Left Ventricular Pressure Ratio Predicts In-Hospital Outcomes in Hospitalized Heart Failure With Reduced Ejection Fraction

Background. Given the risk of hemodynamic compromise in heart failure with reduced ejection fraction (HFrEF) patients undergoing left heart catheterization (LHC), there is a need for a simple parameter that can predict clinical outcomes. We hypothesize that left ventricular pressure ratio (LVPR), calculated as left ventricle systolic/left ventricle end-diastolic pressure, is a strong predictor of hemodynamic collapse in these patients. Methods. Retrospective analysis of consecutive hospitalized HFrEF patients undergoing combined LHC and right heart catheterization (RHC) at a single institution from 2015-2017 was performed. LVPR was compared with standard RHC hemodynamic variables. The primary outcome was in-hospital escalation of therapy, defined as ≥40 mm Hg drop in systolic blood pressure (SBP), SBP ≤90 mm Hg for ≥15 minutes, start or escalation of vasoactive medications, cardiopulmonary resuscitation, or in-hospital death. Receiver-operating characteristic (ROC) analysis and Kaplan-Meier survival analysis were performed for prediction of the primary outcome. Results. A total of 176 patients were included in this study. ROC analysis determined an optimal cut-off value of ≤3.96, which correlated with an area under the curve (AUC) of 0.65 (sensitivity, 45.9%; specificity, 83.2%; correctly classified, 64.9%). AUC was similar to other variables obtained using RHC. In-hospital survival free of escalation of therapy was lower in the low LVPR group vs the high LVPR group (0% vs 33%, respectively; P<.01). Conclusion. LVPR is an easily measured index obtained during LHC that can risk stratify hospitalized patients with HFrEF at the time of LHC. 

J INVASIVE CARDIOL 2021;33(7):E507-E515. Epub 2021 June 16.

Key words: cardiac catheterization, heart failure, invasive hemodynamics

Patients with heart failure and reduced left ventricular ejection fraction (HFrEF) undergoing left heart catheterization (LHC) and percutaneous coronary intervention (PCI) are at high risk for hemodynamic compromise following the procedure.^1-3 Cardiac output has shown variable prognostic value in HFrEF patients, and cardiac power output has been shown to predict in-hospital mortality in patients with acute myocardial infarction and cardiogenic shock.^4-6 Both of these parameters require right heart catheterization (RHC), limiting the ability to rapidly make assessments of HFrEF patients undergoing LHC for coronary angiography and PCI. Although RHC is performed on select patients undergoing LHC, the proportion of patients undergoing this procedure concomitantly is approximately 7%.⁷ Given the risk of adverse events following LHC in patients with HFrEF, there is a need for an easily obtainable parameter that can predict hemodynamic instability and short-term clinical outcomes in patients with HFrEF.⁸

Left ventricular end-diastolic pressure (LVEDP) represents LV filling pressure, and is frequently measured during coronary angiography and intervention. Elevated LVEDP at the time of LHC for patients with ST-segment elevation myocardial infarction (STEMI) is associated with adverse cardiac outcomes.^9,10 The ratio of systolic blood pressure to LVEDP has similarly been shown to predict need for intra-aortic balloon pump and in-hospital mortality in STEMI patients.¹¹ We hypothesized that the index of left ventricular pressure ratio (LVPR), calculated as left ventricular systolic pressure [LVSP]/LVEDP determined at the time of LHC, would predict short-term hemodynamic stability and clinical outcomes in patients with HFrEF. 

This was a retrospective analysis of consecutive hospitalized patients undergoing combined LHC and RHC at a single institution from January 5, 2015 to September 28, 2017. Patients were included if they had a left ventricular ejection fraction ≤50% and excluded if they had prior congenital heart surgery, mechanical aortic valves, LV thrombus, or ventricular septal defect. Patients were also excluded if they had temporary or durable left ventricular assist devices (LVADs), extracorporeal membrane oxygenation, or heart transplantation prior to the procedure. The study was approved by the University of Chicago institutional review board. 

Baseline demographics, clinical characteristics, procedural indications, medication, and hospitalization data were obtained via chart review of the electronic medical record (Epic Systems). Procedural indications were grouped as worsening HF, preoperative evaluation, stable angina or positive stress test, non-STEMI, or STEMI. Hemodynamic tracings were recorded and stored using the Mac-Lab Hemodynamic Recording System (GE Healthcare), reviewed by one of two investigators (SA, MC), and confirmed by the principal investigator (JB). Hemodynamic measurements were recorded at end expiration. Angiographic findings and interventions were obtained from the procedure report and coronary angiogram. LV pressure measurements were performed using either an end-hole or pigtail catheter either before or after coronary angiography, at the discretion of the operator, and ventriculography was not routinely performed. LV pressure measurements were made prior to PCI in patients who underwent PCI. Hemodynamic parameters included in the study were as follows: LVPR, pulse pressure, shock index, cardiac power output, pulmonary artery pulsatility index, and cardiac output. Calculations are as follows: LVPR = peak LVSP/LVEDP; pulse pressure = systolic – diastolic blood pressure; shock index = heart rate/systolic blood pressure; cardiac power output = mean arterial pressure x cardiac output/451; pulmonary artery pulsatility index = (systolic – diastolic pulmonary artery pressure)/right atrial pressure. Cardiac output was calculated using conventional Fick methodology, using estimated oxygen consumption based on body weight.  

The primary outcome was in-hospital escalation of therapy, defined as any one of the following: (1) drop in systolic blood pressure ≥40 mm Hg; (2) systolic blood pressure ≤90 mm Hg for ≥15 minutes; (3) need for starting, or escalating to additional vasoactive medications; (4) cardiopulmonary resuscitation; or (5) in-hospital death. The secondary outcome was all-cause survival at 90 days.

Baseline characteristics, clinical characteristics, and physical characteristics were determined using descriptive statistics. Receiver-operating characteristic (ROC) analysis was used to determine test characteristics of LVPR for determining the primary outcome. The area under the curve (AUC) was determined using non-parametric analysis, and the optimal cut-off value was determined by selecting the cut-point with the highest percentage correctly classified. This was repeated for the other hemodynamic parameters included in the analysis (pulse pressure, shock index, cardiac power output, pulmonary artery pulsatility index, and cardiac output). Kaplan-Meier survival analyses were performed for both the primary and secondary outcomes with respect to LVPR using the cut-point determined as above. Univariate regression analysis was used to determine factors associated with the primary endpoint with a statistical significance declared at a P-value <.05; these factors include patient descriptive values, RHC hemodynamics, and LHC values. These factors were then incorporated into a multivariate analysis, comparing RHC values to LVPR. In the multivariate analysis, variables were excluded in a step-wise manner based on their P-values (absolute value of insignificances), ultimately yielding a set of predictive escalation factors. Because of the relatively small sample size and dependence of several of the calculated variables on each other, we performed two separate multivariate analyses, one involving clinical and RHC variables without LVPR, and another involving clinical variables and LVPR.  Statistical analysis was performed using statistical software StataSE, version 15.0 (StataCorp).

A total of 176 patients were identified for this study and met inclusion criteria. Of these patients, 2 did not have left ventricular pressure measured, leaving 174 patients for analysis, with LV pressure measured after coronary angiography in 122 patients (69.3%). There were no adverse events attributable to the LHC or RHC itself. Based on the primary outcome, ROC analysis determined an optimal cut-off value of LVPR ≤3.96, which correlated with an AUC of 0.65 (sensitivity, 45.9%; specificity, 83.2%; correctly classified, 64.9%) (Figure 1 and Table 1). LVPR had similar AUC to cardiac power output, cardiac output, shock index, pulse pressure, and pulmonary artery pulsatility index (Figure 1). 

Based on the cut-off of LVPR ≤3.96, a total of 51 patients had low LVPR and 123 patients had high LVPR. There were few significant differences in baseline demographics, clinical characteristics, or medication use between the 2 groups. Patients with low LVPR had higher body mass index and body surface area, higher percentage intubated, lower albumin, and lower LVEF, and were more likely to be on a P2Y₁₂ inhibitor at baseline (Table 2). Indications for LHC were mostly for worsening HF (66.7%), followed by preoperative evaluation (18.4%), stable angina or positive cardiac stress test (9.2%), non-STEMI (2.9%), and STEMI (2.9%), with no significant differences in LHC indication between the 2 groups (Table 3). At the time of LHC, 3.4% of patients were on intra-aortic balloon pump support, 16.1% required inotropes, and 8.0% required vasopressors, which were not significantly different between the 2 groups. LHC revealed predominantly non-obstructive coronary artery disease (CAD; 69.5%), with 6.3% of patients undergoing PCI during the initial catheterization, 4.0% with planned staged PCI, and 4.6% with planned coronary artery bypass graft surgery. Baseline hemodynamics revealed significantly higher LVEDP (31.3 ± 4.5 mm Hg vs 19.6 ± 8.0 mm Hg; P<.001), lower cardiac index (2.28 ± 0.63 L/min/m² vs 2.57 ± 0.83 L/min/m²; P=.03), and lower pulmonary artery pulsatility index (2.00 ± 1.07 vs 3.67 ± 4.90; P=.02) in the low LVPR group compared with the high LVPR group, respectively. There were no significant differences in cardiac output, pulse pressure, shock index, or cardiac power output between the 2 groups.

Kaplan-Meier analysis was performed to assess and compare the primary outcome between the low LVPR and high LVPR cohorts. Low LVPR was associated with a lower in-hospital survival free of escalation of therapy compared with high LVPR (0.0% vs 33.0%; 95% confidence interval [CI], 18.0-49.9; Chi²=12.74; P<.01) (Figure 2). In addition, low LVPR was associated with lower 90-day survival compared with high LVPR (74.5% [95% CI, 60.2-84.3] vs 87.8% [95% CI 80.6-92.5]; Chi²=4.38; P=.03) (Figure 3).

Univariate analysis highlighted several factors significantly correlated with the primary outcome (Table 4). Low albumin and hemoglobin, right ventricular end-diastolic pressure, pulmonary artery systolic pressure, pulmonary artery saturation, LVSP, aortic systolic and diastolic pressures, and low LVPR were associated with the primary outcome. In multivariate analysis, only low LVPR (hazard ratio, 1.73; 95% CI, 1.11-2.70; P=.01) and pulmonary artery diastolic pressures (HR, 1.02; 95% CI, 1.00-1.05; P=.01) were associated with the primary outcome. However, the 95% CI of the pulmonary artery diastolic pressure hazard includes 1, and is therefore not as strong of an association as that of low LVPR.  

In this retrospective analysis of 176 consecutive patients with HFrEF undergoing simultaneous LHC and RHC, we found that low LVPR (≤3.96) was a significant predictor of in-hospital escalation of therapy. Furthermore, we demonstrated that the predictive ability of LVPR was better than more conventional measurements, including cardiac power output, Fick cardiac output, shock index, and pulse pressure, all of which, except for pulse pressure, require an additional RHC. Low LVPR remained an important predictor of in-hospital escalation of therapy after multivariate analysis, and was also found to predict 90-day all-cause mortality. 

Recent studies have identified reduced systolic blood pressure, elevated heart rate, increased LVEDP, and low systolic blood pressure/LVEDP ratio as hemodynamic parameters associated with early mortality in patients undergoing PCI. However, these were all measured in patients undergoing acute coronary syndromes or who were in cardiogenic shock.^9,10,12-16 This study emphasizes the predictive ability of LVPR in a larger patient population of those with HFrEF undergoing coronary angiography and possible intervention. The majority of patients in this study (66.7%) underwent simultaneous LHC and RHC for worsening HF. Notably, patients in our study had severe, complex CAD, as evidenced by the low rate of PCI (6.3%) relative to the burden of obstructive CAD (30.5%), which was due to the high rate of ≥2 vessel CAD (17.0%), planned staged PCI procedures (4.0%), referral for coronary artery bypass graft surgery (4.6%), and presence of coronary arteries that had chronic total occlusions (8.5%). Furthermore, unlike prior studies evaluating patients with acute coronary syndromes or in cardiogenic shock, the majority of patients were hemodynamically stable without need for vasoactive medications or intra-aortic balloon pump at the time of cardiac catheterization. Only 5.6% of patients underwent cardiac catheterization for a STEMI or non-STEMI, and just 16.4% were on inotropes, 8.0% on pressors, and 3.4% on intra-aortic balloon pump support at the time of cardiac catheterization. 

Patients with HFrEF are at high risk for complications following PCI, and prior studies evaluating whether these patients, either hemodynamically stable or unstable, would benefit from mechanical circulatory support have had mixed results.^1-3,17-21 Whether hemodynamics can be used to determine which patients benefit from early initiation of mechanical circulatory support is yet to be determined. However, more research is needed to help stratify risk and inform decision making on the choice of mechanical circulatory support, if at all, in patients presenting with cardiogenic shock or for high-risk PCI.³ Sola et al previously demonstrated that LVPR is the superior hemodynamic variable in predicting adverse short-term clinical outcomes in the STEMI population.¹¹ Our study expands on these data by demonstrating that LVPR is the most predictive variable obtained without RHC in patients with left ventricular ejection fraction ≤50%, and a high rate of complex, severe CAD, referred for cardiac catheterization. Prospective studies stratifying patients by LVPR may help to determine appropriate and beneficial use of mechanical circulatory support in this high-risk population. In those HFrEF patients who do not need PCI, either due to non-obstructive CAD or in those awaiting definitive revascularization, LVPR may help identify a high-risk group who may need closer observation. 

Study limitations. First, this is a retrospective study conducted at a single center in patients undergoing coronary angiography and results will need to be confirmed in a larger population. However, our population reflects a common group of patients with HFrEF who were referred for cardiac catheterization for various reasons. Second, similar to other invasive studies, timing of LVPR measurement was not standardized. Finally, although this seems to be a stable population based on low percentage with acute coronary symptoms, inotropes, and mechanical circulatory support, we observed a very high in-hospital event rate. This may be in part due to the high rate of severe, multivessel CAD, as evidenced by the 6.3% PCI rate despite 30.5% noted to have obstructive CAD. The high rate of escalation of therapy highlights the need to stratify these patients into risk using tools such as LVPR.  

LVPR is an easily measured index obtained during LHC. It can risk-stratify hospitalized patients with systolic HF, which can be useful in identifying a high-risk cohort of patients undergoing cardiac catheterization. This measurement may be useful in future studies determining the need for mechanical circulatory support in patients with HFrEF undergoing PCI, or closer monitoring for patients with HFrEF who do not require PCI.