Abelmoschus manihot for Diabetic Nephropathy: A Systematic Review and Meta-Analysis

Diabetic nephropathy (DN) is the leading cause of end-stage renal disease (ESRD). Many trials have shown that Abelmoschus manihot could further improve proteinuria and protect kidney function in patients with DN when added to a renin-angiotensin system (RAS) blocker. A systematic assessment of the efficacy and safety of A. manihot in DN is essential. Eight electronic databases were searched to identify eligible trials published from inception to December 2017. The Cochrane Risk of Bias Tool was used to evaluate the methodological quality of eligible studies. Seventy-two studies with 5,895 participants were identified. The methodological quality of included studies was generally low. The results indicated that, compared to a RAS blocker, combined treatment of A. manihot with a RAS blocker was more effective for 24h urinary protein (24h UP) (mean difference [MD], -0.39 [95% confidence interval [CI], -0.46 to -0.33] g/d; P<0.00001), urinary albumin excretion rate (UAER)(MD, -19.90 [95% CI, -22.62 to -17.18] μg/min; P<0.00001), 24h UP reduction rate (risk ratio [RR], 1.43; 95% CI, 1.26-1.63; P<0.00001), normalization of UAER (RR, 1.48; 95% CI, 1.29-1.70; P<0.00001), and serum creatinine (SCr) (MD, -7.35 [95% CI, -9.95 to -4.76] umol/L; P<0.00001). None of these trials reported the ESRD rate. No statistically significant difference occurred between A. manihot combined with a RAS blocker and a RAS blocker alone in estimated glomerular filtration rate (eGFR) (MD, 4.43 [95% CI, -1.68 to 10.54] mL/min; P=0.16). A. manihot did not increase the rates of adverse drug events. A. manihot in addition to a RAS blocker was effective and safe to further improve proteinuria and protect kidney function in patients with DN. However, due to the generally low methodological quality, significant heterogeneity, and publication bias, high-quality randomized controlled trials are required to confirm these findings before the routine use of A. manihot can be recommended.


Introduction
Approximately 20% to 40% of patients with diabetes mellitus (DM) will develop diabetic nephropathy (DN) [1]. Type 2 diabetes mellitus (T2DM) is the most common cause of chronic kidney disease (CKD) and end-stage renal disease (ESRD) in the developed world and is the second leading cause of ESRD after primary glomerular disease in China [2][3][4]. DM and CKD are independent risk factors of allcause mortality as well as cardiovascular death [5,6]. Diabetic kidney disease (DKD) poses the highest risk for death compared to DM or CKD alone [5,6].
Management of DN requires a multifaceted approach, including a combination of lifestyle modifications and pharmacologic intervention. The effectiveness of current interventions remains limited given the number of patients who continue to have progression of their renal dysfunction, despite blood pressure and glycemic control, and the use of existing renin-angiotensin system (RAS) blockers. Retrospective analyses of clinical studies concerning DN demonstrate a strong relationship between the magnitude of albuminuria reduction and slowing of CKD progression as well as reduced cardiovascular event rates [7][8][9][10][11][12]. An angiotensin-converting enzyme inhibitor (ACEI) combined with an angiotensin receptor blocker (ARB) is not recommended due to the high risk of hyperkalemia and/ or acute kidney injury as well as no benefit in altering the natural 2 Evidence-Based Complementary and Alternative Medicine history of DN [1,10]. Recently, mineralocorticoid receptor antagonist (MRA) in addition to an ACEI/ARB treatment has been studied as a novel approach to further prevent the progression of DN. A meta-analysis by Mavrakanas et al. [13] reported that combined treatment with an ACEI/ARB and an MRA was effective in decreasing albuminuria compared to standard treatment with an ACEI/ARB in DN but increased the risk of hyperkalemia. Therefore, there is an urgent need for a new pharmacologic agent that could be effective and safe to further improve proteinuria and prevent the progression of DN.
Traditional Chinese medicine (TCM) has shown promising effects on the control of proteinuria, protection of renal function, and improvements in patients' clinical symptoms [14]. Abelmoschus manihot has been in use for CKD in China for hundreds of years. Huangkui capsule, a single medicament of TCM extracted from the dry corolla of A. manihot, has been approved by China's State Food and Drug Administration (SFDA) for the treatment of chronic nephritis since 1999. A. manihot can ameliorate proteinuria and protect kidney function in patients with CKD, such as DN, immunoglobulin A nephropathy (IgAN), and membranous nephropathy, and is currently considered an important adjuvant therapy for CKD [15][16][17][18][19][20]. The major biologically active constituents are total flavones of A. manihot (TFA) [21]. Mechanistic studies applying A. manihot to the treatment of CKD suggest that the major effects are associated with improved immunological reaction, inflammation, renal fibrosis, and renal tubular epithelial injury [14,22]. The results of previous meta-analyses preliminarily suggest that A. manihot could improve proteinuria and protect kidney function in patients with DN [16][17][18][19]. However, the evidence was very limited on the effect of A. manihot for DN due to a limited number of trials included, with poor methodological quality. A lot of novel data evaluating A. manihot in DN have been recently published. Therefore, we systematically analyzed the evidence on A. manihot in addition to a RAS blocker therapy in DN, focusing on its effect in albuminuria.

Data Sources and Searches.
This systematic review was reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [23]. See File S1 in the Supplementary Material for the PRISMA 2009 checklist for this article. The review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO registration no. CRD42018087182, available at http://www.crd.york.ac.uk/PROSPERO/display record.php?ID=CRD42018087182). The Cochrane Central Register of Controlled Trials (CENTRAL) on the Cochrane Library, PubMed, EMBASE, Chinese National Knowledge Infrastructure database (CNKI), Chinese Biomedical Literature database (CBM), Chinese Scientific Journal database (VIP), and Wan Fang database were searched to identify eligible trials published from inception to December 15, 2017. Ongoing registered clinical trials were searched at Clinical-Trials.gov (https://www.clinicaltrials.gov). The articles were not restricted based on language. All included studies were subjected to the same quality assessment.
The search terms were as follows: Flos Abelmoschus manihot, Abelmoschus manihot, Abelmoschus moschatus Medicus, Abelmoschus, okra, Huangkui, Huangkui capsule, huangshukui, diabetic nephropathy, diabetes mellitus, diabetic, kidney disease, renal disease, diabetic kidney disease, diabetic renal disease, albuminuria, randomized controlled trial, controlled clinical trial, randomized, randomly, and trial. See File S2 in the Supplementary Material for an example of the full electronic search strategy. Two authors (L. W. Shi and M. Z. Zhang) performed independently the literature search. Disagreements were resolved by discussing with a third party (Q. Ni and L. Feng).

Study Selection.
Eligible trials were listed and assessed independently by two reviewers (L. W. Shi and M. Z. Zhang) using predefined inclusion criteria. Studies were included if they met the following criteria: (1) it was randomized controlled design; (2) patients were with type 1 or type 2 DM and DN (defined as at least 30 mg of albuminuria in a 24h urine collection or urinary albumin excretion rate (UAER) of at least 20 g/min); (3) participants should have received an ACEI or an ARB throughout the study as standard treatment.
To evaluate the effect of concomitant A. manihot, a subset of patients in each study should also receive A. manihot in addition to standard RAS blockade; (4) the primary outcome measures included 24-h urinary protein (24h UP), ESRD rate and estimated glomerular filtration rate (eGFR). The secondary outcome measures were UAER, improvements in 24h UP reduction rate (defined as the proportion of 24h UP decrease in protein excretion ≥50% of the baseline at the end of the study), normalization of UAER (defined as the proportion of UAER <20 g/min upon study completion), serum creatinine (SCr) and adverse drug events (ADEs); (5) the studies included available and relevant data; and (6) the studies were not restricted based on publication language.
Excluded from the meta-analysis were duplicated publications, studies with unavailable or incorrect data, and articles not reporting outcomes of interest. Also excluded were studies enrolling fewer than 10 participants, quasirandomized controlled trials (e.g., allocation using alternation, the sequence of admission, case record numbers), and nonrandomized controlled clinical trials. Studies using combination RAS blockers as background therapy or A. manihot coupled with any other TCM drugs were excluded to avoid confounding information.

Data Extraction and Quality Assessment.
Two authors (X. W. Li and Y. N. Yang) independently extracted information on the patients as well as on the methods, interventions, outcomes, and results using a predesigned data extraction form. The data extraction form included the following items: name of first author, year of publication, total number and number in both groups, gender and mean age, baseline characteristics, method of randomization, allocation concealment, incomplete outcome data, selective reporting, blinding, interventions, and outcomes. The methodological quality of randomized controlled trials (RCTs) was independently assessed by two authors (M. Z. Zhang and Y. Y. Zhang) via the Cochrane Risk of Bias Tool [24]. Each study was respectively categorized as "low risk of bias", "high risk of bias", or "unclear risk of bias". Authors were contacted by e-mail to obtain further data and verify the methodological quality when necessary. The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) methodology was used to assess the quality of the evidence of each outcome. Any disagreement was settled by mutual discussion with a third author (Q. Ni and L. Feng).

Data Synthesis and Analysis.
Dichotomous outcomes were pooled using risk ratio (RR) and 95% confidence intervals (CIs) and continuous outcomes were pooled using mean difference (MD, defined as the difference between study groups at the end of study) and 95% CIs. A random-effects model was used to pool the data. Statistical heterogeneity was assessed with the I-square (I 2 ) statistic [25]. The I 2 statistic of ≤50% referred to low statistical heterogeneity, while >50% was considered as substantial statistical heterogeneity. Publication bias was performed and evaluated using funnel plots, if the group included >10 trials [26]. Sensitivity analysis was assessed by excluding lower quality trials and repeating the meta-analyses to examine the effects of these study subgroups. We had no prespecified plan of subgroup analysis. Meta-analysis was performed by using Review Manager Version 5.3. All tests were 2-tailed, and P<0.05 was considered statistically significant.

Included Studies and Trial Characteristics.
A flow diagram of study selection is shown in Figure 1. During the initial electronic search, 1,114 articles were identified, of which 962 were excluded including duplicates and irrelevant studies. The full texts of the selected 152 trials were retrieved, and after detailed evaluation, 72 RCTs  were finally selected for meta-analysis; of these, 28 met the inclusion criteria from 4 previous meta-analyses [16][17][18][19]. Authors were contacted by e-mail for additional outcome data; however, no reply was received.
The baseline characteristics of DN patients are presented in Table 1. The 72 studies included a total of 5,895 patients followed-up from 4 to 24 weeks. The treatment and control groups consisted of 3,000 and 2,895 patients, respectively. Sample size of the included trials ranged from 40 to 200. The mean age reported for participants in these studies ranged from 36 to 69 years, and the proportion of males ranged from 33.3% to 69.2%. The average baseline protein level in urine was 1.94 g/d (0.14-6.2 g/d). The median follow-up for Evidence-Based Complementary and Alternative Medicine  24h UP was 12 weeks. A. manihot in the form of a Huangkui capsule (Jiangsu SuZhong Pharmaceutical Group Co., Ltd.) was given orally at 3.0 g 3 times daily in one trial [27], orally at 2.5 g 3 times daily in 67 trials [28-41, 43-47, 49-64, 66-85, 87-98], and orally at 2.0 g 3 times daily in 3 trials [42,48,65]. In one study [86], Abelmoschus alcohol extract was given orally at 0.4 g, 3

Risk of Bias Assessment.
A summary of study quality is presented in Figure 2. The methodological quality was generally poor. All trials were reported to be randomized, but only 14 (19.44%) trials [28,30,35,45,47,48,52,61,65,68,75,82,88,94] described adequate sequence generation. None of the included trials mentioned the methods for allocation concealment, the blinding of participants and personnel, and blinding of outcome assessment. Risk of attrition bias (incomplete outcome data) was detected in one [87] of all included trials, with a high risk status. Selective reporting and other potential sources of bias were unclear. Sensitivity analysis was not performed since all included trials were generally of low methodological quality. The funnel plots based on 24h UP, UAER, and SCr were asymmetrical, showing that publication bias might affect the results of this meta-analysis.
The funnel plots constructed for improvements in 24h UP reduction rate and normalization of UAER were both nearly symmetrical, showing that publication bias might not affect the results of this meta-analysis. Funnel plots based on the primary and secondary outcomes are respectively elaborated in Figures 3(a), 3(b), 3(c), 3(d), and 3(e).  Figure 4). There was evidence of significant heterogeneity across these trials (I 2 =98%; P for heterogeneity <0.00001; Figure 4).  Figure 5). There was evidence of significant heterogeneity across these trials (I 2 =89%; P for heterogeneity <0.00001; Figure 5).

Urinary Albumin Excretion Rate (UAER
Again, there was evidence of significant heterogeneity across these trials (I 2 =99%; P for heterogeneity <0.00001; Figure 6). In addition, two [45,91] of 42 trials reported that trial duration per patient was 20 weeks, with 8 weeks of treatment and 12 weeks of follow-up without treatment. The MD of UAER between study groups at the end of follow-up was assessed again and still less in the treatment versus control groups (one trial [45] [32,35,40,43,69,74,75,80,84,86,89] of the included studies reported changes in 24h UP reduction rate. The pooled results showed that A. manihot combined with a RAS blocker therapy was associated with significant improvements in 24h UP reduction rate compared with a RAS blocker alone (RR, 1.43; 95% CI, 1.26-1.63; P<0.00001; Figure 7). The normalization of UAER was reported in 11 trials [29,38,47,50,62,74,77,85,89,94,98] of 72 RCTs. The results showed that combined treatment of A. manihot and a RAS blocker was more effective in normalization of UAER (RR, 1.48; 95% CI, 1.29-1.70; P<0.00001; Figure 8) than a RAS blocker alone. Statistical heterogeneity was low for these outcomes, suggesting a consistent effect size across studies (I 2 =0%; Figures 7 and  8). mol/L; P<0.00001, Figure 9), indicating that A. manihot was associated with improved kidney function. The I 2 statistic based on the data for SCr exhibited significant heterogeneity (I 2 =89%, P<0.00001, Figure 9).
Effects of A. manihot on the likelihood of ADEs are shown in Table 2.

Strength of Evidence.
The GRADE approach was used to assess the quality of the evidence and risk of bias. The results are shown in Table 3. The quality of evidence was generally low.

Summary of Evidence.
This is the first comprehensive systematic review and meta-analysis to assess the effects of A. manihot for DN patients with a diverse range of baseline protein level in urine and kidney function. None of the included trials reported the ESRD rate, and the pooled analysis of 7 trials indicated that there were no statistically significant Heterogeneity: ４；Ｏ 2 = 78.55; ＃ＢＣ 2 = 487.18, df = 55 (P < 0.00001); ) 2 = 89% Figure 9: Effect of Abelmoschus manihot in addition to a renin-angiotensin system blocker therapy on serum creatinine (SCr). differences between A. manihot plus a RAS blocker and a RAS blocker alone on eGFR. Thus, there was limited evidence to make a conclusion on the ESRD rate and eGFR. The results showed that compared to a RAS blocker, combined treatment of A. manihot and a RAS blocker was associated with significant improvement in proteinuria, UAER, and SCr, and the 24h UP reduction rate as well as normalization of UAER. The results also indicated that A. manihot might be generally well tolerated, because A. manihot added to a RAS blocker did not increase the rates of adverse events. However, due to the generally poor methodological quality and significant heterogeneity, there was currently insufficient evidence to support the routine use of A. manihot for DN. If confirmed in larger high-quality studies, these findings indicate that A. manihot might have an important role in improving proteinuria and protecting kidney function.

Limitations.
Although this review is the most comprehensive meta-analysis to date regarding the safety and efficacy of A. manihot in combination with a RAS blocker for DN patients, there are limitations that should be considered. Firstly, the methodological quality of these studies was generally low. Most described randomization poorly. None of the trials described allocation concealment and blinding. Only one [33] used a placebo control. One study [87] was given a grade of high risk for attrition bias (incomplete outcome data) due to the lack of information on how missing data were handled in the analysis. This meta-analysis carried the risk of reporting bias because not all studies reported all outcomes of interest. All the studies were single centered with generally small sample size, which might have resulted in the lack of power. Heterogeneity was significant among these studies, which weakened confidence in the results. Therefore, the results should be interpreted with caution due to the generally low methodological quality and significant heterogeneity.
Secondly, the study periods for all the identified studies were relatively short, resulting in the lack of evidence on the long-term effects of A. manihot for DN. In this systematic review, two studies [45,91] reported that A. manihot was associated with a greater improvement in UAER after 8week therapy, and the effect could persist for 12 weeks after treatment. However, most trials included assessed the shortterm curative effect and did not continue with the followup to investigate the long-term effects of A. manihot on the prognosis of DN. Therefore, long-term studies are required to identify whether A. manihot could further reduce the rate of the ESRD.
Thirdly, close attention should be paid to ADEs. Safety is a fundamental principle for health care. Current evidence indicated that A. manihot combined with a RAS blocker might be relatively safe for DN. Nineteen of the included trials did not clearly provide data for ADEs despite all clear descriptions of great improvements in proteinuria or SCr with A. manihot therapy in this review. Future studies should pay special attention to ADEs of A. manihot. 4.3. Implication for Practice. DM is the most common cause of ESRD in the developed world [2]. In outcome trials of patients with DN, retrospective analyses demonstrate a robust relationship between magnitude of albuminuria reduction and slowing of CKD progression as well as reduced cardiovascular event rates [7][8][9][10][11][12]. The results indicated that A. manihot in addition to a RAS blocker seemed effective and safe, to reduce albuminuria further in patients with DN. However, due to the generally poor quality and significant heterogeneity, high-quality clinical studies are required to confirm these effects.
In this analysis, the results showed that A. manihot added to a RAS blocker could further improve proteinuria and kidney function in DN patients. Four previous meta-analyses [16][17][18][19] of A. manihot for DN preliminarily reported that A. manihot therapy showed great improvements in proteinuria and kidney function, which was consistent with this analysis. The review found that A. manihot for DN was well tolerated with minimal ADEs. Since the Huangkui capsule gained national approval from the China Food and Drug Administration in 1999, there have been no reports of severe ADEs. Previous meta-analyses [16][17][18][19] of A. manihot for DN reported that the most common adverse event was mild to moderate gastrointestinal discomfort; other ADEs such as dizziness, headache, and dry mouth were rarely reported. In this analysis, nine types of adverse events were observed, including gastrointestinal discomfort, dry mouth, headache, dizziness, liver injury, hypoglycemia, hyperkalemia, coughing, and hypotension. Well-tolerated gastrointestinal discomfort was still the most common ADE. Other side effects were not frequently reported. Rates of adverse events were not significantly different between the study groups except for headache, which was reported to occur more commonly in the control group. Although 19 of the included trials provided no data for ADEs, these studies all clearly reported that A. manihot was associated with significant improvements in proteinuria, SCr, and clinical symptoms. If confirmed, these results suggest that A. manihot might be effective and relatively safe for DN.

Conclusions
A. manihot in addition to a RAS blocker appeared to be effective and safe to further improve proteinuria and protect kidney function in patients with DN. However, due to the generally low methodological quality, significant heterogeneity, and publication bias of included articles, highquality clinical studies are required to confirm these findings before the routine use of A. manihot can be recommended.

Data Availability
All relevant data are within the paper and its supporting information files.

Disclosure
The funders did not contribute to study design, collection, analysis, interpretation of data, writing, or submission for publication.