Prognostic Significance of Percutaneous Coronary Intervention for First Acute Myocardial Infarction with Heart Failure: Five-Year Follow-Up Results

Objective The study aimed to investigate the incidence and influencing factors of heart failure after 5 years of percutaneous coronary intervention (PCI) for first acute myocardial infarction. Methods A total of 1235 patients, diagnosed as acute myocardial infarction and treated with PCI in Beijing Anzhen Hospital, Capital Medical University, from January 1, 2014, to December 31, 2014, were enrolled. Based on the exclusion criteria, 671 patients were followed up to obtain echocardiographic results 5 years after the onset of myocardial infarction (from January 1, 2019, to December 31, 2019). Of 671 patients, 62 were lost to follow-up. Finally, 609 patients were recruited in this study. According to the results of the echocardiographic examination, patients were divided into a heart failure group (n = 97) (LVEF < 50%) and a nonheart failure group (n = 512) (LVEF ≥ 50%). The clinical characteristics were compared between the two groups, and the influencing factors of heart failure after 5 years of PCI in patients with acute myocardial infarction were analyzed using logistic regression and receiver-operating characteristic (ROC) analyses. Results Of 609 patients, 97 had heart failure within 5 years after PCI for first myocardial infarction, with an incidence of 15.9%. Multivariate regression analysis finally examined the predictors related to the occurrence of heart failure, including age (aOR, 1.008; 95% confidence interval (CI), 1.054–1.123; P ≤ 0.001), peak troponin I level (aOR, 1.020; 95% CI, 1.006–1.034; P = 0.004), left ventricular ejection fraction (LVEF) (during admission) (aOR, 0.908; 95% CI, 0.862–0.956; P ≤ 0.001), and left ventricular end-diastolic dimension (LVEDD) (at admission) (aOR, 1.136; 95% CI, 1.016–1.271; P = 0.025). Conclusion In this study, the incidence of heart failure (LVEF < 50%) in patients with acute myocardial infarction who underwent PCI was 15.9% at a five-year follow up. Age, peak troponin I level, and LVEDD (at admission) were risk factors for heart failure, while LVEF (at admission) of patients during hospitalization was a protective factor for heart failure.


Introduction
Coronary artery disease (CAD), which is approximately responsible for 9 million deaths in 2016, is one of the leading causes of death globally [1]. Acute myocardial infarction (AMI) is one of the most serious manifestations of CAD. Te increasing use of coronary revascularization and advancement in medical therapy have improved survival rates of patients with AMI [2]. . A large-scale epidemiological survey shows that in China, the in-hospital mortality rate of AMI is approximately 6.58% [3], while most patients have long-term survival after having AMI. Some patients continue to develop heart failure (HF) for several years after percutaneous coronary intervention (PCI), resulting in repeated hospitalizations and even death. It has been proven that the improved treatment of AMI has contributed to the epidemic of HF [4][5][6][7].
Te results of a prospective study in the United Kingdom in 12 years (1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010) suggest that approximately one in four patients will develop HF within 4 years of frst AMI [8]. A Swedish study found that the cumulative risk of HF 5 years after AMI was 21.8% [9]. Tese studies indicate that the incidence of HF after AMI is high. It is necessary to detect and identify patients with AMI who are at risk of progression to HF and provide adequate and even preventive HF treatment as soon as possible.
Relevant studies have identifed risk factors for HF after AMI, such as old age, severity of CAD, infarct size, and adequacy of reperfusion therapy [10,11]. However, there are still insufcient studies on patients with frst AMI. Te prediction of HF after AMI by clinicians is still subjective and empirical. Terefore, the control rate of HF after AMI is still low. In an electronic health record cohort study in England (1998England ( -2010, only 9.2% of patients underwent primary PCI [8]. Moreover, previous studies on HF after AMI reported that it is difcult to ensure that patients receive a timely and uniform rescue process [8,9,12], which is greatly important in the treatment of patients with AMI.
Terefore, this retrospective study aimed to determine the incidence and identify independent predictors of HF after 5 years of PCI of frst AMI. In this study, all patients with frst AMI received PCI in our center, which is one of the largest heart centers in China.

Study Population.
Tis study, following the 2013 Declaration of Helsinki [13], was approved by the institutional review board and deemed exempt from informed consent requirements.
All patients were identifed from a retrospective review of the institution's database with records of detailed information, including baseline characteristics, laboratory examinations, echocardiographic measurements, in-hospital angiographic characteristics, and medications used during hospitalization. From January 1, 2014, to December 31, 2014, 1235 patients diagnosed with AMI including ST-segment elevation myocardial infarction (STEMI) and non-ST segment elevation myocardial infarction (NSTEMI) were treated with PCI in AnZhen Hospital. According to the exclusion criteria, 671 patients were included (Figure 1.). Tese patients were followed up with echocardiography (from January to December 2019) in our hospital to determine cardiac function. During the follow-up, 62 patients were lost. Finally, a total of 609 patients were enrolled in this study. Figure 1 shows the fowchart of the enrolled population in this study.

Data Collection.
Te clinical data recorded in this study included demographic characteristics, laboratory examination during hospitalization, echocardiographic results, details of PCI treatment during hospitalization, and medication regimen. Te demographic data included age, sex, body mass index, and medical history, including smoking, hypertension, diabetes, atrial fbrillation, and stroke. Laboratory examination included peak troponin I (TNI) level, peak creatine kinase-MB (CK-MB), white blood cell (WBC) count, neutrophil count ratio, platelet count, hemoglobin count, high-sensitivity C-reactive protein (Hs-CRP), uric acid, creatinine, cereal third transaminase (ALT), aspartate aminotransferase (AST), total protein, total bilirubin, urea nitrogen, HbA1c, fasting plasma glucose, triglyceride, high-density lipoprotein cholesterol, lowdensity lipoprotein cholesterol, hematocrit, and brain natriuretic peptide (BNP) levels. Echocardiographic results during admission, including left ventricular ejection fraction (LVEF), left ventricular end-diastolic dimensions (LVEDD), left ventricular end-systolic dimension (LVESD), segmental wall motion abnormality, and ventricular aneurysm, were also recorded. All examinations were performed within 24 h of admission. In-hospital angiographic characteristics included AMI type (STEMI/NSTEMI), preinfarction angina, time from attack to PCI (h), thrombolytic therapy, PCI vessels, culprit vessels, single-vessel disease, TIMI fow (before and after PCI), absence of refow, complete revascularization, and use of the intra-aortic balloon pump (IABP) and extracorporeal membrane oxygenation (ECMO). Medication use during hospitalization was also included.

Statistical Analyses.
Te data were analyzed using the SPSS statistical package, version 22.0 (IBM corporation, Armonk, NY). Continuous variables were presented as a mean ± standard deviation or median (range) values and compared using the independent-sample t-test (data with normal distribution) or the Mann-Whitney U test (data with non-normal distribution). Categorical variables were expressed as numbers (percentage) of patients in each group and analyzed using the chi-squared test. Variables with a statistical signifcance between the two groups were incorporated into a single-factor logistic regression analysis to select meaningful predictors and entered into the multifactor logistic regression analysis to fnally obtain independent predictors of HF. Te ROC curve based on the logistic model was established to evaluate the prediction probability. A P value <0.05 (bilateral) was statistically signifcant.

Defnition.
AMI caused by atherothrombotic CAD and usually precipitated by atherosclerotic plaque disruption (rupture or erosion) is designated as type 1 AMI [14]. Chronic total occlusion (CTO) is defned as a coronary lesion with TIMI fow grade 0 of at least three months duration, which is frequently encountered during coronary angiography in patients with coronary artery disease (CAD) [15]. Te defnition of complete revascularization is as follows: noninfarcted vessels with signifcant stenosis (diameter stenosis rate 70%) were also treated during PCI [16]. HF is defned as an ejection fraction (EF) < 50%. Previous AMI was defned as a clear history of NSTEMI or STEMI or a Q-wave on electrocardiography. Previous revascularization is defned as previous PCI or coronary artery bypass grafting or thrombolytic therapy for AMI.
Preinfarction angina was defned as ≥1 episode of angina within 48 h prior to AMI [17]. Culprit vessels of STEMI are determined by the characteristic ST-segment elevation or Qwave formation on electrocardiography. Culprit vessels of NSTEMI were determined by an experienced cardiologist based on angiographic results, electrocardiographic changes, and echocardiographic fndings.

Incidence of HF.
A total of 671 patients met the inclusion criteria. During the follow-up, 62 patients were lost. Finally, 609 patients were included in the statistical analysis. Based on the results of the patient's echocardiography in 2019, 97 patients in the HF group (LVEF < 50%) and 512 patients in the non-HF group (LVEF ≥ 50%) were analyzed. Figure 1 shows the fowchart of the enrolled population in this study.

Baseline Characteristics, Laboratory Examinations, and
Echocardiographic Measurements. Table 1 shows baseline data, laboratory examination, and echocardiographic results of patients in the two groups. In terms of baseline characteristics, patients in the HF group were older and had lower systolic blood pressure than patients in the non-HF group.
Regarding laboratory examinations, many examination results of patients in the HF group were higher than those in the non-HF group, including WBC count, neutrophil count, hematocrit, TNI, CK-MB, Hs-CRP, ALT, total bilirubin, fasting plasma glucose, and BNP levels.

In-Hospital Angiographic
Characteristics. Furthermore, the results of the comparison of angiography and reperfusion therapy during hospitalization between the two groups of patients were as follows (Table 2): frst, the proportion of patients with STEMI in the HF group was signifcantly higher than that in the non-HF group, and the proportion of patients with preinfarction angina in the HF group was signifcantly lower than that in the non-HF group. However, there was no signifcant diference in the time from attack to PCI between the two groups. In terms of target vessels of PCI, patients in the HF group received signifcantly higher percentage of the LAD artery and RCA revascularization than those in the non-HF group, but no signifcant diference was found in the LCX. Te same results were obtained in terms of culprit vessels. In terms of vascular disease severity, single-vessel disease was more likely to develop in patients with HF than in patients without HF. As for TIMI blood fow before PCI, the HF group tended to have more TIMI grade fow 0, while the non-HF group tended to have more TIMI grade fow 3.
Tere was no diference in TIMI fow after PCI between the two groups. Tere were no diferences in absence of refow, stent placement, and complete revascularization between the two groups. Table 3 compares the diference in drug use during hospitalization between patients in the HF group and those in the non-HF group. It was found that all patients used aspirin and either clopidogrel or ticagrelor. Tere was no signifcant diference in the antiplatelet agent and low-molecular-weight heparin (LMWH) use; however, tirofban use was signifcantly higher in the HF group than in the non-HF group. Use of other drugs, beta-blockers and spironolactone, was signifcantly diferent between the two groups, and patients in the HF group tended to use more of both drugs. Te abovementioned diferences were statistically signifcant.

Discussion
With the rapid development of revascularization technology and drug therapy, the main burden in AMI has changed from high in-hospital mortality to a poor prognosis after discharge [4]. Several studies have shown that approximately a quarter of patients develop HF within a few years of hospital discharge [8,9,19]. Predicting the risk of HF after AMI is of great signifcance in the prognosis and treatment of AMI. Compared with many previous studies [8,9,19], a novel defnition of heart failure (EF < 50%) was used in this study. Traditionally, HF has been divided into distinct phenotypes based on the measurement of the left ventricular ejection fraction (LVEF) [18]. Te rationale behind this relates to the original treatment trials in HF that demonstrated substantially improved outcomes in patients with an LVEF < 40% [18]. However, HF spans the entire range of the LVEF (a normally distributed variable), and measurement by echocardiography is subject to substantial variability [18]. Compared with >40% in many previous studies [8,9,19], the defnition of >50% can include more patients. Te patients with HFpEF (those with symptoms and signs of HF, with evidence of structural and/or functional cardiac abnormalities and/or raised natriuretic peptides (NPs), and with an LVEF > 50%) were excluded in this study because its diagnosis is more complicated than the frst two types of HF, and it is not solely dependent on EF examination. Te occurrence of HFpEF in patients with AMI is a complicated process [12]. It involves the diastolic function of the heart, and there may be other specifc pathological mechanisms [20,21]. Te incidence of HF in the study was 15.8% 5 years after frst AMI, which was lower than those in previous studies when the diagnostic criteria for HF were more extensive [8]. Te possible reason was that all patients in this study had frst AMI and received PCI. Tis is also related to the higher level of treatment received by patients.
In this study, patients in the HF group were older, and there are sufcient data to show that advanced age can predict the risk of in-hospital death in patients with AMI [22]. Regarding the relationship between senile HF and AMI, studies have shown that aging may be a cause of HF after ischemia [23,24]. Epidemiological studies have shown that in Europe and America, for 25 years, the number of patients with chronic HF increased by 70-100%. Te reason for this may be that most patients with incipient AMI who received ischemia reperfusion therapy developed chronic HF years later [23,24]. Basic studies have shown that senescence has the efect of inducing apoptosis of cardiomyocytes, which may be the mechanism of late-onset ischemic HF in elderly patients [25,26]. Te results of the multivariate regression analysis in this study showed that age was an independent predictor of HF (OR, 1.008; 95% CI, 1.054-1.123; P ≤ 0.001).
Te infarction area in patients with AMI is closely related to death and long-term prognosis of patients during hospitalization. Since the myocardial size is difcult to regenerate, damaged cells are often replaced by cells without systolic function, so the infarction area can be an important factor in the development of HF. However, due to the high cost and complexity of accurate measurement of the AMI area, such as myocardial nuclide imaging and myocardial magnetic resonance imaging, such tests are often not used in the clinical treatment of patients with AMI. However, a variety of laboratory tests can indirectly or directly refect the size of the infarction area in patients. Some studies have shown that the WBC count, neutrophil count, and CRP level are related to the infammatory response caused by the expansion of the infarction area [27][28][29][30]. In this study, the WBC count, neutrophil count, and CRP level of patients in the HF group were higher than those in the non-HF group, but regression analysis failed to show that these indicators were independent predictors of HF, which may be related to the inability of these test results to accurately evaluate the infarction area.
Moreover, myocardial necrosis indicators (CK-MB and TNI) and liver biomarkers (TB, ALT, and AST) can also refect the area size of myocardial injury. Studies have already confrmed that all myocardial enzyme levels are associated with adverse outcomes of AMI [31]. In this study, all myocardial enzyme levels of patients in the HF group were higher than those of patients in the non-HF group. Multiple factor regression analysis showed that the TNI level is an independent predictor of HF in patients with AMI, and the ROC curve analysis showed that TNI had the best sensitivity and specifcity.
Others may be associated with the prognosis of AMI. However, the laboratory tests failed to predict the efect of HF after AMI occurrence. For example, a study reported that diabetes and stress hyperglycemia are the criteria of AMI in patients at high risk [32]. In this study, although blood glucose levels were higher in patients with HF, the predictive efect regression analysis did not fnd a signifcant diference. Tis may be because the hospital once measured blood glucose levels associated with the patient's stress and diet and did not refect the true blood glucose level in patients [32]. Additionally, statistically signifcant diferences were not found in various lipid tests between groups. Although studies have found that the renal function level in patients with AMI is an independent predictor of the prognosis of patients with AMI [33], this study failed to fnd a correlation between abnormal renal function and HF in the majority of patients with normal renal function. Te abovementioned laboratory examination results need to be clarifed by prospective studies because of the large uncontrollable factors in observational studies.
Liu et al. showed that the level of cardiac function during hospitalization in patients with AMI could predict adverse cardiac events after PCI, suggesting that decreased left ventricular diastolic function and left ventricular dilatation were predictors of adverse events after PCI [34]. In this study, patients in the HF group showed a signifcant diference in LVEF and LVEDD compared with those in the non-HF group, and regression analysis showed that LVEDD was an independent predictor of HF, while LVEF was a protective factor. Te results showed that in patients with severely affected cardiac function after AMI, the risk of development of HF after AMI has increased signifcantly. On the one hand, cardiac function may refect the severity of AMI. On the other hand, in the event of HF, a series of HF neurohumoral factors is activated (e.g., activation of the sympathetic nervous system and renin-angiotensin-aldosterone system). Tese factors may persist after discharge from the hospital. In this study, BNP levels in the HF group were higher than those in the non-HF group, but the predictive efect of BNP was not found in the regression analysis. Tis may be due to the difculty in controlling the infuential factors of laboratory examination results in observational studies, which need to be further clarifed by prospective studies. Regarding ventricular wall motion abnormality and ventricular wall tumors, some studies have found that the score of ventricular wall motion abnormality based on the echocardiographic results is of great signifcance in the determination of AMI severity. However, because its determination is relatively subjective, no signifcant diference was found between the two groups in this study. Similarly, there was no signifcant diference between the HF and non-HF groups. For such indicators, further research is needed to evaluate them on the premise of accurate detection.
It has been reported that patients with angina pectoris before infarction who received thrombolytic therapy had signifcantly lower rates of cardiogenic shock, malignant arrhythmia, and mortality than patients without angina pectoris before infarction [35]. More data indicate that preinfarction angina pectoris has a myocardial protective efect on AMI, and its mechanism may be related to myocardial ischemic preconditioning induced by preinfarction angina pectoris [35]. In this study, the incidence of preinfarction angina in the HF group was signifcantly lower than that in the non-HF group, confrming the results of the abovementioned study.
Te results of this study showed that the type of AMI was not a predictor of HF after AMI, although the HF group had a higher proportion of STEMI than the non-HF group, and despite diferences between the two groups, neither the ofender nor the target vessel was a predictor of HF. Te results are consistent with those of a large study in the UK [8]. Te relationship between the type of AMI and prognosis of AMI has long been controversial [8]. A study in Japan showed that patients with NSTEMI had poorer long-term prognosis than patients with STEMI and had a higher rate of HF after AMI [36]. Tis problem still needs to be further elaborated by prospective studies.
As discussed above, the incidence of HF (LVEF < 50%) in patients with AMI 5 years after PCI was 15.8%, among which age, peak TNI level, and LVEDD were risk factors for HF, while LVEF of patients during hospitalization was a protective factor for HF. In clinical work, patients with AMI with high-risk factors should be paid close attention. In these patients, the changes in cardiac function should be considered, and preventive and anti-HF treatment should be performed, when necessary. We have designed prospective studies to further clarify the fndings of this study and will discuss in further work whether intervention with these risk factors can reduce the risk of HF in patients with AMI.

Limitations
Tere are some limitations in this study. Tis was a retrospective observational study that screened a subset of patients, which may have led to population selection bias. In addition, this study failed to distinguish the types of myocardial infarction, and diferent types may have different mechanisms. Finally, the treatment received by patients after discharge was not followed up accurately, which may afect patient outcomes. We will further design prospective studies and closely follow patients to further explore this issue.

Data Availability
Te data that support the fndings of this study are available from the corresponding author upon reasonable request.

Ethical Approval
Ethical approval is not required for this study.