The Impact of Complete Revascularization in Symptomatic Severe Left Ventricular Dysfunction between Coronary Artery Bypass Graft and Percutaneous Coronary Intervention

Objective The study aimed to compare the clinical outcomes between the patients receiving coronary artery bypass surgery (CABG) or percutaneous coronary intervention (PCI) for the patients with symptomatic severe left ventricular (LV) dysfunction and coronary artery disease (CAD). Methods Between February 2007 and February 2020, a total of 745 patients who received coronary artery angiography for reduced LV ejection fraction (LVEF) < 40% and symptomatic New York Heart Association (NYHA) functional class ≥ 3 were recruited. The patients (N = 236) who were diagnosed with dilated cardiomyopathy or valvular heart disease without coronary artery stenosis, those with prior history of CABG or valvular surgery (N = 59), those who presented ST-segment elevated myocardial infarction (STEMI), those with a CAD and SYNTAX score of  ≦ 22 (N = 175), those who received emergent CABG for coronary perforation (N = 3), and those who had NYHA class ≦ 2 (N = 65) were excluded. Finally, 116 patients with reduced LVEF and those who had a SYNTAX score >22, who received CABG (N = 47) and PCI (N = 69), were recruited for this study. Results There was no significant difference in the incidence values of in-hospital course and those of in-hospital mortality, acute kidney injury, and postprocedural hemodialysis. There was no significant difference in the 1-yearfollow-up of recurrent MI, revascularization, or stroke between the groups. The 1-year heart failure (HF) hospitalization rate was significantly lower in the CABG group than in all patients of the PCI group (13.2% vs. 33.3%; p = 0.035); however, there was no significant difference in the same variable between the CABG group and the complete revascularization subgroup (13.2% vs. 28.2%; p = 0.160). The revascularization index (RI) was significantly higher in the CABG group than in all patients of the PCI group or complete revascularization subgroup (0.93 ± 0.12 vs. 0.71 ± 0.25; p < 0.001) and (0.93 ± 0.12 vs. 0.86 ± 0.13; p = 0.019). The 3-year HF hospitalization rate was significantly lower in the CABG group than in all patients of the PCI group (16.2% vs. 42.2%; p = 0.008); however, there was no difference in the same variable between the CABG group and the complete revascularization subgroup (16.2% vs. 35.1%; p = 0.109). Conclusions In patients with symptomatic (NYHA class ≥ 3) severe LV dysfunction and CAD, CABG brought less HF admission when compared to patients in the PCI group, but this did not differ when compared to the complete revascularization subgroup. Therefore, an extensive revascularization, achieved by CABG or PCI, is associated with a lower HF hospitalization rate during the 3-yearfollow-up period in such populations.


Introduction
Coronary artery disease (CAD) is one of the leading causes of severe left ventricular (LV) dysfunction and mortality in recent years [1][2][3]. With the improvement of healthcare in CAD, the treatment strategies are changing day after day [4]. According to the reports from the American Heart Association, both the annual rate of death due to CAD in the USA and the incidence of LV dysfunction caused by CAD were increasing [5]. Tis fact made us pay more attention to this combination of severe LV dysfunction and CAD.
Current American and European guidelines do not provide precise recommendations for revascularization strategies in patients with CAD and severe LV dysfunction [6,7]. Nevertheless, the exclusion of patients with severe LV dysfunction from the clinical trial made the optimal revascularization of these patients still controversial [8][9][10]. Some cohort studies had concluded that coronary artery bypass surgery (CABG) was associated with a lower mortality rate and a lower major adverse cardiac event (MACE) rate than that with percutaneous coronary intervention (PCI) in patients with severe LV dysfunction CAD [11,12]. However, the evidence was not strong enough to conclude the result of revascularization strategies. Recently, one metaanalysis showed that among the patients with severe LV dysfunction, CABG resulted in a lower mortality rate and an increased risk of stroke [13]. Surprisingly, none of these cohort studies or meta-analyses stressed on patients' clinical conditions such as hemodynamic status, New York Heart Association (NYHA) functional class, or SYNTAX score. Tis retrospective study aimed to explore the optimal strategy for patients with symptomatic severe LV dysfunction with NYHA class ≥3 and CAD.

Materials and Methods
Te study population belonged to the HF registry of Kaohsiung Chang Gung Memorial Hospital. [14] Tis study was approved by the ethical committees of the institutional review board of Chang Gung Memorial Hospital (202100931B0) and the National Institution of Health of Taiwan.

Patients and Groups.
We retrospectively enrolled 745 patients who received coronary artery angiography for reduced LV ejection fraction (LVEF) < 40% between February 2007 and February 2020 (Figure 1). Te patients (N � 236) who were diagnosed with dilated cardiomyopathy or valvular heart disease without coronary artery stenosis, those with prior history of CABG or valvular surgery (N � 59), those with ST-segment elevated myocardial infarction, those with CAD and SYNTAX scores of ≦22 (N � 175), those who underwent emergency CABG for coronary perforation (N � 3), and those with NYHA class ≦2 (N � 65) were excluded. Tere were 116 patients with reduced LVEF and SYNTAX scores of >22 who received CABG (N � 47) and PCI (N � 69) who were enrolled into this study. Te Institutional Review Committee on Human Research at our institution approved the study protocol.

Defnitions.
Our myocardial infarction (MI) defnitions followed the most recent universal defnition of MI [15]. Symptomatic heart failure (HF) was defned, according to the NYHA classifcation, as being in a class of ≥3. Revascularization was defned as any repeat PCI in a target vessel or coronary artery bypass graft (CABG) in a target vessel for lesions with a stenosis of ≧70%. Cardiovascular (CV) mortality was defned as death due to MI, cardiac arrhythmia, or HF. All-cause mortality was defned as death from any cause. Recurrent MI was defned as acute MI occurring 1 month after the index MI. Complete revascularization was defned as the absence of any angiographic signifcant stenosis (>70%) in all three epicardial coronary arteries with a diameter of at least 2.5 mm after PCI.

Study Endpoints.
Te primary composite endpoints of our study were any recurrent MIs, revascularization, sudden death/ventricular arrhythmia, HF hospitalization, stroke or CV mortalities, and all-cause mortality rate during the 1-and 3-yearfollow-up periods. Te secondary composite endpoints included all in-hospital events, such as in-hospital mortality, acute kidney injury, and postprocedural hemodialysis (HD).

Statistical
Analysis. Data were expressed as the mean-± standard deviation for continuous variables or as counts and percentages for categorical variables. Continuous variables were compared using the independent sample t-test or the Mann-Whitney U test. Categorical variables were compared using the chi-square test. Univariate and multivariate cox regression analyses were performed to identify the associations relating to the 1-year CV mortality rate. Correlations between variables were expressed as hazard ratios with 95% confdence intervals. Te Kaplan-Meier curves were created to illustrate the 1-year CV mortality data in each of the groups. All statistical analyses were performed  Te surgical risk and coronary complexity parameters such as the New EuroScore II were higher in the CABG group than in all patients of the PCI group or the complete revascularization subgroup (9.2 ± 7.2 vs. 3.6 ± 2.0; p < 0.001) or (9.2 ± 7.2 vs. 3.5 ± 1.7; p < 0.001). Te SYNTAX score was also higher in the CABG group than in all patients of the PCI group or the complete revascularization subgroup (40.4 ± 10.5 vs. 35.9 ± 8.2; p � 0.010) or (40.4 ± 10.5 vs. 35.4 ± 8.0; p � 0.014). Further analysis revealed no signifcant diference in coronary complexity parameters such as SYNTAX score > 33, left main disease, multivessel disease, two-vessel disease, three-vessel disease, or chronic total occlusion (CTO) between the groups.

Baseline Characteristics.
Te complete revascularization parameter was assessed by the British Cardiovascular Intervention Society's myocardial jeopardy score (BCIS-JS) [16,17] before and after PCI. Te revascularization index was calculated by (preB-CIS-JS-postBCIS-JS)/(preBCIS-JS). Te BCIS-JS before the procedure was also signifcantly higher in the CABG group than in all patients of the PCI group or the complete revascularization subgroup (11.8 ± 0.6 vs. 10.6 ± 1.7; p < 0.001) and (11.8 ± 0.6 vs. 10.4 ± 3.0; p < 0.001). Te BCIS-JS after the procedure was also signifcantly higher in the CABG group than in all patients of the PCI group (0.9 ± 0.4 vs. 3.1 ± 2.8; p < 0.001), but no signifcant diference was observed in the CABG group and the complete revascularization subgroup (0.9 ± 0.4 vs. 1.4 ± 1.3; p � 0.064). Both the revascularization index (RI) was higher than 0.67 in all groups but still signifcantly higher in the CABG group than in all patients of the PCI group or the complete revascularization subgroup (0.93 ± 0.12 vs. 0.71 ± 0.25; p < 0.001) and (0.93 ± 0.12 vs. 0.86 ± 0.13; p � 0.019).
Tere was no signifcant diference in the rate of use of ACEI/ARB/ARNi and MRA between the groups. However, the rate of use of beta-blockers was signifcantly lower in the CABG group than in all patients of the PCI group (62.2% vs. 85.3%; p � 0.007) or the complete revascularization subgroup (62.2% vs. 87.8%; p � 0.007). Also, the rate of use of furosemide was also signifcantly lower in the CABG group than in all patients of the PCI group (31.9% vs. 59.4%; p � 0.005) or the complete revascularization subgroup (31.9% vs. 50.0%; p � 0.090). Tere was no signifcant diference in the mean follow-up duration between the groups. Table 2 demonstrates the echocardiography parameters at the baseline and follow-up. Tere were no signifcant diferences in baseline echocardiography parameters such as LA dimension, LVEF, LVEF < 30%, LVEDV, LVESV, AR grade > 2, MR grade > 2, TR grade > 2, and TRPG between the CABG group and all patients of the PCI group. Among the follow-up echocardiography parameters, there were no signifcant diferences in LA dimension, LVEF, LVEF > 50%, LVEF > 40%, LVEDV, LVESV, AR grade > 2, MR grade > 2, TR grade > 2, and TRPG between the CABG group and all patients in the PCI group. Tere were also no signifcant diferences in a reduction in LVEDV of > 10% or a reduction in LVESV of > 10% between the CABG group and in all patients of the PCI group. Tere were no signifcant diferences in the improvements in LVEF or mean LVEF > 10% between the CABG group and in all patients of the PCI group and the complete revascularization subgroup (14.7 ± 15.5 vs. 15.8 ± 17.3 vs. 18.9 ± 16.6) and (62.8% vs. 64.3% vs. 69.7%). Table 3 illustrates the long-term clinical outcomes of groups. Tere were no signifcant diferences in the incidences of in-hospital course and inhospital mortality, acute kidney injury (AKI), and postprocedural HD. Tere were no diferences in primary composite endpoints at 1-year or 3-yearfollow-up comparing CABG vs. all patients in the PCI group and the complete revascularization subgroup (29.8% vs. 42.0%, p � 0.240; 29.8% vs. 40.5%, p � 0.374) (42.6% vs. 53.6%, p � 0.262; 42.6% vs. 52.4%, p � 0.399). Tere was no statistically signifcant diference in the 1-yearfollow-up between the groups of recurrent MI, revascularization, or stroke, and they were similar; however, the 1-year HF hospitalization rate was signifcantly lower in the CABG group than in all patients of the PCI group (13.2% vs. 33.3%; p � 0.035);   however, the diference in the same parameter between the CABG group and the complete revascularization subgroup was not statistically signifcant (13.2% vs. 28.2%; p � 0.160). Te incidences of sudden death/ventricular arrhythmia, CV mortality, and all-cause mortality were higher in the CABG group; however, the diferences were not statistically signifcant.

Clinical Outcomes.
Tere was no statistically signifcant diference in the 3yearfollow-up between the recurrent MI, revascularization, and stroke groups. Te 3-year HF hospitalization rate was signifcantly lower in the CABG group than in all patients of the PCI group (16.2% vs. 42.2%; p � 0.008); however, the diference in the same parameter between the CABG group and the complete revascularization subgroup was not statistically signifcant (16.2% vs. 35.1%; p � 0.109). Te incidences of sudden death/ventricular arrhythmia, CV mortality, and all-cause mortality were higher in the CABG group, although the diferences were not statistically signifcant. Figure 2 shows the Kaplan-Meier curve illustrating the diferences in the 1-and 3-yearall-cause mortality rates between the groups. Tere was no signifcant diference in the 1-yearall-cause mortality rate between the CABG group and in all patients of the PCI group (p � 0.197) and between the CABG group and the complete revascularization subgroup (p � 0.631). Tere was no signifcant diference in the 3-yearall-cause mortality rate between the CABG group and in all patients of the PCI group (p � 0.298) and between the CABG group and the complete revascularization subgroup (p � 0.741). Figure 3 shows the Kaplan-Meier curve illustrating the diference in the 1-and 3-year HF hospitalization rates between the groups. Tere was some trend lower but without a statistically signifcant diference in the 1-year HF hospitalization rate between the CABG group and in all patients of the PCI group (p � 0.061) or between the CABG group and the complete revascularization subgroup (p � 0.129). Tere was also some trend lower but without a statistically signifcant diference in the 3-year HF hospitalization rate between the CABG group and in all patients of the PCI group (p � 0.050) and between the CABG group and the complete revascularization subgroup (p � 0.144).

Discussion
Te most common cause of severe LV dysfunction is coronary artery disease. Despite numerous studies made, atherogenesis and its progression leading to endothelial injury remain multifactorial which include smoking, hypertensive trigger, and oxidative stress. One of the most defnite progressions of atherosclerosis is coronary artery calcifcation. Coronary artery calcifcation has been used for staging coronary atherosclerosis. Te mechanism of coronary artery calcifcation may lie in the release of apoptotic/necrotic bodies, the release of matrix vesicles, and the diferentiation of pericytes during the plaque calcifcation process [18]. Survivors after adequate treatment of acute coronary syndrome and myocardial infarction may develop severe LV dysfunction due to irreversible myocyte loss with scar formation, hibernating myocardium, or adverse cardiac remodeling [19].
Te meta-analysis results of patients with severe LV dysfunction and CAD who were treated with CABG had a higher risk of stroke during the short-termfollow-up; however, they had a lower risk of death, MI, and repeat revascularization than those treated with PCI during the long-termfollow-up [13]. Randomized controlled trials on revascularization strategies in patients with severe LV dysfunction CAD are rare, except for the Heart Failure Revascularization Trial [20] and Surgical Treatment for Ischemic Heart Failure Extension Study [21]. Both of these randomized controlled trials were not set up by comparing CABG and PCI in patients with severe LV dysfunction. Conversely, most randomized trials only compared CABG with PCI in patients whose LV function was not severely depressed.
Te NYHA classifcation has served as a fundamental tool for the risk stratifcation of HF and for determining the clinical trial eligibility and candidacy for devices and drugs. Recently, one systemic review showed that the NYHA system poorly discriminated against patients with HF across the spectrum of functional impairment [22]. However, other studies showed that the NYHA classifcation, along with other comorbidities, still might be helpful in identifying a subgroup of implantable cardioverter defbrillator carriers with poor prognoses and a higher risk of CV death [23]. To the best of our knowledge, there is no existing study that compares CABG and PCI in patients with symptomatic severe LV dysfunction CAD.
Current evidence from randomized trials suggests that the risk of mortality and MI does not difer signifcantly among various revascularization strategies for multivessel CAD. A meta-analysis showed that complete revascularization of multivessel CAD was associated with a reduction in the rate of MACE due to the reduction in emergency revascularization [24]. However, the importance of complete revascularization had been addressed only to reduce repeat revascularization and subsequent major adverse cardiac and cerebral events in patients with LV dysfunction but not to reduce the long-term mortality rate [25]. One of the most important reasons that complete revascularization cannot achieve long-term clinical outcome improvement is the variety of complete revascularization defnitions. In the COMPLETE trial [26], the defnition of complete revascularization was on the completeness of nonculprit lesion PCI which is defned as the vessel which is larger than 2.5 mm with angiographic more than 70% stenosis or with 50 to 69% stenosis but a fractional fow reserve (FFR) of less than 0.80. Te DANAMI-3-PRIMULTI trial [27] also defned complete revascularization as revascularization of all coronary lesions which are greater than 50% in diameter stenosis and larger than 2.0 mm in diameter which was confrmed by FFR. Another reason that complete revascularization cannot achieve long-term clinical outcome improvement is because the extensive revascularization is not well defned. Te British Cardiovascular Intervention Society's myocardial jeopardy score (BCIS-JS) was developed to defne the completeness of coronary revascularization [16]. Revascularization index (RI) was defned as (pre-BCIS-JS-postBCIS-JS)/(pre-BCIS-JS) which showed that RI < 0.67 is associated with a higher mortality rate. In our study, the RIs were signifcantly lower in all patients of the PCI group and the complete revascularization subgroup compared with the CABG group suggesting that the degree of extensive revascularization is still diferent. In our study, the long-term mortality rate did not difer signifcantly in CABG, all patients in the PCI group, or the complete revascularization subgroup, and even in patients with symptomatic (NYHA class more than 3) and severe LV dysfunction who had less HF admission due to the diferences in complete revascularization.
Elective insertion of the hemodynamic support device in complex high-risk indicated patients (CHIP) may be reasonable as an adjunct to PCI or CABG which was a class IIb recommendation according to guidelines [28]. BCIS-1 study showed that in patients with severe ischemic cardiomyopathy treated with PCI, elective IABP use during PCI was associated with a 34% relative reduction in all-cause mortality compared with unsupported PCI [29]. Te evidence of prophylactic ECMO in CHIP PCI was insufcient which also showed benefts in case series results [30,31]. PROTECT II study showed that in 452 symptomatic patients with complex three-vessel coronary artery disease or unprotected left main coronary artery disease and severe depressed left ventricular function underwent nonemergent PCI, although the 30-day incidence of major adverse events was not different between IABP and Impella 2.5, but improved outcomes were observed for Impella 2.5 group at 90 days with signifcant maintenance of mean arterial pressure during PCI [32,33]. Roma-Verona registry also demonstrated that the Impella-protected PCI could achieve more complete revascularization with a higher revascularization index with signifcant improvement in the left ventricular ejection fraction and survival [34]. Te Impella-supported intervention also showed its beneft not only in CHIP PCI but also in cardiogenic shock patients in the IMP-IT registry [35]. Tis evidence provides us more confdence in extensive revascularization under mechanical support such as IABP, ECMO, or Impella. However, bleeding complications under mechanical support are common. One of the most leading causes of bleeding complications was the cleavage of von Willebrand factor (VWF) when using ECMO or left ventricular assist devices (LVADs). Te basic VWF monomer contains 2050 amino acids making the protein sensitive to changes in fuid fow and shear stress. LVADs have been associated with an increased bleeding risk, commonly attributed to hydrodynamic changes and shear stress induced by continuous fow across the devices, which augmented VWF cleavage also seen in severe LV dysfunction patients [36]. Tese made one-stage surgery such as CABG instead of prolonging the use of mechanical support devices more suitable for clinical decision-making [37].

Limitations.
Our study had several limitations. First, we enrolled patients from 2007 to 2020 due to the limited number of patients, which represented selection bias. Second, the PCI and CABG were all operator-dependent procedures that infuenced the true clinical outcome.
Tird, there was no analysis between drug-eluting stents or bare-metal stents in the PCI group.

Conclusions
In patients with symptomatic (NYHA class ≥ 3) severe LV dysfunction and CAD, CABG brought less HF admission when compared to patients in the PCI group but did not  difer when compared to the complete revascularization subgroup. Terefore, extensive revascularization, achieved by CABG or PCI, is associated with a lower HF hospitalization rate during a 3-yearfollow-up period in such populations.

Data Availability
Te data used to support the fndings of the study can be obtained from the corresponding author upon request.