In recent years, there has been increasing enthusiasm for investigating pharmaceutical strategies to slow the progression from prediabetes (impaired fasting glucose, impaired glucose tolerance [IGT], or elevated glycated hemoglobin (HbA1c) [5.7–6.4%]) to diabetes in hopes that this reduction in progression of dysglycemia would lead to a corresponding reduction in cardiovascular (CV) events. Previous data on use of agents that block the renin-angiotensin system (RAS) have provided conflicting results. Published as a separate paper, because there was no interaction between the two parallel arms of the trial, the Effect of Valsartan on the Incidence of Diabetes and Cardiovascular Events was part of The Nateglinide and Valsartan in Impaired Glucose Tolerance Outcomes Research (NAVIGATOR) trial, designed to prospectively determine whether the angiotensin receptor blocker (ARB) valsartan would reduce the risk of developing diabetes and CV events in patients with IGT.

NAVIGATOR was a prospective, double-blind, international, randomized clinical trial that assigned 9306 patients with IGT in 2×2 randomization to either valsartan or matching placebo and the short-acting oral hypoglycemic nateglinide or matching placebo. Performed in 806 centers in 40 countries, randomization of patients occurred during 2002–2004. The study was sponsored by Novartis, who helped design the study in collaboration with an academic executive committee, and it was monitored by an independent safety committee. In order to be included in NAVIGATOR, potential participants had to have a screening plasma glucose level of 95 mg/dL to 125 mg/dL and a subsequent glucose tolerance test demonstrating a serum glucose of 140 mg/dL to 199 mg/dL 2 hours after a 75-g oral glucose load. Additionally, each patient had to be either at least 55 years of age with at least 1 known CV risk factor or at least 50 years of age with established CV disease. Exclusion criteria were laboratory abnormalities or conditions that could interfere with assessment of the safety or efficacy of a study drug, the use of an angiotensin-converting enzyme (ACE) inhibitor or ARB for the treatment of hypertension (although ACE inhibitors were allowed for other indications), and the use of an antidiabetic medication within the past 5 years. After randomization, valsartan or matching placebo were started at a dose of 80 mg once daily, with an increase after 2 weeks to 160 mg once daily for the remainder of the trial. In addition, all patients were required to participate in a lifestyle intervention program designed to achieve and maintain a 5% weight loss, decrease intake of fat, and increase in physical activity to at least 150 min/wk. After the dose-adjustment phase, patients were seen every 6 months for the first 3 years and then annually until the end of the study. Oral glucose tolerance tests were performed annually.

Initially, there were 2 prespecified co-primary outcomes: the incidence of new-onset diabetes and an extended CV outcome (a composite of death from CV causes, nonfatal myocardial infarction [MI], nonfatal stroke, hospitalization for heart failure, arterial revascularization, or hospitalization for unstable angina). A third co-primary core CV outcome (a composite of death from CV causes, nonfatal MI, nonfatal stroke, and hospitalization for heart failure), which was initially a secondary composite outcome, was added prior to study completion. The difference between the core composite CV outcome and the extended composite CV outcome is that the core outcome did not include unstable angina or revascularization. New-onset diabetes was defined as a fasting plasma glucose of at least 126 mg/dL or a plasma glucose level of at least 200 mg/dL measured 2 hours after an oral glucose load and confirmed within 12 weeks by a glucose tolerance test.

The study was planned to continue until 1774 patients had experienced an extended CV outcome event. Initially, the investigators anticipated a power of 90% to detect a 20% reduction in CV events with valsartan as compared with placebo. However, based on other clinical trial data that became available during the course of the trial, these calculations were revised to assume a difference in event rates of only 12% for the extended CV end point and 18% for the core CV end point. Based on these revised statistics, the study was subsequently assumed to have only a 64% and 74% chance of detecting a statistically significant difference in the extended and core CV end points, respectively. Based on the larger number of anticipated cases of new-onset diabetes, the study was expected to have a power of greater than 99% to detect an 18% difference in the risk of developing diabetes with valsartan as compared with placebo.

Of 43,502 patients who underwent screening, 9518 were randomized to valsartan or placebo. Subsequently, 212 were excluded from further analysis when 10 sites were closed post-randomization for regulatory deficiencies. This left 4631 patients in the valsartan group and 4675 in the placebo group. Baseline characteristics of the 2 groups were similar. The average age of the study cohort was 64 years, 51% were women, and 83% were white. Baseline body mass index adjusted for sex was 31. Nearly 25% of patients had a history of known CV disease (mostly previous MI). Mean estimated glomerular filtration rate was 81 mL/min/1.73m2 with a baseline creatinine of 0.9 mg/dL. On entry, statin use was 34% in both arms and increased to 47% and 49% by the end of the trial in the valsartan and placebo groups, respectively. In addition, 37% of patients in both arms were taking aspirin or another antiplatelet agent, which increased to 46% by the end of the trial.

A history of hypertension was present in 77% of patients, with 75% taking ≥1 antihypertensive medication with a mean baseline blood pressure (BP) of 140/83 mm Hg. The most commonly used antihypertensive medications used at baseline were β-blockers, diuretics, and calcium channel blockers. During the course of the trial, use of these antihypertensive medications became unbalanced, with a statistically significant greater use of β-blockers (42.8% vs 39.7%), diuretics (39.4% vs 34.1%), calcium channel blockers (39.7% vs 33.2%), and ACE inhibitors or ARBs (24.0% vs 20.4%) reported in the placebo group compared with the valsartan group at the end of the trial. At 1 year, 77.6% of patients in the valsartan group were still taking study medication, mostly at a dose of 160 mg daily; after 3 and 5 years, the number of patients still taking valsartan had declined to 72.3% and 66.7%, respectively. BP levels decreased more in the valsartan group than in the placebo group, with a mean (±standard deviation) overall reduction in systolic pressure of 6.3±14.2 mm Hg in the valsartan group as compared with a reduction of 3.8±13.8 mm Hg in the placebo group (between-group difference, 2.8 mm Hg; 95% confidence interval [CI], 2.4–3.2; P<.001) with adjustment for region, CV history, and nateglinide treatment. The mean reduction in diastolic pressure was 4.4±8.4 mm Hg in the valsartan group, as compared with a reduction of 3.0±8.1 mm Hg in the placebo group (difference, 1.4 mm Hg; 95% CI, 1.2–1.7; P<.001). Discontinuation rates (34%) and the incidence of side effects were similar between valsartan and placebo. Weight and waist circumference showed only modest changes from baseline, with a modestly greater effect of weight loss in the placebo group (difference of 0.28 kg; 95% CI, 0.12–0.44; P<.001).

For the primary end points, after 5 years, the cumulative incidence of diabetes was 14% lower with valsartan than with placebo: 33.1% in the valsartan group as compared with 36.8% in the placebo group (hazard ratio [HR] in the valsartan group, 0.86; 95% confidence interval [CI], 0.80–0.92; P<.001). The beneficial effect of valsartan on diabetes incidence was consistent across all prespecified subgroups. Additionally, mean fasting plasma and 2-hour post-load glucose was modestly lower in the valsartan group. However, after mean follow-up of more than 6 years, valsartan, as compared with placebo, did not significantly reduce the incidence of either the extended CV outcome (14.5% vs 14.8%; HR, 0.96; 95% CI, 0.86–1.07; P=.43) or the core CV outcome (8.1% vs 8.1%; HR, 0.99; 95% CI, 0.86–1.14; P=.85). The neutral effect on outcomes was consistent for both CV outcomes and across all prespecified subgroups.

Among patients at least 50 years of age with IGT and either established CV disease or at least 1 of its risk factors, the use of valsartan for a mean of 5 years led to a 14% reduction in the risk of developing diabetes, but did not reduce the rate of CV events as compared with placebo.—The NAVIGATOR Study Group. Effect of valsartan on the incidence of diabetes and cardiovascular events. N Engl J Med. 2010;362:1477–1490.


  1. Top of page
  2. Comment

A recent meta-analysis of post hoc and/or subgroup analysis of previous clinical trials found that use of agents that block the RAS, specifically ACE inhibitor and ARB therapy, decrease the incidence of new-onset diabetes in patients with prediabetes by 27% and 23%, respectively. Yet, the only previous study that prospectively analyzed the rate of diabetes development as a primary end point with RAS blockade, the Diabetes Reduction Assessment With Ramipril and Rosiglitazone Medication (DREAM) study, did not find a significant reduction in new-onset diabetes using up to 15 mg of the ACE inhibitor ramipril. In the present study, NAVIGATOR, use of the ARB valsartan led to a significant 14% reduction in new-onset diabetes, a prespecified co-primary end point. However, before we conclude that ARBs are effective and ACE inhibitors are not effective in reducing progression to diabetes, it is important that we compare the studies more carefully and evaluate the efficacy of valsartan compared with other strategies to prevent incident diabetes.

While making such comparisons can be hazardous, it is important to note that DREAM was a smaller (5269 vs 9306 patients) and shorter (3 vs 5 years) trial. In addition, DREAM included lower-risk patients, more than half of whom did not have hypertension and none of whom had underlying CV disease. As such, DREAM was likely underpowered to detect the modest effect on diabetes incidence seen in NAVIGATOR, an effect that was more modest than originally projected. In fact, the overall magnitude of the statistically significant effect on progression to diabetes seen in NAVIGATOR of 14% is not terribly different from the nonstatistically significant trend of a 9% reduction seen in DREAM. The future meta-analysis to include NAVIGATOR and DREAM will likely demonstrate that ACE inhibitor and ARB use does decrease the incidence of new-onset diabetes, but that the effect is quite modest. It is even possible that NAVIGATOR may have overestimated the magnitude of benefit since some of the increased risk of progression to diabetes seen with placebo may have been due to the greater use of β-blocker and diuretic treatment among those assigned to placebo during the course of the trial. Based on the results of NAVIGATOR, one would need to treat approximately 26 patients with IGT for 5 years to prevent 1 new case of type 2 diabetes, an effect that is much more modest than that seen with other interventions also found to prevent progression to diabetes such as lifestyle modification (58%) in the Diabetes Prevention Program (DPP), metformin (31%) in the DPP, thiazolidinedione (TZD) use in the Troglitzaone in Prevention of Diabetes (TRIPOD) study (55%), and rosiglitazone in the DREAM trial (60%).

Given the fact that diabetes is a potent CV risk factor, it has been hypothesized that if RAS-blocking agents can decrease the risk of progression to diabetes, we will see a corresponding reduction in the rate of CV events. However, in NAVIGATOR, despite a small difference in achieved BP between the study groups (2.8/1.4 mm Hg), no beneficial effect on CV events was evident with valsartan (nor on death with ramipril in DREAM). Subgroup analysis provided in the appendix to the study published online suggested a trend toward benefit in patients with a history of hypertension but no obvious difference between those with and without adequate BP control. Given the initial presumed 20% difference in event rates between groups and the higher-than-expected rate of study medicine discontinuation (34%), loss to follow-up (13%), and ACE and ARB use (24%) in the placebo group, NAVIGATOR may have been underpowered to detect a beneficial effect on CV events. Yet, if this null result was merely a case of the study being underpowered, one would have expected at least a trend toward fewer CV events with valsartan, which was not seen.

There are a number of intriguing potential explanations for the null effect of valsartan on CV events. First, 6 years of follow-up may not be long enough to see a vascular benefit due to less progression from IGT to frank diabetes. Second, the mechanism of action by which RAS-blocking agents decrease progression to diabetes may be independent of insulin sensitivity, which is presumed to be the primary means by which diabetes and vascular pathology interact. Third, increased use of unblinded RAS blockers, β-blockers, diuretics, and calcium channel blockers in the placebo group may have masked any benefit of valsartan over no other treatment. And, fourth, it may be that in a population of patients with IGT or early type 2 diabetes, aggressive BP control by other means, wide-spread statin use (which may have a modestly unfavorable effect on glucose control while favorably effecting overall CV risk), antiplatelet therapy, and early aggressive treatment of hyperglycemia may positively effect vascular physiology to the point that it is very difficult to achieve further reductions in events just by adding one medication and only incrementally and modestly affecting CV risk factors such as BP and glycemic control.

When faced with a patient with prediabetes, defined by the American Diabetes Association (ADA) as a fasting glucose 100 mg/dL to 125 mg/dL, a 2-hour post-load glucose of 140 mg/dL to 199 mg/dL, or an HbA1c of 5.7% to 6.4%, how should these results from NAVIGATOR affect routine clinical care? In patients with prediabetes whose BP is not adequately controlled, who are not yet taking a RAS-blocking agent, and who do not have any contraindication to their use, the next logical step for antihypertensive therapy should be an ACE inhibitor or an ARB in addition to an intensification of therapeutic lifestyle change. Addition of one of these agents in this setting should help control BP and likely have the added benefit of modestly reducing the risk of progressing to type 2 diabetes. However, if the patient with prediabetes already has adequate BP control, the modest effect on diabetes incidence and lack of benefit on CV events seen in NAVIGATOR (and development of diabetes or death in DREAM) do not support adding a RAS-blocking agent. The mainstay of therapy in these patients should be aggressive lifestyle modification, including weight loss and exercise, as well as maintenance of good BP control (and control of other CV risk factors). If drug therapy to reduce the progression to diabetes is considered for such a patient with prediabetes and well-controlled BP, it appears to make more sense to use a medication such as metformin (not approved by the US Food and Drug Administration [FDA] for this indication) in order to prevent diabetes than to add an RAS-blocking agent (regardless of a future FDA approval) based on currently available clinical trial evidence.

Disclosures: Drs Bloch and Basile wish to disclose that they have received research support and consulting fees and serve on the speakers’ bureau for Novartis Pharma, the sponsor of the NAVIGATOR study.