Noninvasive Pressure-Volume Loops Predict Major Adverse Cardiac Events in Heart Failure With Reduced Ejection Fraction

Background Heart failure with reduced ejection fraction (HFrEF) is characterized by ventricular remodeling and impaired myocardial energetics. Left ventricular pressure-volume (PV) loop analysis can be performed noninvasively using cardiovascular magnetic resonance (CMR) imaging to assess cardiac thermodynamic efficiency. Objectives The aim of the study was to investigate whether noninvasive PV loop parameters, derived from CMR, could predict major adverse cardiac events (MACE) in HFrEF patients. Methods PV loop parameters (stroke work, ventricular efficiency, external power, contractility, and energy per ejected volume) were computed from CMR cine images and brachial blood pressure. The primary end point was MACE (cardiovascular death, heart failure (HF) hospitalization, myocardial infarction, revascularization, ventricular tachycardia/fibrillation, heart transplantation, or left ventricular assist device implantation within 5 years). Associations between PV loop parameters and MACE were evaluated using multivariable Cox regression. Results One hundred and sixty-four HFrEF patients (left ventricular ejection fraction ≤40%, age 63 [IQR: 55-70] years, 79% male) who underwent clinical CMR examination between 2004 and 2014 were included. Eighty-eight patients (54%) experienced at least one MACE after an average of 2.8 years. Unadjusted models demonstrated a significant association between MACE and all PV loop parameters (P < 0.05 for all), HF etiology (P < 0.001), left ventricular ejection fraction (P = 0.003), global longitudinal strain (P < 0.001), and N-terminal prohormone of brain natriuretic peptide level (P = 0.001). In the multivariable Cox regression analysis adjusted for age, sex, hypertension, diabetes, and HF etiology, ventricular efficiency was associated with MACE (HR: 1.04 (95% CI: 1.01-1.08) per-% decrease, P = 0.01). Conclusions Ventricular efficiency, derived from noninvasive PV loop analysis from standard CMR scans, is associated with MACE in patients with HFrEF.

Regardless of the underlying etiology, however, mitochondrial dysfunction and hence myocardial energetic deficit are nearuniversal findings in HFrEF, 1 indicating the central role of the myocardial energy metabolism in the progression of HF.The dramatic improvements in long-term morbidity and mortality seen in the wake of the modern paradigm for medical HF treatment largely arise from augmented myocardial energetics secondary to reduced cardiac loading. 2,3Unloading the heartimproving its chronic working conditions-is key to managing HFrEF.
While risk in HFrEF can be assessed using several established imaging measures of left ventricular (LV) function, including the load-dependent parameters left ventricular ejection fraction, 4 longitudinal strain, 5,6 and atrioventricular plane displacement, 7 the prognostic value of these methods likely reflects their intrinsic but imperfect relationship to how efficiently the heart functions as a pump.
Assessing cardiac thermodynamic efficiency 8,9 more directly through pressure-volume (PV) loop analysis 10,11 has potential added benefit in HF, as it provides unique windows into physical quantities such as ventricular stroke work (SW), efficiency, external power, and myocardial O 2 consumption.
While current HF guidelines acknowledge the need for improved assessment of patient-specific hemodynamic measures, 3 PV loop measurement has historically required invasive measurements of LV pressure, limiting applications.
3][14][15] This method enables the study of unique cardiac functional parameters without the risk, discomfort, and cost associated with invasive measurements.Moreover, as the method only requires time-resolved CMR short-axis images and a concurrent brachial blood pressure recording, it is well suited to retroactively examine long-term outcomes in previously studied HF patients.Whether noninvasive PV loops can add prognostic information in HFrEF is currently unknown.
Here we aimed to investigate whether noninvasively computed PV loops can predict major adverse cardiac events (MACE) in HFrEF patients.CMR IMAGING.All patients had undergone imaging at either 1.5-or 3.0T (both Philips Achieva).The imaging protocols included short-and long-axis (2ch, 3ch, and 4ch) cine images acquired using balanced steady-state free precession imaging, and late  CMR IMAGE ANALYSIS.Left ventricular endocardial borders were contoured in the short-axis stack over the cardiac cycle using Segment 4.0 R11026 (Medviso), as previously described. 14,16LGE images were visually evaluated for the presence and distribution of scar by the attending physician according to clinical routine and interpreted in light of the clinical data to categorize patients as either ischemic cardiomyopathy (ICM) or nonischemic dilated cardiomyopathy (NIDCM).
PV loops were computed using a previously published 12 and validated [13][14][15]17 model. In ummary, the model uses a digitized time-varying elastance function 16 to estimate dynamic LV pressures in relation to LV volumes from CMR images.The elastance model is scaled in time so that the minimal LV volume (endsystolic time frame) coincides with the middle of the downslope of the elastance function.The elastance curve is also scaled in amplitude to match the LV peak pressure (LVP systole , approximated from brachial blood pressure using a previously described expression) 18 and LV end-diastolic pressure (EDP).EDP is necessarily estimated by the user but within reasonable limits has little effect on the parameters used here; 12,15 we set the EDP to 7 mmHg for all data sets.
From the resulting PV loops, we readily obtained the following parameters for evaluation: SW (J), potential energy (PE) (J), ventricular efficiency (VE) (%), mean external power (J/s), contractility (mm Hg/mL), energy per ejected volume (J/mL), and arterial elastance (mm Hg/mL).Figure 1 shows how the different parameters are derived from a PV loop.Patient characteristics including medications are summarized in Table 1.
PV LOOP PARAMETERS. Figure 2 shows representative PV loops from three patients, one from each VE tertile.Table 1 shows PV parameters for all patients.
Comparison of PV parameters between NIDCM and ICM groups revealed no statistically significant differences (P > 0.05 for all).
CLINICAL OUTCOMES.Patients were followed for an average of 4.5 years.The primary outcome occurred in 88 individuals after an average of 2.8 years; the most frequently observed was all-cause HF hospitalization (n ¼ 49), followed by cardiovascular death (n ¼ 23), revascularization (n ¼ 8), AMI (n ¼ 6), cardiac arrest (n ¼ 1), and ventricular tachycardia (n ¼ 1).
UNADJUSTED AND ADJUSTED MODELS.Unadjusted Cox regression (Table 2) found significant associations with the primary end point for HF etiology, log N-terminal prohormone of brain natriuretic peptide, EF, atrioventricular plane displacement, global longitudinal strain, as well as SW, VE, external power, contractility, and energy per ejected volume.Figure 3 shows Kaplan-Meier plots for EF and VE.In stepwise multivariable Cox regression, the concordance improved significantly for each step from M1 to M3 P-V with the likelihood ratio test.In M3 P-V , only VE and etiology of HF remained significant, with a concordance of 0.66 (Table 3, Figure 4).

DISCUSSION
The study demonstrates that VE, derived from noninvasive PV loop analysis of a standard CMR scan at rest, is associated with MACE in heart failure  RELATION TO EARLIER WORK.The discovery of the end-systolic PV relationship (ESPVR) by Suga et al 22,23 was a key development for PV analysis, as the maximal value of the time-varying elastance of the ventricle (E max ) is a load-independent index of contractility.These findings were later verified in man by Grossman et al 24 who also demonstrated how V 0 can be measured from variable ventricular loading.
Sagawa et al 25  While our survival model provided good separation between low-and high-risk groups, the addition of stress imaging might further improve its accuracy.
4][35] Dobutamine stress has been shown to unmask energetic deficit across a spectrum of cardiac ailments, including HFrEF. 1,36,37Similarly, noninvasive PV loop analysis has been employed to study VE and contractility in HFrEF 17 as well as in AMI. 38These studies underscore the value of PV loop analysis in quantifying dynamic changes to ventricular thermodynamic efficiency resulting from either pharmacological intervention or pathophysiological processes.
LGE has been repeatedly shown to predict outcomes in heart failure, including adverse remodeling, morbidity, and mortality in NIDCM, 39,40 where an ischemic LGE pattern is associated with worse outcomes. 41Most studies on the prognostic value of LGE in DCM have excluded ICM patients. 39In our mate-  first-event analyses. 43The downside is that such statistical models are inherently more complex and may produce unwanted side effects when using composite end points such as MACE.As this was the first attempt at evaluating the predictive value of PV loop analysis, we opted for a simpler survival analysis.Our study sample consisted of HF patients undergoing treatment in the preneprilysin antagonist, pre-sodium/glucose cotransporter 2 inhibitor era.
Relatively low event rates indicate patients were generally well managed on therapy available at the time.Follow-up studies on more contemporary patient groups will likely see lower event rates than ours, which may be mitigated using multi-event survival analysis and/or a larger cohort.The predictive accuracy of PV loop analysis may also be affected in the presence of modern drugs such as sodium/glucose cotransporter 2 inhibitor, whose primary mechanism of action in HFrEF remains unclear. 44Regardless, we acknowledge that our study was limited in size, and results should be reproduced in a larger cohort before clinical implementation is considered.
Ongoing work using noninvasive PV loop analysis has shown myocardial infarction acutely depresses contractile function in relation to the extent of ischemia (myocardium at risk). 45Long-term adaptive remodeling in ICM involves ventricular dilatation and compensatory hypertrophy of viable myocardium.

H
eart failure with reduced ejection fraction (HFrEF) is defined by impaired systolic function and typically presents with ventricular dilatation, both heralds of myocardial energy waste.Untreated, HFrEF progresses in a 'downward spiral' of forward failure, organ hypoperfusion, reflexive renin-angiotensin-aldosterone system stimulation, fluid accumulation, increased preload, further decline in systolic function, and so on.This progressive decline reflects the body's inherent tendency to prioritize cardiac output and hence cerebral perfusion over long-term cardiac health.The clinical outcome for HFrEF patients depends largely upon whether the therapeutic strategy successfully targets the specific pathophysiological mechanisms causing forward failure in the individual patient.
METHODS STUDY SAMPLE.The study was approved by the Swedish Ethical Review Authority and the Swedish National Board of Health and Welfare and was conducted in accordance with the Declaration of Helsinki.This was a retrospective analysis of data acquired from patients with clinically diagnosed heart failure who underwent CMR examination at our center between 2004 and 2014.From the records, 287 potential patients with complete CMR data sets, a left

FIGURE 1
FIGURE 1 Left Ventricular PV Loop Computed From CMR Images and Brachial Blood Pressure gadolinium enhancement (LGE) images in the same views.Sequence-specific parameters are given in the Supplemental Appendix.A brachial blood pressure measurement was performed using a calibrated automatic cuff when resting in the supine position in conjunction with the CMR scan.
FOLLOW-UP.Follow-up data were obtained from the Cause of Death Registry and the Patient Registry (both Swedish National Board of Health and Welfare) and from manual review of electronic patient records for 5 years after CMR.The primary outcome was MACE within the 5-year follow-up period.MACE was defined as cardiovascular death (caused by acute myocardial infarction (AMI), heart failure, or ventricular arrhythmia), or all-cause HF hospitalization, AMI, revascularization (endovascular or coronary bypass graft), cardiac arrest, pulmonary edema, heart transplantation, sustained ventricular tachycardia or ventricular fibrillation, and LV assist device treatment.A complete list of the International Classification of Diseases diagnosis codes used to generate MACE is provided in the Supplemental Appendix.All events, including recurring events such as HF hospitalization were noted, and elapsed time between CMR examination and first MACE was used as input for survival analyses.Patients were censored if they reached the end of the study period (December 31, 2019) without an event, or at a time of noncardiovascular death.STATISTICAL ANALYSIS.Data distributions were assessed for normality using histograms.Continuous data were tested using Student's t-test or Mann-Whitney U test and are presented as mean AE SD or median (IQR) as appropriate.Categorical data were tested using the chi-square test.Statistical significance was assigned at the P < 0.05 level.Time to first event survival analyses were conducted in a stepwise manner using both a-priori based and data-driven variable selections.First, unadjusted Cox regressions were performed to highlight clinical and imaging parameters of interest.Second, we performed stepwise multivariable Cox regressions.Parameter multicollinearity was assessed using a Spearman correlation matrix, and only pairs of variables with a correlation coefficient <0.8 were allowed in the multivariable

FIGURE 2
FIGURE 2 Noninvasive Pressure-Volume Loops in Heart Failure introduced the ratio of systolic pressure to end-systolic volume (SP/ESV index, oftentimes using peak SP) as a simplified indicator of ventricular contractility, while later validation suggested its use for clinical studies of heart function.26Subsequent studies found SP/ESV, peak SP/ESV, or ESPVR predictive for long-term outcomes in various patient cohorts.[27][28][29][30]While we found ESPVR to be significantly associated with MACE in unadjusted Cox regression, other PV loop parameters demonstrated better Wald test and P values, and contractility was therefore not entered into the multivariable model.In summary and coherent with our findings, the inability of the failing ventricle to adapt from rest to stress conditions carries prognostic information.The main difference between these studies and ours is that we analyzed rest imaging only.8

FIGURE 4
FIGURE 4 Concordance and Added Value of Iterative Survival Models M1-M3 P-V for Predicting MACE

FIGURE 5
FIGURE 5 Effects of Estimated Vs Real V 0 on Ventricular Efficiency

TABLE 2
Univariate Cox Regression Model: Association Between Baseline Clinical and P < 0.05 are shown in bold.n,sample size per analysis for variables available in <100% of subjects.AVPD ¼ atrioventricular plane displacement; GFR ¼ glomerular filtration rate; LGE ¼ late gadolinium enhancement; NT-proBNP = N-terminal prohormone of brain natriuretic peptide.patientswith reduced ejection fraction.This is the first study investigating the prognostic value of CMRderived PV loop parameters in HFrEF patients.In this patient group, PV loop analysis may help improve risk stratification and hence clinical management at minimal expense and with no added risk or discomfort for the patient.CLINICAL IMPLICATIONS.The immediate availability of PV loop analysis from routine CMR images andFIGURE 3 Kaplan-Meier Plots of Event-Free Survival by Tertiles of Ejection Fraction and Ventricular Efficiency Shaded areas indicate 95% CIs for the survival function.VE ¼ ventricular efficiency.

TABLE 3
Multivariable Cox Proportional Hazards Models: Association Between Baseline Clinical and Cardiovascular Magnetic Resonance Characteristics and Major