Exercise Intolerance in Heart Failure With Preserved Ejection Fraction: Diagnosing and Ranking Its Causes Using Personalized O2 Pathway Analysis

Abstract

Background: Heart failure with preserved ejection fraction (HFpEF) is a common syndrome with a pressing shortage of therapies. Exercise intolerance is a cardinal symptom of HFpEF, yet its pathophysiology remains uncertain.

Methods: We investigated the mechanism of exercise intolerance in 134 patients referred for cardiopulmonary exercise testing: 79 with HFpEF and 55 controls. We performed cardiopulmonary exercise testing with invasive monitoring to measure hemodynamics, blood gases, and gas exchange during exercise. We used these measurements to quantify 6 steps of oxygen transport and utilization (the O2 pathway) in each patient with HFpEF, identifying the defective steps that impair each one's exercise capacity (peak Vo2). We then quantified the functional significance of each O2 pathway defect by calculating the improvement in exercise capacity a patient could expect from correcting the defect.

Results: Peak Vo2 was reduced by 34±2% (mean±SEM, P<0.001) in HFpEF compared with controls of similar age, sex, and body mass index. The vast majority (97%) of patients with HFpEF harbored defects at multiple steps of the O2 pathway, the identity and magnitude of which varied widely. Two of these steps, cardiac output and skeletal muscle O2 diffusion, were impaired relative to controls by an average of 27±3% and 36±2%, respectively (P<0.001 for both). Due to interactions between a given patient's defects, the predicted benefit of correcting any single one was often minor; on average, correcting a patient's cardiac output led to a 7±0.5% predicted improvement in exercise intolerance, whereas correcting a patient's muscle diffusion capacity led to a 27±1% improvement. At the individual level, the impact of any given O2 pathway defect on a patient's exercise capacity was strongly influenced by comorbid defects.

Conclusions: Systematic analysis of the O2 pathway in HFpEF showed that exercise capacity was undermined by multiple defects, including reductions in cardiac output and skeletal muscle diffusion capacity. An important source of disease heterogeneity stemmed from variation in each patient's personal profile of defects. Personalized O2 pathway analysis could identify patients most likely to benefit from treating a specific defect; however, the system properties of O2 transport favor treating multiple defects at once, as with exercise training.

Keywords: cardiac output; comorbidity; diagnosis; diffusion; microcirculation; systems biology; taxonomy.

Conflict of interest statement

Disclosures: The authors have no conflicts to disclose.

© 2017 American Heart Association, Inc.

Figures

Figure 1. The O2 Pathway

A schematic depicting the sequence of O2 transport and utilization steps from mouth to mitochondria. Abbreviations: PIO2 stands for partial pressure of inspired O2, Q cardiac output, DM skeletal muscle diffusion capacity for O2, DL lung diffusion capacity for O2, V̇A alveolar ventilation, Hb hemoglobin concentration, vmax mitochondrial respiration capacity.

Figure 2. Peripheral Oxygen Extraction in HFpEF

A, Antagonism between convective (cardiac) and diffusive delivery of O2. The black point represents a typical HFpEF patient from this study (median V̇O2), plotted at her CPET-derived values of peak exercise V̇O2, ΔAVO2, and cardiac output (Q). The black and red curves display calculated values of this patient's V̇O2 and ΔAVO2 respectively, over a range of Q values. “Complete O2 extraction” indicates when the calculated venous pO2 falls within 1mmHg of mitochondrial pO2. B, Observed HFpEF ΔAVO2 vs control predicted ΔAVO2. The red point represents mean ΔAVO2 and Q in this study's HFpEF patients. The black curve depicts the mean ΔAVO2 vs Q relationship in the control population. The open circle denotes the predicted control value of ΔAVO2 at the mean value of Q found in HFpEF. C, Published HFpEF ΔAVO2 vs control predicted ΔAVO2. Mean ΔAVO2 values from previously published HFpEF cohorts (blue points, annotated by first author) are plotted alongside the ΔAVO2 vs Q curve belonging to our control population. As in panel B, the red point denotes the mean HFpEF ΔAVO2 measured in this study. All ΔAVO2 values in panels B and C were scaled to the mean Hb level found in our controls. D, DM in controls vs HFpEF. Error bars depict 95% confidence intervals. E, Mean convective and diffusive components of O2 transport in HFpEF vs controls. The lines of O2 convection (curved) depict V̇O2 as a function of venous pO2 for fixed values of Q and arterial pO2. The lines of O2 diffusion (straight) depict V̇O2 as a function of venous pO2 for fixed values of DM (line slope) and arterial pO2. The intersection of these lines determines the V̇O2 actually achieved.

Figure 3. Simple vs compound mechanisms of exercise intolerance

A, Simulated HFpEF patients (y-axis), each with a single O2 pathway defect. The x-axis lays out each of the six distinct O2 pathway parameters. The color of each tile indicates the severity of the defect, which is expressed as a percentage of the reference value (Supplemental Material). O2 pathway parameters ≥ 80% of the reference value are considered normal and depicted as white tiles. Patients are sorted by the magnitude of their O2 pathway defect. B, HFpEF patients from this study and their CPET-derived O2 pathway defects. Patients are ordered by Q-defect. Abbreviations: Q stands for cardiac output, DM skeletal muscle diffusion capacity for O2, DL lung diffusion capacity for O2, V̇A alveolar ventilation, Hb hemoglobin concentration, vmax mitochondrial respiration capacity.

Figure 4. Causal influence of each O2 pathway parameter on exercise capacity

A, V̇O2 Deficit Recovery (VDR) coefficients for each O2 pathway step in HFpEF. Error bars depict 95% confidence intervals. B, Influence of O2 pathway background on VDR coefficients. The total height of each stack of bars reflects the VDR that results from correcting all associated parameters displayed under the stack. The height of each individual bar in the stack represents the VDR of one parameter after first correcting the parameters associated with the bars beneath it; this VDR value sits atop of each bar. Bars are color coded to a unique parameter(s) (Q = red, DM = blue, Hb+V̇A+DL = yellow). All VDR values reflect an average over all HFpEF patients. C, VDRQ vs Q-defect for each HFpEF patient. Circled patients, E and F, are analyzed in panels E and F. D, VDRDm vs DM-defect for each HFpEF patient. In panels C and D, Q and DM-defects are expressed as: 100% - ratio of the CPET-derived parameter to a reference value. E-F, Influence of O2 pathway background on the VDR coefficients for two individual patients with a similar Q-defect (circled in panel C). The meaning of these charts is identical to panel B, except that each one now characterizes an individual patient.

Figure 5. Subtyping HFpEF by Exercise Pathophysiology

A, Scatterplot of HFpEF patients by O2 pathway defects, Q versus DM. Illustrative taxonomy of HFpEF based on shared O2 pathway defects. B, Scatterplot of HFpEF patients by VDR coefficients (VDRQ vs VDRDm). Illustrative taxonomy of HFpEF based on shared susceptibility to O2 pathway therapy (Q- vs DM-therapy). “Combination therapy” represents a subgroup of patients for whom a meaningful increase in V̇O2 would require correcting multiple O2 pathway parameters.