The global population with diabetes has increased from 151 million to 537 million people worldwide in the past two decades, 90–95% of whom have T2D1. T2D is characterized by a progressive deterioration of beta-cell function, which is associated with a continuous decline in insulin secretion function and is commonly related to insulin resistance2. Despite the availability of antidiabetic medications from nine different classes, only 49.2% of patients with T2D in China achieved the control target of a glycated hemoglobin (HbA1c) level of <7.0%3. Therapeutics such as metformin, glucagon-like peptide-1 (GLP-1) receptor agonists (liraglutide) and insulin (insulin glargine) failed to achieve durable improvements in beta-cell function in drug-naïve patients with T2D after the drug withdrawal4. Patients with T2D show marked delay in early-phase insulin secretion in response to glucose5,6. Targeting the earlier phases of insulin secretion, by improving glucose sensitivity, could be a valuable approach to improve beta-cell function7. New antidiabetic agents that directly target the driver of early-phase insulin secretion in glucose-stimulated insulin secretion (GSIS) are needed to optimize diabetes management8,9,10.

The first report of glucokinase was published in 1968 by Matschinsky, which identified glucokinase’s critical role in hepatic glucose metabolism and also its presence in insulin-producing islets of Langerhans of obese mice11. In beta cells, glycolysis is a transducer for triggering physiological GSIS12. Glucokinase is the first rate-limiting enzyme of glycolysis expressed in pancreatic beta cells and plays a central role in the GSIS-triggering process13. Glucokinase is recognized as a glucose sensor due to its regulation of set points in glucose homeostasis11,12,14,15. Inactivating and activating glucokinase genetic mutations cause critical impacts on the threshold of GSIS16. Homozygous and heterozygous inactivating mutations in the glucokinase gene are associated with the development of permanent neonatal diabetes, a life-threatening disease requiring intensive and lifelong insulin treatment17, and maturity-onset diabetes of the young type 2, which shows reduced insulin secretion in response to glucose stimulation and decreased hepatic glycogen content18,19, respectively. By contrast, activating mutations lead to clinical hyperinsulinemia, the most common cause of persistent hypoglycemia in newborns20. Glucokinase is also present in pancreatic alpha cells and intrinsically regulates glucagon secretion21,22,23. The majority of glucokinase is found in hepatocytes, where it is tightly regulated by the glucokinase regulatory protein in response to hepatic glucose changes and it controls hepatic glucose uptake and glycogen synthesis24. In patients with T2D, the reduction in glucokinase expression and activity in the pancreas and liver might explain the impaired GSIS, insulin resistance and loss of glycemic control. The reduction in glucokinase expression in the pancreas and liver has been reported by Matschinsky and Haeusler25,26. A substantial decrease in pancreas glucokinase activity was found in T2D islets, which showed impaired GSIS function25. A reduction of glucokinase expression in the range of 50–60% was found in the livers of patients with T2D whose HbA1c was above 7.0%26. Decreased expression of glucokinase in the liver and reduction of hepatic glycogen contents were associated with hypermethylation in the hepatic glucokinase gene promoter and were involved in increasing the age-dependent susceptibility to hepatic insulin resistance and diabetes27. Some evidence was also found for a defect in hepatic glucokinase activity in T2D28,29.

Glucokinase activators (GKAs) are a class of small molecules that bind to an allosteric site on glucokinase and aim to restore glucokinase function in patients with T2D. GKAs can be classified as pancreas and liver dual-acting GKAs30,31,32 and liver-selective GKAs33. Additionally, GKAs can be divided into full GKAs31,32,34 and partial GKAs35,36. Challenges emerged through phase 2 trials during the clinical development of GKAs for T2D37,38,39,40,41,42, such as problems related to hypoglycemia, unsustainable efficacy and other adverse reactions38 (Supplementary Table 143,44,45,46,47. Over the past 5 years, considerable progress has been made in developing GKAs, such as dorzagliatin, which showed good efficacy and safety profiles along with a low incidence of hypoglycemia in a 12-week study of patients with T2D48. GKAs have also been developed for diseases other than T2D, including TTP399 for type 1 diabetes (T1D)49 and AZD1656 for reducing mortality in patients with diabetes who were infected with severe acute respiratory syndrome coronavirus 2 (ref. 50).

Dorzagliatin is an orally bioavailable, dual-acting full GKA (molecular weight, 462.94 g mol−1) that activates pancreatic and hepatic glucokinase in a glucose-dependent manner to improve glycemic control in patients with T2D. Dorzagliatin demonstrated positive effects on recovering the number of insulin-secreting cells in the pancreas and on the glucokinase expression at both mRNA and protein levels in the livers of diabetic rats51. Multiple phase 1 trials34,52,53 and one phase 2 trial48 of dorzagliatin have been completed in China and the United States. In addition to effectively reducing 24-h plasma glucose levels, dorzagliatin improved the GSIS in patients with T2D who were treated for 28 days, as indicated by a significant increase in the early-phase insulin secretion index (ΔC30/ΔG30) and homeostasis model assessment 2 of beta-cell function (HOMA2-β) from baseline in a phase 1 trial52. Patients who were treated with dorzagliatin monotherapy of 75 mg twice a day for 12 weeks in a phase 2 trial showed significant improvement in the glucose disposition index and reductions in insulin resistance indicated by homeostasis model assessment 2 of insulin resistance (HOMA2-IR), which were sustained at 1 week after drug withdrawal48. This trial (SEED) is a phase 3 placebo-controlled study of dorzagliatin in drug-naïve patients with T2D, comprising 24 weeks of double-blind treatment followed by 28 weeks of open-label treatment to evaluate the efficacy and long-term safety of dorzagliatin.


Patient demographics and clinical characteristics

Patients were recruited from 18 July 2017 to 28 February 2019, and the last patient visit was conducted on 6 March 2020. Of the 975 patients screened, 463 eligible patients were randomly assigned to one of two treatment groups: 310 patients received dorzagliatin 75 mg twice a day, and 153 patients were assigned to the placebo group (Fig. 1). A total of 457 patients received at least one dose of the study drug and had HbA1c measurements at baseline and at least one post-baseline time point. Among all of the patients who underwent randomization, 401 (86.6%) patients completed the 24 weeks of double-blind treatment (281 (90.6%) in the dorzagliatin group and 120 (78.4%) in the placebo group). All patients were switched to dorzagliatin treatment after the 24-week double-blind period. Subsequently, 351 (75.8%) patients completed 52 weeks of treatment (246 (79.4%) from the original dorzagliatin group and 105 (68.6%) from the original placebo group; Fig. 1 and Extended Data Fig. 1). Among the 463 randomized patients, 301 (65%) were male and 162 (35%) were female. The average age of the study patients was 53.3 ± 9.7 years, and the average duration of disease was 11.7 ± 8.7 months. The patients had an average body mass index (BMI) of 25.5 ± 2.9 kg/m2 and a mean HbA1c value of 8.4% ± 0.7%. The demographic and baseline characteristics were similar between the two groups (Table 1).

Fig. 1: SEED study patient disposition.
figure 1

Flow diagram of SEED study patient disposition throughout the trial. The full analysis set (FAS) included 457 patients, who received at least one dose of the study drug and had at least one posttreatment measurement of the primary endpoint during the double-blind treatment period.

Table 1 Demographic and baseline characteristics

Efficacy outcomes

The primary outcome at week 24 was achieved. The HbA1c level was reduced from baseline by 1.07% (95% confidence interval (CI): −1.19%, −0.95%) in the dorzagliatin group and by 0.50% (95% CI: −0.68%, −0.32%) in the placebo group (estimated treatment difference (ETD), −0.57%; 95% CI: −0.79%, −0.36%; P < 0.001; Fig. 2a, Table 2 and Supplementary Table 2).

Fig. 2: Primary and secondary efficacy endpoints.
figure 2

a, The least-squares (LS) mean changes in the HbA1c level from baseline at week 24 in patients who received either dorzagliatin or placebo. The ETD and corresponding 95% CI were estimated using an MMRM without missing-value imputation in the FAS (dorzagliatin, n = 306; placebo, n = 148); P < 0.001. b, HbA1c level over 24 weeks in patients who received either dorzagliatin or placebo at each visit. c, The LS mean change in 2h-PPG from baseline. ETD and 95% CI were estimated in the FAS using a mixed model of repeated measure (MMRM) (dorzagliatin, n = 293; placebo, n = 133). d, The LS mean change in FPG from baseline. ETD and 95% CI were estimated in the FAS using an MMRM (dorzagliatin, n = 306; placebo, n = 148). e, The mean value of HbA1c over 52 weeks at each visit. The FAS comprised all randomized patients who took at least one dose of the study drug and had at least one posttreatment measurement of the primary endpoint during the double-blind treatment period. All statistical tests were two sided at a significance level of 0.05, and no adjustments were made for multiplicity. Data in a, c and d are presented as LS means ± s.e., and data in b and e are presented as means ± s.e.

Table 2 Changes in efficacy endpoints from baseline to week 24 and week 52

The secondary outcome showed that the reduction in HbA1c levels in the dorzagliatin group started at week 4, reaching the maximum reduction at week 12 (Fig. 2b, Table 2 and Supplementary Table 2). The effects of dorzagliatin on reducing 2-h post-challenge glucose (2h-PPG) and fasting plasma glucose (FPG) levels at week 24 compared to placebo are also shown (Fig. 2c,d, Table 2 and Supplementary Tables 3 and 4). A greater decrease in the 2h-PPG level was observed after treatment with dorzagliatin than with placebo (−50.9 mg dl−1 versus −9.0 mg dl−1; ETD, −41.94 mg dl−1; 95% CI: −57.06 mg dl−1, −26.82 mg dl−1; Fig. 2c, Table 2 and Supplementary Table 4). In addition, the FPG level showed a greater reduction in the dorzagliatin group than in the placebo group (−10.4 mg dl−1 versus −4.7 mg dl−1; ETD, −5.94 mg dl−1; 95% CI: −13.68 mg dl−1, 1.98 mg dl−1; Fig. 2d, Table 2 and Supplementary Table 3).

At week 24, 42.5% of the patients in the dorzagliatin group versus 17.3% in the placebo group achieved an HbA1c level of <7.0% (odds ratio (OR), 4.20; 95% CI: 2.51, 7.02). HbA1c levels of <7.0% were achieved by week 8 in 40.8% of patients on dorzagliatin versus 10.0% of patients on placebo, respectively, and were sustained until week 24 (Table 2).

The exploratory endpoints included the composite endpoint (an HbA1c level of <7.0% without hypoglycemia and no weight gain) and HOMA2-β at week 24. The composite endpoint in the dorzagliatin group was 29.4% at week 24, compared with 13.3% in the placebo group (OR, 2.87; 95% CI: 1.66, 4.96; Table 2). In addition, HOMA2-β showed consistent improvement at week 24 in the dorzagliatin group compared with the placebo group (+2.56 change from baseline with dorzagliatin versus −0.72 with placebo; ETD, 3.28; 95% CI: 0.44, 6.11; Table 2).

Moreover, from week 24 to week 52, in the open-label treatment period, the reduction in HbA1c observed in the dorzagliatin group during the double-blind period of the trial was sustained with reductions of 1.15% ± 1.10% and 1.11% ± 1.00% from baseline at week 24 and week 52, respectively. A decrease of 1.27% ± 1.02% in HbA1c levels from baseline was also observed in the original placebo group after the switch to dorzagliatin treatment at week 52 (Fig. 2e and Supplementary Table 5).

Post hoc analyses were performed to explore outcomes other than primary and secondary endpoints. The time to first achieving an HbA1c level of <7.0% was shorter in the dorzagliatin group than in the placebo group (12.1 weeks versus not estimable, respectively; Table 2).

Safety outcomes

During the 24-week double-blind treatment period, at least one adverse event was reported in 240 of 310 patients (77%) in the dorzagliatin group and 103 of 153 patients (67%) in the placebo group (Table 3). Most of the adverse events were mild and resolved during treatment, and the investigators considered them unrelated to the drug (Supplementary Table 8). No clustering of serious adverse events in any organ system occurred (Supplementary Table 9). In addition, no severe hypoglycemia events or drug-related serious adverse events were observed in the dorzagliatin group. The adverse events that occurred in at least 5% of the patients were upper respiratory tract infection, hyperlipidemia, protein in urine, abnormal hepatic function and hypertension, in order, which are mostly not related to investigational drug judged by investigators (Table 3). During the open-label treatment period, the rates of both adverse events and serious adverse events remained the same in terms of type and incidence as those that occurred during the double-blind treatment period. No deaths were reported during the study (Supplementary Table 8).

Table 3 Adverse events and hypoglycemic events during the 24-week double-blind treatment period

Clinically significant hypoglycemia (blood glucose level < 54 mg dl−1) was reported in 1 (0.3%) of the 310 patients in the dorzagliatin group during the 24-week double-blind period and none of the 281 patients during the 28-week open-label period (Table 3 and Supplementary Table 8), whereas it was reported in none of the 153 patients in the placebo group during the 24-week double-blind period and 1 (1%) of the 120 patients during the 28-week open-label period (Table 3 and Supplementary Table 8). The incidence of exposure-adjusted events for clinically significant hypoglycemia over the 52 weeks was 0.006 events per patient year.

There were no significant differences in blood pressure or BMI between the two groups. All laboratory indicators related to liver and kidney function were within the normal ranges during the study. Regarding blood lipids, total cholesterol, low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) were also within the normal range. At the same time, triglyceride level showed a minor increase at week 24 without further change noted over the 52 weeks of treatment (Table 4). The incidence of hyperlipidemia was 12% in the dorzagliatin group and 11% in the placebo group. Only one case of a hyperlipidemia adverse event in the dorzagliatin group was judged by investigators to be related to dorzagliatin (Table 3).

Table 4 Changes in selected laboratory results and vital signs


The SEED study was a phase 3 trial to evaluate the efficacy and long-term safety of dorzagliatin, a GKA, as monotherapy in drug-naïve patients with T2D. The population enrolled in the SEED study had a new diagnosis of T2D, were on a diet and exercise regimen for at least 3 months and had not previously received any antidiabetic treatment. The baseline characteristics of the participants reflect a group of patients with T2D primarily with impaired pancreatic beta-cell function rather than insulin resistance with an increased postprandial glucose and relatively high HbA1c (baseline of 8.4%). The SEED study achieved its primary efficacy endpoint of a reduction in HbA1c compared to placebo in a 24-week double-blind period.

Dorzagliatin had a good safety profile during the 52-week treatment period, with no drug-related serious adverse events or severe hypoglycemia that required medical assistance. Clinically significant hypoglycemia (blood glucose < 54 mg dl−1) occurred in one patient (0.3%) treated with dorzagliatin over 24 weeks, and the incident was mild in nature.

GSIS plays a central role in the glucose regulation of pancreatic beta cells. Despite an initial failure with GKAs in phase 2 clinical trials, such as piragliatin37 and MK-0941 (ref. 38), it has led to the understanding that glucokinase plays a glucose-sensing role in GSIS. The failure was due to clinical hypoglycemia induced by the disruption of the GSIS threshold. Allosteric GKAs developed for diabetes therapeutics can change the kinetic parameters of the glucokinase, such as the Hill coefficient, the changes of which can result in the alteration of its glucose-dependent activity, showing different clinical outcomes during their development. The excessive decrease in the Hill coefficient by GKAs might be the primary cause of a reduced GSIS threshold and an increased incidence of clinical hypoglycemia13,38,54. MK-0941 changed the GSIS threshold to 45 mg dl−1 of glucose, which is similar to the value under the effects of activating glucokinase mutations31,55. The chemical structure of dorzagliatin is markedly different from that of MK-0941, and it has shown minimum disruption of Hill coefficient values of glucokinase. In a phase 2 study, dorzagliatin demonstrated improved glucose dependency of insulin secretion and increased glucose sensitivity with low hypoglycemia risks48. Among the partial GKAs, AZD1656 and PF-04937319 are the two leading candidates evaluated in phase 2 trials. PF-04937319 is currently being studied in China in phase 3 trials56, and AZD1656 has shown encouraging outcomes for the reduction of overall coronavirus disease 2019 mortality in patients with diabetes in the ARCADIA study50. A liver-selective GKA, TTP399, which is chemically modified to optimize its uptake into hepatocytes and prevent its entry into pancreatic beta cells, has been developed for T2D and was shown to achieve sustained glycemic control without dyslipidemia with 6 months of treatment33. TTP399 also demonstrated benefits in patients with T1D when used as an adjunct treatment with insulin49.

Dorzagliatin also showed other clinical outcomes in different study populations through several phase 1 trials. In a combination study of dorzagliatin with sitagliptin in patients with T2D in the United States, active GLP-1 was increased in the dorzagliatin plus sitagliptin regimen compared to dorzagliatin alone, while GLP-1 secretion in the dorzagliatin alone was markedly enhanced compared with the combination therapy. In addition, dorzagliatin showed a synergistic glycemic reduction with the addition of sitagliptin57. Improved postprandial glucose control was observed using dorzagliatin as an add-on therapy to empagliflozin, an SGLT2 inhibitor, in patients with T2D in the United States57. Dorzagliatin was also studied in a population with end-stage renal disease and a healthy control group to evaluate the effect of renal impairment on pharmacokinetics and safety. The results suggested that dorzagliatin can be safely used in patients with T2D at all stages of renal impairment without a need for dose adjustment58.

The current study has several limitations. First, the SEED study focused on patients likely in the early stage of T2D, selected from newly diagnosed drug-naïve patients with T2D whose blood glucose was not controlled by diet and exercise regimens. Another phase 3 trial, the DAWN study59, was completed in patients with moderate T2D who had inadequate glycemic control using metformin alone. Therefore, the conclusions may not be generalizable to patients with severe T2D, and more clinical studies are needed. Second, both phase 3 studies were carried out in China; dorzagliatin has been evaluated in four phase 1 studies in the United States, including the pharmacological evaluation of combined dorzagliatin with metformin, sitagliptin and empagliflozin. Thus, long-term clinical studies in Europe and the United States are needed to evaluate the clinical benefit of dorzagliatin to the different T2D populations compared to Chinese populations. Third, according to ethical requirements, the duration of placebo observation in a diabetes monotherapy trial was limited to no more than 24 weeks; patients in the placebo group crossed over to dorzagliatin treatment during the extended 28-week open-label treatment period. No control group was designed for the entire 52-week period in this study. The total observation time in SEED was 52 weeks, and more time will be needed to observe dorzagliatin’s effects on cardiovascular and renal outcomes. No comparison with other hypoglycemic drugs was established in the SEED study, which will need to be considered in future trials. By the end of the 24-week double-blind treatment period, the proportion of participants missing HbA1c measurements was approximately 12%. Sensitivity analyses were performed to verify that the missing data had no impact on the primary analysis. Finally, the SEED study detected hypoglycemia by self-monitoring of blood glucose and laboratory glucose testing as well as self-reported symptoms associated with hypoglycemia. If a patient experienced a mild decrease in blood glucose but no overt symptoms before the detection time point, hypoglycemia will not have been detected.

In conclusion, we report that dorzagliatin as monotherapy for the treatment of drug-naïve patients with T2D is safe and well tolerated. Dorzagliatin treatment resulted in significant improvement in glycemic control starting at treatment week 4. These improvements were maintained throughout the 24-week double-blind treatment period and were sustained during a 28-week open-label extension out to 52 weeks of treatment.


Trial design

The SEED trial had a 2-week screening period and a 4-week single-blind placebo run-in period, a 24-week double-blind placebo-controlled treatment period followed by an extended 28-week open-label treatment period, and a 1-week safety follow-up period (Extended Data Fig. 1).


Patients were recruited at the study site by investigators. The patients were required to sign an informed consent form before any study procedures or screening for eligibility. Patients were not paid for taking part in the study, but the patients were compensated for the travel to site and blood draw at the site. During the study, patients were paid for completing different procedures at each visit. If the patients suffer any trial-related injury during the trial period, they can receive treatment at study hospital, and the sponsor will cover the relevant medical costs as well as financial compensation as defined in the relevant laws and regulations of China. Individuals (18 to 75 years old) were included in the trial if they met the following criteria: diagnosis of T2D; undergoing a diet and exercise regimen for at least 3 months; no antidiabetic pharmacotherapy before screening; HbA1c levels between 7.5% and 11.0%; a BMI between 18.5 and 35.0 kg/m2 at screening; and willingness to provide written informed consent and comply with the study protocol.

Individuals were excluded if they met the following criteria: T2D medical history of more than 3 years when signing the informed consent form; treatment with any antidiabetic drugs for more than 14 d within 3 years before screening; severe hypoglycemia with no known cause within 3 months if hypoglycemia occurred three or more times within 1 month before screening; fasting C-peptide level of <1.0 ng ml−1 (normal range: 0.81–3.85 ng ml−1) at screening; history of diabetic ketoacidosis, diabetes lactic acidosis or hyperosmotic nonketotic diabetic coma; diagnosis of T1D. Other exclusion criteria were: major cardio-cerebrovascular diseases within 6 months before screening; unstable or rapidly progressive kidney disease; active liver diseases; psychiatric diseases; hemoglobinopathy; immunocompromised participants; any type of malignancy; unstable endocrine system diseases; history of drug abuse; received oral or injected corticosteroids treatment within 1 year at screening; alcohol intake >2 units a day or >14 units a week; ALT or AST > 2.5 times the upper limit of normal (ULN); or a serum total bilirubin level > 1.5 × ULN; serological evidence of hepatitis virus infection at screening; eGFR < 60 ml min−1/1.73m2; triglyceride level > 5.7 mmol l−1; anemia; and any accompanied diseases, treatment or participant status that may impede completion of the whole study or potentially affect the interpretation of efficacy and safety data.

Trial procedures

At week 3 of the 4-week run-in period, the patients were reevaluated before randomization to confirm eligibility. Eligibility criteria included HbA1c levels between 7.5% and 10.0% and FPG levels between 126.0 and 239.4 mg dl−1 at randomization. Diet and exercise counseling was provided throughout the study.

Patients were randomly assigned at a 2:1 ratio to receive dorzagliatin or placebo during the double-blind period. Randomization and drug dispensation were performed with an interactive web response system (IWRS; Medidata RAVE RTSM, v2020.1.1). A stratified randomization method with the permuted block randomization algorithm was used. The blocks were dynamically allocated to each site and stratum from the randomization list. A unique ID number was provided by the vendor and marked on the medication box. Using central randomization, randomization codes were assigned to eligible participants by the IWRS system based on stratification factors (baseline HbA1c level ≤ 8.5% or >8.5%) and the block size. During the double-blind treatment, the random allocation sequences were concealed from patients, investigators and other study members until week 24. All patients received dorzagliatin twice a day during 28-week open-label period. The blinding was maintained throughout the entire 52 weeks of treatment. The placebo tablets had the same size, color, odor and appearance as the investigational drug.

Efficacy endpoints

The primary efficacy endpoint was the change from baseline in the HbA1c level at week 24. Key secondary efficacy endpoints included the changes from baseline in 2h-PPG, FPG levels at week 24, HbA1c level at each visit (except week 24) and the percentage of patients who reached an HbA1c level of <7.0% at week 24. Additional efficacy endpoints included the composite endpoint (HbA1c level of <7.0% without hypoglycemia and no weight gain) as well as the change from baseline in HOMA2-β and HOMA2-IR. During the study, HbA1c and FPG levels were measured and evaluated at visits 4 to 15. At visits 4, 7, 10 and 15, 2h-PPG was measured by a standardized mixed-meal tolerance test (MMTT).

Safety endpoints

Safety assessments were completed at each study visit. Blood and urine were sampled at each visit. Adverse events and serious adverse events were assessed throughout the trial. Hypoglycemic episodes were classified according to the American Diabetes Association (ADA) definitions. Vital signs and clinical laboratory test results were assessed, and physical examinations were performed. All laboratory tests were conducted in the central laboratory. All samples were collected at the site.

Trial oversight

The trial was conducted in accordance with the principles of the Declaration of Helsinki, Good Clinical Practice guidelines, and laws and regulations in China. The study protocol was amended once during study enrollment. The important changes to the protocol are listed as follows, which mainly involve exclusion criteria and randomization criteria: exclude patients treated with any antidiabetic drugs within 3 months before screening; round original fasting C-peptide values to 1 decimal place; short-term external use of corticosteroids treatment within 1 year at screening was allowed and inhaled glucocorticoid treatment was forbidden; stable coronary heart disease was allowed; exclusion and randomization criteria related to blood pressure and antihypertensive drugs were redefined as systolic blood pressure ≥ 160 mm Hg or diastolic blood pressure ≥ 100 mm Hg at screening, or who added/changed antihypertensive drugs or adjusted dose within 4 weeks before screening; added electrocardiogram (ECG) examination at visit 3 and evaluation of ECG results to the randomization criteria; added a visit window period of ±3 d for visit 2 to match actual operational needs; added the 30-min testing point for the MMTT to optimize the study protocol; removed the rule that patients need not be in fasting status at visit 2; and clarified samples collected was blood sample in workflow. Minor protocol revisions involve wording, consistency and accuracy. The trial protocol and amendments were approved by the local ethics committees of all study sites: Affiliated Drum Tower Hospital, Medical School of Nanjing University, Nanjing, China; Zhongshan Hospital, Fudan University, Shanghai, China; Nanjing First Hospital, Nanjing, China; Jingzhou Hospital Affiliated to Yangtze University, Jingzhou, China; The First People’s Hospital of Changde City, Changde, China; Jinan Central Hospital Affiliated to Shandong First Medical University, Jinan, China; The First Hospital of Shanxi Medical University, Taiyuan, China; Zhuzhou Central Hospital, Zhuzhou, China; The Second Hospital of Jilin University, Changchun, China; Chenzhou First People’s Hospital, Chenzhou, China; Taihe Hospital, Hubei University of Medicine, Shiyan, China; Tongji Hospital of Tongji University, Shanghai, China; The First Affiliated Hospital of Anhui Medical University, Hefei, China; The Second Affiliated Hospital of Nanjing Medical University, Nanjing, China; Wuxi People’s Hospital, Wuxi, China; Changsha Central Hospital, Changsha, China; The First Hospital of Jilin University, Changchun, China; Affiliated Hospital of Jiangsu University, Zhenjiang, China; West China Hospital, Sichuan University, Chengdu, China; The First Affiliated Hospital of Shandong First Medical University & Shandong Provincial Qianfoshan Hospital, Jinan, China; The Affiliated Hospital of Qingdao University, Qingdao, China; Siping Hospital of China Medical University, Siping, China; The Affiliated Hospital of Guizhou Medical University, Guiyang, China; The 960th Hospital of the PLA Joint Logistics Support Force, Jinan, China; Southern Medical University Nanfang Hospital, Guangzhou, China; The First Affiliated Hospital and College of Clinical Medicine of Henan University of Science and Technology, Luoyang, China; PLA Rocket Force Characteristic Medical Center, Beijing, China; Sanya Central Hospital, Sanya, China; The Second Affiliated Hospital of Hainan Medical University, Haikou, China; Chongqing Red Cross Hospital, Chongqing, China; Chongqing University Three Gorges Central Hospital, Chongqing, China; Tianjin Medical University General Hospital, Tianjin, China; The Third Medical Center of PLA General Hospital, Beijing, China; Peking Union Medical College Hospital, Beijing, China; Tongji Hospital, Tongji Medical College of HUST, Wuhan, China; The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China; Jiangxi Pingxiang People’s Hospital, Pingxiang, China; The Second Hospital of Dalian Medical University, Dalian, China; The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China; Xuzhou Central Hospital, Xuzhou, China; Shanghai Changzheng Hospital, Shanghai, China; and China-Japan Friendship Hospital, Beijing, China. Written informed consent was obtained from all patients before initiating any trial-related procedures. The trial was also conducted in accordance with the Chinese Diabetes Society guidelines, which require physicians to provide education and to strictly enforce improved exercise and dietary control as well as self-monitoring of blood glucose (at least twice per week) when treating patients with T2D.

Statistical analyses

We hypothesized that dorzagliatin would show superiority to placebo in decreasing HbA1c levels in patients after 24 weeks of treatment. For the primary endpoint HbA1c, we calculated that a total sample size of 450 patients would provide the trial with 91.4% power to detect a difference of 0.4% between the dorzagliatin group and the placebo group in a 2:1 ratio of allocation at a two-sided significance level of 0.05, assuming a standard deviation of 1.2%. The FAS included all randomized patients who received at least one dose of the study drug and had at least one posttreatment measurement of the primary endpoint during the double-blind treatment period. All major protocol deviations were reviewed in the blind-data-review meeting to determine whether a participant with any major protocol deviations should be excluded from the per-protocol set. The safety set included all randomized patients who received at least one dose of the study drug. Because this trial had only one confirmatory hypothesis test, that is, to test the null hypothesis of no difference in the primary endpoint between dorzagliatin and placebo, there was no adjustment for multiplicity.

The primary analysis method for the primary endpoint was in the FAS using an MMRM without missing-value imputation, which included treatment group, scheduled visit, the interaction of treatment group with the scheduled visit, pooled site and HbA1c baseline value as fixed effects. In the MMRM, data collected after initiation of prohibited antidiabetic medications were handled as missing data. The LS means of each treatment group, the ETD between dorzagliatin and placebo and its 95% CI were calculated. A sensitivity analysis was performed with the primary endpoint in the FAS using an ANCOVA with factors of treatment group, pooled site, interaction of treatment group with the pooled site (the interaction effect was removed if not significant on a two-sided alpha level of 0.1) and a covariate of HbA1c baseline value. In the ANCOVA model, missing values were imputed using the LOCF method, which was applied after excluding data collected after initiation of prohibited antidiabetic medications. Another sensitivity analysis was performed with the primary endpoint using the same MMRM in the per-protocol set.

The secondary efficacy endpoints, including 2h-PPG and FPG, were assessed using the same MMRM for the primary endpoint. Another secondary efficacy endpoint of the percentage of patients who reached an HbA1c level of <7.0% and an exploratory efficacy endpoint of the composite endpoint (HbA1c < 7.0% without hypoglycemia and no weight gain) were estimated based on the data imputed by the LOCF approach in the FAS. The OR and 95% CI between two treatment groups were estimated using the logistic regression model with categorical independent variables of the treatment group, pooled site, interaction of the treatment group with the pooled site (the interaction effect was removed from the model if not significant at a two-sided alpha level of 0.1) and a continuous independent variable of baseline HbA1c level. Another two exploratory efficacy endpoints of the change from baseline in the HOMA2-β and HOMA2-IR were analyzed using the ANCOVA model with factors of treatment group, pooled site, interaction of treatment group with the pooled site (the interaction effect was removed if not significant on a two-sided alpha level of 0.1) and baseline value of the analyzed variable. A post hoc analysis was conducted by using the Kaplan–Meier method to analyze the time to first reach an HbA1c level of <7.0% over 24 weeks. Furthermore, as a post hoc analysis to explore the in-group reduction in HbA1c from baseline at week 24 and week 52, paired t-tests based on observed measurements of HbA1c were carried out to compare baseline and week 24 or week 52 in each of dorzagliatin and placebo groups.

The incidence of treatment-emergent adverse events between the first intake of double-blind study drug and the seventh day after the last dose was summarized by the treatment group and compared between the two groups. The incidence of hypoglycemic events was also summarized by the treatment group and compared between treatment groups, including those spontaneously reported hypoglycemic events by patients and others reported at the on-site visit. The electronic data capture software (Medidata RAVE, Classic Rave v2018.2.4) was used for clinical trial patient data collection. Additional details regarding the statistical analysis are provided in the statistical analysis plan included in the Supplementary Information. Commercially available software (SAS v9.4, SAS Institute) was used for analyses.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.