Use and effectiveness of dapagliflozin in patients with type 2 diabetes mellitus: a multicenter retrospective study in Taiwan

Data source and study population

This was a multicenter retrospective observational study (ClinicalTrials.gov identifier NCT03084965) enrolling patients with T2DM exposed to dapagliflozin after dapagliflozin reimbursement in Taiwan since May 2016. Medical information was extracted manually by chart review and recorded in a standardized case report form after obtaining written informed consent from the patients. The final protocol of this study, including the final version of the subject informed consent form, had been approved by the Ethics Committee/Institutional Review Board (IRB)/Independent Ethics Committee (IEC) of each sites (Institutional Review Board Changhua Christian Hospital, Research Ethics Committee China Medical University & Hospital, The Institutional Review Board Chung Shan Medical University Hospital, Institutional Review Board Chia-Yi Hospital, Taipei City Hospital Research Ethics Committee, Chang Gung Medical Foundation Institutional Review Board, Kaohsiung Medical University Chung-Ho Memorial Hospital, Institutional Review Board Chi Mei Medical Center, Institutional Review Board of the E-DA Hospital, The Institutional Review Board of Shin Kong Wu Ho-Su Memorial Hospital, Institutional Review Board, Tri-Service General Hospital, Research Ethics Review Committee Far Eastern Memorial Hospital, Institutional Review Board of the Cathay General Hospital, TMU-Joint Institutional Review Board, National Taiwan University Hospital Research Ethics Committee, Institutional Review Board MacKay Memorial Hospital).

Patients were eligible for inclusion if they (1) were diagnosed with T2DM and aged ≥20 years old, (2) initiated 5 mg or 10 mg of dapagliflozin after May 1st 2016, either as add-on therapy to existing OAD(s) with or without insulin, or as switch therapy from another OAD, and (3) completed follow-up of at least 6 months regardless of continuation on dapagliflozin therapy. Data were excluded if they (1) received other SGLT2 inhibitors prior to the initiation of dapagliflozin, (2) had a diagnosis of type 1 diabetes, or (3) were included in other clinical trials concurrently during the retrospective data collection period.

Assessment and outcome measures

Patient baseline demographics and clinical characteristics were recorded at the time of dapagliflozin initiation. Reasons for starting/switching to dapagliflozin, rate and reasons of dapagliflozin discontinuation were collected. Measurements of HbA1c, body weight, BP, fasting plasma glucose (FPG), and lipid profile (total cholesterol, high-density lipoprotein cholesterol [HDL-C], low-density lipoprotein cholesterol [LDL-C], and triglycerides) were obtained at baseline, 3 and 6 months to assess for changes. Further subgroup analyses were performed to assess changes in HbA1c by baseline HbA1c, manner of dapagliflozin initiation (as add-on or switch), BMI, age, and dosage of dapagliflozin. Changes in body weight were also analyzed by subgroups of baseline BMI, HbA1c, age, and dosage of dapagliflozin. Patients who have one or more times record using 5 mg dapagliflozin during the follow up were defined as 5 mg group. The proportion of patients within each glycemic level (HbA1c <7%, 7–8%, 8–9%, and >9%) was evaluated at baseline and 6 months.

Statistical analysis

Descriptive statistics were provided for all variables. Continuous variables were presented with mean ± standard deviation, and numbers or percentages were used for categorical variables. Changes from baseline of the variables were analyzed using evaluable data at the respective time points. Mean changes of each variable from baseline to Month 3, baseline to Month 6, and between 3 and 6 months were evaluated by paired t-test and Wilcoxon signed-rank test in the overall cohort and subgroups. Differences in HbA1c reduction between dapagliflozin add-on and switch groups at Month 3 and Month 6 were analyzed by Wilcoxon rank-sum test. For changes in HbA1c and body weight among other subgroups at Month 3 and Month 6, the Kruskal-Wallis test was used. Patient distributions of changes in HbA1c and body weight at Month 6 were displayed by scatter plots for total, add-on, and switch groups. Baseline factors associated with dapagliflozin response in HbA1c were examined by univariate and multivariate logistic regression analyses. The cutoff for dapagliflozin responders was determined by using the median of change in HbA1c in patients with evaluable data at Month 6. Regarding the missing data, patients without the respective data were excluded in each analysis. For instance, when analyzing change from baseline to Month 3, only patients with both baseline and Month 3 data were included. Statistical significance was set at p value <0.05. All statistical analyses were conducted using SAS version 9.3.

Baseline characteristics

A total of 1960 patients were eligible, with a mean age of 57.8 ± 11.5 years, and 52% were male. The baseline clinical and laboratory characteristics are shown in Table 1. The mean HbA1c was 8.8 ± 1.4%, and body weight was 75.2 ± 15.8 kg. Baseline systolic blood pressure (SBP) and diastolic blood pressure (DBP) was 136.6 ± 17.6 and 77.8 ± 11.3 mmHg, respectively. Metformin was the most prescribed antihyperglycemic agent (92.4%), followed by sulfonylurea (70.7%) and dipeptidyl peptidase-4 (DPP4) inhibitor (53.5%). Over half of the patients were on either dual (38.2%) or triple (36.2%) therapy prior to initiating dapagliflozin. A large proportion of patients (78.9%) initiated dapagliflozin at the higher dose (10 mg). Regarding the manner of initiation, more than half (53.8%) of the patients were switched to dapagliflozin, while the others (46.2%) started dapagliflozin as add-on therapy. Main reasons for starting dapagliflozin were for its HbA1c lowering efficacy (81.2%), followed by less concern for weight gain (26.8%) and lower risk of hypoglycemia (11.4%).

Effectiveness of dapagliflozin on clinical and laboratory parameters

Table 2 shows changes in clinical and laboratory parameters after dapagliflozin initiation at 3 and 6 months. Compared with baseline, statistically significant reductions in HbA1c were observed both at Month 3 (−0.68%; 95% confidence interval [CI] -0.74, −0.62, p < 0.001) and at Month 6 (−0.73%; 95% CI −0.80, −0.67, p < 0.001). Considering patients with evaluable HbA1c data at baseline and Month 6, the proportion of patients achieving the glycemic target (HbA1c <7%) was 6% at baseline, and it increased to 19% by Month 6 after initiating dapagliflozin. Moreover, the proportion of patients who were poorly controlled (HbA1c >9%) decreased from 34.7% at baseline to 15.9% at Month 6 (Fig. S1A). Improvements in FPG (−28.3 mg/dL), BMI (-0.60), body weight (−1.61 kg), and SBP/DBP (−3.6/−1.4 mmHg) at Month 6 were also significant (all p < 0.001, Table 2). Aside from BP reduction that plateaued at Month 3, small but significant improvements were observed in HbA1c, FPG, BMI, and body weight from Month 3 to Month 6. Small changes in lipid profiles were also noted (Table 2).

Subgroup analyses: HbA1c and body weight

We performed analyses to examine the effects of dapagliflozin treatment on HbA1c and body weight among different subgroups. For HbA1c, a statistically significant trend of greater reduction was observed in patients with higher baseline HbA1c from baseline to 3 and 6 months (Fig. 1A). In patients with baseline HbA1c between 8–9% and those >10%, a significant greater HbA1c reduction was found at Month 6 compared with Month3. Patients who received dapagliflozin as add-on therapy had a significantly greater reduction in HbA1c (−0.82%) than those who were switched from one antihyperglycemic agent to dapagliflozin at Month 6 (−0.66%, p = 0.002, Fig. 1B). Considering patients with evaluable HbA1c data at baseline and Month 6, the proportion of patients achieving the glycemic target (HbA1c <7%) were 5% and 6% for dapagliflozin add-on and switch groups at baseline, and they subsequently increased to 23% and 15% by Month 6, respectively (Fig. S1B). When stratified by baseline BMI, no significant difference in HbA1c reduction was found across subgroups (Fig. S2A). Similarly, no difference among age subgroups was found for HbA1c reduction (Fig. S2B). When stratified by dosage of dapagliflozin, patients receive 10 mg dapagliflozin had significantly greater reduction in HbA1c (−0.74%) than those who receive 5 mg at month 6 (−0.67%, p < 0.001, Fig. S2C).

For body weight, treatment with dapagliflozin showed a statistically significant trend of greater weight loss with increasing baseline BMI from baseline to 3 and 6 months (Fig. 2). Among patients with evaluable data at both Month 3 and Month 6, further weight reduction was significant at Month 6 compared with Month 3 across all BMI categories, except for those with baseline BMI ≥35. When stratified by baseline HbA1c, significantly greater weight reductions were found in patients with lower baseline HbA1c throughout the study, despite similar baseline body weight among HbA1c categories (Fig. S3A). There was no difference in weight reduction across age subgroups at Month 3, but a significant difference was found at Month 6 (Fig. S3B). Comparing data between 3 and 6 months, significant reductions in weight were found in groups aged 40–65 and 65–75 years. When stratified by dosage of dapagliflozin, patients receive 10 mg dapagliflozin had significantly greater reduction in body weight (−1.65 kg) than those who receive 5 mg at month 6 (−1.43 kg, p = 0.011, Fig. S3C).

In addition, patients were stratified into four groups according to the number of antidiabetic therapies at baseline: monotherapy, dual therapy, triple therapy, and >3 drugs combination. Across these subgroups, no significant differences were observed in both HbA1c and body weight reductions (Fig. S2D; Fig. S3D).

Relationship between changes in HbA1c and body weight

Patient distributions of changes in HbA1c and body weight after initiating dapagliflozin for 6 months were presented in scatter plots (Fig. 3). Of the 1,094 patients with evaluable data, 79.9% and 77.6% of them experienced reductions in HbA1c and body weight, respectively. Moreover, 64.0% of patients showed simultaneous reductions in both outcomes (Fig. 3A). For patients in the add-on (n = 529) and switch (n = 565) groups, 69.2% and 59.4% had a reduction in both HbA1c and body weight, respectively (Figs. 3B and 3C). In terms of clinical meaningful change, 74.4% and 74.2% of patients had at least 0.5% and 1% reduction in HbA1c for 6 months, respectively. 37.1% of patients had at least 3% weight loss for 6 months. The baseline characteristics of at least 1% reduction in HbA1c, 3% weight loss and both were show in the supplement Tables 1–3, respectively.

Baseline factors associated with dapagliflozin response in HbA1c

Baseline factors associated with dapagliflozin response in HbA1c at Month 6 were shown in Table 3. Responders and non-responders were determined by using the median of change in HbA1c (−0.60%; min, max [−6.6, 3.2]) as the cutoff.

Univariate logistic regression analysis indicated that higher baseline HbA1c, FPG, add-on dapagliflozin, use of insulin were significantly associated with dapagliflozin response in HbA1c reduction, while being on dual therapy at the time of dapagliflozin initiation was significantly associated with less HbA1c reduction. In multivariate logistic regression analysis, significant associations with dapagliflozin response in HbA1c reduction were found in patients with higher baseline HbA1c (odds ratio [OR] 2.10; 95% CI [1.79–2.47], p < 0.001), and those who received dapagliflozin as add-on therapy (OR 1.60; 95% CI [1.19–2.14], p = 0.002).

Discontinue rate and reasons for discontinuation

Among eligible patients, the total number of discontinuation at Month 3 and Month 6 were 153 (7.8%) and 247 (12.6%), respectively. The main reasons for the discontinuation were inadequate HbA1c control (n = 83; 4.23%), intolerance (n = 46; 2.35%), or poor compliance to the current regimens (n = 9; 0.46%). The most commonly reported reasons for intolerance were genital or urinary tract infections (n = 15), frequency of urination (n = 8), and vaginal itching (n = 5).