The use of sodium glucose co-transporter 2 inhibitors (SGLT2i) has been associated with increased serum ketone levels. However, most previous studies included subjects who were either insulin or even drug naïve with relatively short duration of diabetes. It is well known that insulin deficiency increases the risk of developing ketoacidosis with SGLT2 inhibitors. Moreover, since the glucose-lowering effect of SGLT2 inhibitors is at its maximum at 3 to 6 months after use, the extent of increase in serum ketone levels and its clinical relevance with chronic use of SGLT2 inhibitors, especially among insulin-treated patients that often have longer duration of diabetes and potentially more insulin deficient than those who are insulin naive, have not been clearly defined. Therefore, the investigators perform this randomised study to evaluate the effect of SGLT2 inhibitors on serum ketone levels among Chinese patients with T2DM inadequately controlled with insulin therapy.

Sodium glucose co-transporter 2 (SGLT2) inhibitors introduce a novel approach of glycaemic control in type 2 diabetes (T2DM). Inhibition of SGLT2 causes glycosuria and lowers blood glucose levels regardless of insulin sensitivity and beta cell function. It has recently been shown that SGLT2 inhibition is efficacious and safe not only in diabetic patients with normal renal function but also in patients with chronic kidney disease stage 3a [estimated glomerular filtration rate (eGFR) 45-59 mL/min/1.73m2]. While the clinical efficacy has been well proven by various randomized controlled trials, the significance of increased serum ketone levels after SGLT2 inhibition, however, remains to be elucidated. Certainly, the risk of ketoacidosis, albeit small, has raised considerable concern among both patients and clinicians. On the other hand, although still controversial at this stage, an alternate fuel hypothesis has emerged that tries to explain the cardiovascular benefits observed with SGLT2 inhibitors.

Several mechanisms have been proposed to explain the increased serum ketone levels after SGLT2 inhibition. In patients who are on background insulin therapy, reduced insulin dose, hoping to minimize risk of hypoglycaemia during concomitant use of SGLT2 inhibitors, could increase lipolysis and hepatic ketogenesis. In addition, even among those who are insulin naïve, the use of SGLT2 inhibitors might decrease renal clearance of ketone bodies, or increase ketone production through augmented glucagon to insulin ratio. Recent studies had also demonstrated that SGLT2 inhibitors shifted substrate utilization from glucose to lipid oxidation, thereby contributing to increased ketones production. In a study involving 9 subjects with T2DM treated with dapagliflozin, plasma ketone levels increased significantly from 0.05 mmol/L to 0.19 mmol/L over 2 weeks. In another study of 66 subjects with T2DM treated with empagliflozin, plasma ketone levels did not rise after a single dose administration but increased statistically from 0.02 mmol/L to 0.06 mmol/L after 4 weeks. Importantly, both studies included subjects who were either insulin or even drug naïve with relatively short duration of diabetes. It is well known that insulin deficiency increases the risk of developing ketoacidosis with SGLT2 inhibitors. Moreover, since the glucose-lowering effect of SGLT2 inhibitors is at its maximum at 3 to 6 months after use, the extent of increase in serum ketone levels and its clinical relevance with chronic use of SGLT2 inhibitors, especially among insulin-treated patients that often have longer duration of diabetes and potentially more insulin deficient than those who are insulin naive, have not been clearly defined. Therefore, the investigators perform this randomised study to evaluate the effect of SGLT2 inhibitors on serum ketone levels among Chinese patients with T2DM inadequately controlled with insulin therapy.

Inclusion Criteria:

• Chinese
• Aged 21 to 75 both inclusive
• Type 2 diabetes on single or two doses of insulin therapy with or without metformin, which include intermediate acting human insulin, premixed human insulin or insulin analogues
• On stable insulin doses, as defined by less than 10% changes in total daily insulin dose within 3 months prior to randomization
• Suboptimal glycaemic control with baseline HbA1c ≥8.0% and ≤10.5%, taken within 2 months prior to randomization
• Body mass index between 21 and 40 kg/m2

Exclusion Criteria:

• Type 1 diabetes mellitus
• History of ketoacidosis
• Concurrent use of sulphonylurea or glucagon like peptide-1 receptor (GLP1) agonists
• Prior use of SGLT2 inhibitors, DPP4-inhibitors or GLP1 agonists within 3 months of randomization
• History of intolerance to SGLT2 inhibitors or DPP4-inhibitors
• Concurrent use of loop diuretics
• eGFR <45 ml/min/1.73m2 within 3 months prior to randomization
• History of acute or chronic pancreatitis
• History of benign or malignant pancreatic tumours
• History of bladder cancer
• Alcohol or drug abuse
• Pregnant or nursing women
• Women at childbearing age not using and refused to start chemical or mechanical contraception after randomization
• Severe liver disease with elevated plasma alanine aminotransferase (ALT) of more than five times the upper limit of normal, taken within 3 months prior to randomization
• Active or history of malignancy within 5 years prior to randomization
• Hospitalization for acute illness within 3 months prior to randomization
• Severe mental disorder
• Unable to understand written patient information and to give informed consent
• Ongoing participation in other clinical intervention trials
• Other unspecified concomitant conditions that deemed unsuitable for study participation upon professional judgments by principal investigators

Individual participant data will be shared upon special request to principal investigator.