Value of exercise echocardiography in heart failure with preserved ejection fraction: a substudy from the KaRen study

Abstract

Background

KaRen is a multicentre study designed to characterize and follow patients with heart failure and preserved ejection fraction (HFpEF). In a subgroup of patients with clinical signs of congestion but left ventricular ejection fraction (LVEF) >45%, we sought to describe and analyse the potential prognostic value of echocardiographic parameters recorded not only at rest but also during a submaximal exercise stress echocardiography. Exercise-induced changes in echo parameters might improve our ability to characterize HFpEF patients.

Method and results

Patients were prospectively recruited in a single tertiary centre following an acute HF episode with NT-pro-BNP >300 pg/mL (BNP > 100 pg/mL) and LVEF > 45% and reassessed by exercise echo-Doppler after 4–8 weeks of dedicated treatment. Image acquisitions were standardized, and analysis made at end of follow-up blinded to patients' clinical status and outcome. In total, 60 patients having standardized echocardiographic acquisitions were included in the analysis. Twenty-six patients (43%) died or were hospitalized for HF (primary outcome). The mean ± SD workload was 45 ± 14 watts (W). Mean ± SD resting LVEF and LV global longitudinal strain was 57.6 ± 9.5% and −14.5 ± 4.2%, respectively. Mean ± SD resting E/e′ was 11.3 ± 4.7 and 13.1 ± 5.3 in those patients who did not and those who did experience the primary outcome, respectively (P = 0.03). Tricuspid regurgitation (TR) peak velocity during exercise were 3.3 ± 0.5 and 3.7 ± 0.5 m/s (P = 0.01). Exercise TR was independently associated with HF-hospitalization or death after adjustment on baseline clinical and biological characteristics.

Conclusion

Exercise echocardiography may contribute to identify HFpEF patients and especially high-risk ones. Our study suggested a prognostic value of TR recorded during an exercise. That was demonstrated independently of the value of resting E/e′.

Heart failure with preserved ejection fraction, Echocardiography, Right ventricle, Strain

Introduction

Heart failure (HF) with preserved ejection fraction (HFpEF) is a complex pathophysiological entity. Echocardiographic parameters offer a key tool for syndrome diagnosis as indicated in the new ESC guidelines.¹ HFpEF is defined as an association of typical HF signs and symptoms, normal or preserved left ventricular ejection fraction (LVEF) and normal or small LV volumes, pertinent structural heart disease [LV hypertrophy/left atrial (LA) enlargement], and evidence of diastolic dysfunction.¹ Only a few papers have proposed exercise echocardiography as a relevant diagnostic tool in HfpEF.^2–5 The relevance of echocardiographic parameters that could be recorded during an exercise remains an issue particularly pregnant in this complex HFpEF syndrome. A strong correlation between E/e′ and physical activity has been demonstrated in a large series of patients including patients with HFpEF.⁶ We have previously reported that longitudinal systolic and diastolic LV as well as right ventricular (RV) functions assessed during a submaximal exercise stress echocardiography can distinguish symptomatic HFPEF patients from matched normal controls.⁷ In the present study, we sought to evaluate the value of submaximal exercise-echocardiographic parameters that are usually recorded at rest and that one can record during an exercise. The exercise echocardiography was systematically performed in stable state in the weeks following an acute HF episode to characterize and potentially predict the long-term clinical outcome in patients with HFpEF.

Methods

The design of the KaRen study has been published elsewhere.⁸ KaRen was designed to enrol patients showing HF symptoms who attended the emergency ward, and to follow them up 4–8 weeks after the acute episode. The inclusion criteria were as follows: (i) acute presentation at hospital admission with clinical HF signs and symptoms, according to the Framingham criteria; (ii) BNP > 100 pg/mL or NT-pro-BNP > 300 pg/mL; (3) LVEF > 45% by echocardiography within the first 72 h.

In this exercise, echocardiography substudy, measurements were carried out in a subgroup of patients after 4–8 weeks following an acute HF exacerbation. All patients enrolled in Rennes were invited to participate in the submaximal exercise echo study. After ensuring, they were haemodynamically stable (no functional or clinical signs of acute HF decompensation and no argument for any symptomatic coronary artery disease) and they had no neurological or orthopaedic limitation, patients who agreed underwent a semi-supine exercise test. Treatments including beta-blockers were not modified for the test. All patients had to be in sinus rhythm at the time of the exercise stress echocardiography but patients could have been identified as paroxysmal atrial fibrillation patients.

The echo-protocol was always the same, all the exams being performed by the same investigator (E.D.). A particular attention was brought to get the Doppler recording of the tricuspid regurgitation (TR) at rest and during exercise. All the parasternal and apical views were used to get this maximal velocity as a first thing at each step of the standardized echocardiographic protocol we were and we are using after our initial experience.⁷ All the measurements were performed according to the guidelines, offline by a dedicated physician working at the echo core Lab (CIC-IT 804, Rennes, France). This analysis was done afterwards, blinded from any clinical consideration. The results of the exercise echocardiography were not provided to the clinician.

Standardized submaximal exercise testing

After a clinical examination, arterial blood pressure (BP) measurement (Dinamap Procare Auscultatory 100), 12-lead electrocardiogram (ECG), and resting transthoracic echocardiography (Vivid 7, General Electric Healthcare, Horten, Norway), patients underwent a standard supine exercise (slightly on the left side) echocardiography on a tilting table using an electromagnetic cycle ergometer (Ergometrics). Exercise testing started at an initial workload of 30 W, with increase to 45 (if the exercise capacity was too weak) and 60 and 90 W every 3 min according to individual patient's capabilities. The pedalling rate was of 60 rpm. The ECG was recorded continuously, and BP was measured every 2 min during both exercise and recovery. BP, ECG, and echocardiographic images were acquired at rest and at a predefined maximal heart rate range (HR, 100–120/min), with at least three beats recorded. According to current practice, the physician was standing closed to the patient performing the images, assessing the clinical, ECG, and BP adaptation of the patient to the exercise. Thus, exercise testing could have been interrupted promptly in the event of typical chest pain, constraining breathlessness, dizziness, muscular exhaustion, drop in BPs or severe hypertension (systolic BP ≥ 250 mmHg), or significant ventricular arrhythmia. The test was considered abnormal if the patient presented one or more of the following criteria: angina, evidence of shortness of breath at a low workload level (<50 W), dizziness, syncope, or near-syncope, ≥2 mm ST segment depression compared with baseline, rise in systolic blood during exercise <20 mmHg, fall in systolic BP during exercise, or complex ventricular arrhythmias. Exercise duration was designed to be 8–10 min for every patient. Enough time was set to record a complete echocardiographic evaluation at each step.

Two-dimensional and tissue Doppler echocardiography

All patients underwent detailed echocardiographic examination at rest and at the maximal workload sustained during exercise using a Vingmed Vivid™ 7 or e9 (GE Healthcare, Horten, Norway). The position of the patient was constant from baseline to the end on the tilt table and always the same inclination. LV end-systolic and end-diastolic volumes, as well as LVEF, were measured using the modified biplane Simpson's method from the apical four- and two-chamber views. Left and right atrial volumes were calculated using the biplane area–length method from the apical four- and two-chamber views and indexed to the body surface area.⁹ The early filling (E) and atrial (A) peak velocities, as well as the deceleration time of early filling and isovolumic relaxation time, were measured from transmitral flow. All measurements were averaged over three beats (3–5 according to the homogeneity of the results).

Peak mitral annular myocardial velocity of the LV septal and lateral walls were recorded (and averaged) using the real-time pulse-wave tissue Doppler method, which allowed for measuring the mean peak systolic (s′), early diastolic (e′), and late diastolic (a′) velocities.¹⁰ LV filling pressure was calculated as the ratio of early mitral diastolic inflow velocity to early diastolic mitral annular velocity (averaged from the septal and the lateral side of the mitral annulus) (E/e′).¹⁰ Peak annular RV free-wall velocities (RV s′ and RV e′ for peak systolic and early diastolic velocities, respectively) were measured using the same method. Tricuspid annular peak systolic excursion was calculated using M-mode echocardiography.¹¹ Peak systolic pulmonary arterial pressure (PAP) was estimated using the Bernouilli formula according to the tricuspid maximal jet velocity. TR maximal velocity was used as a surrogate marker of PAP according to recommendation.¹² That was predefined because the assessment of right atrial pressure could be challenging at rest but was supposed even more questionable during the exercise.

Speckle tracking

LV longitudinal strains were assessed using the speckle tracking method.¹³ The apical four-, two-, and three-chamber images were analysed offline by tracing the endocardium in end-diastole, and the thickness of the region of interest was adjusted so as to include the entire myocardium. The software automatically tracked myocardial deformation on the subsequent frame, and the results were displayed graphically. End-systolic peaks were automatically considered for the measurement of global longitudinal strain (GLS) and maximal peaks were manually tracked for the look for mechanical dyssynchrony. The intraobserver and interobserver variabilities as well as repeatability were previously reported.⁷ The GE healthcare EchoPAC BT 12 was used.

Follow-up

After the ‘4–8 weeks scheduled visit’, patients were prospectively followed via phone call or by means of correspondence with their physicians every 6 months for at least 18 months. The follow-up was closed for all patients on 31 October 2012. A dedicated research team blinded to the initial visit and 4–8 week visit (registry unit from the French Society of Cardiology)⁸ documented hospitalizations and cause of hospitalization, death, as well as causes of death.

Definition of cardiovascular events and primary clinical end-point

The primary clinical endpoint was time to first heart failure hospitalisation or all-cause death. Heart failure had to be defined as primary diagnosis in the patient file for adjudicating hospitalisation as HF related hospitalisation.

Statistical analysis

Continuous variables were reported using central tendency and dispersion measurements. Qualitative variables were expressed as frequency and percentage. The primary outcome was defined as either death or readmission for HF whichever came first (censoring applied at the date of hospitalization for patients for hospitalized for HF and who subsequently died). We used the Cox proportional hazard model to analyse ‘time to event’ data. The assumption that covariates exhibited a linear form was checked using the shape of parameter estimate plots by mid-point quartile intervals, and the pattern of martingale residual plots as a function of the corresponding covariate. Departure from the proportional hazards assumption was investigated using time-dependent explanatory variables, in addition to a plot of the scaled Schoenfeld residuals as a function of time.

To develop a prediction model and to test whether some exercise parameters added some incremental information to parameters measured at rest, we started with selected echo parameters measured at rest (those associated with the outcome in univariate Cox regression analysis at a P < 0.10), and then added selected exercise parameters (those associated with the outcome in univariate Cox regression analysis at a P < 0.10 and not highly correlated with parameters measured at rest, Spearman coefficient <0.7). A stepwise backward selection retained parameters associated with outcome at a P-value of <0.05. We then adjusted those selected echo parameters on clinical and basic laboratory data (those associated with the outcome in univariate Cox regression analysis at a P-value of <0.10 among age, gender, medical history, drug use, creatinine, and NT-pro-BNP levels).

Considering the limited available data and that some echo parameters had some missing data, we did multiple imputations using the Monte Carlo Markov Chain method.

All analyses used procedures available in SAS software, version 9.3 (SAS Institute, Cary, NC, USA).

Ethics for the substudy

A specific authorization has been obtained for the KaRen exercise-echocardiographic substudy (authorization 0820-679). A specific informed consent has been signed by the included patients.

Results

From December 2008 to January 2012, 60 of the 203 patients included at the Rennes University Hospital in the KaRen registry were enrolled in the substudy. The reason for not participating in the substudy were significant orthopaedic or neurologic limitation (n = 45), suboptimal quality of resting echocardiography (missing data) (n = 33) and refusal to participate (n = 65). There were some minor differences in baseline clinical characteristics between patients participating and patients non-participating in the substudy (that have been reported elsewhere¹⁴) with a higher proportion of males (63.3%, P = 0.0015) and a higher resting SBP (138 ± 23 mmHg, P = 0.01) in the substudy population. Mean age was 74.8 ± 7.4 years and 28 patients (46.7%) had history of atrial fibrillation or flutter. The workload sustained during exercise was 45 ± 14 W (30–90 watts). There was no broad QRS. The exercise time ranged between 6 and 12 min (Table 1). There was strictly no argument for any coronary artery disease, with no chest pain, no segmental wall motion abnormality observed during the exercise. No ECG significant change was, as well, observed. The amount of exercise was limited in these elderly patients with HFpEF. Therefore, the changes observed during the exercise were limited but considered significant because they appeared after a very short and limited exercise.

Table 1

Main baseline clinical characteristics of patients included in the ‘KaRen exercise stress echocardiography’ substudy; Comparison with the whole cohort enrolled in Rennes-KaRen centre

Label	Substudy (N = 60)	KaRen in Rennes (N = 203)	P-value
Age (years)	74.8 ± 7.4	76.8 ± 9.6	0.1782
Female gender, n (%)	20 (36.4)	76 (50.7)	0.0690
NYHA I/II/III/IV, n	3/47/5/0	11/87/35/8	0.0067
Weight (kg)	73.1 ± 17.8	78.6 ± 19.8	0.0739
BMI (kg/m²)	27.8 ± 5.9	29.3 ± 6.4	0.1325
Overweight/obese, n (%)	15 (27.8)/20 (37.0)	46 (31.5)/61 (41.8)	0.5029
Diagnosis of HF prior to enrolment, n (%)	24 (43.6)	58 (38.9)	0.5426
History of coronary artery disease, n (%)	22 (36.6)	38 (25.5)	0.7982
History of AF/flutter, n (%)	28 (46.7)	96 (64.4)	0.1279
Arterial hypertension, n (%)	51 (85)	119 (79.9)	0.9831
Type 2 diabetes, n (%)	13 (23.6)	33 (22.1)	0.8214
NT pro-BNP (ng/L), median (p25–p75)	2067 [1419–4969]	257 [1313–5216]	0.7211
Haemoglobin (g/L)	122.4 ± 14.9	119.5 ± 18.6	0.2544
Creatinine (µmol/L)	102.5 ± 50.1	110.7 ± 58.5	0.3544

BMI, body mass index; HF, heart failure.

During follow-up (median time of 523 days), 7 patients died and 21 were hospitalized for decompensated HF. In total, 26 patients (43%) had a primary outcome event when compared with 49% in the whole KaRen cohort.¹⁵

Echocardiographic predictors

Table 2 displays demographic data in each group (those with vs. those without events) and Table 3 the mean values of echo parameters in each group as well as univariate P-values for Cox regression models. E/e′ at rest was associated with the primary endpoint and remained the only resting echo parameter significantly associated in the multivariable model (Table 4, Figures 1 and 2).

Table 2

Clinical parameters and medical history according to heart failure or death occurrence during follow-up and univariate measure of association

Parameters	Patients without event at the end of follow-up (N = 34)	Patients hospitalized for HF or who died (N = 26)	P-value
Gender, female, n (%)	14 (41.2)	8 (30.8)	0.3099
Age, years	73.8 ± 7.5	76.2 ± 7.2	0.3263
BMI, n (%)
Normal	7 (21.2)	5 (19.2)	0.3836
Overweight	12 (36.4)	7 (26.9)
Obese	14 (42.4)	14 (53.8)
Prior diagnosis of HF, n (%)	13 (38.2)	10 (38.5)
History of CAD, n (%)	13 (38.2)	9 (34.6)	0.9433
History of AF/flutter, n (%)	14 (41.2)	14 (53.8)	0.2680
Arterial hypertension, n (%)	26 (76.5)	25 (96.1)	0.0657
Baseline medication, n (%)
Statin	23 (67.6)	22 (84.6)	0.2726
Beta-blocker	27 (79.4)	17 (65.4)	0.2188
ACE	13 (38.2)	16 (61.5)	0.0795
NT-pro-BNP (ng/L)	2107 [353–24 975]	1994 [305–9352]	0.9989
Creatinine, µmol/L
At baseline	99 ± 41	114 ± 49	0.0792
At 4–8 weeks	99 ± 35	121 ± 48	0.0038

Values for continuous parameters are mean ± SD or median [range]; P-values are from univariate Cox regression Wald test.

Table 3

Echocardiography parameters (mean ± SD): description and univariate association with the risk of heart failure or death

Parameters	Patients without event at the end of follow-up (N = 34)	Patients hospitalized for HF or who died (N = 26)	P-value (raw data)	Missing (%)	P-value (imputed data)
Geometry
SBP	143 ± 24	157 ± 24	0.0169	<5
RR	906 ± 165	1019 ± 200	0.0217	<5
IVS	11.8 ± 2.4	11.7 ± 2.7	0.7884
LVED diameter	50.2 ± 6.5	52.7 ± 5.9	0.1640
LVES diameter	35.6 ± 7.3	37.6 ± 7.0	0.6030
Systolic function
SV	64.9 ± 23.6	72.3 ± 24.1	0.0575
LVEF (%)	58.5 ± 9.5	56.2 ± 9.5	0.4822
s′	7.09 ± 1.46	6.50 ± 1.52	0.1618
GLS	14.7 ± 4.0	14.2 ± 4.4	0.5193
Septal LS (%)	13.4 ± 4.7	13.9 ± 5.1	0.2841
Lateral LS (%)	14.4 ± 6.0	16.0 ± 7.0	0.4386
Diastolic function
LAVI (mL/m2)	45.5 ± 15.7	50.7 ± 19.6	0.1080	<2
E-dt (ms)	213 ± 92	202 ± 70	0.5833	<2
E/e′	11.3 ± 4.7	13.1 ± 5.3	0.0329	<5
e′	8.5 ± 2.8	7.4 ± 2.6	0.3727
Asynchronism
MIT/RR	0.57 ± 0.21	0.50 ± 0.10	0.1173	6.66	0.2064
LVPEI (ms)	85.1 ± 21.3	83.8 ± 25.7	0.9468
Septo-lateral delay DTI (ms)	38.8 ± 53.1	45.8 ± 52.8	0.8701
Delay IV (ms)	−11.1 ± 23.5	−9.5 ± 16.5	0.6101	13.3	0.3248
Right ventricle
RVPEI (ms)	95.7 ± 21.8	89.6 ± 21.4	0.2257	13.3	0.2413
RAVI (mL/m2)	32.1 ± 13.4	40.5 ± 19.3	0.0249	8.33	0.0760
TR Vmax (cm/s)	2.74 ± 0.47	2.99 ± 0.72	0.0892	33.3	0.0910
RV Sa (cm/s)	11.7 ± 2.8	11.6 ± 3.3	0.8045	10.0	0.7740
TAPSE (mm)	19.4 ± 4.5	20.1 ± 6.7	0.8019	6.66	0.7366
RV Ea (cm/s)	10.6 ± 4.0	10.6 ± 3.3	0.6363	10.0	0.6098
Exercise
LVES diameter	35.3 ± 6.4	36.3 ± 8.2	0.4894	11.7	0.4285
SV	62.2 ± 23.0	66.9 ± 23.1	0.2057
s′	7.07 ± 2.06	6.70 ± 2.7	0.1883	<2
E/e′	12.6 ± 6.4	18.3 ± 14.0	0.0413	5	0.0843
e′	10.6 ± 0.3.8	8.4 ± 3.5	0.0867
GLS	15.6 ± 4.0	15.9 ± 4.2	0.9380
RV Sa (cm/s)	13.6 ± 3.8	12.6 ± 4.1	0.2052	13.3	0.3086
TR Vmax (cm/s)	3.35 ± 0.47	3.72 ± 0.53	0.0097	13.3	0.0637
SBP	166 ± 30	178 ± 29	0.1306	8.33	0.5662
Work load (W)	45.7 ± 13.6	43.8 ± 13.4	0.5921
Reserve
LVEF	−0.35 ± 7.47	1.65 ± 8.13	0.3028
e′	−1.87 ± 2.41	−0.78 ± 2.91	0.1788
E/e′	1.72 ± 4.77	3.40 ± 12.9	0.4971	8.33	0.4720
s′	0.04 ± 2.07	0.05 ± 1.80	0.4603	<2
2DS	0.82 ± 2.75	1.53 ± 3.16	0.3850

LS, longitudinal strain; RV, right ventricular; TR, tricuspid regurgitation; LA, left atrial; E-dt, mitral inflow E-wave deceleration time; RA, right atrial; Vol, volume; RVPEI, right ventricular pre-ejection interval; LVPEI, left ventricular pre-ejection interval; IV, interventricular; GLS, global longitudinal strain; MIT, mitral inflow duration; RR, cycle length; e′, early diastolic pulsed tissue Doppler peak velocity; s′, systolic pulsed tissue Doppler peak velocity.

Reserve = difference between the measurement made during exercise and the one performed at rest (in other words: reserve = delta exercise − rest value).

Table 4

Multivariable estimates for the risk of death or hospitalization for heart failure related on echocardiography parameters

	Model 1 (full model)	Model 2 (final model)	Model 3 (full model)	Model 4 (final model)	Model 5 (final model)
	P-value (imputed data)	P-value (raw data)	P-value (imputed data)	P-value (imputed data)	P-value (raw data)	HR (95% CI)a
Resting echo
SBP, mmHg	0.0196	0.0078	0.1517
RR	0.3161
SV	0.8708
E/e′	0.0596	0.0153	0.0128	0.0020	0.0086	1.76 (1.15–2.68)
RAVI (mL/m2)	0.1315
TR Vmax (cm/s)	0.2406
Exercise echo
E/e′			0.6618
e′			0.5642a
TR Vmax (cm/s)			0.0056	0.0016	0.0015	2.07 (1.32–3.25)

Model 1 included all resting echo parameters and Model 2 retained only statistically significant parameters at 0.05 level; Model 3 included those selected parameters plus exercise echo parameters (as E/e′ and e′ were highly correlated, they were not entered simultaneously); Model 4 retained only statistically significant parameters at 0.05 level.

^aHR for 5 units increase of E/e′ and for 0.5 cm/s increase of TR V_max; adjusted on clinical parameters (history of hypertension, creatinine level and ACE inhibitor use/Model 5) did not substantial affect estimates.

Figure 1

Kaplan–Meier curve for (A) E/e′ ratio measured at rest and (B) TR maximal velocity recorded during the exercise.

Figure 2

Graphical presentation of (A) LVEF measured at rest and during standardized exercise stress echocardiography, (B) GLS measured under the same conditions, (C) E/e′ ratio and (D) TR maximal velocity (Tric Regurg).

TR maximal velocity was the only parameters recorded during exercise that remained significantly associated with the primary endpoint in the multivariable model (Table 4). Re-running the final model on raw data (without imputation of missing values) and adjusting on clinical (history of hypertension and ACE inhibitor use) and biological (creatinine level) parameters showed very similar estimates (Table 4).

Discussion

E/e′ at rest and estimated PAP by TR maximal velocity measured during standardized exercise has a predictive value in our HFpEF population. These two parameters may help to better define the prognosis of HFpEF individuals. They seem crucial for best characterizing HFpEF patients.

E/e′

E/e′ was a key parameter proposed in the 2007 diagnostic algorithm,^10,16 while LVEF and structural heart disease were introduced in the 2012 ESC guidelines.¹ With regard to supposed E/e′ robustness for estimating left heart filling pressures, studies in elderly populations or in dilated or hypertrophic cardiomyopathy patients have highlighted the necessary prudency in the use of this ratio for estimating filling pressures.^17–19 The E/e′ value for estimating filling pressure during exercise also has been challenged.²⁰ However, studies have found a relationship between exercise E/e′, exercise capacity, and invasive left ventricular end-diastolic pressure recorded during exercise.⁶ To date, E/e′ measured at rest was not shown to be the best parameter associated with HFpEF prognosis,^21,22 whereas the change in E/e′ from rest to exercise was reported to be correlated with prognosis in one study.²³ In this study, 197 patients with Type 2 diabetes mellitus were followed for 57 months, and the incidence death or hospitalization for heart failure was 9.1%.

Of note is that E/e′ can be measured during or after exercise. E/e′ > 14.5 was shown to be an independent predictor of outcome in a study involving 522 unselected patients referred for exercise echocardiography.²⁴ Yet, this study was not focused on heart failure and HFpEF patients. Our study, which used a prospective design and was part of a large prospective registry, clearly demonstrated the prognostic value of E/e′ recorded before any exercise.

Estimated pulmonary artery pressure using echocardiography

In addition to E/e′, we showed an independent prognostic value of TR maximal velocity recorded during exercise emphasizing the value of exercise stress echocardiography in HFpEF patients. Previous reports have already demonstrated the test's diagnostic value^2,3,5 and its limits,²⁵ whereas its prognostic value was less obvious. Estimation of PAP during exercise using echocardiography has already been reported to be associated with prognosis, especially in heart valve diseases.^26,27 The prevalence of pulmonary hypertension as estimated at rest by echocardiography, its impact on functional status, and its impact on prognosis have already been discussed.²⁸ Lam et al. reported pulmonary hypertension in 83% with a median systolic PAP of 48 mmHg²⁹ in a community study involving 244 HFpEF patients with a mean age of 76 years. The pathophysiological background of pulmonary hypertension, a reflection of left ventricular increased filling pressures, as well as its rapid and critical increase during exercise in HFpEF patients, also has been demonstrated.²⁸ Borlaug et al. in HFpEF subjects experienced significantly greater exercise-induced increases in mean PAP than subjects with non-cardiac dyspnoea despite achieving lower peak cardiac outputs.^28,30 In the same study, exercise PASP, non-invasively recorded, identified HFpEF with 96% sensitivity and 95% specificity using a cut point of 45 mmHg. Exercise PASP outperformed resting PASP, natriuretic peptide levels, and echocardiographic indicators for diagnosing HFpEF in Borlaug et al. experience.^30,31 Thus, high filling pressures with the propensity for rapid increments during exercise provide a key prognostic information and potentially a theoretical therapeutic target. But, despite diuretics no specific treatment can currently be recommended in spite of high expectations for sildenafil that have been frozen by the Relax trial.^30,32–34

Limitations

Several limitations have to be highlighted. Inviting elderly HFpEF patients to participate in exercise stress echocardiography remains a challenge. Exercise stress echocardiography may be of interest in difficult cases, although patients must be fit enough to undergo the test. Therefore, this testing is unlikely to be considered a key tool for assessing all HFpEF patients despite its prognostic usefulness, as revealed by our study data. The hand grip would be easier to perform, but bears its own imperfections, such as lack of standardization and insufficient reproducibility. Dobutamine stress echocardiography could be performed for researching ischaemia but probably not more as Dobutamine will start by inducing a decrease in pre and afterload. The workload used during the exercise was low but stable allowing the images acquisitions. Also, the experience shows us that when performing an exercise stress echocardiography, the heart maladjustment to changes in loading condition is most of the time observed at a low level of exercise even in valvular heart diseases.²⁷

Conclusions

Exercise echocardiography may contribute to identify HFpEF patients and especially high-risk ones. Our study suggested a prognostic value of TR recorded during an exercise. That was demonstrated independently of the value of resting E/e′.

Funding

We would like to thank Medtronic Europe, France, and Sweden. Our thanks also go to the French Federation (FFC), French Society of Cardiology (SFC), and Swedish Society of Cardiology for their dedicated grants that made the KaRen study feasible.

Acknowledgements

We also thank the research nurses: Marie Guinoiseau, Rennes University Hospital, Valerie Le Moal, Rennes University Hospital. The French Cardiac Society: Anissa Bouzamondo, Genevieve Mulak, and Elodie Drouet.

Conflict of interest: There are no commercial products involved in this study. However, to the extent that findings in KaRen may affect the use of heart failure drugs or devices, we disclose the following: L.H.L.: research grants and/or speaker and/or consulting honoraria from AstraZeneca, Novartis, Boston Scientific, and St Jude Medical; C.L.: principal investigator of REVERSE, a CRT study sponsored by Medtronic research grants, speaker honoraria, and consulting fees from Medtronic, speaker honoraria and consulting fees from St Jude Medical; E.D.: speaker honoraria and consulting fees from Novartis, Bristol-Myer-Squibb; J-.C.D.: research grants, speaker honoraria and consulting fees from Medtronic and St Jude Medical.