Prevention of diabetes

Diabetes affects one in 20 adults worldwide and 333 million cases are projected worldwide by 2025.¹ Treatment can prevent some of the microvascular and macrovascular complications, but diagnosis is often delayed until complications present,² so attention has focused on prevention and early screening. Two strategies currently exist for reducing the onset of diabetes—lifestyle interventions and drugs.

The Diabetes Prevention Program Research Group study found that lifestyle interventions delivered over 2.8 years reduced the incidence of diabetes by 58%.³ A similar reduction in risk was found in a Finnish study of 522 people at risk.⁴ The problem is that these interventions are labour intensive—one study needed 16 one to one sessions delivered by case managers to achieve target weight reduction and exercise levels.³ Although lifestyle interventions produce successful results in research settings, they are difficult to replicate even in well funded healthcare systems.

Considerable interest has focused on the prevention of diabetes with drugs. For instance, the Diabetes Prevention Program Research Group study found a 31% reduction in the incidence of diabetes with metformin at 2.8 years.³ Previously troglitazone was shown to be effective in controlling blood sugar levels but was removed from the market because of serious liver toxicity.⁵ In people with obesity orlistat has been shown to reduce the risk of diabetes by 37% when compared with placebo.⁶

More recently came the publication of the diabetes reduction assessment with ramipril and rosiglitazone medication (DREAM) trial.⁷,⁸ In this trial, which cost $25m (£13m; €20m), 5269 people over 30 years with impaired fasting glucose or impaired glucose tolerance, or both, and no previous cardiovascular disease were randomised to receive either rosiglitazone 8 mg daily or placebo, or ramipril 15 mg daily or placebo. The primary outcome was a composite of incidence of diabetes or death over a three year median follow-up period.

The trial was well executed; randomisation produced equivalent groups via a concealed computer telephone group. Although the drop out rate was high at around 25%, the analysis was on an intention to treat basis. At the end of the study 306 (11.6%) of the patients taking rosiglitazone developed diabetes compared with 686 (26%) of those given placebo (hazard ratio 0.40, 95% confidence interval 0.35 to 0.46, P < 0.0001). Ramipril did not reduce the risk of diabetes.

These results are promising, but they should be interpreted with caution. The mean fasting plasma concentration of glucose in both groups at baseline was 5.8 mmol/l, whereas the two hour impaired glucose tolerance test had a value of 8.7 mmol/l. The study population was therefore composed predominantly of people with impaired glucose tolerance rather than those with abnormal fasting glucose. Fasting glucose concentrations rather than impaired glucose tolerance are usually used to screen for diabetes in the United Kingdom. Secondly, the rationale for using a composite end point of death and diabetes is unclear. Several considerations should be taken into account when using a composite end point.⁹ These include whether the component outcomes carry similar weight of importance to patients; and whether the component outcomes are likely to have similar relative risk reductions. This is not the case for death rates, which were similar in both groups and therefore should be analysed separately. Furthermore, despite the population being at low risk of heart failure (10 year risk 0.33%) a significant increase (0.4%) in heart failure was seen in the rosiglitazone group compared with placebo (7.03, 1.60 to 30.9, number needed to harm at three years 250). The use of thiazolidinediones is increasingly recognised as being associated with fluid retention and heart failure, and this is more common when they are combined with insulin.⁵ Also the drug showed no clear benefit on patient relevant outcomes at three years—the rate of all cardiovascular events was higher in the intervention group (1.37, 0.97 to 1.94, P = 0.08).

A key question that remains is whether rosiglitazone prevents the onset of type II diabetes or merely lowers blood sugar concentrations in patients with new onset diabetes. As in the metformin trial in 2002, this can only be resolved by analysis after a washout period, which is promised later in the year.

The finding that rosiglitazone can prevent diabetes in people at risk of getting diabetes needs to be tempered with possible adverse effects of the drug, such as heart failure, and perhaps the risk of medicalising a lifestyle issue. Providing drugs is expensive, even without the additional costs of monitoring their side effects and treating them. Lifestyle interventions seem to work but they are difficult to replicate. What we still need are pragmatic lifestyle interventions that can delay the onset of diabetes. We need to decide whether we want to spend more on drugs for prevention rather than on lifestyle measures and public health strategies to reduce the burden of chronic disease.