The Evaluation of Effective Drugs for the Treatment of Non-Alcoholic Fatty Liver Disease: A Systematic Review and Network Meta-Analysis

Purpose: Non-alcoholic fatty liver disease (NAFLD) and steatohepatitis are two forms of fatty liver disease with benign and malignant nature, respectively. These two conditions can cause an increased risk of liver cirrhosis and hepatocellular carcinoma. Given the importance and high prevalence of NAFLD, it is necessary to investigate the results of different studies in related scope to provide a clarity guarantee of effectiveness. Therefore, this systematic review and meta-analysis aim to study the efficacy of various medications used in the treatment of NAFLD. Methods: A systematic search of medical databases identified 1963 articles. After exclusion of duplicated articles and those which did not meet our inclusion criteria, eta-analysis was performed on 84 articles. Serum levels of alanine aminotransferase (ALT), aspartate amino transferase (AST) were set as primary outcomes and body mass index (BMI), hepatic steatosis, and NAFLD activity score (NAS) were determined as secondary outcomes. Results: Based on the P-score of the therapeutic effects on the non-alcoholic steatohepatitis (NASH), we observed the highest efficacy for atorvastatin, tryptophan, orlistat, omega-3 and obeticholic acid for reduction of ALT, AST, BMI, steatosis and NAS respectively. Conclusion: This meta-analysis showed that atorvastatin. life-style modification, weight loss, and BMI reduction had a remarkable effect on NAFLD-patients by decreasing aminotransferases.


Introduction
The non-alcoholic fatty liver disease (NAFLD) covers a broad spectrum of liver disease ranging from pure fatty liver to non-alcoholic steatohepatitis (NASH) and finally cirrhosis. It occurs in patients without a significant alcohol intake. 1 Although the pathogenesis of NAFLD is not wellunderstood, some studies support the "two-hit theory. " While the "first hit" involves insulin resistance and results in the collection of fat in the liver, the "second hit" consists of oxidative stress resulting in lipid peroxidation, hepatocellular degeneration, cell death, hepatic stellate cell (HSC) activation, and fibrogenesis. 2,3 End-stage liver disease and hepatocellular carcinoma secondary to NAFLD are the second leading cause of liver transplantation in the United States. This is an alarming trend that NAFLD is replacing hepatitis C as the most common indication for liver transplantation during the future decades. [4][5][6] As a result of obesity epidemics, the prevalence of NAFLD, as a common liver disease, has dramatically increased. Estimations reveal an incidence of up to 20%-30% for NAFLD in Western and 5-18% in Asian countries. 7 Due to the inappropriate diet and sedentary lifestyle, the increased prevalence of NAFLD throughout the world is considered a common clinical concern over time. 8 Weight loss, achieved either by a hypocaloric diet alone or in combination with increased physical activity, reduces hepatic steatosis a combination of a hypocaloric diet (daily caloric reduction of 500-1000 kcal), and moderate-intensity exercise is presumably to provide the best likelihood of improving steatosis. However, a more significant weight loss (7%-10%) is required to improve the majority of the histopathological features of NASH. 9 Although NAFLD is associated with components of metabolic syndrome, the presence of an increasing number of metabolic disorders, including insulin

Inclusion, exclusion and eligibility criteria
We included publications that: had studied the various medical interventions for the treatment of NAFLD, were published in English language, were published to January 2019 and also paperers which were presented in different conferences.
We excluded: articles that were published in languages other than English, animal studies, research studies with

Select studies
The selected articles were processed in 3 states by the subject experts: The titles of all review and incompatible articles were excluded. Abstracts and full texts of the articles were studied. The selected studies were evaluated by the two experts for recognizing the risk of bias using Joanna Briggs Institute tool for critically appraising the clinical trials; the controversies between them were resolved by a third expert. 17 The critical appraisal form of the select studies is shown in Table S1.  The extracted information was summarized in the data extraction table (Table 2). Extracted data were included: first author, publication year, country, sample size, and number of participants over. Finally, the Endnote X5 software was used to manage the references.

Publication bias
The funnel plot (Figure 1) illustrates the publication bias of the studies that have been entered into the network meta-analysis. This symmetry shows that there has been evidence of publication bias in this meta-analysis.

Primary and secondary outcomes in the study
The primary outcome measure in this study was the mean changes in ALT and AST levels. Changes in BMI, NAS, and steatosis were considered as secondary outcomes.
The evaluation of effective drugs for the treatment of non-alcoholic fatty liver disease Advanced Pharmaceutical Bulletin, 2020, Volume 10, Issue 4 545 Y = Yes, N = No, U = Unclear; JBI critical appraisal checklist for randomized controlled trials: Q1 = Was true randomization used for assignment of participants to treatment groups?; Q2 = Was allocation to treatment groups concealed?; Q3 = Were treatment groups similar at baseline?; Q4 = Were participants blind to treatment assignment?; Q5 = Were those delivering treatment blind to treatment assignment?; Q6 = Were outcome assessors blind to treatment assignment?; Q7 = Were treatment groups treated identically other than the intervention of interest?; Q8 = Was follow-up complete, and if not, were strategies to address incomplete follow-up utilized?; Q9 = Were participants analyzed in the groups to which they were randomized?; Q10 = Were outcomes measured in the same way for treatment groups?; Q11 = Were outcomes measured in a reliable way?; Q12 = Was appropriate statistical analysis used?; Q13 = Was the trial design appropriate, and any deviations from the standard RCT design (individual randomization, parallel groups) accounted for in the conduct and analysis of the trial?

Statistical methods
Investigation of stability and similarity assumptions in intervention network: To assess the similarity hypothesis, the baseline characteristics of the participants were assessed, and trials with no similarity hypothesis were included in the network meta-analysis.
The heterogeneity between studies was analyzed via the Cochran Q test and I-square statistic.
Rankings of treatments P-scores were used to rank for each treatment. P-scores are calculated on the basis of defect estimation and standard error estimates in the grid. The wake score is between zero and one, and the closer the wake of a cure to a cure, the better the cure. We have utilized the GRADE approach to calculate the level of evidence; the level of evidence has been categorized to very low, low, moderate and high taking bias, inconsistency, indirectness and imprecision into account.

Results and Discussion
Search results and basic characteristics of studies entered into a meta-analysis of networks. The data extraction is represented in Table S1.
After excluding duplicate papers, and papers that did not meet the inclusion criteria we entered 84 to the final meta-analysis. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) diagram ( Figure  2) illustrates the number of the articles screened for final number of analyzed articles.

Network meta-analysis results for primary outcomes Comparison of the effects of treatments on ALT
After selecting studies, 84 studies were included in the meta-analysis ALT outcome. Heterogeneity was not significant between studies (Q-value = 52.83, P value = 0.002, I-square = 55.9). Figure 3 shows the network of interventions introduced into the network meta-analysis for ALT outcome.
Based on the ratings given the therapeutic effect of ALT on the follow-up score, it is evident that treatment with Atorvastatin had the most significant impact on ALT compared to other investigated drugs. Table 3 shows the treatment ratings by use of P-score for ALT.
The forest plot has also combined preliminary results with studies performed on the network meta-analysis in comparison to the placebo treatment ( Figure 4). The level of evidence for the impact of atorvastatin vs. placebo on ALT levels has been calculated as low using GRADE approach.   Comparison of treatment effects on AST After selecting studies, 71 cases were included in network meta-analysis for AST. Heterogenicity was significant between studies (Q-value = 261.46, P value <0.001, I-square = 86.4). Figure 5 represents the network of interventions introduced into the network meta-analysis for the AST outcome. Based on the ratings given the therapeutic effect on AST, it was found that treatment with tryptophan had the best change over AST in comparison to the other investigated drugs. Table 4 shows the treatment ratings by use of the P-score. The level of evidence for impact of tryptophan vs. placebo on AST levels has been calculated as low using GRADE approach.

Network meta-analysis results for secondary outcomes Comparison of the effects of the treatment on BMI
After selecting the studies, 48 cases were included in the meta-analysis. Heterogeneity was not significant between studies (Q-value = 15.98, P value = 0.31, I-square = 0.00).
The network of interventions introduced into the network meta-analysis for BMI outcomes is demonstrated in Figure 6.
Based on the given ratings related to therapeutic effects on BMI, it was found that treatment with Orlistat had the best efficiency in BMI over the other drugs studied. Table  5 shows the treatment ratings by use of the P-score. The level of evidence for the impact of orlistat vs. placebo on BMI has been calculated as low using GRADE approach.

Comparison of the effects of the treatment on steatosis
After selecting the studies, 14 studies were included in the meta-analysis. Heterogenicity was not significant between studies (Q-value = 0.43, P value = 0.51, I-square = 0.00). Table 4. Treatment ratings by use of the P-score on AST

Rank
Drug P-score (fixed) P-score (random) Rank Drug P-score (fixed) P-score (random) Based on the given ratings associated with the therapeutic effects of steatosis, it was concluded that treatment with Omega-3 had the best efficiency over steatosis in comparison to the other investigated drugs. Table 6 shows the ratings of treatments using the P-score. The forest plot has also been shown to combine preliminary results with studies performed on the meta-analysis of networks compared with placebo treatment (Figure 7). The level of evidence for impact of Omega-3 vs. placebo on steatosis has been calculated as low using GRADE approach.

Comparison of the treatment effects on the NAS
After selecting studies, 15 cases were included in the meta-analysis. Heterogeneity was not significant between studies (Q-value = 1.15, P value = 0.28, I-square = 12.8). The network of interventions introduced into the network meta-analysis for the outcome of NAS are shown in Figure 8.  Table 5. Treatment ratings by use of the P-score

Rank
Drug P-score (fixed) P-score (random) Rank Drug P-score (fixed) P-score(random) Based on the given ratings pertinent to the therapeutic effects on NAS, it was observed that treatment with betaine had the best efficiency over NAS compared to the other investigated drugs. The forest plot has also been shown to combine preliminary results with studies performed on the meta-analysis of networks compared with placebo treatment (Figure 9). The level of evidence for impact of betaine vs. placebo on NAS has been calculated as low using GRADE approach (Table 7).
Based on our search of medical databases and although there is a meta-analysis conducted on the topic of effective medication on NAFLD conducted by Sridharan et al our research is the first network meta-analysis that has been conducted on this topic among adult population so far. Additionally, our systematic review and network metaanalysis is different than the previous work as this article has also included children's studies that could potentially have confounding factors as NAFLD in children ma arise from genetic origins. 103 Another major difference that is noted in the conduct of the present stud and the research Table 6. Ratings of treatments using the P-score for steatosis

Rank
Intervention P-score (fixed) P-score (random)   Sridharan and colleagues has put overall response rate to medication and AST, ALT, NAS, BMI, lipid profiles were set as secondary outcome.
We have implicated the primary impact of different medical therapies and their effect on liver transaminases, as our primary outcome of NAFLD. We showed that atorvastatin, colesevelam and ezetimibe had the best outcome compared to the SOC on ALT, and tryptophan and melatonine had the best impact on lowering AST while in the study of Sridharan and colleagues, the best medication to reduce ALT is reported to be with vitamin D followed by gemfibrozil and ipragliflozin. 104 Based on the given ratings pertinent to the therapeutic effects on NAS, it was observed that treatment with Betaine had the best efficiency over NAS compared to the other investigated drugs.
Sridharan et al have determined the better efficacy for elafibranor. 104 While Sridharan et al had reported the best outcome from telmisartan, in the current investigation, Telmisartan had lower efficacy than SOC on. 104 High-quality clinical trials are needed to prove the efficacy of atorvastatin in NAFLD.
As we expected, life-style modification shown to be effective in reducing aminotransferases. Alongside the use of medicinal compound. Of these, orlistat was the first effective drug followed by liraglutide. Given to the fact that treatment of NAFLD can be associated with complications including NAS, we illustrated that betaine, metformin, omega 3, liraglutide and colesevelam were most effective drugs in reduction of NAS. Of these, efficacy of betaine and liraglutide had similarities to the results of previous meta-analyses. Omega 3 were shown to be best therapeutic choice in hepatic steatosis.
In previous meta-analysis, pentoxifylline was one of the drugs that had a remarkable effect on reducing liver fibrosis. Compared to our findings, although pentoxifylline were a weaker agent in reducing NAS, it worked well. 104 According to the guidelines of American Association for Liver Diseases, vitamin E administered at a daily dose of 800 IU/day improves liver histology in nondiabetic adults with biopsy-proven NASH and therefore may be considered for this patient population. 9 Based on the results of our meta-analysis, vitamin E had no significant impact on the reduction of aminotransferases compared to SOC in the context of decline hepatic steatosis, nevertheless, has not been shown to be effective in reducing NAS.
We observed no obvious effect for polyunsaturated fatty acids (PUFA) for both primary and secondary outcomes, therefore, its administration in NAFLD is not recommended.
Ultimately for NAS, our meta-analysis revealed the best response from Betaine. It is also recommended to include the betaine in larger clinical trials to further study its antisteatotic properties.
We encountered some limitation in performing current systematic review, heterogenicity of studies and the lack of studies that had investigated the combined medications are some of them.
It is suggested to do other meta-analyses to update previous studies, which had reduced the amount of drug diversity. A general consensus about some drugs and patient's treatment process is needed in prospective studies as well.

Conclusion
All in all, our study shows a higher efficacy for the reduction in liver transaminases for atorvastatin. Although lifestyle modification, weight loss and BMI reduction are all effective in improving the primary and secondary outcomes, our network meta-analysis showed the greater efficacy for tryptophan, orlistat, omega-3 and betaine to improve AST, BMI, steatosis and NAS. Considering these findings, we recommend randomized clinical trials to examine these medical modalities with placebo and each other.

Ethical Issues
This systematic review and meta-analysis has been evaluated and approved by regional ethics at Tabriz university of Medical Sciences, Faculty of Medicine. Additionally, all ethical considerations were meet in all levels of systematic search according to systematic review and meta-analysis

Conflict of Interests
None to declare.