Long-term efficacy and safety of Huangqi (Radix Astragali Mongolici)-based Traditional Chinese Medicine in diabetic peripheral neuropathy: a Meta-analysis of randomized controlled trials

Abstract

OBJECTIVE:

To assess the long-term effectiveness of Huangqi (Radix Astragali Mongolici, HQ)-based Traditional Chinese Medicine (TCM) in the treatment of diabetic peripheral neuropathy (DPN).

METHODS:

Nine databases were searched to retrieve available randomized controlled trials that compared HQ-based TCM and Western Medicines in the treatment of DPN. The methodological quality of the included studies was assessed using the Cochrane bias risk tool, and RevMan 5.4 was used for data analysis. The effect estimates of interest were risk ratio (RR), mean difference (MD) or standardized mean difference (SMD) with 95% confidence interval (CI).

RESULTS:

The results from 48 available studies assessing 3759 patients demonstrated that cases administered HQ-based TCM [RR = 1.30, 95% CI (1.21, 1.40), P < 0.000 01] or HQ-based TCM combined with Western Medicines [RR = 1.25, 95% CI (1.19, 1.31), P < 0.000 01] exhibited higher total efficacy rates than individuals who received Western Medicine alone. The results showed that the HQ-based TCM group had decreased Toronto Clinical Scoring System scores [MD =－1.50, 95% CI (－1.83, －1.17), P < 0.000 01], and reduced serum interleukin 6 [SMD = －0.57, 95% CI (－0.87, －0.27), P = 0.0002] and tumor necrosis factors-α levels [SMD = －0.60, 95% CI (－0.95, －0.25), P = 0.0009]. In addition, both HQ-based TCM and HQ-based TCM combined with Western Medicine increased nerve conduction velocity and decreased glycaemia compared with Western Medicine alone. In terms of blood lipids, oxidative stress and adverse drug reactions, there were no significant differences between the HQ-based TCM groups and the Western Medicine control group.

CONCLUSION:

The current Meta-analysis revealed that HQ-based TCM yields higher efficacy and safety than Western Medicine alone for the treatment of DPN, although further well-designed RCTs are required to validate these findings.

1. INTRODUCTION

Diabetic peripheral neuropathy (DPN), a major chronic complication of diabetes, represents the most destructive type of peripheral neuropathy.¹ DPN affects nearly half of diabetic patients with a long disease course, resulting in high morbidity and mortality.^2,3 The global prevalence of DPN is estimated at 15.3-72.3/100 000 person-years.⁴ Its clinical symptoms include pain, paraesthesia and sensory loss in the extremities.^5,6 Complications of end-stage DPN, such as foot ulceration and limb amputation, lead to substantial increases in medical expenses and reductions in quality of life.⁷ Currently, the pathogenesis of DPN is multifactorial, including endoplasmic reticulum stress, DNA damage, mitochondrial dysfunction and apoptosis.⁸ Previous treatments have been based on aldose reductase inhibitors, antioxidants, neurotrophic factors, Gamma-aminobutyric acid analogues and cellular Metabolism agonists.^{9,⇓,⇓,⇓-13} However, these therapies are not effective in a good number of patients. Due to the limitations of present therapies, complementary and alternative medicines are urgently needed.

In recent years, Traditional Chinese Medicine (TCM) has been considered a major complementary strategy for treating DPN in China. Numerous studies have demonstrated that TCM has obvious advantages in the treatment of DPN. Huangqi (Radix Astragali Mongolici, HQ) is derived from Astragalus membranaceus (Fisch.), a Fabaceae considered a primary tonic herb in TCM for more than 2000 years, is broadly utilized for treating DPN clinically in China.^14,⇓-16 Previous research has indicated that HQ is the most frequently used TCM for treating DPN and is also a part of the ancient classic formulae Huangqi Guizhi Wuwu decoction (黄芪桂枝五物汤) and Buyang Huanwu decoction (补阳还五汤), which have been applied for DPN treatment since ancient times.^17,18 Both preparations use HQ as the Monarch (Jun) medicine.

However, according to TCM theory, HQ is a tonify-Qi medicine, not a medicine that promotes blood circulation and removes blood stasis recommended in guidelines.^19,20 In addition, the use of TCM requires a slow adjustment process, especially for chronic Metabolic diseases, in which long-term safety and efficacy are more clinically meaningful. However, the effectiveness of the long-term use of HQ-based TCM therapies for DPN has not been reported. Furthermore, multiple clinical studies have been published in recent decades.^21,22 Thus, the current Meta-analysis aimed to comprehensively analyse the data of these studies to provide insights regarding the efficacy and safety of HQ-based TCM in the treatment of DPN.

2. MATERIALS AND METHODS

The current study was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines²³ and has been registered with PROSPERO (No. CRD42022321654).

2.1. Study selection

Five English and four Chinese databases were systematically searched for all reports published prior to January 31st, 2023. The databases included the Cochrane Library, PubMed, Embase, Web of Science, Clinical Trial Registries, the Chinese Biomedical Literature database (CBM), China National Knowledge Infrastructure (CNKI), Chongqing VIP (CQVIP), and Wanfang. In addition, relevant articles and reference lists were manually searched.

Take search strategy in PubMed for example. The search terms included “Astragalus propinquus [MeSH Terms]” OR “astragalus” OR “radix astragali” OR “huang qi” OR “huangqi” were used to get the result #1 and #2 about HQ. Secondly, combination of “Medicine, Chinese Traditional [MeSH Terms]” OR “Chinese medicine” OR “traditional Chinese medicine” OR “Chinese herb*” OR “oriental medicine” OR “oriental traditional medicine” OR “medicinal plant*” OR “herb*” OR “decoction” OR “tang” were used to get the result #3 about TCM. Thirdly, “Diabetic Neuropathies [MeSH Terms]” OR “diabetes peripheral ne*” OR “DPN” OR “Diabetic complication” OR “diabetic microvascular complication” OR “diabetic polyneuropathy” were used to get the result #7 about DPN. Combining the above results, the final result was gotten.

MeSH terms and free terms were utilized for retrieval. Only Chinese and English publications were included. The complete search strategy is shown in supplementary Table 1. Two investigators (PING Jing and HAO Hongzheng) selected studies and assessed the risk of bias in each trial independently.

2.2. Inclusion criteria

(a) Randomized controlled trials (RCTs). (b) Patients clearly diagnosed with diabetic peripheral neuropathy. (c) Treatment in the trial group consisted of an oral preparation of HQ-based TCM or HQ-based TCM combined with Western medicinal products. The pre-paration should display all names of TCM components. (d) The control group was treated with conventional Western Medicine. (e) Long-term therapy (12 weeks or more) and ≥ 30 cases per group. (f) For multiple studies reporting the same data, the most recent report was considered.

2.3. Exclusion criteria

(a) Nonclinical studies, including animal or cell experiments. (b) Case reports, conference abstracts or reviews. (c) Treatment in the trial group comprised other TCM approaches, e.g., “acupuncture” and “moxi-bustion”. (d) Treatment in control cases comprising TCMs. (e) Cases with serious diseases such as cardiovascular and respiratory system diseases.

2.4. Data collection and quality assessment

Two reviewers (PING Jing, HAO Hongzheng) used a three-stage process to select studies. First, duplicated studies were excluded. Second, title and abstract screening was performed. Finally, the full texts of the retained reports were read. Any disagreements were resolved by discussion with a third investigator (CHENG Gang). The following data were extracted from the included studies: publication year, first author, study design, sample size, patient features and outcomes.

Additionally, two investigators evaluated the method-logical quality of all included studies independently based on the Cochrane risk of bias tool across 7 domains: (a) random sequence generation; (b) allocation concealment; (c) blinding of participants and personnel; (d) blinding outcome assessment; (e) incomplete outcome data; (f) selective reporting; (g) other bias. For each study, these domains were graded as high, unclear or low risk of bias. Review Manager 5.4 was utilized for analysis. Disagreements were resolved by consulting methodologists.

2.5. Outcome definition

The primary outcomes were as follows: (a) Total efficacy rate (TER) referred to the “Guidance principle of clinical study on new drug of Traditional Chinese Medicine”.²⁴ (b) NCV included peroneal motor (PMNCV), peroneal sensory (PSNCV), median motor (MMNCV) and median sensory (MSNCV) NCV, all expressed in m/s. (c) The Toronto Clinical Scoring System (TCSS), a method that combines both symptom and sign assessments, is considered to accurately evaluate DPN. (d) Adverse drug reactions (ADRs) were assessed by measuring the incidence rates of adverse reactions, gastrointestinal reactions, liver or kidney dysfunction, dizziness and other ADRs reported.

The secondary outcomes were as follows: (a) Glycaemia was defined as fasting blood glucose (FBG), 2 h postprandial blood glucose (PBG) and HbA1c. (b) Blood lipid profile, including total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) amounts. (c) Inflammatory status was assessed by detecting serum interleukin 6 (IL-6), tumour necrosis factor-α (TNF-α) and hs-CRP (hyper sensitive C-reactive protein) levels. (d) Oxidative stress was assessed by detecting serum superoxide dismutase (SOD) and malondialdehyde (MDA) levels.

2.6. Statistical analysis

Review Manager 5.4 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark) was utilized in this Meta-analysis. Risk ratios (RRs) and corresponding 95% confidence intervals (CIs) were applied to assess dichotomous variable data, while mean differences (MDs) and standardized mean differences (SMDs) with 95% CIs were used to assess continuous variables. Statistical heterogeneity of therapeutic effects among trials was evaluated by the I ² statistic; I ² > 50% indicated significant heterogeneity. The random effects model was adopted for Meta-analysis. For all outcomes, P < 0.05 indicated statistical significance. Egger's linear regression test and contour-enhanced funnel plots were applied to examine potential publication bias for the main outcomes with Stata version 16.0 (StataCorp, College Station, TX, USA).²⁵ The robustness of the results was examined via sensitivity analysis, including trials with a low risk of bias in randomization.

2.7. Quality of evidence

The quality of evidence was evaluated by the Grading of Recommendations Assessment, Development and Evaluation approach. This assessment of the quality of evidence included risk of bias, heterogeneity, indirectness, imprecision and publication bias, which were rated as high, moderate, low or very low.

3. RESULTS

3.1. Study selection

A total of 12 636 articles were initially retrieved. After screening the titles and removing duplicate studies, 4220 records were retained. After screening the titles and/or abstracts, 1699 full texts were assessed, and 1647 were excluded based on the above eligibility criteria. Ultimately, 48 eligible trials were included in this study (Figure 1).

Figure 1. Flow diagram of study selection for the Meta-analysis.

CNKI: China national knowledge infrastructure database; CBM: Chinese biomedical literature database; VIP: China science and technology journal database; RCT: randomized controlled trial.

The 48 trials were all RCTs, whose features are listed in Table 1. They were all carried out in China with Chinese as the publication language. Sample sizes were 60-159, and 3759 patients were included in total, with 1908 and 1851 in the test and control groups, respectively. Seventeen trial groups received HQ-based TCM alone, and the remaining 31 groups received HQ-based TCM combined with Western Medicine.

Table 1.

Characteristics of the eligible studies in our research

Study	No. (T/C)	Age		Intervention		FBG (mmol/L)		2h PBG (mmol/L)		HbA1c (%)		Duration	Outcome
		Trial	Control	Trial	Control	Trial	Control	Trial	Control	Trial	Control
Cheng RD 201634	108 (72/36)	57.278±9.205	57.03±9.52	Yiqi Huoxue Tongluo prescription	Vitamin B12 5 Capsules tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②④⑧
Dai AQ et al 202035	106 (53/53)	52.22 ±3.63	52.13± 3.52	Yiqi Huayu Tongbi prescription	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①⑥
Ding Q 202136	70 (35/35)	61.54 ±8.35	61.94± 11.36	Compound Qiying granules	Epalrestat 50 mg tid	8.31±3.36	7.50±3.57	8.11± 1.70	8.69±2.06	Nr.	Nr.	12 w	①②④⑦
Ding YM et al 201037	62 (32/30)	56.12 ±3.87	58.32± 2.78	Yiqi Yangxue Huoxue prescription	Vitamin B12 0.5 mg tid	8.23±1.69	9.11±1.85	11.58± 2.32	10.01±1.25	8.9± 1.05	8.2± 1.89	3 m	①②④
Du T 201938	70 (35/35)	60.52 ±8.90	59.65± 9.14	Yiqi Huoxue Tongluo prescription	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	7.05± 0.58	7.11± 0.56	12 w	①②⑧
Du ZL et al 201239	60 (30/30)	Nr.	Nr.	Modified Buyang Huanwu decoction	Vitamin B12 1 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①⑧
					Vitamin B1 20 mg tid
Feng HY 202040	63 (34/29)	58.68 ±7.50	59.84± 7.00	Tongluo Zuzuo prescription	Vitamin B12 0.5 mg tid	6.52±0.71	6.58±0.65	9.61± 0.88	9.51±0.69	6.96± 0.72	6.89± 0.67	12 w	①③④⑧
Han YB 201441	72 (36/36)	53.08 ±8.07	53.69± 8.11	Quyidi capsules	Vitamin B12 0.5 mg tid	6.58±0.20	6.56±0.21	9.34± 0.23	9.42±0.21	7.03± 0.20	7.02± 0.19	12 w	①②④⑤⑦⑧
Hou N 201742	60 (30/30)	54.43 ±7.48	54.93± 7.25	Xiaoke Tongbi capsules	Vitamin B12 0.5 mg tid	6.07±0.37	6.09±0.39	8.70± 0.34	8.67±0.36	6.59± 0.19	6.61± 0.16	12 w	①②④
Hou Y et al 201243	75 (3936)	55.38 ±10.63	56.44± 11.70	Yiqi Yangyin Huoxue prescription	Vitamin B12 0.5 mg tid	11.34±2.86	10.85±2.61	16.48± 2.52	15.80±2.77	9.49± 1.54	9.28± 1.71	3 m	①④⑤⑥
Lan BY 201644	60 (30/30)	53.27 ±8.36	50.67± 8.91	Xiaoke Bitong capsules	Epalrestat 50 mg tid	5.99±0.66	6.24±0.52	9.05± 0.58	9.26±0.67	6.58± 0.20	6.67± 0.14	12 w	①②④⑧
Lei LL 202045	60 (30/30)	37.71 ±12.54	37.45± 12.32	Huangqi Guizhi Wuwu decoction	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②
Lin CY 202146	60 (30/30)	54.30 ±10.3	53.53± 8.50	Yiqi Zhuyu Tongluo decoction	α-Lipoic acid 0.2 mg tid	6.97±0.59	6.65±0.58	9.84± 0.71	9.75±0.68	7.36± 0.36	7.37± 0.42	12 w	①②④⑤⑦⑧
Wang HZ 200947	60 (30/30)	55.3± 7.9	55.9± 7.6	Xiaoke Xubi decoction	Vitamin B12 0.5 mg tid	7.14±0.32	7.14±0.38	9.53± 0.49	9.83±0.43	7.05± 0.30	7.10± 0.31	12 w	①⑧
Wang JJ 201748	62 (32/30)	59.00±11.54	60.07± 13.29	Huangqi Guizhi Wuwu decoction and Siteng Yixian decoction	Vitamin B12 0.5 mg tid	8.37±1.26	8.27±1.07	12.69± 1.83	12.20±1.86	8.45± 0.93	8.62± 1.04	12 w	①⑧
Zhang SL et al 201349	100 (50/50)	58.1± 11.5	57.9± 10.8	Matong granules	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②⑦
Zhang YX 201450	74 (38/36)	55.23± 4.71	58.68± 3.22	Tangbikang prescription	Vitamin B12 0.5 mg tid	7.43± 1.71	7.67±1.25	9.79± 2.40	9.76±2.36	6.94±0.73	6.74± 0.60	12 w	①②④⑤⑧
Bao WP et al 202251	60 (30/30)	54.30± 12.61	57.40± 10.74	Jiangtang Tongbi decoction+C	Epalrestat 50 mg tid	10.00± 2.50	9.70±2.60	14.30± 3.00	13.60±3.60	9.50±1.40	10.10± 2.20	12 w	①③④
Ding HY 201852	92 (46/46)	55.91± 4.23	56.75± 3.63	Modified Huangqi Guizhi Wuwu decoction+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②⑧
Fan LH 201453	60 (30/30)	59.33± 7.46	60.58± 7.81	Qigui Tangtongning granules+C	Alprostadil 10 μg qd	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①④⑤⑧
Jia ZH 202154	90 (45/45)	60.16± 8.03	59.84± 7.98	Yiqi Huoxue Tangtong prescription+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	8.61±1.02	8.54± 0.97	12 w	①④
Li AQ et al 202255	120 (60/60)	50.40± 5.59	50.98± 5.22	Wenjing Huoxue prescription+C	Vitamin B12 0.5 mg tid	6.17± 0.84	6.23±0.64	10.65± 0.93	10.34±0.97	7.24±0.76	7.12± 0.92	12 w	①②③⑦⑧
Li WD et al 201656	90 (45/45)	62.19± 2.93	68.68± 3.96	Modified Huangqi Guizhi Wuwu decoction+C	Vitamin B12 0.5 mg tid	8.45± 0.53	8.36±0.66	12.74± 0.80	12.79±0 84	8.76±0.58	8.58± 0.62	3 m	①②④⑤
Li XF 201457	60 (30/30)	59.83± 7.29	60.03± 9.41	Wenyang Tongluo prescription+C	Vitamin B12 0.5 mg tid；Epalrestat 50 mg tid	9.31± 2.09	9.10±1.91	13.54± 2.69	13.40±2.34	9.67±1.67	9.40± 1.78	12 w	①②④⑥⑧
Lin ZH et al 201558	102 (50/52)	62.88	64.58	Qutan Huoxue Tongluo prescription+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②⑧
Liu XJ et al 201959	70 (35/35)	Nr.	Nr.	Self-made Huangqi Mugua prescription+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②
Lu SM et al 201360	120 (60/60)	55.23± 4.71	58.68± 3.22	Tangbikang decoction+C	Vitamin B12 0.5 mg bid	8.59± 1.55	8.61±1.55	12.79± 2.40	12.76±2.36	9.14±1.17	9.05± 1.16	12 w	①②④⑤
					Vitamin B1 100 mg tid
					Vitamin B6 100 mg tid
Lu WF et al 202161	96 (48/48)	64.76± 9.17	64.76± 9.17	Compound Qidan prescription+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②③⑦
Ning RZ et al 202062	60 (30/30)	62.53± 3.67	61.43± 4.29	Modified Buyang Huanwu decoction+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①⑧
Ouyang CL 202263	60 (30/30)	61.27± 8.95	62.63± 8.75	Juanbi Tongluo oral liquid+C	Vitamin B12 0.5 mg tid Epalrestat 50 mg tid	9.01± 1.42	8.61±2.21	15.20± 2.52	14.91±3.70	8.97±1.29	8.95± 1.67	12 w	①③④⑤⑧
Pan XJ 201764	60 (3030)	50.74± 9.25	49.62± 9.83	Modified Buyang Huanwu decoction+C	Vitamin B12 0.5 mg tid	7.62± 0.17	7.59±0.23	10.36± 0.48	10.57±0.53	7.75±0.49	7.81± 0.52	12 w	①②③④⑧
Qian LW et al 201165	80 (40/40)	57.5± 7.9	57.2± 8.4	Yiqi Huoxue Tongbi prescription+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	3 m	①②
Sun JJ et al 202166	60 (30/30)	56.38± 9.32	57.89±9.27	Shenqi Jiangtang granules+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	6 m	①②③⑧
Sun XY et al 202067	60 (30/30)	66.83± 2.12	65.93±3.24	Mudan granules+C	Vitamin B12 0.5 mg tid	6.90± 0.67	7.09±0.66	11.46± 3.26	10.19±2.72	7.26±0.49	7.25±0.63	12 w	①⑧
Tian SL 201968	60 (30/30)	Nr.	Nr.	Qigui Jiedu Tongluo decoction+C	Vitamin B12 0.5 mg tid	9.91± 1.11	9.60±0.88	11.93± 1.35	11.60±0.88	8.38±0.89	8.32±1.07	12 w	①④⑤⑧
Wang ZF et al 201069	159 (80/79)	62.68± 7.3	62.78±7.57	Huangqi Guizhi Wuwu decoction+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②
Wei Q et al 201870	60 (30/30)	65.1±6.3	64.5± 5.8	Taditional Chinese Medicine+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	8.37±0.98	8.44±1.39	12 w	①④
Wu Y et al 201571	120 (60/60)	Nr.	Nr.	Buyang Huanwu decoction+C	Vitamin B12 0.5-1 mg qd	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②
Xu S 202172	60 (30/30)	61.07± 6.87	61.87±6.68	Yiqi Yangyin Zhuyu decoction+C	Vitamin B12 0.5 mg tid	7.20± 0.93	7.11±1.01	11.08± 1.44	11.26±1.58	7.37±1.24	7.28±0.91	12 w	①②③④
Xu XJ 201973	60 (30/30)	52.67± 8.96	56.63±8.18	Yiqi Yangyin Tongluo prescription+C	Vitamin B12 0.5 mg tid	5.97± 0.75	6.15±0.59	8.31± 0.68	8.10±0.55	9.75±2.09	9.55±1.45	12 w	①②③④⑤⑧
Yan J et al 202174	98 (49/49)	57.24± 10.83	56,55±11.22	Zicui Juanbi prescription+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②⑦
Yang WJ et al 202175	60 (30/30)	52.8± 7.46	55.3± 6.417	Bushen Tongluo prescription+C	Vitamin B12 0.5 mg tid	6.02± 0.56	6.14±0.65	8.76± 0.70	8.63±0.88	8.48±0.68	8.50±0.47	12 w	①③④⑧
Yang XQ et al 201676	96 (48/48)	55.3± 11.4	56.6± 10.8	Huangqi Guizhi Wuwu decoction+C	Epalrestat 50 mg tid	8.60± 1.40	8.80±1.20	14.40± 2.20	14.60±2.30	8.20±1.60	8.50±1.30	12 w	②④
					Vinpocetine 30 mg qd
Zhang Y et al 200877	60 (30/30)	60.13± 10.38	60.87±11.51	Huoxue Jiangtang capsules+C	Vitamin B12 0.5 mg tid	10.30± 2.00	10.00±1.70	15.90± 2.70	14.50±2.30	Nr.	Nr.	3 m	①②④
Zhao W et al 200878	60 (30/30)	58.3± 5.4	57.3± 4.0	Tangluoan capsules+C	Vitamin B12 0.5 mg tid	6.30± 0.70	6.20±0.80	8.70± 0.70	8.40±0.70	Nr.	Nr.	3 m	①②④⑧
Zhao YY 201979	80 (40/40)	52.04± 12.02	53.58±11.80	Qigui Tongluo granules+C	α-Lipoic acid 0.3 mg bid	8.54± 0.78	9.02±1.12	12.43± 1.30	11.09±0.66	7.80±0.74	7.55±0.42	12 w	①②④⑦
Zhou YH 201580	102 (52/50)	56.4± 10.1	54.9± 9.8	Xiaoma Zhitong granules+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	①②③⑦
Zhou ZN et al 201581	122 (64/58)	63.6±9.7	65.2± 10.7	Xiaoke Tongbi prescription+C	Vitamin B12 0.5 mg tid	Nr.	Nr.	Nr.	Nr.	Nr.	Nr.	12 w	②

Notes: Nr.: not report. ①: Total efficacy rate; ②: Nerve conduction velocity; ③: Toronto Clinical Scoring System; ④: Glycemia level; ⑤: Blood lipids profile; ⑥: Inflammatory factors; ⑦: Oxidative stress indicators; ⑧: Adverse drug reactions. w: week; m: month.

Next, the study quality was assessed for all trials. All articles indicated that a randomization method was used. A total of 28 trials utilized random sequence generation and were considered to have a low risk of bias for this domain; the other included studies only mentioned “random” and were thus considered to have an unclear risk of bias for this domain. In total, 3 studies^34,36,48 indicated blinded allocation, with only Ding³⁶ using sealed envelopes as the method of allocation concealment, so these studies were considered to have a low risk of bias for this domain. The remaining trials were considered to have an unclear risk of bias for this domain. Concerning the implemented blinded method, 3 studies^34,36,48 indicated using the blinding method; other studies did not disclose whether investigators, patients or evaluators were blinded, and thus, they were considered to have an unclear risk of bias for this domain. Next, reporting bias was examined. The prescriptions used in 34 studies were self-made prescriptions, indicating a certain level of reporting bias, and these studies were considered to have an unclear risk of bias for this domain. The other studies were considered to have a low risk of bias for this domain. No follow-up bias or other biases were detected, and all trials were considered to have a low risk of bias (supplementary Figures 1, 2).

Figure 2. Forest plot for total efficacy rate analysis.

3.2. Total efficacy rates

Total efficacy rates were evaluated in 40 included trials, including 14 that assessed HQ-based TCM mono-therapy in comparison with Western Medicine controls and 26 that assessed HQ-based TCM combined with Western Medicine in comparison with Western Medicine alone. The HQ-based TCM [n = 948, RR = 1.30, 95% CI (1.21, 1.40), P < 0.000 01; heterogeneity: P = 0.88, I ² = 0%] and HQ-based TCM + Western Medicine [n = 2178, RR = 1.25, 95% CI (1.19, 1.31), P < 0.000 01; heterogeneity: P = 0.22, I ² = 17%] groups had markedly higher efficacy rates than the Western Medicine alone group (Figure 2).

3.3. Nerve conduction velocity

PMNCV was assessed in 26 included trials, including 6 that assessed HQ-based TCM monotherapy and 20 that HQ-based TCM combined with Western Medicine. HQ-based TCM and HQ-based TCM + Western Medicine had markedly better PMNCV than Western Medicine controls [n = 436, RR = 1.88, 95% CI (0.33, 3.43), P = 0.02; heterogeneity: P = 0.02, I² = 63%] and [n = 1773, RR = 3.14, 95% CI (2.55, 3.74), P < 0.000 01; heterogeneity: P < 0.000 01, I² = 70%], respectively (Figure 3).

Figure 3. Forest plot for PMNCV analysis.

PMNCV: peroneal motor nerve conduction velocity.

For PSNCV and MMNCV, both HQ-based TCM and HQ-based TCM + Western Medicine also showed had significantly better NCV than Western Medicine controls. However, for MSNCV, HQ-based TCM alone showed no significant difference compared with Western Medicine controls [n = 274, MD = 0.41, 95% CI (－1.00, 1.82), P = 0.57; heterogeneity, P = 0.37, I ² = 4%], while HQ-based TCM combined with Western Medicine showed significantly better NCV than Western Medicine controls [n = 1305, MD = 3.01, 95% CI (2.14, 3.89), P < 0.000 01; heterogeneity P < 0.000 01, I² = 84%]. It is worth noting that for sensory nerve conduction velocity, HQ-based TCM combined with Western Medicine had significantly higher effectiveness than HQ-based TCM alone (P = 0.001 for PSNCV and P = 0.002 for MSNCV), while for motor nerve conduction velocity, both subgroups were comparable (P = 0.14 for PMNCV and P = 0.27 for MMNCV; supplementary Table 2 and supplementary Figures 3-5).

3.4. Toronto clinical scoring system (TCSS) scores

TCSS scores were evaluated in 11 included trials. The test groups showed markedly lower TCSS scores than the Western Medicine control groups [n = 785, MD = －1.50, 95% CI (－1.83, －1.17), P < 0.000 01; heterogeneity: P = 0.12, I² = 35%] (supplementary Figure 6).

3.5. Glycaemia

Twenty-six trials reported glycaemia. The results demonstrated that compared with the Western Medicine control group, both HQ-based TCM and HQ-based TCM + Western Medicine groups had significantly reduced fasting blood glucose [MD = －0.42, 95% CI (－0.64, －0.20), P = 0.0002] and [MD = －0.59, 95% CI (－0.84, －0.33), P < 0.000 01], 2 h postprandial blood glucose [MD = －0.45, 95% CI (－0.71, －0.19), P = 0.0007] and [MD = －1.10, 95% CI (－1.55, －0.65), P < 0.000 01], and HbA1c levels [MD = －0.35, 95% CI (－0.52, －0.19), P < 0.0001] and [MD = －0.50, 95% CI (－0.82, －0.18), P = 0.002]. Notably, there were significant differences between the HQ-based TCM and HQ-based TCM + Western Medicine groups in 2 h postprandial blood glucose levels (supplementary Figures 7-9).

3.6. Blood lipids

Nine included trials reported blood lipid levels. The results demonstrated that in comparison with the Western Medicine control group, both HQ-based TCM and HQ-based TCM + Western Medicine groups had reduced LDL-C levels [MD = －0.33, 95% CI (－0.60, －0.06), P = 0.02] and [MD = －0.48, 95% CI (－0.93, －0.04), P = 0.03]. However, the remaining blood lipid indicators, including TC, TG and HDL-C, showed no significant differences between the control group and the HQ-based TCM or HQ-based TCM + Western Medicine group (supplementary Figures 10-11).

3.7. Inflammatory factors

Inflammatory factors were assessed in 4 included studies. The trial group showed markedly reduced levels of inflammatory factors compared with the Western Medicine control group in terms of IL-6 [n = 92, SMD = －0.57, 95% CI (－0.87, －0.27), P = 0.0002], TNF-α [n = 152, SMD = －0.60, 95% CI (－0.95, －0.25), P = 0.0009] and hs-CRP [n = 122, SMD = －1.01, 95% CI (－1.65, －0.37), P = 0.002] (supplementary Figure 12).

3.8. Oxidative stress indicators

Oxidative stress indicators were assessed in 6 included studies. The results demonstrated that compared with the Western Medicine control group, both HQ-based TCM and HQ-based TCM + Western Medicine groups had no significant differences in serum contents of oxidative stress indicators, including SOD [n = 85, SMD = 0.30, 95% CI (－0.21, 0.81), P = 0.25] and [n = 182, SMD = 0.86, 95% CI (－0.47, 2.18), P = 0.21] and MDA [n = 85, SMD = －0.19, 95% CI (－0.96, －0.57), P = 0.62] and [n = 130, SMD = －1.06, 95% CI (－2.74, 0.63), P = 0.22] (supplementary Figure 13).

3.9. Adverse drug reactions

ADRs were assessed in 26 of the 48 analyzed trials. Eighteen trials reported no ADRs during treatment. Eight studies reported specific adverse reactions, mainly abdominal pain and diarrhoea, nausea and vomiting and dizziness. The HQ-based TCM and Western Medicine control groups had no significant differences in ADRs (Figure 4).

Figure 4. Forest plot for the analysis of adverse drug reactions.

Other adverse reactions included mild elevation of transaminases in 1 case,⁵⁸ anaphylaxis in 2 cases^{52, 66}), facial swelling in 1 case,³⁴ decreased appetite in 1 case ⁶⁶ and hypoglycaemia in 1 case.⁵² All ADRs abated spontaneously, with no incidence differences among groups.

3.10. Sensitivity analysis

Sensitivity analysis was assessed by omitting one study at a time to assess the main outcomes for data robustness, including TER, PMNCV, PSNCV, MMNCV, MSNCV and TCSS score. Sensitivity analysis revealed that no significant changes were detected in various outcomes, suggesting the reliability of our findings.

3.11. Publication bias

Publication bias was evaluated by Egger’s test, which detected no significant publication bias in PMNCV (P = 0.5524), PSNCV (P = 0.5363), MMNCV (P = 0.3716), MSNCV (P = 0.1531) or TCSS score (P = 0.4086). However, TER data had overt publication bias (P = 0.0000; Table 2). Then, further analysis was performed by combining nonparametric trim-and-fill analysis with contour-enhanced funnel plots to analyse publication bias for TER. The results showed that missing data all fell within the range of no statistical significance (P > 0.1), suggesting publication bias and the possibility that negative results were not reported (supplementary Figure 14).

Table 2.

Results of Egger’s test

Outcome	Egger’s test (P value)
Total efficacy rate	0.0000
peroneal motor nerve conduction velocity	0.5524
peroneal sensory nerve conduction velocity	0.5363
median motor nerve conduction velocity	0.3716
median sensory nerve conduction velocity	0.1531
TCSS score	0.4086

Notes: TCSS: Toronto Clinical Scoring System.

3.12. Quality of evidence

The quality of evidence was assessed according to GRADE criteria. Among the included RCTs, only 3 studies reported allocation concealment and blinding information; thus, all outcomes were downgraded one level in the risk of bias domain. The total efficacy rate was assessed as low-quality evidence, owing to the existence of obvious publication bias. Median nerve MNCV and median nerve SNCV were evaluated as low-quality evidence because of clinical heterogeneity among studies. Peroneal nerve MNCV, peroneal nerve SNCV and TCSS score were assessed as moderate-quality evidence (Table 3).

Table 3.

Quality of evidence of primary outcomes

Outcome	Certainty assessment								No. of patients			Effect Size		Certainty
	No. of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias		Intervention group	Control group		Relative effects (95% CI)	Absolute effects (95% CI)
TER	40	RCTs	Seriousa	Not serious	Not serious	Not serious	Seriousc		1381/1574 (87.7%)	1060/1552 (68.3%)		RR 1.25 (1.21 to 1.30)	171 per 1000 ( 143 to 205 )	⨁⨁◯◯ low
PMNCV	26	RCTs	Seriousa	Not serious	Not serious	Not serious	Not serious		1111	1098		-	MD 2.94 more (2.39 more to 3.48 more)	⨁⨁⨁◯ Moderate
PSNCV	24	RCTs	Seriousa	Not serious	Not serious	Not serious	Not serious		1073	1026		-	MD 2.71 more (2.21 more to 3.22 more)	⨁⨁⨁◯ Moderate
MMNCV	23	RCTs	Seriousa	seriousb	Not serious	Not serious	Not serious		994	955		-	MD 3.16 more (2.45 more to 3.87 more)	⨁⨁◯◯ low
MSNCV	19	RCTs	Seriousa	seriousb	Not serious	Not serious	Not serious		792	787		-	MD 2.68 more (1.88 more to 3.48 more)	⨁⨁◯◯ low
TCSS	11	RCTs	Seriousa	Not serious	Not serious	Not serious	Not serious		396	389		-	MD 1.5 fewer (1.83fewer to 1.17 fewer)	⨁⨁⨁◯ Moderate

Notes: TER: Total efficacy rate; PMNCV: peroneal motor nerve conduction velocity; PSNCV: peroneal sensory nerve conduction velocity; MMNCV: median motor nerve conduction velocity; MSNCV: median sensory nerve conduction velocity; TCSS: Toronto Clinical Scoring System; RCT: randomized controlled trial; TCM: Traditional Chinese Medicine; RR: Risk ratio; MD: mean difference; CI: confidence interval; ^a: Most information is from studies at unclear risk of bias; ^b: exist clinical heterogeneity; ^c: exist obviously publication bias. Patient or population: Diabetic Peripheral Neuropathy; Setting: Randomized Controlled Trials; Intervention: HQ-based TCM or HQ-based TCM combined with Western Medicines; Comparison: Western Medicines; ⨁: High quality ofevidence; ◯: Low quality of evidence.

4. DISCUSSION

As an important part of complementary and alternative medicines, TCM, especially HQ-based TCM, is widely administered to DPN patients in China. To assess whether HQ-based TCM has clinical therapeutic effectiveness, 48 RCTs assessing 3759 DPN cases were Meta-analysed. This represents the first systematic review and Meta-analysis evaluating the long-term efficacy and safety of HQ-based TCM in DPN.

As shown above, HQ-based TCM and HQ-based TCM combined with Western Medicine showed superior effects compared with Western Medicine with respect to elevating peroneal nerve and median nerve conduction velocity. Furthermore, the above two treatment approaches effectively reduced the TCSS score, which indicated significantly improved clinical symptoms. In terms of glycaemia, both HQ-based TCM and HQ-based TCM combined with Western Medicine markedly reduced glycaemia-related indicators, including FBG, 2 h postprandial blood glucose and HbA1c, especially 2 h postprandial blood glucose, for which HQ-based TCM combined with Western Medicine showed a more obvious advantage. However, in terms of blood lipids, no significant differences were found in comparison with control cases in the HQ-based TCM or HQ-based TCM + Western Medicine group, except in LDL-C levels. Overall, the effect on blood lipids was not as significant as the effect on glycaemia. In terms of inflammatory factors, both HQ-based TCM alone and HQ-based TCM combined with Western Medicine obviously reduced serum IL-6 and TNF-α levels in DPN patients, while in terms of oxidative stress indicators, neither group showed an advantage compared with the Western Medicine control group in decreasing serum SOD and MDA levels in DPN patients. Combined with previous mechanistic analysis, the main regulatory role of HQ-based TCM may be to regulate blood glycaemia rather than blood lipid levels. In addition, a further analysis of regulatory sites based on known data demonstrated that HQ-based TCM mainly affects inflammatory factors, which may indicate a possible involvement of the Advanced Glycation End Product (AGE)-Receptor for AGE pathway in diabetic complications. On the other hand, the effects on oxidative stress indicators were not obvious, inferring that the regulated pathway may not be associated with reactive oxygen species, as shown in supplementary Figure 15 (quoted from a reference except for the red analysis part).⁸

In terms of clinical safety, treatment durations in the included trials were 3 months or more. Among all included studies, 26 assessed ADRs, including 18 trials that reported no adverse reactions in both trial and control groups during the whole treatment process. The remaining 8 studies reported concrete adverse reactions such as abdominal pain, diarrhoea, nausea/vomiting, dizziness and mild elevation of transaminases. There were no significant differences compared with control cases. Our results indicated that HQ, as the Monarch (Jun) medicine of the TCM preparation, has good safety in the long-term treatment of DPN.

According to the theory of Chinese medicine, DPN is considered a chronic disease that leads to “deficiency”. Qi deficiency leads to weak blood flow and blockage of blood vessels. Blood stasis can block the transportation of Qi, blood, body fluids and other small molecules, thus forming a vicious cycle and affecting each other. Qi deficiency is the internal and main reason, and blood stasis is the outer symptom. If Qi deficiency is not replenished, only methods promoting blood circulation could deplete Qi and blood, while only invigorating Qi without blood-activating medicines could even worsen blood stasis.²⁶ HQ is considered one of the best medicines for tonifying Qi and is listed first as a monarch medicine. Its significance may be to follow the guidance of the TCM theory of “treating Qi first, treating blood follow, Qi moves blood”.²⁷

Three previous articles discussed the role of TCM in DPN from the perspective of TCM but were all limited by the number of trials and sample sizes, with no discussion on safety and mechanism.^28,⇓-30 They reviewed from the perspective of a broad category of TCMs. Our research focuses on HQ as a monarch medicine. The name HQ comes from Shen Nong Ben Cao Jing, the earliest known book on Chinese medicine. Therefore, HQ has a history of use for thousands of years. Contemporary pharmacological studies have shown that HQ components include polysaccharides, flavonoids, saponins, and so on, which have immunomodulatory, antihyperglycaemic, anti-inflammatory, antiviral and other pharmacological properties.³¹ Astragalus polysaccharides markedly decrease glycaemia and serum triglyceride, carbohydrate and low-density lipoprotein amounts in rats with experimental diabetes. Ryu and collaborators revealed that HQ exerts anti-inflammatory effects in zymosan air-pouch mice by decreasing inducible nitric oxide synthase, cyclooxygenase-2 (COX-2), IL-6 and IL-1β levels.^32,33 In animal and human studies, HQ-based Chinese medicines have been shown to markedly alleviate DPN, corroborating our findings.

The limitations of this study should be mentioned. First, only reports published in Chinese and English were included, and those written in Japanese, Korean and other languages might have not been considered. In addition, all assessed RCTs were carried out in China and published in Chinese. Second, only 28 trials overtly reported randomization techniques, while the status of other trials was “unclear”. In the majority of included studies, the risk of bias was high for the allocation concealment and blinding domains, which may potentially cause implementation and selective biases. Third, Egger’s test and nonparametric trim-and-fill analysis with contour-enhanced funnel plots revealed possible publication bias in the total efficacy rate, suggesting that trials with negative results might have been obtained but not published. Therefore, the latter outcome should be interpreted with caution. Fourth, significant heterogeneity (I ² > 50%) was detected among trials for multiple outcomes, including glycaemia, blood lipids, inflammatory factors and oxidative stress indicators. HQ is seldom utilized as monotherapy and is generally used in combination with other medicinal plants; this also increases clinical heterogeneity, which may weaken the current findings. We look forwards to more well-designed studies to support our findings. Despite these limitations, we still believe this Meta-analysis was meaningful, providing insights into the possible benefits of HQ-based TCM in DPN. Further well-designed trials following CONSORT guidelines are warranted to verify the effectiveness of HQ-based TCM.

In conclusion, the present systematic review and Meta-analysis revealed that HQ-based TCM preparations are overtly superior to Western Medicine controls as monotherapies in total efficacy rate and NCV. In addition, HQ-based TCM can significantly reduce glycaemia and IL-6 and TNF-α levels in DPN patients, with good safety in the long-term treatment process. However, due to the above limitations, further large well-designed RCTs in multiple countries are warranted to promote the development and application of HQ-based TCM clinically.

5. SUPPORTING INFORMATION

Supporting data to this article can be found online at http://journaltcm.cn.