Introduction
Approximately 30 million people have diabetes in the United States [
1]. The majority of people with diabetes (~ 90%) are classified as having type 2 diabetes mellitus (T2DM), which is characterized by insulin resistance, beta-cell dysfunction, and hyperglycemia [
2,
3]. Chronic uncontrolled hyperglycemia in people with T2DM is associated with an increased risk of debilitating and potentially life-threatening micro- and macrovascular complications, including myocardial infarction (MI), stroke, end-stage renal disease (ESRD), and blindness, as well as premature death [
4,
5]. Consequently, the economic burden of diabetes in the United States is staggering; in 2012, the estimated cost of diagnosed diabetes was $245 billion, of which 18% was attributed to prescription medications to treat complications of diabetes and 12% was attributed to the costs of anti-hyperglycemic agents (AHAs) and diabetes-testing supplies [
6].
Controlling hyperglycemia can decrease the risk of complications in people with T2DM [
5]. The American Diabetes Association (ADA) encourages a patient-centered approach to T2DM management, with consideration of the overall risk profile (i.e., beyond glycemic control) when identifying treatment targets [
6]. Lifestyle modifications that result in healthier eating habits and increased physical activity are an important component of T2DM management that can lead to weight loss and improved glycemic control [
7]. When pharmacologic intervention is needed, metformin is generally preferred; however, most patients will require a second medication soon after metformin is initiated in order to meet treatment targets [
7]. Selection of a second-line therapy is at the discretion of the treating physician, with consideration of patient preferences, health history, and risk of side effects (e.g., hypoglycemia). Appropriate choices for combination therapy with metformin include sulfonylurea, thiazolidinedione, dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium glucose co-transporter 2 (SGLT2) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, and basal insulin [
6]. However, with emerging data demonstrating the cardioprotective effects of newer AHA classes (i.e., GLP-1 receptor agonists and SGLT2 inhibitors), some diabetes management guidelines have been revised to preferentially recommend the initiation of AHAs with demonstrated cardioprotective benefits [
8].
Agents that inhibit SGLT2 are the newest class of AHAs approved for the treatment of T2DM. These drugs work to lower glucose by reducing the renal threshold for glucose (RT
G), which increases urinary glucose excretion (UGE); SGLT2 inhibition is also associated with a mild osmotic diuresis and a net loss of calories that lead to blood pressure (BP) reduction and weight loss [
9]. Because this mechanism of action works independently of insulin, SGLT2 inhibitors are complementary to other AHA classes (including insulin) and have an inherently low risk of hypoglycemia [
9]. To date, four SGLT2 inhibitors—canagliflozin, dapagliflozin, empagliflozin, and ertugliflozin—have been approved for the treatment of T2DM in the United States, and the American Association of Clinical Endocrinologists (AACE) endorses the use of SGLT2 inhibitors as the first adjunctive oral AHA for combination therapy with metformin [
8].
Canagliflozin and dapagliflozin were the first SGLT2 inhibitors approved in the United States. In phase 3 studies, canagliflozin and dapagliflozin improved glycemic control, lowered body weight, and reduced BP in patients with T2DM on a variety of background AHAs, including metformin, with greater improvements generally seen with the highest approved doses of each drug (canagliflozin 300 mg and dapagliflozin 10 mg) [
10,
11]. There have not been any head-to-head clinical trials to evaluate the efficacy and safety of canagliflozin versus dapagliflozin in patients with T2DM, and, to date, the only head-to-head study of any SGLT2 inhibitors is a phase 1 study comparing the pharmacodynamic properties of canagliflozin 300 mg versus dapagliflozin 10 mg in healthy individuals [
12]. In this study, canagliflozin 300 mg lowered RT
G to a greater extent than dapagliflozin 10 mg, resulting in ~ 25% more UGE over 24 h; in addition, canagliflozin 300 mg, but not dapagliflozin 10 mg, also delayed glucose reabsorption and decreased postprandial glucose [
12]. Furthermore, unlike canagliflozin 100 mg and dapagliflozin 10 mg, canagliflozin 300 mg has been shown to reduce postprandial plasma glucose by transiently inhibiting intestinal SGLT1 [
13]. In the absence of head-to-head clinical data in patients with T2DM, indirect results obtained using Bayesian network meta-analysis (NMA) have been used to compare the efficacy of canagliflozin and dapagliflozin as an add-on to metformin; the NMA results indicated that better HbA1c lowering was achieved with canagliflozin 300 mg versus dapagliflozin 10 mg [
14‐
17]. The pharmacodynamic differences between canagliflozin and dapagliflozin may account for differences in efficacy and thus the differences in the occurrence of clinical outcomes with both agents.
Because T2DM is chronic and progressive, the financial burden over the long run is substantial. Efficient use of available health care resources requires economic assessment of the available treatment strategies, including estimation of cost-effectiveness, in order to inform the decision-making process [
18]. Although drug acquisition costs are an important consideration, cost-effectiveness calculations must also capture the cost offsets and improved welfare that are associated with the better patient outcomes achieved with improved management of T2DM over time. Owing to the impracticality of obtaining evidence over a sufficient duration to capture the full impact of interventions over time, economic modeling is widely used as a method to generate such evidence, thus enabling assessment of the impact of alternative interventions [
19,
20]. Economic simulations have been used to evaluate the cost-effectiveness of canagliflozin and dapagliflozin versus other AHA classes [
21‐
28]. For example, canagliflozin 100 and 300 mg have demonstrated cost-effectiveness in second- and third-line therapy versus the DPP-4 inhibitor sitagliptin 100 mg in Mexico [
21] and Canada [
23]. Dapagliflozin has also been found to be cost-effective as monotherapy and in second-line therapy versus sulfonylurea, DPP-4 inhibitors, and acarbose in Nordic countries (i.e., Denmark, Finland, Norway, and Sweden) [
22], the United Kingdom [
24,
25], Greece [
26], and China [
27,
28]. There have been few reports of head-to-head cost-effectiveness comparisons of SGLT2 inhibitors [
29‐
33]. Cost-effectiveness evaluations of canagliflozin versus dapagliflozin in the United Kingdom, Ireland, and Spain showed that canagliflozin was generally cost-effective compared with dapagliflozin as monotherapy and in second-line therapy with metformin [
29‐
33]; however, similar analyses have not been conducted in the United States setting to the best of our knowledge.
As described above, the differential glucose-lowering efficacy seen for canagliflozin and dapagliflozin may impact health outcomes, making it suitable for comparison via economic simulations. The purpose of the analysis reported in the present paper was to compare the cost-effectiveness of canagliflozin 300 mg versus dapagliflozin 10 mg in patients with inadequate glycemic control on metformin monotherapy over 30 years from the perspective of the third-party payer in the United States health care system.
Discussion
Economic simulation results suggest that canagliflozin 300 mg dominated dapagliflozin 10 mg as add-on to metformin in the United States from the perspective of the third-party payer. QALY gains and lower costs were observed with canagliflozin versus dapagliflozin in the majority of cohort replications, indicating a reasonable level of certainty.
The improvements in health outcomes and cost offsets seen with canagliflozin were largely attributable to better control of key biomarkers, including HbA1c, weight, and SBP. Of particular importance was the better glycemic control with canagliflozin 300 mg compared with dapagliflozin 10 mg, which was modeled based on previously reported NMA results that were used in the absence of head-to-head clinical trial data [
14,
17]. Note that other NMAs using different methodologies [
14‐
16] have reported similar findings to the NMA utilized in this study. The previously reported pharmacodynamic data that demonstrated greater UGE with canagliflozin 300 mg versus dapagliflozin 10 mg [
12] support the findings from these NMAs. The greater HbA1c lowering with canagliflozin was associated with a delay in intensification with insulin rescue, which not only led to greater QALY gains via fewer hypoglycemic events and avoided weight gain, but also yielded substantial cost offsets, primarily related to the acquisition costs of insulin.
Results from each sensitivity analysis corroborated the base case findings, as canagliflozin dominated dapagliflozin in scenarios that may be more reflective of the patient experience in the United States. Given the emphasis on patient-centered care, a higher HbA1c target may be appropriate in some patients; in the sensitivity analyses that used treatment intensification thresholds of 7.5% or 8.0%, canagliflozin continued to dominate dapagliflozin. With the assumption that the clinical effects from the trials would be transferable to a different patient population in a nonrandomized setting, canagliflozin also dominated dapagliflozin when patient characteristics from a real-world United States patient population were used. This suggests that canagliflozin 300 mg may be a cost-effective treatment option in actual clinical practice. Despite the chronic nature of T2DM, better health outcomes and lower costs of canagliflozin treatment were realized even over the shorter time horizons of 10 and 20 years, with canagliflozin dominating dapagliflozin. In addition, assuming the use of a lower-cost insulin treatment regimen as rescue therapy had no notable impact on the simulation results, as canagliflozin 300 mg continued to dominate dapagliflozin 10 mg.
Economic simulation modeling is a widely used tool (endorsed by the ADA Consensus Panel [
20]) that can be used to generate evidence needed to help make informed decisions about clinical outcomes, costs, and QALYs of competing T2DM treatments. A limitation of the analysis is the lack of head-to-head clinical evidence for canagliflozin and dapagliflozin in patients with T2DM. Therefore, treatment effects were sourced from a previously reported Bayesian NMA of 26-week data from clinical trials that included studies of canagliflozin and dapagliflozin in combination with metformin [
14,
17]. The NMA results used to inform these simulations were robust, and the greater efficacy of canagliflozin versus dapagliflozin used as inputs in the model was in line with observations from similar analyses [
15,
16,
62]. Of note, the use of 26-week data for the NMA is supported by the clinical data for each individual drug, as the nadir in glycemic efficacy was observed near this time and was shown to be sustained for 52 weeks in longer studies [
51,
52,
63]. For comparison, a NMA using 52-week data from add-on to metformin studies showed similar reductions in HbA1c (– 0.76% with canagliflozin 300 mg and – 0.48% with dapagliflozin 10 mg) with a considerably smaller network, and using these values as treatment effects in the cost-effectiveness simulations produced results consistent with the simulations based on the 26-week data (data not shown). Because an evidence-based approach was used to reduce bias in the model, conservative modeling assumptions were used for parameters that lacked reliable data sources or had relatively higher levels of uncertainty.
The model accounts for the potential impacts of AEs that are known to be related to the mechanism of SGLT2 inhibition. However, the comparative risk associated with canagliflozin and dapagliflozin treatment on amputation, which was identified as a safety risk in the CANagliflozin cardioVascular Assessment Study (CANVAS) Program [
64], was not modeled owing to the lack of comparable data for dapagliflozin. Analysis of safety data from the Multicenter Trial to Evaluate the Effect of Dapagliflozin on the Incidence of Cardiovascular Events (DECLARE) will be important to confirm or refute whether amputation is a class effect [
65], and emerging safety data from this study will be incorporated into future modeling exercises.
This analysis is strengthened by the use of ECHO-T2DM, which has demonstrated good model validity for established effects of AHAs on intermediate biomarkers such as glucose, body weight, BP, and renal function [
34], but it does not yet account for the potential direct cardioprotective effects of SGLT2 inhibitors that have been reported recently [
64,
66]. In the CANVAS Program, canagliflozin was associated with a reduced risk of cardiovascular death, nonfatal MI, or nonfatal stroke compared with placebo in patients with T2DM and established cardiovascular disease or cardiovascular risk factors [
64]. Similar observations in patients with established cardiovascular disease from the Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients (EMPA-REG OUTCOME) and the CVD-REAL observational study suggest that cardioprotection is likely to be a class effect for SGLT2 inhibitors [
66,
67]. Cardiovascular outcomes data from the CANVAS Program and EMPA-REG OUTCOME, together with forthcoming data from DECLARE, will help to inform any potential SGLT2 inhibition-mediated cardioprotective effects in updates to the ECHO-T2DM model. Including assumptions regarding this anticipated class effect in future modeling exercises will be particularly important for comparisons with other AHA classes.