POTENTIAL ASSOCIATION BETWEEN VERY LOW DIASTOLIC BLOOD PRESSURE AND INCREASED CARDIOVASCULAR RISK IN HYPERTENSIVE PATIENTS WITH KNOWN CORONARY ARTERY DISEASE
The treatment of hypertension significantly reduces the risk of cardiovascular (CV) events including stroke, ischemic heart disease, myocardial infarction (MI), and heart failure. The current goal from the Seventh Report of the Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7) is to reduce systolic blood pressure (BP) to <140 mm Hg and diastolic BP to <90 mm Hg. There has been considerable debate for decades, however, whether reducing diastolic BP too much might increase the risk of CV events—the so‐called J‐curve phenomenon. The theory is that since the coronary arteries are perfused during diastole, reducing the diastolic BP to levels <80–85 mm Hg could adversely affect coronary perfusion and result in myocardial ischemia. If such a J‐shaped relationship exists, it should be most evident in patients with established coronary artery disease (CAD). It is also possible, however, that patients who develop low diastolic BP are predisposed to an increase in CV events due to their underlying condition; ie, it is not the treatment per se but the underlying condition causing the low diastolic BP that predisposes to an increase in CAD events. A secondary analysis of the International Verapamil‐Trandolapril Study (INVEST) offered an opportunity to look for a J‐curve phenomenon.
INVEST, a clinical trial using a prospective, randomized, open label with blinded end point design enrolled 22,576 patients from 14 countries with clinically stable CAD and hypertension. They were randomly assigned to a verapamil sustained release‐ or atenolol‐based antihypertensive regimen. Trandolapril or hydrochlorothiazide was added to each regimen as necessary to reach BP goals. In accordance with JNC VI, which was in effect at the time of the study, a goal BP <140/90 mm Hg or <130/85 mm Hg in people with diabetes or renal disease was targeted. The prespecified primary outcome was a composite of the first occurrence of all‐cause death, nonfatal MI, and nonfatal stroke by intention‐to‐treat analysis. Prespecified main secondary outcomes included each of these components individually. For the present analysis, additional outcomes included fatal and nonfatal MI, fatal and nonfatal stroke, and average final on‐treatment systolic and diastolic BP. The association between on‐treatment systolic and diastolic BP, regardless of treatment assignment, and the incidence of the primary end point was determined using a proportional hazard model (ie, determining whether there was a J‐curve). Subsequently, further analyses were performed to identify any clinically relevant interactions between baseline characteristics and the relationship between BP and the primary end point, assuming a quadratic relationship exists between BP values and outcome. Follow‐up data were available for 97.5% of the original participants.
At study close (average of 2.7 years), there was no statistical difference in the incidence of the primary end point, prespecified secondary end points, or on‐treatment systolic or diastolic BP between the verapamil‐ or atenolol‐based strategies. The incidence of the primary end point was related to both on‐treatment systolic and diastolic BP in a J‐shaped pattern. Similar curves were seen regardless of treatment allocation. In the unadjusted models, the lowest incidence of the primary end point occurred at an on‐treatment BP of approximately 119/84 mm Hg, with an increase in events below that value. Those with diastolic BPs <70 mm Hg experienced adverse cardiovascular outcomes at the same rate as those with readings >100 mm Hg. Subjects with an on‐treatment diastolic BP <70 mm Hg made up only 10.5% of subjects but accounted for 19.6% of events.
The J‐curve was evident for the primary end point, as well as the secondary outcomes, fatal and nonfatal MI, and fatal and nonfatal stroke, although to a lesser extent for stroke. The ratio between MI and stroke remained constant throughout most BP strata, but increased with progressive decline in diastolic BP, suggesting that compromised coronary perfusion resulting from low diastolic BP could be a factor contributing to this observation. In the models adjusted for important baseline characteristics, however, the relation between systolic BP and outcome failed to show evidence of a substantial J‐curve; for diastolic BP, the J‐curve was drastically attenuated, with the nadir moved to 74 mm Hg as the shape of the curve became much more shallow. A history of coronary revascularization was associated with a diminution in the J‐curve between decreasing diastolic BP and events.
In this post hoc analysis of hypertensive patients with CAD, the authors concluded that there is a J‐curve between systolic and diastolic BP and the incidence of CV disease events; however, in models adjusted for important baseline characteristics, the increased risk for CV disease events at lower on‐treatment BP was significantly attenuated.—Messerli FH, Mancia G, Conti CR, et al. Dogma disputed: can aggressively lowering blood pressure in hypertensive patients with coronary artery disease be dangerous? Ann Intern Med. 2006;144:884–893.
COMMENT
There has been considerable debate for decades within the hypertension community as to whether or not the J‐curve actually exits. The only major prospective controlled clinical trial that attempted to answer this question was the Hypertension Optimal Treatment (HOT) trial. It randomized more than 18,000 patients with hypertension to 1 of 3 targeted diastolic pressures and demonstrated no evidence of a J‐curve. The lowest incidence of CV events was seen with a mean achieved diastolic BP of 82.6 mm Hg, with no additional benefit or harm observed with lower values. Since the actual diastolic BPs achieved in each group within HOT were separated by only a few mm Hg and few patients actually achieved values <80 mm Hg, the authors could not rule out that a J‐curve phenomenon existed at lower BPs. As more than half of the more than 22,576 subjects randomized in INVEST achieved a diastolic BP <80 mm Hg, an excellent opportunity existed to conduct this post hoc analysis.
Another question that has clouded the J‐curve phenomenon is whether the observation applies predominantly to individuals with isolated systolic hypertension (ISH). The major intervention studies in ISH have included subjects with baseline systolic BPs of 150–160 mm Hg and diastolic BPs <90 mm Hg. While the Systolic Hypertension in the Elderly Program (SHEP) demonstrated a marked decrease in CV disease events with antihypertensive treatment, there appeared to be a paradoxical relationship between reduction in diastolic BP and increase in CV disease events that became significant at values <65 mm Hg; these were increased more in patients with diastolic BPs <55 mm Hg. The treated subset who achieved diastolic BPs <55 mm Hg, however, did not do worse or better in terms of CV disease events compared with those on placebo. JNC 7 recommended treating patients with ISH to a systolic BP of <140 mm Hg regardless of achieved diastolic BP, but did express appropriate concern that patients with occlusive CAD may be at increased risk of coronary events if the diastolic BPs are too low. If a J‐curve phenomenon exists in people with hypertension, it should be most apparent in patients with stable CAD such as those in INVEST.
The present secondary analysis of INVEST suggests that there is a J‐curve between achieved BP and the risk of CV disease events in patients with established CAD and hypertension. There are 3 reasons, however, why it should not change current clinical practice or guidelines. First, as a post hoc analysis, these results should be considered only exploratory and hypothesis generating. To change clinical practice, a properly designed prospective study specifically addressing this question needs to be completed. We should remain skeptical of the post hoc findings of this or any clinical study. Second, these data pertain only to the relatively small subset of patients with both hypertension and CAD and should not be extrapolated to the millions of patients with essential hypertension and no underlying clinical CAD as defined in the original INVEST paper (MI occurring ≥3 months previously, angiographic CAD of at least 50% narrowing in ≥1 vessel, classic angina pectoris, or abnormalities on 2 different noninvasive tests including electrocardiography, echocardiography, or stress testing). Third, the present report examining the relationship between events and on‐treatment BP is a nonrandomized post hoc analysis without a true control group. The authors themselves point out that it is impossible to identify causation in this type of analysis.
Although there may be an association between low achieved diastolic BP and an increase in events, one cannot say that aggressive antihypertensive therapy directly resulted in this increase. Data from the Framingham Heart Study demonstrate that in older patients, untreated diastolic BP is inversely associated with the risk of CV events; in other words, lower diastolic BP is associated with higher CV risk. Since the mean age in INVEST was 66 years, it is not surprising that given similar systolic BP, a lower diastolic BP (wider pulse pressure) would be associated with an increased risk of CV events. The subset of subjects who had the highest risk of events, those with on‐treatment diastolic BP <70 mm Hg, also had a mean systolic BP of approximately 130 mm Hg or below. As these were all treated hypertensives, the lower mean systolic BP in subjects with diastolic BP <70 mm Hg suggests that they were being treated by their physicians to more aggressive BP goals, likely because they had other high‐risk characteristics. In fact, patients with lower diastolic BPs were older and had higher rates of previous MI, heart failure, and diabetes than those with higher diastolic BPs. After adjusting for important baseline characteristics, the shape of the curve became extremely shallow, and any increased risk of events associated with lower on‐treatment BP was small and not likely to be of clinical significance.
This trial highlights the need for future trials to clarify BP goals in patients with hypertension and stable CAD. The findings that the ratio of MI to stroke increased at low diastolic BPs and that coronary revascularization attenuated the increased risk of CV events associated with low diastolic BP support the biologic plausibility of the J‐curve hypothesis. For now, we should continue to strive to control systolic BP to <140 mm Hg in most hypertensives and <130 mm Hg in high‐risk patients such as those with diabetes and chronic kidney disease. We should also be aware that a low diastolic BP may be a marker of increased CV risk in patients with hypertension and CAD as well as in those with ISH. Future clinical trials should be designed to prospectively determine the optimal diastolic BP for clinical benefit. Pending results of these future trials, our commitment should be to control BP to the targets outlined in JNC 7.