The Effect of Non-Pharmacological and Pharmacological Interventions on Measures Associated with Sarcopenia in End-Stage Kidney Disease: A Systematic Review and Meta-Analysis

This systematic review and meta-analysis provides a synthesis of the available evidence for the effects of interventions on outcome measures associated with sarcopenia in end-stage kidney disease (ESKD). Thirteen databases were searched, supplemented with internet and hand searching. Randomised controlled trials of non-pharmacological or pharmacological interventions in adults with ESKD were eligible. Trials were restricted to those which had reported measures of sarcopenia. Primary outcome measures were hand grip strength and sit-to-stand tests. Sixty-four trials were eligible (with nineteen being included in meta-analyses). Synthesised data indicated that intradialytic exercise increased hand grip strength (standardised mean difference, 0.58; 0.24 to 0.91; p = 0.0007; I2 = 40%), and sit-to-stand (STS) 60 score (mean difference, 3.74 repetitions; 2.35 to 5.14; p < 0.001; I2 = 0%). Intradialytic exercise alone, and protein supplementation alone, resulted in no statistically significant change in STS5 (−0.78 s; −1.86 to 0.30; p = 0.16; I2 = 0%), and STS30 (MD, 0.97 repetitions; −0.16 to 2.10; p = 0.09; I2 = 0%) performance, respectively. For secondary outcomes, L-carnitine and nandrolone-decanoate resulted in significant increases in muscle quantity in the dialysis population. Intradialytic exercise modifies measures of sarcopenia in the haemodialysis population; however, the majority of trials were low in quality. There is limited evidence for efficacious interventions in the peritoneal dialysis and transplant recipient populations.


Introduction
Sarcopenia, originally believed to be a condition related to age, is the term used to indicate a progressive reduction in muscle strength, quantity or quality, and function, (hand grip strength (HGS) and the following sit-to-stand tests (STS), 5, 30, and 60). The secondary outcomes were muscle quality and quantity (assessed by magnetic resonance imaging (MRI), dual-energy X-ray absorptiometry (DEXA), bioelectrical impedance analysis (BIA), and computed tomography (CT) imaging), physical performance (assessed by the short physical performance battery (SPPB), the timed-up-and-go test (TUG), 400 m walk test, and gait speed), and sarcopenia health-related quality of life as assessed by the SARQoL questionnaire.

Trial Design
Trials included in this review had to have adhered to the following trial designs: parallel-group randomised controlled trials (allocation at individual or cluster levels) or crossover randomised trials.

Search Strategy
Searches were conducted to identify any relevant completed or ongoing systematic reviews using the following resources: Cochrane, PROSPERO, and the National Health Service Centre for Reviews and Dissemination (Health Technology Assessment (HTA) and Database of Abstracts of Reviews of Effects (DARE)). The following bibliographical databases and trial registers were searched for completed and ongoing trials: MED-LINE, EMBASE, CINAHL, Cochrane Central Register of Controlled Trials (CENTRAL), ClinicalTrials.gov, and the ISCRTN Registry. British Library (ETHOS), OpenGrey, and Conference Proceedings Citation Index (Web of Science™ Core Collection) were searched for unpublished data. All databases were searched from inception to 19 July 2021, and no limits on language were set. Database searches were supplemented with internet searches (e.g., Google Scholar), and contact with the Physical Activity and Wellbeing Kidney Research Study Group (in the United Kingdom). An example of a full search strategy for MEDLINE, EMBASE, and CINAHL databases is presented in Tables S1 and S2. Other databases were searched by using different combinations Wof these search terms. Search results were compiled using the web-based screening and data extraction tool Covidence (Veritas Health Innovation Ltd., Melbourne, Australia) as recommended by the Cochrane Collaboration. Duplicate citations were removed, and title and abstracts were screened independently by two reviewers against the inclusion criteria (if there was disagreement, Wthen this was settled through the use of a third reviewer). Full-text articles of trials not excluded based on title or abstracts were retrieved and assessed by two reviewers. Conference abstracts and trials included on registries only (e.g., ClinicalTrials.gov) were excluded.

Selection Criteria, Data Extraction, and Quality Appraisal
We developed, tested, and refined a structured data collection form based on the Cochrane Data Extraction Template for interventions. For each included trial, information on trial methods, participants, interventions/comparator, and outcomes was extracted and cross-checked by one reviewer (DSM). Risk of bias for each trial was assessed using the Cochrane Risk of Bias Tool across five domains. Each domain was classified as adequate, unclear, or inadequate, with risk of bias for each trial to be classified using the following criteria: (1) low risk of bias (all criteria are deemed adequate), (2) moderate risk of bias (one criterion graded as inadequate or two graded as unclear), and (3) high risk of bias (more than one criterion is deemed inadequate, or more than two are graded unclear). Funnel plots were used to visually assess publication bias in the meta-analyses performed for the primary outcome only. Formal testing for plot asymmetry would only be performed where the meta-analysis contains more than ten trials [13].

Data Synthesis
Where means and standard deviation of outcome measures were not available, they were estimated from medians and interquartile ranges [14]. Gait speed data were converted from cm/s to m/s for one trial [15], and were provided by the authors for another [16]. HGS was converted from lbs to kg for one trial [17]. Data for mid-arm muscle area (MAMA) were subtracted for one trial [18] using Web-Plot Digitizer version 4.5 [19] and 95% confidence intervals were converted to standard deviations [13]. A meta-analysis was performed for trials that reported the same outcome measures using a generic inverse variance random effects method via Review Manager (RevMan) version 5.3.26 (The Cochrane Collaboration, 2020). Primary and secondary measures of efficacy were treated as continuous data and interpreted as either difference in means or standardised mean difference dependent on the methods of measurement. Analysis was based on the final (post-intervention) values only (at last follow-up) with the exception of mean change data from two trials [15,20]. Statistical heterogeneity was interpreted using the I 2 value. Data were not pooled (or subgroup analysis was considered) if I 2 > 40% (this is the threshold to which heterogeneity is considered important). Separate analysis was performed for each type of population (dialysis and transplant) and each non-pharmacological and pharmacological intervention. We had prospectively planned a network meta-analysis (NMA); however, this was not possible as a result of a limited number of trials for each population reporting the same sarcopenia-associated outcome. In addition, variances between the delivered interventions within the included trials suggested that the transitivity assumption (needed for NMA) was unlikely to be met.  (Tables 1-3), with 19 trials being included in meta-analyses. Eleven conference abstracts were excluded at the full-text screening stage (due to insufficient information). There were 54 trials in the dialysis population (43 in the haemodialysis, 7 in the peritoneal dialysis, and 4 trials containing both dialysis populations) (Tables 1 and 2). In total, 23, 20, and 8 trials tested exercise, nutritional supplement, and pharmacological interventions, respectively. Two trials tested both exercise and pharmacological interventions [15,21], and one trial tested an exercise and a nutritional intervention [22]. There were ten eligible trials in the transplant recipient population ( Table 3). The most prevalent measurements of muscle strength, muscle quality/quantity, and physical performance in the ESKD population were HGS (n = 26), lean whole body mass (LBM) (n = 29), and gait speed (n = 15), respectively. There were no trials identified that included conservative management participants, and no trial reported the SARQoL questionnaire as an outcome (Tables 1-3). Twenty-nine trials (45%) reported an a priori power calculation.

Risk of Bias
Risk of bias summaries are provided in Figures A1-A3. Only 10 (16%) of the included trials were rated as having an overall low risk of bias. Funnel plots are provided in Figure A4 (for the analyses presented in Figures 2-5). There was no observation of publication bias.

Nutritional Interventions
Data from two trials (110 participants) [53,64] investigating the effect of Vitamin D (cholecalciferol) on HGS were available, but there was considerable heterogeneity between the trials (I 2 = 60%). Neither trial [53,64] reported any significant change with Vitamin D. Other interventions including L-carnitine [18,67] and keto acid supplementation [58] appeared to have no effect in the dialysis population.

Pharmacological Interventions
Three trials reported measuring HGS following the administration of growth hormone [48,55,56] in the haemodialysis population, but the data were not suitable for metaanalysis. Individual data from two of these trials showed no statistically significant increase [48,55]. Two trials investigated the effect of anabolic steroid supplementation on HGS, one reported a significant increase [63], whilst there was no change reported in the other [20].

Pharmacological Intervention
One trial reported a lack of effect of anabolic steroids on STS5 [15].

Nutritional Interventions
Synthesised data from two trials [18,43] including 108 haemodialysis participants indicated that L-carnitine supplementation significantly increased MAMA (MD, 3.10 cm 2 ; 0.92 to 5.28; p = 0.005; I 2 = 0%) (Figure 7). One of these trials [18] also reported data for LBM, skeletal muscle mass, and appendicular lean mass (ALM) with no statistically significant change in these outcomes following L-carnitine supplementation. Synthesised data from two trials [22,66] in the haemodialysis population (98 participants) indicated that oral whey protein supplementation resulted in no statistically significant effect on LBM (MD, −1.55 kg; −4.25 to 1.14; p = 0.26; I 2 = 0%) ( Figure S1). Data were reported on LBM from trials investigating a number of heterogeneous nutritional interventions (see Table 2). Individual results from these trials reported statistically significant increases in LBM following water-soluble vitamin supplementation [44], amino acid supplementation [54], and creatine supplementation [60]. Other trials reported data for ALM [49], MAMA [44], and FFM [45,51,62] and individually reported no significant changes (see Table 2 for interventions). For the peritoneal dialysis population, data from three trials were available reporting on the effect of protein supplementation on mid-arm muscle circumference (MAMC) [50,61,65]; there was heterogeneity between trials (p = 45%). One trial reported a statistically significant increase in MAMC (along with LBM) [65], whilst there was no change for this variable in the other two trials [50,61].

Pharmacological Interventions
Synthesised mean change data from two trials [15,20] investigating the effect of nandrolone decanoate (an anabolic steroid) on LBM showed a statistically significant increase (MD, 3.10 kg; 2.12 to 4.08; p < 0.044; I 2 = 0%) (Figure 8). One of these trials [15] also reported a significant increase in MT-CSA, and another has shown an increase in FFM following oxymetholone [63]. Mean change data were available for LBM from three trials [48,52,57] investigating the effect of growth hormone. There was considerable heterogeneity between trials (I 2 = 75%). Two trials reported significant increases in LBM following growth hormone injections compared to placebo [48,52]. In two trials investigating the effect of early steroid withdrawal in transplant recipients there was no effect of this on LBM [75,78].

Pharmacological Interventions
Data (which were unsuitable for meta-analysis) were reported for two trials investigating the effect of human growth hormone [48,55]. Only one trial reported a significant increase in gait speed following the administration of growth hormone [55]. Another trial found a lack of effect following anabolic steroid supplementation [15].

Exercise Intervention
Synthesised data from two trials (69 haemodialysis participants) [29,33] for TUG reported no significant effect (MD, −1.05 s; −2.12 to 0.02; p = 0.06; I 2 = 0%) ( Figure 11) following intradialytic exercise. Moreover, a supervised programme of exercise performed on non-dialysis days significantly improved TUG [26]. Programmes of home-based walking [29,36] and intradialytic exercise [16] did not significantly improve SPPB [16,36] or TUG [29]. In contrast, one trial [23] in the peritoneal dialysis population and another in transplant recipients [71] demonstrated significant increases in TUG following programmes of exercise. Figure 11. Effect of intradialytic exercise on timed-up-and-go score (s) in individuals receiving haemodialysis. Data are expressed as mean difference and 95% CI. * Data for exercise and control groups only [29,33].

Discussion
This is the first review that has aimed to synthesise the effect of non-pharmacological and pharmacological interventions for sarcopenia outcomes (using the most up-to-date and widely accepted definition [1]) in the ESKD population. The main findings of this review were that intradialytic exercise significantly improved measures of muscle strength (HGS and STS60) and physical performance as measured by gait speed. However, the majority of trials included in the review were considered to be at high risk of bias. There was some evidence that programmes of exercise in transplant recipients may improve STS scores. The evidence for nutritional and pharmacological interventions was less clear, with some tentative evidence that L-carnitine and nandrolone decanoate may have favourable effects on muscle quantity (MAMA and LBM, respectively) in individuals receiving haemodialysis. There was a lack of evidence for efficacious interventions to treat sarcopenia in the transplant and peritoneal dialysis population, and there were no included trials in those individuals with ESKD receiving conservative management.
A recent systematic review exploring the effect of exercise interventions on objective physical function in the ESKD population [79] reported that the majority of included trials reported a significant improvement in STS and HGS, although unlike the present review they were not able to perform a meta-analysis for these outcomes. This is in agreement with another review [80] that demonstrated that exercise training in the haemodialysis population was able to increase muscle strength. Our review confirms that exercise is efficacious at modifying outcomes associated with sarcopenia; however, the evidence for pharmacological and nutritional interventions is less clear. This review included trials with a number of heterogeneous nutritional and pharmacological interventions with a lack of evidence for their efficacy on measures of sarcopenia. However, this is with the exception of synthesised data for L-carnitine and nandrolone-decanoate showing modifications to MAMA and LBM. However, it is unclear whether changes to these outcomes would translate to improvement in muscle strength and function.
Sarcopenia is highly prevalent in CKD [3], particularly for those with the advanced stages of the disease (ESKD) [6]. It is associated with hard endpoints including cardiovascular events and mortality [2,5]. With prevalence of ESKD projected to increase [81], identifying effective interventions for the treatment of sarcopenia is particularly relevant. Therefore, the finding of this review, that intradialytic exercise improves HGS and gait speed, has clinical significance. A low walk (gait) speed has been shown to be associated with mortality in 752 individuals receiving dialysis [82], with a walk speed of >0.6 m/s associated with greater survival [82]. Another study [83] has also reported that both low gait speed and HGS are predictors of cardiovascular events and all-cause mortality in individuals receiving haemodialysis [83]. This supports the recent shift from low muscle mass to low muscle strength as a key characteristic for the diagnosis of sarcopenia [1], as low muscle strength appears to be better at predicting outcomes [3,84]. Furthermore, muscle strength (STS and HGS) can be easily evaluated in the clinical setting (outpatient clinics and dialysis units, etc.). The evidence from this review that intradialytic exercise increases muscle strength, coupled with recent RCT data [16] (that this mode of exercise improves cardiovascular health and is safe), suggests that the methods of implementation should be considered as outlined in the recent Clinical Practice Guideline for Exercise and Lifestyle in CKD [85].
It is believed that increasing protein intake may be an effective countermeasure to sarcopenia for individuals with CKD. This is highlighted by the recommendation of increased intake (compared to the general population) for individuals with ESKD in the updated KDOQI Clinical Practice Guideline for Nutrition in CKD [86]. However, the present review found limited current RCT evidence for the efficacy of protein supplementation for sarcopenia in CKD, a point that has recently been highlighted by others [6]. Protein without an adequate exercise stimulus often provides little benefit, although notably the largest RCT to date in the ESKD population investigating the combined effect of exercise and protein supplementation found no effect on muscle strength or function [22]. This review identified a limited number of trials in the peritoneal dialysis and transplant recipient population. Given the positive effects that we have seen for exercise interventions (particularly for muscle strength in the haemodialysis population), it would be prudent to test these in future RCTs involving other ESKD populations. A recent review article [6] has highlighted a number of pharmacological interventions as having the potential to mitigate sarcopenia in the CKD population. However, this review found no evidence for the benefit of pharmacological interventions on muscle strength. There was some indication from synthesised data that nandrolone-decanoate increases LBM and individual data from two trials show that growth hormone may improve LBM. Whether these changes may improve outcome is unlikely. A previous trial of nandrolone decanoate in individuals with rheumatoid arthritis found an increase in LBM but no accompanying change in muscle strength [87]. Properly powered (<50% of the included trials reported an a priori sample size calculation) trials are required to test both the efficacy and safety of pharmacological and nutritional interventions in the ESKD population. This should enable a wide range of evidence-based therapeutics to be available in line with a personalised medicine approach to tackling sarcopenia. Lastly, although we have shown that exercise programmes may be an effective countermeasure to sarcopenia in the ESKD population, there remains a lack of evidence for these interventions on associated hard endpoints such as cardiovascular events and mortality. Despite the inclusion of 64 trials in the review, only a small number of these were able to be included in meta-analyses (with only fifteen trials being included in analyses for the primary outcome (muscle strength)) and the majority were assessed as having a high risk of bias.

Conclusions
Currently, exercise appears to be the strongest therapeutic intervention for sarcopenia in the end-stage kidney disease population. There is a lack of proven efficacy for nutritional and pharmacological interventions.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/nu14091817/s1, Table S1: Full search strategy for MEDLINE and CINAHL databases; Table S2: Full search strategy for EMBASE databases; Figure S1: Effect of whey protein supplementation on lean body mass (kg); Figure S2: Effect of whey protein supplementation on timed-up-and-go in individuals receiving haemodialysis. Data Availability Statement: Data will be made available upon a request to the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
Risk of bias for included exercise, nutrition, and pharmacological intervention trials in the dialysis population, and included transplant trials (assessed using the Cochrane Risk of Bias tool). Unclear risk of bias is indicated by "?", low risk of bias "+", high risk of bias "-".