Unique study design had each patient act as their own control
Trial statistician Beverley Shields, PhD, of the University of Exeter, U.K., reported the results. The 525 participants with type 2 diabetes were aged 30-79 years and had A1c levels above 58 mmol/mol (7.5%) but not greater than 110 mmol/mol (12.2%) with metformin alone or combined with a sulfonylurea. Just over half (58%) had a BMI above 30 kg/m2 and 52% had an eGFR greater than 90 mL/min/1.73m2.
Each participant received each of the three medications as second- or third-line oral therapy in random order – in one of six possible sequences – for 16 weeks each, with no washout period in between (to prevent dropouts due to hyperglycemia). Thus, each participant acted as their own control.
A total of 458 participants completed all three study periods.
The drugs work differently in different patient groups
Without stratification by patient type, there was no overall difference in A1c reduction between the three therapies, with all achieving about 59-60 mmol/mol (7.5-7.6%) from a baseline average of 69 mmol/mol (8.9%).
But when stratified by BMI, A1c was 1.48 mmol/mol higher with pioglitazone versus sitagliptin in the group with BMI less than 30 kg/m2 and 1.44 mmol/mol lower with pioglitazone versus sitagliptin in the group with BMI greater than 30 kg/m2, giving a significant overall difference of 2.92 mmol/mol (P = .003).
By eGFR stratification, A1c was 1.74 mmol/mol lower with sitagliptin than canagliflozin in the 60-90 mL/min/1.73m2 group and 1.08 mmol/mol higher in the greater than 90 mL/min/1.73m2 group, giving a significant difference of 2.83 mmol/mol (P = .002).
“So, if we were to treat the patients with the drug that is optimal for their strata ... this would lead to a benefit of about 3 mmol/mol compared to if those patients were treated with the other drug,” Dr. Shields said.
By BMI, there were no significant differences by drug or strata for tolerability, defined as staying on drug for at least 12 weeks (P = .2), nor in the percentage of patients reporting at least one hypoglycemic episode (P = .6).
However, pioglitazone was associated with higher weight gain in both BMI groups, resulting in a 0.93 kg difference overall (P < .001), although it was higher in the higher BMI group (1.9 vs. 0.97 kg).
Similarly, by eGFR there were no differences in tolerability or hypoglycemic episodes between sitagliptin and canagliflozin (P = .09 and P = .6, respectively). And here, there were no differences in weight (P = .6).
Patients compared their own experiences with each drug
Patients were asked about their drug preferences after being reminded about their own changes in A1c and weight with each one. The result was a split: 25.8% picked pioglitazone, 34.8% sitagliptin, and 38.7% canagliflozin.
Looking at study outcomes by therapy, pioglitazone had the lowest rate of nontolerability but the highest weight gain, sitagliptin had the highest nontolerability but the lowest number of side effects, while canagliflozin had the highest number of reported side effects but the lowest weight gain.
Patients’ preferred drugs were associated with the lowest A1c and the fewest side effects for each group. Interestingly, pioglitazone was associated with the highest weight on therapy regardless of preference, so that even those who preferred pioglitazone had a higher weight than they did with the other two drugs.
In response to an audience question about durability of the results given the relatively short trial periods, Mr. Hattersley said: “We’re following up these patients who have chosen their drug, and on the whole, their primary care doctor agreed with them. So we’re following that up as a prospective cohort. We’re looking at tolerance and response and also to see if they’re still happy with that drug. That will be a future analysis.”
The TriMASTER data will be submitted for publication soon.
TriMASTER was funded by the UK Medical Research Council. Mr. Hattersley and Dr. Shields have reported no relevant financial relationships. Dr. Kistorp has reported receiving honoraria from and/or is on advisory boards for AstraZeneca, Novo Nordisk, Boehringer Ingelheim, MSD, Otsuka Pharma, and Chiesi.
A version of this article first appeared on.