Dapagliflozin in combination therapy for treating type 2 diabetes

Clinical effectiveness

4.5 The Committee considered the evidence on the clinical effectiveness of dapagliflozin compared with other antidiabetic therapies, noting that most of the data came from the network meta-analyses submitted by the manufacturers. The Committee noted that, although the WinBUGs programme code used to run the original network meta-analyses provided in the manufacturers' submission differed from the code recommended by the NICE DSU in their technical support document, the manufacturers had also provided revised network meta-analyses that were based on the recommended code. The Committee also noted that the results of the manufacturers' revised network meta-analyses were similar to those from the original analyses. The Committee concluded that the results of the manufacturers' revised network meta-analyses provided an appropriate basis for making decisions about the clinical effectiveness of dapagliflozin and other antidiabetic therapies.

4.6 The Committee discussed the outcomes collected in the clinical trials and network meta-analyses, noting that the primary outcomes were intermediate rather than clinical outcomes. The Committee noted that studies including the UKPDS had then been used to provide a link between these intermediate outcomes and long-term clinical outcomes including micro- and macrovascular complications. The Committee heard from the clinical specialists that there was some uncertainty about the impact of HbA_1c reduction on longer-term macrovascular complications. The Committee also heard from the manufacturers that follow-up data were available for the clinical trials of dapagliflozin but that, because most of the trials of other antidiabetic drug therapies were of shorter duration, the clinical-effectiveness data used in the cost-effectiveness analysis had to be based on the short-term clinical trial data. The Committee concluded that, despite some uncertainty about the impact of HbA_1c reduction on longer-term macrovascular complications, it was prepared to accept the link between the intermediate outcomes collected in the clinical trials and the longer-term clinical outcomes.

4.7 The Committee considered the clinical effectiveness of dapagliflozin in dual therapy for people whose type 2 diabetes is inadequately controlled by metformin alone. The Committee noted that the evidence came from 3 clinical trials and a network meta-analysis. The Committee also noted that only 1 of the clinical trials of dapagliflozin had an active comparator (sulfonylureas), and that the clinical trial results were based on a relatively small number of patients who were given dapagliflozin at its licensed dose. However, on the basis of these clinical trial results, the Committee considered dapagliflozin to have greater efficacy than sulfonylureas for the outcomes of weight loss and systolic blood pressure reduction and similar efficacy for HbA_1c reduction. The Committee concluded that, on the basis of the results of the network meta-analyses (see sections 3.10 and 3.45), dapagliflozin in dual therapy as add-on to metformin appeared to provide similar glycaemic control to other antidiabetic drugs but may result in greater weight loss.

4.8 The Committee further considered the clinical effectiveness of dapagliflozin as dual therapy, noting that the manufacturers had not provided data on dapagliflozin as add-on therapy to a sulfonylurea, despite clinical trial data being available. The Committee accepted that most of the patients would start on metformin monotherapy, but noted the evidence provided by the clinical specialists that a proportion of patients who cannot tolerate metformin or for whom it is contraindicated would receive sulfonylurea monotherapy. It noted that the clinical effectiveness of dapagliflozin as an add-on to a sulfonylurea appeared to be consistent with its effectiveness when used as an add-on to metformin. The Committee concluded that, because the manufacturers had not provided clinical evidence on dapagliflozin as an add-on to a sulfonylurea, it could not make recommendations on this combination regimen.

4.9 The Committee considered the clinical effectiveness of dapagliflozin for people whose type 2 diabetes is inadequately controlled by insulin, noting that the evidence came from 2 clinical trials and a network meta-analysis. The Committee noted that both trials were placebo controlled and that 1 of these was of 12 weeks' duration only. Again, the Committee noted that the clinical trial results for dapagliflozin were based on a relatively small number of patients who were treated with dapagliflozin at its licensed dose. Further, the Committee noted that the network meta-analysis excluded trials of GLP‑1 analogues because they were not comparable to other trials included in the analysis and therefore consideration of the full range of possible comparators was restricted by the available evidence. The Committee concluded that, on the basis of the results of the network meta-analyses (see sections 3.12 and 3.45), dapagliflozin as add-on therapy to insulin appeared to have greater efficacy than DPP‑4 inhibitors for the outcome of weight loss and similar efficacy for HbA_1c reduction.

4.10 The Committee considered the evidence on the clinical effectiveness of dapagliflozin in triple therapy. The Committee noted that dapagliflozin is currently being studied as a triple therapy add-on to 2 other oral agents and that, in the absence of any other currently available clinical-effectiveness data, the manufacturers provided a post-hoc analysis of pooled data from a subset of older patients with type 2 diabetes and cardiovascular disease recruited in 2 trials of dapagliflozin as an add-on to metformin and a sulfonylurea. The Committee also noted that no direct head-to-head studies comparing dapagliflozin with other antidiabetic drugs currently exist and that the clinical-effectiveness data used to indirectly compare dapagliflozin with other antidiabetic drugs were taken from a previously published systematic review that had not been updated since 2009. It was aware of the limitations of these analyses highlighted by the manufacturer and therefore concluded that significant caution should be taken when interpreting the results of these preliminary analyses on the clinical effectiveness of dapagliflozin in the triple therapy setting.

4.11 The Committee considered the adverse events associated with dapagliflozin. It noted that common adverse events included urinary tract and genital infections and that these events were more common in women than in men. However, the Committee heard from the manufacturers that the recurrence of these events in the clinical trials was low. It also heard from the manufacturers that, because of the mechanism of action of dapagliflozin, the clinical trials had actively looked for such infections and that only a small proportion of these infections needed treatment. The Committee also noted that the incidence of hypoglycaemia was low when dapagliflozin was added to metformin and that the currently available evidence suggested that dapagliflozin was not associated with increased risk of cardiovascular events. However, it was aware that regulatory agencies had identified some uncertainty about the risk of some cancers associated with dapagliflozin. The Committee heard from the patient experts that adverse events were a concern for patients with type 2 diabetes if they result in the need for additional drug therapies, especially for patients who are already receiving many drug therapies for their condition. However, it also recognised that a new drug therapy that was associated with a lower risk of hypoglycaemia than some other existing therapies would also be valued by patients for whom driving might be a significant factor in their lifestyle or livelihood. The Committee concluded that the adverse-events profile of dapagliflozin was different from those of other antidiabetic therapies and that it was important to examine these adverse events when considering the manufacturers' economic model.

Cost effectiveness

4.12 The Committee considered the cost effectiveness of dapagliflozin as an add-on to metformin and insulin and as triple therapy in the manufacturers' submission, and the critique and exploratory analyses provided by the DSU and the ERG. The Committee noted that the manufacturers had provided a revised economic model in order to address concerns raised by the DSU about the original model and that the DSU considered that their concerns had been addressed. However, it also noted that the DSU and the ERG had identified a number of errors in the revised model which were subsequently addressed by the DSU in its exploratory analyses. The Committee concluded that the manufacturers' revised economic model with the subsequent amendments made by the DSU was acceptable for assessing the cost effectiveness of dapagliflozin in combination therapy for treating type 2 diabetes.

4.13 The Committee discussed the validation report provided by the manufacturers which compared the results from the revised model with the results that would have been obtained using the CORE diabetes model, which has been used in previous appraisals of treatments for type 2 diabetes (such as NICE's technology appraisal guidance on liraglutide and exenatide in combination with oral antidiabetic therapy). The Committee noted that the results generated from the CORE diabetes model were comparable to those obtained from the manufacturers' original and revised economic models for the dual therapy and insulin add-on therapy analyses. The Committee concluded that the results of the validation exercise with the CORE diabetes model provided reassurance about the integrity of the results obtained from the manufacturers' revised economic model.

4.14 The Committee discussed the cost-effectiveness analyses presented by the manufacturers, noting that these included a more restricted set of comparators than were specified in the scope. It was aware that GLP‑1 analogues had been included in the network meta-analysis for dual therapy but then subsequently excluded from the cost-effectiveness analysis. The Committee considered that it would have been more appropriate for all treatments in the network meta-analysis to have been included in the cost-effectiveness analysis. However, it noted the comments from the ERG and clinical specialists that GLP‑1 analogues were used in dual therapy on a restricted basis. On balance, the Committee concluded that the manufacturers had included an adequate range of comparators for the cost-effectiveness analysis of dapagliflozin in dual therapy as an add-on to metformin.

4.15 The Committee discussed the clinical-effectiveness data that were applied in the economic models. The Committee noted that the DSU had completed analyses of dual therapy add-on to metformin using different sources of clinical-effectiveness data. One analysis used the 24‑week network meta-analysis data for dapagliflozin, DPP‑4 inhibitors and thiazolidinediones and presented a separate comparison using head-to-head data from study 4 for dapagliflozin and sulfonylureas. The other analysis considered all treatments in a single analysis using the 52‑week network meta-analysis data. It heard from the manufacturers that, for the metformin add-on analyses, the trials of other antidiabetic therapies as add-on to metformin used a fixed dose but the trials of sulfonylureas did not have a stable dose over 24 weeks. Therefore, trials of sulfonylureas as an add-on to metformin were excluded from the 24‑week network meta-analysis. The Committee discussed which set of analyses was the most appropriate, noting that the estimates of efficacy differed between analyses. It considered that it was more appropriate to use a single source as was available in the 52‑week network meta-analysis, but was aware of the limited number of trials informing this analysis. The Committee noted that the 24‑week network meta-analysis only excluded sulfonylureas, and that the evidence from the clinical specialists suggested that dapagliflozin would be used where sulfonylureas were not appropriate. On this basis the 24‑week network meta-analysis data were appropriate. The Committee concluded that the results of the revised 24‑week network meta-analysis provided the most appropriate clinical-effectiveness data for the dual therapy analyses.

4.16 The Committee discussed the manufacturers' assumptions about the decision to switch or intensify treatment in the model, noting that this was based on baseline HbA_1c levels taken from the clinical trials and network meta-analysis. The Committee noted that the HbA_1c threshold levels for switching treatment in the original dual therapy and triple therapy analyses were above those recommended in NICE's guideline on type 2 diabetes and therefore may not reflect UK clinical practice. However, the Committee noted that the DSU's revised analyses applied an HbA_1c threshold for switching treatment that is currently recommended in NICE's guideline on type 2 diabetes. The Committee heard from the DSU that the results from the revised model were sensitive to the timing of treatment switching in the model which was dependent on the relationship between HbA_1c at the start of treatment, treatment-related changes in HbA_1c levels and the HbA_1c threshold levels for switching treatment. The Committee concluded that HbA_1c threshold levels for switching treatment as recommended in NICE's guideline on type 2 diabetes were appropriate to use in the economic modelling and as a basis for decision-making.

4.17 The Committee discussed the manufacturers' approach to modelling changes in body weight. The Committee noted that in the revised model the effect of treatment on changes in weight was applied gradually over the course of the first year, and considered that this was more plausible than the original model in which the effect of treatment on changes in weight was applied immediately. The Committee noted that, for treatments associated with weight loss, the manufacturers made assumptions about how long weight loss was maintained in the model (weight plateau), and about how long it took for the weight to increase to its baseline level after the plateau (loss of effect). The Committee understood that the changes made by the DSU meant that for treatments associated with weight loss, the weight profiles of the treatment groups now converged over time, but that for treatments associated with weight gain, differences in weight were maintained over the model time horizon. The Committee acknowledged that unpublished data from the clinical study of dapagliflozin and sulfonylureas as add-on to metformin provided by the manufacturers showed that patients who remained on dapagliflozin treatment without switching to other treatments maintained their weight loss over 4 years. However, the Committee considered that uncertainty remained about the effects of stopping treatment with dapagliflozin and the impact on weight gain. Therefore, it concluded that the scenario analysis conducted by the DSU, which involved the convergence of differences in weight profiles between treatment groups at the time of switching to the last line of treatment, was more appropriate for decision-making.

4.18 The Committee considered the utility values applied in the model, noting that in all analyses the majority of the QALY gains associated with dapagliflozin arose from the direct impact of weight change on health-related quality of life rather than from a reduction of diabetic complications and other adverse events. The Committee noted that utility values associated with changes in BMI were taken from a study commissioned by the manufacturers and that the methods by which these values were obtained were not in line with the NICE reference case for measuring and valuing health effects. The Committee also noted that this study produced different utility values associated with a 1-unit increase or decrease in BMI and that these were larger than other utility values that were identified in the literature. The Committee acknowledged that the manufacturers presented scenario analyses using alternative utility values for weight change and that these resulted in higher ICERs for the metformin and insulin add-on analyses. The Committee also noted that the loss in utility associated with a 1-unit increase in BMI (−0.0472) was similar to the loss in utility associated with a myocardial infarction (−0.055), which may not be credible. The Committee concluded that the utility values associated with changes in weight may have been too large and that the values (±0.0061 per BMI unit decrease or increase) applied in the manufacturers' scenario analyses and DSU analyses were more reasonable.

4.19 The Committee considered the utility values associated with hypoglycaemic events. The Committee heard from the ERG that they considered that the loss in QALYs associated with hypoglycaemic events may have been too large. The Committee was also aware that the loss in utility associated with severe hypoglycaemic events (−0.047) was higher than that applied in the economic model of third-line therapy with insulins, thiazolidinediones or exenatide in NICE's guideline on type 2 diabetes (−0.010). However, the Committee noted that after the publication of this guideline, the Driving and Vehicle Licensing Agency issued new regulations for people who have experienced a severe hypoglycaemic event in the previous 12 months. Therefore, the Committee acknowledged that any loss in utility associated with severe hypoglycaemic events may be higher in people for whom driving might be a significant factor in their lifestyle or livelihood. The Committee noted that the DSU had completed analyses that included both the higher and lower estimates of loss of utility associated with hypoglycaemic events, and that these had made small differences to the estimates of the ICER. The Committee therefore concluded that the utility values associated with hypoglycaemic events were not a critical factor in the decision-making.

4.20 The Committee considered the utility values applied to urinary tract and genital infections in the model, noting that the loss in utility associated with these events was much smaller than the loss in utility associated with other adverse events. The Committee considered that it was likely that there would be a greater loss in utility associated with these events than had been proposed by the manufacturers. The Committee also noted that the study commissioned by the manufacturers to examine the impact of weight change on health-related quality of life had also estimated the impact of urinary tract and genital infections, although these data were not presented in the manufacturers' submission. The Committee noted that in scenario analyses the manufacturers had applied a range of estimates for the loss in utility associated with urinary tract and genital infections. It was also aware that the results of the revised analyses were not sensitive to changes in these utility values. The Committee concluded that, although the loss in utility associated with urinary tract and genital infections was likely to be greater than that proposed by the manufacturers, it was satisfied that this did not significantly impact on the relative cost effectiveness of dapagliflozin as dual therapy or add-on to insulin.

4.21 The Committee was aware that the ERG had proposed alternative estimates for some costs, including drug acquisition costs for pioglitazone and the costs associated with diabetic complications. The Committee noted that pioglitazone is now off-patent and that the latest acquisition costs are substantially lower than those presented in the manufacturers' submission. The Committee acknowledged that the manufacturers were unable to provide this estimate in their submission, but considered that the DSU estimate of an average annual cost of £69.09 was reasonable. The Committee also noted that the manufacturers' revised model did not correctly adjust the annual inpatient and non-inpatient costs (estimated as £483 in the UKPDS 65 study) for people who did not experience a macro- or microvascular diabetic complication. The Committee concluded that it was appropriate to consider the latest acquisition cost of pioglitazone and that the manufacturers' revised model should be amended to correctly account for the annual costs incurred by people who did not experience a macro- or microvascular diabetic complication.

4.22 The Committee considered the most plausible ICERs for dapagliflozin as dual therapy in combination with metformin. The Committee considered that, on the basis of clinical specialist opinion that suggested that the use of pioglitazone in UK clinical practice was decreasing, a thiazolidinedione was not a key comparator in the dual therapy setting. The Committee also noted the evidence from the clinical specialists supported by the manufacturers that, in clinical practice, dapagliflozin would predominantly be used in combination with metformin when a sulfonylurea is not appropriate. Therefore, the Committee also considered that sulfonylureas were not a relevant comparator in the dual therapy setting. The Committee considered the DSU deterministic analysis and scenario analyses, which included the convergence of differences in weight between treatment groups at the time of switching to the last line of treatment. It noted that these showed that DPP‑4 inhibitors were associated with higher costs and QALYs than dapagliflozin, but that these differences were small. It noted further that in the DSU probabilistic sensitivity analysis these differences were even smaller. The Committee noted that the differences in QALYs were largely explained by the changes in health-related quality of life (utility) associated with changes in weight (BMI). Overall, the Committee concluded that because of the small differences in costs and QALYs between dapagliflozin and DPP‑4 inhibitors, dapagliflozin in a dual therapy regimen in combination with metformin could be recommended as a treatment option for people with type 2 diabetes that is inadequately controlled with metformin alone if it is used in the same scenario as described for the use of DPP‑4 inhibitors in NICE's guideline on type 2 diabetes.

4.23 The Committee considered the most plausible ICERs for dapagliflozin as add-on to insulin. It noted that in all the analyses conducted by the DSU the estimate of the ICER for dapagliflozin compared with DPP‑4 inhibitors was below £20,000 per QALY gained. The Committee considered that, in comparison to DPP‑4 inhibitors, dapagliflozin had been shown to be a cost-effective use of NHS resources. The Committee recommended dapagliflozin as a treatment option for people with diabetes inadequately controlled by insulin with or without other oral antidiabetic drugs.

4.24 The Committee discussed the results of the manufacturers' revised base-case analyses for dapagliflozin as triple therapy add-on to metformin and a sulfonylurea. It noted that the sequence of treatments in the manufacturers' revised economic model had been amended so that the approach was consistent with the dual therapy and insulin add-on analyses, with patients in the model starting treatment with triple add-on therapy. The Committee noted that in both the manufacturers' original and revised triple therapy analyses, dapagliflozin dominated other comparator drug therapies, meaning that dapagliflozin was associated with lower costs and higher QALYs than the comparators. However, the Committee noted that the clinical-effectiveness data applied in the triple therapy model were based on an indirect comparison of pooled data of 2 trials of dapagliflozin and a separate systematic review of other antidiabetic drug therapies conducted in 2009. The Committee was also aware that dapagliflozin is currently being studied as a triple therapy add-on to 2 other oral agents. The Committee considered that the cost-effectiveness analyses should be considered as exploratory in nature. The Committee concluded that dapagliflozin as triple therapy in combination with metformin and a sulfonylurea should not be recommended for treating type 2 diabetes except as part of the ongoing clinical trials.

Summary of Appraisal Committee's key conclusions