Propranolol or Hydrochlorothiazide Alone for the Initial Treatment of Hypertension IV. Effect on Plasma Glucose and Glucose Tolerance VETERANS ADMINISTRATION COOPERATIVE STUDY GROUP ON ANTIHYPERTENSIVE AGENTS SUMMARY To evaluate the short-term and long-term effectiveness of propranolol and hydro- chlorothiazide monotherapy for hypertension, 683 hypertensive men were randomized to treatment with either propranolol or hydrochlorothiazide. Both drugs increased the average fasting plasma glucose level by approximately 5 mg/dl after 10 weeks (p < 0.001) and 1 year (p < 0.001) of treatment, but the elevation persisted only in the propranolol-treated group 1 month after discontinuing the year- long treatment (p < 0.01). A subset of 191 patients had 2-hour glucose tolerance tests. Hydrochloro- thiazide increased the average 2-hour oral glucose tolerance test result by 18.0 mg/dl after 10 weeks (p < O.OOl), an increase significantly higher than that induced by propranolol (p < 0.012). After 1 year of treatment, however, propranolol also increased the average 2-hour oral glucose tolerance test result (p < 0.05) and there was no significant difference between drugs. The hyperglycemic effects were dose-related, which suggests that both drugs should be administered at their lowest effective dosage. The clinical importance of the persistent fasting plasma glucose elevation in propranolol- treated patients 1 month after discontinuing treatment is unknown. (Hypertension 7: 1008-1016, 1985) KEY WORDS o hyperglycemia o side effects I T has been known for well over 20 years that thiazide diuretics elevate blood glucose levels and impair glucose tolerance in a substantial pro- portion of patients treated with therapeutic doses for hypertension. `I 2 These effects have been noted during both short-term and long-term treatment3, 4 and repre- sent a complicating factor in managing hypertensive patients with borderline or frank diabetes. 5 In addition, on the suspicion that the vascular lesions of diabetes might be related to the blood glucose leve1,6-8 it has been suggested that treatment with thiazide diuretics could enhance the vasculopathic potential of hyperten- sion by increasing blood glucose levels.4 P-Blockers have been suggested as an alternative to thiazides at the first level of the stepped-care scheme. With respect to this consideration, the development of diabetes has been mentioned as one of the specific disadvantages of thiazides as compared to P-blockers.9 From the Cooperative Studies Program, Medical Research Ser- vice of the Veterans Administration, Washington, DC. The article was written by Drs. Elf A. Ramirez and Frederick N. Talmers. Supported by a grant from Ayerst Laboratories, Inc.. New York. New York. Address for reprints: Eli A. Ramirez. M.D., F.A.C.P., Veterans Administration Medical Center, G.P.O., Box 4867. San Juan. Puerto Rico 00936. Received July 30, 1984; accepted May 29, 1985. o hypertensive therapy The main purpose of the present study was to evaluate the short-term and long-term effectiveness of propran- 0101 (PROP) as compared to hydrochlorothiazide (HCTZ) in the monotherapy of hypertension. Herein, we report on that aspect of the study that compares the effects of both drugs on fasting plasma glucose levels and glucose tolerance. Materials and Methods The design of this study has been described pre- viously.`", " Briefly, this was a seven-hospital, cooper- ative, double-blind, randomized clinical trial. Hyper- tensive men, aged 2 1 to 65 years, were admitted to the trial after appropriate informed consent if their sitting diastolic blood pressure (DBP) was in the range of 95 to 114 mm Hg after a minimum of 2 weeks without antihypertensive therapy. Subjects were excluded if they had a severe complication of hypertension, sec- ondary hypertension, a serious systemic disease, or a contraindication to treatment with PROP or HCTZ. Patients with diabetes mellitus were excluded if the diabetes was unstable, was of preadult onset, or re- quired treatment with insulin. No selection was at- tempted or made on the basis of family or personal history of diabetes, dietary intake, or other treatment for diabetes. 1008 HYPERGLYCEMIA WITH PROPRANOLOL AND HYDROCHLOROTHIAZIDE 1009 There was a prerandomization, single-blind placebo period of 4 weeks to obtain baseline data and to deter- mine eligibility, defined as an average DBP in the range of 95 to 114 mm Hg on two consecutive visits and pill counts of the placebo within a designated ac- ceptable range. Eligible patients were randomly as- signed in a double-blind fashion to receive either PROP or HCTZ to start the titration dose finding peri- od (phase A). At a maximum of seven clinic visits 1 or 2 weeks apart during a IO-week period, daily doses were titrated from 80 mg to 160,240,320,480, or 640 mg for PROP and from 50 mg to 100 or 200 mg for HCTZ. Both drugs were given in equal, twice-daily divided doses. The titration was continued until either a DBP below 90 mm Hg was achieved or there were side effects. No specific dietary restrictions were pre- scribed. To enter the long-term treatment period (phase B) of the study, the patient had to achieve either a DBP below 90 mm Hg on two consecutive phase A visits or an average DBP during the last two consecu- tive visits of phase A below 100 mm Hg and at least 6 mm Hg below the baseline average. The long-term treatment phase (phase B) consisted of 12 visits at 4-week intervals for a total of 48 weeks of continuous treatment. Additional adjustments to dosage were permitted during the long-term treatment phase according to DBP, compliance, and side effects. Subjects could be terminated for severe side effects or for lack of desig- nated antihypertensive effectiveness within specified periods. After the long-term treatment phase, there were two weekly visits for dose tapering and two fur- ther weekly visits for the final placebo period. Laboratory studies including the standard urinaly- sis, complete blood count, and blood chemistry tests using automated analyzers were done just before be- ginning the dosage titration phase, after its completion (phase A), at the end of the long-term treatment phase (phase B), and at the end of the 2-week final placebo phase. Two of the seven participating hospitals (Allen Park, MI; San Juan, Puerto Rico) volunteered to per- form 2-hour oral glucose tolerance tests (2-hour OGTT) at randomization, at the end of phase A, and at the end of phase B. Throughout the study, the labora- tories at all the participating hospitals satisfied the quality control requirements of the American College of Pathologists. The patients who had a 2-hour OGTT received a 300-g carbohydrate diet for 3 days before the test and were instructed to take only water after 1900 hours on the night before the test. The next morning, a fasting blood sample was drawn and, if it was less than 150 mg/dl, the subject was given 7 fl oz of Glucola contain- ing 75 g of glucose. A repeat plasma glucose sample was drawn 2 hours later. True glucose values were determined using the glucose oxidase method. This study was designed by a committee that includ- ed biostatisticians, some of whom participated in the analysis of the data and in the monitoring of the study. Paired and unpaired Student's t tests, Pearson r corre- lations, chi-square tests, and analyses of covariance were used to assess statistically significant differences (defined as p < 0.05) between groups of data. Results Of the 906 subjects entering the study, 683 (75.4%) were randomized; 340 to PROP and 343 to HCTZ. The most common causes for prerandomization dropout were noncompliance and blood pressure above or below the randomization criteria. Only four subjects were taking oral hypoglycemic agents at randomiza- tion; one was assigned to PROP and three to HCTZ. Blacks constituted 57.6% of the patients randomized to PROP and 56.0% of those randomized to HCTZ; the difference was not significant. Some clinics had a higher percentage of blacks than others, but the racial composition between the drug groups was not signifi- cantly different within any of the clinics. Phase A was completed by 610 subjects; 298 re- ceived PROP and 3 12 received HCTZ. More subjects receiving HCTZ (128) than PROP (102) completed phase A before the full 10 weeks was over, because they achieved earlier a DBP less than 90 mm Hg on two consecutive visits; therefore, the average duration of phase A was shorter for HCTZ-treated than for PROP-treated subjects. Of the 6 10 subjects who com- pleted phase A, only 491 entered the study early enough to be eligible for long-range treatment. Of this number, 394 met the criteria for entering the long-term treatment period (phase B) of the study; 182 received PROP and 2 12 received HCTZ. Phase B was complet- ed by 125 subjects receiving PROP and by 177 receiv- ing HCTZ. The baseline characteristics of all randomized pa- tients are shown in Table 1; there were no significant differences between the two groups. The baseline var- iables were also studied for the subgroups of subjects that completed phase A and phase B. There were no significant differences between these subgroups either, except for a lower baseline serum calcium level in the PROP-treated subgroup that completed phase B @ = 0.037). Reasons for terminations were classified as either administrative (e.g., failure to return to clinic, stop- ping treatment) or medical (e.g., inadequate DBP con- trol, side effects). The administrative terminations were equally divided between both groups, but the medical terminations, particularly inadequate DBP control, were more frequent in the PROP-treated group. The total number of terminations in the PROP- treated subjects compared with the HCTZ-treated sub- jects were 42 to 3 1 during phase A and 57 to 35 during phase B. This aspect of the study has been described in previous communications. `OS `I No subject was terminated for a diabetes-related reason. Only one subject was started on a regimen of oral hypoglycemic agents de novo during the study. As the study progressed, the total number of subjects re- ceiving oral hypoglycemic agents was very small: one at the end of phase A, two at phase B, and one at the final placebo phase. All had been randomized to HCTZ. 1010 HYPERTENSION VOL 7, No 6, NOVEMBER-DECEMBER 1985 TABLE 1. Comparison of Baseline Characteristics of the Patient Populations Randomized to Propranolol and H~drochl~rothiazide af All Hospitals ~opranolol No. of Variable subjects Mean f SD Age (~6 340 49.61t9.8 Black (%) 340 57.6 Weight (lb) 340 191.5 + 33.4 Diastolic blood (mm Hg) pressure 340 101.6t4.6 Systolic blood pressure (mm Hg) 340 146.0r 14.4 Creatinine (mg/dl) 339 1.18?0.20 Potassium (m&/L) 339 4.2 LO.39 Uric acid (mgldl) 335 6.4L 1.31 Cholesterol (mg/dl) 335 221.1247.4 Triglycerides (mgidl) 336 165.8t: 140.6 Calcium (mgfdl) 257 9.5rt;oso Fasting plasma glucose fmgidl) 337 lOO.O+. 22.7 There were no signi~cant differences between groups. Hydrochlorothiazide No. of subjects Mean I SD 343 49.8~9.9 343 56.0 343 189.4 t 30.9 343 101.3 k4.5 343 146.5 I 15.8 339 1.16+0.21 342 4.3 10.68 342 6.5-c 1.45 336 224.3247.0 338 184. I +- 197.8 259 9.4rto.90 343 100.4rt:25.2 The average dose of medication found necessary to control blood pressure in each group at the end of phase A was 268 mgiday for PROP and 93 mglday for HCTZ. During phase B, dosage increases to maintain blood pressure control were necessary in 37.6% of PROP-treated and in 20.9% of HCTZ-treated patients; dosage decreases were necessary in 0.8% of PROP- treated and in 4% of HCTZ-treated patients. At the two participating hospitals that performed the 2-hour OGTT, 96 subjects were randomized to PROP and 95 to HCTZ. Whites outnumbered blacks 53 to 43 in the group randomized to PROP and 56 to 39 in the group randomized to HCTZ. Eight-eight subjects re- ceiving PROP and 89 receiving HCTZ completed phase A, and 38 subjects receiving PROP and 49 re- ceiving HCTZ completed phase B. The baseline characteristics were studied at the two hospitals for the randomized groups and for the sub- groups that completed phase A and phase B. There were no significant differences between them. Howev- er, there were several statistically significant differ- ences in the baseline characteristics between the sub- jects at the two hospitals (Table 2). The subjects at Allen Park were significantly younger and heavier and had significantly higher average serum cholesterol and serum calcium levels than those at San Juan. On the other hand, the San Juan subjects had significantly higher systolic blood pressure and DBP, serum triglyc- eride levels, fasting plasma glucose (FPG) levels, and a-hour OGTT results. Fasting Plasma Glucose Levels Table 3 shows the average FPG level at all the hospi- tals for all phases and the differences between each phase and the baseline value of the same group of patients. For all subjects, the FPG level increased sig- TABLE 2. Significant* Baseline Characteristic Differences at the Tw Hospitals that Performed ~-HOW Oral Glucose Tolerance Tests Allen Park, Michigan San Juan, Puerto Rico No. of No. of Variable subjects Mean It SD subjects Mean + SD P Age (yr) 91 49.0-c 10.2 100 52.3-cl.7 0.014 Black (%) 73.6 15.0 Weight (lb) 91 190.1?30.7 100 178.0t25.8 0,003 Diastolic blood (mm Hg) pressure 91 102.4+5.2 100 103.8+3.6 0.047 Systolic blood pressure (mm Hg) 91 139.62 12.0 100 149.1r11.5 0.000 Cholesterol (mg/dl) 91 216.7&39.7 97 203.8T48.4 0.048 Triglycerides (mg/dl) 91 157.52 123.5 100 240.62 192.5 0.000 Calcium (mg/dl) 91 9.6eO.5 100 9.4+0.5 0.029 Fasting plasma glucose (mg/dl) 90 93.71- 17.5 loo 107.8+ 27.4 O.ooO 2-hour OMIT (mgldl) 88 105.3 2 39.0 95 141.5-c65.1 0.000 OGTT = oral glucose tolerance test. *p < 0.05. HYPERGLYCEMIA WITH PROPRANOLOL AND HYDROC~LOROTHlAZlDE 1011 TAMLE 3. Averuge Fasting Plasma Glucose Values at AII Hospitals for the Various Phases All patients Propranolol HCTZ No. of No. of No. of p (between Variable subjects Mean 2 SD subjects Mean + SD subjects Mean 2 SD drugs) Phase A 534 104.Ok21.8 259 102.3+20.2 275 105.61t23.1 NS A baseline 532 4.Ok 20.3* 257 4.02 17. I* 275 4.0r22.9: NS Phase B 294 106.2 2 26.6 120 105.9k21.8 174 104.4+ 29.5 NS A baseline 293 5.4*25.1* 119 6.4? 18.4* 174 4.7 c28.9$ NS A from phase A 256 1.6+20.0 103 3.1 -t 17.9 153 0.6rt21.4 NS Final placebo 265 102.3k23.2 110 105.91-22.9 155 99.7223.2 0.032 A baseline 264 0.7k23.7 109 4.62 18.3t 155 -2.Ok26.5 <0.017 Chance represents the difference between a phase and the respective group baseline average. HCT% = ~~d~o~hlo~othia~ide; NS = not s&&cant @ > 0.k) *p < 0.001, tp < 0.01, $p < 0.05, within groups. nificantly from the respective group baseline, in both phase A and phase B @ < 0.00 1). Each drug group by itself also showed significant FPG level increases from its baseline in both phases. The average increases in FPG level were relatively small: 4.0 mgidl for each drug in phase A, and 4.7 mg/dl for PROP and 6.4 mgidl for HCTZ in phase B. When the drug groups were compared with each other, neither the average values in each phase nor the changes between phases were significantly different. However, compared with their respective group baseline, the FPG level re- mained significantly elevated in the 109 PROP-treated subjects who completed the final placebo period, whereas the FPG level in the HCTZ-treated group re- turned to its baseline. The difference was also signifi- cant between drugs, between both the levels @ = 0.032) and the changes from baseline @ < 0.017). The number of subjects who crossed diabetes diag- nostic limits was studied according to accepted stan- dards that define the FPG level as definitely normal below 115 mg/dl and definitely abnormal above 139 mgidl . `* The percentage change within each of these categories in those subjects who completed each phase is shown in Figure 1. Of 119 subjects taking PROP who completed phase B, there were 12% fewer with a definitely normal FPG level than at baseline @ = 0.016). There were no other statistically significant changes in the percentages. Two-Hour Glucose Tolerance Test Table 4 shows the average 2-hour OG'IT results for the two hospitals at randomization and at each phase. The differences between each phase and the baseline average of the respective group of subjects are also shown. For all subjects, there were significant average increases from baseline to phase A @ = O.OOS), from baseline to phase B @ = 0.002), and from phase A to phase B @ = 0.003). Values in HCTZ-treated subjects increased significantly from baseline both to phase A @ = 0.001) and to phase B @ = 0.024), while values in PROP-treated subjects increased significantly only from baseline to phase B @ = 0.014). Analyses of between-drug differences indicated that the HCTZ- treated subjects exhibited a significantly higher level fPG< 115 mgldl FPG=-139 mB/dt 40+- O PROPRANOLOL O ~YDROCHLOROTH~~IDE 257 275 119 174 109 155 PHASE A 8 FINAL P~CEBO A B FINAL PLACEBO CHANGES FROM BASELINE WITHIN GROUPS: * =p<.O2 FIGURE 1. Changes in the prevalence of,fasting plasma glu- case (FPG) levels less than I I5 mgldl and more than 139 mgidl in those subjects completing phase A, phase B, andfinal piace- bophase. N = number of subjects in each group. Data arefrom all seven hospitals. @ = `0.007) and a greater change from baseline @ = 0.012) than the PROP-treated subjects during phase A. There were no significant differences between drugs during phase B. The number of patients who crossed diabetes diag- nostic limits was studied according to accepted stan- dards that define the 2-hour OGTT as definitely normal below 140 mg/dl and definitely abnormal above 199 mg/dl . `* The percentag e change in each of these cate- gories in those subjects who completed each phase is shown in Figure 2. In 80 HCTZ-treated subjects com- pleting phase A, there were 21% fewer with a definite- ly normal @ = O.OOOl), and 11% more with a defi- nitely abnormal @ = 0.027), 2-hour OGTT result than at baseline. Both differences were also significant be- tween drugs Cp = 0.0002 and p = 0.039 respective- ly). During phase B the only significant difference was in the 45 HCTZ-treated subjects: 22% fewer subjects had a definitely normal 2-hour OGTT result than at baseline (p = 0.016). However, this difference was not significant between drugs. 1012 HYPERTENSION VOL 7, No 6, NOVEMBER-DECEMBER 1985 TABLE 4. Average Z-Hour Oral Glucose Tolerance Test Values at Randomization and the Various Phases, Combinedfor the Two Hospitais All patients Propranolol HCTZ No. of No. of No. of p (between Variable subjects Mean + SD subjects Mean t SD subjects Mean 2 SD dws) R~dom~zation 209 123.0_~55.6 90 119.5148.3 93 128.5k64.3 NS Phase A 156 131.4r58.5 15 118.5 245.4 81 143.2266.6 0.008 A baseline 154 9.5 143.F 74 0.4236.2 80 18.0t48.4.l' 0.012 Phase B 80 155.8-c90.9 35 147.7279.2 45 162.1 k99.4 NS A baseline 79 22.5?61.9* 34 18.1*40.8$ 45 25.9k74.35 NS A from phase A 75 13.7 -c54.8* 32 18.4k59.9 43 10.8'51.3 NS Change represents the difference between a phase and the respective group baseline average. HCTZ = hydrochlorothiazide; NS = not significant @ > 0.05). *p < 0.01, tp < 0.001, $p < 0.05, within groups. 2 HR KillclOD mg/dl 2 HR OGlT 3r 199 mQ!dl J--J-- ,0 b FROPRANOLOl _HYDA~CHLORO~I~~~E 14 80 3445 A 8 PHASES A 8 CHANGES ROM ~~NEWI~IN GROUPS:* =p-= cb: tyi-= 001 CHANGES BElVREN DRUGS: t-p<& _-o-.001 FKXJRE 2. Changes in prevalence of2-hour oral glucose tol- erance test (OGU) results less than 140 mgidl of glucose and more than 199 mgidl of glucose in those putients completing phases A and B. N = number of subjects in each group. Data are from two hospitals (Allen Park, MI and San Juan, PRJ. Relationship with Drug Dosage Figure 3 shows the average changes in piasma glu- cose level from the respective group baseline through phases A and B according to dosage. Subjects that were taking more than 320 mgidl of PROP had a significantly greater increase from baseline in FPG level @ < 0.001) and in 2-hour OGTT results @I < 0.05) during phase B than did those who were taking a lower dosage. There were no significant differences between these dosage subgroups from baseline to phase A. Subjects taking 100 or 200 mg of HCTZ per day had a significantly greater increase in FPG level from base- line during both phase A @I < 0.00 1) and phase B 0) < 0.05) than did those who were taking 50 mg of HCTZ per day. The 2-hour OGTT result increases were great- er in the higher dosage subgroup during both phases, but the differences were not significant. However, there were only seven subjects in the low dosage sub- group at phase B. FPG PFw +30 I 7 N 118 139 63 56 135 139 91 79 PHASES A B A B OIFFERENCES BEIWEEN DOSAGE LEVELS .=PCG5 t =P-=Wl 2HROGn PROP , wli! A B A B O LOW DOSAGE RANGE !!!@ = mgfday FIJXTJ 50 Ea H'GH tgE luaGE 480~ 100-200 FIGURE 3. Changes in fasting plasma glucose (FPG) levels and 2-hour oral glucose tolerance test (OGTT) results in those subjects completing phase A and phase B grouped by low and high dosage of propranolo~ (PROP) and ~l~ldrochloroth~az~de (HC7Z). N = number of subjects in each group. The FPG data are from all hospitals; 2-hr OG7T data are from two hospitals (Allen Park, MI and San Juan, PR), Other Analyses Analyses of covariance were performed to test the significance of the differences between the drugs in terms of plasma glucose level after adjustment for the effect of interhospital differences, serum triglyceride levels, body weight, and race. The results for values at the various phases and the interphase changes are shown in Table 5. The differences remained significant for FPG level during phase A and for the change from baseline to final placebo phase. They also remained significant for the 2-hour OGTT during phase A and for the change from baseline to phase A. The relation between the plasma glucose level and serum potassium concentration was studied during each phase and for the interphase changes. Table 6 shows that there was a significant inverse correlation between serum potassium concentration and FPG level in HCTZ-treated subjects during phase A and phase B and for most interphase differences. However, there HYPERGLYCEMIA WITH PROPRANOLOL AND HYDROCHLOROTHIAZIDE 1013 TABLE 5. Analysis of Covariance Between Plasma Glucose Level and Other Selected Variables Demonstrating Signiji- cant Associations After Adjustment for D#erences Due to the Other Variables Baseline Phase Phase B Final Variables Baseline A B Phase A Phase B to A placebo FPG - ail hospitals No. of subjects 845 520 286 532 293 240 264 Hospital