LAUSR

Recent evidence implicates the endothelium as a central mediator in the development of heart failure with preserved ejection fraction (HFpEF).1 Co-morbidities such as hypertension, diabetes mellitus and obesity induce an inflammatory state and reduce nitric oxide (NO) bioavailability.2 This NO shortage modulates diastolic function and ventricular stiffness and reduces peripheral vasodilatory capacity—both physiological hallmarks of HFpEF.3 Reduced NO bioavailability can be clinically measured as endothelial dysfunction. In HFpEF patients, most studies agree brachial artery endothelial function is not different from controls, but microvascular function is reduced.4 Repair of deficient endothelium is possible through endothelial progenitor cells (EPC), circulating bone marrow-derived cells that secrete vascular growth factors and are recruited rapidly by injury or exercise.5 Angiogenic T cells (TA), a subpopulation of T lymphocytes with high angiogenic properties, have been shown to help proliferate EPC and mature endothelial cells in vitro.5 Exercise training is recommended to improve aerobic capacity and quality of life in HFpEF patients,6 but the mechanisms underlying these improvements are largely unknown. In patients with heart failure and reduced ejection fraction (HFrEF), known to have severe endothelial dysfunction, levels of circulating EPC are reduced and TA are dysfunctional. The latter could be corrected by a single exercise bout.7 Data in HFpEF are lacking. We hypothesized that HFpEF patients are characterized by microvascular endothelial dysfunction, coinciding with reduced numbers of circulating EPC and TA. Also, we postulated that a single maximal exercise bout recruits EPC and TA, in analogy to HFrEF patients. Power analysis based on previous literature8 revealed that 26 patients per group would detect a difference in endothelial function with 80% power. From 2014 to 2015, we recruited ambulatory and clinically stable HFpEF patients according to criteria for an ongoing HFpEF clinical trial.9 Inclusion criteria were (i) signs or symptoms of heart failure, New York Heart Association (NYHA) class II or III; (ii) left ventricular ejection fraction ≥50%; (iii) echocardiographic E/e’ ratio >15 or E/e’ 8–15 and plasma brain natriuretic peptide >80 pg/mL; (iv) structured exercise <2× 30 min per week. Additionally, we recruited ageand sex-matched healthy volunteers (HV). Volunteers were required to be sedentary, asymptomatic, free from cardiovascular disease and diabetes, and not taking cardiovascular drugs, to limit the confounding effect of co-morbidities. Cardiac structural or functional abnormalities were excluded by electrocardiogram and echocardiography. All participants provided written informed consent. This study abides to the Declaration of Helsinki and was approved by the ethics committee of the Antwerp University Hospital. Endothelial function was assessed by peripheral arterial tonometry (PAT) at the fingertip (EndoPAT, Itamar Medical, Caesarea, Israel). After 5 min of baseline measurement, a blood pressure cuff was inflated at the forearm during 5 min to 100 mmHg above systolic blood pressure, and subsequently released causing an endothelium-dependent reactive hyperaemia. Dedicated software calculated the PAT ratio and reactive hyperaemia index (RHI). A RHI below the median has been described to predict a worse prognosis in HFpEF patients.10 Exercise capacity was assessed by a symptom-limited maximal cardiopulmonary exercise test (CPET) using a ramp protocol of 20 W+10 W/min on a bicycle ergometer (Ergoline-Schiller).11 Endothelial function was reassessed after CPET. Blood was drawn before and 10 min after CPET, and flow cytometry for EPC and TA was performed on a FACSCanto II flow cytometer (BD Biosciences). EPC were defined as CD34+KDR+CD45dim cells, TA were defined as CD3+CD31+CD184+ cells. Analysis and gating strategy have been published previously.5 Baseline comparisons were performed using Welch two-sample t-test (continuous variables), Wilcoxon rank sum test (skewed continuous variables) and Pearson’s Chisquared test (categorical variables). Spearman coefficients (rho) were used for correlations. Linear mixed models were used to analyse repeated measures data, with Holm correction for post-hoc multiple comparisons. A multiple linear regression model was used to assess determinants of peak oxygen uptake (peak VO2), NYHA class and E/e’ adding all traditional cardiovascular risk factors as covariates and subsequently removing those not contributing to the analysis. A P-value of <0.05 was considered significant. All data were analysed using R v3.4.3. Healthy volunteers and HFpEF patients were well matched for age and gender (HV: 73± 6, HFpEF: 74± 7 years, P= 0.622; HV: 62%, HFpEF: 62% female, P=1.0) Several co-morbidities were more prevalent in HFpEF patients, including hypertension (85%), hyperlipidaemia (77%), obesity (46%) and diabetes (31%, all P< 0.05). HFpEF patients were characterized by diastolic dysfunction [E/e’ HV: 10.1 (9.3–11.7), HFpEF 16.5 (14.0–20.0), P< 0.001] and structural cardiac changes [left atrial volume HV: 22.9 (17.3–27.5), HFpEF 43.2 (33.5–49.4) mL/m2, P< 0.001]. Exercise capacity was reduced in HFpEF patients compared to HV [peak VO2: HV: 23.3 (21.6–29.0), HFpEF 17.5 (13.6–19.1) mL/kg/min, P< 0.001]. Baseline endothelial function was impaired in HFpEF patients: the PAT ratio was consistently lower in HFpEF patients (Figure 1A). RHI, adjusting PAT measurements for systemic effects and baseline variation, was significantly reduced (P= 0.036, Figure 1B). Likewise, the amount of circulating EPC and TA was significantly lower in HFpEF compared to HV (EPC P= 0.025; TA P= 0.047, Figure 1C and 1D). A single exercise bout decreased RHI in HV, while it did not aggravate the pre-existent endothelial dysfunction in HFpEF patients (Figure 1E and 1F). No exercise-induced changes were seen in EPC (Figure 1G), whereas circulating TA significantly increased with exercise in both groups (Figure 1H). RHI showed an inverse relation with E/e’: better endothelial function predicted better diastolic function (β =−7.53, adjusted R2 = 0.20, P= 0.001, multiple linear regression corrected for age). Endothelial function