High Prevalence of Exercise-induced Heart Failure with Normal Ejection Fraction in Post-heart Transplant Patients

Aim. Post-heart transplant patients are at increased risk of diastolic dysfunction. The aim of this study was to assess the prevalence of isolated only exercise-induced heart failure with normal ejection fraction (HFNEF) in heart transplant recipients. Methods and Results. To determine pulmonary capillary wedge pressure (PCWP) at rest and during exercise, 81 patients after orthotopic heart transplantation with normal left ventricular ejection fraction (LVEF) underwent exercise right heart catheterization with simultaneous exercise echocardiography. Based on PCWP values, the patients were divided into three groups. Twenty-one patients had no evidence of HFNEF (PCWP at rest < 15 mmHg, maximal PCWP during exercise < 25 mmHg, prevalence 26%). Forty-seven subjects were found to have only exercise-induced HFNEF (PCWP at rest < 15 mmHg, maximal PCWP during exercise ≥ 25 mmHg, prevalence 58%). Thirteen patients had HFNEF already at rest (PCWP ≥ 15 mmHg at rest, prevalence 16%). Of the noninvasive parameters obtained at rest, multivariate regression analysis identified LV mass index adjusted for allograft age to be an independent predictor of exercise-induced HFNEF. Conclusions. In heart transplant recipients with normal LVEF, there is a high prevalence of exercise-induced HFNEF. LV mass index adjusted for allograft age is predictive of exercise-induced HFNEF.


INTRODUCTION
Heart failure with normal left ventricular ejection fraction (HFNEF) is a frequent disease affecting up to 50% of patients with clinical features of heart failure 1 .Even if their risk of death is lower than that of patients with heart failure and reduced left ventricular ejection fraction (LVEF), the absolute mortality of patients with HFNEF is still high 2 and requires new approaches in the prevention and treatment to improve prognosis.A basic step in the diagnostics of HFNEF is evidence of an increase in the LV filling pressure (LVFP) suggesting clinically significant LV diastolic dysfunction 3 .However, a significant proportion of patients suffer from heart failure symptomatology only during exercise.Borlaug et al. 4 demonstrated the need for exercise to diagnose HFNEF in 58% of patients with unexplained dyspnea and normal hemodynamics at rest.The exercise may be particularly useful in post heart transplant patients with unexplained exertional dyspnea or fatigue.Heart transplant recipients frequently suffer from a number of risk factors of diastolic dysfunction such as hypertension, LV hypertrophy, myocardial fibrosis, diabetes mellitus, obesity, post-transplant vasculopathy, etc. and, therefore, are at a high risk of exercise-induced HFNEF.The aim of our study was to assess the prevalence of isolated, only exercise-induced HFNEF in patients after orthotopic heart transplantation and to determine independent predictors of exercise HFNEF development.

Patient population
The study comprised 81 patients after orthotopic heart transplantation, who were referred to St. Anne's Hospital (Brno, Czech Republic) for post-transplant cardiac examination and gave their informed consent to the examinations, including simultaneous exercise right heart catheterization and echocardiography.All patients were regularly followed up at this hospital within the long-term programme of care for patients after heart transplantation.Coronary angiography was performed 12 months after the transplantation and repeated later if clinically indicated.The inclusion criteria were: a time interval ≥ 6 months after heart transplantation; a sinus rhythm on the electrocardiogram; no history of myocardial infarction or angina pectoris after heart transplantation and a standard transthoracic echocardiogram demonstrating a normal LVEF (≥ 50%), no significant pericardial effusion, and no mitral regurgitation other than trivial.The study complied with the Declaration of Helsinki and was approved by the ethics committee at St. Anne's Hospital in Brno.

Study protocol
Initially, standard transthoracic echocardiography was performed to determine whether the patients fulfilled echocardiographic inclusion criteria.On the following day, supine resting, exercise and recovery echocardiography and right heart catheterization with pulmonary capillary wedge pressure (PCWP) measurements were performed simultaneously.Examinations were performed in the morning after at least a 30-min rest.Measurements of PCWP, heart rate and blood pressure were initially performed at rest in patients with legs in a horizontal position and then repeated at rest four to five min after patient's leg elevation due to preparation for cycling.The same measurements were made during exercise at the end of each workload, at the time of termination of exercise (peak exercise), and at the end of each min of the postexercise recovery period until PCWP normalization.At both resting positions, at peak exercise and at the end of post-exercise recovery period, echocardiographic recordings of transmitral and mitral annular tissue Doppler velocities were obtained simultaneously with PCWP measurements.In patients referred for endomyocardial biopsy, several specimens were collected after finishing the study.On the day of catheterization, the morning medication was omitted.

Right heart catheterization
A 7F Swan-Ganz thermodilution catheter (model 131HF7, Baxter Healthcare Corporation, Irvine, CA, USA) was inserted into the pulmonary capillary wedge position.The correct balloon position was verified by the presence of characteristic wedge pressure waveforms.PCWP was measured with a zero level at the midaxillary line.PCWP, heart rate, and systemic blood pressure measurements were obtained using a multiparametric module Ultraview SL (TM) 91496 (Spacelabs Healthcare, Issaquah, WA, USA).PCWP was averaged over pressure waveform data obtained during a 12-s interval and expressed as a mean.

Echocardiography
Echocardiographic examinations were performed using Vivid E9 (GE Healthcare, Wauwatosa, WI) with an M5S transducer.At rest, grey-scale two-dimensional images were recorded from the parasternal short axis views at the base, at the level of papillary muscles and at the apex, as well as from the apical two-, three-, and four-chamber views.Three to five consecutive cardiac cycles in each view were digitally stored.Aortic and transmitral flows were recorded using pulsed-wave Doppler echocardiography.Pulsed-wave Doppler tissue imaging (DTI) of mitral annular motion was performed in the apical four-chamber view.A sample volume of 6.0 mm was placed on septal and lateral mitral annular corners.All Doppler recordings were done during shallow respiration or end-expiratory apnea.During the exercise, LV filling was monitored in the apical four-chamber view.The acquisition of peak exercise images started at the time of termination of cycling.

Exercise protocol
Graded supine bicycle ergometry limited by the onset of symptoms was performed starting at 25 W for two min.The load was then increased in increments of 25 W at twomin intervals until the occurrence of the first symptom (dyspnea or fatigue).All exercise tests were performed on an ergometer Ergoline GmbH (type er900L, Bitz, Germany).The patients were lying and cycling with their trunk in a horizontal position with their legs slightly elevated.

Echocardiographic parameters analyzed
Measurements were performed according to recommendations of the American Society of Echocardiography 5 .The data were analyzed offline using EchoPAC PC versions 108.1.5-110.1.1 (GE Vingmed Ultrasound A/S, Horten, Norway).LV mass was estimated using Devereux formula 6 .LV and left atrial (LA) volumes were calculated by means of the biplane method of disks 5 using apical four-and two-chamber views.From the conventional pulsed-wave Doppler recordings, the peak early diastolic transmitral flow velocity (E), the peak late diastolic transmitral flow velocity (A), and the deceleration time of E wave (DT) were measured.From DTI, peak systolic and early diastolic mitral annular velocities at the septal corner and at the lateral corner were measured and the values were averaged (s' and e').All echocardiographic parameters were obtained as a mean of three to six consecutive heart cycles.The analyses were performed by one experienced observer (JM), who was blinded to PCWP values.The variability of resting and exerciseinduced Doppler results of this reader has already been published 7 .Mean absolute differences of intra-observer repeated measurements for both resting and exercise E, e', s' did not exceed 5%.

Definition of heart failure with normal left ventricular ejection fraction
HFNEF is defined by the presence of a history of exertional dyspnea and/or by the evidence of a low exercise tolerance (< 100 W using the above described exercise protocol), by the finding of non-dilated left ventricle (enddiastolic volume index < 97 mL/m 2 ) with normal LVEF > 50%, and by the evidence of diastolic dysfunction 3 indicated by invasively measured PCWP elevation.PCWP at rest ≥ 15 mmHg and/or maximal exercise PCWP ≥ 25 mmHg were considered elevated and indicative of HFNEF (ref. 4 ).

Statistical analysis
The baseline clinical characteristics, echocardiographic data, PCWP, and hemodynamic results were analyzed descriptively and compared between groups.Univariate and multivariate logistic regressions were used to determine predictors and independent predictors of exercise-induced HFNEF.Standard measures of summary statistics were used to describe the primary data: relative and absolute frequencies for categorical variables, arithmetic mean supplied with standard error of the mean (SEM) for continuous variables.Most variables did not present normal distribution (Shapiro-Wilk's test), therefore, nonparametric tests were applied.For the comparison of all three groups in continuous parameters, the Kruskal-Wallis test was conducted.For the detailed mutual comparison of the groups, the Wilcoxon Rank Sum test was performed.The Chi-square test was applied for categorical data.To determine independent predictors of the isolated, only exercise-induced HFNEF, univariate and multivariate regression analyses were performed in patients with a normal PCWP at rest (< 15 mmHg).Clinical data and resting echocardiographic and hemodynamic parameters were included.For parameters with potential predictive power (providing at least P<0.10 in univariate logistic regression), various multivariate models were used.Odds ratios (OR), their 95% confidence intervals, and p-values are presented.Receiver operating characteristic (ROC) curves were constructed to identify sensitivities, specificities, areas under the curve (AUC), and the optimal cut-off values of independent predictors of the presence of exercise-induced HFNEF.Results with a P-value< 0.05 were considered statistically significant.

Prevalence of isolated only exercise-induced heart failure with normal ejection fraction
Out of the 87 initially screened patients, 81 subjects fulfilled the inclusion criteria and were included in this study.According to the resting (patients in a horizontal position) and maximal exercise PCWP values, the patients were divided into three groups.Group A comprised 21 patients without HFNEF (PCWP at rest < 15 mmHg, maximal exercise PCWP < 25 mmHg), Group B included 47 patients with exercise-induced HFNEF (PCWP at rest < 15 mmHg, maximal exercise PCWP ≥ 25 mmHg), and Group C comprised 13 patients with HFNEF at rest (PCWP at rest ≥ 15 mmHg).All Group B and C patients reported a history of exertional dyspnea and/or had a low exercise tolerance due to fatigue and/or shortness of breath.The distribution of patients in relation to HFNEF diagnosis is demonstrated in Fig. 1.
All patients were transplanted using the bicaval technique 8 and were on a combination of at least two immunosuppressive drugs.At the time of this study, endomyocardial biopsy was performed in 57 (70%) patients; according to the modified ISHLT histological classification of cellular rejection 9 , 27 patients had grade 0, 21 patients grade 1A, seven patients grade 1B, and two patients had grade 2 rejection.The distribution of rejection episodes in individual groups at the time of evaluation is demonstrated in Table 1.Sixty-one patients underwent coronary angiography (16 Group A, 34 Group B, and 11 Group C patients).Only four of them (one Group A, two Group B, and one Group C patients, P = NS) were found to have significant coronary vasculopathy (luminal diameter narrowing of at least one major coronary artery ≥ 50%).

Clinical characteristics of Group A, B, and C patients
Table 1 summarizes the baseline clinical characteristics of Groups A, B, and C. Group A patients were less frequently on angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers than Group C patients.There were no significant differences between Groups A and B. Table 2 shows two-dimensional echocardiographic parameters related to the diagnostics of HFNEF or diastolic LV function as well as the most important pulsedwave and tissue Doppler resting and exercise parameters.Groups A, B, and C differed in many Doppler-derived parameters reflecting differences in LV diastolic function.Complete or nearly complete fusion of early and late diastolic velocities or the inferior quality of exercise images did not allow the assessment of exercise E, DT, s', and e' in 13, 24, 12, and 13 transplanted subjects, respectively.Table 3 presents the hemodynamic and PCWP data obtained at rest and maximal or peak exercise results.As expected, patients with HFNEF (Groups B and C) had significantly higher LV filling pressures both at rest and mainly during exercise compared to Group A.

Predictors of isolated exercise-induced heart failure with normal ejection fraction
In the clinical setting, patients with PCWP elevation present even at rest (Group C in our study) do not require exercise to diagnose HFNEF.In contrast, the exercise is important in patients suspected of having HFNEF with a normal or borderline PCWP at rest.Thus, the ability of clinical variables and of hemodynamic and echocardiographic parameters obtained at rest to predict the exerciseinduced elevation of PCWP ≥ 25 mmHg was analyzed in Groups A and B patients with resting PCWP < 15 mmHg.In a multivariate regression model without adjustment for parameters, only PCWP was found to be a significant independent predictor of exercise HFNEF (odds ratio 1.32, 95% confidence interval 1.06-1.66,P=0.014).
After adjustment for allograft age, LV mass index (optimal cut-off 100 g/m 2 with a sensitivity of 76.2% and a specificity of 57.5%, AUC 0.66) in addition to PCWP  (optimal cut-off 10 mmHg with a sensitivity of 71.5% and a specificity of 59.6%, AUC 0.69) were proved to be independent predictors.Results of the univariate and multivariate regression analyses after adjustment for donor heart age are demonstrated in Table 4.

DISCUSSION
Our study has provided some important facts and has one priority.To our knowledge, it is the first study to describe the high prevalence of isolated only exerciseinduced HFNEF reaching 58% in a cohort of 81 post-heart transplant patients with normal LVEF.This finding explains their low exercise tolerance due to exertional dyspnea and/or fatigue.Of the noninvasive parameters obtained at rest, multivariate regression analysis identified LV mass index adjusted for allograft age to be the independent predictor of exercise-induced HFNEF.

Aspects of left ventricular filling pressure elevation and development of heart failure with normal ejection fraction in post-heart transplant patients
The most important aspects of LVFP elevation and development of HFNEF in the general population are hypertensive LV hypertrophy, myocardial fibrosis, vas-  1.
Data is presented as mean ± SEM.P: statistical test comparing all three groups; * P < 0.05 vs patients with the absence of HFNEF; ** P < 0.01 vs patients with the absence of HFNEF; † P < 0.05 vs patients with HFNEF at rest; † † P < 0.01 vs patients with HFNEF at rest.cular endothelial dysfunction, and changes in titin-based regulations of diastolic function 10 .All of these aspects may also play an important role in post-heart transplant patients being frequently influenced by risk factors of diastolic dysfunction such as hypertension, diabetes mellitus, obesity, and metabolic syndrome [11][12][13][14][15][16] .As demonstrated previously 11 and confirmed in this study, systemic hypertension is a very common finding in patients after heart transplantation.Hypertension is frequently associated with LV hypertrophy which results in diastolic dysfunction 17,18 .Severe LV hypertrophy is known to reduce subendocardial coronary flow reserve which can lead to subendocardial ischemia.Long-lasting hypertension is accompanied by a progression of myocardial fibrosis resulting in an increased myocardial stiffness.Moreover, there are also specific factors directly related to heart transplantation that can further induce or accelerate diastolic dysfunction in cardiac allograft recipients [19][20][21][22][23][24][25][26][27] .These include myocardial denervation, volume overload due to a mismatch between the size of donor and recipient hearts 21 , the effect of allograft ischemic time during the surgery 22,26 , pericardial constriction because of pericardial effusion 25 , acute cardiac rejection 20,23,27 , myocardial ischemia due to significant post-transplant vasculopathy, and immune heart injury associated with persistent tumor necrosis factor-alpha expression 28 .Serious risk factors for diastolic dysfunction are also progressive myocardial hypertrophy and fibrosis due to immune injury 28 , long donor  1 and 2. Statistical data presentation as in Table 2.For the parameter LVMI unit 5 g/m 2 was used in order to present odds for exercise-induced HFNEF if the value of LVMI is increased by 5 g/m 2 .For all the other parameters unit 1 was used.Abbreviations as in Tables 2 and 3.
heart ischemic time during surgery 26 , and due to healing of reccurent episodes of myocardial rejection 24 .Thus, owing to multiple potential risk factors, the post-heart transplant patients are at a high risk of both subclinical and manifest diastolic dysfunction and heart failure.Some of these risk factors, mainly early post-transplant immune reaction and the size of transplanted heart, are more important early after transplantation and are unlikely to significantly influence hearts late after transplantation.In the late post-transplant period, as found in the majority of our patients, myocardial ischemia due to vasculopathy and mainly progressive fibrosis and LV hypertrophy associated with repeated episodes of rejection and long-last-ing hypertension likely become of greater importance.In our patients with PCWP < 15 mmHg at rest, i.e. in those at risk of only exercise-induced LVFP elevation, resting PCWP and LV mass index both adjusted for donor heart age independently predicted exercise-induced HFNEF.LV mass index a quantitative measure of the severity of LV hypertrophy can easily be obtained noninvasively using standard echocardiography and its relationshisp to diastolic dysfunction is well established 17,18,29 .Thus, its elevation in transplanted patients with unexplained exertional dyspnea or fatigue may suggest the need for exercise to diagnose HFNEF.

Left ventricular diastolic dysfunction in patients after orthotopic heart transplantation
To date, a number of authors have described the presence of diastolic dysfunction both early and late after orthotopic heart transplantation 23,24,27,[30][31][32][33][34][35][36] , frequently accompanied by PCWP elevation 23,24,27,31,[33][34][35][36] .Impaired LV diastolic function in heart transplant recipients was found even if LVEF was normal 30,32 and if there was no or only 1A myocardial rejection 30 .Diastolic dysfunction was attributed to both the worsening of relaxation 37 and an increase in myocardial stiffness 22,35 .The common presence of structural post-transplant myocardial damage in the absence of myocardial rejection and without reduction of LVEF was also suggested by worsening of LV longitudinal systolic function which precedes the decrease in LVEF (ref. 11,30).As LV systolic and diastolic functions are closely coupled 38,39 , worsening of the LV longitudinal systolic function can contribute to LV diastolic dysfunction.
There have been several reports on post-transplant exercise-induced LVFP changes 19,21,37,[40][41][42] .However, these studies included small numbers of patients (the largest comprising 30 patients) and none assessed the prevalence of exercise-induced HFNEF.Rudas et al. 40 described in 20 post-heart transplant recipients without rejection a significant increase in PCWP from 13 ± 4 mmHg to 27 ± 7 mmHg during supine graded exercise; results very similar to those found in our study.PCWP exercise-induced increase was attributed to abnormal left-shifted and steep diastolic pressure-volume relation in transplanted patients indicating abnormal cardiac compliance.Hosenpud et al. 21found PCWP increase from 10 ± 3 mmHg to 20 ± 6 mmHg during supine exercise in 23 patients with a normal LVEF at one year after heart transplantation.The greatest resting and exercise LVFP had patients with little or no change in end-diastolic volume on exercise.Paulus et al. 37 studied 27 heart transplant recipients with a normal LVEF, who were free of rejection and of significant graft atherosclerosis.Supine exercise resulted in a smaller acceleration of LV relaxation than in a normal control group exercising to the same heart rate.There was a significant association between elevation in LV end-diastolic pressure (from 14 ± 4 mmHg to 25 ± 7 mmHg) and slower LV isovolumic relaxation 37 .Thus, elevated LVFP during exercise can result not only from abnormal passive LV diastolic properties 40 and preload reserve 21 , but also from altered LV relaxation 37 .To date, however, no study has focused on the prevalence of exercise-induced HFNEF in cardiac allograft recipients and on two-dimensional and Doppler echocardiographic predictors of exercise HFNEF.Our study, which included 81 post-heart transplant patients demonstrated a high prevalence of exercise-induced HFNEF.This diagnosis was carefully documented by the presence of exertional dyspnea and/or low exercise tolerance, by the evidence of non-dilated left ventricle with a normal LVEF, and by invasively measured PCWP both at rest and during exercise.The diagnosis of latent, only exercise-induced HFNEF can optimize patient management and prevent further heart failure progres-sion.However, the exact clinical significance of this early HFNEF diagnosis has yet to be determined.

Study limitations
The sample of our patients after orthotopic heart transplantation was narrowed by the study inclusion criteria that primarily focused on the diagnostics of HFNEF.A study of unselected post-heart transplant patients would probably result in a different exercise-induced HFNEF prevalence.However, in view of a good LVEF in the majority of transplanted patients at the late follow-up, the high prevalence of exercise-induced HFNEF described in our study is undoubtedly a reality.Based on the results, we cannot determine the exact contribution of coronary vasculopathy and of acute allograft rejection to exerciseinduced HFNEF.Since 25% of patients did not undergo coronary angiography and others had a long-time interval (> one year) between this study and the last coronary angiography, we cannot exclude significant but clinically silent allograft vasculopathy in these patients resulting in a silent exercise-induced myocardial ischemia and diastolic dysfunction.Similarly, 30% of patients did not undergo endomyocardial biopsy at the time of measurement.This was mainly in patients with a long interval since the transplantation and with a low probability of rejection.However, a subclinical rejection at the time of study cannot be completely excluded in these patients.

CONCLUSIONS
In heart transplant recipients with normal LVEF, there is a high prevalence of exercise-induced HFNEF.The LV mass index adjusted for allograft age is predictive of exercise-induced HFNEF.

Fig. 1 .
Fig. 1.Distribution of patients in relation to heart failure with normal ejection fraction (HFNEF).
ACEi, angiotensin-converting enzyme inhibitor; AT II, angiotensin II receptor blocker; HFNEF, heart failure with normal ejection fraction; OHT, orthotopic heart transplantation; PCWP, pulmonary capillary wedge pressure.Data is presented as mean ± SEM or number (%).P, statistical test comparing all three groups; † † P < 0.01 vs patients with HFNEF at rest.
DT, deceleration time of E wave; E, peak early diastolic transmitral velocity; e', peak early diastolic mitral annular velocity; EDVI, end-diastolic volume index; ESVI, end-systolic volume index; ele, measured at rest with leg elevation; exe, measured at peak exercise; LAVI, left atrial volume index; LVEF, left ventricular ejection fraction; LVMI, left ventricular mass index; rest, measured at rest with legs in a horizontal position; RWT, relative wall thickness; s', peak systolic mitral annular velocity; Other abbreviations as in Table
BP, blood pressure; max, maximal PCWP during exercise.Other abbreviations as in Tables

Table 4 .
Univariate and multivariate logistic regression, including potential exercise-induced HFNEF predictors adjusted for allograft age.