Comparison with Literature
Four phase 3 randomised trials initially compared DPP-44i to placebo and established its cardiovascular safety, though it remained unclear whether it had a better cardiovascular safety profile than other anti-diabetics9, 21–23. Additionally, there is always a challenge in translating trial evidence into clinical practice and those recruited to trials are not always reflective of patients in the “real world” utilising these treatments. For example, in the TECOS cardiovascular trial for the DPP-4 inhibitor, sitagliptin, the patients recruited had better controlled diabetes at baseline (compared to “real world” populations).22 They had a HbA1c between 48 mmol/mol (6.5%) and 64 mmol/mol (8.0%) and were excluded if they had a history of two or more episodes of hyperglycaemia in the previous 12 months.22 In our study, the mean Hba1c at baseline was 74.4mmol/mol in the cohort, and 2.2% recorded a history of hypoglycaemia prior to initiation during which period they had only been on metformin for glycaemic control. Such a profile is more typical of patients that receive these add-on treatments in clinical practice.
In addition to placebo-controlled trials, the CAROLINA trial was also undertaken which was the first head-to-head trial of a DPP-4i, linagliptin against sulphonylurea, glimepiride. It was undertaken among a cohort of less advanced, type 2 diabetes patients.9 In CAROLINA, outcomes were compared among those prescribed linagliptin vs glimepiride, that were already on metformin or another first-line treatment; a more accurate reflection of clinical practice. They found that risk of MACE (defined as 3P-MACE including cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke) was similar with both treatments; HR = 0.98 [95% C.I., 0.84–1.14]. They also found no increased risk of hospitalisation for heart failure with DPP-4i, a cited concern previously.9, 24 Despite our use of an altered definition for MACE that our real-world dataset could support, our findings were similar in this population free of MACE at baseline; HR, 0.92 [95% C.I., 0.77–1.10]. CAROLINA excluded patients on other DPP-4is or sulphonylureas, on insulin and with more established cardiovascular disease.9 In our study, we examined the DPP-4 class as whole against sulphonylureas in real-world patients both with and without a history of MACE. We demonstrated similar cardiovascular event rates with both treatments, among those with a history of MACE at baseline as well as without. We also found similar outcomes with both treatments for all-cause mortality, and across several vulnerable subgroups by HbA1c, socioeconomic status and a history of SMI and chronic kidney disease.
Previous database studies undertaken using data from clinical practice and registries where DPP-4i and sulphonylureas have been compared have also shown similar risks of cardiovascular outcomes with both. However a major limitation of these studies has been limited follow-up, samples size, inadequate comparators and most commonly, an inability to adjust for key confounders such as HbA1c, weight and smoking status at baseline.11, 12, 25
There have been several updates to treatment algorithms in recent years for T2DM, following emergence of newer DPP-4i, SGLT-2 and GLP-1 analogues. Agents in the latter two classes have also demonstrated evidence of cardiovascular benefits in Phase 3 clinical trials,10 however choosing add-on therapy to metformin has become increasingly complex for prescribers. Indeed, the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) guidelines break down recommendations for add-on according to costs, risk of adverse events, as well as history of complications. For example, it specifically recommends preferential use of GLP-1 analogue or SGLT-2 inhibitors with proven cardiovascular benefit as add-on to metformin, among those with established or at high-risk of cardiovascular disease. Contrastingly, UK NICE guidance does not discriminate between add-on treatments except for GLP-1 analogues which are only recommended in those with a BMI ≥ 35kg/m2 or as an alternative to those needing insulin.26 This may explain why trends in UK clinical practice still suggest, that in those individuals both with and without baseline cardiovascular disease, DPP-4i and sulphonylureas remain the most commonly prescribed add-on treatments to metformin for the last decade.2 As a result, we were unable to extend our analysis as intended, to examine risk of MACE among those prescribed SGLT-2i and GLP-1 analogues as add-on to metformin as well. Due to small sample sizes, there was insufficient power to make meaningful conclusions from UK primary care data with regards to their longer-term effectiveness in this context.5
Several studies have compared DPP-4i vs sulphonylureas for other outcomes and these have consistently demonstrated similar glycaemic reductions, weight neutral/loss with DPP-4i and gain with sulphonylureas and a consistently higher risk of hypoglycaemias with sulphonylureas.3 Indeed in CAROLINA, 1 or more episodes of hypoglycemia occurred in 320 (10.6%) participants in the linagliptin group but in 1132 (37.7%) in the glimepiride group, highlighting a significant disparity.9 Previous concerns with DPP-4i regarding pancreatitis have also been largely allayed in recent studies, and more recently, evidence supporting their glycaemic efficacy in older adults has emerged as well.4 This study has now shown, using real-world data, that cardiovascular outcomes in clinical practice appear similar with both treatments. Despite this growing evidence supporting benefits of DPP-4i mainly from a safety and tolerability point of view, they do however remain more costly than sulphonylureas, meaning setting-specific economic evaluations remain essential in guiding selection.
Strengths and Limitations
There are several strengths to this study. The study used real-world data from UK primary care where prescribing for T2DM and add-on therapy are managed, ensuring data are comprehensive and reflect true clinical practice. We examined cardiovascular outcomes, among a cohort of patients in routine clinical practice who were already on metformin and at similar stages in their disease management trajectories, thus reducing the potential for confounding by indication.27 This makes populations prescribed sulphonylurea and DPP-4i more similar at baseline, as was found in our analysis of baseline characteristics. There have been few head-to-head clinical trials that have compared the effects of different anti-diabetics on cardiovascular event rates in T2DM, among both patients with and without established cardiovascular disease, hence this study addresses an importance evidence gap. Despite growing evidence to support cardiovascular benefits of some SGLT-2 inhibitors and GLP-1 analogues in those with higher baseline cardiovascular risk,10, 28 DPP-4i and sulphonylureas remains the most widely prescribed treatments in the UK due to both cost and practical reasons, given need for subcutaneous administration for GLP-1 analogues.1 Hence, evidence comparing DPP-4 inhibitors and sulphonylureas remains clinically highly relevant. Our samples size allowed for robust outcome and exploratory subgroup analyses.
There are notable limitations. We were unable to analyse cardiovascular outcomes for other add-on treatments to metformin such as SGLT-2 inhibitors and GLP-1 analogues due to lack of sample size. Our subgroup analysis, which raised the possibility of a protective cardiovascular effect with DPP-4i in older adults and those with BMI < 25kg/m2, needs further investigation as the sample size is small, but does warrant further investigation given use of anti-diabetics is common in both these groups. As with any observational study, despite extensive adjustment for potential confounders including key ones excluded in several previous studies, a risk of residual and unmeasured confounding remains.