A Calcium Antagonist vs a Non–Calcium Antagonist Hypertension Treatment Strategy for Patients With Coronary Artery Disease:  The International Verapamil-Trandolapril Study (INVEST): A Randomized Controlled Trial FREE

The prevalence of coronary artery disease (CAD) is increasing, as are major CAD risk factors (hypertension, aging, diabetes, obesity, and inactivity).¹ Blood pressure is important in the progression of CAD, yet no large trials have evaluated blood pressure management in only patients with CAD. Thus, blood pressure management in CAD patients² must be guided by data from relatively small subsets of high-risk patients from other trials.^3- 6

Current trends in hypertension management emphasize multidrug regimens rather than monotherapy. Combinations of antihypertensive drugs with complementary actions may minimize adverse effects and reduce clinical outcomes by improving blood pressure control and organ protection.^7- 10 β-Blockers are effective in hypertension treatment and reduce incidence of death and reinfarction in patients who have had a myocardial infarction (MI).¹¹ Along with diuretics, β-blockers became the standard of care for hypertensive CAD patients.¹² However, β-blockers may be less effective antihypertensive agents in older patients, who are also more likely to have CAD.^13- 14 The possibility that other antihypertensive regimens, particularly those containing calcium antagonists and/or angiotensin II active agents, might be as or more effective than β-blocker and/or diuretic regimens has not been convincingly demonstrated.^15- 16 Previous trials were performed predominantly in populations with low frequencies of CAD and used dihydropyridine calcium antagonists.^{5- 6,16- 17} A recent trial in high-risk hypertensive patients showed that a combination of an angiotensin II-receptor blocker and a diuretic was more effective than a combination of a β-blocker and a diuretic.¹⁴

Heart rate–reducing nondihydropyridine calcium antagonists, on the other hand, have rarely been studied in large randomized hypertension trials,^18- 21 although verapamil appears to reduce the risk of death and reinfarction in acute CAD trials.²⁰ The combination of a nondihydropyridine calcium antagonist and an angiotensin-converting enzyme (ACE) inhibitor may provide better blood pressure control and organ protection than monotherapies.^22- 25 Many recent trials^{8,10,14,24,26- 30} indicate that drugs influencing the actions of angiotensin II can be beneficial in high-risk patients, but no hypertension trial has prospectively used these agents for CAD patients with diabetes, renal impairment, or heart failure.

We designed a randomized trial, the International Verapamil-Trandolapril Study (INVEST), to compare outcomes in older hypertensive patients with CAD treated with a calcium antagonist strategy (CAS; verapamil sustained release [SR]) or a non–calcium antagonist strategy (NCAS; atenolol). Because most older hypertensive patients require more than 1 agent to adequately control blood pressure, INVEST was intended to compare multidrug strategies rather than individual agents.

The INVEST design and methods have been published.³¹ INVEST was an international, multicenter study with a prospective, randomized, open blinded end-point evaluation design³² conducted according to principles of the Declaration of Helsinki. The institutional review boards and ethics committees at participating sites approved the protocol and patients provided written informed consent.

We tested the hypothesis that risk for adverse outcomes is equivalent to a verapamil SR–based regimen compared with an atenolol-based regimen. Clinically stable CAD patients with hypertension were randomly assigned to either verapamil SR or atenolol for blood pressure treatment according to the sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC VI) (target: systolic blood pressure [SBP] <140 and diastolic blood pressure [DBP] <90 mm Hg or SBP <130 mm Hg and DBP <85 mm Hg when diabetes or renal impairment is present).⁹ Addition of trandolapril and/or hydrochlorothiazide was recommended when necessary to achieve blood pressure goals. Trandolapril also was recommended for patients with heart failure, diabetes, or renal insufficiency. Thus, this was not simply a comparison of verapamil SR with atenolol because it was anticipated that few patients would be treated with only those drugs. Ultimately, it was expected that most would be using the combination of verapamil SR plus trandolapril or atenolol plus hydrochlorothiazide.

Patient inclusion and exclusion criteria have been described previously.³¹ Briefly, patients were eligible if they were aged 50 years or older and had documented CAD, with essential hypertension as defined by JNC VI⁹ requiring drug therapy. Documented CAD was defined as any of the following: remote (≥3 months prior to enrollment) confirmed MI, coronary angiogram with more than 50% narrowing of at least 1 major coronary artery, diagnosis of classic angina pectoris, or concordant abnormalities on 2 different types of signals (electrocardiograms, echocardiograms, and/or radionuclide scans) from stress tests provided that 2 different signals showed findings consistent for ischemia (eg, ST-segment depression and/or perfusion defects by radionuclide, and/or wall-motion abnormalities by echocardiogram or radionuclide). Patients with heart failure classes I through III were included. Patients taking β-blockers within 2 weeks of randomization or taking β-blockers for an MI that occurred in the previous 12 months were excluded to avoid withdrawal phenomena in patients randomized to the CAS group.

Following validity checks of eligibility data, an Internet-based management system automatically randomized each patient to a treatment strategy. The randomization scheme used a standard C routine and blocked by site using randomly permuted block sizes of 4 and 6. The randomization result was automatically stored in the central database as part of the patient's record and was also returned to the site investigator for electronic signature of strategy drugs in accordance with the protocol.

Figure 1 outlines the protocol-recommended treatment schedule for each strategy to achieve JNC VI blood pressure targets.⁹ The blood pressure target was determined from a mean of 2 sitting cuff blood pressure measurements as described in JNC VI.⁹

Patients allocated to the CAS group were given 240 mg/d of verapamil SR while patients allocated to the NCAS group were given 50 mg/d of atenolol (step 1). If patients did not achieve target blood pressure, in step 2 the CAS group also could receive trandolapril (an ACE inhibitor) and the NCAS group also could receive hydrochlorothiazide. The rationale for this was to maximize use of the combination of calcium antagonist and ACE inhibitor while minimizing diuretic use for the CAS group and maximizing use of the combination of β-blocker and diuretic for NCAS group. In step 3, doses were increased in both groups. In step 4, the CAS group also could receive hydrochlorothiazide and the NCAS group also could receive trandolapril. Trandolapril was recommended for all patients with renal impairment, diabetes, or heart failure.⁹ If the dose was not well tolerated or the target blood pressure was not achieved, verapamil SR could be titrated to between 120 and 480 mg/d and atenolol could be titrated to between 25 and 200 mg/d. The recommended starting dose for trandolapril was 2 mg/d and it could be titrated to between 0.5 and 8 mg/d. For patients in the CAS group, a fixed combination was available for verapamil SR and trandolapril in doses of 180 mg/d and 2 mg/d, respectively; 240 mg/d and 1 mg/d; and 240 mg/d and 4 mg/d. The recommended starting dose for hydrochlorothiazide was 25 mg/d and it could be titrated between 12.5 and 100 mg/d. Doses greater than 25 mg of hydrochlorothiazide were provided to limit the need for nonstudy diuretics in patients with heart failure or edema. If the blood pressure goal was not achieved and adverse effects had not occurred, doses were titrated to those shown in Figure 1 before a patient was moved to the next step.

Additional nonstudy antihypertensive drugs, except β-blockers for CAS patients and calcium antagonists for NCAS patients, could be added when needed to reach blood pressure targets or minimize adverse effects. Patients were considered to have crossed over from their randomized treatment strategy if they received a β-blocker during the trial and were in the CAS group or received a calcium antagonist and were in the NCAS group. Standard of care nonpharmacological JNC VI guidelines⁹ and secondary prevention according to the National Cholesterol Education Program were provided online to physicians, which could be printed and given to patients.

Protocol visits were scheduled every 6 weeks for the first 6 months and then biannually until 2 years after the last patient was enrolled. Patients were assessed for response to treatment, occurrence of symptoms, treatment compliance, and adverse effects at each visit and at study close as detailed elsewhere.³¹

Patient follow-up was complete when a final assessment form was received via the online data system or a death report was received. For all patients not completing the final assessment visit, lost to follow-up, or withdrawn, data were censored according to last visit date.

The primary outcome was the first occurrence of death (all-cause), nonfatal MI, or nonfatal stroke by intention-to-treat analysis. The MI and stroke definitions are detailed on the INVEST Web site.³³ These 3 components individually were the main secondary outcomes. Additional outcomes included time to most serious event (ranked from death as most serious, to MI, to stroke as least serious), cardiovascular death (definite or presumed), angina, cardiovascular hospitalizations, blood pressure control, cancer, Alzheimer disease, Parkinson disease, and gastrointestinal tract bleeding.³⁴ Shortly after the study started, new information became available on the potential for ACE inhibitors to prevent or delay the onset of diabetes.^8,10 Accordingly, at the recommendation of the independent data safety and monitoring committee, new diagnosis of diabetes was added as an outcome early in the recruitment phase of the study.

Outcomes such as death, MI, stroke, and cardiovascular hospitalization were reported within 24 hours using the online adverse event reporting system and then appropriate documentation was gathered. Adverse experiences were collected from responses to open, active questioning not restricted to those events known to be associated with the drugs taken. Three members of the events committee, masked to treatment assignment, confirmed all outcome events by reviewing documentation and other pertinent patient records. The data safety and monitoring committee reviewed efficacy and safety data at regular intervals throughout the trial.

It was decided a priori that a 20% difference in primary outcome between the treatment strategies would be clinically relevant³¹ using the intention-to-treat population. Therefore, the equivalence bound for the risk ratio was a confidence interval (CI) of 1.20 to 0.83. We assumed an annual primary outcome rate of no less than 2%,³¹ an α of .05 (2-sided), and 90% power when estimating the number of patients required. On this basis, a tentative sample size of 27 000 patients was calculated, with an anticipated yearly drop-out rate of 5% to 10%. Because the enrollment period was longer than initially planned, patient-years of follow-up were greater than those used for initial power estimates. At the recommendation of the INVEST study biostatisticians and the data safety and monitoring committee, the steering committee reduced the sample size to 22 000 patients.

All of the main analyses were completed as specified in the protocol with the intention-to-treat population, including patients withdrawn or lost to follow-up censored at the time of the last visit (unless the patient was known to be dead based on death records). One planned interim analysis was performed in August 2001 and the prespecified stopping rules³³ were not met.

The final significance level for the primary outcome, adjusted for the single interim analysis, was P = .04806 for a 2-sided test. For the secondary outcomes of death, nonfatal MI, and nonfatal stroke, a Bonferroni adjustment was made to the same P = .04806 significance level (P = .02 for each outcome). All other analyses are reported at the P<.05 significance level. Kaplan-Meier survival analysis was used to assess time to first event for the primary outcome and the main secondary outcomes. The primary outcome was analyzed both unadjusted and adjusted for 5 prespecified covariates: age, race, sex, previous MI, and prior heart failure. Standard relative risk (RR) estimates and 95% CIs were also calculated.

χ² Analysis was used to compare CAS with NCAS on percentage occurrence of different outcomes. Cox proportional hazard models were used to evaluate potential interactions in the reported prespecified subgroup analyses (by baseline characteristic). All data were captured and stored in database tables (Version 7.1, Oracle, Redwood Shores, Calif). Data management and statistical analyses were performed using SAS statistical software (Version 8.2, SAS Institute Inc, Cary, NC). The database was maintained at the University of Florida, Division of Biostatistics, Gainesville.

The pilot phase (30 selected sites) started in September 1997. Full-scale site recruitment and patient enrollment began in January 1998, and patient follow-up was completed on February 14, 2003. A total of 22 576 patients at 862 sites in 14 countries provided informed consent, satisfied administrative requirements, and completed randomization; 11 267 were assigned to the CAS group and 11 309 to the NCAS group (Figure 2). A total of 594 patients had all assigned drugs withdrawn due to an adverse experience. A total of 568 patients failed to return for final assessment and did not appear in death searches (withdrawals or lost to follow-up). These latter patients were censored at the time of their last visit. Mean follow-up was 2.7 years (range, 1 day to 5.4 years) in each strategy. A total of 30 829 patient-years were accumulated in the CAS group and 31 006 patient-years in the NCAS group.

At baseline, patient characteristics were well-balanced (Table 1). The study population included a large proportion of elderly, Hispanic, diabetic, and female patients. Blood pressure levels were similar between groups (Table 2). Overall, only 4267 patients (18.9% of all patients) had controlled blood pressure.

At 24 months, 6391 (81.5%) of CAS patients were taking verapamil SR and 6083 (77.5%) of NCAS patients were taking atenolol (Table 3). As expected from the recommended order of additional drug treatment by strategy, usage of trandolapril and hydrochlorothiazide differed significantly (P<.001). The distribution of number of study drugs used was similar between strategies as was the distribution of total antihypertensive medications. At 24 months, only 2.1% of patients in each group (CAS, 145; NCAS, 141) were taking no antihypertensive medications. At final assessment, nonstudy antihypertensive drug use was observed in 5873 patients (43%) in both strategies (Table 4). As expected, calcium antagonist use was more frequent in the CAS group and β-blocker use was more frequent in the NCAS group (Table 4). Crossover to β-blocker use in the CAS group (373 [5.5%]) was less than crossover to calcium antagonist use in the NCAS group (479 [7.0%]). This difference persisted over the entire duration of follow-up, β-blocker use at any time in the CAS group was 1305 (11.6%) of 11 267 and calcium antagonist use in the NCAS group was 1862 (16.5%) of 11 309 (P<.001). Nonstudy diuretic use was also more frequent in the NCAS group. The percentage of patients taking antidiabetic medications was significantly lower in the CAS group (23.2%; n = 1574) compared with the NCAS group (24.7%; n = 1682) (P = .04). The frequencies of other medication use were similar between strategies (Table 4).

Figure 3 presents SBP and DBP data by treatment strategy over 48 months (error bars in the upward direction represent 1 SD for CAS and bars in the downward direction represent 1 SD for NCAS). Mean (SD) SBP reduction at 24 months was 18.7 (22.2) mm Hg in the CAS group compared with 19.0 (22.6) mm Hg in the NCAS group (P = .41). The mean (SD) DBP reduction at 24 months was 10.0 (12.4) mm Hg in the CAS group compared with 10.2 (12.4) mm Hg in the in the NCAS group (P = .26). A reduction of 90% of the maximum achieved in SBP and 100% in DBP occurred in the first 6 months of treatment; the reductions were maintained throughout the trial. Mean resting heart rate at 24 months was significantly lower (P<.001) in NCAS patients (69.2/min) compared with CAS patients (72.8/min).

Figure 4 presents the primary outcome and Figure 5 presents the patients with events comprising the primary and other outcomes. A total of 2456 events were reported (CAS, 1214; NCAS, 1242) and the events committee confirmed 2380 of the events (96.9%; CAS, 1171; NCAS, 1209). Sites reported that 2333 patients (CAS, 1153; NCAS, 1180) experienced an event in the primary outcome cluster during follow-up and the events committee confirmed that an event had occurred in 2269 of those patients (97.3%; CAS, 1119; NCAS, 1150). Death (all-cause) occurred in 1766 patients (CAS, 873; NCAS, 893); 304 were nonfatal MIs (CAS, 151; NCAS, 153); and 279 were nonfatal strokes (CAS, 131; NCAS, 148). Of the 1766 confirmed deaths, 862 were classified as definitely or presumed cardiovascular (CAS, 431; NCAS, 431); 701 were noncardiovascular (CAS, 350; NCAS, 351); and 203 could not be classified (CAS, 92; NCAS, 111). Of the 1563 classifiable deaths, 862 (55.2%) were cardiovascular. The analyses reported herein were performed only on events confirmed by the events committee, but analyses of site-investigator reported events yielded similar results (data not shown).

Kaplan-Meier analysis (unadjusted) of time to first primary outcome event demonstrated no difference comparing the CAS group with the NCAS group for a primary outcome (Figure 4; RR, 0.98 [95% CI, 0.90-1.06]). A sensitivity analysis in which the 568 patients who were lost to follow-up or withdrew were all presumed to have died produced an RR of 1.00 (95% CI, 0.94-1.08). When adjusted for the prespecified covariates of age, race, sex, previous MI, and previous heart failure, the CAS and NCAS groups were not different (hazard ratio [HR], 0.98; 95% CI, 0.91-1.07; P = .69). Other outcomes were also similar in frequency between strategies (Figure 5). Time to death (all cause) did not differ between treatment groups (P = .72), nor did time to nonfatal MI (P = .95), or time to nonfatal stroke (P = .33). Time to the most serious event also did not differ between treatment groups (P = .58). Fatal and nonfatal MI occurred in 452 CAS patients (4.01%) and 441 NCAS patients (3.90%) (RR, 1.03; 95% CI, 0.90-1.17). Fatal and nonfatal stroke occurred in 176 CAS patients (1.56%) and 201 NCAS patients (1.78%) (RR, 0.88; 95% CI, 0.72-1.07). Subgroup analyses by baseline characteristics showed consistency for the primary outcome in both high- and low-risk subgroups (Figure 6). Of particular note were the similar event rates for each strategy among patients with prior MI as well as those with prior coronary revascularization. The exception was patients with prior heart failure, for which those assigned to the NCAS strategy appeared to have fewer events (P = .03 for interaction). Also important was the marked difference in the event rate of 14.3% (913/6400) for those with diabetes compared with 8.4% (1356/16 176) for those without diabetes.

The effect of the treatment strategies using an overall SBP control goal of less than 140 mm Hg and DBP control goal of less than 90 mm Hg was similar. A total of 5625 patients (71.7%) in the CAS group and 5553 (70.7%) in the NCAS group achieved overall blood pressure control at 24 months (P = .18). Based on JNC VI blood pressure goals, SBP control was achieved by 65.0% of CAS patients (n = 5093) compared with 64.0% of NCAS patients (n = 5025) (P = .23); DBP control was achieved by 88.5% of CAS patients (n = 6937) compared with 88.1% of NCAS patients (n = 6914) (P = .46).

At baseline, angina was reported in 66.2% of CAS patients (n = 7463) compared with 67.0% of NCAS patients (n = 7582). At 24 months, these percentages decreased to 27.3% in the CAS group (n = 2055) and 28.3% in the NCAS group (n = 2136) (P = .18). Angina and unstable angina were infrequently reported as adverse experiences and rates were similar in both groups (Table 5). At baseline (based on the previous 4 weeks), there was a mean (SD) of 1.5 (2.33) angina episodes/wk in the CAS group and 1.5 (2.43) in the NCAS group. At 24 months, angina episodes decreased in both groups, but the mean (SD) frequency was lower in the CAS group (0.77 [1.31] episodes/wk) compared with the NCAS group (0.88 [1.62] episodes/wk) (P = .02). Revascularization was required in only 2% of patients in each group (Table 5). Nitrate use was the same in each strategy (Table 4).

Analysis of the development of diabetes revealed significant differences between the treatment strategies. Of the 8098 CAS patients without diabetes at entry, 569 (7.03%) were diagnosed as having diabetes during follow-up. Of the 8078 NCAS patients without diabetes at entry, 665 (8.23%) were diagnosed as having diabetes during follow-up (RR, 0.85; 95% CI, 0.77-0.95). Patients in the CAS group were also less likely to die or develop diabetes compared with patients in the NCAS group (1050 [12.97%] vs 1177 [14.57%]; RR, 0.89; 95% CI, 0.82-0.96) and less likely to have an event in the primary outcome cluster or develop diabetes (1185 [14.63%] vs 1313 [16.25%]; RR, 0.90; 95% CI, 0.84-0.97). To explore possible explanations for reduced risk of diabetes, we conducted preliminary analyses adjusting for the 5 prespecified baseline covariates (age, race, sex, prior MI, and prior heart failure) and included factors for average daily dose of add-on medication (trandolapril and/or hydrochlorothiazide). In these analyses, trandolapril appeared to confer a protective effect in the CAS group. Compared with those in the NCAS group not taking either trandolapril or hydrochlorothiazide, those in the CAS group not taking trandolapril had a HR of developing diabetes of 0.95 (95% CI, 0.82-1.10). A 2-mg dose of trandolapril was associated with a HR of 0.86 (95% CI, 0.74-1.00) and a 4-mg dose was associated with a HR of 0.77 (95% CI, 0.62-0.96). In the NCAS group, a 2-mg dose of trandolapril was associated with a HR of 0.99 (95% CI, 0.90-1.08) and a 4-mg dose was associated with a HR of 0.98 (95% CI, 0.82-1.18). On the other hand, hydrochlorothiazide appeared to confer a nonstatistically significant increased risk of diabetes. Compared with those in the NCAS group not taking either trandolapril or hydrochlorothiazide, the addition of 12.5 mg of hydrochlorothiazide was associated with a HR of 1.17 (95% CI, 1.09-1.25) and 25 mg of hydrochlorothiazide was associated with a HR of 1.36 (95% CI, 1.18-1.57). Those in the CAS group not taking hydrochlorothiazide had a HR of 0.95 (95% CI, 0.82-1.10); the addition of 12.5 mg of hydrochlorothiazide was associated with a HR of 1.11 (95% CI, 0.95-1.29) and 25 mg of hydrochlorothiazide was associated with a HR of 1.28 (95% CI, 1.05-1.57).

Both drug combinations were generally well tolerated in each treatment group. Cancer was reported in 192 patients (1.70%) in the CAS group compared with 186 patients (1.64%) in the NCAS group (P = .73). Alzheimer disease, gastrointestinal tract bleeding, and Parkinson disease were reported in 1% or less of patients in each group and incidence did not differ between groups. Patients in the CAS group reported constipation and cough more frequently than patients in the NCAS group, while NCAS patients had more dyspnea, lightheadedness, symptomatic bradycardia, and wheezing (Table 5).

We tested the hypothesis that treatment of hypertensive CAD patients with either a verapamil SR–based strategy (CAS group) or a β-blocker–based strategy (atenolol; NCAS group) would result in equivalent clinical outcomes. Our findings demonstrated that these treatment strategies were equivalent in the prevention of the outcome of all-cause mortality, nonfatal MI, or nonfatal stroke. Furthermore, similar results were observed comparing the treatment strategies for all-cause mortality, cardiovascular death, cardiovascular hospitalization, and blood pressure control. Significant differences were observed between strategies that favored the verapamil SR plus trandolapril strategy (CAS group) for lower angina frequency and new diagnoses of diabetes. There was a significant interaction between treatment group and prior heart failure, suggesting that those randomized to the atenolol plus hydrochlorothiazide strategy (NCAS group) had better outcomes than those randomized to the verapamil SR plus trandolapril strategy (CAS group). Both strategies were well tolerated.

INVEST is the first, to our knowledge, large randomized, prospective trial to focus on CAD patients with hypertension and to follow JNC VI guidelines,⁹ which recommend use of an ACE inhibitor for special populations and lower blood pressure goals than other guidelines. It is important to note that this was not simply a comparison of verapamil SR with atenolol because it was anticipated that few patients would be treated with only those drugs. At study end, most were taking the combination of verapamil SR plus trandolapril (CAS group) or atenolol plus hydrochlorothiazide (NCAS group). Also, the study population included a high percentage of elderly, female, nonwhite, and diabetic patients. Thus, the results reported herein should be clinically applicable.

Although other trials^3,5,16,21 have investigated use of calcium antagonists in hypertensive patients, the frequency of CAD in these trials was too low to reach any relevant conclusions. For example, the Nordic Diltiazem (NORDIL) study demonstrated equivalence between diltiazem and diuretics and/or β-blockers for cardiovascular morbidity and mortality and showed a reduction in incidence of fatal and nonfatal stroke in the diltiazem group, but only a small proportion of those patients (4.5%; n = 496) had coronary heart disease.²¹ Results from several hypertension trials, including LIFE¹⁴ and ALLHAT,¹⁶ have been confounded by differences in achieved blood pressure level, which influences outcomes. In our study, the reductions and achieved levels for SBP and DBP were similar in both treatment groups. Most INVEST patients achieved JNC VI goals for blood pressure control. These findings in patients with CAD extend those from LIFE¹⁴ and ALLHAT,¹⁶ demonstrating that even lower blood pressure targets are achievable with more aggressive management. However, ALLHAT neither tested a β-blocker arm nor used an angiotensin II active agent for organ protection for patients with diabetes, renal impairment, or heart failure. Thus, INVEST results complement ALLHAT by including a β-blocker–based strategy plus organ protection in an elderly population with CAD. The INVEST data also confirm and extend the suggestions of others^7,35 that monotherapy is not necessarily sufficient for optimal treatment of hypertension.

Overall, adverse experiences reported were minimal and similar in frequency between treatment strategies. Previous articles^17,36- 37 have suggested that some calcium antagonists (principally short-acting dihydropyridines) may be associated with an increased risk of cancer, gastrointestinal tract bleeding, and all-cause mortality. Results of ALLHAT,¹⁶ STOP-2,⁵ and INVEST have not confirmed these suggestions. The difference in crossover rates may reflect the consequences of adverse experiences (dyspnea, lightheadedness, symptomatic bradycardia, and wheezing) associated with the combination of atenolol plus hydrochlorothiazide (NCAS group) compared with adverse experiences (constipation and cough) associated with the combination of verapamil SR plus trandolapril (CAS group). The possibility that the higher crossover rate in the atenolol-based strategy is related to previous intolerance or physician bias against β-blockers cannot be excluded, particularly because patients recently taking β-blockers were excluded from the trial. Another possibility is that the differing drug components of CAS (verapamil SR plus trandolapril) or NCAS (atenolol plus hydrochlorothiazide) could have conferred advantages in addition to blood pressure control. The combination of verapamil SR plus trandolapril could result in fewer metabolic complications, as was observed with reduction of new diagnoses of diabetes. The NCAS might have been expected to have advantages in patients with a prior MI and prior coronary revascularization; however, the results observed were similar with both strategies. Our outcome data for patients with prior heart failure, on the other hand, concur with recent trials documenting benefits of β-blockers when added to diuretics and ACE inhibitors,^38- 40 although not all patients in those trials had hypertension. In light of the results reported herein, management of hypertension must focus on the risk profile of the patient and overall treatment regimen rather than a single drug.

There are some limitations to our study. We used blood pressure goals in accordance with JNC VI; however, JNC VII⁷ and epidemiological data⁴¹ indicate that CHD risk increases with SBP level higher than 115 mm Hg so it could be argued that even lower blood pressure targets may be reasonable. More than half the patients required 3 or more antihypertensive drugs to achieve blood pressure control. Better blood pressure control might have been possible if we had included a fourth drug in each of the specified treatment strategies. The large sample size resulted in a statistically significant difference in angina frequency comparing CAS with NCAS, but this difference may not be clinically significant. The decline in angina prevalence and frequency from entry (only 2% underwent revascularization) is clinically important. This, at least in part, is likely due to the decline in both SBP and heart rate. Lastly, although the new diabetes analysis was not planned before the trial started, we added this outcome early in the recruitment phase. Our findings suggest potential clinical implications that require confirmation. Other analyses of INVEST baseline data indicate that Hispanic ethnicity, heart failure, US residency, hypercholesterolemia, left ventricular hypertrophy, stroke and transient ischemic attack, prior coronary revascularization, and body mass index are linked to risk of developing diabetes.⁴² In our preliminary analyses herein, administration of trandolapril appeared to confer some protection, as suggested in previous studies of ACE inhibitors.^8,10,16,43 Hydrochlorothiazide was associated with a nonsignificantly increased risk of developing diabetes, which is also consistent with previous studies (usually a thiazide diuretic with a β-blocker).^14,16,44 Further analyses are required to better understand the complex interactions among drug, dose, and demographic factors. Patients' potassium levels were not collected in this study, so the role that hypokalemia may have played in precipitating hyperglycemia cannot be determined.

In conclusion, our results indicate that lower targets for blood pressure control can be achieved in most hypertensive patients with CAD using a multidrug strategy that includes administration of ACE inhibitors to patients with heart failure, diabetes, or renal impairment. The clinical equivalence of the CAS and NCAS groups in prevention of death, MI, or stroke supports the use of either strategy in clinically stable patients with CAD who require blood pressure control. The decision regarding which drug classes to use in specific CAD patients should be based on additional factors including adverse experiences, history of heart failure, diabetes risk, and the physician's best judgment. The possibility of delaying the emergence of a diabetes diagnosis with a CAS compared with an NCAS requires further investigation.