The U‐shape relationship between pulse pressure level on inpatient admission and long‐term mortality in acute coronary syndrome patients undergoing percutaneous coronary intervention

Abstract The association between pulse pressure and long‐term mortality was investigated among acute coronary syndrome (ACS) patients who received percutaneous coronary intervention (PCI). The study population included 5055 ACS patients in the Department of Cardiology of Beijing Friendship Hospital who were enrolled from January 2013 to July 2019. The median duration of follow‐up was 24 months. Multivariate Cox regression was used to analyze the relationships between PP on inpatient admission and mortalities. Non‐linear associations were studied by restricted cubic splines. Considering the heart function, the analyses were performed in the whole cohort and the LVEF > = 0.5 cohort separately. Subgroup analyses were performed according to the different diagnosis (the myocardial infarction subgroup and the unstable angina pectoris subgroup). When PP was used as categorical variable, the high PP group (≥61 mm Hg) significantly increased the risk of death compared with the intermediate PP group (50–60 mm Hg) in the both cohorts. When PP was used as continuous variable, a U‐shape relationship were found between PP and mortalities in the whole cohort (p (for nonlinearity) = .005 and .003, respectively), with reference PP level of 55 mm Hg. However, this U‐shape relationship disappeared in the LVEF > 0.5 cohort (p (for nonlinearity) = .111 and .117, respectively). The similar results were obtained in MI subgroup. From this study, the U‐shape relationships between PP level and all‐cause and cardiac mortalities were found in ACS patients who underwent PCI. The U‐shape relationships disappeared in the LVEF > 0.5 cohort. The reference PP level was 55 mm Hg.


INTRODUCTION
Pulse pressure (PP) is defined as the difference between systolic blood pressure (SBP) and diastolic blood pressure (DBP). PP is the pulsatile component of blood pressure. 1 Globally, increased PP is related to higher large artery stiffness which leads to an increase in SBP and a decrease in DBP. Increased PP has been reported to be associated with mortality or adverse outcomes in general population, [2][3][4] hypertensive patients, 5 myocardial infarction patients, 6,7 patients with atherothrombosis, 8 and patients undergoing percutaneous coronary intervention (PCI). 9 The previous studies also reported the inversely relationship between PP level and adverse outcomes in patients with heart failure (HF), 10,11 and acute coronary syndrome (ACS). 12,13 Lower PP may indicate the lower cardiac output. For ACS patients, the Ushape relationships between PP level and adverse outcomes were found in some studies based on univariate analysis, 14,15 while the curve often disappeared after multivariable regression analysis. For ACS patients, the prognostic importance of SBP and DBP has been well documented. Nevertheless, the influence of PP on long-term mortality has been controversially discussed. In the present study, we aimed to examine the association between PP level on inpatient admission and long-term mortality in ACS patients who underwent PCI.

Study population
All consecutive in-hospital patients with ACS who underwent PCI in

Clinical outcomes
The outcomes examined in this study included all-cause mortality and cardiac mortality. Cardiac death was defined as death resulting from any cardiac events (eg, myocardial fraction, heart failure, fatal arrhythmia, and sudden death). All-cause death included both cardiovascular deaths and non-cardiovascular deaths.
The patients were followed-up at 1, 3, 6, 12, 24, 36, 48, 60, and 72 months. The outcomes were collected by phone calls or from the inpatient or outpatient medical records during the follow-ups.

Baseline characteristics
Of the 6779 patients with ACS who underwent PCI in the entire cohort, 1724 patients were excluded. The median duration of follow-up was 24 Compared with the low PP group, patients in the high PP group were older, more likely to be female, and had a higher prevalence of coronary artery disease, hypertension, and diabetes mellitus. They also had a higher likelihood of having three arteries or left main coronary artery stenosis after angiography. Because of the higher prevalence of coronary artery disease and hypertension in the high PP group, the patients in the high PP group more frequently took beta blockers, ACE inhibitors/ Angiotensin II receptor blockers, and Calcium channel blockers before admission and after discharge. After discharge, there were no differences in taking antiplatelet and statins between groups.
Compared with the other two groups, the patients in the intermediate PP group had a less incidence of STEMI or NSTEMI. They also had lower levels of creatinine, peak Creatine kinase-MB, peak NT-proBNP, while higher level of albumin and total triglyceride was detected. In the low PP group, both high hemoglobin level and low LVEF level were detected. During the follow-ups, the patients in the intermediate PP group experienced lower incidence of all-cause mortality and cardiac mortality.

Association between PP level and clinical outcomes
Death occurred in 216 patients, and cardiac death was found in 147 patients. Univariate and multivariate Cox regression analyses were carried out to evaluate the influence of selected covariables on the allcause and cardiac mortalities in the whole cohort and LVEF > = 0.5 cohort (Tables S1-S4). Cumulative survival curves for all-cause and cardiac mortalities in PP-dependent groups were shown in Figure 2.    Considering the correlation between levels of PP and SBP, multivariate Cox regression analyses were undertaken to investigate the association of SBP level with mortalities. There was no significant association between SBP level and mortality (Tables S5-S8).
To further analyze the nonlinear relationships between PP level and mortalities, the restricted cubic spline regression line with three knots was used. The nonlinear U-shape relationships between PP level on admission and all-cause and cardiac mortalities were observed in the whole cohort (p (for nonlinearity) = .005 and .003, respectively). However, in the LVEF > = 0.5 cohort, the U-shape relationships were not found (p (for nonlinearity) = .058 and .071, respectively). The reference PP level was equal to median PP level (55 mm Hg). The grey shaded area represented the area between the upper level and the lower level of 95% CI. In the LVEF > = 0.5 cohort, the grey shaded area for lower PP level was shifted down compared in the whole cohort. (Figure 3).
In the LVEF < 0.5 cohort, there was no significant association between PP level on admission and mortalities. Age and NT-proBNP level were identified as independent risk factors for mortalities. level and mortalities were observed in the whole cohort (p (for nonlinearity) = .009 and .008, respectively). However, in the LVEF > = 0.5 cohort, the U-shape relationships were not found (p (for nonlinearity) = .111 and .117, respectively). Details in Tables S9-S12 and Fig-ure S1.

Subgroup analysis
In the UA subgroup, no significant differences in HR were observed between PP groups. No statistical associations were found among PP and mortalities. In previous studies, whether lower PP level or higher PP level was the independent predictor of adverse events, the patients with higher PP level were older, and had higher proportion of females, and greater incidence of hypertension and diabetes mellitus, which were consistent with the present study. 21,22 We also found that patients However, after non-linear analysis, the U-shape relationships between PP level and mortalities were observed in the whole cohort, and the U-shape relationships disappeared in the LVEF > = 0.5 cohort.
The area between the upper and lower level of 95% CI for lower PP In the LVEF > = 0.5 cohort, for the cardiac mortality. Data were fitted by a Cox proportional hazards regression model that was based on restricted cubic splines and adjusted for age, sex, ACS diagnosis, atrial fibrillation, diabetes mellitus, malignancy (not for cardiac mortality), smoking history, percentage of three arteries or left main artery involvement, anemia, heart rate, body mass index, creatinine, albumin, and NT-proBNP peak. Solid black lines represent hazard ratios, and grey shaded areas represent 95% CIs and/or all-cause mortality, especially in patients with HF 10,11,22,31,32 and ACS. 13 In the Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study (EPHESUS) trial, patients with HF and LVEF less than 0.4 were enrolled. It was found that a low PP level was associated with adverse outcome. 10 PP level was reported to be more dependent on LVEF, rather than being a marker for aortic elasticity. 10 A low PP level may indicate the low cardiac output, which is the early sign of cardiogenic shock.
As in the previous study, 13,32 systolic blood pressure demonstrated strong correlation with pulse pressure. In several validated ACS risk scores, SBP-related variables have been included. In the Thrombolysis in Myocardial Infarction (TIMI) risk score for STEMI, low SBP level (< 100 mm Hg) was a strong risk factor for mortality. 33 In the Global Registry of Acute Coronary Events (GRACE) risk score, the lower SBP level attained the higher score, indicating the worse outcome. 34 However, in the present study, the relationships between SBP and mortalities were not found. Similar result about SBP was found in previous study. 32 Maybe PP is more accurate measure of the cardiac index. 35 In that study, the adequacy of cardiac output was assessed reliably by PP, not by SBP. There was a poor correlation between cardiac index and SBP.
After subgroup analyses, the significant association between PP and mortalities were only found in MI subgroup. Patients with MI are more likely to suffer from atherosclerotic diseases. And the cardiac function is more susceptible in the MI patients.
Some previous studies also showed the U-shape relationship between PP level and outcomes in different populations, 14,15,[36][37][38][39] which were mainly formulated by univariate analysis. After adjusted for the confounders, the U-shape curve disappeared. In the present study, the adjusted U-shape relationships were found in the whole cohort with the reference PP level of 55 mm Hg. The reference PP level was similar to that in a previous study. 36

Study limitations
The current study had several limitations. First, this was an observational study conducted in a single-center, restricting the generalization of the results. Second, we could not obtain data related to blood pressure and medication during follow-ups. Third, we did not exclude patients who had aortic valve disease, which might influence PP measurement.

CONCLUSIONS
According to the results of the present study, PP was not only found as an indicator of artery stiffness, but also as an indicator of heart function. In addition to high PP level, low PP level was also associated with increased risk of all-cause and cardiac mortalities in ACS patients undergoing PCI. Although PP level was correlated with SBP level, SBP level was not found to be linked with the risk of all-cause and cardiac mortalities. Further studies are warranted to determine the optimal PP level in ACS patients.

ACKNOWLEDGMENTS
The authors would like to thank Ms. Zhao Guoliang for her support dur-