European Heart Journal (1992) 13 {Supplement G), 68-71
Diabetes, impaired glucose tolerance and insulin resistance with diuretics L. E. RAMSAY, W. W. YEO AND P. R. JACKSON
Royal Hallamshire Hospital, Sheffield, U.K. KEY WORDS: Hypertension, diuretics, glucose tolerance, insulin resistance. There is a definite relation of diuretic treatment to impaired glucose tolerance and biochemical diabetes, and a probable relation to insulin resistance. The effect of diuretics on glucose tolerance is dose-related. Spironolactone does not impair glucose tolerance even at high dosage, but apparent differences between other diuretics may well be due to comparison at doses which are not equivalent. Diuretic-induced changes in carbohydrate metabolism are not conclusively related to altered potassium homeostasis, and impaired glucose tolerance occurs when relatively low doses ofthiazide are combined with potassium-sparing agents. The effect of diuretics on glucose tolerance is largely and possibly wholly reversible. These disturbances of carbohydrate homeostasis have been detected by detailed biochemical testing, and their clinical importance is uncertain. In established diabetes, diuretics have a rapid and substantial adverse effect on metabolic control. In non-diabetic subjects diuretics may rarely cause or trigger a serious hyperosmolar non-ketotic diabetic syndrome. This apart, it is not known whether the metabolic changes cause clinical diabetes or lead to microvascular complications in the long-term. It is now established that biochemical diabetes, glucose intolerance and insulin resistance probably do not increase the risk of coronary heart disease in treated hypertensive patients. Diuretics should be avoided in patients with diabetes, otherwise they remain an excellent choice for first-line antihypertensive therapy. Introduction
When considering the relation of diuretic therapy to diabetes, impaired glucose tolerance and insulin resistance it is important to recognise that uncontrolled observations will be confounded by a well-established association of these metabolic abnormalities with hypertension, which is entirely independent of treatment. For example, meticulous epidemiological and case-control studies documenting the development of diabetes and glucose intolerance in hypertensive men'1'2' have shown that those destined to acquire these abnormalities are characterized by insulin resistance and abnormal carbohydrate homeostasis at the outset. Furthermore, impaired glucose tolerance was shown to antedate the development of hypertension by as long as 18 years'31, so that those in the upper tertiles for blood glucose 60 min after an oral glucose load had a 1 -7-fold risk of becoming hypertensive. Insulin resistance is present even in young people with essential hypertension, and this relation is independent of obesity or the presence of glucose intolerance141. Even more striking evidence of the intimate association of hypertension and impaired glucose intolerance is the fact that the presence of both conditions at the age of 64 can be predicted from birthweight, or from weight at one year151. For example men with birthweight > 2500 g have an odds ratio of 6-6 for development of impaired glucose tolerance when aged 64, and are significantly more likely to develop hypertension'51. During the middle and late years of life blood pressure tends to rise and glucose tolerance to decline, even in the absence of diuretic or other treatment'1-21. Because of this intimate association'61, uncontrolled observations and epidemiological studies Correspondence- University Department of Medicine and Pharmacology, L Floor, Royal Hallamshire Hospital, Glossop Road, Sheffield S10 2JF. U.K. 0195-668X/92/0GO068 + 04 $08.00/0
are not well suited to examining the role of diuretics in causing diabetes, impaired glucose tolerance, or insulin resistance. For example a study of serial glucose tolerance tests over 14 years in diuretic-treated hypertensive patients is difficult or impossible to interpret in the absence of an appropriate control group171. Reliable conclusions are likely to emerge only from observations in unconfounded controlled trials comparing diuretics to placebo or other antihypertensive agents. The controlled trials available were reviewed with the aim of answering some of the relevant questions. Do diuretics cause impaired glucose intolerance?
In the M RC trial in mild hypertension'81 bendrofluazide at high dose (10 mg od), was compared to placebo, and glucose tolerance tests were performed if random blood sugar was >8-3mmol.l~' or fasting blood sugar > 6 0 m m o l . 1~'. Over 3 years the rates of impaired glucose tolerance (per 1000 patient years) were 9-4 (bendrofluazide) vs 2-5 (placebo) for men, and 60 vs 0-8 for women. In the subsequent MRC trial in elderly hypertensive patients'9' rates of impaired glucose tolerance (per 1000 patient years) were 6-9 for hydrochlorothiazide 25 mg plus amiloride 2-5 mg, and 2-7 for placebo. One concern is that these trials were single blind, and although the investigators considered that bias was unlikely to be present'81 this possibility cannot be excluded with certainty. However, broadly similar findings have emerged in double-blind studies. For example, in the EWPHE trial"01 hydrochlorothiazide 25-50mg plus triamterene 50-100 mg daily was compared to placebo over 3 years in elderly hypertensives. Fasting blood sugar was increased significantly by diuretic treatment at 1 year, and remained elevated, by about 6%, after 3 years. In the © 1992 The European Society of Cardiology
Carbohydrate tolerance and diuretics 69
SHEP study1111 chlorthalidone 12-5-250mg daily was compared to placebo in elderly subjects with isolated systolic hypertension. After one year of treatment, blood sugar was significantly higher with thiazide treatment, by 5%. There is therefore ample evidence that long-term treatment with thiazide diuretics results in impaired glucose tolerance in hypertensive patients. Relatively little information is available on the time-course of development of impaired glucose tolerance. In the first MRC trial'8' 30% of the impaired glucose tolerance detected had developed within 3 months of starting bendrofluazide treatment, and 80% within 1 year. Is the effect of thiazides on glucose tolerance dose-related?
There is unequivocal evidence that the effect of thiazides on blood glucose is related to dose, probably in linear fashion. On the other hand, the hypotensive action of thiazides shows little relation to dose. For example, as long ago as 1977 Tweeddale et a/.'12' showed step-wise increases in blood glucose with increasing doses of chlorthalidone, from 25-200 mg daily, each given for 8 weeks. Compared to placebo values, blood glucose was elevated by a mean of 0-29 mmol. 1~' with 25 mg daily, and by 0-80 mmol. I"1 with 200 mg daily. These doses of chlorthalidone had similar effects on supine blood pressure. These findings have been confirmed recently in a placebo-controlled parallel group study of bendrofluazide 1-25-10-0 mg daily for 10 weeks in 257 patients (Fig. I)'13'. Bendrofluazide 10mg daily elevated blood sugar significantly, by 7%, whereas a dose of 1-25 mg daily had no adverse effect on blood glucose (mean fall 2%). The antihypertensive effect of bendrofluazide showed little relation to the dose (Fig. 1). The question arises whether long-term treatment with thiazides at low dosage impairs glucose tolerance. The evidence on this is conflicting, perhaps because of differences between studies in dosage, the individual diuretic used, duration of treatment, and the patient population studied. In an uncontrolled study Berglund et a/.'14' showed no adverse effect in 53 patients of bendrofluazide, at an average dose of 3-5 mg daily for 2 years, as regards fasting blood glucose, blood glucose 60min after a glucose load, or plasma insulin concentration. In another controlled trial in 93 patients, hydrochlorothiazide 25-50 mg daily for 6 months caused no change in fasting blood glucose'15'. On the other hand, significant effects on blood glucose were observed in some of the long-term controlled trials which used relatively low doses of thiazide. Elevated blood glucose or impaired glucose tolerance was reported with hydrochlorothiazide 25-50 mg for 3 years'10'; with hydrochlorothiazide 25 mg daily*9'; and chlorthalidone 12-5-250 mg daily1'''. The evidence available suggest that the relatively low doses used in these long-term trials do alter glucose homeostasis, and similar studies with even lower doses would be valuable. Are there differences between diuretics?
There have been many suggestions in the literature that longer-acting agents such as chlorthalidone may have a
25
2-5
5-0
IOO
Bendrofluazide dott (mg)
Figure I Changes in blood sugar and blood pressure with different doses of bendrofluazide, each given for 10 weeks. The changes are from placebo values, with n = 50-52 for each dose. (Derived from Carlsen et a/.|l3r).
more pronounced effect on glucose tolerance, and that shorter-acting diuretics, such as loop diuretics, have a less marked effect'1617'. This is difficult to confirm from controlled trials. There have been suggestions that indapamide may be free of any adverse effect on glucose tolerance. However, it is probable that any difference from other medium-potency diuretics is dose-related'16'. Doses of indapamide sufficiently high to cause electrolyte disturbance (mean 3-5 mg daily) caused deterioration in fasting blood glucose, glucose tolerance and glycosylated haemoglobin in diabetic patients'16'. As noted above, very low doses of other thiazides, e.g. bendrofluazide 1 -25 mg daily (Fig. l'131), have no discernible adverse effect on blood glucose. The potassium-sparing agents, amiloride, spironolactone and triamterene appear to have no adverse effect on glucose tolerance. In the case of amiloride and triamterene this may be due to their small effect on plasma or extracellular fluid volume. However, there is good evidence that spironolactone has no adverse
70 L. E. Ramsay et al.
effect on glucose tolerance even at doses sufficiently high to have antihypertensive and volume-depleting effects equivalent to those of the thiazides1'81. Is glucose intolerance related to potassium changes?
and Tuck1221. It is well established that diuretic treatment has a rapid and substantial adverse effect on metabolic control in diabetes, as shown by increases in blood sugar and glycosylated haemoglobin1'617'221. Beta-blocker therapy may also have an adverse effect on glucose homeostasis, which is additive to that of thiazides1231.
The fact that spironolactone has no effect on glucose homeostasis might suggest that potassium depletion has some role in causing the metabolic disturbance, and there Do diuretic-induced changes in glucose tolerance matter? is some experimental evidence for this ll6 ' l7 l However It is important to recognise that the extensive literature data from controlled trials give limited support to this hypothesis. For example, use of the potassium-sparing on diuretics and carbohydrate metabolism has focused agents amiloride, in the MRC trial in the elderly191, and almost entirely on changes detected only by detailed triamterene in EWPHE1'01, combined with relatively low biochemical testing, namely biochemical diabetes, doses of thiazide have failed to prevent impaired glucose impaired glucose tolerance, and more recently insulin tolerance, as has been discussed earlier. Furthermore, resistance. Information on the clinical importance of such high doses of potassium chloride in patients with thiazide- changes is lacking. For example, it is not known whether induced hypokalaemia had no important effect on blood they cause symptoms, or lead to the retinal, renal or glucose or serum insulin when fasting or 1 and 2 h after an peripheral nerve complications observed in naturallyoccurring diabetes. The only clinical presentation of oral glucose load1"'. diabetes clearly linked to diuretic treatment is the hyperosmolar non-ketotic syndrome. Several case reports and Do diuretics impair sensitivity to insulin? small series of this rare but serious condition have implior loop diuretics as a common cause or The general subject of insulin resistance has been cated 1thiazide 7 241 1201 trigger ' . To put this in context, we have seen one reviewed by Moller and Flier , and the effect of diuretics by Yudkin161. Clinical trial evidence on the effect of patient with diuretic-related hyperosmolar non-ketotic diuretic treatment on insulin resistance is remarkably diabetes in the last 15 years despite extensive use of limited considering the large volume of comment and diuretics in our general medical and hypertension practice. speculation on the topic. Hydrochlorothiazide, at an Biochemical diabetes, glucose intolerance and insulin average dose of 40 mg daily for 18 weeks, increased insulin resistance have received attention mainly because of conresistance substantially when compared to captopril'2'1. cern that these metabolic phenomena may increase the Insulin resistance was probably increased in absolute risk of coronary heart disease. However, the results of terms, although it should be noted that the study was not numerous long-term controlled trials of antihypertensive strictly placebo controlled. However, it is probable that treatment do not support such concerns. An overview of the changes in blood glucose and glucose tolerance 14 outcome trials in hypertension, with treatment observed in the studies reviewed above do reflect changes based mainly on thiazides at high dosage, showed a in insulin sensitivity. 14% reduction in coronary events'251. This decrease in coronary morbidity was not significantly different to that anticipated from epidemiological studies. Since then two Are diuretic effects on glucose tolerance reversible? additional major outcome trials pertinent to these conThere is relatively little evidence on reversibility. In the cerns have been published, namely SHEP1"1 and the MRC trial in mild hypertension181 the biochemical abnor- Medical Research Council trial in the elderly191. Both trials mality was reported to have resolved in 60% of affected examined therapy based on thiazide diuretics, and in each patients within one year. This figure takes no account trial diuretic treatment was associated with significant of impaired glucose tolerance observed during placebo glucose intolerance. Despite this, thiazide-based treattreatment, and is likely to have underestimated the degree ment reduced coronary events significantly, by 25% in of reversibility. Murphy et a/.'71 showed some reversibility SHEP1"1 and by 44% in the MRC trial191. These findings of impaired glucose tolerance when diuretics were discon- may dispel concern about the safety of thiazides during tinued after 14 years of therapy, but in only a very small long-term treatment. group of patients and without appropriate controls. In the crossover study of Pollare et a/.'2'1 a significant carryover effect was observed which the authors attributed to per- Implications for use of thiazides sistence of the influence of hydrochlorothiazide on insulin sensitivity. If correct, this suggests that the effect is not Despite the well-documented association between immediately reversible. diuretic treatment and biochemical impaired glucose tolerance, a thiazide at low dosage remains an excellent choice for first-line treatment in patients who do not have What effect does diuretic treatment have in established diabetes or gout. However, diuretics should be avoided in diabetes? diabetic patients whenever possible, although they may The effects of diuretics and other antihypertensive have to be used if the blood pressure cannot be controlled drugs on diabetic patients have been reviewed by Stern satisfactorily with alternative antihypertensive agents.
Carbohydrate tolerance and diuretics 71
References [1] Skarsfors ET, Lithell HO, Selinus I, Aberg H. Do antihypertensive drugs precipitate diabetes in predisposed men? Br Med J 1989; 298: 1147-52. [2] Skarsfors ET, Selinus K.I, Lithell HO. Risk factors for developing non-insulin dependent diabetes: a 10 year follow up of men in Uppsala. Br Med J 1991; 303: 755-60. [3] Salomaa VV, Strandberg TE, Vanhanen H, Naukkarinen V, Sarna S, Miettinen TA. Glucose tolerance and blood pressure: long term follow up in middle aged men. Br Med J 1991; 302: 493-6. [4] Ferrannini E, Buzzigoli G, Bonadonna R el al. Insulin resistance in essential hypertension. N Engl J Med 1987; 317: 350-7. [5] Hales CN, Barker DJP, Clark PMS et al. Fetal and infant growth and impaired glucose tolerance at age 64. Br Med J 1991; 303: 1019-22. [6] Yudkin JS. Hypertension and non-insulin dependent diabetes. BrMedJ 1991; 303: 730-1. [7] Murphy MB, Lewis PJ, Kohner E, Schumer B, Dollery CT. Glucose intolerance in hypertensive patients treated with diuretics: a fourteen-year follow-up. Lancet 1982; 2: 1293-5. [8] Medical Research Council Working Party on Mild to Moderate Hypertension. Adverse reactions to bendrofluazide and propranolol for the treatment of mild hypertension. Lancet 1981; 2: 539-43. [9] MRC Working Party. Medical Research Council trial of treatment of hypertension in older adults: principal result. Br Med J 1992; 304:405-12. [10] Amery A, Birkenhager W, Brixko P el al. Glucose intolerance during diuretic therapy in elderly hypertensive patients. A second report from the European Working Party on high blood pressure in the elderly (EWPHE). Postgrad Med J 1986; 62: 919-24. [11] SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension. Final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA 1991; 265: 3255-64. [12] Tweeddale MG, Ogilvie RI, Ruedy J. Antihypertensive and biochemical effects of chlorthalidone. Clin Pharmacol Ther 1977; 22: 519-27.
r'
[13] Carlsen JE, Kober L, Torp-Pedersen C, Johansen P. Relation between dose of bendrofluazide, antihypertensive effect, and adverse biochemical effects. Br Med J 1990; 300: 975-8. [14] Berglund G, Andersson O, Larsson O, Wilhelmsen L. Antihypertensive effect and side-effects of bendroflumethiazide and propranolol. Acta Med Scand 1976; 199:499-506. [15] Helgeland A, Strommen R, Hagelund C, Tretli S. Enalapril, atenolol, and hydrochlorothiazide in mild to moderate hypertension. Lancet 1986; 1: 872-5. [16] Feely J, O'Byrne S. Effects of drugs on glucose tolerance in non-insulin-dependent diabetics (part 1). Drugs 1990; 40: 6-18. [17] Dollery CT. Diabetogenic effect of long-term diuretic therapy. In: Lant AF, Wilson GM, eds. Modern Diuretic Therapy in the Treatment of Cardiovascular and Renal Disease. Amsterdam: Excerpta Medica, 1973: 320-8. [18] Schersten B, Thulin T, Kuylenstierna J et al. Clinical and biochemical effects of spironolactone administered once daily in primary hypertension. Multicenter Sweden Study. Hypertension 1980; 2: 672-9. [19] Kaplan NM, Carnegie A, Raskin P, Heller JA, Simmons M. Potassium supplementation in hypertensive patients with diuretic-induced hypokalaemia. N Engl J Med 1985; 312: 746-9. [20] Moller DE, Flier JS. Insulin resistance—mechanisms, syndromes, and implications. N Engl J Med 1991; 325: 938-48. [21] Pollare T, Lithell H, Berne C. A comparison of the effects of hydrochlorothiazide and captopril on glucose and lipid metabolism in patients with hypertension. N Engl J Med 1989; 321:868-73. [22] Stern N, Tuck M. Drug therapy of hypertension in diabetic patients. J Hum Hypertens 1991; 5: 295-305. [23] Dornhorst A, Powell SH, Pensky J. Aggravation by propranolol of hyperglycaemic effect of hydrochlorothiazide in type II diabetics without alteration of insulin secretion. Lancet 1985; 1: 123-6. [24] Fonseca V, Phear DN. Hyperosmolar non-ketotic diabetic syndrome precipitated by treatment with diuretics. Br Med J 1982; 284: 36-7. [25] Collins R, Peto R, MacMahon S et al. Blood pressure, stroke, and coronary heart disease. 2. Short-term reduction in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet 1990; 335: 827-38.
Diabetes, impaired glucose tolerance and insulin resistance with diuretics L. E. RAMSAY, W. W. YEO AND P. R. JACKSON
Royal Hallamshire Hospital, Sheffield, U.K. KEY WORDS: Hypertension, diuretics, glucose tolerance, insulin resistance. There is a definite relation of diuretic treatment to impaired glucose tolerance and biochemical diabetes, and a probable relation to insulin resistance. The effect of diuretics on glucose tolerance is dose-related. Spironolactone does not impair glucose tolerance even at high dosage, but apparent differences between other diuretics may well be due to comparison at doses which are not equivalent. Diuretic-induced changes in carbohydrate metabolism are not conclusively related to altered potassium homeostasis, and impaired glucose tolerance occurs when relatively low doses ofthiazide are combined with potassium-sparing agents. The effect of diuretics on glucose tolerance is largely and possibly wholly reversible. These disturbances of carbohydrate homeostasis have been detected by detailed biochemical testing, and their clinical importance is uncertain. In established diabetes, diuretics have a rapid and substantial adverse effect on metabolic control. In non-diabetic subjects diuretics may rarely cause or trigger a serious hyperosmolar non-ketotic diabetic syndrome. This apart, it is not known whether the metabolic changes cause clinical diabetes or lead to microvascular complications in the long-term. It is now established that biochemical diabetes, glucose intolerance and insulin resistance probably do not increase the risk of coronary heart disease in treated hypertensive patients. Diuretics should be avoided in patients with diabetes, otherwise they remain an excellent choice for first-line antihypertensive therapy. Introduction
When considering the relation of diuretic therapy to diabetes, impaired glucose tolerance and insulin resistance it is important to recognise that uncontrolled observations will be confounded by a well-established association of these metabolic abnormalities with hypertension, which is entirely independent of treatment. For example, meticulous epidemiological and case-control studies documenting the development of diabetes and glucose intolerance in hypertensive men'1'2' have shown that those destined to acquire these abnormalities are characterized by insulin resistance and abnormal carbohydrate homeostasis at the outset. Furthermore, impaired glucose tolerance was shown to antedate the development of hypertension by as long as 18 years'31, so that those in the upper tertiles for blood glucose 60 min after an oral glucose load had a 1 -7-fold risk of becoming hypertensive. Insulin resistance is present even in young people with essential hypertension, and this relation is independent of obesity or the presence of glucose intolerance141. Even more striking evidence of the intimate association of hypertension and impaired glucose intolerance is the fact that the presence of both conditions at the age of 64 can be predicted from birthweight, or from weight at one year151. For example men with birthweight > 2500 g have an odds ratio of 6-6 for development of impaired glucose tolerance when aged 64, and are significantly more likely to develop hypertension'51. During the middle and late years of life blood pressure tends to rise and glucose tolerance to decline, even in the absence of diuretic or other treatment'1-21. Because of this intimate association'61, uncontrolled observations and epidemiological studies Correspondence- University Department of Medicine and Pharmacology, L Floor, Royal Hallamshire Hospital, Glossop Road, Sheffield S10 2JF. U.K. 0195-668X/92/0GO068 + 04 $08.00/0
are not well suited to examining the role of diuretics in causing diabetes, impaired glucose tolerance, or insulin resistance. For example a study of serial glucose tolerance tests over 14 years in diuretic-treated hypertensive patients is difficult or impossible to interpret in the absence of an appropriate control group171. Reliable conclusions are likely to emerge only from observations in unconfounded controlled trials comparing diuretics to placebo or other antihypertensive agents. The controlled trials available were reviewed with the aim of answering some of the relevant questions. Do diuretics cause impaired glucose intolerance?
In the M RC trial in mild hypertension'81 bendrofluazide at high dose (10 mg od), was compared to placebo, and glucose tolerance tests were performed if random blood sugar was >8-3mmol.l~' or fasting blood sugar > 6 0 m m o l . 1~'. Over 3 years the rates of impaired glucose tolerance (per 1000 patient years) were 9-4 (bendrofluazide) vs 2-5 (placebo) for men, and 60 vs 0-8 for women. In the subsequent MRC trial in elderly hypertensive patients'9' rates of impaired glucose tolerance (per 1000 patient years) were 6-9 for hydrochlorothiazide 25 mg plus amiloride 2-5 mg, and 2-7 for placebo. One concern is that these trials were single blind, and although the investigators considered that bias was unlikely to be present'81 this possibility cannot be excluded with certainty. However, broadly similar findings have emerged in double-blind studies. For example, in the EWPHE trial"01 hydrochlorothiazide 25-50mg plus triamterene 50-100 mg daily was compared to placebo over 3 years in elderly hypertensives. Fasting blood sugar was increased significantly by diuretic treatment at 1 year, and remained elevated, by about 6%, after 3 years. In the © 1992 The European Society of Cardiology
Carbohydrate tolerance and diuretics 69
SHEP study1111 chlorthalidone 12-5-250mg daily was compared to placebo in elderly subjects with isolated systolic hypertension. After one year of treatment, blood sugar was significantly higher with thiazide treatment, by 5%. There is therefore ample evidence that long-term treatment with thiazide diuretics results in impaired glucose tolerance in hypertensive patients. Relatively little information is available on the time-course of development of impaired glucose tolerance. In the first MRC trial'8' 30% of the impaired glucose tolerance detected had developed within 3 months of starting bendrofluazide treatment, and 80% within 1 year. Is the effect of thiazides on glucose tolerance dose-related?
There is unequivocal evidence that the effect of thiazides on blood glucose is related to dose, probably in linear fashion. On the other hand, the hypotensive action of thiazides shows little relation to dose. For example, as long ago as 1977 Tweeddale et a/.'12' showed step-wise increases in blood glucose with increasing doses of chlorthalidone, from 25-200 mg daily, each given for 8 weeks. Compared to placebo values, blood glucose was elevated by a mean of 0-29 mmol. 1~' with 25 mg daily, and by 0-80 mmol. I"1 with 200 mg daily. These doses of chlorthalidone had similar effects on supine blood pressure. These findings have been confirmed recently in a placebo-controlled parallel group study of bendrofluazide 1-25-10-0 mg daily for 10 weeks in 257 patients (Fig. I)'13'. Bendrofluazide 10mg daily elevated blood sugar significantly, by 7%, whereas a dose of 1-25 mg daily had no adverse effect on blood glucose (mean fall 2%). The antihypertensive effect of bendrofluazide showed little relation to the dose (Fig. 1). The question arises whether long-term treatment with thiazides at low dosage impairs glucose tolerance. The evidence on this is conflicting, perhaps because of differences between studies in dosage, the individual diuretic used, duration of treatment, and the patient population studied. In an uncontrolled study Berglund et a/.'14' showed no adverse effect in 53 patients of bendrofluazide, at an average dose of 3-5 mg daily for 2 years, as regards fasting blood glucose, blood glucose 60min after a glucose load, or plasma insulin concentration. In another controlled trial in 93 patients, hydrochlorothiazide 25-50 mg daily for 6 months caused no change in fasting blood glucose'15'. On the other hand, significant effects on blood glucose were observed in some of the long-term controlled trials which used relatively low doses of thiazide. Elevated blood glucose or impaired glucose tolerance was reported with hydrochlorothiazide 25-50 mg for 3 years'10'; with hydrochlorothiazide 25 mg daily*9'; and chlorthalidone 12-5-250 mg daily1'''. The evidence available suggest that the relatively low doses used in these long-term trials do alter glucose homeostasis, and similar studies with even lower doses would be valuable. Are there differences between diuretics?
There have been many suggestions in the literature that longer-acting agents such as chlorthalidone may have a
25
2-5
5-0
IOO
Bendrofluazide dott (mg)
Figure I Changes in blood sugar and blood pressure with different doses of bendrofluazide, each given for 10 weeks. The changes are from placebo values, with n = 50-52 for each dose. (Derived from Carlsen et a/.|l3r).
more pronounced effect on glucose tolerance, and that shorter-acting diuretics, such as loop diuretics, have a less marked effect'1617'. This is difficult to confirm from controlled trials. There have been suggestions that indapamide may be free of any adverse effect on glucose tolerance. However, it is probable that any difference from other medium-potency diuretics is dose-related'16'. Doses of indapamide sufficiently high to cause electrolyte disturbance (mean 3-5 mg daily) caused deterioration in fasting blood glucose, glucose tolerance and glycosylated haemoglobin in diabetic patients'16'. As noted above, very low doses of other thiazides, e.g. bendrofluazide 1 -25 mg daily (Fig. l'131), have no discernible adverse effect on blood glucose. The potassium-sparing agents, amiloride, spironolactone and triamterene appear to have no adverse effect on glucose tolerance. In the case of amiloride and triamterene this may be due to their small effect on plasma or extracellular fluid volume. However, there is good evidence that spironolactone has no adverse
70 L. E. Ramsay et al.
effect on glucose tolerance even at doses sufficiently high to have antihypertensive and volume-depleting effects equivalent to those of the thiazides1'81. Is glucose intolerance related to potassium changes?
and Tuck1221. It is well established that diuretic treatment has a rapid and substantial adverse effect on metabolic control in diabetes, as shown by increases in blood sugar and glycosylated haemoglobin1'617'221. Beta-blocker therapy may also have an adverse effect on glucose homeostasis, which is additive to that of thiazides1231.
The fact that spironolactone has no effect on glucose homeostasis might suggest that potassium depletion has some role in causing the metabolic disturbance, and there Do diuretic-induced changes in glucose tolerance matter? is some experimental evidence for this ll6 ' l7 l However It is important to recognise that the extensive literature data from controlled trials give limited support to this hypothesis. For example, use of the potassium-sparing on diuretics and carbohydrate metabolism has focused agents amiloride, in the MRC trial in the elderly191, and almost entirely on changes detected only by detailed triamterene in EWPHE1'01, combined with relatively low biochemical testing, namely biochemical diabetes, doses of thiazide have failed to prevent impaired glucose impaired glucose tolerance, and more recently insulin tolerance, as has been discussed earlier. Furthermore, resistance. Information on the clinical importance of such high doses of potassium chloride in patients with thiazide- changes is lacking. For example, it is not known whether induced hypokalaemia had no important effect on blood they cause symptoms, or lead to the retinal, renal or glucose or serum insulin when fasting or 1 and 2 h after an peripheral nerve complications observed in naturallyoccurring diabetes. The only clinical presentation of oral glucose load1"'. diabetes clearly linked to diuretic treatment is the hyperosmolar non-ketotic syndrome. Several case reports and Do diuretics impair sensitivity to insulin? small series of this rare but serious condition have implior loop diuretics as a common cause or The general subject of insulin resistance has been cated 1thiazide 7 241 1201 trigger ' . To put this in context, we have seen one reviewed by Moller and Flier , and the effect of diuretics by Yudkin161. Clinical trial evidence on the effect of patient with diuretic-related hyperosmolar non-ketotic diuretic treatment on insulin resistance is remarkably diabetes in the last 15 years despite extensive use of limited considering the large volume of comment and diuretics in our general medical and hypertension practice. speculation on the topic. Hydrochlorothiazide, at an Biochemical diabetes, glucose intolerance and insulin average dose of 40 mg daily for 18 weeks, increased insulin resistance have received attention mainly because of conresistance substantially when compared to captopril'2'1. cern that these metabolic phenomena may increase the Insulin resistance was probably increased in absolute risk of coronary heart disease. However, the results of terms, although it should be noted that the study was not numerous long-term controlled trials of antihypertensive strictly placebo controlled. However, it is probable that treatment do not support such concerns. An overview of the changes in blood glucose and glucose tolerance 14 outcome trials in hypertension, with treatment observed in the studies reviewed above do reflect changes based mainly on thiazides at high dosage, showed a in insulin sensitivity. 14% reduction in coronary events'251. This decrease in coronary morbidity was not significantly different to that anticipated from epidemiological studies. Since then two Are diuretic effects on glucose tolerance reversible? additional major outcome trials pertinent to these conThere is relatively little evidence on reversibility. In the cerns have been published, namely SHEP1"1 and the MRC trial in mild hypertension181 the biochemical abnor- Medical Research Council trial in the elderly191. Both trials mality was reported to have resolved in 60% of affected examined therapy based on thiazide diuretics, and in each patients within one year. This figure takes no account trial diuretic treatment was associated with significant of impaired glucose tolerance observed during placebo glucose intolerance. Despite this, thiazide-based treattreatment, and is likely to have underestimated the degree ment reduced coronary events significantly, by 25% in of reversibility. Murphy et a/.'71 showed some reversibility SHEP1"1 and by 44% in the MRC trial191. These findings of impaired glucose tolerance when diuretics were discon- may dispel concern about the safety of thiazides during tinued after 14 years of therapy, but in only a very small long-term treatment. group of patients and without appropriate controls. In the crossover study of Pollare et a/.'2'1 a significant carryover effect was observed which the authors attributed to per- Implications for use of thiazides sistence of the influence of hydrochlorothiazide on insulin sensitivity. If correct, this suggests that the effect is not Despite the well-documented association between immediately reversible. diuretic treatment and biochemical impaired glucose tolerance, a thiazide at low dosage remains an excellent choice for first-line treatment in patients who do not have What effect does diuretic treatment have in established diabetes or gout. However, diuretics should be avoided in diabetes? diabetic patients whenever possible, although they may The effects of diuretics and other antihypertensive have to be used if the blood pressure cannot be controlled drugs on diabetic patients have been reviewed by Stern satisfactorily with alternative antihypertensive agents.
Carbohydrate tolerance and diuretics 71
References [1] Skarsfors ET, Lithell HO, Selinus I, Aberg H. Do antihypertensive drugs precipitate diabetes in predisposed men? Br Med J 1989; 298: 1147-52. [2] Skarsfors ET, Selinus K.I, Lithell HO. Risk factors for developing non-insulin dependent diabetes: a 10 year follow up of men in Uppsala. Br Med J 1991; 303: 755-60. [3] Salomaa VV, Strandberg TE, Vanhanen H, Naukkarinen V, Sarna S, Miettinen TA. Glucose tolerance and blood pressure: long term follow up in middle aged men. Br Med J 1991; 302: 493-6. [4] Ferrannini E, Buzzigoli G, Bonadonna R el al. Insulin resistance in essential hypertension. N Engl J Med 1987; 317: 350-7. [5] Hales CN, Barker DJP, Clark PMS et al. Fetal and infant growth and impaired glucose tolerance at age 64. Br Med J 1991; 303: 1019-22. [6] Yudkin JS. Hypertension and non-insulin dependent diabetes. BrMedJ 1991; 303: 730-1. [7] Murphy MB, Lewis PJ, Kohner E, Schumer B, Dollery CT. Glucose intolerance in hypertensive patients treated with diuretics: a fourteen-year follow-up. Lancet 1982; 2: 1293-5. [8] Medical Research Council Working Party on Mild to Moderate Hypertension. Adverse reactions to bendrofluazide and propranolol for the treatment of mild hypertension. Lancet 1981; 2: 539-43. [9] MRC Working Party. Medical Research Council trial of treatment of hypertension in older adults: principal result. Br Med J 1992; 304:405-12. [10] Amery A, Birkenhager W, Brixko P el al. Glucose intolerance during diuretic therapy in elderly hypertensive patients. A second report from the European Working Party on high blood pressure in the elderly (EWPHE). Postgrad Med J 1986; 62: 919-24. [11] SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension. Final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA 1991; 265: 3255-64. [12] Tweeddale MG, Ogilvie RI, Ruedy J. Antihypertensive and biochemical effects of chlorthalidone. Clin Pharmacol Ther 1977; 22: 519-27.
r'
[13] Carlsen JE, Kober L, Torp-Pedersen C, Johansen P. Relation between dose of bendrofluazide, antihypertensive effect, and adverse biochemical effects. Br Med J 1990; 300: 975-8. [14] Berglund G, Andersson O, Larsson O, Wilhelmsen L. Antihypertensive effect and side-effects of bendroflumethiazide and propranolol. Acta Med Scand 1976; 199:499-506. [15] Helgeland A, Strommen R, Hagelund C, Tretli S. Enalapril, atenolol, and hydrochlorothiazide in mild to moderate hypertension. Lancet 1986; 1: 872-5. [16] Feely J, O'Byrne S. Effects of drugs on glucose tolerance in non-insulin-dependent diabetics (part 1). Drugs 1990; 40: 6-18. [17] Dollery CT. Diabetogenic effect of long-term diuretic therapy. In: Lant AF, Wilson GM, eds. Modern Diuretic Therapy in the Treatment of Cardiovascular and Renal Disease. Amsterdam: Excerpta Medica, 1973: 320-8. [18] Schersten B, Thulin T, Kuylenstierna J et al. Clinical and biochemical effects of spironolactone administered once daily in primary hypertension. Multicenter Sweden Study. Hypertension 1980; 2: 672-9. [19] Kaplan NM, Carnegie A, Raskin P, Heller JA, Simmons M. Potassium supplementation in hypertensive patients with diuretic-induced hypokalaemia. N Engl J Med 1985; 312: 746-9. [20] Moller DE, Flier JS. Insulin resistance—mechanisms, syndromes, and implications. N Engl J Med 1991; 325: 938-48. [21] Pollare T, Lithell H, Berne C. A comparison of the effects of hydrochlorothiazide and captopril on glucose and lipid metabolism in patients with hypertension. N Engl J Med 1989; 321:868-73. [22] Stern N, Tuck M. Drug therapy of hypertension in diabetic patients. J Hum Hypertens 1991; 5: 295-305. [23] Dornhorst A, Powell SH, Pensky J. Aggravation by propranolol of hyperglycaemic effect of hydrochlorothiazide in type II diabetics without alteration of insulin secretion. Lancet 1985; 1: 123-6. [24] Fonseca V, Phear DN. Hyperosmolar non-ketotic diabetic syndrome precipitated by treatment with diuretics. Br Med J 1982; 284: 36-7. [25] Collins R, Peto R, MacMahon S et al. Blood pressure, stroke, and coronary heart disease. 2. Short-term reduction in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet 1990; 335: 827-38.
