Chinese Herbal Injections for Primary Nephrotic Syndrome in Adults: A Systematic Review and Network Meta-Analysis

Primary nephrotic syndrome (PNS) is a common renal disease that presents with heavy proteinuria and hypoalbuminemia. Despite notable advances in its treatment, some patients show poor responses and clinical outcomes when treated with conventional Western medicine (WM). Chinese herbal injections (CHIs) have been reported to have beneficial effects for PNS. The aim of the present study was to comprehensively determine the efficacy and safety of CHIs for PNS in adults using a network meta-analysis approach. PubMed, Embase, the Cochrane library, and four Chinese databases were systematically searched to identify randomized controlled trials (RCTs) using CHIs for treatment of PNS published before June 1, 2019. Quality assessment of the identified RCTs was performed according to the Cochrane Handbook. Pooled odds ratios (OR) or mean differences (MD) with corresponding 95% confidence intervals (CI) were calculated for discrete or continuous variables, respectively. The primary outcome was complete/total remission and secondary outcomes were serum albumin and urinary protein excretion. The surface under the cumulative ranking curve (SUCRA) value and cluster analyses were used to rank treatment by probability. Eighty-five studies involving 11 CHIs and 5801 subjects were included. Compared with WM alone, CHI plus WM showed an improved complete/total remission rate as well as higher serum albumin and lower 24-hour urinary protein excretion, except in the following: Yinxingye injection plus WM did not improve the total remission rate, and Dengzhanhua or Xueshuantong injection plus WM did not lower the 24-hour urinary protein excretion. Either Danhong (DH) or Dengzhanhua (DZH) injection plus WM was the preferable treatment for PNS based on SUCRA and cluster analyses of clinical remission and adverse events. However, considering that literature in this area is limited, these results need further validation. CHIs administered as adjuvants to WM showed favourable outcomes for PNS. DH + WM and DZH + WM might be the potential optimal therapies for PNS.


Introduction
Primary nephrotic syndrome (PNS) is a prevalent glomerular disease characterised by excessive proteinuria and hypoalbuminemia, manifesting with histological differences, such as minimal change disease and membranous nephropathy [1]. Renin-angiotensin system (RAS) blockade and/or immunosuppressive agents are recommended treatments. However, patients show varying responses; 20%-50% of patients continue with substantial proteinuria, which is associated with higher risk of end stage kidney disease (ESKD) [2][3][4]. Additionally, failure of clinical remission is associated with increased risk of various complications, such as infections and thromboembolisms. Moreover, recommended treatments for PNS can also cause unwanted side effects [5][6][7]. Traditional Chinese medicine (TCM) has been documented to have a beneficial effect for PNS diagnosed based on individual symptoms, but not the pathological type; thus, TCM might be a potential adjuvant or alternative treatment for PNS [8][9][10]. Chinese herbal injections (CHIs) are innovative formulations of herbs with high bioavailability and rapid action that are widely administered in China [11]. Several traditional pairwise metaanalyses have suggested that CHIs might be effective treatments for PNS [12,13]. However, since pairwise metaanalyses only directly compare two interventions, a comparison of the therapeutic effects of many different CHIs for PNS has not been performed. Network meta-analysis (NMA) can give a unified, coherent analysis of direct and indirect evidence as well as rank the probability of optimal treatment. erefore, in this study, NMA was used to determine the relative efficacy and safety of different CHIs for PNS and predict the best candidate treatment.

Compliance with Ethical Standards.
is work was performed according to the PRISMA extension statement for network meta-analyses (Table S1). e protocol of the present NMA was registered in PROSPERO: CRD42019133746. Because all the data were based on previously published literature, ethical approval and informed consent were not applicable.

Literature Searching. PubMed, Embase, Cochrane
Central Register of Controlled Trials (CENTRAL), and four Chinese databases (the China National Knowledge Infrastructure (CNKI) database, WangFang database, China Science and Technology Journal (VIP) database, and China Biology Medicine (CBM) database) were searched for eligible studies published prior to June 1, 2019. Medical subject headings and free-text searches were used by combining the following three domains without language restrictions: CHI, nephrotic syndrome, and randomized controlled trial (RCT). e specific search terms are shown in Table S2.

Inclusion and Exclusion Criteria.
e inclusion criteria were identified by the Participants, Interventions, Comparisons, Outcomes, and Study design (PICOS) framework: (1) participants: adults diagnosed with PNS; (2) interventions and comparisons: the experimental group was administered a CHI or CHI plus conventional Western medicine ((WM) defined as RAS inhibitors, steroids, immunosuppressive agents, and symptomatic treatments that were the same in both groups), and the control group was treated with WM alone or another CHI with/without WM; (3) outcomes: the primary outcomes were complete remission (CR) or total remission (TR), which were assessed according to the definition provided in each study; secondary outcomes were 24-hour urinary protein excretion, serum albumin, total cholesterol (TC), triglycerides, serum creatinine, adverse reactions (ADRs), ESKD, and all-cause mortality; and (4) study design: RCT. e exclusion criteria were as follows: (1) children with nephrotic syndrome (NS); (2) no data available for analysis; or (3) secondary NS.

Data Extraction and Quality Assessment.
We extracted the following data from each included study: first author; publication year; location; baseline information for all groups (sample size, age, and gender); diagnostic criteria for PNS; details of intervention and control; duration of followup; outcomes; and sources of publication.
e Cochrane Handbook assessment tool for risk of bias (version 5.1.0) was used to assess the quality of each trial. Data extraction and quality assessment were independently performed by two investigators each, and disagreements were solved by a third reviewer.

Statistical
Analysis. All statistical analyses were performed with STATA software (version 14.0) using the network command [14]. Pooled odds ratios (OR) or mean differences (MD) with 95% confidence intervals (CIs) were calculated for discrete or continuous data, respectively. e random-effects network meta-analysis was conducted to compare all classes of CHIs for each prespecified outcome under the frequentist framework. Inconsistency tests were performed to explore the network heterogeneity between designs; P < 0.05 was considered to be significant heterogeneity. e loop-specific approach was used to compare the difference between direct and indirect comparisons within triangular loops in a network; random-effects pairwise metaanalyses were performed and presented as main results if significant inconsistency existed. e surface under the cumulative ranking curve (SUCRA) value was calculated to rank the interventions. Cluster analysis was utilized to comprehensively compare the efficacy and safety of interventions. Publication bias was tested by the comparison-adjusted funnel plot, Begg's test, and Egger's test. We performed sensitivity analyses by comparing the results between random-and fixed-effects models.

Characteristics of Eligible Studies.
A total of 3556 publications were retrieved from electronic databases after removing duplications. Eighty-five studies were identified for meta-analyses according to the inclusion and exclusion criteria. Details of the literature selection process are shown in Figure  ere were 5801 participants (3331 males and 2470 females) enrolled in the 85 RCTs. All experimental groups (3155 participants) received CHIs plus WM and control groups (2967 participants) received WM alone or another CHI with WM. Characteristics of the selected studies are depicted in Table S3. e network graphs for each outcome are presented in Figure 2.

Risk of Bias Assessment.
irteen of the 85 studies described appropriate methods for generating random sequences; thus, their selection bias risk was considered low. e remaining studies were classified as unclear risk because they only mentioned "random" selection. None of the studies reported the processes used for allocation concealment or blinding. us, allocation concealment was considered as high risk. Performance bias and detection bias Studies included in network meta-analysis (n = 85)  were classified as unclear risk due to insufficient information. All studies had complete data, so the attrition bias was evaluated as low risk. Nine studies did not report all the outcomes described in the methods section, so their reporting bias was deemed high risk. e other biases were classified as unclear risk because there were no available details on which to evaluate the risk of bias. In summary, the quality of included RCTs was poor ( Figure 3). NMA results showed that all the groups of PNS patients receiving CHI + WM had significantly better TR rates than those receiving WM alone (OR, 2.92-6.54), except for the YXY + WM group. SXT + WM had a significantly higher TR rate than FFDS + WM (OR, 2.65; 95% CI, 1.33-5.28), and FFDS + WM had a significantly lower TR rate than DZH + WM (OR, 0.35; 95% CI, 0.13-0.98) and DH + WM (OR, 0.50; 95% CI, 0.27-0.91) ( Table 1).
Due to inconsistency in the urinary protein excretion data (P � 0.01), the pairwise meta-analyses are shown as main results.
e results suggest that all groups of PNS patients receiving CHIs + WM had significantly lower urinary protein than groups receiving WM alone (MD,− 0.  Table 3; overall, no serious adverse events were reported. We next performed NMA for total ADR and found there were no significant differences between any of the groups, except the YXY + WM group, which showed lower ADR than the CXQ + WM group (Table S5). ere were no significant differences between serum creatinine levels in any of the groups (Table S6). None of the studies reported on the incidence of ESKD or all-cause mortality.

Rank Probability Based on SUCRA and Cluster Analysis.
e rank of interventions for each outcome, based on SUCRA, is presented in Table 4. Compared to WM alone, combining WM with CHIs showed better clinical benefits. Cluster analyses showed that DH + WM and DZH + WM might be the preferred interventions for PNS based on the outcomes of efficacy and safety (CR, TR, and ADR) (Figure 4). Moreover, SK + WM, SXT + WM, and YXY + WM also showed preferred efficacy outcomes; however, this should be interpreted with caution due to the low number of studies reporting results with YXY + WM and SXT + WM (1 and 3, respectively).

Inconsistency
Test. Inconsistency tests suggested that direct and indirect comparisons were consistent for the outcomes of CR, TR, serum albumin, triglycerides, and serum creatinine. However, several loops showed inconsistency for the outcomes of TC and urinary protein excretion (Table S7), so pairwise meta-analyses were used as the main results for these two outcomes.

Publication Bias.
e comparison-adjusted funnel plot was used to evaluate publication bias according to the primary outcome of CR. Because the funnel plot was relatively symmetrical ( Figure 5), we next used Begg's test and Egger's test to check the publication bias quantitatively, which both showed no statistical differences (Begg's test, z � 0.80, P � 0.42; Egger's test, t � 1.74, P � 0.09).

Sensitivity Analysis.
We performed sensitivity analysis of the primary outcome of clinical remission by comparing the results of random-effects models with fixed-effects models. e sensitivity analysis showed a stable result.

Discussion
Although many new immunological treatments, such as rituximab, have been used in PNS, some patients showed unsatisfactory responses and subsequent unfavourable clinical outcomes [15,16]. An increasing amount of evidence indicates that TCM herbal formulations, because of their diversified effects, might be an effective complementary or alternative regimen for chronic kidney disease, especially primary glomerular disease [17]. Injection of herbal formulations is widely prescribed in China, especially for hospitalised patients; however, the efficacy and safety of these herbal injections are not clear.
is is the first NMA based on RCTs to assess the efficacy and safety of CHIs for treating PNS in adults. In the present NMA, the use of eleven CHIs as adjuvants to WM in PNS was comprehensively compared based on the outcomes of clinical remission, urinary protein excretion, serum albumin, TC, triglycerides, serum creatinine, and ADRs. e treatments evaluated in this study were CXQ + WM, DS + WM, DSCX + WM, DH + WM, DZH + WM, FFDS + WM, HQ + WM, SK + WM, SXT + WM, YXY + WM, and XST + WM. e results suggested that CHIs plus conventional Western pharmaceutical agents are associated with significantly better measured PNS outcomes than WM alone (the effect on serum creatinine was not significantly different between the groups). Among the treatments including CHIs, DH + WM and DZH + WM might be the preferable therapies, according to the results of SUCRA and cluster analyses.
According to TCM theory, the herbs used to formulate the included injections identified in this study are bloodactivating and stasis-removing drugs, with the exception of Huangqi (Astragali Radix), which is a Qi-invigorating herb.
is indicates the importance of blood stasis syndrome and Qi deficiency syndrome in PNS, which is consistent with previous studies [18]. According to TCM theory, Qi deficiency is involved in the pathogenesis of nephrotic syndrome, and blood stasis acts as a key pathological factor to further complicate the condition [8]. A TCM syndrome study in PNS showed that Qi deficiency is one of the most common TCM syndrome types, whose score correlated significantly with higher urine protein/creatinine ratio and TC level [19]. Moreover, 91.19% of patients with primary glomerular disease were diagnosed as blood stasis syndrome, of which PNS had a higher syndrome score [20]. Additionally, the blood stasis syndrome score was positively correlated with urine protein ration, triglyceride, and cholesterol. e results also indicated that the more severe the proliferative and sclerotic renal pathologies were, the higher the score was [21]. Similar results were found in IgA nephropathy [22]. In this context, the crucial TCM concept for treating PNS is to invigorate Qi, activate blood, and remove stasis, which might increase the rationality and applicability of our results. Idiopathic membranous nephropathy, which, besides IgA nephropathy, is a prevalent pathological type of PNS in China [23], has been shown to positively respond to the Qi-invigorating and stasis-removing method; moreover, the most frequently used Chinese medicines have been reported to be Astragali Radix, Angelicae Sinensis Radix, Chuanxiong Rhizoma, and Salvia Miltiorrhiza [24].
Previous studies have suggested that the preferred CHIs have positive effects on glomerular disease, such as reducing proteinuria and blood lipid. In addition, the therapeutic mechanisms for these effects have been partly revealed through experimental studies. For example, several pairwise meta-analyses have shown the beneficial effects of DZH (Breviscapus) as an antiproteinuria, in improving albumin     Evidence-Based Complementary and Alternative Medicine level, and in lowering cholesterol and triglyceride for hypertension-induced renal injury and diabetic nephropathy; the mechanism may be correlated with synergetic suppression of increased oxidative stress [25][26][27]. Recent studies have suggested that DH can effectively decrease urine protein excretion by multiple pathways, as well as confer lipid-lowering effects [28][29][30]. DH is a herbal formulation of Danshen (Salvia Miltiorrhiza) and Honghua (Carthami Flos). eir main components, such as salvianolic acid, have been shown to have a podocyte-protection effect in an adriamycin-induced nephrotic syndrome rat model and a mouse podocyte injury cell model, which may be one of the mechanisms of DH's renoprotective effects [31]. In the present study, SK and SXT also showed preferred efficacy outcomes. SK is a Chinese medical-standardised product extracted from Dahuang (Rhei Radix), Huangqi (Astragali Radix), Danshen (Salvia Miltiorrhiza), and Honghua (Carthami Flos), formulated according to TCM theory and   [32,33]. SXT is extracted from Shuizhi (Hirudo) and Dilong (Pheretima), which are powerful blood-activating and stasis-removing herbs in TCM theory. Pharmacological studies have shown that Shuizhi plays a role in anticoagulation, preventing platelet aggregation and improving hemorheological parameters, which might reflect the mechanism through which SXT is beneficial for PNS [34]. In addition, the Dilong has been shown to reduce glomerular injury and urinary albumin excretion, possibly via suppression of mesangial matrix expansion and activation of matrix metalloproteinase-2 [35]. Apart from efficacy, the safety of CHI in the treatment of PNS is an important issue. Although CHIs, including those evaluated in this study (such as Danshen, Xueshuantong, and Shuxuetong injection), have been used in China for more than half a century, research assessing their safety should be systematically conducted because of their higher risk of adverse effects compared to other forms of TCM medications and the higher percentage of new serious adverse events compared to conventional injections [36]. Few studies included in this NMA focused on adverse reactions; therefore, more evidence is needed to verify the safety of these CHIs.

Evidence-Based Complementary and Alternative Medicine
Potential limitations should be considered when interpreting the results of this study. First, significant heterogeneity and inconsistency were observed when performing comparisons for urinary protein excretion and cholesterol; therefore, we could only conduct pairwise meta-analyses for these outcomes and rank results could not be obtained. Second, the enrolled population was exclusively Chinese; thus, whether our results are applicable to a more general population remains unknown. ird, although all patients were diagnosed as having PNS with the same criteria, the pathological types varied, which may have caused potential heterogeneity. Fourth, the different dosage and duration for CHIs may have led to potential bias. Because the methodological quality of the included RCTs was low and few studies reported adverse events, further high-quality RCTs are required to validate the efficacy and safety of CHIs. We also propose some suggestions towards RCTs on CHI for the treatment of PNS. First, RCTs should be registered in advance to ensure the transparency of the trial process. Second, researchers should report the trial process in as much detail as possible according to the RCT reporting standard, in order to improve the quality, credibility, applicability, and value of the literature. Regarding its design and contents, the present study has two particular strengths. First, it conducted a comprehensive literature search and a comparison for eleven CHIs in treating PNS. Second, an eligibility criterion was formulated before the NMA to reduce the clinical heterogeneity as far as possible. Finally, the results demonstrated the beneficial effects in multiple aspects and provided several clinical suggestions for treatment of PNS.

Conclusions
Taken together, the current work demonstrates that adjuvant therapy with CHIs could have favourable clinical outcomes for PNS; DH + WM and DZH + WM might be the potential optimal PNS treatments. Due to the limited evidence, highquality studies with reasonable design are needed to confirm the efficacy of CHI. Moreover, their safety also requires a systematical assessment.

Evidence-Based Complementary and Alternative Medicine
Data Availability e data used to support the findings of this study are included within the supplementary materials file.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.