November 16, 2006 — Fasting glucose (FG) levels increase in older adults with hypertension regardless of whether they are treated with chlorthalidone, amlodipine, or lisinopril, according to the results of an analysis from the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) reported in the November 13 issue of the Archives of Internal Medicine. Although the risk of developing FG levels higher than 125 mg/dL was modestly greater with chlorthalidone, there was no conclusive or consistent evidence that this diuretic-associated increase in risk for diabetes increased the risk for clinical events.

"Many patients with hypertension eventually develop diabetes," Robert Phillips, MD, author of an editorial accompanying the article, and director of the Heart and Vascular Center of Excellence at the University of Massachusetts Memorial Medical Center in Worcester, told Medscape. "Physicians and others interested in the individual and public health are guided by the principle of primum non nocere — first, do no harm. The initial reports from the ALLHAT trial showed that there was more diabetes associated with thiazide-diuretic treatment, and the authors sought to determine if this was harmful."

The objective of this analysis was to compare the effect of first-step antihypertensive drug therapy with thiazide-type diuretic, calcium-channel blocker (CCB), or angiotensin-converting enzyme (ACE) inhibitor on FG levels and to determine the risks for cardiovascular and renal disease associated with elevated FG levels and incident diabetes mellitus in these 3 treatment groups. The investigators performed post hoc subgroup analyses from ALLHAT among nondiabetic participants who were randomized to treatment with chlorthalidone (n = 8419), amlodipine (n = 4958), or lisinopril (n = 5034). Mean follow-up was 4.9 years.

"Mean fasting glucose levels increased during ALLHAT follow-up in all treatment groups," corresponding author Barry R. Davis, MD, PhD, of the University of Texas School of Public Health, Coordinating Center for Clinical Trials in Houston, told Medscape. "There was no significant association of fasting glucose level change at 2 years with subsequent coronary heart disease [CHD], stroke, cardiovascular disease, total mortality, or end-stage renal disease."

At year 2, the chlorthalidone group had the greatest increase in mean FG levels (+8.5 mg/dL [0.47 mmol/L] vs +5.5 mg/dL [0.31 mmol/L] for amlodipine and +3.5 mg/dL [0.19 mmol/L] for lisinopril). The odds ratios for developing diabetes at 2 years were 0.55 with lisinopril vs chlorthalidone (95% confidence interval [CI], 0.43 - 0.70), and 0.73 with amlodipine vs chlorthalidone (95% CI, 0.58 - 0.91; P <.01). Incident diabetes at 2 years was not significantly associated with clinical outcomes, except for CHD (risk ratio [RR], 1.64; P = .006), but RR was lower and nonsignificant in the chlorthalidone group (1.46; P = .14).

"The diuretic did not appear to be the culprit responsible for the increased CHD risk associated with new-onset DM [diabetes mellitus]," Dr. Phillips says. "Rather surprisingly, development of DM during treatment with ACE inhibitors was associated with an increased risk of CHD and congestive heart failure, and development of DM during treatment with CCB was associated with increased mortality. We need to be cautious about the clinical implications, but I do believe that they suggest that diuretic-induced hyperglycemia is relatively benign. Conversely, if diabetes develops on an ACE or CCB, it suggests that that patient has developed insulin resistance despite being on drugs that tend to improve insulin sensitivity, and hence the consequences are more dire."

Strengths of this study, according to Drs. Davis and Phillips, are the large sample size providing much greater statistical power to recognize associations and differences between medications and the relatively long duration of follow-up. Limitations are the retrospective design, and follow-up possibly insufficient to detect the adverse effects of thiazide-induced hyperglycemia.

"Considerable attention was paid to quality assurance, and all laboratory tests were done in a certified central laboratory," Dr. Davis says. "Limitations of the study are that no FG measurements during follow-up were available for nearly half of the cohort, owing primarily to participants not fasting prior to venipuncture. Also, we did not obtain measures of overall glucose control, such as glycosylated hemoglobin, and we did not obtain information on treatment of diabetes."

Dr. Phillips also notes that because of its design, ALLHAT did not study the typical types of combinations of hypertensive medications that patients frequently take, limiting generalizability of the results. However, the International Verapamil-Trandolapril Study showed that an atenolol-hydrochlorothiazide based strategy was as effective as the verapamil-trandolapril strategy in preventing death and cardiovascular outcomes in this high-risk population, even though there was more incident diabetes associated with the former regimen.

"I think that [the ALLHAT analysis] will help physicians to feel more comfortable about using low-dose thiazide therapy as either the initial drug choice as or part of combination therapy," Dr, Phillips said. "If hyperglycemia or frank diabetes does develop, many physicians would stop the diuretic and substitute another agent to see if the hyperglycemia would resolve. One can't argue with that approach, but actually we do not know if it is the correct action to take."

"These results should reinforce the recommendations of the Joint National Committee on Hypertension that diuretics should be considered as a first-line treatment of hypertension and should be part of any multidrug regimen for hypertension," Dr. Davis concluded.

This study was supported by the National Heart, Lung, and Blood Institute. The ALLHAT investigators disclosed receiving contributions of study medications from Pfizer (amlodipine and doxazosin), AstraZeneca (atenolol and lisinopril), and Bristol-Myers Squibb (pravastatin), and financial support from Pfizer. Some of the authors have disclosed various financial relationships with Takeda, GlaxoSmithKline, Merck, BioMarin, Proctor & Gamble, Bristol-Myers Squibb, Pfizer, Abbott Laboratories, AstraZeneca, Boehringer Ingelheim, Forest Pharmaceuticals, Novartis, Reddy, Sankyo, SmithKline Beechman/Glaxo Wellcome, Bertek, DepoMed, Medco, and/or Wyeth. Dr. Phillips has disclosed no relevant financial relationships.

Arch Intern Med. 2006;2174-2176, 2191-2201.

Clinical Context

Study Highlights

• Patients eligible for study participation were at least 55 years old and either had blood pressure values of at least 140/90 mm Hg or were receiving treatment of hypertension with fewer than 3 medications. Subjects also had at least 1 risk factor for CHD and a baseline serum glucose level less than 110 mg/dL.
• Participants were randomized to receive treatment with chlorthalidone, amlodipine, or lisinopril to reduce blood pressure to less than 140/90 mm Hg.
• Serum glucose was measured at baseline and at years 2, 4, and 6. The main study outcome was the relationship between antihypertensive assignment, serum glucose at follow-up, and the incidence of diabetes (defined by serum glucose > 125 mg/dL).
• 18,411 patients in the ALLHAT trial did not have diabetes at baseline, and 53.2% of these subjects had FG levels examined at subsequent visits. The mean age of participants was 66 years, and the mean serum FG level was 93 mg/dL among all treatment groups.
• The mean follow-up time was 4.9 years. Rates of maintenance of blinded medication use at year 4 were 76.3%, 84.7%, and 83.2% among the lisinopril, amlodipine, and chlorthalidone groups, respectively.
• Glucose levels rose in all treatment groups with time. During the first 2 years, the mean increase is serum glucose levels was 8.5, 5.5, and 3.5 mg/dL among the chlorthalidone, amlodipine, and lisinopril groups, respectively. The difference between chlorthalidone and the other antihypertensive drugs in this outcome was significant and remained relatively stable at 4 years.
• Compared with the chlorthalidone group, the odds ratios for developing diabetes were 0.55 and 0.73 in the lisinopril and amlodipine groups, respectively.
• Despite the increase in serum glucose and risk for diabetes linked with chlorthalidone therapy, subjects with these outcomes did not have a significantly increased risk for CHD or mortality vs subjects receiving chlorthalidone whose glucose remained stable during study therapy. Conversely, incident diabetes increased the risk for CHD and heart failure among subjects receiving lisinopril. The overall risk for CHD was also increased among all subjects who developed incident diabetes.

Pearls for Practice

• High-dose thiazide diuretics can promote sudden death among patients with hypertension and electrocardiogram abnormalities, but they can also safely reduce blood pressure and left ventricular mass at lower dosages among patients with hypertension.
• In the current analysis of ALLHAT, chlorthalidone was associated with a more significant increase in serum glucose level and incident diabetes than amlodipine and lisinopril. However, incident diabetes among subjects receiving chlorthalidone did not increase the risk for CHD or mortality.

CME Test