Insulin Action and Hepatic Glucose Cycling in Essential
Hypertension
D.P. Rooney, R.D.G. Neely, C.N. Ennis, N.P. Bell, B. Sheridan, A.B. Atkinson, E.R. Trimble, and P.M. Bell Peripheral insulin resistance is a feature of essential hypertension, but there is little information about hepatic insulin sensitivity. To investigate peripheral and hepatic insulin sensitivity and activity of the hepatic glucose/glucose &phosphate (G/GGP) substrate cycle in essential hypertension, euglycemic glucose clamps were performed in eight untreated patients and eight matched controls at insulin infusion rates of 0.2 and 1.0 mU * kg-’ . min-‘. A simultaneous infusion of (23H)- and (6’H)glucose. combined with a selective detritiation procedure, was used to determine glucose turnover, the difference being G/GGP cycle activity. Endogenous hepatic glucose production (EGP) determined with (6’H)glucose was similar in hypertensive and control groups in the postabsorptive state (11.0 + 0.3 v 10.9 2 0.3 pmol * kg-’ . min-‘) and with the 0.2 mU insulin infusion (4.9 + 0.5 v 4.0 f 0.8 Pmol * kg-’ * min-‘). With the 1.0 mU insulin infusion, glucose disappearance determined with (6’H)glucose was lower in the hypertensive group (21.8 * 2.4 Y 29.9 % 2.4 Pmol . kg-’ - min-‘, P < 601). G/GGP cycle activity was similar both in the postabsorptive state (2.2 f 0.4 v 2.7 + 0.4 Pmol * kg-’ . min-‘) and during insulin infusion (0.2 mU, 2.5 f 0.3 v 2.9 f 0.4; 1.0 mu, 4.7 f 0.3 v 5.3 + 1.1 Pmol - kg-’ * mitt- ’ for hypertensive and control groups, respectively). Following 12 weeks’ therapy with cyclopenthiazide 500 Pg/d in the hypertensive group, there were increases in fasting glucose (5.2 + 0.2 to 5.6 f 0.2 mmol - L-l, P < .05), fasting insulin (8.5 f 1.8 to 11.1 f 3.1 mU * L-‘, P < 95). and postabsorptive endogenous glucose production (11.0 + 0.3 to 12.3 2 0.4 pmol * kg-’ . min-‘. P < .05). EGP during the 0.2-mU infusion and glucose turnover with the l.O-mU infusion did not change. In conclusion, peripheral but not hepatic insulin sensitivity is diminished in essential hypertension. Activity of the G/GGP substrate cycle is similar in hypertensive and control groups. Cyclopenthiazide therapy causes postabsorptive hyperglycemia, but insulin sensitivity is not affected during hyperinsulinemia. Copyright 0 1992 by WA Saunders Company
A
N ASSOCIATION between insulin resistance and hypertension has been recognized for many years.’ A recent population study has demonstrated that this association is independent of obesity, glucose intolerance, age, and antihypertensive medication.’ Several studies have shown that subjects with essential hypertension have consistently higher plasma glucose and insulin levels during oral glucose tolerance tests compared to matched controls?” Of special interest is the study by Ferrannini et al using the glucose clamp technique in which untreated non-obese patients with essential hypertension were found to have peripheral insulin resistance.6 However, the plasma insulin levels achieved in this study (60 mU . L-l) did not permit detailed investigation of the possibility of impaired hepatic insulin sensitivity, because at this insulin concentration, hepatic glucose production is maximally suppressed.’ The term glucose cycling has been used to describe several pathways in glucose metabolism which can “recycle” glucose. These include the Cori cycle: concurrent glycogenesis and glycogenolysis,9 and the substrate or “futile” cycles of the glycolytic pathway.“’ The glucose + glucosedphosphate -+ glucose (G/G6P) cycle is one substrate cycle, and evidence of significant activity has been reported in man.” Following uptake from the plasma by the liver, glucose molecules can be phosphorylated, then dephosphorylated, and can subsequently reenter the plasma, resulting in no net metabolism of glucose. Increased G/G6P cycling, by reducing net hepatic glucose uptake, may contribute to hyperglycemia and has been reported in insulin-resistant conditions such as non-insulin-dependent diabetes mellitus” and acromegaly.13 The activity of the G/G6P cycle in essential hypertension has not been investigated previously. The demonstration of insulin resistance in essential hypertension also focuses attention on the choice of antihypertensive agent, because many of those currently available are reported to affect glucose tolerance.14 Thiazide diuretics are widely prescribed and may impair glucose tolerMetabolism, Vol41, No 3 (March), 1992: pp 317-324
ance,15 decrease insulin sensitivity,16 and increase the risk of developing diabetes in the longer term.‘7.‘8However, there is little information on the effects of thiazide diuretics on hepatic insulin action in essential hypertension. The aims of the present study were to determine if altered hepatic insulin sensitivity is a component of the impaired insulin action in essential hypertension and to assess G/G6P substrate cycling in essential hypertension. We used a low-dose insulin infusion, during which it was anticipated that partial suppression of hepatic glucose production would occur, as well as a high-dose insulin infusion, when it was expected that hepatic glucose production would be maximally suppressed. Finally, we investigated the effects in the hypertensive patients of a 1Zweek period of standard antihypertensive therapy with cyclopenthiazide on insulin action and G/G6P substrate cycling. MATERIALS AND METHODS Subjects
We studied eight non-obese patients with recently diagnosed, untreated essential hypertension and eight matched controls (Table 1). All patients and controls were of white western European origin. Hypertension was defined as a supine, outpatient blood pressure greater than 160/95 mm Hg, recorded using a random zero sphygmomanometer on three separate occasions by the same
From the Sir George E. Clark Metabolic Unit, Regional Endocrine Laboratoty, Royal Victoria Hospital, Belfast, Northern Ireland; and the Depattment of Clinical Pathologv, The Queen’s Universiry of Belfast, Northern Ireland. D.P.R. was supported by a Royal Kctoria Hospital Research Fellowship. Support aIso received from the British Diabetic Association and Mason Medical Research Foundation, Berkshire, UK Address reprint requests to P.M. Bell, MD, Consultant Physician, Sir George E. Clark Metabolic Unit, Royal Victoria Hospital, Belfast BT12 6BA, Northern Ireland. Copyright Q 1992 by W.B. Saunders Company 00260495/92/4103-0016$03.00l0
317
ROONEY ET AL
318
Table 1. Clinical Characteristics of the Hypertensive and Control Subjects (Mean k SEM)
Patients Sex (M/F) Age
(vr)
Body mass index (kg
m-‘)
PostControls Cyclopenthiazide
612
6/Z
53 + 2
53 * 1
24.4 2 1.1 25.2 _t 0.9
24.6 -c 1.3
Blood pressure (mm Hg) Systolic
175 ‘- 4
134 + 3
Diastolic
108 2 2
80 + 2
93 * 2
. L-‘) (mmol . L-‘)
148 2 4
Serum potassium fmmol
4.2 i 0.1
4.3 + 0.1
3.9 i 0.1
Fasting triglycerides
1.1 & 0.1
1 .O ? 0.1
1.7 * 0.3
physician. Neither hypertensive nor control subjects had a personal or family history of diabetes mellitus. All subjects had a similar dietary history, consumed less than eight units of alcohol per week, and had a similar sedentary life-style. Hypertensive subjects had a full clinical history and examination to exclude, in particular, past or present symptoms of renal disease, renal bruits, or palpable kidneys. Routine urine testing, serum potassium, urea, and creatinine levels were normal in each case. Urinary catecholamines were also normal in all cases. In the absence of clinical and biochemical features of renal disease, renal vascular causes of hypertension are exceptionally rare in those over 40 years of age.” Approval for the study was granted by the Ethical Committee of The Queen’s University of Belfast, and written informed consent was obtained in each case. Experimental Design At study entry, a standard 75-g oral glucose tolerance test was performed on each subject to establish normal glucose tolerance. Hepatic and peripheral insulin sensitivity were assessed using the euglycemic glucose clamp technique. Hypertensive subjects then commenced cyclopenthiazide therapy, 500 kg/d. Blood pressure and serum electrolyte concentrations were checked monthly. (Potassium supplements were not prescribed, as we wished to study the effects of cyclopenthiazide when administered alone.) After 12 weeks of cyclopenthiazide therapy, a second glucose clamp study was performed in each hypertensive subject, using an identical protocol to the first glucose clamp. The last dose of the drug was administered on the day preceding the second glucose clamp. Glucose Clamp Studies The euglycemic glucose clamp technique used to determine insulin sensitivity has been described previously.m Briefly, subjects were admitted at 8:00 AM after an overnight fast (12 hours). A plastic cannula (l&gauge, Venflon Virgo, Helsingborg, Sweden) was inserted into a left antecubital vein for all infusions. Insulin (Humulin S, Eli Lilly, Basingstoke, UK) was administered sequentially at 0.2 mU kg-’ min-’ for the first 2-hour period (chosen to produce partial suppression of hepatic glucose production) and at 1.0 mU kg-’ . min-’ for the second 2-hour period. Plasma glucose concentration was maintained at the fasting level by a variable exogenous infusion of 20% glucose. Arterialized blood samples were obtained at 5-minute intervals for glucose estimation, from a cannula (21-gauge, Venflon Viggo) placed retrogradely in a dorsal vein of the right hand, which was maintained at 55°C in a thermostatically controlled perspex enclosure (Automation Division, Ashby Institute, The Queen’s University of Belfast, Northern Ireland). Samples for plasma glucose were centrifuged immediately at the bedside and analyzed using a glucose oxidase method (Beckman Glucose Analyser II, Beckman RIIC, High Wycombe, UK).
Glucose turnover in the fasting state and during hyperinsulinemia was determined by the isotope dilution technique” using (2>H)- and (6’H)glucose tracers (Amersham International PLC, Aylesbury, UK, all hypertensive and five control clamps; material supplied by New England Nuclear Research Products Division, DuPont UK, Stevenage, Herts., was used for three controls). Since use of unpurified tracer can lead to underestimation of glucose turnover, all isotopes were purified by high-pressure liquid chromatography (HPLC) before infusion, as previously described.‘* These tracers were administered simultaneously as a primed (25 FCi) constant (0.25 FCiimin) infusion for 2 hours before the start of the insulin infusion (to allow isotope equilibration) and continued throughout the 4-hour glucose clamp period. Samples for determination of plasma glucose specific activity were taken at IO-minute intervals from -30 to 0 minutes. 90 to 120 minutes, and 210 to 240 minutes, collected in iced lithium heparin fluoride tubes, and separated within 15 minutes. Samples for insulin, free fatty acids (FFA), glycerol, and P-hydroxybutyrate were collected in plain glass tubes. Samples for pyruvate and lactate were collected in tubes containing 8% perchloric acid. Samples were separated as soon as clotting was complete. All samples were stored at -20°C until analysis. Analytical Techniques Plasma for glucose specific activity was deproteinized using Ba(OH), and ZnSO, by the method of Somogyi.” After centrifugation, the supernatant was passed sequentially through anion (AGIX8, Biorad Laboratories, Watford, UK) and cation (AG50W-X8, Biorad) exchange columns. The eluate was lyophilized and reconstituted in 500 FL of 133 mmol L-’ phosphate buffer (pH 7.4). A modification of the selective enzymatic detritiation method of Issekutz was used to determine separately (2jH)- and (6’H)glucose radioactivity.‘4 During this process, (2’H)glucose is detritiated while tritium in (6’H)glucose is left attached. Aliquots (200 FL) of reconstituted plasma extract were mixed in glass liquid scintillation vials with 500 WL of a buffered solution containing 1.2 IU hexokinase. 5 IU phosphoglucose isomerase, MgCl, (6 mmol L-l). and adenosine triphosphate (ATP) (5.6 mmol L-‘) (all reagents Sigma Chemical, Dorset, UK). Vials were incubated at 37°C for 2 hours and the mixture lyophilized. Following reconstitution in 1 mL 1N H,SO, and addition of 10 mL biofluor (Dupont, Stevenage, UK), radioactivity was counted in a liquid scintillation counter. Total tritiated glucose radioactivity was determined by processing 200 FL of the original reconstituted material in parallel with the samples above, with 500 FL of buffer in place of the enzyme solution. Subtraction of the (6’H)glucose radioactivity from total radioactivity yielded (23H)glucose radioactivity. External standards of (23H)glucose and (63H)glucose from each infusate were added to patient plasma (taken prior to tracer infusion) and processed in parallel with each patient assay. The results were used to calculate the degree of detritiation of the two isotopes for each assay. Detritiation of (23H)gIucose was 96 & 0.3%, whereas detritiation of (63H)glucose was only 0.1 2 0.3% (interassay coefficients of variation [CVs], 1.0% and 1.1%. respectively). Serum insulin was determined by a radioimmunoassay with insulin antibody precipitate” (interassay CV, 5.0%). Serum FFA
(Wako Chemicals, Neuss, Germany; interassay CV, 4.2%), serum total triglyceride (Wako Chemicals; interassay CV, 2.4%). P-hydroxybutyrate (Randox Laboratories, Crumlin, UK; interassay CV, 1.2%), serum glycerol (Randox Laboratories; interassay CV, 5.2%), serum pyruvate (Sigma Chemical, Dorset, UK, interassay CV, 2.1%), and serum lactate (Sigma Chemical; interassay CV, 2.5%) were all assayed by commercial kits using enzymatic meth-
319
INSULIN ACTION IN ESSENTIAL HYPERTENSION
ods and spectrophotometry. All CVs are given for the relevant ranges measured in this project.
IO
mmol/l
1
Calculation-s The non-steady-state equations of Steele:’ as modified by De Bodo et al% (assuming a pool fraction value of 0.65 and an extracellular volume of 190 mL. kg“), were used to determine rates of glucose appearance (Ra) and disappearance (Rd) separately for (2’H)- and (63H)glucose (Ra2/Ra6 and Rd2/Rd6, respectively), Endogenous (predominantly hepatic origin) glucose production (EGP) was calculated for (63H)glucose by subtraction of the exogenous glucose infusion rate (EGIR) required to maintain euglycemia from Ra6. It is known that when glucose is labeled with tritium in position 2 of the carbon ring, the label is lost in. the rapidly reversible reaction between glucose-6-phosphate and fructose-6-phosphate?’ Hence, (23H)glucose does not retain its radioactivity during G/G6P cycling, but instead may reappear in the plasma following hepatic uptake as unlabeled glucose. When the isotope dilution technique is employed using this form of labeled glucose, the measured rate of endogenous glucose appearance is assumed to represent total hepatic glucose output, ie, glucose derived from gluconeogenesis, glycogen breakdown, and glucose cycling. (63H)glucose instead retains its radioactivity throughout glucose cycling, and measurements of Ra using this label are taken to accurately reflect true hepatic glucose production?* Thus, the magnitude of the G/G6P substrate cycle was measured as the difference Ra2-Ra6 (ie, total hepatic glucose output minus true hepatic glucose production) and is expressed in absolute terms of km01 . kg’ . min-’ of glucose and as a percentage of glucose appearance determined with (63H)ghrcose (Ra6).
Statistical Methods Areas under the glucose and insulin curves were calculated for the glucose tolerance tests and analyzed by paired, two-tailed Student’s t tests. Paired t tests were also used for the comparison of glucose turnover values during the clamp studies and basal metabohte concentrations. The slopes of simple regression lines were calculated for metabolite concentrations during hyperinsulinemia (serum FFA, B-hydroxybutyrate, and glycerol concentrations were first subjected to natural logarithmic transformation). Regression line slopes were compared by paired, two-tailed Student’s t tests. P values less than .05 were taken as significant. Data in the text and figures are given as mean 2 SEM. RESULTS
Oral Glucose Tolerance Tests In the hypertensive (H) and control (C) groups, fasting venous plasma glucose (H 4.8 + 0.3 v C 4.4 ? 0.1 mmol . L-l) and serum insulin (H 11.3 + 1.6 v C 10.5 f 1.9 mu. L-l) were similar. Following 75-g oral glucose challenge, glucose tolerance was normal in all subjects (Fig 1). Although mean serum insulin and plasma glucose levels were consistently slightly higher in the hypertensive group, areas under the respective curves were not significantly different in hypertensive and control subjects. Glucose Clamp Studies Fasting arterialized venous plasma glucose levels were similar in both groups (H 5.2 + 0.2 v C 5.4 f 0.1 mmol . L-l). Plasma glucose levels during the clamp studies were (H)
4 120 ca,D
O
3o
TimeGYmin)
@O
mu/I 1001 80-
60. 4020-
Fig 1. Plasma glucose (a) and serum insulin (b) concentrations during 75-g oral glucose tolerance test in hypertensive (-O-1 and control I-O-) subjects.
5.2 ? 0.2 mmol * L-’ (mean CV, 2.7% ? 0.2%) and (C) 5.3 + 0.1 mm01 . L-l (mean CV, 3.3% * 0.2%). The EGIR required to maintain euglycemia during the last 30 minutes of the 0.2 mU . kg-’ . min-’ insulin infusion was not significantly different in hypertensive and control subjects (H 6.2 ? 0.3 v C 8.3 + 1.3 kmol . kg-’ . min-‘). During the 1.0 mU . kg-’ . min-’ insulin infusion, EGIR was significantly less in the hypertensive group (H 27.5 * 3.3 v C 38.1 + 2.6 kmol . kg-’ . min-‘, P < .005). EGP determined with (63H)glucose (Fig 2) was similar in hypertensive and control subjects in the postabsorptive state (H 11.0 + 0.3 v C 10.9 2 0.3 umol * kg-’ . min-‘, respectively) (Fig 2a) and also during the final 30 minutes of the 0.2 mU . kg-’ . min-’ insulin infusion period when partial suppression of EGP had occurred (H 4.9 f 0.5 v C 4.0 f 0.8 kmol . kg-’ . min-‘) (Fig 2b). During the final 30 minutes of the 1.0 mU . kg-’ . min-’ insulin infusion period, calculated EGP appeared to be completely suppressed, in that negative values were obtained in both groups (H -5.8 + 3.4 v C -8.2 + 2.6 pmol . kg-’ . min-‘) (Fig 2~). Total glucose disappearance of (63H)glucose during the 0.2 mU . kg-’ . min -’ insulin infusion was similar in hypertensive and control subjects (H 11.3 + 0.5 v C 12.4 ? 0.7 mU . kg-’ . min-‘), but during the 1.0 mU . kg-’ . min-’ insulin infusion period (Fig 2d) it was significantly
ROONEY ET AL
320
(a) 16
“I
(b)
nd/kg/min
16
umdlkglmin 1
12
63H
23~
(d) 10
umdlkgfmin
40
umd/kg/min
1
-10’
Fig 2. EGP ([2’H]and [G3Hlglucose) in the postabsorptive state (a), with the 0.2-mU insulin infusion (b), and with the l.O-mU insulin infusion (cf. Rate of glucose disappearance with the 1.0 mU insulin infusion (d). 0, Hypertensive; q. control; 8, postthiazide.
63H
23~
less in the hypertensive than in the control group (H 21.8 & 2.4 v C 29.9 ? 2.4 kmol . kg-’ . min-‘, P < .OOl). Glucose Cycling In the postabsorptive state (Fig 3a), the rate of G/G6P cycling was similar in both hypertensive and control groups (H 2.2 ? 0.4 v C 2.7 -C 0.4 Fmol . kg-’ . mini’; H 20% 2 3% v C 25% & 4%). During the 0.2 mU . kg-’ . min-’ infusion period, G/G6P cycling, either absolute or expressed as a percentage of (Ra6), did not change (H 2.5 f 0.3 and C 2.9 r 0.4 umol . kg-’ . min-‘; H 22% +- 2% and C 24% * 4%). However, expressed as a percentage of EGP (Fig 3b), glucose cycling increased significantly in both groups, compared with the postabsorptive state (P < .OOl). During the l.O-mU insulin infusion period, absolute G/G6P cycling (H 4.7 ? 1.3 and C 5.3 + 1.1 umol . kg-‘. min-‘, P < .05 for both groups) increased compared with the postabsorptive period, but remained at a similar percentage of Ra6 (H 21% -r- 5% and C 19% -r- 5%). There were no significant differences in G/G6P cycling between hypertensive and control groups at the 0.2- or l.O-mU insulin infusion levels. Metabolites
Postabsorptive levels of serum FFA, p-hydroxybutyrate, glycerol, lactate, and pyruvate were similar in hypertensive and control groups (Fig 4). In response to insulin, serum FFA, glycerol, and P-hydroxybutyrate concentrations were rapidly suppressed in both groups to a similar extent. Serum lactate and pyruvate levels did not alter significantly in either group from basal levels (Fig 4b and c).
Post-Cyclopenthiazide Therapy
Systolic and diastolic blood pressure decreased following 12 weeks’ cyclopenthiazide therapy (from 175 + 4/108 2 2 mm Hg to 148 + 4193 * 2 mm Hg, P < .OOl). Fasting plasma glucose, insulin, and triglyceride concentrations increased (from 5.2 ? 0.2 to 5.6 ? 0.2 mmol . L-‘, P < .05, 8.5 ? 1.8 to 11.1 2 3.1 mu. L-‘, P < .05. and 1.1 * 0.1 to 1.7 ? 0.3 mmol . L-‘, P < .Ol, respectively) (Table 1). Serum potassium did not change significantly (pre-cyclopenthiazide 4.2 * 0.1 mmol/L-’ v post-cyclopenthiazide 3.9 ? 0.1 mmol/L-‘). Postabsorptive EGP determined with (63H)glucose showed a slight, but significant, increase after cyclopenthiazide (from 11.0 c 0.3 to 12.3 +- 0.4 umol . kg-’ min. ‘, P < .05) (Table 2). However, the suppression of EGP by insulin during the euglycemic hyperinsulinemic clamp periods (Fig 2) and glucose disappearance (Rd6) at the 1.0 mU . kg-1 . min-’ insulin infusion period (from 21.8 +_2.4 to 21.7 ? 2.3 pmol . kg-’ . min-‘) did not change. Activity of the G/G6P substrate cycle was not affected by cyclopenthiazide treatment, either in the postabsorptive state (2.5 f 0.4 prnol . kg-’ . min-‘; 21% + 4%) or during the low (3.1 + 0.3 pmol . kg-’ . min-‘; 26% t 2%) and high (4.8 ? 0.9 kmol . kg-’ . min-‘; 22% * 3%) insulin infusion clamp periods (Fig 3). Postabsorptive serum FFA concentrations following cyclopenthiazide were similar to pretreatment values, despite an elevation in serum insulin concentration. During hyperinsulinemia, FFA suppression following cyclopenthiazide was impaired compared with pretreatment (P < .05) (Fig 4a). Compared with pretreatment, serum lactate and pyru-
INSULIN ACTION IN ESSENTIAL HYPERTENSION
321
8. % EGP
1
:::::::::::::::::::::::: :::::::::::::::::::::::: ..........” ........... .::::::::::::::::.::::::: .*..a :::::::::::::::::::::::: (a)
O Hycertensives
Control
Post Cycle.
% EGP
801
Fig 3. Glucose cycle activity (expressed as a percentage of EGP) in the postabsorptive state (a) and with the 0.2-mU insulin infusion (b). 8, Hypertensive; @I,control;
vate concentrations were increased in the postabsorptive state (P < .OOl) (Fig 4b and c) and remained higher during subsequent insulin infusion. Postabsorptive serum glycerol concentrations and glycerol suppression during hyperinsulinemia were unchanged by cyclopenthiazide (Fig 4d). Serum P-hydroxybutyrate concentrations following cyclopenthiazide were lower in the postabsorptive state and during hyperinsulinemia, although this difference was not statistically significant. DISCUSSION
Our results demonstrate that hepatic insulin sensitivity is preserved in essential hypertension. Postabsorptive EGP was similar in both the hypertensive and control groups and was suppressed to a similar extent by a small increment in plasma insulin concentration. At this insulin concentration (20 mu/L), glucose disappearance rates were not greatly changed from baseline, which is in keeping with the observation that stimulation of glucose utilization is less sensitive to insulin than suppression of EGP.’ At higher insulin concentrations, EGP appeared to be completely suppressed, in that negative glucose production rates were obtained. Given the difficulty of interpreting EGP at high insulin concentrations when negative results are obtained,
the conclusion that hepatic insulin sensitivity is preserved relies critically on the data obtained at low insulin concentrations. In studies of this type with relatively small numbers, the possibility of a type II error has to be considered. Taking a difference of 2 umol/kg/min as likely to be of biological significance, the power of our study to exclude a type II error is 76% (confidence limits, 1.1 to 2.0 Fmolikgi min). For a 3 umol/kg/min difference, the power is 97%. Therefore, we believe than an important defect in hepatic insulin action is unlikely to have been overlooked. In addition, we have confirmed that the defect in insulin action is located in the peripheral tissues. The rate of glucose disappearance, under conditions of rapid glucose turnover at the higher plasma insulin level, was 30% less in the hypertensive group. The majority of the insulinmediated increase in glucose disappearance during euglycemic glucose clamp studies can be accounted for by disposal in skeletal muscle,” strongly suggesting this is the site of defective insulin action in essential hypertension. The nature of this defect is unknown. Possible explanations include either restricted access of insulin and glucose to insulin-sensitive tissues caused by degenerative changes in blood vessels,‘” or insulin receptor and postreceptor abnormalities in the tissues themselves.3o Increased G/G6P cycle activity may be an important contributing mechanism to impaired suppression of hepatic glucose production in non-insulin-dependent diabetes mellitus.‘* We have demonstrated that G/G6P cycling in the postabsorptive state during hyperinsulinemia is not different in hypertensives and controls. Given similar hepatic insulin sensitivity, this is perhaps not surprising. However, several points of interest emerge from our calculations of G/G6P cycle activity, which are free from any possible errors generated by use of impure glucose tracers.” First, the magnitude of the G/G6P cycle in the postabsorptive state shows considerable heterogeneity among individuals with normal glucose tolerance (range, 0.9 to 5.0 pmol . kg-’ . min-I). This finding agrees with the results of Karlander et al in similarly aged normal subjects.28 It is possible that G/G6P cycling in man is higher than our estimates determined using (23H)- and (63H)glucose, if detritiation of (2’H)glucose is only 80% in vivo, as has been suggested recently.” Second, insulin does not appear to exert control over cycle activity. Cycling in absolute terms continued unabated at 20 mu. L-’ and increased slightly at 80 mU . L-‘. Assuming G/G6P cycling measured by our technique reflects mainly activity of hepatic glucokinase and glucose-6-phosphatase, continued cycle activity, at insulin concentrations assumed to suppress EGP maximally, suggests that even though glucose production from gluconeogenesis and glycogen breakdown is minimal, glucose may continue to leave the liver by means of recycling following uptake. When EGP is low, such cycling will represent a significant proportion of the hepatic glucose output. Finally, it should be mentioned that differences in glucose turnover rates determined with (23H)- and (63H)glucase will overestimate G/G6P cycling if significant glycogen breakdown is associated with or follows glycogen synthesis. This is because (63H)glucose that has been incorporated
322
0.6
ROONEY ET AL
rnmol/l
1.6.
0.4
0.8
umolil
200
mmolll
160.
umol/l
I
umolll 1
0
60
Time (min)
180
240
(4
(d) -
120 Time (rain)
Fig 4. Change in serum FFA (a), lactate (b), pyruvate (c), glycerol (d), and B_hydroxybutyrate , Hypertensive; -0., control; -A-, post-thiazide.
into glycogen can be released back to plasma without loss of labek3’ whereas (23H)glucose will lose its label during glycogen synthesis.” Thus, the difference in glucose turnover rates determined with (23H)- and (63H)glucose will reflect cycling through glycogen as well as G/G6P substrate
(e) concentrations during glucose clamp studies.
cycling.” However, under hyperinsulinemic, euglycemic conditions, it is likely that glycogen breakdown is minimal, and therefore, the difference between the isotopically determined turnover rates is predominantly a consequence of G/G6P cycling.
Table 2. Results Summary for the Euglycemic Glucose Clamp Studies in Hypertensive Subjects, Control Subjects, and Hypertensive Subjects Post-Cyclopenthiazide
Fasting glucose (mmol
. L-‘)
Mean clamp glucose (mmol Fasting serum insulin (mU
. L-l) ’ L-l)
(Mean lr SEM)
Hypertensive
Control
Hypertensive
Subjects
Subjects
Post-Cvclopenthiazide
Subjects
5.2 2 0.2
5.4 2 0.1
5.2 -t 0.2
5.3 -t 0.1
5.6 r 0.2* 5.5 -c 0.2
8.5 r 1.8
8.4 ? 2.0
11.1 f 3.1’
20.6 + 2.0
20.0 + 1.8
24.1 + 3.0
77.0 + 6.0
80.0 -c 6.9
80.0 * 8.0
11.0’
0.3
10.9 + 0.3
12.3 + 0.4,
4.9 * 0.5
4.0 + 0.8
4.2 2 0.4
21.8 r 2.4
29.9 + 2.4t
21.7 -e 2.3
27.5 + 3.3
38.7 + 2.7t
28.0 r 3.4
Serum insulin during 0.2-mU infusion (mu
L-‘)
Serum insulin during 1.0.mU infusion (mU
L-‘)
Postabsorptive hepatic glucose production (6?-i)glucose (pmol . kg-’ . min’) Hepatic glucose production (63H)glucose 0.2-mU insulin infusion (pmol
. kg-’ . min.‘)
Glucose disappearance rate (Rd6) 1.O-mU insulin infusion (pmol kg-’ EGIR l.O-mU insulin infusion
min-‘1
(pmol . kg-‘. min’) *P < .05 Pre-cyclopenthiazide v post-cyclothiazide therapy. tP .05 Hypertensive v control groups.
INSULIN ACTION IN ESSENTIAL
323
HYPERTENSION
In our hypertensive patients, insulin-mediated glucose disposal following 12 weeks’ cyclopenthiazide treatment was not significantly different from pretreatment. This is in contrast to a study by Pollare et al using hydrochlorothiazide 25 mg/d, in which there was a decrease in peripheral insulin sensitivity.16 Metabolic effects of thiazide diuretics have been attributed to associated hypokalemia,” and low potassium concentrations impair nonoxidative glucose disposal as a consequence of diminished glycogen synthetase activity.” The decrease in serum potassium in our study was of similar magnitude to that of Pollare et al, although it did not reach statistical significance. Interestingly, in the Pollare et al study, lower absolute potassium concentrations resulted, and while this may be related to differences in laboratory measurement, it is possible that potassium concentrations in our study, although reduced, did not become low enough to affect muscle insulin sensitivity. Unlike our patients, those of Pollare et al had received previous antihypertensive treatment, which may have included potassium-decreasing diuretics. An alternative explanation of the difference between the two studies is the use of different thiazide diuretics. However, we can say that decreasing blood pressure with conventional doses of diuretics did not ameliorate the insulin resistance of essential hypertension. We have demonstrated significantly increased postabsorptive EGP, plasma glucose, and serum insulin levels following cyclopenthiazide. This may represent induction of a degree of hepatic insulin resistance (seen only at low insulin levels), even though suppression of EGP by exogenous insulin was not affected. Such an effect of a thiazide diuretic has not previously been reported in subjects with normal glucose tolerance and emphasizes the importance, in studies focusing on the effects of drug therapy on insulin resistance, of examining possible changes in hepatic insulin sensitivity. Cyclopenthiazide therapy resulted in changes in lipid metabolism. In addition to causing an increase in serum triglyceride, the suppression of serum FFA by insulin during the glucose clamp studies was impaired. There are two main possibilities to account for the latter. Cyclopenthiazide may cause antagonism of the antilipolytic activity of insulin on adipose tissue, or it may alter clearance of FFA from plasma. Serum glycerol levels reflect the extent of triglyceride lipolysis and were not significantly altered after cyclopenthiazide therapy, either in the postabsorptive state
or during the hyperinsulinemic clamp studies. This suggests that the serum FFA findings were due to reduced clearance from plasma, although because postabsorptive serum insulin was higher following cyclopenthiazide, some impairment in insulin antilipolytic activity cannot be completely excluded. It is also possible that reesterification of fatty acid in the adipocyte using new glycerol-3-phosphate derived from glycolysis? is impaired by thiazides. Reduced triglyceride/ fatty acid cycling in this way would result in blunted suppression of FFA concentration without affecting liberation of glycerol. Serum lactate and pyruvate, both in the postabsorptive state and during hyperinsulinemia, were also increased after cyclopenthiazide therapy. These findings were probably not a consequence of enhanced FFA provision, because even though FFA oxidation can indirectly promote lactate formation from pyruvate as a result of pyruvate dehydrogenase inhibition,” the serum lactate to pyruvate ratio and l3-hydroxybutyrate concentrations were not significantly altered. It is instead more likely that the increase in serum lactate and pyruvate after cyclopenthiazide was a result of either enhanced cellular release or greater nonoxidative glycolysis, even though overall glucose utilization was not altered. Increased hepatic glucose production may result from enhanced gluconeogenesis. Lactate and pyruvate are gluconeogenic substrates,3s and gluconeogenesis may be stimulated by FFA.39 Thus, increased serum levels of these metabolites following cyclopenthiazide therapy may be one mechanism to account for the increased EGP. In conclusion, we have demonstrated that in patients with essential hypertension, significant peripheral, but not hepatic, insulin resistance is present. Increased G/G6P substrate cycling is not a feature of insulin resistance in essential hypertension, and G/G6P cycling in hypertensive and normotensive subjects is similar. Finally, cyclopenthiazide therapy over a 1Zweek period results in increased postabsorptive endogenous glucose production and altered lipid metabolism, but causes no change in insulin sensitivity during hyperinsulinemia.
ACKNOWLEDGMENT
We acknowledge with gratitude the help of Dr C. Patterson, Department of Community Medicine and Medical Statistics, The Queen’s University of Belfast, in the analysis of data, and also the secretarial assistance of May Weller and Marie Loughran.
REFERENCES
1. Welborn TA, Breckenridge A, Rubinstein AH, et al: Serum insulin in essential hypertension and in peripheral vascular disease. Lancet 1:1336-1337,1966 2. Modan M, Halkin H, Almog S, et al: Hyperinsulinemia: A link between hypertension, obesity and glucose intolerance. J Clin Invest 75:809-817,1985 3. Singer P, Godicke W, Voight S, et al: Postprandial hyperinsulinemia in patients with mild essential hypertension. Hypertension 7:182-186,1985 4. Shen DC, Shieh SM, Fuh MT, et al: Resistance to insulinstimulated glucose uptake in patients with hypertension. 4 Clin Endocrinol Metab 66:580-583,1988
5. Swislocki AL, Hoffman BB, Reaven GM: Insulin resistance, glucose intolerance and hyperinsulinemia in patients with hypertension. Am .I Hypertens 2:419-423,1989 6. Ferrannini E, Buzzigoli G, Bonnadonna R, et al: Insulin resistance in essential hypertension. N Engl J Med 317:350-357, 1987 7. Rizza RA, Mandarin0 LJ, Gerich JE: Mechanisms of insulin resistance in man. Assessment using the insulin dose-response curve in conjunction with insulin-receptor binding. Am J Med 70:169-176,198l 8. Reichard G, Moury N, Hochella N, et al: Quantitative estimation of Cori cycle in humans. J Biol Chem 258:495-501,1963
324
9. Newgard CB, Foster DW, McCarty JD: Evidence for suppression of hepatic glucose-6-phosphatase with carbohydrate feeding. Diabetes 33:192-195,1984 10. Katz J, Rognstad R: Futile cycle in the metabolism of glucose. Cm-r Top Cell Regul 10:237-289,1976 11. Bell PM, Firth RG, Rizza R: Effects of hyperglycemia on glucose production and utilization in humans. Measurement with (23H)-, (33H)-, and (6%)glucose. Diabetes 35:642-648,1986 12. Efendic S, Karlander S, Vranic M: Mild type II diabetes markedly increases glucose cycling in the postabsorptive state and during glucose infusion irrespective of obesity. J Clin Invest 81:1953-1961,1988 13. Karlander S, Vranic M, Efendic S: Increased glucose turnover and glucose cycling in acromegalic patients with normal glucose tolerance. Diabetologia 29:778-783,1986 14. Ritchie CM, Atkinson AB: Towards better management of the diabetic patient with raised blood pressure. Diab Med 3:301305,1986 15. Breckenridge A, Welborn TA, Dollery CT, et al: Glucose tolerance in hypertensive patients on long-term diuretic therapy. Lancet 1:61-64,1967 16. Pollare T, Lithe11H, Berne C: A comparison of the effects of hydrochlorothiazide and captopril on glucose and lipid metabolism in patients with hypertension. N Engl J Med 321:868-873,1989 17. European Working Party on Hypertension in the Elderly: Glucose intolerance during diuretic therapy. Lancet 1:681-684, 1978 18. Bengtsson C, Blohme G, Lapidus L, et al: Do antihypertensive drugs precipitate diabetes? Br Med J 289:1495-1497,1984 19. Atkinson AB, Kellett RG: Value of intravenous urography in investigating hypertension. J R Co11Physicians Lond 8:175-181, 1970 20. Bell PM, Firth RG, Rizza RA: Assessment of insulin action in insulin-dependent diabetes mellitus using (6”C)glucose, (33H)glucase and (23H)glucose. J Clin Invest 78:1479-1486,1986 21. Steele R, Wall JS, De Bodo RC, et al: Measurement of site and turnover rate of body glucose pool by the isotope dilution method. Am J Physiol187:15-251956 22. Neely RDG, Rooney DP, Atkinson AB, et al: Underestimation of glucose turnover determined using (6’H)glucose in nonsteady state. The role of a tritiated tracer impurity. Diabetologia 33:681-687,199O 23. Somogyi M: Determination of blood sugar. J Biol Chem 160:69-73, 1945 24. Issekutz B: Studies on hepatic glucose cycles in normal and methylprednisolone treated dogs. Metabolism 26:157-170,1977
ROONEY ET AL
25. Hales CN, Randle PJ: Immunoassay of insulin with insulinantibody precipitate. Biochem J 88:137-146,1963 26. De Bodo RC, Steele R, Altszuler N, et al: On the hormonal regulation of carbohydrate metabolism: Studies with Cl4 glucose. Recent Prog Horm Res 19:445-448,1963 27. Katz J, Dunn A: Glucose 2-t as a tracer for glucose metabolism. Biochemistry 6:1-5, 1967 28. Karlander S, Roovete A, Vranic M, et al: Glucose and fructose 6-phosphate cycle in humans. Am J Physiol 251:E530E536,1986 29. De Fronzo RA, Jacot E, Jequier E, et al: The effect of insulin on the disposal of intravenous glucose: Results from indirect calorimetry and hepatic and femoral venous catheterization. Diabetes 30:1000-1007, 1981 30. Reaven GM, Chang H, Hoffman BB, et al: Resistance to insulin-stimulated glucose uptake in adipocytes isolated from spontaneously hypertensive rats. Diabetes 30~1155-1160,1989 31. Wajngot A, Chandramouh V, Schumann WC, et al: Testing of the assumptions made in estimating the extent of futile cycling. Am J Physiol256:E668-E675, 1989 32. Dunn A, Katz J, Golden S, et al: Estimation of glucose turnover and recycling in rabbits using various (‘H, “C)glucose labels. Am J Physiol230:1159-1162,1976 33. Butler PC, Rizza RA: Contribution to postprandial hyperglycemia and effect on initial splanchnic glucose cycling in glucoseintolerant or NIDDM patients. Diabetes 40:73-81, 1991 34. Hellerman JH, Elahi D, Andersen DK, et al: Prevention of the glucose intolerance of thiazide diuretics by maintenance of body potassium. Diabetes 32:106-111,1983 35. Bergstrom J, Hultman E: The effect of thiazides, chlorthalidone and furosemide on muscle electrolytes and muscle glycogen in normal subjects. Acta Med Stand 180:363-376, 1966 36. Randle PJ, Garland PB, Hales CN, et al: The glucose fatty-acid cycle. Its role in insulin sensitivity and the metabolic disturbances of diabetes mellitus. Lancet 1:7285-7289,1963 37. Garland PB, Newsholme EA, Randle PJ: Regulation of glucose uptake by muscle. Effects of fatty acids and ketone bodies and of alloxan-diabetes and starvation on pyruvate metabolism and on lactate/pyruvate and L-glycerol-3-phospbate/dihydroxy acetone phosphate concentration ratios in rat heart and rat diaphragm muscles. Biochem J 93:665-678, 1964 38. Ross BD, Hems R, Krebs HA: The rate of gluconeogenesis from various precursors in the perfused rat liver. Biochem J 102:942-951, 1967 39. Williamson JR, Kreisberg RA, Felts PW: Mechanism for the stimulation of gluconeogenesis by fatty acids in perfused rat liver. Proc Nat1 Acad Sci USA 56:247-254. 1966