Data from observational studies involving more than 1

Transcription

1 What Is the Optimal Blood Pressure in Patients After Acute Coronary Syndromes? Relationship of Blood Pressure and Cardiovascular Events in the Pravastatin or Atorvastatin Evaluation and Infection Therapy Thrombolysis in Myocardial Infarction (PROVE IT-TIMI) 22 Trial Sripal Bangalore, MD, MHA; Jie Qin, MS; Sarah Sloan, MS; Sabina A. Murphy, MPH; Christopher P. Cannon, MD; for the PROVE IT-TIMI 22 Trial Investigators Background Aggressive blood pressure (BP) control has been advocated in patients with acute coronary syndrome, but few data exist in this population relative to cardiovascular outcomes. Methods and Results We evaluated 4162 patients enrolled in the PRavastatin Or atorvastatin Evaluation and Infection Therapy Thrombolysis In Myocardial Infarction (PROVE IT-TIMI) 22 trial (acute coronary syndrome patients randomized to pravastatin 40 mg versus atorvastatin 80 mg). The average follow-up BP (systolic and diastolic) was categorized into 10-mm Hg increments. The primary outcome was a composite of death due to any cause, myocardial infarction, unstable angina requiring rehospitalization, revascularization after 30 days, and stroke. The secondary outcome was a composite of death due to coronary heart disease, nonfatal myocardial infarction, or revascularization. The relationship between BP (systolic or diastolic) followed a J- or U-shaped curve association with primary, secondary, and individual outcomes, with increased events rates at both low and high BP values, both unadjusted and after adjustment for baseline variables, baseline C-reactive protein, and on-treatment average levels of low-density lipoprotein cholesterol. A nonlinear Cox proportional hazards model showed a nadir of 136/85 mm Hg (range 130 to 140 mm Hg systolic and 80 to 90 mm Hg diastolic) at which the incidence of primary outcome was lowest. The curve was relatively flat for systolic pressures of 110 to 130 mm Hg and diastolic pressures of 70 to 90 mm Hg. Conclusions After acute coronary syndrome, a J- or U-shaped curve association existed between BP and the risk of future cardiovascular events, with lowest event rates in the BP range of approximately 130 to 140 mm Hg systolic and 80 to 90 mm Hg diastolic and a relatively flat curve for systolic pressures of 110 to 130 mm Hg and diastolic pressures of 70 to 90 mm Hg, which suggests that too low of a pressure (especially 110/70 mm Hg) may be dangerous. Clinical Trial Registration URL: Unique identifier: NCT (Circulation. 2010;122: ) Key Words: blood pressure hypertension prognosis acute coronary syndrome Data from observational studies involving more than 1 million individuals without preexisting vascular disease have indicated that deaths due to both ischemic heart disease and stroke increase progressively and linearly with blood pressure (BP). 1 Consequently, the notion that lower is better 2 has been popular for management of hypertension. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure states, The relationship between BP and risk of cardiovascular events is continuous, consistent, and independent of other risk factors. 3 Subsequently, a BP of 120/80 mm Hg has been considered optimal 4 or normal. 3 Clinical Perspective on p 2151 However, this linear theory has been challenged for nearly 3 decades, especially for diastolic pressure. 5 8 Physiologically, a J- or U-shaped curve phenomenon would be expected to exist in vital components such as BP and other biological systems, with an increased mortality exhibited at both ends of the spectrum. The linear relationship might hold true for the Received September 1, 2009; accepted September 2, From the New York University School of Medicine (S.B.), New York, NY, and TIMI Study Group (J.Q., S.S., S.A.M., C.P.C.), Brigham and Women s Hospital, Harvard Medical School, Boston, Mass. Guest Editor for this article was Bernard R. Chaitman, MD. Presented in part at the Annual Scientific Session of the American College of Cardiology, Orlando, Fla, March 30, Reprint requests to Dr Christopher P. Cannon, TIMI Study Group, Division of Cardiovascular Medicine, Brigham and Women s Hospital, 350 Longwood Ave, First Floor, Boston, MA cpcannon@partners.org 2010 American Heart Association, Inc. Circulation is available at DOI: /CIRCULATIONAHA

2 Bangalore et al J- or U-Shaped Curve for Blood Pressure After ACS 2143 general population, but in patients with stable coronary artery disease, the relationship between BP and cardiovascular outcomes has been shown in some studies to follow a J- or a U-shaped curve, with higher event rates at very low and very high BP. 5,7,9 14 However, this is controversial. The Seventh Report of the Joint National Committee states, There is no definitive evidence of an increase in risk of aggressive treatment (a J-curve) unless the diastolic BP is lowered to 55 or 60 mm Hg by treatment. 3 In the American Heart Association scientific statement on Treatment of Hypertension in the Prevention and Management of Ischemic Heart Disease, a target of 130/80 mm Hg has been recommended in patients at high risk of coronary artery disease and acute coronary syndromes (ACS), but it was acknowledged that there were limited data to support this recommendation. 15 We aimed to analyze what data exist for the target range of BP in patients after ACS. Methods Patient Population and Study Design We analyzed patients enrolled in the PRavastatin Or atorvastatin Evaluation and Infection Therapy Thrombolysis In Myocardial Infarction (PROVE IT-TIMI) 22 trial 16,17 which was an international, multicenter, randomized, double-blind, 2 2 factorial design trial of men or women at least 18 years old hospitalized with ACS (either myocardial infarction [MI], with or without ST-segment elevation, or high-risk unstable angina) in the preceding 10 days who were randomly assigned to receive pravastatin 40 mg or atorvastatin 80 mg once daily and to receive gatifloxacin or placebo. The protocol required that the baseline total cholesterol level be 240 mg/dl in statin-naive patients or 200 mg/dl in patients previously given a statin. Patients were managed with standard medical and interventional treatment for ACS. Follow-Up Patients were followed up for 18 to 36 months (average 24 months), with visits at 30 days, 4 months, and every 4 months thereafter. At each visit, vital signs and information on clinical end points, adverse events, and concurrent medication use were collected. During each visit, BP was recorded, and blood samples were collected and analyzed at a central laboratory. BP management was at the discretion of the treating physician. For the present analysis, average follow-up systolic and diastolic pressures were calculated for each patient by use of all postbaseline results up to the last visit before the date of primary outcome or the end of follow-up in those patients without events. The baseline value was substituted for patients with no postbaseline data. The risk of cardiovascular outcomes was then evaluated as a function of BP. Study Outcomes Primary and secondary outcomes measures in the present analysis were the same as for the main PROVE IT-TIMI 22 trial. 16,17 The primary outcome was the time from randomization to the first occurrence of death due to any cause, MI, documented unstable angina requiring hospitalization, revascularization with either percutaneous coronary intervention or coronary artery bypass graft performed more than 30 days after randomization, and stroke. The secondary outcome was a composite of death due to coronary heart disease, nonfatal MI, or revascularization after 30 days. Tertiary outcomes consisted of all-cause mortality, death due to coronary heart disease, and nonfatal MI considered as separate outcomes. Statistical Analyses BP values were categorized in 10-mm Hg increments for association with clinical outcomes. Patient groups were compared by 1-way analysis of variance (ANOVA) or Kruskal-Wallis rank test for continuous variables and 2 test for categorical variables. The decision to use the average on-treatment follow-up BP category was based on the following: We created separate models (unadjusted and adjusted) using baseline pressure, follow-up pressure, and average on-treatment follow-up pressures and calculated the predictive value of the models using the concordance index (C statistic). The end of follow-up (adjusted C statistic of 0.62 for systolic BP and 0.63 for diastolic BP) and average follow-up pressure variables (adjusted C statistic of 0.62 for systolic and 0.63 for diastolic BP) were higher than those for the baseline BP variables (adjusted C statistic of 0.60 for systolic and 0.60 for diastolic pressure), which suggests a higher predictive value than with the baseline BP variables. Because the average follow-up BP variable represents the effect of pressure over a period of time rather than at 1 point in time, we considered this to be more important for prediction of long-term events and used this for the rest of the analyses. However, we performed similar modeling (for events) with baseline pressure variables. We hypothesized that if a J- or U-shaped relationship was found with both baseline and average follow-up BP variables and outcomes, it was likely due to reverse causality (with low BP being a mere marker of ill health). If, however, a J- or U-shaped relationship was found with average follow-up BP but not at baseline, the BP itself was more likely to contribute to increased events at follow-up. Univariate Cox proportional hazards analysis was performed to assess the risk of outcomes for each 10-mm Hg increment in BP. Multivariable Cox proportional hazards analysis was performed that included BP category as the major factor, with adjustment for age, sex, smoking, baseline body mass index, history of hypertension, diabetes mellitus, history of coronary artery bypass graft surgery, coronary angioplasty, angina pectoris, cerebrovascular disease, peripheral arterial disease, heart failure, arrhythmia, baseline C-reactive protein level, average follow-up low-density lipoprotein levels, and treatment effect. This was done for the whole cohort and for the 2 treatment groups separately in accordance with the intention-to-treat principle. The adjusted hazard ratio for each category of systolic or diastolic BP was calculated in reference to the systolic BP group in which the event rate was lowest (nadir BP calculated by the delta method as outlined below), for which the hazard ratio was considered as 1. In addition, nonlinear Cox proportional hazards models were estimated with mean BP as a continuous variable and with the square of BP. Unadjusted hazard ratios were calculated on the basis of univariate Cox proportional hazards analysis that included BP and BP squared only. Adjusted hazard ratios were calculated on the basis of multivariate Cox proportional hazards analysis that included the major predictors BP and BP squared adjusted for the confounding variables listed previously. Overall BP effects were examined on the basis of a likelihood ratio test that compared a full model that included both linear and quadratic mean BP terms plus other covariates (in adjusted analysis only) and a reduced model without the linear and quadratic mean BP terms. Nadir BP was calculated on the basis of the delta method, which is equal to the coefficient of the linear term divided by 2 times the coefficient of the quadratic term. Furthermore, interactions between treatments by mean BP and BP squared were examined on the basis of the likelihood ratio test by comparison of a full model that included both linear and quadratic mean BP terms plus treatment plus other covariates (in adjusted analysis) versus a reduced model without the 2 interaction terms. A P value of 0.05 was considered statistically significant for all tests. All analyses were performed with Stata software version 9.2 (College Station, Tex). Results Patients A total of 4162 patients who had been hospitalized for an ACS within the preceding 10 days were randomized to pravastatin 40 mg or atorvastatin 80 mg per day. The main results of the trial have been discussed elsewhere. 17

3 2144 Circulation November 23, 2010 Table 1. Demographic and Baseline Characteristics by Mean Systolic BP Categories Mean Systolic BP During Follow-Up, mm Hg 100 to 110 (n 370) 110 to 120 (n 985) Parameter 100 (n 65) 160 (n 63) P* Age, mean (SD), y 56.5 (11.7) 54.5 (9.9) 54.4 (10.4) 57.8 (11.0) 61.0 (10.8) 63.2 (10.8) 62.5 (11.0) 67.2 (11.0) Men, n (%) 51 (68.9) 323 (85.2) 820 (82.2) 951 (79.4) 657 (74.6) 297 (73.5) 113 (70.2) 38 (57.6) White, n (%) 69 (93.2) 347 (91.6) 924 (92.7) 1098 (91.6) 786 (89.2) 361 (89.6) 132 (82.0) 57 (86.4) BMI, mean (SD), kg/m (6.1) 28.2 (4.9) 29.1 (5.6) 29.5 (5.6) 30.3 (5.7) 29.8 (5.8) 31.7 (7.5) 29.2 (6.1) Never smoked, n (%) 20 (27.0) 93 (24.5) 226 (22.7) 299 (25.0) 252 (28.6) 121 (29.9) 48 (29.8) 26 (39.4) Hypertension, n (%) 20 (27.0) 84 (22.2) 323 (32.4) 588 (49.1) 562 (63.8) 315 (78.0) 137 (85.1) 61 (92.4) Diabetes mellitus, n (%) 9 (12.2) 36 (9.5) 115 (11.5) 206 (17.2) 191 (21.7) 102 (25.2) 53 (32.9) 21 (31.8) FH of CAD, n (%) 43 (59.7) 173 (47.4) 503 (52.7) 618 (54.3) 436 (53.6) 200 (52.5) 69 (46.9) 30 (51.7) MI, n (%) 7 (9.5) 54 (14.2) 159 (15.9) 231 (19.3) 164 (18.6) 101 (25.0) 38 (23.6) 14 (21.2) CABG, n (%) 6 (8.1) 34 (9.0) 69 (6.9) 134 (11.2) 119 (13.5) 60 (14.8) 20 (12.4) 11 (16.7) Angioplasty, n (%) 13 (17.8) 75 (19.8) 215 (21.6) 329 (27.5) 262 (30.0) 140 (35.0) 54 (34.0) 23 (34.8) Angina pectoris, n (%) 13 (18.3) 73 (19.4) 198 (20.1) 258 (21.9) 188 (21.8) 101 (25.5) 55 (34.4) 19 (28.8) Peripheral vascular disease, n (%) 5 (6.8) 16 (4.2) 39 (3.9) 53 (4.4) 62 (7.0) 37 (9.2) 18 (11.2) 10 (15.1) CHF, n (%) 1 (1.3) 15 (4.0) 20 (2.0) 31 (2.6) 32 (3.6) 23 (5.7) 11 (6.9) 3 (4.5) Arrhythmia, n (%) 0 (0.0) 21 (5.6) 45 (4.5) 83 (7.0) 69 (7.9) 43 (10.7) 7 (4.4) 6 (9.1) Prior stroke/tia, n (%) 3 (4.0) 5 (1.3) 37 (3.7) 63 (5.3) 55 (6.2) 31 (7.7) 20 (12.4) 9 (13.6) Chronic renal failure, n (%) 0 (0.0) 24 (6.4) 80 (8.1) 93 (7.8) 82 (9.3) 69 (17.1) 18 (11.3) 6 (9.1) Baseline Characteristics Tables 1 and 2 show baseline characteristics by systolic and diastolic pressure categories, respectively. Patients with low systolic pressure were a lower-risk cohort who were more likely to be younger, male, leaner, and to have never smoked; they were less likely to have risk factors (hypertension, diabetes, prior MI, coronary artery bypass graft surgery, angioplasty, peripheral arterial disease, heart failure, stroke or transient ischemic attack, or chronic renal failure); and they had lower levels of baseline low-density lipoprotein cholesterol/high-density lipoprotein cholesterol/triglycerides but higher levels of baseline C-reactive protein than patients with high systolic pressure (Table 1). In contrast, patients with low diastolic pressure were older and more likely to be female and hypertensive and to have prior coronary artery bypass graft surgery, heart failure, and peripheral arterial disease than patients with high diastolic pressure. BP and Primary Outcome Among the 4162 patients, 1000 (24%) reached the primary outcome. The relationship between systolic pressure and the 120 to 130 (n 1182) 130 to 140 (n 878) 140 to 150 (n 399) 150 to 160 (n 158) Baseline heart rate, mean 71.6 (13.5) 69.2 (11.9) 68.9 (11.1) 69.5 (12.1) 69.3 (12.1) 69.3 (12.6) 70.2 (12.8) 68.2 (12.6) (SD), bpm Baseline LDL-C, mean (SD), mg/dl (30.4) (31.3) (29.9) (29.7) (29.8) (30.8) (29.0) (29.9) Baseline TG, mean (SD), mg/dl (67.2) (173.2) (84.1) (99.9) (101.2) (92.1) (79.1) (83.7) Baseline HDL-C, mean (SD), mg/dl 38.6 (10.4) 39.0 (10.1) 38.8 (10.3) 40.0 (11.0) 40.8 (11.1) 40.9 (10.3) 41.9 (12.9) 44.5 (14.2) Baseline CRP, mean (SD), mg/dl 3.01 (4.35) 2.65 (3.22) 2.15 (2.79) 2.19 (2.92) 2.45 (3.13) 2.61 (3.33) 2.64 (3.08) 2.74 (3.69) Aspirin, n (%) 20 (27.0) 86 (22.7) 298 (30.0) 431 (36.0) 338 (38.4) 178 (44.1) 71 (44.1) 28 (42.4) ACEi/ARB, n (%) 11 (14.9) 50 (13.2) 139 (14.0) 276 (23.0) 245 (27.8) 165 (40.8) 69 (42.9) 41 (62.1) CCB, n (%) 6 (8.1) 24 (6.3) 103 (10.3) 184 (15.4) 182 (20.7) 120 (30.1) 52 (32.3) 23 (34.8) blocker, n (%) 10 (13.5) 52 (13.7) 182 (18.3) 284 (23.7) 259 (29.4) 134 (33.2) 57 (35.4) 26 (39.4) BMI indicates body mass index; FH, family history; CAD, coronary artery disease; MI, myocardial infarction; CABG, coronary artery bypass graft surgery; CHF, congestive heart failure; TIA, transient ischemic attack; LDL-C, low-density lipoprotein cholesterol; TG, triglycerides; HDL-C, high-density lipoprotein cholesterol; CRP, C-reactive protein; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; and CCB, calcium channel blocker. *P values are based on 1-way analysis of variance for continuous variables (ie, age, BMI, and LDL-C) and Pearson s 2 test for the remaining categorical variables. Baseline medication usage 2 weeks before qualifying event before randomization. incidence of primary outcome followed a J- or U-shaped shaped curve, with an increased event rate at low and high systolic pressures (Figure 1A). After adjustment for baseline covariates, treatment effect, baseline C-reactive protein, and average on-treatment low-density lipoprotein cholesterol levels, compared with the reference group (BP 130 to 140 mm Hg), the risk of primary outcome increased 4.9-fold in the group with systolic BP 100 mm Hg and by 1.2-fold in the group with systolic BP 160 mm Hg (Figure 1A). A nonlinear Cox proportional hazards model with systolic pressure on a continuous scale ( 2 49, P ) identified a nadir of 136 mm Hg at which the event rate was the lowest. This was true for the overall cohort (Figure 1A) and separately for the 2 treatment cohorts (data not shown). The J- or U-shaped relationship was more pronounced with average follow-up BP than with baseline BP variables (Figure 1B). The relationship between diastolic pressure and the incidence of primary outcome also followed a J- or U-shaped curve, with increased event rates at low and high diastolic pressures (Figure 1C). After adjustment for baseline covariates and for treatment

4 Bangalore et al J- or U-Shaped Curve for Blood Pressure After ACS 2145 Table 2. Demographic and Baseline Characteristics by Mean Diastolic BP Categories Mean Diastolic BP, mm Hg Parameter 60 (n 123) 60 to 70 (n 907) 70 to 80 (n 2049) 80 to 90 (n 947) 90 to 100 (n 115) 100 (n 18) P* Age, mean (SD), y 61.7 (11.8) 60.3 (11.9) 58.0 (11.2) 56.7 (10.1) 54.4 (10.4) 54.9 (11.5) Men, n (%) 76 (61.8) 690 (76.1) 1602 (78.2) 770 (81.3) 97 (84.3) 15 (83.3) White, n (%) 114 (92.7) 834 (91.9) 1873 (91.4) 851 (89.9) 87 (76.3) 14 (77.8) BMI, mean (SD), kg/m (6.1) 28.4 (5.4) 29.4 (5.5) 30.7 (5.9) 32.1 (7.3) 30.3 (5.4) Never smoked, n (%) 24 (19.5) 231 (25.5) 512 (25.0) 276 (29.1) 33 (28.7) 8 (44.4) Hypertension, n (%) 57 (46.3) 352 (38.8) 975 (47.6) 591 (62.4) 98 (85.2) 17 (94.4) Diabetes mellitus, n (%) 30 (24.4) 156 (17.2) 364 (17.8) 157 (16.6) 23 (20.0) 3 (16.7) FH of CAD, n (%) 62 (53.0) 441 (51.6) 1014 (52.4) 491 (54.6) 56 (52.3) 7 (50.0) MI, n (%) 35 (28.5) 174 (19.2) 359 (17.5) 171 (18.1) 26 (22.6) 3 (16.7) CABG, n (%) 20 (16.3) 122 (13.4) 221 (10.8) 79 (8.3) 10 (8.7) 1 (5.6) Angioplasty, n (%) 44 (36.1) 237 (26.2) 535 (26.2) 257 (27.3) 33 (29.2) 5 (27.8) Angina pectoris, n (%) 30 (25.0) 223 (24.9) 424 (21.0) 196 (21.0) 27 (23.9) 5 (27.8) Peripheral vascular disease, n (%) 15 (12.2) 75 (8.3) 108 (5.3) 34 (3.6) 8 (7.0) 0 (0.0) CHF, n (%) 11 (9.0) 46 (5.1) 57 (2.8) 19 (2.0) 2 (1.7) 1 (5.6) Arrhythmia, n (%) 4 (3.2) 74 (8.2) 131 (6.4) 60 (6.4) 4 (3.5) 1 (5.6) Prior stroke/tia, n (%) 9 (7.3) 49 (5.4) 106 (5.2) 49 (5.2) 7 (6.1) 2 (11.1) Chronic renal insufficiency, n (%) 9 (7.3) 84 (9.3) 171 (8.4) 93 (9.9) 11 (9.6) 4 (22.2) Baseline heart rate, mean (SD), bpm 70 (12) 69 (12) 69 (12) 70 (12) 72 (11) 73 (15) Baseline LDL-C, mean (SD), mg/dl (33.7) (30.6) (29.1) (31.3) (30.2) (27.7) Baseline TG, mean (SD), mg/dl (59.7) (129.2) (95.1) (98.4) (81.5) (188.1) Baseline HDL-C, mean (SD), mg/dl 42.5 (12.9) 40.4 (11.3) 39.8 (10.6) 39.8 (10.8) 40.3 (11.1) 38.2 (9.7) Baseline CRP, mean (SD), mg/dl 2.99 (4.29) 2.44 (3.12) 2.32 (2.95) 2.28 (3.06) 2.50 (3.10) 1.46 (1.78) Aspirin, n (%) 44 (35.8) 319 (35.2) 717 (34.5) 331 (34.9) 44 (38.3) 4 (22.2) ACEi/ARB, n (%) 38 (30.9) 171 (18.8) 463 (22.6) 271 (28.6) 42 (36.5) 11 (61.1) CCB, n (%) 17 (13.8) 156 (17.2) 313 (15.3) 180 (19.0) 22 (19.1) 6 (33.3) blocker, n (%) 33 (26.8) 209 (23.0) 458 (22.4) 269 (28.4) 28 (24.3) 7 (38.9) Abbreviations as in Table 1. *P values are based on 1-way analysis of variance for continuous variables (ie, age, BMI, and LDL-C) and Pearson s 2 test for the remaining categorical variables. Baseline medication usage 2 weeks before qualifying event before randomization. effect, compared with the reference group (BP 80 to 90 mm Hg), the risk of primary outcome increased 3.7-fold in the group with diastolic BP 60 mm Hg and 2.1-fold in the group with diastolic BP 100 mm Hg (Figure 1C). A nonlinear Cox proportional hazards model with diastolic BP on a continuous scale ( 2 52, P ) identified a nadir of 85 mm Hg at which the event rate was the lowest (Figure 1C). This was true for the overall cohort (Figure 1C) and separately for the 2 treatment cohorts (data not shown). The J- or U-shaped relationship was more pronounced with average follow-up BP than with baseline BP variables (Figure 1D). BP and Secondary Outcome Among the 4162 patients, 856 (21%) patients reached the secondary outcome. A similar J- or U-shaped relationship with secondary outcome was found for both systolic ( 2 37, P ; Figure 2A) and diastolic ( 2 47, P ; Figure 2C) pressure in both the unadjusted and adjusted models, with a nadir at 137/84 mm Hg. However, as with the primary outcome, the J- or U-shaped relationship was more pronounced with average follow-up BP variables than with baseline BP variables (Figure 2B and 2D). BP and Tertiary Outcomes Among the 4162 patients, 119 (2.9%) died; 53 (1.3%) of these were cardiovascular deaths, 260 (6.2%) had a nonfatal MI, and 40 (0.96%) had a stroke. A J- or U-shaped relationship was found for both systolic ( 2 10, P 0.007; Figure 3A) and diastolic ( 2 11, P ; Figure 3C) pressure with all-cause mortality in both the unadjusted and adjusted models. The J- or U-shaped relationship was found with average follow-up BP variables but not with baseline BP variables (Figure 3B and 3D). A J- or U-shaped relationship with cardiovascular mortality was also found for both systolic ( 2 6, P 0.041; Figure 4A) and diastolic ( 2 14, P ; Figure 4C) pressure in both the unadjusted and adjusted models. The J- or U-shaped relationship was found with average follow-up BP variables but not with baseline BP variables (Figure 4B and 4D). Similarly, a J- or U-shaped relationship with nonfatal MI was found for both systolic ( 2 26, P ; Figure 5A) and diastolic ( 2 24, P ; Figure 5C) pressure in both the unadjusted and adjusted models, with a nadir at 134/84 mm Hg. The J- or U-shaped relationship was found with average follow-up BP variables but not with baseline BP variables (Figure 5B and 5D).

5 2146 Circulation November 23, 2010 Figure 1. BP and primary outcome. A, Incidence and adjusted risk of primary outcome as a function of average follow-up systolic BP categories. B, Adjusted risk of primary outcome as a function of baseline or average follow-up systolic BP categories. C, Incidence and adjusted risk of primary outcome as a function of average follow-up diastolic BP categories. D, Adjusted risk of primary outcome as a function of baseline or average follow-up diastolic BP categories. DBP indicates diastolic BP; SBP, systolic BP; and HR, hazard ratio. All of the above relationships were true for the overall cohort and separately for the 2 treatment cohorts (data not shown). The event rate for stroke (0.96%) was too small to derive any meaningful relationship between BP and stroke. The incidence ratio of nonfatal MI to stroke remained constant for a wide range of BPs; however, at lower diastolic pressures, the incidence of nonfatal MI was much higher than the incidence of stroke, which implies that a compromised coronary circulation resulting from low diastolic pressures could be a more important factor for MI than for stroke. Sensitivity Analysis To evaluate the effect of pulse pressure, we performed a sensitivity analysis that controlled for it. The results were unchanged when the analysis was repeated after controlling for pulse pressure (data not shown). We also evaluated the joint distribution of systolic and diastolic pressures in a number of ways. There was only a modest correlation between the 2 variables (r 0.58, 95% CI 0.56 to 0.60). The mean diastolic pressures for each systolic pressure category (Figure 1A) showed no distribution of higher diastolic pressures for low systolic categories and vice versa; the same was true for diastolic BP categories (Figure 1C). In a model designed to predict whether the risk of events was higher in those with low systolic pressure or diastolic pressure or in those with both low systolic and diastolic pressures, the test for interaction was not significant, and the risk of events was comparable for all 3 categories. Discussion The present analysis of a high-risk post-acs population enrolled in the PROVE IT-TIMI 22 trial showed that a J- or U-shaped relationship existed between BP and the risk of cardiovascular outcomes such that there was an exponential increase in event rates at high and low BP values. The relationship was true for both systolic and diastolic BP. The analyses identified a BP nadir of 136/85 mm Hg at which the event rate was the lowest; however, the curve was relatively flat for systolic pressures of 110 to 130 mm Hg and diastolic pressures of 70 to 90 mm Hg. ABP 110/70 mm Hg was associated with an increased risk of cardiovascular events, which suggests that a BP that is too low indentifies a subset of patients with poor prognosis.

6 Bangalore et al J- or U-Shaped Curve for Blood Pressure After ACS 2147 Figure 2. BP and secondary outcome. A, Incidence and adjusted risk of secondary outcome as a function of average follow-up systolic BP categories. B, Adjusted risk of secondary outcome as a function of baseline or average follow-up systolic BP categories. C, Incidence and adjusted risk of secondary outcome as a function of average follow-up diastolic BP categories. D, Adjusted risk of secondary outcome as a function of baseline or average follow-up diastolic BP categories. SBP indicates systolic BP; HR, hazard ratio; and DBP, diastolic BP. Shifting Paradigm of Intensive BP Management In patients treated for hypertension, guidelines recommend a target of 140/90 mm Hg, with lower targets ( 130/ 80 mm Hg) for special populations such as those with diabetes, renal impairment, or coronary artery disease. 3,18,19 However, the evidence to support a lower BP target is not robust. In the Action to Control Cardiovascular Risk in Diabetes BP trial, 20 there was no reduction in the rate of fatal and nonfatal major cardiovascular outcomes with intensive management (target systolic pressure 120 mm Hg) compared with a standard target of 140 mm Hg among high-risk persons with type 2 diabetes mellitus. Similarly, in a Cochrane meta-analysis of 7 randomized trials, Arguedas et al 21 showed no difference in cardiovascular morbidity and mortality with intensive BP management (target 135/85 mm Hg) compared with standard targets ( 140 to 160/ 90 to 100 mm Hg). The results of the present analyses extend the above observation to the highest-risk cohort, those with established coronary artery disease after ACS, and suggests not only a flat curve for BP 110 to 130/70 to 90 mm Hg but also a higher risk of cardiovascular events at lower pressures. This suggests that the paradigm of lower is better in cardiovascular medicine is not applicable to BP control beyond a certain target. BP and the J-Curve Phenomenon The J- or U-shaped curve phenomenon with BP has been explored for more than 3 decades. Most of the prior studies were retrospective analyses or post hoc analyses of randomized trials of hypertensive cohorts. In patients without known coronary artery disease, the existence of a J- or U-shaped curve has not been shown consistently. Stewart 8 showed a J- or U-shaped relationship for MI and cardiovascular death with a nadir at a diastolic pressure of 100 to 109 mm Hg. The Australian National BP Trial 22 echoed these findings, with a nadir at 85 to 89 mm Hg of diastolic pressure. Conversely, an analysis of the International Prospective Primary Prevention Study in Hypertension trial failed to show a J- or U-shaped curve. 23 More recently, an analysis of the Hypertension Optimal Treatment trial showed a J-shaped curve for both systolic and diastolic pressure for cardiovascular death, with a nadir at 139/86 mm Hg, 24 similar to the nadir of 137/ 84 mm Hg shown in the present study. A meta-analysis of individual patient data from 7 randomized trials showed a J- or U-shaped curve for all-cause and cardiovascular mortality, with a nadir at 156/84 mm Hg. However, in a recent randomized trial of nondiabetic hypertensive subjects, tight control of BP (mean 132/77 mm Hg) was associated with a significant

7 2148 Circulation November 23, 2010 Figure 3. BP and all-cause mortality. A, Incidence and adjusted risk of all-cause mortality as a function of average follow-up systolic BP categories. B, Adjusted risk of all-cause mortality as a function of baseline or average follow-up systolic BP categories. C, Incidence and adjusted risk of all-cause mortality as a function of average follow-up diastolic BP categories. D, Adjusted risk of all-cause mortality as a function of baseline or average follow-up diastolic BP categories. SBP indicates systolic BP; HR, hazard ratio; and DBP, diastolic BP. reduction in cardiovascular events compared with usual control (mean 136/79 mm Hg). 25 Of note, the above studies enrolled very few or no patients with known coronary artery disease. The optimal BP for patients with coronary artery disease, especially post-acs patients, is not well defined. A recent American Heart Association statement suggests a target of 130/80 mm Hg in this cohort. 15 Insights from the Comparison of Amlodipine versus Enalapril to Limit Occurrences of Thrombosis study (CAMELOT), which enrolled patients with normal BP (baseline 129/78 mm Hg), appear to suggest that further BP reduction might be beneficial 26 ; however, the optimal target is less well defined. Few studies have evaluated the J- or U-shaped curve phenomenon in patients with hypertension and coronary artery disease. Cruickshank et al 5 showed a J-shaped relationship between cardiovascular death and diastolic BP, but only in patients with known coronary artery disease, with a nadir at 85 to 90 mm Hg. Similarly, data from the Framingham Heart Study also showed a J- or U-shaped relationship for cardiovascular mortality, but only in patients with prior MI. 27 In the International Verapamil-SR Trandolapril study of patients with hypertension and coronary artery disease, a J- or U-shaped relationship was shown between BP and cardiovascular outcomes, 7 and the relationship was stronger in patients without prior angioplasty, which suggests that patients with prior angioplasty tolerated low diastolic pressures better than those without. In the recently published Ongoing Telmisartan Alone and in Combination With Ramipril Global End-Point Trial (ONTARGET) of patients with atherosclerotic disease or diabetes with organ damage, a J- or U-shaped relationship with cardiovascular mortality and MI was shown for both systolic and diastolic pressure. 28 The present data are concordant with prior studies showing a J- or U-shaped relationship; however, the present study differs from prior studies in many respects. Unlike the prior studies in hypertensive cohorts, the present study is a post hoc analysis of patients after ACS. Unlike other hypertension studies, the treatment of hypertension and the choice of medications were left to the discretion of the managing physician. The present study also showed that the degree of statin therapy did not mitigate the J- or U-shaped curve phenomenon. The J- or U-shaped relationship persisted after we controlled for baseline C-reactive protein and average on-treatment low-density lipoprotein cholesterol levels. Pathophysiological Mechanisms for J-Curve Phenomenon Several pathophysiological mechanisms have been proposed to explain the existence of a J- or U-shaped curve. The

8 Bangalore et al J- or U-Shaped Curve for Blood Pressure After ACS 2149 Figure 4. BP and cardiovascular mortality. A, Incidence and adjusted risk of cardiovascular mortality as a function of average follow-up systolic BP categories. B, Adjusted risk of cardiovascular mortality as a function of baseline or average follow-up systolic BP categories. C, Incidence and adjusted risk of cardiovascular mortality as a function of average follow-up diastolic BP categories. D, Adjusted risk of cardiovascular mortality as a function of baseline or average follow-up diastolic BP categories. SBP indicates systolic BP; HR, hazard ratio; CV, cardiovascular; and DBP, diastolic BP. results of the present analyses can be explained by any these mechanisms either singly or in combination. It has been hypothesized that the J- or U-shaped curve may be an epiphenomenon of more severe underlying chronic illness (including cancer) or underlying severe cardiac illness (like heart failure), thereby increasing mortality. However, only 1% to 5% of patients in the present study had a history of heart failure at baseline, with a lower percentage in those with lower systolic BP. In addition, the J- or U-shaped relationship between BP (both systolic and diastolic) and outcomes persisted even after we controlled for baseline covariates. Moreover, the results are from data derived from a randomized trial in which patients with debilitating illness (for example, advanced cancer with a survival of 2 years) are excluded. In addition, if the J- or U-shaped curve were due to reverse causality (low BP variables being a mere marker of ill health), the relationship should have been seen with both baseline and average follow-up BP variables. In our analyses, the J- or U-shaped relationship was found with average on-treatment BP variables but not at baseline, which suggests that decreased BP per se might be a contributor to the increased events. However, other unmeasured indicators of poor health were not taken into consideration in the present analysis. The J- or U-shaped curve may represent an epiphenomenon of increased arterial stiffness, and thus, a low diastolic BP might be a marker for high pulse pressure and an increase in mortality. 29 In our analyses, we noticed a J- or U-shaped curve phenomenon not just for diastolic but also for systolic BP, for which the pulse pressure theory would not be applicable. A low diastolic BP may compromise coronary perfusion in subjects with coronary heart disease, because coronary perfusion occurs in diastole. In the present analysis, all patients were post-acs patients, and the incidence ratio of nonfatal MI to stroke increased at low diastolic pressures, which emphasizes that at low BP, MI is more likely than stroke. This can be explained by impaired coronary perfusion and is a more likely explanation given that we observed a more pronounced J- or U-shaped curve effect with follow-up BP variables than with baseline BP variables. However, the present study does not prove a causal relationship between low pressure (either systolic or diastolic) and an increased risk of events, although the differential effect on outcomes between baseline and follow-up BP values appears to suggest that a low BP might be causally related to outcomes.

9 2150 Circulation November 23, 2010 Figure 5. BP and nonfatal myocardial infarction. A, Incidence and adjusted risk of nonfatal myocardial infarction as a function of average follow-up systolic BP categories. B, Adjusted risk of nonfatal myocardial infarction as a function of baseline or average follow-up systolic BP categories. C, Incidence and adjusted risk of nonfatal myocardial infarction as a function of average follow-up diastolic BP categories. D, Adjusted risk of nonfatal myocardial infarction as a function of baseline or average follow-up diastolic BP categories. MI indicates myocardial infarction; SBP, systolic BP; HR, hazard ratio; and DBP, diastolic BP. Study Limitations The present study was a post hoc analysis that evaluated the relationship between BP and cardiovascular events in a coronary artery disease population with tight control of cholesterol levels, and hence, the results cannot be extrapolated to other populations. We did not adjust our analyses for all possible confounders, especially those that are predictors of poor health, such as socioeconomic status, job stress, and mental health. We also did not adjust our analyses for dosage of antihypertensive agents received. Finally, therapies that reduce systolic BP usually reduce diastolic pressure as well, which makes it difficult to determine precisely whether differences in event rates observed at the lower range were caused by reduced systolic or diastolic BP or their combination. Conclusions In post-acs patients, a J- or U-shaped curve association existed between BP and the risk of future cardiovascular events, with the lowest event rates in the BP range of approximately 130 to 140/80 to 90 mm Hg and a relatively flat curve for pressures of 110 to 130/70 to 90 mm Hg, which suggests that too low of a pressure (especially 110/ 70 mm Hg) may be dangerous. Thus, although our data generally support the lower is better approach, this only goes so far. The present findings provide support for the Seventh Report of the Joint National Committee s guideline recognition of a possible increased risk when diastolic pressures are lowered to 60 mm Hg. Our findings are consistent with a recent randomized trial 20 and a large observational study 30 in stable patients that showed no benefit of more intensive BP management (beyond standard lowering of systolic BP to 140 mg/dl) but also extend the observation to a high-risk group of post-acs patients. Sources of Funding PROVE IT-TIMI 22 was supported by a research grant from Bristol-Myers-Squibb and Sankyo. Disclosures Dr Cannon has received research grants/support from Accumetrics, AstraZeneca, GlaxoSmithKline, Intekrin Therapeutics, Merck, and Takeda; has served on advisory boards (but donated the funds therefrom to charity) for Alnylam, Bristol-Myers Squibb/Sanofi Partnership, and Novartis; and has received honoraria for independent educational conferences from Pfizer and AstraZeneca; he also

10 Bangalore et al J- or U-Shaped Curve for Blood Pressure After ACS 2151 holds an equity interest in Automedics Medical Systems. The remaining authors report no conflicts. References 1. Lewington S, Clarke R, Qizilbash N, Peto R, Collins R. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet. 2002;360: Anderson TW. Re-examination of some of the Framingham bloodpressure data. Lancet. 1978;2: Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr, Roccella EJ. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report. JAMA. 2003;289: The sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Arch Intern Med. 1997;157: Cruickshank JM, Thorp JM, Zacharias FJ. Benefits and potential harm of lowering high blood pressure. Lancet. 1987;1: Lindholm L, Lanke J, Bengtsson B, Ejlertsson G, Thulin T, Schersten B. Both high and low blood pressures risk indicators of death in middle-aged males: isotonic regression of blood pressure on age applied to data from a 13-year prospective study. Acta Med Scand. 1985;218: Messerli FH, Mancia G, Conti CR, Hewkin AC, Kupfer S, Champion A, Kolloch R, Benetos A, Pepine CJ. Dogma disputed: can aggressively lowering blood pressure in hypertensive patients with coronary artery disease be dangerous? Ann Intern Med. 2006;144: Stewart IM. Relation of reduction in pressure to first myocardial infarction in patients receiving treatment for severe hypertension. Lancet. 1979;1: Protogerou AD, Safar ME, Iaria P, Safar H, Le Dudal K, Filipovsky J, Henry O, Ducimetiere P, Blacher J. Diastolic blood pressure and mortality in the elderly with cardiovascular disease. Hypertension. 2007;50: Kaplan N. J-curve not burned off by HOT study: Hypertension Optimal Treatment. Lancet. 1998;351: Weinberger MH. Do no harm: antihypertensive therapy and the J curve. Arch Intern Med. 1992;152: Staessen JA. Potential adverse effects of blood pressure lowering: J-curve revisited. Lancet. 1996;348: Farnett L, Mulrow CD, Linn WD, Lucey CR, Tuley MR. The J-curve phenomenon and the treatment of hypertension: is there a point beyond which pressure reduction is dangerous? JAMA. 1991;265: Waller PC, Isles CG, Lever AF, Murray GD, McInnes GT. Does therapeutic reduction of diastolic blood pressure cause death from coronary heart disease? J Hum Hypertens. 1988;2: Rosendorff C, Black HR, Cannon CP, Gersh BJ, Gore J, Izzo JL Jr, Kaplan NM, O Connor CM, O Gara PT, Oparil S. Treatment of hypertension in the prevention and management of ischemic heart disease: a scientific statement from the American Heart Association Council for High Blood Pressure Research and the Councils on Clinical Cardiology and Epidemiology and Prevention. Circulation. 2007;115: Cannon CP, McCabe CH, Belder R, Breen J, Braunwald E. Design of the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE IT)-TIMI 22 trial. Am J Cardiol. 2002;89: Cannon CP, Braunwald E, McCabe CH, Rader DJ, Rouleau JL, Belder R, Joyal SV, Hill KA, Pfeffer MA, Skene AM. Intensive versus moderate lipid lowering with statins after acute coronary syndromes. N Engl J Med. 2004;350: Williams B, Poulter NR, Brown MJ, Davis M, McInnes GT, Potter JF, Sever PS, Thom SM. British Hypertension Society guidelines for hypertension management 2004 (BHS-IV): summary. BMJ. 2004;328: Whitworth JA World Health Organization (WHO)/International Society of Hypertension (ISH) statement on management of hypertension. J Hypertens. 2003;21: The ACCORD Study Group. Effects of intensive blood-pressure control in type 2 diabetes mellitus. N Engl J Med. 2010;362: Arguedas JA, Perez MI, Wright JM. Treatment blood pressure targets for hypertension. Cochrane Database Syst Rev. 2009;(3):CD Untreated mild hypertension: a report by the Management Committee of the Australian Therapeutic Trial in Mild Hypertension. Lancet. 1982;1: The IPPPSH Collaborative Group. Cardiovascular risk and risk factors in a randomized trial of treatment based on the beta-blocker oxprenolol: the International Prospective Primary Prevention Study in Hypertension (IPPPSH). J Hypertens. 1985;3: Hansson L, Zanchetti A, Carruthers SG, Dahlof B, Elmfeldt D, Julius S, Menard J, Rahn KH, Wedel H, Westerling S; HOT Study Group. Effects of intensive blood-pressure lowering and low-dose aspirin in patients with hypertension: principal results of the Hypertension Optimal Treatment (HOT) randomised trial. Lancet. 1998;351: Verdecchia P, Staessen JA, Angeli F, de Simone G, Achilli A, Ganau A, Mureddu G, Pede S, Maggioni AP, Lucci D, Reboldi G. Usual versus tight control of systolic blood pressure in non-diabetic patients with hypertension (Cardio-Sis): an open-label randomised trial. Lancet. 2009; 374: Nissen SE, Tuzcu EM, Libby P, Thompson PD, Ghali M, Garza D, Berman L, Shi H, Buebendorf E, Topol EJ. Effect of antihypertensive agents on cardiovascular events in patients with coronary disease and normal blood pressure: the CAMELOT study: a randomized controlled trial. JAMA. 2004;292: D Agostino RB, Belanger AJ, Kannel WB, Cruickshank JM. Relation of low diastolic blood pressure to coronary heart disease death in presence of myocardial infarction: the Framingham Study. BMJ. 1991;303: Sleight P, Redon J, Verdecchia P, Mancia G, Gao P, Fagard R, Schumacher H, Weber M, Bohm M, Williams B, Pogue J, Koon T, Yusuf S. Prognostic value of blood pressure in patients with high vascular risk in the Ongoing Telmisartan Alone and in combination with Ramipril Global Endpoint Trial study. J Hypertens. 2009;27: Kannel WB, Wilson PW, Nam BH, D Agostino RB, Li J. A likely explanation for the J-curve of blood pressure cardiovascular risk. Am J Cardiol. 2004;94: Cooper-DeHoff RM, Gong Y, Handberg EM, Bavry AA, Denardo SJ, Bakris GL, Pepine CJ. Tight blood pressure control and cardiovascular outcomes among hypertensive patients with diabetes and coronary artery disease. JAMA. 2010;304:61 8. CLINICAL PERSPECTIVE Although aggressive blood pressure (BP) control has been advocated in patients with diabetes and more recently in those with acute coronary syndromes, recent trials and observational studies in stable diabetic patients have challenged this concept by finding that more intensive BP lowering was not superior to standard BP lowering. We found in a population of 4162 post acute coronary syndrome patients that a J- or U-shaped curve existed between BP and the risk of future cardiovascular events, with the lowest event rates in the BP range of approximately 130 to 140/80 to 90 mm Hg. Thus, our findings are consistent with recent studies in stable patients but extend the observation to high-risk post acute coronary syndrome patients. As such, this study helps provide evidence that clinicians treating hypertension should aim for a systolic BP 140 mm Hg but not 110 mm Hg.