, with the shoulders facing forward without rotation, the elbows flexed to 90°, and the wrists in neutral flexion for the sitting position HGS measurements. The participants were encouraged to grasp the device strongly three times at intervals of 30 s in each position and each hand. The largest HGS value for each method was used in the analysis.

Body composition

The participants were fasted at least 8 h before the body composition test but could take essential medications with a small amount of water (<100 mL). Body composition was measured by using an InBody S10 (Seoul, South Korea), an instrument based on bioelectrical impedance analysis.

Laboratory tests

Parameters including IS, TNF-α, IL-6, myostatin, intact parathyroid hormone (iPTH), and 25-OH-vitamin D levels in serum were measured at the central laboratory institution (Seoul Clinical Laboratories, Yongin-si, Gyeonggi-do, Korea). The TNF-α, IL-6, and myostatin levels were assessed by an enzyme-linked immunosorbent assay (ELISA) using HSTA00D (Human TNF-α Quantikine HS ELISA), HS600C (Human IL-6 Quantikine HS ELISA Kit), and DGDF80 (GDF-8/Myostatin Quantikine ELISA Kit, R&D Systems). The levels of iPTH and 25-OH-vitamin D were measured by electrochemiluminescence immunoassay and chemiluminescence immunoassay using a Cobas E801 analyser (Roche Diagnostics GmbH, Mannheim, Germany). The serum total IS levels were measured using a high-performance liquid chromatography-fluorescence detector (HPLC-FLD, Agilent 1260 series; Agilent Technologies, Santa Clara, CA, USA). Other blood and urine data were measured at each research institution.

Questionnaire survey

We used the Charlson comorbidity index to assess the comorbidity status of the participants. We used version 1.3 of the kidney disease QoL (KDQOL) short form to evaluate health-related QoL. A short, self-administered version of the international physical activity questionnaire (IPAQ) was used to assess the physical activity level of the individual over the past 7 days.

Statistical methods

We performed the primary analysis by intention-to-treat (ITT; including data on the patients who were randomly assigned to a group and underwent any study outcome evaluations) and per-protocol (PP; including data on the patients who were randomly assigned to a group and completed all the data collections without major protocol deviations) methods. We conducted all statistical analyses by using PASW advanced statistics (SPSS Inc, Chicago, IL, USA) version 20.0. The data are reported as the mean, standard deviation, or percentage frequency.

We used Student's t-test (or Mann–Whitney U tests) and paired t tests (or Wilcoxon signed-rank tests) for the continuous variables depending on whether the data were normally distributed, and the χ2chi-square test for the categorical variables. The between-group differences in the outcome measures after the intervention were assessed by using repeated measure analysis of variance. Multiple regression analysis was conducted to determine the statistical significance of each variable with respect to the outcomes. The Spearman rank test or Pearson product–moment correlation coefficient analysis was used to analyse the associations between the clinical data and outcome measures. We considered P < 0.05 (two sided) statistically significant.

Results Study population and baseline characteristics

The first participant was enrolled on November 11, 2018, and the last follow-up was performed on June 16, 2020. A total of 150 patients were randomly assigned to the CON and REN groups and underwent any study outcome evaluations (ITT population). In total, 124 patients completed the follow-up and all the study outcomes (PP population) (Figure 1). The mean AST-120 compliance rates of the ITT and PP populations were 85.9% and 87.3%, respectively. The baseline characteristics were not significantly different between the two treatment groups in the ITT analysis (Table 1) and PP analysis.

Flow chart of study participant enrolment, randomization, and analysis.

Table 1. Baseline characteristics of intention-to-treat population Total (n = 150) CON (n = 75) REN (n = 75) P valuea Age (years) 65.0 ± 10.8 65.9 ± 10.7 64.1 ± 10.8 0.318 Sex (men) 97 (64.7%) 52 (69.3%) 45 (60.0%) 0.232 Diabetes mellitus 76 (50.7%) 36 (48.0%) 40 (53.3%) 0.514 Modified CCI score 3.9 ± 1.9 3.8 ± 1.8 3.9 ± 2.0 0.667 Haemoglobin (g/dL) 12.3 ± 2.1 12.3 ± 2.0 12.4 ± 2.1 0.826 Albumin (g/dL) 4.3 ± 0.4 4.3 ± 0.4 4.3 ± 0.3 0.242 Calcium (mg/dL) 9.1 ± 0.6 9.1 ± 0.6 9.1 ± 0.5 0.707 Phosphorus (mg/dL) 3.5 ± 0.6 3.5 ± 0.6 3.5 ± 0.6 0.483 hs-CRP (mg/dL) 0.5 ± 1.2 0.4 ± 1.0 0.5 ± 1.4 0.599 Blood urea nitrogen (mg/dL) 32.0 ± 11.0 32.3 ± 11.8 31.6 ± 10.3 0.694 Creatinine (mg/dL) 2.1 ± 0.7 2.1 ± 0.7 2.1 ± 0.8 0.668 Total CO2 (mmol/L) 23.8 ± 3.5 23.9 ± 3.0 23.7 ± 3.9 0.806 Sodium (mmol/L) 140 ± 2 140 ± 2 140 ± 2 0.860 Potassium (mmol/L) 4.8 ± 0.6 4.9 ± 0.6 4.8 ± 0.6 0.651 Chloride (mmol/L) 106 ± 4 106 ± 4 106 ± 4 0.888 iPTH (ng/mL) 86.5 ± 67.9 88.8 ± 80.6 84.2 ± 52.6 0.683 eGFR (CKD-EPI) 33.8 ± 12.5 33.0 ± 12.0 34.6 ± 12.9 0.442 Pr/Cr ratio 1.3 ± 1.5 1.5 ± 1.7 1.2 ± 1.2 0.299 Data are expressed as numbers (percentages) for categorical variables and means ± standard deviations for continuous variables. The P values were tested with t test for continuous variables, and Pearson χ2 test or Fisher exact test was used to analyse categorical variables. Abbreviations: CCI, Charlson comorbidity index; CKD-EPI, Chronic Kidney Disease-Epidemiology Collaboration; eGFR, estimated glomerular filtration rate; hs-CRP, high sensitivity-C-reactive protein; iPTH, intact parathyroid hormone; Pr/Cr, spot urine protein to creatinine. a Between the CON and the REN. Indoxyl sulfate level

In the ITT analysis, the serum IS levels in the CON and REN groups were 0.50 ± 0.42 and 0.50 ± 0.46 mg/dL at baseline, 0.50 ± 0.43 and 0.41 ± 0.40 mg/dL at 24 weeks, and 0.42 ± 0.36 and 0.32 ± 0.32 mg/dL at 48 weeks, respectively (Figure 2A). The serum IS level decreased over 48 weeks in both groups, and the interaction of the treatment group with IS level was weak (Ptime < 0.001, PtimeXtreat = 0.065). However, in the PP analysis, the interaction of the treatment group with IS level was statistically significant (PtimeXtreat = 0.046).

Changes of indoxyl sulfate level (A), gait speed (B), and standing handgrip strength (C) from baseline to 24 and 48 weeks.Data were expressed as mean and standard error. *P < 0.05 vs. baseline, ¶P < 0.01 vs. baseline, #P < 0.05 vs. 24 weeks. ITT, intention-to-treat; PP, per-protocol.

Primary outcome—gait speed

The mean dynamic start gait speeds in the CON and REN groups are presented in Table 2 and Figure 2B. A difference of gait speed ≥0.1 m/s between the two groups was not observed during the study period (1.07 ± 0.28 and 1.04 ± 0.31 m/s at baseline, 1.11 ± 0.28 and 1.06 ± 0.31 m/s at 24 weeks, and 1.10 ± 0.30 and 1.08 ± 0.32 m/s at 48 weeks in the CON and REN groups, respectively). The mean dynamic-start gait speed in the REN group increased from baseline to 48 weeks (1.04 ± 0.31 to 1.08 ± 0.32 m/s, P = 0.019), and PP analysis showed similar statistically significant results (1.03 ± 0.30 to 1.08 ± 0.33 m/s, P = 0.018) (Figure 2B).

Table 2. Comparison of gait speed and handgrip strength between treatment groups Baseline 24 weeks 48 weeks CON REN P value CON REN P value CON REN P value ITT population Dynamic-start gait speed (mean, m/s) 1.07 ± 0.28 1.04 ± 0.31 0.484 1.11 ± 0.28 1.06 ± 0.31 0.310 1.10 ± 0.30 1.08 ± 0.32* 0.764 Static-start gait speed (mean, m/s) 1.05 ± 0.26 1.01 ± 0.22 0.398 1.05 ± 0.21 1.03 ± 0.22 0.494 1.04 ± 0.23 1.04 ± 0.22* 0.896 Dynamic-start gait speed (faster, m/s) 1.10 ± 0.29 1.06 ± 0.32 0.432 1.13 ± 0.28 1.08 ± 0.32 0.328 1.11 ± 0.31 1.10 ± 0.34* 0.873 Static-start gait speed (faster, m/s) 1.08 ± 0.28 1.05 ± 0.23 0.432 1.07 ± 0.22 1.05 ± 0.23 0.579 1.06 ± 0.25 1.07 ± 0.24* 0.820 Standing HGS (kg) 28.7 ± 8.75 28.6 ± 9.24 0.959 27.9 ± 7.85 27.0 ± 8.22* 0.495 28.6 ± 8.33 27.9 ± 8.67 0.657 Sitting HGS (kg) 28.6 ± 8.75 28.0 ± 9.07 0.719 27.5 ± 7.70 26.5 ± 8.06* 0.465 28.2 ± 7.93 27.5 ± 9.09 0.648 PP population Dynamic-start gait speed (mean, m/s) 1.08 ± 0.29 1.03 ± 0.30 0.324 1.10 ± 0.28 1.06 ± 0.31 0.475 1.10 ± 0.30 1.08 ± 0.33* 0.717 Static-start gait speed (mean, m/s) 1.06 ± 0.26 1.00 ± 0.21 0.217 1.05 ± 0.22 1.03 ± 0.22 0.632 1.04 ± 0.23 1.04 ± 0.23* 0.958 Dynamic-start gait speed (faster, m/s) 1.11 ± 0.30 1.05 ± 0.31 0.301 1.12 ± 0.29 1.08 ± 0.32 0.507 1.12 ± 0.31 1.11 ± 0.34* 0.825 Static-start gait speed (faster, m/s) 1.09 ± 0.28 1.03 ± 0.22 0.217 1.07 ± 0.22 1.06 ± 0.23 0.710 1.06 ± 0.25 1.07 ± 0.24* 0.889 Standing HGS (kg) 29.5 ± 8.80 28.7 ± 9.17 0.634 28.3 ± 7.69 27.3 ± 8.42* 0.513 28.9 ± 8.14 28.2 ± 8.68 0.668 Sitting HGS (kg) 29.2 ± 8.87 28.3 ± 7.69 0.463 27.7 ± 7.61 26.8 ± 8.23* 0.525 28.4 ± 7.73 27.7 ± 9.11 0.659 Data are expressed as means ± standard deviations. The comparison between groups was tested with t test and comparison with baseline value was tested with paired t test. * P < 0.05 vs. baseline value. Abbreviations: HGS, handgrip strength; ITT, intention-to-treat; PP, per-protocol.

AST-120 also showed a beneficial effect on the change of static-start gait speed during the study period. The faster static-start gait speed in the CON and REN groups changed over 48 weeks by −0.024 ± 0.204 and 0.04 ± 0.152 m/s (P = 0.049) according to the ITT analysis and −0.025 ± 0.205 and 0.038 ± 0.152 m/s (P = 0.058) according to the PP analysis, respectively. Multiple regression analysis showed that AST-120 tended to increase the static-start gait speed, although change was marginally statistically significant after adjustments for age, sex, primary disease, and physical activity (95% confidence interval 0.000–0.128, P = 0.051 and 95% confidence interval −0.003 to 0.128, P = 0.061 for ITT and PP analysis, respectively) (Table S1).

Secondary outcomes Handgrip strength and muscle mass

Hand grip strength at each time point is presented in Table 2 and Figure 2B. Standing HGS decreased from baseline to 24 weeks in both groups and did not change during the latter 24 weeks (28.7 ± 8.75 and 28.6 ± 9.24 kg at baseline, 27.9 ± 7.85 and 27.0 ± 8.22 kg at 24 weeks, and 28.6 ± 8.33 and 27.9 ± 8.67 kg at 48 weeks in the CON and REN groups, respectively). The interaction of the treatment group with standing HGS was not significant (PtimeXtreat = 0.875). Standing HGS changed similarly in men and all participants (Supporting Figure S1A). In women, there was no significant change in standing HGS over the study period (Figure S1B). PP analysis showed similar results. In addition, sitting HGS showed similar results in the two treatment groups and in both sexes.

In the ITT analysis, the skeletal muscle index (SMI) in the CON and REN groups was 7.8 ± 1.2 and 7.6 ± 1.3 kg/m2 at baseline, 8.0 ± 1.1 and 7.6 ± 1.3 kg/m2 at 24 weeks, and 7.9 ± 1.2 and 7.6 ± 1.3 kg/m2 at 48 weeks, respectively. The interaction of the treatment group with SMI was not significant (PtimeXtreat = 0.616). There was no significant change in SMI from baseline to 48 weeks in either group. PP analysis showed similar results of SMI.

Sarcopenic components and sarcopenia

The proportions of patients with low muscle mass (SMI), a slow gait speed (mean dynamic-start), weak standing HGS, and sarcopenia according to the 2019 Asian Working group for Sarcopenia (AWGS) in the CON and REN groups are presented in Figure 3A. The proportion of patients with low muscle mass or sarcopenia at baseline was larger in the REN group than in the CON group, but the difference between the two groups attenuated over the study period (at baseline, 4.0% vs. 18.9% (P = 0.004) and 2.7% vs. 13.5% (P = 0.017); at 24 weeks, 2.9% vs. 13.6% (P = 0.021) and 1.4% vs. 10.5% (P = 0.029); and at 48 weeks, 7.6% vs. 12.9% (P = 0.319) and 4.5% vs. 8.1% (P = 0.482), respectively). The PP and ITT analysis showed similar results (Figure 3B).

The proportions of low muscle mass (SMI), slow gait speed, weak handgrip strength, and sarcopenia according to 2019 AWGS in the ITT (a) and PP (B) groups. *P < 0.05. AWGS, Asian working group for sarcopenia; GS, gait speed; HGS, handgrip strength; ITT, intention-to-treat; PP, per-protocol; MM, muscle mass; SP, sarcopenia.

Quality of life scores

There were no significant differences in the QoL scores at baseline between the two groups in the ITT analysis (Table 3). The overall health ratings improved from baseline to 24 and 48 weeks in both groups. Bodily pain improved in the REN group from baseline to 48 weeks. According to the kidney disease-specific scores, the symptoms/problems and cognitive function in the REN group improved from baseline to 48 weeks. The quality of social interactions in the CON group was poorer at 48 weeks than at baseline. The trends for the overall health ratings, bodily pain, symptoms/problems, cognitive function, and quality of social interactions were similar between the ITT and PP analysis (Table S2). The vitality in the REN group was improved while the kidney disease effects in the CON group were poorer at 48 weeks than at baseline in the PP analysis. Interaction between time and group was evident only in symptoms/problems, cognitive function, and kidney disease effects, but not in overall health rating, bodily pain, vitality, and quality of social interaction in both the ITT and PP analyses (Tables 3 and S2).

Table 3. Comparison of quality of life scale scores between treatment groups in intention-to-treat population Short form-6 scale Baseline 6 M 12 M P for interaction CON REN P CON REN P CON REN P Time Group Time–group PF 80.4 ± 18.9 76.0 ± 23.7 0.250 78.8 ± 20.3 78.0 ± 20.4 0.805 78.5 ± 25.2 79.8 ± 18.7 0.746