Development of Diabetes With Thiazide Diuretics: The Potassium Issue

There is currently a major debate occurring among hypertension specialists concerning the significance of the development of diabetes following the administration of thiazide diuretics. This debate has been refueled by the findings of the Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial (ALLHAT), ¹ where new‐onset diabetes occurred in 11.6% of people receiving the thiazide‐type diuretic chlorthalidone, compared with 9.8% receiving the calcium channel blocker (CCB) amlodipine, and 8.1% receiving the angiotensin‐converting enzyme (ACE) inhibitor lisinopril. These were statistically significant differences. Other studies have also found absolute differences of 1% and 3% in new‐onset diabetes between ACE inhibitors, CCBs, and diuretics. ²

Some authors have minimized the glucose changes seen in ALLHAT because patients with diabetes, as well as nondiabetics, had significant reductions in cardiovascular (CV) outcomes. ² A recent report by Verdecchia ³ found that hypertensive patients at greater CV risk because of higher blood pressure, more left ventricular hypertrophy, and signs of the metabolic syndrome who developed new‐onset diabetes secondary to a diuretic‐based treatment regimen experienced CV events at rates similar to those who had long‐standing diabetes. The study included only 63 CV disease events. In contrast, however, a 14‐year follow‐up of the Systolic Hypertension in the Elderly Program (SHEP) study ⁴ concluded that diabetes developing during thiazide‐type diuretic therapy did not have significant long‐term association with CV or total mortality. Those who developed diabetes while on chlorthalidone therapy had a better prognosis than people with diabetes on entry. Because of these differences in outcome, many authors urge some caution when using thiazide‐type diuretics, especially in patients who have diabetes or develop diabetes during therapy. What has largely been avoided in this debate is the longestablished relationship between hypokalemia and the development of glucose intolerance. A major question is whether diabetes associated with thiazides could have been ignored in these major trials by maintaining serum K⁺ in the high‐normal range. The first reports of glucose intolerance associated with thiazides appeared in the 1950s and 1960s, shortly after these agents were introduced. ⁵ , ⁶ , ⁷ It was soon noted that there was a relationship between hypokalemia and glucose intolerance. ⁸ , ⁹ Experimentally induced hypokalemia, even in normal subjects, was found to promote glucose intolerance. ¹⁰ It has been known since at least 1964 that glucose intolerance, and the loss of insulin sensitivity following diuretic therapy, could be minimized or prevented with K⁺ treatment. ¹¹ Rapoport and Hurd ¹¹ evaluated 16 patients with either a family history of diabetes or mild diabetes (fasting blood glucose <130 mg/dL) not requiring drug therapy. Seven of 16 patients developed glucose intolerance following diuretic treatment. Baseline K⁺ was 4.7 mEq/L; this decreased to 3.4 mEq/L following thiazide treatment. The hypokalemia was associated with increased glucose intolerance. Replacing K⁺ led to an increase in potassium to 3.9 mEq/L; this change minimized the glucose intolerance. Notably, in the nine patients who did not develop glucose intolerance on the thiazide‐type diuretic, serum K⁺ declined only to 4.2 mEq/L (from 4.7 mEq/L at baseline). Gorden ¹² induced hypokalemia in normal volunteers and found that K⁺ replacement could correct the glucose intolerance that developed in these subjects. Larger studies have shown an association between hypokalemia and glucose intolerance secondary to thiazide diuretics, with the highest fasting blood sugars seen when serum K⁺ was less than 3.9 mEq/L. ¹³ , ¹⁴ , ¹⁵

The exact mechanism by which hypokalemia may cause glucose intolerance is still not known, but it is probably related to defects in insulin release or insulin sensitivity. It is interesting to note that ACE inhibitors and angiotensin II receptor antagonists (ARBs) have the lowest rates of new‐onset diabetes and have been advocated as drugs that improve insulin sensitivity. At the same time, the use of these drugs results in an increase in serum K⁺. Evidence of the importance of hypokalemia in the development of diabetes comes from the Valsartan Antihypertensive Long‐term Use Evaluation (VALUE) study, ¹⁶ designed to evaluate whether for the same level of blood pressure control, a regimen beginning with the ARB valsartan would be more effective than one initiated with the long‐acting dihydropyridine CCB amlodipine in reducing CV morbidity and mortality. In this trial, the incidence of new‐onset diabetes was 3.3% less in the valsartan‐based regimen. As the study was not designed to investigate the mechanism of this finding, we cannot conclude that this finding was the result of a beneficial effect of valsartan, a detrimental effect of amlodipine, or some combination of both. An equal percentage of both arms in the VALUE trial, however, had a thiazide‐type diuretic added to their regimen. As hypokalemia was significantly more common in patients on amlodipine, it appears that the ARB may have ameliorated the K⁺ loss associated with concomitant diuretic therapy. It therefore seems that if hypokalemia is avoided, the occurrence of new‐onset diabetes will be decreased.

Thiazide‐type diuretics are necessary to achieve blood pressure control in the majority of patients with hypertension. ¹⁷ All the clinical hypertension treatment trials that have been diuretic‐based have reported a reduction in CV events comparable to or better than other agents, despite the possible negative effects on glucose metabolism. These agents should remain an important part of our antihypertensive medication arsenal. Low doses (12.5 mg chlorthalidone or 25 mg hydrochlorothiazide) should continue to be recommended as initial therapy, but some patients may require higher doses. ¹⁸ The clinician should routinely measure serum K⁺ and replace even modest losses with a K⁺ supplement, use a K⁺‐sparing agent, or combine a diuretic with an ACE inhibitor or ARB. Because serum K⁺ is not a good indicator of total body K⁺, it may be more appropriate to maintain K⁺ in the high‐normal range from the beginning of treatment, with the target for serum K⁺ at or above 4.0 mEq/L.

Data from studies over the past 40 years suggest that K⁺ replacement should minimize the occurrence of new‐onset diabetes observed with thiazide diuretics. Thiazide‐type diuretics should continue to be used in most patients with hypertension, but special attention should be paid to K⁺ levels.