Chinese Herbal Medicine for Type 2 Diabetes Mellitus With Nonalcoholic Fatty Liver Disease: A Systematic Review and Meta-Analysis

Objectives: To evaluate the efficacy and safety of Chinese herbal medicine (CHM) for type 2 diabetes mellitus (T2DM) with nonalcoholic fatty liver disease (NAFLD) with current evidence. Methods: This study was registered in PROSPERO as CRD42021271488. A literature search was conducted in eight electronic databases from inception to December 2021. The primary outcomes were lipid indices and liver functions, including triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), alanine transaminase (ALT), and aspartate transaminase (AST). Review Manager 5.2 and Stata v14.0 were applied for analysis. Results: The research enrolled 18 RCTs with 1,463 participants. Results showed CHM combined with western medicine (WM) was more effective than WM alone in TG (weighted mean differences (WMD) = −0.35.95% confidence interval (CI) [−0.51, −0.19], p < 0.0001), TC (WMD = −0.58.95%CI [−0.80, −0.36], p < 0.00001), LDL-C (WMD = −0.37, 95%CI [−0.47, −0.26], p < 0.00001), HDL-C (WMD = 0.20, 95%CI [0.10, 0.29], p < 0.0001), ALT (WMD = −4.99, 95%CI [−6.64, −3.33], p < 0.00001), AST (WMD = −4.76, 95%CI [−6.35, −3.16], p < 0.00001), homeostatic model assessment of insulin resistance (WMD = −1.01, 95%CI [−1.22, −0.79], p < 0.00001), fasting blood glucose (WMD = −0.87, 95%CI [−1.13, −0.61], p < 0.00001), 2-h postprandial glucose (WMD = −1.45.95%CI [−2.00, −0.91], p < 0.00001), body mass index (WMD = −0.73.95%CI [−1.35, −0.12], p = 0.02), and overall effective rate (risk ratio (RR) = 1.37.95%CI [1.29, 1.46], p < 0.00001). Conclusion: The CHM in combination with WM seems to be more beneficial in T2DM with NAFLD patients in improving lipid and glucose metabolism, liver function, and insulin resistance as well as improving overall efficiency and reducing body weight. Given the poor quality of reports from these studies and uncertain evidence, these findings should be interpreted cautiously. Systematic Review Registration: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021271488, identifier CRD42021271488.

resistance as well as improving overall efficiency and reducing body weight. Given the poor quality of reports from these studies and uncertain evidence, these findings should be interpreted cautiously.

INTRODUCTION
Nonalcoholic fatty liver disease (NAFLD), also known as metabolism-associated fatty liver disease (MAFLD), represents a clinical syndrome characterized by steatosis and accumulation of fat in liver parenchymal cells (Eslam et al., 2020). NAFLD includes simple fatty liver, nonalcoholic steatohepatitis (NASH), and cirrhosis (Younossi et al., 2018). The prevalence of NAFLD is as high as 75% in population diagnosed with type 2 diabetes mellitus (T2DM) (Hua et al., 2014;Younossi et al., 2016). T2DM and NAFLD can accelerate disease progression in a reciprocal manner, thereby becoming an important social public health burden (Loria et al., 2013;Mantovani et al., 2018). NAFLD could increase glycemic excursions, making it more challenging to glycemic control glycemic (Bril and Cusi, 2017), as a result significantly increase the risk of macrovascular and microangiopathy (Hazlehurst et al., 2016;Lomonaco et al., 2016). T2DM, on the other hand, exacerbates the risk of liver fibrosis and cancer in NAFLD (Targher et al., 2021). The coexistence of the two pathologies  leads to a significant increase in the risk of aggravated metabolic disorders and cardiovascular disease, culminating in death related to liver disease (Portillo-Sanchez et al., 2015). Most T2DM patients with NAFLD experience "late detection, late treatment, and difficulty in recovery" because of the lack of typical or serious clinical symptoms or signs in the early stages, thereby impacting the physical and mental health and quality of life (Hu et al., 2017). Obesity and insulin resistance (IR) are major pathogenic drivers in NAFLD and T2DM (Buzzetti et al., 2016;Wei et al., 2021), and these two pathological conditions usually coexist (Tilg et al., 2017;Targher et al., 2018). Due to the prevalence of T2DM with NAFLD and the many potential health risks associated with their coexistence, active and effective prevention measures should be employed to protect this population (Wong et al., 2018;Lee et al., 2020). At present, there is no effective pharmacotherapy for NAFLD approved by the international authorities, regardless of T2DM status. (European Association for the Study of the Liver, 2016). Lifestyle modifications are key to the clinical management of NAFLD across the disease spectrum (Nguyen and George, 2015). Both T2DM and NAFLD promote deterioration of each other physiologically and pathologically, and there are many restrictions in treatment, thereby precluding an effective interventional strategy. Therefore, alternative therapy is urgently needed.

Search Strategies
We performed a comprehensive search of eight electronic databases from inception to December 2021, including PubMed, Cochrane Library, Embase, Web of Science, China National Knowledge Infrastructure (CNKI), Wanfang Database, China Biomedical Medicine database (CBM), and the VIP information resource integration service platform (cqvip). In addition, the Chinese Clinical Trial Registry (CHiCTR) (http://www.chictr.org.cn/index. aspx) and ClinicalTrials.gov were also searched for randomized controlled trials (RCTs) that were either ongoing or completed but unpublished. We included all RCTs that examined the efficacy of CHM in the management of T2DM with NAFLD. Search terms included were as follows: "Traditional Chinese Medicine," "Traditional Tongue Diagnosis," "Zhong Yi Xue," "Chung I Hsueh," "Diabetes Mellitus, Type 2," "Diabetes Mellitus, Noninsulin-Dependent," "Stable Diabetes Mellitus," "Diabetes Mellitus, Type II," and "Maturity-Onset Diabetes," etc. Comprehensive search strategies for the databases are shown in the Supplementary Files (Supplementary Table S1). No restrictions were applied on language.

Inclusion and Exclusion Criteria
All RCTs evaluating the effects of CHM on T2DM with NAFLD were included in the meta-analysis. The inclusion criteria were as follows: 1) Study design: RCTs; 2) Participants: patients with a definite diagnosis of T2DM with NAFLD and no limitations relating to gender, nationality, ethnicity, and education level. 3) Interventions: patients in the intervention group should receive CHM (including decoction, pills, and granules, regardless of duration) plus WM, and the control group should be treated with WM the same as the intervention group; 4) Outcomes: the primary outcomes of the study were triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), alanine transaminase (ALT), and aspartate transaminase (AST). The secondary outcomes included were homeostatic model assessment of insulin resistance (HOMA-IR), fasting blood glucose (FBG), 2-h postprandial glucose (2hPG), body mass index (BMI), overall effective rate, and adverse effects. Studies that met any of the following criteria were excluded: 1) non-RCTs, such as retrospective studies, animal experiments, case reports, reviews, and conference abstracts. 2) Patients received other TCM interventions, including acupuncture, massage, or moxibustion. 3) Studies that lacked sufficient details on outcomes.

Study Selection and Data Extraction
Three investigators (LL, ZX, and X-YZ) searched and screened for appropriate studies according to the predefined inclusion and exclusion criteria. In the case of discrepancies, the final decision was made through consensus agreement. To manage literature, Endnote V.X9 software was used. Two reviewers (HW and XL) independently extracted relevant data from the eligible studies using standardized extraction forms, including: the first author, year of publication, country, sample size, average age, gender, duration of disease, interventions, details of CHM (name of the prescription and composition), adverse events, and outcomes. The extracted data were cross-checked by HW and XL, and a third reviewer (SY) was available to resolve any conflicts.

Risk of Bias Assessment
Two reviewers (CX and X-GZ) independently assessed the risk of bias according to the Cochrane Collaboration's Risk of Bias tool (Higgins et al., 2019), which included the following criteria: random sequence generation, allocation concealment, incomplete data, blinding, selective reporting, and other bias. The results were judged as 'low,' 'high,' or 'unclear,' and any disagreements were resolved by the third investigator (SY).

Data Synthesis and Statistical Analysis
Review Manager 5.2 was applied to analyze and assess the effect of CHM on T2DM with NAFLD patients from the aspects of lipid indices, liver functions, insulin, glycemic indices, and so on. For dichotomous data, a risk ratio (RR) with a 95% confidence interval (CI) was used to measure the results. Continuous variables, such as TG, TC, ALT, AST, FBG, and BMI, were evaluated by weighted mean differences (WMDs) and 95%CI. The heterogeneity of data was investigated by the X 2 test and I 2 test. A fixed effects model was applied if there was homogeneity (p > 0.05, I 2 <50%) (Higgins et al., 2003); otherwise, the random effects model was used. A p-value of less than 0.05 was considered statistically significant. To explore the potential sources of heterogeneity, the factors that contributed to heterogeneity were analyzed through subgroup analysis. In addition, publication bias was assessed by funnel plots and investigated statistically by Egger's test with Stata v14.0.

Sensitivity Analysis
To assess the robustness and reliability of the combined results in meta-analysis, we used sensitivity analysis as an important method. Sensitivity analysis was conducted by excluding individual studies in-turn and re-performing the meta-analysis of the remaining studies. We evaluated whether the results obtained were significantly different from those before the exclusion to ensure the robustness of the results.

Risk of Bias Assessment
The results of the risk of bias assessment are shown in Figure 2 and Figure 3. Of the 18 included studies, 8 studies (Wang, 2010;Guan et al., 2018;Liu, 2019;Wang, 2019;Zou, 2019;Li et al., 2020;Fan et al., 2021;Pu et al., 2021) were classified as low risk of bias because they used the random number table for randomization. Two studies (Zheng et al., 2016;Wang et al., 2018) reported randomization according to the intervention, so they were at a high risk of bias. The other 8 studies (Kang et al., 2016;Xia, 2017;Chen, 2018;Wang, 2018;Li, 2019;Liu, 2020;Wang, 2020;Chen M. L. et al., 2021) claimed to use randomization but did not report details of randomization methodology and were therefore marked as "unclear risk." Except for one study (Fan et al., 2021), the allocation concealment was unclear for the remaining studies. None of the experiments reported blinding of the participants or researchers. Thus, all studies were classified as having a high risk of bias in this aspect. However, these tests were considered to use objective outcome measures. With regards to other biases, none of the studies provided sufficient information that could be used in determining the presence of other significant risks of bias and thus assessed as "unclear risk."  Figure 4). With regards to subgroup analysis, there was no significant difference between different intervention durations, different types of hypoglycemic drugs, and different control treatments (p for interaction = 0.99, 0.95, and 0.67, respectively) ( Table 3, Supplementary Figure S1).

Total Cholesterol
In total, 15 studies ( Figure 4). Subgroup analyses were carried out according to different intervention durations and types of hypoglycemic drugs, and control treatments showed no significant difference in intervention effect between groups (p for interaction = 0.62, 0.22, and 0.51, respectively), and significant heterogeneity was seen ( Table 3, Supplementary Figure S1).

Sensitivity Analysis
After excluding each study, we found no significant changes in the results, and all the results showed good agreement. For further validation, we used STATA v14.0 and performed sensitivity analysis of TG, ALT, FBG, and overall effective rate; the results were considered robust (Supplementary Figure S3).

Summary of the Main Results
For this study, a total of 783 relevant articles were retrieved, and 18 articles were included in the meta-analysis after  Abbreviations: CI, confidence interval; MD, mean difference; RR, risk ratio; a, the risk of bias assessment is mostly "unclear risk" in articles; b, ihere is serious heterogeneity among the studies included in the analysis of this outcome. screening. The main findings of our meta-analysis showed that when compared with WM therapy alone, a combination of CHM and WM therapy was effective for improvement of lipid and glucose metabolism, liver function, insulin resistance, and body mass. The high level of heterogeneity could be attributable to different interventions. The subgroup analysis was performed based on different intervention durations and different types of WM to explain or reduce the degree of associated heterogeneity and obtain a more reliable conclusion. Subgroup analyses showed that intervention duration was primarily responsible for the positive changes in HDL-C and BMI. A few studies did not report adverse effects. The methodological quality of the included studies was poor by the risk assessment of bias. Sensitivity analysis indicated that the results were robust.

Quality of Evidence
In this study, we used GRADEpro to assess the quality of evidence. The assessment showed that the evidence of all the results was of low quality, except that the evidence quality of the overall effective rate was moderate ( Table 4). The decreased certainty of the evidence was mainly attributed to the low methodological quality and the high level of heterogeneity among the studies. Therefore, the results of this study should be applied cautiously to clinical practice, and more high-quality RCTs are needed to evaluate efficacy.

Frequency Distribution Analysis of Chinese Herb Medicines
A total of 70 single CHMs were recorded, sorted by frequency of occurrence from high to low, thereafter listed the CHMs with a frequency of more than 5 times ( For Coptis chinensis Franch., which has the highest frequency of occurrence, pharmacological studies have shown that berberine is the main active ingredient, which has definite effects of reducing hepatocyte lipid accumulation, anti-inflammation, and anti-fibrosis , lowering blood glucose and improving insulin resistance (Chen Y. et al., 2021), and providing typical multi-target and multi-system pharmacological effects on T2DM with NAFLD. Berberine can regulate hepatic metabolism by protecting the intestinal mucosal epithelial barrier, thereby regulating the microenvironment of the intestinal microbiota and changing microbiota-derived metabolites such as shortchain fatty acids and secondary bile acids (Betrapally et al., 2017;Sun et al., 2017;Tian et al., 2019). Berberine can also promote GLP-1 secretion to improve glucose metabolism by up-regulating the related expression of proglucagon and prohormone convertase mRNA (Yu et al., 2010). With regards to Pueraria montana var. thomsonii (Benth.), its main active ingredients are isoflavones, including puerarin and daidzin. Studies have shown that puerarin can repair liver injury and reduce dyslipidemia caused by liver fat deposition by inhibiting IκBα/NF-κB p65 signaling axis activity (Hu et al., 2021). Furthermore, it can improve insulin resistance and oxidative stress by affecting insulin receptor signaling pathways and adjusting the structure of the intestinal microbiome (Zhang H. M. et al., 2019).

Strengths and Limitations for Research
This study was conducted in strict accordance with the method of systematic review, and we interpreted the results cautiously to avoid confusion while ensuring accuracy. We found that combined CHM therapy might improve lipid and glucose metabolism, liver function and insulin resistance, and reduce body weight as well as increase overall effective rate, better than conventional therapy. Therefore, the results may provide new treatment opportunities, new ideas, and new directions for the study of T2DM with NAFLD.
However, this review has a few limitations that warrant discussion. For example, due to the incomplete nature of information provided by most articles as well as the flawed study design, the overall methodological quality of the included studies is poor, which may lead to overestimation of efficacy. Therefore results should be interpreted cautiously. Second, while we performed the subgroup analysis, the source of heterogeneity could not be determined completely. Heterogeneity could have emerged due to the different composition and dose of CHM used in interventions and different dosage forms of CHM (such as decoction, tablets, granules, and pills). In addition, since some studies did not report adverse effects in the aftermath of CHM treatment, the safety associated with CHM remains unclear, and further studies are needed to confirm it. Finally, T2DM with NAFLD as a metabolic disease, lifestyle intervention (such as physical exercise and diet) and pharmacological treatment are the important therapy modalities and have a great impact on values of the biomarkers evaluated. In this study, lifestyle intervention and pharmacological treatment were only briefly statistically analyzed in Table 1, so subsequent studies can further explore the effects of different physical exercises, diets, and pharmacological treatments on the basis of this study, for example, physical exercise, to analyze whether physical exercise is combined, and the specific method and duration of physical exercise. This allows a more detailed and in-depth discussion of T2DM with NAFLD. In the meantime, it is necessary and important to compare whether the effect of two herbs differ on T2DM with NAFLD and T2DM without NAFLD. This is also a significant direction in our subsequent study.

CONCLUSION
In summary, the CHM in combination with WM seems to be more beneficial in T2DM with NAFLD patients in improving lipid and glucose metabolism, liver function, and insulin resistance as well as improving overall efficiency and reducing body weight. Given the poor quality of reports from these studies and uncertain evidence, these findings should be interpreted cautiously. Future RCTs with larger samples and higher quality should be conducted to provide more accurate and complete data to support and validate the clinical efficacy and safety of CHM in the treatment of T2DM with NAFLD.

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.

AUTHOR CONTRIBUTIONS
SP and YL conceived and designed the study. LL, ZX, XYZ, and SY conducted this meta-analysis. XL and HW drafted the manuscript. XGZ and CX revised this article. YL supervised all aspects of this study. All authors contributed to the article and approved the submitted version.