Histological analysis of hypoglycemic agents on liver fibrosis in patients with non-alcoholic fatty liver disease: a systematic review.

To the Editor: Non-alcoholic fatty liver disease (NAFLD) includes a spectrum of disease: simple steatosis, non-alcoholic steatohepatitis (NASH), and the related fibrosis and cirrhosis.[1] The treatment of NAFLD should aim to reverse or prevent liver fibrosis, which is associated with liver-specific morbidity or mortality, rather than just to reduce circulating liver enzyme activities and improve NASH. In addition to progressing to severe liver diseases, NAFLD is often associated with metabolic disorders such as insulin resistance (IR) and type 2 diabetes (T2D),[1] the coexistence of which accelerates the progression of NAFLD and increases the associated cardiovascular risk.[2] Thus, the prevention, slowing, or reversal of liver fibrosis is important in patients with NAFLD and T2D. Hypoglycemic agents that also improve liver fibrosis will be of use in such patients, and numerous studies are being conducted regarding the effects of hypoglycemic agents on the liver fibrosis and exploring the possible molecular targets.[3] The gold-standard method for the evaluation of liver fibrosis is the histological assessment of liver biopsies, but this is an invasive technique with limited use clinically. Instead, a number of non-invasive methods are now considered to be acceptable substitutes, including vibration-controlled transient elastography (VCTE) and magnetic resonance imaging (MRI).[2] To date, with a lack of histological evidence, the optimal choice of hypoglycemic agent for those patients remains subjective. In the present study, we aimed to collate the evidence obtained in prospective randomized controlled trials (RCTs) of the effects of hypoglycemic agents on liver fibrosis in NAFLD (evaluated by histology, VCTE, or MRI), to provide clinicians with more information regarding the options for the management of patients with NAFLD and T2D. We searched the literature published in English between 1990 and 2021 in the PubMed database, after which we screened the articles obtained with inclusion and exclusion criteria. Then, 14 clinical trials were divided into three categories, according to the method used to evaluate liver fibrosis. After that, the relevant information was extracted, and the outcomes and the bias were analyzed. Finally, we deliberately get the conclusion about the effect of each hypoglycemic agent according to the strength of evidence, considering the significance of the used indicators and the bias of the study. The data were presented as: (1) the percentage of participants who benefited from the intervention (showed as average %), and (2) the degree of their improvement if any (showed as an average reduction in histological scores). The cut point of the P-value, the estimation of sample size, and the specific method of statistical analysis varied in different studies. Although several of the RCTs were of the same agent, we could not combine the data for statistical analysis because of differences in the glucometabolic state and the baseline liver fibrosis stage of the participants. The details of the literature review methodology, inclusion and exclusion criteria, data collection, and bias analysis are presented in Supplementary 1, https://links.lww.com/CM9/B642; the systemic review process is presented in Supplementary Figure 1, https://links.lww.com/CM9/B642; and the analysis about the risk of bias of included trials is presented in Supplementary Table 1, https://links.lww.com/CM9/B642. Aithal et al[4] published the results of an RCT of the use of pioglitazone in patients with NAFLD but without? diabetes in 2008. More participants in the pioglitazone group than in the placebo group showed improvements in fibrosis score (29% vs. 20%, P = 0.05). Thus, the authors concluded that 12 months of pioglitazone treatment in patients with NAFLD but without diabetes significantly improves liver fibrosis, according to histologic examination. Cusi et al[5] published the results of an RCT of the use of pioglitazone in patients with NAFLD and T2D or prediabetes in 2016. A larger reduction in fibrosis score was found in the pioglitazone group than in the placebo group (-0.5 vs. 0, P = 0.04), suggesting that pioglitazone improves early liver fibrosis in these patients. In addition, more of the participants in the pioglitazone group showed the resolution of NASH. The overall incidence of adverse events did not differ between the groups in both RCTs, the results of which suggest that pioglitazone improves liver fibrosis in patients with NAFLD, whether or not they have T2D or prediabetes, and is well tolerated. Sanyal et al[6] published the results of an RCT of the use of pioglitazone and vitamin E in 247 patients with biopsy-confirmed NASH but without T2D in 2010. The pioglitazone group showed a higher incidence of improvement in liver fibrosis than the placebo group (44% vs. 31%, P = 0.12) and a trend toward a larger change in fibrosis score (-0.4 vs. -0.1, P = 0.10). Furthermore, more of the participants in the pioglitazone group showed a resolution of NASH with larger changes in NAFLD activity score (NAS). These findings differ from those reported by Aithal et al[4] with respect to the effects of pioglitazone in patients with NAFLD but without? T2D. The reason for this is unclear but might be related to the earlier fibrosis stage, lack of hyperglycemia, and/or the lack of diet or exercise regimens in this study. The results of the Liraglutide Safety and Efficacy in Patients with NASH (LEAN) trial were published in 2016 by Armstrong et al.[7] The liraglutide group showed less deterioration in liver fibrosis than the placebo group (9% vs. 36%, P = 0.04), although there was no difference in the number of participants in which improvements occurred. More participants in the liraglutide group showed resolution of NASH, but without significant difference with respect to the changes in NAS. Thus, the results of this trial suggest that liraglutide might delay the progression of liver fibrosis in patients with NAFLD, whether or not they have T2D, but not reverse it. There are three clinical trials showing separately that pioglitazone in combination with vitamin E, rosiglitazone, and semaglutide do not improve liver fibrosis in NAFLD, but do improve NASH. Other three trials studied metformin and sitagliptin, but failed to support their beneficial effects on NAFLD. The detailed analysis about the study design, the patients' characteristic, and the outcome of each study was presented in Supplementary 2, https://links.lww.com/CM9/B642 and was summarized in Supplementary Table 2, https://links.lww.com/CM9/B642. There are three clinical trials evaluating the NASH or liver fibrosis by VCTE and one by MRI. The hypoglycemic included metformin, empagliflozin, dapagliflozin, and sitagliptin [Supplementary 3 and Supplementary Table 3, https://links.lww.com/CM9/B642]. In four RCTs, the effect of pioglitazone on liver fibrosis was assessed histologically in patients with NASH, including those with or without T2D or prediabetes. Three of these showed that pioglitazone reverses, improves, or delays the progression of liver fibrosis. Although one study showed that a combination of pioglitazone and vitamin E did not improve liver fibrosis, the effect of pioglitazone alone was not studied; therefore, a negative conclusion was not entirely justified. Thus, there is substantial evidence that pioglitazone can improve or reverse liver fibrosis in such patients. The dose administered was 30 mg/day in the participants without diabetes and 45 mg/day in those with T2D. In conclusion, the evidence accumulated to date suggests that pioglitazone is the best hypoglycemic agent for use in patients with NAFLD who do or do not have T2D or prediabetes, in the absence of contraindications such as severe osteoporosis or heart failure. Two histologic studies showed that metformin does not improve liver fibrosis or NASH in patients with NAFLD but without T2D. However, one RCT showed that metformin in combination with diet therapy improves liver fibrosis and steatosis in patients who also had T2D or prediabetes, assessed using VCTE. The differences in the methods used to evaluate NAFLD and in the participants in the two trials may explain their discrepant findings. Specifically, (1) metformin might protect the liver of patients with NAFLD principally by maintaining glucose homeostasis, such that has a more marked effect in patients with more severe abnormalities in glucose metabolism. (2) The baseline liver fibrosis stage might have differed in the three trials, but we could not accurately compare the differences in the liver fibrosis stage that had been assessed histologically or using liver stiffness measurement (LSM). (3) Metformin in combination with dietary therapy might have more substantial effects. (4) The doses of metformin used in the trials and the mean follow-up periods differed. (5) The trials had different sample sizes, and only 23 of the 63 participants underwent liver biopsy at the end of one, which might have led to bias. In summary, metformin might not be effective against liver fibrosis in patients with NAFLD but without T2D. For patients with NAFLD and T2D, metformin was shown to be effective using VCTE, but not histologically. Histologic evidence showed that, irrespective of the presence or absence of T2D, liraglutide delayed the progression of liver fibrosis and improved NASH, but overall, the data suggest that it is inferior to pioglitazone. Neither rosiglitazone nor semaglutide slows the progression of liver fibrosis but both improve NASH. According to VCTE, empagliflozin improves liver fibrosis and NASH, in patients with NAFLD but no diabetes. However, there were no significant changes in other indicators during the trial. Dapagliflozin improves substantial liver fibrosis and NASH in patients with NAFLD and T2D. The follow-up period was the same length as that in the empagliflozin trial, but the participants had more advanced liver fibrosis, which might have permitted greater improvement in LSM. Thus, their effects in patients with or without T2D, and at various stages of liver fibrosis, requires further histological evaluation. Sitagliptin did not improve liver fibrosis or NASH in patients with NAFLD and T2D, neither when assessed histologically nor using MRI. Although the participants had relatively severe fibrosis at baseline, sitagliptin still failed to have beneficial effects. Therefore, there is no evidence that sitagliptin protects the liver of patients with NAFLD and T2D. Although there is high prevalence of both NAFLD and T2D, the sample sizes of the trials performed to date have been limited, mainly for the fact that liver biopsy, as an invasive diagnostic tool, is not accepted by most patients with "benign" liver lesions. Moreover, the risks associated with this procedure and its cost restrict its clinical application. Thus, it is difficult to evaluate the effects of drugs accurately in clinical trials. The resulting small sample sizes, especially with respect to second biopsies at the end of studies, and differences in the evaluation methods used, mean that the reproducibility of the trial results requires verification. Thus, more trials, targeting specific patient groups, for example with different metabolic states and fibrosis stages, are needed to clarify the effects of hypoglycemic agents on liver fibrosis in NAFLD. Admittedly, as non-invasive substitutes for biopsy, the VCTE and MRI are valuable in clinical practice. However, attention should also be paid to the incongruous conclusion emerged in the trials using different evaluation methods, the compliance between which need to be further addressed. Although these limitations render the results of single trials less reliable, we focused on the use of the gold standard method (histology) for the evaluation of liver fibrosis in NAFLD, and have analyzed and summarized all the evidence regarding the effects of hypoglycemic agents, to make our conclusions more reliable. We hope that this review will provide useful evidence for the more reasoned management of patients with NAFLD and T2D. Acknowledgments We thank Mark Cleasby, PhD from Liwen Bianji (Edanz) (www.liwenbianji.cn) for editing the language of a draft of this manuscript. Funding This work was supported by a grant from the 1.3.5 Project for Disciplines of Excellence, West China Hospital, Sichuan University (No. ZYGD 18017). Conflicts of interest None