Changes in HbA1c and weight, and treatment persistence, over the 18 months following initiation of second-line therapy in patients with type 2 diabetes: results from the United Kingdom Clinical Practice Research Datalink

Type 2 diabetes (T2D) is a complex chronic condition characterised by increased blood glucose levels and associated with micro- and macrovascular complications. In 2015, an estimated 415 million people globally had diabetes, with the vast majority (~ 90%) having T2D [1]. Optimal glycaemic control (i.e. attaining recommended treatment targets) has been shown to reduce the incidence of diabetes-related complications, microvascular disease and myocardial infarction in the large, long-term United Kingdom Prospective Diabetes Study [2, 3].

Along with lifestyle modifications, current guidelines recommend the use of metformin as the preferred first-line glucose-lowering therapy in patients with T2D [4]. Metformin has a low risk of hypoglycaemia and is weight neutral, with no increased risk of or benefit against adverse cardiovascular events [5, 6]. However, in view of its progressive nature, the majority of patients with T2D require treatment intensification from metformin monotherapy to achieve and maintain recommended HbA1c targets [6]. Second-line agents approved for dual therapy with metformin in the UK include sulfonylureas (SUs), thiazolidinediones (TZDs), dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium–glucose cotransporter-2 (SGLT-2) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists and basal insulin analogues. The National Institute for Health and Care Excellence (NICE) guidelines and the position statement of the American Diabetes Association and the European Association for the Study of Diabetes recommend that initial treatment intensification after failure of metformin monotherapy should consist of dual therapy with one of the following agents: DPP-4 inhibitors, pioglitazone, SUs or SGLT-2 inhibitors [4, 6]. The choice of second-line agent should be based upon patient-specific considerations to minimise side effects whilst reducing HbA1c levels [7].

Established agents such as SUs are highly efficacious but are associated with weight gain and may cause hypoglycaemia [8]. TZDs do not induce hypoglycaemia, but tend to cause weight gain and fluid retention, and have been associated with an increased risk of heart failure [6, 9]. Unlike SUs and TZDs, DPP-4 inhibitors do not lead to weight gain and a have minimal risk for hypoglycaemia, comparable to metformin [10, 11]. Likewise, SGLT-2 inhibitors do not induce hypoglycaemia and bring the added benefits of reductions in body weight and blood pressure [12‐14]. Results from the EMPA-REG OUTCOME [15] and CANVAS [16] randomised controlled trials and the CVD-REAL observational studies [17‐19] have shown a reduced risk for major adverse cardiovascular events, hospitalisation for heart failure or death in patients with T2D treated with SGLT-2 inhibitors, with additional studies ongoing [20]. GLP-1 receptor agonists are associated with low risk for hypoglycaemia and significant reductions in body weight [21]. GLP-1 receptor agonists are not generally recommended as a second-line therapy in the UK [22], and their effect on adverse cardiovascular outcomes is inconsistent. While liraglutide has been associated with a reduction in adverse cardiovascular outcomes in patients with T2D at high cardiovascular risk, a neutral effect on cardiovascular outcomes has been observed with other GLP-1 receptor agonists [23‐26]. Of all the glucose-lowering agents, insulin has the greatest HbA1c-lowering potential, although it carries the highest risk of hypoglycaemia and is associated with weight gain [6, 21].

The NICE guidelines and the position statement of the American Diabetes Association and the European Association for the Study of Diabetes recommend that therapies should be reviewed every 3–6 months, and therapy escalated if HbA1c targets are not met. Despite these recommendations, therapeutic inertia is still common, with the average time to initiate second-line therapy in the UK being between 1.6 to 2.9 years after HbA1c levels had reached those recommended for treatment intensification (HbA1c > 7.5% [58 mmol/mol]) [27]. Likewise, in patients from North America, Europe and Israel receiving one oral anti-diabetic drug, second-line treatment intensification occurs a median of 0.3–2.7 years following above-target HbA1c levels [28]. This may carry grave consequences for patients who have poor glycaemic control for extended periods and may result in higher risk of cardiovascular events and other late-stage diabetes complications [29]. On the other hand, patients who receive earlier treatment intensification have greater HbA1c reductions and a higher probability of attaining target HbA1c levels, as well as a lower risk of micro- and macrovascular complications, compared with patients with delayed intensification [2, 3, 30‐32].

Real-world clinical-practice data are an important source of evidence for clinicians considering second-line therapies for the treatment of T2D. This study assessed the clinical outcomes of patients initiated on second-line glucose-lowering therapies, using data from the UK Clinical Practice Research Datalink (CPRD). The impact of these second-line therapies on HbA1c, weight and treatment persistence was evaluated over 18 months in patients who had received first-line metformin monotherapy. The data in this manuscript were accepted as an abstract for - and presented at - the 53rd Annual Meeting of the European Association for the Study of Diabetes [33].

This study used a large, representative population of UK patients to evaluate the use of second-line therapies after metformin monotherapy in a real-world clinical practice setting.

Unlike most clinical trials, which have durations of 6–12 months, this observational study followed patients for 18 months, with 6-month follow-up intervals, and compared the effect of initiation of different second-line therapies on three clinically relevant criteria: changes in HbA1c and weight, and treatment persistence. When these three criteria were evaluated as a composite outcome representative of treatment success at 18 months, more patients (36.5%) on metformin plus an SGLT-2 inhibitor achieved HbA1c reductions of ≥ 0.5% (5.5 mmol/mol), body weight loss ≥ 2 kg and treatment persistence for 18 months, compared with all other therapies. Patients who received combination therapy with metformin plus a DPP-4 inhibitor achieved the second highest composite outcome, with 17.1% of patients achieving treatment success.

In this study, all second-line therapies reduced HbA1c levels, with a mean reduction of 1.2% (13.4 mmol/mol) at 18 months. HbA1c reductions achieved at 6 months following second-line treatment initiation correlated with HbA1c levels at baseline: patients with lower baseline levels experience smaller HbA1c reductions than patients who had higher baseline levels. This observation is in line with previous studies, which report a positive correlation between baseline HbA1c and subsequent changes in HbA1c levels [28, 35]. The greatest reductions in HbA1c at 6 months were observed in patients on metformin plus an SU or on metformin plus an SGLT-2 inhibitor, although at 18 months, these reductions were observed only in patients on metformin plus an SGLT-2 inhibitor. It is known that SUs are highly efficient at rapidly reducing HbA1c levels in patients with T2D, with only insulin acting on HbA1c faster [6]. However, many clinical trials comparing SU to other therapies have shown the effects of SUs on HbA1c to be short-lived [36‐38]. The sustained reduction in HbA1c levels observed in patients receiving metformin plus an SGLT-2 inhibitor has been previously reported in a meta-analysis [39]. However, the higher HbA1c levels at baseline and the reduced number of patients who had HbA1c measurements at 18 months in the CPRD database may have contributed to the effect seen.

The effect of a particular treatment on weight is also an important factor when deciding upon an appropriate second-line agent, as obesity is often associated with T2D [40]. Overall, modest weight gain was seen in patients who received the combination of metformin plus an SU, whereas weight loss was observed for patients treated with metformin plus either a DPP-4 inhibitor or an SGLT-2 inhibitor, with significantly greater weight loss observed in patients receiving an SGLT-2 inhibitor. While DPP-4 inhibitors are generally considered to be weight neutral [6], reductions in BMI following treatment with a DPP-4 inhibitor have been observed in a previous database analysis in Italy [41]. Weight reductions comparable to those reported here following treatment with a DPP-4 inhibitor (mean −1.58 kg at 12 months) were also seen in a previous analysis of a US database, in which patients achieved a mean weight change from baseline of −1.26 kg at 12 months [42]. Previous studies have demonstrated the add-on benefit of SGLT-2 inhibitors by promoting weight loss via increased glucose excretion [43‐46]. The weight reductions seen in this analysis following initiation with an SGLT-2 inhibitor were higher than those observed in clinical trials (mean weight reduction at 18 months of 4.2 kg versus 2.2–3.4 kg) [43] and continued at each time point up to 18 months, which may reflect a greater likelihood of treatment persistence in those who lose weight. Patients who received an SGLT-2 inhibitor-based treatment generally had a higher weight at baseline, suggesting selection of this therapy in more obese patients due to its established effects on weight, consistent with NICE guidelines [4, 47]. Given the demonstrated correlation between baseline weight and weight loss, this may also partly account for the greater weight loss observed with SGLT-2 inhibitors in this analysis. Additionally, the low number of patients who had weight measurements at 18 months in the SGLT-2 inhibitors group could be considered a possible confounder. Perhaps expectedly, no overall change in weight was observed in patients included in the other group, as this group included both patients who received agents that are known to cause weight gain (including insulins and TZDs) and patients who received agents associated with weight loss (including GLP-1 receptor agonists). The weight changes presented in this analysis were adjusted for baseline values to allow comparisons between treatment groups. However, the effect of second-line treatment on weight has been associated with ethnicity, with Asian populations showing a greater response to agents such as DPP-4 inhibitors [48]. Despite information on ethnic groups not being available in the CPRD, the Asian population in the UK represents approximately 7% of the general population [49] and, thus, this population effect should not be considered a potential bias.

A pooled analysis of previous clinical and observational studies reported that treatment persistence in patients with T2D receiving oral anti-diabetic agents ranged from 41.0 to 81.1%, and treatment discontinuation estimates ranged from 9.9% to 60.1% [50]. In the current study, treatment persistence was highest in patients receiving treatment with metformin plus an SGLT-2 inhibitor at all time points, with 57.7% of patients continuing with the index treatment at 18 months. By comparison, 53.2% of patients who received metformin plus either an SU or a DPP-4 inhibitor and 47.2% of patients on other treatments continued with their index therapies, respectively. In line with this, a previous study has shown greater treatment persistence in patients receiving SGLT-2 inhibitors versus SU [51]. The higher rates of treatment persistence seen with an SGLT-2 inhibitor may be associated with the steady decrease in HbA1c levels and weight reductions and reflect increased patient satisfaction with these therapies. Although the reasons for patient discontinuation, switching or intensification of therapy were not reliably recorded in the UK CPRD, and information on adherence to medication and completion of each course of treatment were limited to the available records, in clinical practice common reasons for discontinuation or change of treatment include undesirable side effects (such as weight gain or risk of hypoglycaemia) and inefficacy of treatment requiring intensification or change of therapy [52].

The median time to initiation of second-line treatment did not differ substantially among the different treatment groups investigated, with patients initiating second-line therapy a median of 2.3 years after metformin initiation. Therapeutic inertia remains an issue in clinical practice, with a previous analysis of the UK CPRD database observing a delay of 1.6–2.9 years before second-line treatment intensification in patients with suboptimal glycaemic control (study cohort of 81,573 patients) [27]. Another analysis of the UK CPRD (n = 6710) observed that only 39.5% of patients had their treatment intensified within the year following metformin monotherapy treatment failure (HbA1c ≥ 7% [53 mmol/mol]), and 24% of patients had no evidence of treatment intensification throughout the ~4.3 years of the study [31]. In an analysis of a large US insurance claims database (n = 11,525), which included a cohort of patients with T2D and poor glycaemic control (HbA1c ≥ 8% [64 mmol/mol]), more than half (52%) of the patients did not receive treatment intensification within 12 months of treatment failure [53]. Another study using a US claims database that included patients with T2D identified a mean time to treatment intensification after HbA1c levels were above the target of over 700 days [54]. A recent systematic review that assessed data from patients with T2D from North America, Europe and Israel showed that the median time to treatment intensification after HbA1c levels were above target was 0.3–2.7 years and 1.3–4.9 years, in patients receiving one or two oral anti-diabetic drugs respectively [28].

As an observational study, this analysis had a number of limitations. Firstly, as it aimed to represent a population on-treatment, patients at each time point who discontinued, switched or intensified the index second-line treatment, as well as those who did not have HbA1c levels or body weight data available, were censored from the study. The number of patients evaluated for changes in HbA1c levels and weight were, therefore, smaller at later time points. This is particularly noticeable for the analysis at 18 months, which may limit the generalisability of the results. Secondly, only patients whose first-line therapy was metformin monotherapy were included. Although these patients accounted for the majority of those identified initiating a second-line therapy in the CPRD, consistent with guidelines and clinical practice [4, 6], this patient population may limit the generalisability of these findings to patients who receive a different first-line therapy. Thirdly, consideration should be given to the limited use of some of the more recent glucose-lowering therapies, such as DPP-4 inhibitors and SGLT-2 inhibitors, which have not been used as widely, as is the case for older, more established therapies. Moreover, these more recent therapies were not recommended by NICE guidelines as a treatment choice for treatment intensification after metformin until December 2015, close to the cut-off date for inclusion in this study. Consequently, fewer patients were exposed to these agents, limiting the analyses and interpretations of these results. And fourthly, given the low number of patients identified who received therapy combinations other than metformin plus an SU, a DPP-4 inhibitor or an SGLT-2 inhibitor or received second-line therapy without metformin, these patients were combined into the other treatment group, which prevented reliable analysis of clinical outcomes for each individual treatment strategies.

In addition to the above, observational studies using databases such as the UK CPRD inherently have a number of limitations, including potential issues relating to the quality of the data for key clinical measurements and possible misclassification of exposure and outcome. Hypoglycaemia events were not reliably captured in the UK CPRD, which prevents the assessment of risk of hypoglycaemia for second-line therapy in this analysis. The reports for events such as diabetic ketoacidosis or urinary tract infections are also limited and, therefore, not reported in this analysis. These safety events have been formally evaluated in other post-authorisation safety studies specifically designed to that end, and their incidence evaluated in an independent meta-analysis [55] and in clinical practice [56]. In addition, there is the possibility of confounding by indication, as patients were not randomised to receive the treatments; rather, treatments were selected by the physician based on clinical judgement according to specific patient characteristics. This non-random selection resulted in the differences between the baseline characteristics of the treatment groups. While multivariate regression was undertaken to adjust for the measured differences, the possibility of residual confounding contributing to the findings remains.

In this population-based cohort, all second-line therapies initiated after metformin monotherapy improved glycaemic control, with the greatest HbA1c reductions seen in patients with higher baseline HbA1c levels. In addition, significant reductions in weight were observed for patients treated with metformin plus either a DPP-4 inhibitor or an SGLT-2 inhibitor. The lowest incidence of second-line therapy change was seen in patients treated with metformin plus an SGLT-2 inhibitor. Patients treated with metformin plus an SGLT-2 inhibitor were most likely to achieve a composite of HbA1c reduction ≥ 0.5% (5.5 mmol/mol), weight reduction ≥ 2 kg and continued treatment at 18 months. This highlights the potential benefits of treating patients with more recent agents, including SGLT-2 inhibitors. The ongoing, randomised, pragmatic trial (DECIDE; NCT02616666) aims to evaluate the comparative effectiveness between the SGLT-2 inhibitor dapagliflozin and standard of care over 2 years in patients with T2D, and should provide further data on this therapeutic class [57].