Efficacy and Safety of Mulberry Twig Alkaloids Tablet for the Treatment of Type 2 Diabetes: A Multicenter, Randomized, Double-Blind, Double-Dummy, and Parallel Controlled Clinical Trial

According to the International Diabetes Federation’s IDF Diabetes Atlas (9th edition), an estimated 463 million people are currently living with diabetes globally, and ∼90% of these individuals have type 2 diabetes (T2D). The number of people with diabetes is projected to reach 578 million by 2030 (1). The standardized incidence of diabetes in the Chinese adult population is 10.9–11.2%; however, only 32.2–49.0% of these individuals receive treatment, and approximately one-half of all treated patients do not achieve satisfactory glycemic control (2,3). Although new hypoglycemic agents have recently been introduced to the market (4–9), these drugs still do not meet the clinical demand for glycemic control.

The use of herbal medicines for the prevention and treatment of T2D is well documented in China and other countries. Herbal medicines often have a wide range of pharmacological effects. There has been an increasing interest in the use of herbal medicines for the treatment of diabetes (10,11). Previous studies have demonstrated the hypoglycemic effect of some herbs derived from parts of the mulberry tree (12,13). The main components of mulberry twigs include alkaloids, flavonoids, polysaccharides, coumarin, amino acids, and organic acids (12,14). Studies have shown that mulberry twig alkaloids (Sangzhi alkaloids [SZ-A]) are the active component of alkaloids extracted and isolated from the Chinese herbal medicine “mulberry twig” and are mainly composed of 1-deoxyrijolymycin (1-DNJ), fagomine (FA), 1,4-dideoxy-1,4-imino-d-arabitol (DAB), and other polyhydroxyalkaloids. The total content of alkaloids is not less than 50% in SZ-A. Among the active components of alkaloids, the content of 1-DNJ, FA, and DAB account for >90%, and 1-DNJ is the predominant alkaloid (Supplementary Material 1). SZ-A has been shown to reduce fasting and nonfasting glucose levels and prolong peak glucose concentrations after sucrose or starch load in alloxan-induced diabetic mice and rats (15). In addition, animal experiments have revealed that SZ-A improves insulin resistance, increases basal insulin levels, and enhances glucose-stimulated insulin secretion (16,17). Pharmacological studies have suggested that the beneficial effects of SZ-A in the treatment of diabetes may be mediated through several therapeutic targets. Of these, intestinal α-glucosidase is the most well-characterized target of SZ-A.

People in East Asia usually rely on starchy foods as their main source of calories; therefore, α-glucosidase inhibitors are often used as the first-line drug for the treatment of patients with diabetes in this region (18–20). However, these drugs are associated with a high incidence of gastrointestinal disorders (GDs), such as abdominal distension, abdominal pain, and diarrhea, because of low selectivity to α-glucosidase (5,21,22). In vitro experiments have shown that SZ-A imparts a significant inhibitory effect on sucrase and maltase, which is equal to or slightly stronger than that of acarbose. However, the inhibitory effect of SZ-A on amylase is much weaker than acarbose. This suggests that SZ-A has stronger disaccharidase selectivity and may help in reducing GDs, such as flatulence (23–26).

A placebo-controlled, randomized, double-blind, and multicenter trial showed that change in glycosylated hemoglobin (HbA_1c) was significantly decreased compared with baseline after treatment for 16 weeks, and the effect was much better than that of placebo (data not published). In this study, we evaluated the efficacy and safety of SZ-A in the treatment of patients with T2D, using acarbose as positive control, to provide evidence for its clinical application. To our knowledge, this is the largest randomized controlled trial of SZ-A in patients with T2D in the world. In addition, it is the largest randomized controlled trial of natural botanical drugs for diabetes that have been marketed globally.

This multicenter, randomized, double-blind, double-dummy, and parallel controlled clinical trial was conducted across 23 institutions in China. The lead-in period was 4 weeks, and the treatment period was 24 weeks. It used a positive-controlled (acarbose), multicenter, phase III noninferiority clinical research design, and each center competed for the entry.

Patients who qualified on the basis of 1999 World Health Organization diagnostic criteria for T2D and who had not received antidiabetic medical therapy before enrollment or who had received antidiabetic medical therapy for no more than 3 months at any time in the past were eligible for participation. The inclusion criteria were 1) age range of 18–70 years, 2) provision of written informed consent, 3) HbA_1c level of 7.0–10.0% (53–86 mmol/mol), 4) fasting blood glucose (FBG) <13 mmol/L (<234 mg/dL), and 5) BMI of 19–30 kg/m².

The main exclusion criteria were 1) >2.5 mmol/L (>45 mg/dL) variation in FBG level between the first follow-up and second follow-up; 2) allergy or intolerance to α-glucosidase inhibitors; 3) history of severe diabetic complications (the proliferative stage of diabetic retinopathy, diabetic nephropathy stage V, diabetic ketoacidosis, diabetic hyperosmolar coma, and diabetic lactic acidosis); 4) combined therapy with drugs that affect glucose metabolism, such as antidiabetic herbal medicine or glucocorticoids; 5) hyperlipidemia with a history of irregular intake of lipid-lowering drugs; 6) chronic gastrointestinal dysfunction, obvious digestive and absorptive disorders, or endocrine diseases such as hyperthyroidism, hypercortisolism, and acromegaly; 7) medical conditions that may be aggravated by flatulence (e.g., severe hernia, ileus, postoperative intestinal surgery, intestinal ulcers); 8) serious heart disease, myocardial infarction, unstable angina pectoris, or chronic cardiac insufficiency (New York Heart Association class III and IV); poorly controlled blood pressure (BP) (systolic BP ≥160 mmHg or diastolic BP ≥100 mmHg); 9) impaired liver function (serum ALT or AST >2× the upper limit of normal [ULN] reference range or total bilirubin >2× ULN) or kidney function (serum creatinine >1× ULN or creatinine clearance <60 mL/min); 10) serious medical conditions likely causing terminal illness during the treatment and follow-up periods; 11) mental or neurological disorders and inability to express wishes correctly; 12) participation in clinical trials of other drugs or medical devices in the immediately preceding 3 months; 13) substantial alcohol consumption (episodic drinking of >5 units of alcohol for men and >4 units for women at once [1 unit = 10 g of pure ethanol]) or drug abuse or addiction; 14) pregnancy; and 15) unlikelihood of completing the expected course of treatment and follow-up (on the basis of the expert team’s opinion).

The study (version number HY-SZIIIa-Pro-131020-01) protocol was approved by the ethics review committee of the Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, and other participating institutions. Written informed consent was obtained from all participants before their enrollment. The trial was performed in accordance with the principles of the Declaration of Helsinki.

The data analysis system for electronic data capture and clinical trial central randomization was used for randomization and dispensing of drugs. Electronic emergency letters were set up, and each center competed for enrollment.

A double-blind, double-dummy design was adopted to ensure that the two groups of drugs were consistent in terms of usage, dosage, and appearance. Because there are significant differences in appearance, smell, flavor, and quantity between the trial drug SZ-A and the control drug acarbose, in consideration of the scientificity of clinical study, the placebo of SZ-A and acarbose tablets were prepared to ensure the similarity of treatment between SZ-A and acarbose treatment groups in terms of usage, dosage, and appearance. To maintain similarity in taste and smell between SZ-A tablets and placebo, 1/20th of the dosage of SZ-A tablet was added in the preparation process of the placebo, which is a clinically ineffective dose. Acarbose placebos were prepared according to the shape, color, and weight of acarbose. The preparation of placebos absolutely satisfies the requirement of double blinding in a clinical trial.

All participants received one SZ-A placebo tablet and acarbose placebo tablet three times a day for a 4-week lead-in period. Subsequently, participants in the SZ-A group were provided an SZ-A tablet (50 mg) and acarbose placebo tablet, and those in the acarbose group were provided an acarbose tablet (50 mg) and SZ-A placebo tablet for 4 weeks. In the subsequent treatment period, the doses of SZ-A tablets and SZ-A placebo tablets were doubled. All the tablets were administered orally with the first bite of food (Supplementary Materials 2 and 3 and Supplementary Fig. 1).

SZ-A tablets (50 mg/tablet; active ingredient SZ-A, mainly 1-DNJ) were produced by Beijing Wehand-Bio Pharmaceutical Co., Ltd. SZ-A placebo tablets (2.5 mg/tablet; active ingredient SZ-A, mainly 1-DNJ; blank excipient added) and acarbose placebo tablets were both manufactured by the same company. Acarbose tablets (50 mg/tablet) were produced by Bayer Healthcare Pharmaceutical Inc. (Berlin, Germany). To render the tablets indistinguishable in terms of appearance, weight, and taste, the SZ-A placebo tablets contained a 5% dose of SZ-A, while the acarbose placebo tablet contained the corresponding blank excipient to mimic the acarbose tablet.

The primary efficacy end point was the change in HbA_1c compared with baseline after treatment for 24 weeks (Supplementary Material 4). The secondary efficacy end point included the proportion of patients with HbA_1c <7% (<53 mmol/mol) after 24 weeks of treatment as well as changes in FBG, 1-h postprandial blood glucose (1h-PBG), 2-h PBG (2h-PBG), area under the curve of the PBG (AUC_{0–2 h}) (Supplementary Material 5), body weight, and BMI compared with baseline. Safety indicators included the incidence of adverse events (AEs), severe AEs (SAEs), treatment-related AEs (TAEs), and GDs (Supplementary Material 6).

A reduction in HbA_1c level after treatment was the primary efficacy end point; noninferiority design was adopted with the unilateral test α = 0.025, β = 0.2 (power = 80%). On the basis of the phase II clinical study of SZ-A (the relevant results have not been published yet), the expected variation in HbA_1c was –0.82% (–9.0 mmol/mol) in the experimental group and –0.84% (–9.2 mmol/mol) in the acarbose group; the expected common SD was 0.88%, and the noninferior standard δ was 0.3% (3 mmol/mol). According to the 2007 “drug registration management approach” of traditional Chinese medicine, the minimum sample size required for a phase III clinical trial group is 300. Accounting for a dropout rate of 20%, with the ratio of 3:2 between the treatment group and acarbose group, the total number of patients was finally determined to be 600, including 360 in the SZ-A group and 240 in the acarbose group.

SAS 9.3 software was used for all statistical analyses. The noninferiority test was represented by a bilateral 95% CI of the difference between the groups. Bilateral tests were used for all statistical tests, and two-sided P < 0.05 was considered indicative of statistical significance. Descriptive statistics were generated for demographic data and baseline analysis. Continuous variables are presented as the mean, SD, and 95% CI. Categorical variables are presented as frequency and percentage. Inferential statistical results are listed as descriptive results.

The primary efficacy end point was analyzed by covariance, with groups as fixed effects. The differences between the two groups and the bilateral 95% CIs were calculated. The upper limit of the bilateral 95% CI and the preset noninferior standard of 0.3% (3 mmol/mol) were compared to determine whether the experimental group was noninferior to the acarbose group. The primary end point was analyzed using full analysis set (FAS) and per-protocol set (PPS) as well as the intention-to-treat population.

Diachronic analysis was performed for the proportion of patients with HbA_1c <7% (<53 mmol/mol), and the between-group difference was assessed using the χ² test. Changes in HbA_1c, FBG, 1h-PBG, 2h-PBG, AUC_{0–2 h}, body weight, and BMI from baseline were analyzed over time, and the between-group differences were assessed using the t test. Paired t test was used to compare the pre- and posttreatment levels of various parameters. Secondary efficacy end points were analyzed by FAS.

AEs are coded according to the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use International Medical Dictionary for Regulatory Activities. The incidence rates of AEs, SAEs, TAEs, and GDs were calculated, and the between-group differences were assessed using the χ² test. Safety indicators were analyzed by security data set.

A total of 600 patients were recruited, and 517 completed the study. The dropout rates of the SZ-A and acarbose groups were 13.89% and 13.75%, respectively. Approximately 50 patients in the SZ-A group and 33 in the acarbose group dropped out between 25 March 2014 and 15 July 2015 (Supplementary Fig. 2). HbA_1c and other parameters, such as age, sex, course of diabetes, FBG, 1h-PBG, 2h-PBG, weight, and BMI, did not differ between the two groups. The baseline characteristics of the study population are summarized in Table 1.

In the FAS analysis, the mean change in HbA_1c level at 24 weeks (24–0 weeks) in the SZ-A and acarbose groups was –0.93% (95% CI –1.03 to –0.83) (–10.2 mmol/mol [–11.3 to –9.1]) and –0.87% (–0.99 to –0.76) (–9.5 mmol/mol [–10.8 to –8.3]), respectively; covariance analysis revealed no significant between-group difference (P = 0.368). The mean difference between the SZ-A and acarbose groups (least squares mean difference) was –0.05% (95% CI –0.18 to 0.07) (–0.5 mmol/mol [–2.0 to 0.8]). According to the noninferior standard of 0.3% (3 mmol/mol), the SZ-A group was not inferior to the acarbose group. PPS and intention-to-treat analyses generated similar results (Fig. 1A and Supplementary Material 7).

There was no significant between-group difference with respect to HbA_1c levels at baseline. At the 8-, 16-, and 24-week follow-up evaluations, HbA_1c levels were all significantly lower than baseline (P < 0.001) in both the SZ-A and the acarbose groups; however, there was no significant between-group difference (P > 0.05) (Fig. 1B).

Taking HbA_1c <7% (<53 mmol/mol) as the target of treatment, the number of patients who achieved the target showed a gradual increase with treatment. At 24 weeks, 153 patients in the SZ-A group had achieved the target with a rate of reaching the standard of 47.7%, and 100 patients achieved the target in the acarbose group with a rate of reaching the standard of 45.0%. There was no significant between-group difference (P = 0.548) (Fig. 1C).

Changes in FBG, 1h-PBG, 2h-PBG, and AUC_{0–2 h} from baseline showed no significant between-group differences (Table 2). After 8, 16, and 24 weeks of treatment, FBG, 1h-PBG, 2h-PBG, and AUC_{0–2 h} in both groups showed a significant decrease from the respective baseline levels (P < 0.001 for all). However, there were no significant between-group differences at the indicated time points with the exception of 2h-PBG at 8 weeks and 1h-PBG at 16 weeks (Fig. 1D–G).

After 24 weeks of treatment, changes in body weight in the SZ-A group (–0.68 ± 2.23 kg [95% CI –0.94 to –0.42]) did not significantly differ from those in the acarbose group (–0.76 ± 2.46 kg [–1.11 to –0.42]; P = 0.698). BMI in the SZ-A group changed by –0.25 ± 0.81 kg/m² (–0.34 to –0.16), while that in the acarbose group changed by –0.28 ± 0.90 kg/m² (–0.41 to –0.16) compared with the respective baseline levels; the between-group difference was not statistically significant (P = 0.664).

In the SZ-A group, 189 AEs (incidence rate 53.7%) and 19 SAEs (incidence rate 5.4%) were recorded during the treatment. Among these, 54 (15.3%) were TAEs, but none were serious TAEs. In the acarbose group, 135 AEs (incidence rate 57.2%) and 9 SAEs (incidence rate 3.8%) were recorded. Among these, 67 (28.4%) were TAEs (Supplementary Materials 8–10). There were no significant between-group differences with respect to the incidence of AEs or SAEs (P = 0.402 and P = 0.377, respectively). The incidence of TAEs in the SZ-A group was significantly lower than that in the acarbose group (P < 0.001) (Fig. 2A).

GDs were the main TAEs in both groups. There were 43 cases of GDs in the SZ-A group (incidence rate 12.2%), which accounted for 79.6% of TAEs in this group. There were 58 cases of GDs in the acarbose group (incidence rate 24.6%), which accounted for 86.6% of TAEs. The incidence of GDs in the acarbose group was significantly higher than that in the SZ-A group (P < 0.001) (Fig. 2A). Diachronic analysis showed that newly reported GDs occurred mainly at 4–8 weeks, and the incidence showed a gradual decrease with treatment. However, the incidence of newly reported GDs in the SZ-A group was lower than in the acarbose group at all time points; the between-group differences in this respect were statistically significant at 4 and 8 weeks (P = 0.033 and P = 0.013, respectively) (Fig. 2B).

There were seven patients with cardiovascular and cerebrovascular SAEs, including four in the SZ-A group (incidence rate 1.1%) and three in the acarbose group (incidence rate 1.3%), with no significant between-group difference (P = 1.000). According to the determination of researchers and independent experts in the condition of unexposed blindness, none of the SAEs mentioned above had anything to do with therapeutic drugs. No hypoglycemia events occurred in either group during the trial.

Previous studies have demonstrated diverse pharmacological effects of SZ-A, such as regulation of blood glucose concentration, insulin secretion, and status of insulin resistance (15–17). The inhibitory effect of SZ-A on α-glycosidase may account for its hypoglycemic effect (23,25). Acarbose, a type of α-glycosidase inhibitor, can reduce HbA_1c by 0.5–1.0% (5.5–10.9 mmol/mol) with a curative effect. It is widely used for the treatment of T2D, especially in regions with a predominantly cereal diet (18,27–32). Therefore, acarbose was selected as positive control. We adopted a double-blind and double-dummy study design to obtain more robust results and to minimize potential bias. With respect to randomization, each participating center competed for inclusion through the central stochastic system.

This study observed that SZ-A imparts a good hypoglycemic effect. After treatment for 24 weeks, the primary efficacy end point HbA_1c decreased by 0.93% (10.2 mmol/mol), which was comparable to that achieved with acarbose (50 mg t.i.d.). The UK Prospective Diabetes Study demonstrated that the risks of diabetes-related end events, diabetes-related death, diabetes-related myocardial infarction, and diabetes-related microvascular complications will decrease by 21%, 21%, 14%, and 37%, respectively, along with the reduction of HbA_1c by 1% (10.9 mmol/mol) (33). According to contemporary clinical guidelines, HbA_1c <7% (<53 mmol/mol) can effectively reduce the occurrence of chronic complications of diabetes and reduce the incidence of hypoglycemia (18,20,27,30). The proportion of patients who achieved HbA_1c <7.0% (<53 mmol/mol) in the SZ-A group (47.7%) was similar to that in the acarbose group (45.0%).

The above results indicate that the hypoglycemic effect of SZ-A is comparable to that of acarbose. The results of the current study are consistent with those of the single-center study we previously reported (34). Weight gain is an important issue in the treatment of diabetes; effective weight management is of utmost significance for the achievement of therapeutic effect (8,20,27,30). Our findings suggest that SZ-A has no negative effect on weight compared with acarbose.

Safety is an important consideration during the evaluation of new drugs. In this study, there was no significant between-group difference with respect to the incidence of AEs or SAEs. Furthermore, the incidence of TAEs in the SZ-A group was significantly lower than in the acarbose group. GDs are the main side effects of acarbose. In our study, the incidence of GDs in the acarbose group (24.6%) was similar to that in previous reports (5,21,22,28,29,31,32,35). The incidence of GDs in the SZ-A group was 12.2%, which was much lower than the acarbose group, showing a significant difference. In vitro experiments have demonstrated a strong inhibitory effect of SZ-A on maltase and sucrase, which is similar to that of acarbose. However, SZ-A showed no inhibitory effect on amylase at 100 g/mL, while acarbose had an obvious inhibitory effect on α-amylase (IC₅₀ 1.74–2.4 g/mL). It can be concluded that SZ-A has stronger selectivity in inhibiting maltase and sucrase. The main active ingredients of SZ-A, namely, 1-DNJ, FA, and DAB, are more likely to form hydrogen bonds with the active structural domains of maltase and sucrase, which explains their greater affinity for disaccharidases (36,37). These findings may partly explain the underlying mechanism of the reduced incidence of GDs in the SZ-A group in our study.

In previous studies, after oral intake of 40, 200, and 1,000 mg/kg of SZ-A, the main components of 1-DNJ, FA, and DAB showed absorption saturation with an increase in dose; the absolute bioavailability was 33.3–72.4%, 29.9–77.5%, and 40.1–78.2%, respectively. After oral administration, these drugs are mainly distributed in the gastrointestinal tract, partially absorbed into the blood, and excreted by the kidneys in their original form rather than metabolized by the liver; therefore, these have no obvious effect on hepatic enzymes. The 24-h urine and feces recovery of 1-DNJ, FA, and DAB was nearly 100%, indicating no significant accumulation of SZ-A in the body (23). In our study, we did not find an incidence of TAEs, such as abnormal liver and kidney function, in the SZ-A group, which may be related to the plasma pharmacokinetic characteristics of SZ-A. However, the effects of SZ-A on liver and renal function still need to be paid attention in clinical application.

Some limitations of our study should be considered while interpreting the results. First, the observation period was only 24 weeks; a longer observation period is more conducive in further evaluating the efficacy and safety of SZ-A. Second, SAEs related to cardiovascular and cerebrovascular diseases were observed in both the SZ-A and the acarbose groups. However, these SAEs were determined by researchers and independent experts who were blinded to the group identity and were found to be unrelated to drug therapy; moreover, there was no significant between-group difference with respect to the incidence of SAEs. The Acarbose Cardiovascular Evaluation trial has confirmed that acarbose is similar to placebo without any adverse effects on cardiovascular end points (38). Therefore, we hypothesize that SZ-A may not produce adverse cardiovascular and cerebrovascular effects; however, further studies are required to evaluate its impact on cardiovascular and cerebrovascular end events. Third, because this study excluded patients with irregular drug intake regardless of lipid-lowering or glucose-lowering ones, it is not known whether the effect of SZ-A in these patients was consistent with that in this study population. In addition, animal studies have demonstrated other pharmacological effects of SZ-A, such as hypolipidemic effect, improvement in basal insulin level, alleviation of insulin resistance, and insulin secretion stimulated by glucose (16,17). Future studies should explore the extensive targets of SZ-A in multiple tissues and organs. Furthermore, phase IV clinical studies should observe the impact of SZ-A on metabolic syndrome, weight control, and cardiovascular risk and complications and evaluate its comprehensive benefits during long-term treatment of patients with T2D.

In summary, in this study, SZ-A and acarbose showed similar effects in reducing HbA_1c. The effect of SZ-A on body weight was equal to that of acarbose without hypoglycemia. In addition, SZ-A caused fewer TAEs and GDs than acarbose, suggesting good tolerance in patients with T2D.

Acknowledgments. The authors thank Prof. Qingshan Zheng, Pharmaceutical Clinical Research Center of Shanghai University of Traditional Chinese Medicine, for statistical support. In addition, the authors thank the participants, experimenters, and researchers for their participation. The authors thank LetPub (https://www.letpub.com) for linguistic assistance during the preparation of this manuscript.

Funding. This study was supported by the National Science and Technology Major Project (2013ZX09101005).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.Q. contributed to drafting and revising the manuscript. L.Q., X.L., G.T., Ga.Z., Q.W., X.H., Y.C., Y. Liu, Z.S., C.X., Y.Q., H.M., Y.Z., Zi. Li, S.Y., Xu.Z., J.Y., G.C., Yi Li, G.Y., J.H., X.W., Zh. Li, Yu. Li, Xiu.Z., Gu.Z., L.C., W.H., M.Y., C.L., Xia.Z., and H.J. contributed to the clinical data collection. X.L. and G.T. designed the study and contributed to the final approval of the manuscript. Y. Liu and Z.S. contributed to the study conception. G.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.