DEVELOPMENT OF NEW‐ONSET DIABETES MELLITUS IS ASSOCIATED WITH AN INCREASED RISK FOR CARDIAC MORBIDITY: THE VALUE TRIAL
In patients with type 2 diabetes mellitus (DM), new‐onset diabetes (NOD) is a frequently reported secondary or post hoc outcome measure in many recent hypertension trials. The prognostic significance of NOD in this clinical setting continues to be hotly debated. The Valsartan Antihypertensive Long‐term Use Evaluation (VALUE) trial provided another opportunity to determine the effect of NOD on cardiovascular (CV) outcomes occurring in initially nondiabetic patients randomized to different antihypertensive therapies.
VALUE was an investigator‐designed, prospective, multicenter, international, double‐blind, randomized, parallel group trial that included 15,245 patients with treated (92%) or untreated hypertension and predefined combinations of CV risk factors. Eligibility was determined by a complex algorithm based on age, sex, CV disease, and CV risk factors. Patients were randomized to either the angiotensin receptor blocker (ARB) valsartan or the calcium channel blocker (CCB) amlodipine as initial antihypertensive therapy, with both randomized groups receiving a thiazide‐type diuretic and then non‐ARB or CCB additional therapy added sequentially as necessary to achieve the goal blood pressure (BP) of <140/90 mm Hg. The VALUE study was end point driven and continued until 1450 patients experienced the primary composite end point, defined as time to the first CV event. Fatal or nonfatal myocardial infarction (MI), fatal or nonfatal heart failure (HF), and fatal or nonfatal stroke were prespecified secondary end points. All‐cause mortality and NOD were other prespecified end points. Patients were followed for 4 to 6 years (average 4.2 years). There was no statistically significant difference in the incidence of the primary composite end point between the treatment allocation groups. The risk of NOD was 23% lower in the valsartan‐based group compared with the amlodipine‐based group.
For the current post hoc analysis, the 15,245 patients were divided into 3 groups: baseline DM, NOD, and no diabetes at any point in the trial (never DM). The diagnosis of diabetes was changed in 1999, and for the final several years of the trial was defined as the presence of at least 1 of the following: (1) use of any antidiabetic treatment, (2) fasting blood glucose of at least 126 mg/dL (1999 World Health Organization [WHO] criteria), or (3) blood glucose of at least 200 mg/dL 2 hours after oral intake of 75 g of glucose. Hazard ratios (HR) were developed for the risk of each prespecified study end point using Cox regression models adjusted for age, diabetes status, left ventricular hypertrophy, coronary heart disease, and randomized study group. The baseline DM and NOD groups were both compared with the never DM group for the various end points. In addition, baseline and in‐trial use of aspirin, β‐blocker, diuretic, combination of β‐blocker and diuretic, statin, and randomized study medications were compared within the 3 groups.
Of the 15,245 original patients, 5250 (34.2%) met criteria for baseline DM. Of the remaining 9995 nondiabetic participants, 1298 (8.5%) developed NOD during the course of the trial. The average incidence of NOD was 3.1% per year, which was higher than that seen in most other BP trials, suggesting a higher‐risk study population. As compared with never DM, NOD and baseline DM patients had a higher baseline mean body mass index (BMI) (27.7, 29.6, and 29.9 kg/m2 for never DM, NOD, and baseline DM, respectively) and glucose values (96.8 mg/dL, 105.8 mg/dL, and 166.5 mg/dL for never DM, NOD, and baseline DM, respectively). Baseline BP was similar between the 3 groups as was the achieved systolic BP (137.8 mm Hg, 138.1 mm Hg, and 139.7 mm Hg for never DM, NOD, and baseline DM, respectively). Over the course of the trial, β‐blocker, diuretic, β‐blocker and diuretic, statin, and amlodipine use was higher for NOD than never DM patients, while the use of valsartan study medication was less in the same comparative groups. Changes in serum lipid values occurring during the trial were not presented.
As compared with never DM, baseline DM was associated with an almost 2‐fold increased risk for both the primary composite end point (HR, 1.98 [1.78–2.19, P<.0001]) and cardiac morbidity, defined as nonfatal MI or HF, (HR, 2.20[1.95–2.49, P<.0001]). In addition, baseline DM was associated with an increased risk for all‐cause mortality (HR, 1.41 [1.28–1.56, P<.0001). As compared with never DM, NOD was not associated with a significant risk for the primary composite end point (HR, 1.10 [0.90–1.33, P=.3447]). It was, however, associated with an increased risk of cardiac morbidity (defined above), which was mainly related to the development of HF (HR, 1.43 [1.16–1.77, P=.0008]). Interestingly, it was also associated with a decreased risk of all‐cause mortality (HR, 0.61 [0.48–0.77, P=.0001]).
The authors conclude that high‐risk hypertensive patients with baseline DM on entry into the VALUE trial have a higher risk of cardiac morbidity and mortality over the 4.2 years of follow‐up than patients who do not go on to develop diabetes. Individuals who develop NOD during the trial have cardiac morbidity (nonfatal MI and HF) intermediate between those with baseline DM and those who never go on to develop diabetes.—Aksnes TA, Kjeldsen SE, Rostrup M, et al. Impact of new‐onset diabetes mellitus on cardiac outcomes in the Valsartan Antihypertensive Long‐term Use Evaluation (VALUE) trial population. Hypertension. 2007;50:467–473.
COMMENT
In multiple observational studies, the presence of type 2 DM confers an increased risk for subsequent CV disease. Available classes of antihypertensive medication appear to have differing effects on glucose metabolism and the subsequent development of NOD. Thiazide‐type diuretics and most conventional β‐blocker therapy increase the risk of developing NOD, while the effect of CCB therapy is neutral. Angiotensin‐converting enzyme (ACE) inhibitor and ARB therapy appear to modestly decrease the risk of NOD. Analysis of trials of ARB and ACE inhibitor therapy, including VALUE, the Losartan Intervention for Endpoint Reduction in Hypertension (LIFE) trial, the Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial (ALLHAT), the Candesartan in Heart Failure: Assessment of Reduction in Mortality and Morbidity (CHARM)‐Preserved, and the Heart Outcomes Prevention Evaluation (HOPE) study all suggest that initiating antihypertensive therapy with renin‐angiotensin system‐blocking agents decreases the risk of NOD. In each of the above‐mentioned trials, however, NOD was investigated as a secondary or post hoc end point. Interestingly, in the one trial designed to prospectively evaluate this effect as a primary end point, the Diabetes Reduction Assessment With Ramipril and Rosiglitazone Medication (DREAM) trial, use of the ACE inhibitor ramipril failed to show a decrease in the risk of NOD over 3.5 years of follow‐up.
The prognostic implication of this apparent differential effect on NOD among various classes of antihypertensive agents remains an important controversy in the management of hypertension. For example, in ALLHAT and the Systolic Hypertension in the Elderly (SHEP) trials, initial therapy with a thiazide‐type diuretic was associated with an increased risk of NOD compared with other classes of antihypertensive medications (ALLHAT) or placebo (SHEP). In these studies, however, the medication induced changes in serum glucose did not appear to negatively impact the risk of CV events. This has led some authors to suggest that the clinical implication of medication‐induced NOD is not the same as “naturally occurring” NOD.
The present post hoc analysis from VALUE provides little additional clarity in helping to resolve this important controversy. While this report does demonstrate that in‐trial thiazide‐type diuretic and β‐blocker use was greater in the NOD group, in VALUE, thiazide‐type diuretic use was not randomly assigned and was similarly used in both the valsartan and amlodipine group, making causation difficult to interpret. In addition, it does not appear that all baseline variables were appropriately corrected for in this analysis. Patients who developed NOD in VALUE had higher BMI and fasting glucose values on entry. While baseline BMI was adjusted for in the regression analysis, it is not clear whether baseline fasting glucose was; the methods section states only that “diabetes status” was included in the analysis.
Of similar importance, baseline lipid values are not reported or adjusted for in the analysis. While the BP was only modestly higher in the NOD group, mean fasting glucose in patients who developed NOD was in the impaired fasting glucose range on entry into the trial. In addition, BMI in those who developed NOD was similar to those with baseline DM on entry. It is therefore likely that baseline lipid values including high‐density lipoprotein cholesterol and triglycerides, which are associated with the metabolic syndrome, may have differed at baseline but were not measured in the trial. Multiple studies have previously demonstrated that patients with impaired fasting glucose and the metabolic syndrome have a risk of CV events that is intermediate between those with established type 2 DM and those with normal fasting sugars. Accordingly, from the present VALUE post hoc analysis it is difficult to determine whether the increased risk of cardiac morbidity seen with NOD is due to baseline differences in metabolic risk factors or to changes in glucose metabolism seen with antihypertensive therapy.
The authors of this report have focused on a single end point, cardiac morbidity, which was the only end point that was significantly increased with NOD. While reported in the original results of the VALUE trial, cardiac morbidity, grouped here as first incident nonfatal MI or HF, was not originally a specific predefined primary or secondary end point. The primary outcome of the present post hoc analysis, which was a composite of fatal and nonfatal CV events, did not show a significant increase in the NOD group. Of the prespecified secondary end points (fatal and nonfatal MI, HF, and stroke), only HF was seen significantly more often with NOD (although MI nearly reached statistical significance). Since there was essentially no difference in stroke rates between the NOD and never DM groups, if stroke had been included in the cardiac morbidity end point, it is not likely to have reached statistical significance. In fact, another prespecified end point, cardiac mortality, was actually less frequent in the NOD group. While the authors try to explain this finding by suggesting either a survivor effect (ie, patients who died earlier in the study may not have lived long enough to develop diabetes) or that patients with NOD were treated more aggressively with other risk reduction strategies, including greater use of aspirin, β‐blocker, diuretic, β‐blocker and diuretic, and statin‐based therapy, given the post hoc nature of this analysis, these explanations should be considered conjecture.
Given the limitations of the data described above, this report from the VALUE investigators offers little additional information in helping the clinician choose initial antihypertensive therapy in a patient at risk for the development of type 2 DM. It and a recent report from the Ongoing Telmisartan Alone and in Combination With Ramipril Global Endpoint Trial (ONTARGET) both suggest that NOD is associated with a greater risk for the development of clinical HF. Once a patient develops DM, the currently recommended BP goal is <130/80 mm Hg, which was not achieved in either ONTARGET or VALUE. Achieving effective BP control is especially important for preventing clinical HF.
Achieving effective BP control continues to remain more important than the choice of any particular antihypertensive agent. In those with a compelling indication, specific antihypertensive agents should be used. While we await the results of future clinical trials, in patients currently at high risk for the development of NOD, including those with impaired fasting glucose, it appears reasonable to avoid thiazide‐type diuretics and traditional β‐blockers in the minority of patients whose BP can be controlled on monotherapy. In these patients, use of ACE inhibitor and ARB therapy may have an advantage in decreasing the subsequent risk for NOD. When BP is not effectively controlled, however, addition of a thiazide‐type diuretic is clearly indicated since the benefits of achieving BP control appear to outweigh any negative effects on glucose metabolism. Further clarification of the comparative value of antihypertensive combinations comparing use of a glucose neutral agent such as a CCB with a thiazide‐type diuretic will come from the Avoiding Cardiovascular Events Through Combination Therapy in Patients Living With Systolic Hypertension (ACCOMPLISH) trial. In this trial, high‐risk hypertensive patients are initially randomized to a fixed‐dose combination of either an ACE inhibitor/thiazide‐type diuretic or an ACE inhibitor/CCB. For now, lifestyle measures including weight loss and exercise as well as the importance of addressing global CV risk through the use of statin and aspirin therapy may be a more effective strategy to improve vascular health and limit insulin resistance in patients with hypertension than restricting the use of any particular antihypertensive agent, regardless of their small effect on plasma glucose.
Disclosure:
Drs Bloch and Basile both report receiving research support, consulting fees, and honoraria from Novartis Pharmaceuticals, maker of valsartan.