For most patients with stable ischemic cardiomyopathy (SICM), PCI neither improves LVEF nor heart failure outcomes. We sought to ascertain if specific subgroups of SICM patients might have a LVEF and clinical benefit from PCI. From a cohort of 1702 consecutive SICM patients with LVEF ≤50% treated between 2009 and 2023, patients were randomly selected for screening to meet final inclusion criteria including target vessel subtending ≥20% of the LV, hibernation+ischemia>scar, timely pre- and post PCI echos and no intercurrent treatment that might influence change in LVEF, until a cohort of 200 patients was available. Parsimoniously selected variables were then assessed for correlation with change in LVEF, with 1000-fold bootstrapping to minimize overfitting. Correlation of change in LVEF with freedom from cardiac death or heart failure admission was assessed with Cox multivariable analysis. Mean age was 69 ± 10 yrs, 76% were male, 52% had diabetes and baseline LVEF obtained a median 9 days before PCI was 35% ± 11%. Post-PCI LVEF obtained at a median 132 days was 39% ± 11%. Median follow-up was 35 months. After consideration of medication usage, the presence of an ICD, baseline LVEF >35%, prior CABG and scar all were independently and negatively correlated with improvement in LVEF (p ≤0.035). Patients with none of these factors had an improvement in LVEF of 5.5% ± 7.4%. Post-PCI LVEF and change in LVEF were independently correlated with reduced risk of clinical events (p = 0.003 and p = 0.032, respectively). If validated, these data will change the paradigm that no patients with SICM have a heart failure or mortality benefit from PCI.
Left ventricular function, as ascertained by LVEF, has been shown to be a powerful predictor of clinical outcomes in many settings. ,,, Impaired LVEF may be due to scarring, hibernation, or ischemia. Hibernating myocardium is characterized by a putatively protective hypocontractile response to sublethal reduction in coronary blood flow, , although when prolonged the reduced flow can lead to the myocardial cells “sleeping to death.” Improved contractility with dobutamine is the most well accepted marker of myocardium likely to recover with reperfusion, although recent data also suggests that fractional flow reserve (FFR) can be predictive.
In the important REVIVED-BCIS Trial, which showed that the typical patient with SICM derived no survival and heart failure-related benefit from PCI, scar was most often assessed with MRI. This leaves open the possibility that a select minority of patients still might benefit. In this study we sought to further identify correlates of improved LVEF after PCI, and to evaluate the effect of change in LVEF on subsequent risk of cardiac death or hospitalization for heart failure.
Methods
Patient population
Screening for suitable candidates for the study was performed using our IRB-approved PCI Database which collects ACC/NCDR and other data on consecutive patients undergoing PCI at our institution. The screening flow diagram to acquire 200 qualifying patients is shown in Figure 1 . Information regarding baseline and long-term outcome data collection in this Registry has been published.
Project overview and patient identification.
Hypotheses and outcome measures
A priori, we hypothesized that one or more of an intentionally limited lists of potential covariates would be associated with improvement in LVEF after PCI. Secondarily, we hypothesized that improvement in LVEF would correlate with reduced risk of cardiac death or hospitalization for heart failure during follow-up.
Analysis and reporting
Qualifying echocardiograms were required to be within 6 months before and up to 18 months after the PCI, without intercurrent intervention such as CRT or TAVR that might have affected LVEF. Echocardiograms were initially read by the Cleveland Clinic Echocardiography staff. For our analysis, based on our lab’s data on reproducibility of different measurements of LVEF, we prioritized noncontrast 3D if available, followed by the mean of apical 2 and 4 chamber measurements (with contrast when available.).
Echocardiograms, masked to temporal order, were then overread (SGE) and if a possible difference in LVEF≥4% was of concern, they were forwarded to Echo Core Lab staff (BA) and consensus reached. The mean±SD interobserver difference between observers SGE and BA was 3.6% ± 3.2%, Core laboratory reading for all echocardiograms was not feasible due to cost constraints.
-
1. Stress test images were reviewed in detail for ischemia, hibernation and scar (when available) as reported by imaging staff cardiologists by one of the authors (SGE), and reinterpretation by the imaging staff sought if necessary. We assessed LV scar either with PET scan or MRI. For the former, we defined scar as a matched perfusion defect on rest and post-regadenoson imaging. Using MRI, it was defined as delayed enhancement after injection of gadolinium chelate.
Angiographic analysis was performed by the Cleveland Clinic Core Angiographic Laboratory using CAAS-PIE II, version 8.1 (Pie Medical Imaging BV, Maastricht, The Netherlands) using both edge detection and video densitometric measurement. Territory revascularized was assessed by the Duke Jeopardy Score and considering published artery length perfusion territory correlations.
All of these tests were read without knowledge of the primary outcome result.
The Registry Database was further supplemented by EMR (supplemented by telephone contact, as necessary) review for: 1) estimated time of LV dysfunction, categorized by a single physician (SGE) for consistency, primarily using data from sequential imaging studies (which were common), supplemented by clinical notes and labs (if a reasonable estimate could not be made, this field was left empty); 2) heart failure medications and treatments (CRT, ICD) at the times of the pre and post-PCI echocardiograms, and 6 months after the second echocardiogram; 3) clinical follow-up as of June 1, 2024.
The statistical analysis plan was finalized prior to data evaluation. Sample size based upon need for a conservative ∼15 endpoints per candidate variable. Data are presented as mean ± SD or median and IQR, as appropriate. Statistical testing including multiple linear regression with bootstrapping analyses was performed using SPSS (IBM, version 28) and SYSTAT software (The System for Statistics, version 13.2. SPSS, Chicago, USA. Candidate variables are shown in Figure 2 , chosen on the basis of literature review and clinical experience. As LVEF recovery might be time dependent, time from PCI to follow-up echo was analyzed first, and then would be forced into the multiple regression analysis if found to correlate with change in LVEF. Medications were individually assessed for increase in dose between the time of pre and post-PCI and change in LVEF. In the primary analysis, outlier delta EF data >2SD were Winsorized to lessen their effect on modelling. Parsimonious polynomial regression was used to refit data when nonlinear relations were observed. Multicollinearity assessed using variance inflation factor, with removal of variables least well supported in the relation to LVEF improvement. Models were rerun with 1000 bootstraps and bootstrapped p values are presented. Left ventricular end-systolic volume (LVESV) was analyzed in a similar manner, as another measure of LV function. The correlation of change in LVEF with freedom from cardiac death or hospitalization for heart failure was assessed with Cox regression analysis, adjusting for age, CKD, diabetes, target vessel, sex and NTproBNP.
Candidate covariates of LVEF improvement.
Further secondary analyses were prespecified: (1) analysis limited to patients with baseline EF≤35%; (2) analysis stratifying patients by use or nonuse of invasive physiology; (3) analysis limited to patients without ≥ 30 mmHg difference in systolic BP between pre and post-PCI echocardiograms; (4) analysis of correlation between QCA and LVEF recovery for patients with single lesion PCI.
Results
The mean age of the treated cohort was 69 yrs, most were male, half had diabetes and estimated LV scar for those with available images to review was 6% ± 7% of the LV (see Table 1 ). Their baseline LVEF was 35% ± 11%. PCI was performed on 2.0 ± 1.1 lesions, most commonly in the LAD, and they subtended an estimated 42% of the LV myocardium (Jeopardy score 5.1 ± 2.2). LV mechanical support used was in 5.5% of PCIs.
Table 1
Baseline characteristics
| All patients ( n = 200) | |
|---|---|
| Age (yrs) | 69 ± 10 |
| CKD≥3 (%) | 21.0 |
| Diabetes (%) | 52.0 |
| Diameter stenosis (%) | 61 ± 19 |
| Hypertension (%) | 84.5 |
| Jeopardy Score | 5.1 ± 2.2 |
| LVEF, pre (%) | 35 ± 11 |
| LVESV (ml-3, pre) | 113 ± 57 |
| Male (%) | 75.5 |
| NTproBNP (median, IQR) | 2875 (904, 8069) |
| Post- PCI | |
| Jeopardy Score | 0 (0,2) |
| SYNTAX score | 0 (0,5) |
| Prior CABG (%) | 22.0 |
| Prior MI (%) | 51.5 |
| Race | |
| Caucasian (%) | 85.0 |
| Black (%) | 10.5 |
| Other (%) | 4.5 |
| Smoking, current (%) | 9.5 |
| Stress testing, images available (%) | 77.0 |
| PET | 50.5 |
| MIBI | 16.0 |
| DSE | 5.5 |
| MRI | 5.0 |
| Outside/images not available | 23.0 |
| For patients with images available | |
| Scar, any (%) | 48.0 |
| Scar, % LV | 6 ± 7 |
| Treatments (%) | |
| ACEI/ARB | 55.0 |
| Aldactone (%) | 20.5 |
| Beta blockers (%) | 83.5 |
| Sacubitril/valsartan (%) | 11.5 |
| CRT (%) | 5.5 |
|
ICD (%) need a horizontal line betwseen ICD and Vessel
Vessel(s) treated (%) |
11.0 |
| LMT | 9.5 |
| LAD | 70.0 |
| LCX | 28.0 |
| RCA | 24.5 |
For comparison, all patients with SICM and LVE F ≤ 50% treated 2009 to 2023 ( n = 4784) had mean age of 68 yrs, 73.7% were male, 50% were diabetic, mean LVEF was 38±10%, LAD was the target in 49.3% and 1.5 ± 1.1 lesions were treated.
Echocardiograms obtained were at a median 9 (IQR 3,32) days before, and 132 (IQR 92,212) days after PCI. Median follow-up was 35 months (3.5% of patients were lost to follow-up). Mean follow-up LVEF was 39% ± 11%, and 44% of patients had an increase in LVEF ≥5%. LVESV improved by 13.3 ± 40.8 mL 3. Heart failure medication use is shown in Table 2 .
Table 2
Heart failure medications (% alive patients)
| Pre-PCI | Post-PCI post | Late follow up | |
|---|---|---|---|
| ACE/ARB | |||
| High dose * | 15.0 | 11.3 | 11.6 |
| Low dose | 40.0 | 47.5 | 42.0 |
| Aldactone | 21.0 | 27.3 | 33.1 |
| Beta-blocker | |||
| High dose † | 32.5 | 32.5 | 37.7 |
| Low dose | 51.0 | 57.5 | 52.2 |
| Empagliflozin | 2.0 | 14.6 | 18.8 |
| Sacubitril/valsartan | |||
| High dose ‡ | 0.0 | 5.0 | 8.7 |
| Low dose | 11.5 | 8.7 | 8.7 |
| Total number of HF medications | 1.8 ± 1.0 | 2.1 ± 1.0 | 2.1 ± 1.0 |
Predictors of LVEF Recovery: Presence of an ICD, baseline LVEF, prior CABG, and target vessel-associated scar all significantly and negatively impacted improvement of LVEF (bootstrapped p <0.05) (see Tables 3 and 4 ) (as time from PCI to follow up echo was not correlated with change in LVEF, that initial analysis step was not required.) The results of a simplified score for recovery based on these 4 elements is shown in Figure 3 . Change in LVEF did not correlate with time from PCI to follow-up echo ( r = 0.08, p = 0.26). Improvement in LVESV correlated with initial LVEF<35% (p = 0.002), lack of prior bypass surgery (p = 0.014) and no ICD (p = 0.049).
Table 3
Correlates of improvement in LVEF (Univariate p values)
| All LVEF | Winsorized LVEF | |
|---|---|---|
| CKD≥3 | 0.58 | 0.44 |
| Diabetes | 0.48 | 0.60 |
| Estimated duration of LV hibernation | ||
| Clinical | 0.21 | 0.22 |
| From Pre-PCI echo to PCI | 0.22 | 0.25 |
| ICD already in place | 0.007 | 0.004 |
| Interim addition or dose escalation of HF medications | ||
| ARNI | 0.99 | 0.92 |
| ACE/ARB | 0.26 | 0.22 |
| Beta blocker | 0.41 | 0.35 |
| Prior CABG | 0.007 | 0.008 |
| Scar in territory to be revascularized | 0.068 | 0.045 |
| Severity of LV dysfunction | ||
| LVEF | 0.002 | 0.003 |
|
LVESV
Territory revascularized |
0.14 | 0.18 |
| Jeopardy score | 0.23 | 0.25 |
| LAD | 0.019 | 0.022 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
