Patient preference for second- and third-line therapies in type 2 diabetes: a prespecified secondary endpoint of the TriMaster study

Abstract

Patient preference is very important for medication selection in chronic medical conditions, like type 2 diabetes, where there are many different drugs available. Patient preference balances potential efficacy with potential side effects. As both aspects of drug response can vary markedly between individuals, this decision could be informed by the patient personally experiencing the alternative medications, as occurs in a crossover trial. In the TriMaster (NCT02653209, ISRCTN12039221), randomized double-blind, three-way crossover trial patients received three different second- or third-line once-daily type 2 diabetes glucose-lowering drugs (pioglitazone 30 mg, sitagliptin 100 mg and canagliflozin 100 mg). As part of a prespecified secondary endpoint, we examined patients' drug preference after they had tried all three drugs. In total, 448 participants were treated with all three drugs which overall showed similar glycemic control (HbA1c on pioglitazone 59.5 sitagliptin 59.9, canagliflozin 60.5 mmol mol^-1, P = 0.19). In total, 115 patients (25%) preferred pioglitazone, 158 patients (35%) sitagliptin and 175 patients (38%) canagliflozin. The drug preferred by individual patients was associated with a lower HbA1c (mean: 4.6; 95% CI: 3.9, 5.3) mmol mol^-1 lower versus nonpreferred) and fewer side effects (mean: 0.50; 95% CI: 0.35, 0.64) fewer side effects versus nonpreferred). Allocating therapy based on the individually preferred drugs, rather than allocating all patients the overall most preferred drug (canagliflozin), would result in more patients achieving the lowest HbA1c for them (70% versus 30%) and the fewest side effects (67% versus 50%). When precision approaches do not predict a clear optimal therapy for an individual, allowing patients to try potential suitable medications before they choose long-term therapy could be a practical alternative to optimizing treatment for type 2 diabetes.

Extended Data Figure 1

Bar chart showing, at baseline, how patients ranked the importance of 5 particular attributes when considering a glucose lowering treatment. Data presented as proportions of patients choosing each level of importance for each of the 5 attributes.

Extended Data Figure 2

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) with proportions experiencing the side effects at baseline shown by black bars.

Extended Data Figure 3

a) HbA1cs and b) Side effects on the three drugs, split by preferred therapy on 1^st ranking before being fed back HbA1c and weight information. Bars represent the mean and error bars represent the 95% confidence intervals.

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 (see methods). Participant preference collected at study end.

Figure 2

Flow diagram showing participant flow through the study, with full breakdown of numbers on each drug at each stage and exclusions. P=pioglitazone, S=sitagliptin, C=canagliflozin.

Figure 3

a) Mean HbA1c b) mean number of side effects and c) weight at end of treatment period for each of the three therapies (pioglitazone in blue, sitagliptin in yellow, canagliflozin in red) split by the participants’ preferred drug on final ranking (shown by light background - pioglitazone in light blue, sitagliptin light yellow, canagliflozin in pink). Error bars represent 95% confidence intervals.