Management of Type 2 Diabetes—Reflecting on the PREDICTIVE Study Results

The incidence of diabetes is increasing year on year, approaching pandemic levels. Globally, the disease is estimated to affect 366 million people by 2030, an increase from the approximately 171 million people with diabetes in 2000.¹ In the US, by 2050 the disease is expected to affect 48.3 million people—more than double the figure for 2005.²

The incidence of diabetes is increasing year on year, approaching pandemic levels. Globally, the disease is estimated to affect 366 million people by 2030, an increase from the approximately 171 million people with diabetes in 2000.¹ In the US, by 2050 the disease is expected to affect 48.3 million people—more than double the figure for 2005.²

Currently, type 1 diabetes accounts for approximately 5–10% of all diagnosed cases of diabetes in the US, with type 2 diabetes accounting for the majority of the remaining 90–95%.³ Type 1 diabetes is characterized by the auto-immune destruction of the insulin-producing beta-cells in the pancreas, and usually requires insulin replacement therapy from the time of diagnosis. The pathophysiology of type 2 diabetes is more complex: it is a progressive metabolic disorder in which glycemic control deteriorates over time, combining initial insulin resistance with a progressive deterioration in beta-cell function. Obesity is an important risk factor in the development of type 2 diabetes, and the significant increase in childhood obesity has led to an alarming increase of type 2 diabetes in younger adults, as well as in children and adolescents. By 2050, the subgroup with young-onset diabetes in the US is estimated to more than double, with 5.6 million individuals affected.⁴

Poor glycemic control is associated with major health risks, including cardiovascular disorders, kidney failure, blindness, and lower limb amputations; these risks increase with increasing duration of diabetes. The increase in the number of younger individuals being diagnosed with diabetes is extremely worrying, as it is likely that these serious comorbidities will increase in their future. Adolescents and young people with type 2 diabetes are now already presenting with hypertension and evidence of early nephropathy at the time of their initial diagnosis, and are therefore at greatly increased risk for future premature cardiovascular morbidity and mortality.⁵

The current diabetes epidemic involving increasingly younger individuals therefore represents a major health challenge that will require new treatment strategies to prevent the current rising prevalence of type 2 diabetes shortening the lives of millions.

Current Treatment Guidelines
The American Diabetes Association (ADA) recommends that glycated hemoglobin (HbA_1c) levels be as close as possible to the normal level of <6.0% without risking significant hypoglycemia.⁶ Maintenance of blood glucose levels within acceptable levels protects against long-term diabetic complications, and lower than current recommended glycemic targets substantially decrease the risk for microvascular complications.^6–8

Consensus guidelines from the ADA/European Association for the Study of Diabetes (ADA/EASD) and the American College of Endocrinology/ American Association of Clinical Endocrinologists (ACE/AACE) currently recommend sustaining a long-term HbA1c target level of <7.0 and ≤6.5%, respectively.^6,7 There is, however, increasing acceptance that in individuals with long-standing diabetes and those who already have established cardiovascular disease, aggressive glycemic management can increase the risk for hypoglycemia without established benefit on cardiovascular outcomes.

All current guidelines recommend promoting and supporting lifestyle changes around diet and exercise at the time of the initial diagnosis of type 2 diabetes. For the majority of patients, who will be overweight or obese, dietary advice to help them reduce their bodyweight and limit any future weight gain is an essential first management step. However, the progressive nature of type 2 diabetes inevitably results in oral agents being required early in the management of the disease, with dose escalation and the addition of further oral therapies required over time. Eventually—within six to 10 years of diagnosis—the majority of patients require insulin therapy in addition to their oral agents.
The ideal insulin therapy should mimic endogenous insulin secretion, with meal-related bursts of insulin on a background of low interprandial basal secretion. The introduction of rapid-acting and long-acting insulin analogs to cover the prandial and basal periods have helped to provide a more physiological insulin profile. The use of these analogs has also brought clinical advantages. The rapid-acting insulins have not only a shorter duration of action than conventional insulin formulations, but also a faster onset, allowing the rapid-acting insulin to be administered just prior to the meal. The short duration of action results in fewer hypoglycemic episodes prior to the next meal.^9,10 The basal insulin analogs, with their longer duration of action and reduced peak effects, decrease the risk for nocturnal and between-meal episodes of hypoglycemia compared with conventional insulin formulations. The basal analogs are also more predictable than conventional basal insulin, making titration of the basal insulin dose against the fasting blood glucose safer and easier.^11,12

An insulin regimen using basal insulin only combined with oral agents can offer good glycemic control, provided the patient has sufficient endogenous insulin of his/her own to inhibit gluconeogenesis and limit post-prandial glucose levels. However, due to disease progression the ability of a patent’s endogenous insulin production, even in the presence of an oral insulin secretagog, to adequately control post-prandial glucose levels usually requires the addition of rapid-acting insulin at meal times. The pre-mixed biphasic insulin analog preparations offer a good alternative strategy to separate basal and bolus insulin preparations. However, the pre-mixed biphasic insulin analogs provide less flexibility around meal timing than the basal–bolus regimes. Any given insulin regime needs to be individually tailored to a patient to allow him/her to reach his/her recommended glycemic targets while minimising the risk for hypoglycemia.

Observational Studies versus Randomized Controlled Clinical Trials
The guidelines published by the ADA/EASD are based on controlled clinical trials, as are the ACE/AACE guidelines. To date, the influence of randomized controlled trials (RCTs) compared with observational studies has formed the evidence base for these guidelines, with RCTs ranked as level of evidence (LOE) 1 or 2 depending on size, and observational studies ranked as LOE 3.

In the ranking of evidence, LOE 1 data encompass the results from prospective RCTs involving a large number of subjects who are representative of the targeted patient population with results that are relevant to the rest of the target population, as well as data from multicenter trials, meta-analyses of RCTs, and ‘all-or-none’ evidence.¹³ LOE 2 data comprise conclusive results from RCTs with limited subject numbers or a subject population that is less representative of the target population. LOE 3 evidence includes the clinical findings from nonrandomized studies, studies lacking controls, and non-interventional or observational studies. According to the AACE, LOE 3 evidence necessitates interpretation in the context of other forms of evidence; it does not provide compelling, conclusive evidence when presented on its own. Finally, LOE 4 refers to expert consensus, experience-based opinion, information, or claims that remain scientifically unproven. There is debate around the value of observational studies compared with RCTs. Historically, observational studies have been criticized for overestimating treatment effects, although more recent analysis shows that this is not always the case.^14,15 However the publication of many poorly designed observational studies resulting in incorrect conclusions has left many purists questioning the reliability of any observational research. While most investigators would argue that evidence from RCTs is stronger and less prone to bias than that from non-randomized observational studies, most clinicians would also recognize the clinical importance of prospectively collected data from observational studies; after all, clinical observation forms a fundamental part of good clinical practice, and as such well-designed observational studies can provide clinically relevant data on very large numbers of patients over short periods of time that can complement the data obtained from randomized clinical trials. The evaluation of new treatments benefits from the evidence derived from both RCTs and observational studies, especially as the population of patients included in RCTs is often not highly representative of the population that will be exposed to that drug or treatment postlicensing. ^16,17 Results from well-designed observational studies can provide valuable clinical data that randomized clinical trials are unable to provide. Therefore, when evaluating clinical outcomes, results from both of these methodologies should be considered.
The PREDICTIVE/PREDICTIVE 303 Studies
PREDICTIVE
The Predictable Results and Experience in Diabetes through Intensification and Control to Target: An International Variability Evaluation (PREDICTIVE) study is a large, multinational observational study designed to assess the safety and efficacy of the long-acting insulin analog insulin detemir (Levemir®, Novo Nordisk) in patients with type 1 or type 2 diabetes.

Previous open-label, randomized, parallel-group clinical trials evaluating insulin detemir as part of a basal–bolus regimen showed that this formulation had greater predictability in terms of glycemic control, consistent reductions in hypoglycemia, and less weight gain compared with neutral protamine Hagedorn (NPH) insulin.^18–20 In a study treating insulin-naïve patients with type 2 diabetes who had suboptimal glycemic control on oral antidiabetic drug (OAD) regimens with either insulin detemir or NPH insulin, a similar percentage of patients reached an HbA_1c of 7.0%, but more patients treated with insulin detemir achieved this target without hypoglycemia (26 versus 16%). The risk of total hypoglycemia and nocturnal hypoglycemia with insulin detemir compared with NPH insulin use was reduced by 47 and 55%, respectively. Mean weight gain was also lower with insulin detemir: 1.2kg compared with 2.8kh with NPH insulin.²¹

The PREDICTIVE study enrolled over 30,000 men and women with type 1 or type 2 diabetes globally. These patients were initiated with insulin detemir with or without OADs as part of routine clinical care with the aim of improving glycemic control, and were followed up for 12–52 weeks. Primary study end-points were serious adverse drug reactions, including major hypoglycemic episodes, while secondary end-points evaluated overall and nocturnal hypoglycemia, HbA_1c levels, fasting glucose levels, within-subject fasting glucose variability, and weight change. Serious adverse drug reactions, inclusive of major hypoglycemic episodes, were reported by approximately 1% of patients, although the mean incidence of hypoglycemic episodes was significantly reduced at 14-week follow- up in both type 1 and type 2 diabetes patients.²² Both type 1 and type 2 patients exhibited significant improvements in glycemic control, with reductions in mean HbA_1c (p<0.0001), fasting glucose (p<0.0001), and within-patient fasting glucose variability (p<0.0001) under insulin detemir treatment. A post hoc subgroup analysis of 2,377 insulin-naïve type 2 diabetes patients on OADs followed up for a mean of 14.4 weeks found that individuals taking insulin detemir experienced a small but significant reduction in mean bodyweight by 0.7kg compared with baseline (p<0.0001).²³ This weight loss was even more pronounced for heavier patients: those with a higher body mass index (BMI) experienced a greater weight reduction. The ability to improve glycemic control without inducing weight gain has obvious benefits, as many type 2 diabetes patients are overweight or obese. The lack of weight gain after the initiation of insulin detemir contrasts with other conventional insulin therapy that does lead to weight gain, potentially further increasing the risk for cardiovascular disease.

Subgroup analyses have assessed treatment safety and efficacy in type 2 diabetes patients previously receiving basal therapy with NPH insulin or insulin glargine in combination with OADs and who were subsequently transferred to insulin detemir and followed up over a period of 12–14 weeks.^24,25 Glycemic control improved in both the NPH insulin and insulin glargine groups following transfer to insulin determir, as indicated by reductions in HbA_1c and fasting plasma glucose (FPG) levels. Total and nocturnal hypoglycemic events decreased, and both groups experienced a small but significant reduction in bodyweight. Notably, a three-month subgroup investigation of insulin detemir in German patients with type 2 diabetes treated with OAD who were insulin-naïve or transferred from NPH insulin or insulin glargine found that there was no weight gain associated with insulin detemir, and looking at all patient groups together there was an average weight loss of 0.9kg (p<0.0001).²⁶ The advantages in improving glycemic control and reducing hypoglycemia risk and bodyweight observed in insulin-naïve patients treated with insulin detemir were also observed when patients transferred from other basal insulins to insulin detemir in combination with OADs.

PREDICTIVE 303
The PREDICTIVE 303 study was a prospective, randomized, open-label, multicenter phase IV study in 5,604 type 2 diabetes patients in primary care settings across the US designed to investigate the use of a self-titration dosing algorithm compared with the conventional standard-of-care physician-driven dosing adjustment.²⁶ Patients were randomized to one of two dosing algorithms: either the 303 algorithm, in which patients adjusted their insulin detemir dose every three days depending on their mean FPG values (303 algorithm group), or physician adjustment of insulin detemir dose according to the standard of care (standard-of-care group). According to the 303 algorithm, FPG <80mg/dl would require a dose reduction by three units, FPG between 80 and 110mg/dl would require no change, and FPG >110mg/dl would require an increase by three units. After 26 weeks, mean reduction in HbA_1c from baseline was similar in both the 303 algorithm and the standard-of-care groups, but did not reach optimal levels, at 7.6±1.3 and 7.7±1.3%, respectively. Both groups experienced significant decreases in the rate of overall hypoglycemia, but it is proposed that the slightly higher rate of hypoglycemia in the algorithm group is due to the more aggressive adjustments in insulin dosing. In spite of this, a higher percentage of patients in the 303 algorithm group remained on their once-daily insulin detemir dosing (91%) compared with the standard-of-care group (85%). No significant weight gain was observed in either group. A post hoc sub-analysis of insulin-naïve OAD-treated patients in the PREDICTIVE 303 study demonstrated analogous results, with similar reductions in HbA_1c from baseline, and although both groups achieved significant reductions in FPG from baseline, patients in the 303 algorithm group experienced greater reductions.²⁷ Hypoglycemia rates were similarly low in both groups, and adherence was high, with 95 and 92% of patients in the 303 algorithm and standard-of-care groups, respectively, using determir once daily.
Implications of Study Results
The PREDICTIVE study was designed to assess the safety and efficacy of insulin detemir in a ‘real-world’ setting, beyond the restrictions imposed by an RCT. Like any observational study, however, the PREDICTIVE study has its limitations. Following patients over such a short period suggests that a longer-term follow-up is necessary to ensure that the observed glycemic and metabolic benefits can be maintained. The study design also lacked standardized treatment regimens, pre-defined end-points, and, most importantly, a control group. However, there were also many advantages to the study design: the very large patient population and study performance within medical practices in the PREDICTIVE study offer a realistic representation of diabetes patients and the routine clinical care of such patients. The results obtained from the study suggest that the benefits seen by patients in the clinical trial setting are also achievable in standard clinical practice.

The clinical results of improved glycemic control and reduced hypoglycaemia and weight gain support available data from RCTs with insulin detemir in a basal–OAD regimen in comparison with NPH insulin and insulin glargine.^20,21,28,29 With the PREDICTIVE results paralleling those of prior RCTs, these results are unlikely to be attributed to a study-related effect, which some observational studies have been prone to in the past. Subgroup analyses have also suggested that transferring to a once-daily regimen of insulin detemir from OADs only or another basal insulin regimen can not only improve glycemic control, but also do so with a reduced risk for hypoglycemia and without weight gain—and in some cases with weight loss—such that the efficacy of insulin detemir is apparent in both insulinnaïve and -experienced type 2 diabetes patients. The weight-sparing effect of insulin detemir is especially significant in patients with type 2 diabetes, as such a high proportion of these patients are obese and likely to require large doses of insulin.

Consistent with efficacy and safety data of insulin detemir in previous clinical trials and the PREDICTIVE study, results from PREDICTIVE 303 demonstrated significant reductions in HbA_1c and FPG from baseline in insulin-naïve type 2 diabetes patients receiving insulin detemir, regardless of whether patients applied the 303 algorithm or adhered to physician directions. The inability to reach optimal levels of HbA_1c after 26 weeks may be due to the fact that PREDICTIVE 303 was not an enforced treat-to-target study; other treat-to-target studies with insulin detemir have shown reductions in HbA_1c up to 1.8%.²¹ Patients self-titrating their insulin according to the 303 algorithm appeared to take a more aggressive approach than their standard-of-care counterparts. However, the simplicity and high adherence to the algorithm presents an opportunity to patientsand physicians alike in terms of improving glycemic control with once-daily insulin detemir through active participation in managing the disease. As one of the largest observational studies in diabetes to date, it is hoped that the information provided in the PREDICTIVE/PREDICTIVE 303 studies will be used as a complement to RCT data for use in clinical decisions.

The recent publication of two large-scale clinical intervention studies has also added to our knowledge base of the role of intensive glycemic control strategies in type 2 diabetes and the prevention of cardiovascular events. The Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial³⁰ and the Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) trial³¹ were both randomized studies designed to evaluate tight glycemic control via intensive treatment on vascular outcomes. Both studies tested the respective strategies in type 2 diabetes patients who were considered to be at high risk and both had long-term follow-ups: a median of five years for ACCORD and a median of 3.5 years for ADVANCE. However, neither study proved a mortality benefit of intensive glycemic control and, more importantly, in the ACCORD trial intensive therapy was associated with an increase in cardiovascular deaths and all-cause mortality.

The strategies used in the two trials differed considerably. The intensive regimen cohort in the ACCORD trial had a higher proportion of patients treated with insulin (77%), and over 90% of the patients randomized to the intensive regimen used thiazolidinediones. In comparison, in the ADVANCE trial over 90% of patients on the intensive strategy used a modified-release form of the sulfonylurea—gliclazide—and fewer patients used insulin. Of note, although analyses in the ACCORD study found no association between increased mortality and hypoglycemia, the intensive treatment group in this study reported a much higher rate of hypoglycemic events.³⁰ Current guidelines suggest the adoption of less intensive glycemic goals in patients with severe or frequent hypoglycemia.⁸ Additionally, in the ACCORD trial intensive therapy was associated with weight gain, although it is unclear whether these patients had higher rates of cardiovascular events. While these two trials did not show a positive correlation between intensive glycemic control and reduction in cardiovascular events, the data from the ADVANCE trial support the benefits of tight glycemic control on microvascular complications, with a significant reduction in nephropathy.■