Safety and Efficacy of Spironolactone in Dialysis-Dependent Patients: Meta-Analysis of Randomized Controlled Trials

Background Patients with end-stage renal disease (ESRD) are characterized with high risk of heart failure. Although mineralocorticoid receptor antagonists have beneficial effect on relieving cardiac fibrosis and, thus, reduce the incidence of cardiovascular disease and cardiac death, the therapeutic benefits and adverse effects are still controversial. We conducted a meta-analysis to measure the safety and efficacy of spironolactone in patients undergoing dialysis. Methods A systematic search for randomized controlled trials (RCTs) was performed in PubMed, Embase, and Cochrane databases. Primary outcomes included changes in all-cause mortality (ACM), serum potassium concentration, incidence of hyperkalemia and gynecomastia (GYN). Secondary outcomes included changes in blood pressure (BP), left ventricular mass index (LVMI) and left ventricular ejection fraction (LVEF). Subgroup analysis and sensitivity analysis were further conducted. This research was registered with PROSPERO (International Prospective Register of Systematic Reviews; No. CRD42021287493). Results Fifteen RCTs with 1,258 patients were enrolled in this pooled-analysis. Spironolactone treatment significantly decreased ACM (RR = 0.42, P < 0.0001), CCV (RR = 0.54, P = 0.008) and LVMI (MD = −6.28, P = 0.002), also increased occurrence of GYN (RR = 4.36, P = 0.0005). However, LVEF (MD = 2.63, P = 0.05), systolic BP (MD = −4.61, P = 0.14) and diastolic BP (MD = −0.12, P = 0.94) did not change between two groups after treatment. Although serum potassium concentration was increased (MD = 0.22, P < 0.0001) after spironolactone supplement, the risk of hyperkalemia remained unchanged (RR = 1.21, P = 0.31). Further subgroup analysis found more obvious advantageous as well as disadvantageous effects in Asian subjects than European or American ones. Also, with more than 9 months of treatment duration, patients achieved more favorable influence than shorter duration. Conclusions These results highlight the therapeutic effects of spironolactone on cardiovascular indexes, including ACM, CCV, and LVMI. However, the unignorable increase of GYN incidence and serum potassium level indicate that close monitor in dialysis-dependent patients, especially Asian patients, is essential.


INTRODUCTION
Heart failure, which often occurs in patients with chronic kidney disease (CKD), may contribute to high cardiovascular morbidity and mortality (1,2). Among all causes of death in patients undergoing dialysis, sudden cardiac death is the leading one, accounting for 25% of all-cause mortality (ACM) (3). Hypertension and left ventricular hypertrophy, which is directly associated with the risk for sudden cardiac death and ACM, occur in more than 70% patients with long-term dialysis (4,5).
Aldosterone has been implicated as an important factor to keep cardiovascular homeostasis. As a pleiotropic hormone, aldosterone also can regulate various tissues, such as heart, kidney and liver, through activating the mineralocorticoid receptors (MRs) (6). In the presence of impaired renal function, the renin-angiotensin-aldosterone system (RAAS) are always activated abnormally, mediating high blood pressure and cardiac fibrosis (7). Former researches proved that utilization of mineralocorticoid receptor antagonists (MRA) can mitigate the deleterious effects on cardiovascular system and thus, improving the prognosis of patients with end-stage renal disease (ESRD) (8).
The role of MRAs therapy as a neurohormonal antagonist has been studied by prior studies, however, had various outcomes. Flevari et al. (9) found significant increased sodium potassium level and decreased blood pressure after MRAs treatment, while Lin et al. (10) and Gross et al. (11) suggested unchanged serum potassium and blood pressure, respectively. Previous metaanalyses have been limited by small number of clinical trials (12) or results for single system (13). Thus, spironolactone is not widely understood in subjects undergoing dialysis. These differences in adverse effect and efficacy prompted us to conduct a meta-analysis to determine the changes in dialysis-dependent patients after spironolactone supplement. Also, we will further explore the effect of some factors (e.g., country, dosage, intervention duration) on the results through subgroup analysis.

METHODS
This meta-analysis was conducted in accordance with the preferred reporting items for systematic reviews and metaanalyses (PRISMA) guideline, and was registered with PROSPERO (International Prospective Register of Systematic Reviews; No. CRD42021287493).

Search Strategy and Data Sources
Literature published up to October 2021 in PubMed, Embase, and Cochrane databases, without time or language restriction, were searched. The search strategies are provided in Appendix A. Also, the reference lists of review articles and original studies were manually searched for additional eligible reports.

Selection Criteria
Studies were considered to be eligible if they met the following criteria: (1) randomized-control study on humans; (2) dialysis patients (3) patients in the intervention group, were treated with spironolactone, while the control subjects received placebo or standard treatment. Exclusion criteria were: (1) compared different dosages of spironolactone; (2)

Data Extraction and Quality Assessment
The data extraction from each studies includes study characteristics (year of publication, country, randomization method, type of study, sample size, duration and follow-up period) and patient characteristics (age, sex, and dialysis type).

Statistical Analyses
Review Manager (RevMan, version 5.4; the Nordic Cochrane Center, the Cochrane Collaboration, Copenhagen, Denmark) and Stata/SE (version 15.1; StataCorp LP, College Station, TX) were used for the analysis. Two authors extracted raw data from individual studies and then calculated pooled risk ratios (RRs) for dichotomous outcomes and mean differences (MDs) for continuous ones, and corresponding 95% confidence intervals (95% CIs) for each outcomes. The outcomes are presented as SMD if continuous indexes were measured in different methods. For research with more than one intervention groups, we split the shared control group into several groups with smaller sample size, and thus, included two or more comparisons (14). A fixedeffects model was used to perform meta-analysis, and a randomeffects model was applied when severe heterogeneity was present. Heterogeneity of included studies was quantified by Q test and I 2 statistic. High heterogeneity was defined as p < 0.1 for Q statistic or I 2 > 50%. To identify the source of heterogeneity, subgroup analysis was further conducted, according to country, dosage, and length of follow-up. Besides, two reviewers performed additional sensitivity analyses to explore the impact of a single article on the results. Cochrane Collaboration methodology (14) was used to assess included studies for bias.

Search Results
The process of study selection is schematically presented in the flowchart (Figure 1). Overall, the search terms identified 511 references. Of these, the reviewers excluded 259 articles from the initial screening. Subsequently, the majority of articles (217 articles) were excluded after reading the title and the abstract. After assessing the remaining 36 full text articles, we eliminated 23 additional articles since they failed to meet the inclusion criteria. Therefore, 15 randomized controlled trials (in 13 articles) were included in the meta-analysis.

Search Characteristics and Quality
The studies enrolled 1,258 patients, with the mean age spanned 52.92 ± 6.90 to 70.45 ± 9.70 years. Of these articles, three were conducted in China (10,15,16), two in the United States (11,17), two in Grace (9,18), two in Japan (19,20), two in Iran (21,22), one in Brazil (23) and one in Chile (24). In terms of ethnicity, seven RCTs were performed in Asians and 6 articles (8 RCTs) included non-Asians. Nine RCTs were performed in patients on maintenance HD, 3 articles involved PD patients, and other 3 researches reported data in both HD and PD population ( Table 1). Figure 2 summarizes the risk of bias of 13 RCTs. Eight studies and 3 studies provided information of random sequence or allocation concealment, respectively. Eleven studies were tripleblinded and 1 had a single-blind design.

Subgroup Analysis
Further subgroup analysis based on country was performed. In Asian subgroup, the pooled analysis found significant difference in serum potassium level, SBP, ACM, LVMI, LVEF, and incidence of CCV and GYN between experimental and control patients. Besides, spironolactone supplementation did not cause a consistent change in DBP and hyperkalemia occurrence. However, in European and American subjects, meta-analysis showed that above mentioned indexes did not confer difference after spironolactone application (Figure 8).  To eliminate the effect of duration, we also took subgroup analysis based on duration period. As shown in Figure 9, although longer intervention times (>9 months) had significant benefits on cardiac function (lower ACM and CCV occurrence, decreased LVMI, and increased LVEF), side effects (GYN, hyperkalemia, serum potassium) were significantly increased. However, in the subgroup with a treatment duration of less than or equal to 9 months, neither cardiac-related efficacy nor side effects were significant (Figure 9).

DISCUSSION
In this meta-analysis of 15 RCTs of 1,258 dialysis-dependent patients, we suggested that spironolactone treatment affected various parameters associated with cardiovascular system. Spironolactone decreased ACM, SBP, LVMI. Although the serum potassium level significantly increased, spironolactone did not elevate the occurrence of hyperkalemia. The further subgroup analysis implied the intergroup differences across countries. Spironolactone treatment had more obvious efficacy (decreased ACM, CCV, LVMI, and SBP, also increased LVEF) and more severe adverse effects (increased GYN occurrence and serum potassium) on Asian patients. However, spironolactone did not affect these markers in European and American population. Seven RCTs used spironolactone for more than 9 months, appearing more effective in reducing ACM and CCM as well as increasing incidence of GYN and hyperkalemia than those RCTs with shorter application duration. Hypertension often exists in patients with ESRD, accompanied by activated RASS and secondary hyperaldosteronemia (25). Aldosterone adversely affects blood pressure through both cellular and nervous mechanism. In blood vessels, mineralocorticoid not only decreases deformability of endothelial cells (26), but also suppresses bioactivity of nitric oxide in smooth muscle cells (27), therefore regulates blood pressure. Meanwhile, aldosterone receptors in the central nervous system may enhance sympathetic activity and thus contribute to hypertension (28). Although our meta-analysis didn't find antihypertensive effect of spironolactone, the high heterogeneity promoted us to perform sensitive analysis, which found data provided by Ni et al. (15) and Flevari et al. (9) were the source of heterogeneity. It may because all subjects included in these two studies were hypertensive, while other researches had a lower proportion of participants with hypertension (37.5 to 87.6%). Patients with hypertension, especially resistant hypertension, may be particularly prone to salt and water retention, therefore, more susceptible to spironolactone (29). Abnormal level of serum aldosterone promotes the production of profibrotic TGF-βsignaling and following cardiac fibrosis (30). Thus, using aldosterone antagonist can relief cardiac fibrosis in nephrectomized animals (31,32). Similarly, we also observed rate of LVEF and LVMI significantly changed after MRAs treatment. Spironolactone may also benefit CCV system through following pathways: (1) blocking aldosterone effect on collagen formation, therefore, inhibiting left ventricle remodeling (33); (2) the antihypertensive effect of spironolactone improves vascular endothelial function (34); (3) preventing peritoneal inflammation and fibrosis thereby maintaining peritoneal function (35).
Our research also noted the safety of spironolactone in dialysis-dependent patients, with spironolactone treatment showing a tendency to increase serum potassium concentration, but an unchanged incidence of hyperkalemia. Hyperkalemia, which is well-recognized as adverse effect of MRAs, always prevent the physician from applying spironolactone. Notably, in our pooled-analysis, most of patients who dropped out of the study because of hyperkalemia were hemodialysis (HD) patients, not PD patients. These results are similar to previous findings showing the greater removal of potassium in PD than HD subjects (36,37). Since dialysis patients are more dependent on dialysis rather than kidney to excrete potassium, using spironolactone maybe safer in dialysis patients than in nondialysis ones. GYN also appeared with spironolactone application (38), and a significant occurrence of GYN was noted in treated group. The risk of GYN may be minimized if patients used low dose of spironolactone or switched to selective MRAs, such as eplerenone (19).
In subgroup analysis, the effects, including efficacy and side effects, of spironolactone were more significant in Asian patients. Actually, this kind of racial difference existed in several clinical studies on MRAs. Vardeny et al. reported that non-African Americans might have greater beneficial effect from spironolactone supplement than African Americans (39). Besides the unclear genetic mechanism of racial disparity, different selection of ACEI or ARB by ethnic groups may also cause intersubgroup heterogeneity. The high incidence of cough among Asian patients makes them prone to choose ARB rather than ACEI (40,41), which may contribute to different results on efficacy and side effects between Asian and non-Asian subgroups (42,43). Besides, relatively longer intervention duration in Asian populations (3 ∼ 36 months) than in European and American ones (0.5 ∼ 24 months) should also be taken into account. Till now, the safe duration of spironolactone treatment is still controversial. Spironolactone exerts cardio-protective effects by inhibiting MRs on the one hand and reduces potassium excretion owing to Na + /K + pump inhibition on the other hand. These double-edged sword effects do not occur simultaneously. Spironolactone elevates serum potassium at an early stage, while cardioprotective effects appear later (44), suggesting close laboratory surveillance for patients newly initiating therapy with MRAs. From this pooled-analysis, longer duration (more than 9 months) was related to an increased LVEF, decreased LVMI, ACM, and CCV, however, the incidence of GYN and ACM also raised. We also did further subgroup-analysis based on dosage, and surprisedly, the results indicated that the lower dosage (≤25 mg) had more obvious side effect and efficacy than higher dosage (50 mg) (Supplementary Figure 2). This may due to the possibility that Asians are more sensitive to drugs and therefore, based on former experience, all Asian groups have chosen smaller doses. Thus, optimal dosage in terms of safety and efficacy remains a critical question that needs to be addressed. Epidemiological evidence suggests that diabetes mellitus is one of the most common modifiable risk factors for CCV and ACM (45). However, all but one article [the study by Vukusich et al. (24)] included only non-diabetic patients, other 12 articles included diabetic and non-diabetic patients. Thus, we failed to perform a diabetes-based subgroup analysis.
Besides, as an early sensitive indicator, carotid intima-media thickness has been used as a surrogate endpoint for CCV and ACM to access the efficacy of certain interventions in the past few years (46). Therefore, further studies limited to non-diabetic patients with reported data on carotid intima-media thickness are recommended.
The strength of this study is the strict selection range that only includes RCTs. Meanwhile, without the restrictions of followup duration and renal replacement type, we maximized the collected information while selection bias or other potential bias were minimized. Moreover, to the best of our knowledge, this is the first meta-analysis to provide evidence of racial differences in spironolactone use in dialysis-dependent patients. Our results should also be interpreted within the context of several limitations. These include the relatively small sample size and various duration of treatment period. Since spironolactone is an MRA, future studies should report serum as well as urine aldosterone level. Most RCTs included are single-center trials. However, the racial factor is one of the sources of intergroup heterogeneity, thus, multi-central research with patients from various continents is needed in following studies. Moreover, Ito et al. (19) proposed inconsistent changes in kidneyrelated indicators after spironolactone use in men and women,  suggesting further researches to report outcomes according to gender.

CONCLUSIONS
Patients undergoing dialysis can achieve cardiac benefit (LVEF, LVMI, CCV, and ACM) after spironolactone treatment, while the risk of GYN and serum potassium concentration also increased. However, spironolactone does not affect the incidence rate of hyperkalemia. Asian patients can achieve more obvious benefit, although more side effects, from spironolactone than European and American subjects. The use of spironolactone for more than 9 months can have a pharmacological effect compared to a shorter course of treatment. Studies with   Frontiers in Medicine | www.frontiersin.org larger scales and multi-countries are advocated to further evaluate the balance of efficacy and adverse effects in spironolactone use.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.