Mechanistic insights into diuretic-induced insulin resistance.

The incidence of diabetes mellitus and hypertension continues to rise worldwide. The proportion of patients with hypertension at risk for developing diabetes mellitus is also growing secondary to aging and increased obesity rates.1 Several guidelines recommend thiazide diuretics as either first-line or add-on antihypertensive therapy to achieve blood pressure goals.2 Concern over negative metabolic effects associated with thiazide diuretics, however, dates back 3 decades.3 A substantial fraction of patients with hypertension have additional cardiovascular risk factors, and many have elevated fasting glucose and are at risk for developing diabetes mellitus.4 Impaired fasting glucose itself increases the risk for cardiovascular events.5 Any medication that worsens insulin sensitivity, ie, thiazide diuretics or most -blockers will hasten the development of diabetes mellitus in those with impaired fasting glucose.6 Large observational studies demonstrate that thiazide diuretics and most -blockers increase the incidence of new-onset diabetes mellitus compared with renin-angiotensin system (RAS) blockers or calcium channel blockers.7 To further support this observation, a network-based meta-analysis of hypertensive agents showed that RAS blockers were the agents least likely to be associated with the development of diabetes mellitus, whereas thiazides had a higher incidence of diabetes mellitus compared with placebo.7 The mechanism traditionally associated with this increased risk of diuretic-associated diabetes mellitus is a reduction in serum potassium. A meta-analysis of 59 studies involving 83 thiazide diuretic treatment arms found a significant correlation between the degree of diureticinduced hypokalemia and an increase in plasma glucose.8 Moreover, there is evidence that prevention of hypokalemia with K supplementation or potassium-sparing agents lessens the degree to which plasma glucose is increased consequent to diuretic therapy.8 The mechanism of this glucose increase by diuretics may relate to insulin secretion. Mechanisms related to insulin release were reviewed recently, and it was noted that hyperkalemia stimulates insulin secretion and induces cellular uptake of potassium.9 This suggests that low plasma potassium could impair insulin secretion and thereby increase plasma glucose. Ironically, the significant hypokalemia associated with hyperaldosteronism is not associated with hyperglycemia. The presence of insulin resistance and an impaired glucose response to an oral glucose load, however, are reported in hyperaldosteronism.9 Thus, the exact relationship between hypokalemia and worsening of insulin resistance is unclear but appears most pronounced in those with preexisting impaired glucose tolerance and not all people. Given this background, combining an agent that reduces potassium loss, ie, an RAS blocker with a thiazide diuretic, should reduce the risk of new-onset diabetes mellitus. Unfortunately, the Study of Trandolapril/Verapamil SR and Insulin Resistance failed to support this hypothesis. It demonstrated a 4-fold increase in diabetes mellitus at 1 year in comparison with a fixed-dose combination of an RAS blocker with a calcium channel blocker.10 This result was not attributable to differences in serum potassium between groups, because serum potassium values were 4.0 mEq/L in both groups. Thus, mechanisms other than changes in potassium may be operative to worsen glycemic control and are summarized elsewhere.9 One mechanism proposed for the prevention of worsening glycemic control by RAS blockers is their peroxisome proliferator-activated receptorstimulating effects; however, this was not observed in this or any other trial, because candesartan had a neutral effect on glucose.11 Moreover, the peroxisome proliferator-activated receptorstimulating effect observed by some RAS blockers appears relevant only in animal models or at a cellular level.12 The current study by Eriksson et al13 provides a potentially novel mechanism by which diuretics worsen insulin resistance. Twenty-six obese, hypertensive subjects were randomly assigned to candesartan, hydrochlorothiazide (HCTZ), or placebo (in random sequence), each for 12 weeks, using a 3-way crossover design. Insulin sensitivity and secretion, hepatic fat accumulation, inflammatory markers, and the ratio of subcutaneous:visceral abdominal fat were measured. Insulin sensitivity was assessed using a hyperinsulinemic, euglycemic clamp. Significant reductions in insulin sensitivity were present with HCTZ compared with candesartan. Serum potassium levels were within the normal range in all of the groups but 0.3 mEq/L lower among those randomly assigned to HCTZ. The authors reported that differences in potassium level beThe opinions expressed in this editorial are not necessarily those of the editors or of the American Heart Association. From the Pritzker School of Medicine, Department of Medicine, Hypertensive Diseases Unit, Section of Endocrinology, Diabetes, and Metabolism, University of Chicago, Chicago, Ill. Correspondence to George L. Bakris, Hypertensive Diseases Unit, University of Chicago School of Medicine, 5841 S Maryland Ave, MC 1027, Chicago, IL 60637. E-mail [email protected] (Hypertension. 2008;52:1009-1011.) © 2008 American Heart Association, Inc.