Meta-Analysis Based on Clinical RCTs: The Effect of Molecular Epimerism on the Safety of Glycyrrhizic Acid

Objective To carry out the meta-analysis on the clinical safety of glycyrrhizic acid and the influencing factors between 18α-glycyrrhizinate (18α-GL) and 18β-glycyrrhizinate (18β-GL). Methods Magnesium isoglycyrrhizinate injection was used as the representative preparation of 18α-GL, and compound glycyrrhizin injection was used as the representative preparation of 18β-GL. The clinical control trial of magnesium isoglycyrrhizinate injection and compound glycyrrhizin injection was searched in a computer, which was published from January 2006 to December 2019 on the databases such as PubMed, China National Knowledge Infrastructure (CNKI), China Science and Technology Journal Database (CSTJ), and Wanfang Medical Network (Wanfang Data). The data associated with adverse drug reactions (ADRs) were extracted. RevMan5.3 was used for statistical analysis. Results Finally, 24 studies were included, and 2757 patients were involved, of which the experimental group was mainly treated with magnesium isoglycyrrhizinate, while the control group was mainly treated with compound glycyrrhizin. The results showed that the occurrence of ADRs was significantly lower in the experimental group than that in the control group, and the difference between two groups was statistically significant (RR = 0.26, 95% CI = (0.18, 0.38), P < 0.00001). There was no heterogeneity among the studies (I2 = 0%, P=1.00). Conclusion Compared with 18β-GL, 18α-GL had a lower incidence of adverse reactions and better clinical safety.


Introduction
Glycyrrhizin (GL) is one of the active ingredients in licorice, and it has been widely used in clinical diseases of abnormal liver function. Based on epimerism, GL is classified into 18α-glycyrrhizinate (18α-GL) and 18βglycyrrhizinate (18β-GL) [1]. e difference between the efficacy and safety of these two epimers has become the focus of clinical attention. Currently, the representative 18α-GL drug on the market is magnesium isoglycyrrhizinate (MI), and the representative 18β-GL drug is compound glycyrrhizin (CG). Existing studies with small samples have shown that MI and CG have differing conclusions of the adverse drug reactions (ADRs). e studies of Ye et al. [2] and Mao et al. [3] showed that there were no differences in ADRs between these two drugs. Many other controlled studies only reported the number of ADR cases from these two drugs and did not perform statistical analyses. Animal studies have shown that the two isomers are different in cortisol competition for 11βhydroxysteroid dehydrogenase [4]; conversely, 11βhydroxysteroid dehydrogenase was directly associated with pseudo-aldosteronism-related symptoms. However, whether the difference in the structure of these two preparations has led to the occurrence of ADR difference has not been reported clinically. erefore, in this study, meta-analysis was used to explore their relationship of ADRs and 11β-hydroxysteroid dehydrogenase in MI and CG in order to provide reference for clinical safe medication and new production preparations.  [5]. Lowquality or missed information literature studies were removed. Two evaluators evaluated the literature independently and cross-checked. When there were divergences, the consensus was reached through discussion.

Statistical Analysis.
A meta-analysis was carried out by RevMan5.3 software. e publication bias was tested by Stata software. e development of ADRs was evaluated using the relative risk (RR) and 95% confidence interval (CI). Heterogeneity was examined using a homogeneity test (chisquare test). I 2 ≤50% indicated that the heterogeneity might not be important or that there was moderate heterogeneity. e fixed-effect model was used for the meta-analysis. I 2 >50% indicated that heterogeneity was evident. If interference could not be excluded, the random-effect model was used for analysis.

Literature Retrieval Results.
Databases were searched comprehensively, and then, the retrieval results were crosschecked ( Figure 1). At the end, 55 articles were chosen for the meta-analysis. Preliminary screening was performed on the titles and abstracts, and two non-RCTs [6,7] and two articles with nonmatched drugs were removed [8,9]. After reading the entire articles, one article with repeated data [10], one retrospective study [11], 20 articles that did not report ADR data [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31], three articles that only reported the number of ADRs but did not report symptoms [32][33][34], and two articles that had inconsistent ADR numbers and symptoms were excluded [35,36]. Finally, 24 articles that conformed to the inclusion criteria were included in this study [3,. ese 24 articles involved a total of 2,757 patients with 1,520 cases in the MI group and 1,237 cases in the CG group. e MI doses in these studies ranged from 50 to 200 mg, and the CG doses ranged from 60 to 200 mg. e MI preparations in these studies were from original manufacturers. Nine studies had labeled the origins of the CG injections, and five studies did not indicate the origin; therefore, a statistical analysis could not be performed (Table 1).

Evaluation of Literature Quality and the Heterogeneity Test.
ree [37,38,57] of the 24 articles described the method of randomization. None of the articles mentioned allocation concealment. One article used a double-blinded method. One article was published in a Chinese-English journal [3], and the other articles were all published in Chinese journals. e determination of ADRs was based on clinical symptoms and laboratory biochemical indicators; therefore, it was objective. Data in all articles were complete, and there was no loss to follow-up. Because most of the articles were published in Chinese journals, publication bias was evaluated using a funnel plot. e results are shown in Figure 2. e funnel plot was tested by Begg's test (Figure 3). e results showed that z � 0.64 (continuity corrected), pr > |Z | � 0.519 (continuity corrected), and 0.519>0.05, indicating that the results were not statistically significant. All of them are uniformly distributed in Figure 4. erefore, no publication bias was found by Begg's test. e results of the literature quality evaluation are shown in Figure 5.

3.3.
e Conditions of ADR Development. e specific conditions of the number and symptoms of ADRs in the articles are shown in Table 2.
e overall analytic results of the 24 studies are shown in Figure 6. ey all reported ADR development in both the MI  Evidence-Based Complementary and Alternative Medicine group and the CG group (n � 1520/1237). To assess ADR development, the meta-analysis was performed using the fixed-effect model because the heterogeneity among all studies might not be important (I 2 ≤ 50%). e results showed that the RR of ADR development in the MI group was significantly lower than that in the CG group (RR � 0.26, 95% CI � (0.18, 0.38), P < 0.00001).

Discussion
e mechanism of ADR development is very complicated. Molecular heterogenicity is one of the reasons for the difference ADR occurrence in glycyrrhizin preparations. MI is a single-ingredient preparation, in which the non-8α-GL content is only 1%. CG, as a representative drug of secondgeneration glycyrrhizic acid, mainly contains 18β-GL and a small amount of 18α-GL, and the proportion of 18β-GL ranged between 95% and 99%. is study compared the highpurity 18α-GL preparation of MI with the clinically extensively used CG preparation that mainly contained 18β-GL.
rough a comprehensive analysis of the conditions of ADR development in many studies, the results showed that these two preparations had significant differences in the ADR.

Clinical Presentations of ADRs.
e statistical analysis of this study showed that the ADRs of GL primarily manifested as lower limb edema, facial edema, increased blood pressure, palpitation, and dizziness and headache. A total of 112 cases accounted for 78.87% of the 142 ADR cases. ese frequent ADRs were all associated with pharmacological and toxicological effects of GL and were pseudo-aldosteronism reactions. Pseudo-aldosteronism reactions accounted for 67.86% of ADRs in the MI group and 81.58% of ADRs in the CG group. With MI as the 18α-GL preparation, pseudoaldosteronism was still the main type of ADR reported; however, its incidence was lower than that from the CG presentation, which mainly contained 18β-GL. ese results were consistent with the study of Li et al. [60].

e Mechanism of Differences.
e current view considers that the pseudo-aldosteronism reaction of GL is associated with the drug itself or its metabolic product [61]. Different structures of GL also have metabolic products with different structures in the body. e metabolic product of 18α-GL is 18α-glycyrrhetinic acid, and the metabolic product of 18β-GL is 18β-glycyrrhetinic acid. Existing preclinical studies have shown that the incidence of pseudo-aldosteronism reactions produced by 18α-glycyrrhetinic acid was lower than that produced by 18β-glycyrrhetinic acid. A possible reason is that 18α-glycyrrhetinic acid selectively inhibits type 1 11β-hydroxysteroid dehydrogenase, whereas 18βglycyrrhetinic acid has inhibitory effects on both type 1 and type 2 11β-hydroxysteroid dehydrogenase [4]. ese results suggest that differences in the clinical ADRs between these two preparations might also be caused by differences in their molecular conformation. e specific clinical, pharmacological, and toxicological mechanisms and the strength of action require further experimental studies.

4.3.
e Influence of the Experimental Design Program. e meta-analysis results were influenced by the quality and number of patients in the included articles. None of  Evidence-Based Complementary and Alternative Medicine the articles included in this study described allocation concealment, which might influence the subjective judgments of doctors and patients, particularly the subjective symptoms of ADRs such as dizziness, headache, and palpitation.
Data Availability e datasets used and/or analyzed during the present study are available from the corresponding author upon reasonable request.

Conflicts of Interest
e authors declare that there are no conflicts of interest.

Authors' Contributions
Hongdou Chen and Menglei Wang participated in the design of the study. Fangfang Zheng and Menglei Wan were involved in writing of the paper and statistical analysis. Other authors participated in the analysis of the literature. All authors read and approved the final manuscript. Hongdou Chen and Fangfang Zheng contributed equally to this work.  Figure 6: Results of the meta-analysis of ADR development. 6 Evidence-Based Complementary and Alternative Medicine