Patient stratification for determining optimal second-line and third-line therapy for type 2 diabetes: the TriMaster study

Abstract

Precision medicine aims to treat an individual based on their clinical characteristics. A differential drug response, critical to using these features for therapy selection, has never been examined directly in type 2 diabetes. In this study, we tested two hypotheses: (1) individuals with body mass index (BMI) > 30 kg/m², compared to BMI ≤ 30 kg/m², have greater glucose lowering with thiazolidinediones than with DPP4 inhibitors, and (2) individuals with estimated glomerular filtration rate (eGFR) 60-90 ml/min/1.73 m², compared to eGFR >90 ml/min/1.73 m², have greater glucose lowering with DPP4 inhibitors than with SGLT2 inhibitors. The primary endpoint for both hypotheses was the achieved HbA1c difference between strata for the two drugs. In total, 525 people with type 2 diabetes participated in this UK-based randomized, double-blind, three-way crossover trial of 16 weeks of treatment with each of sitagliptin 100 mg once daily, canagliflozin 100 mg once daily and pioglitazone 30 mg once daily added to metformin alone or metformin plus sulfonylurea. Overall, the achieved HbA1c was similar for the three drugs: pioglitazone 59.6 mmol/mol, sitagliptin 60.0 mmol/mol and canagliflozin 60.6 mmol/mol (P = 0.2). Participants with BMI > 30 kg/m², compared to BMI ≤ 30 kg/m², had a 2.88 mmol/mol (95% confidence interval (CI): 0.98, 4.79) lower HbA1c on pioglitazone than on sitagliptin (n = 356, P = 0.003). Participants with eGFR 60-90 ml/min/1.73 m², compared to eGFR >90 ml/min/1.73 m², had a 2.90 mmol/mol (95% CI: 1.19, 4.61) lower HbA1c on sitagliptin than on canagliflozin (n = 342, P = 0.001). There were 2,201 adverse events reported, and 447/525 (85%) randomized participants experienced an adverse event on at least one of the study drugs. In this precision medicine trial in type 2 diabetes, our findings support the use of simple, routinely available clinical measures to identify the drug class most likely to deliver the greatest glycemic reduction for a given patient. (ClinicalTrials.gov registration: NCT02653209 ; ISRCTN registration: 12039221 .).

Extended Figure 1

Distribution of side effects experienced on each of the three study drugs (pioglitazone represented by blue bars, sitagliptin by yellow bars, and canagliflozin by red bars) for all instances where people tried the therapy (n=469 pioglitazone, n=474 sitagliptin, n=474 canagliflozin). Proportions experiencing the side effects at baseline shown by black bars.

Extended Figure 2

Scatterplots showing a) difference in on-treatment HbA1c between pioglitazone and sitagliptin (negative values favour pioglitazone, positive values favour sitagliptin) against BMI, and b) difference in on-treatment HbA1c between sitagliptin and canagliflozin (negative values favour sitagliptin, positive values favour canagliflozin) against eGFR. Line of best fit shown for each plot.

Figure 1

Study design for the TriMaster three-treatment, three-period crossover trial of pioglitazone, sitagliptin, and canagliflozin. Six sequences represent the 6 possible treatment orders for pioglitazone (P), canagliflozin (C) and sitagliptin (S). No washout between treatment periods.

Figure 2. The two main hypotheses being tested in TriMaster

Figure 3

Trial profile (CONSORT diagram): patient flow through the stages of the crossover trial and eligibility for primary analysis. Numbers presented for each visit are the numbers assigned each drug at that time. For an HbA1c to be valid for primary analysis, it needed to be taken after at least 12 weeks of therapy (exclusions indicated by <12wks), and participants needed to have at least 80% adherence on the therapy (exclusions indicated by adh<80%). P= pioglitazone, S=sitagliptin, C=canagliflozin

Figure 4

Effect of stratification on treatment response. Point estimates represent the mean difference in HbA1c between the two therapies for a) hypothesis 1 - pioglitazone and sitagliptin, with stratification by obesity (n=141 BMI<=30; n=215 BMI >30), and b) hypothesis 2 – canagliflozin and sitagliptin, with stratification by renal function (eGFR) (n=163 eGFR 60-90; n=179 eGFR>90). Error bars represent 95% confidence intervals. Overall difference between strata determined from drug*strata interaction in mixed effects analysis adjusting for period.