Article Text

Download PDFPDF

Pathophysiological rationale and diagnostic targets for diastolic stress testing
  1. Tamás Erdei1,
  2. Svend Aakhus2,
  3. Paolo Marino3,
  4. Walter J Paulus4,
  5. Otto A Smiseth2,
  6. Alan G Fraser1
  1. 1Wales Heart Research Institute, Cardiff University, Cardiff, UK
  2. 2Department of Cardiology, Oslo University Hospital, Rikshospitalet, Oslo, Norway
  3. 3Cardiology Clinic, Universita Piemonte Orientale, Novara, Italy
  4. 4Department of Physiology, Institute for Cardiovascular Research, VU University Medical Center, Amsterdam, The Netherlands
  1. Correspondence to Professor Alan G Fraser, Wales Heart Research Institute, School of Medicine, Cardiff University, Heath Park, Cardiff CF14 4XN, UK; fraserag{at}


Cardiopulmonary functional reserve measured as peak oxygen uptake is predicted better at rest by measures of cardiac diastolic function than by systolic function. Normal adaptations in the trained heart include resting bradycardia, increased LV end-diastolic volume and augmented early diastolic suction on exercise. In normal populations early diastolic relaxation declines with age and end-diastolic stiffness increases, but in healthy older subjects who have exercised throughout their lives diastolic function can be well preserved. The mechanisms by which LV diastolic filling and pressures can be impaired during exercise include reduced early diastolic recoil and suction (which can be exacerbated by increased late systolic loading), increased preload and reduced compliance. Abnormal ventricular-arterial coupling and enhanced ventricular interaction may contribute in particular circumstances. One common final pathway that causes breathlessness is an increase in LV filling pressure and left atrial pressure. Testing elderly subjects with breathlessness of unknown aetiology in order to detect worsening diastolic function during stress is proposed to diagnose heart failure with preserved EF. In invasive studies, the most prominent abnormality is an early and rapid rise in pulmonary capillary wedge pressure. A systematic non-invasive diagnostic strategy would use validated methods to assess different mechanisms of inducible diastolic dysfunction and not just single parameters that offer imprecise estimates of mean LV filling pressure. Protocols should assess early diastolic relaxation and filling as well as late diastolic filling and compliance, as these may be affected separately. Better refined diagnostic targets may translate to more focused treatment.

View Full Text

Statistics from


Many patients with heart failure have symptoms only during exercise but diagnostic recommendations are based on tests performed at rest. Patients with heart failure with preserved EF (HFPEF) have normal global LV systolic function at rest, but they can also have normal or only mildly abnormal global diastolic function at rest, for their age.1 Analysing changes in diastolic function during exercise or loading interventions should identify which mechanisms are most affected in patients.

We discuss the physiological responses of LV diastolic function to dynamic exercise, pharmacological stress and acute changes in loading, and then review how these can be impaired in patients with HFPEF. We summarise echocardiographic measurements that can be used to assess diastolic function during stress.

Normal diastolic responses to stress

Dynamic exercise

During physical exertion, cardiac output is augmented by increasing heart rate and stroke volume. Sinus tachycardia abbreviates diastolic filling so more blood must be transferred from the left atrium (LA) to the LV in a shorter time if stroke volume is to be maintained or increased.

In healthy subjects, the onset of LV untwisting and myocardial relaxation (‘regional’ diastole) precedes aortic valve closure (and the start of ‘global’ diastole). Early relaxation creates an expansion wave that travels as a rarefaction from the LV to the aorta where it decelerates flow and reduces pressure, causing the aortic valve to close. The mitral valve opens when LV early diastolic pressure falls below LA pressure. On exercise, diastolic filling is accelerated. Sympathetic stimulation causes a downward shift of the early diastolic portion of the LV pressure-volume loop. Myocardial relaxation is shortened due to increased calcium reuptake by the sarcoplasmic reticulum. Elastic recoil and LV untwisting are increased and the LV pressure decays faster in early diastole. These mechanisms increase LV suction2 so the transmitral pressure gradient and early diastolic filling are increased without any increase in LA pressure.

During exercise the isovolumic relaxation time shortens, the PR interval shortens (by 7 ms for every increase in heart rate of 10 bpm) and the optimal atrioventricular (AV) delay is shorter. The velocities of early mitral filling (E wave; figure 1, upper panels) and diastolic lengthening of the LV (mitral annular e′ velocity; figure 2, upper panels) increase. With advancing age, the resting isovolumic relaxation time is prolonged, the proportion of filling during early diastole declines and the atrial filling fraction increases. In elderly subjects, heart rate and PR interval are independent determinants of LV diastolic filling, and at higher heart rates the atrial filling fraction is increased.

Figure 1

Mitral inflow recorded by pulsed Doppler echocardiography at rest and during submaximal semisupine bicycle exercise in a 63-year-old healthy subject and in a 67-year-old patient with heart failure with preserved EF (HFPEF). E, peak early diastolic velocity of mitral flow. Note that in the healthy subject (‘Normal’) E more than doubles during submaximal exercise (heart rate 104/min; increase from baseline +51 bpm). In the patient (‘HFPEF’) E increases by only 11% indicating poor stroke volume reserve (at heart rate 94/min; increase +44 bpm), partly compensated by increased duration and integral of flow during atrial contraction.

Figure 2

Regional (longitudinal) myocardial function at rest and during submaximal semisupine bicycle exercise in a 63-year-old healthy subject and in a 67-year-old patient with heart failure with preserved EF (HFPEF). The traces show velocity profiles of long-axis myocardial function obtained by processing of digitally stored colour velocity loops of apical four-chamber images of the LV, obtained in the same subjects as shown in figure 1. Vs, peak velocity in systole; Ve, peak velocity in early diastole; bs, basal septal site; bl, basal lateral wall. Note that during submaximal exercise Ve increases in the normal subject, while in the patient it does not change but the velocity during atrial filling increases. Changes in systolic function (Vs) parallel those in diastolic function.

In contrast to early diastolic relaxation, LV (late) diastolic compliance may be relatively unaffected by ageing. In healthy older subjects, a given volume load caused a similar increase in mean pulmonary capillary wedge pressure (9±2 mm Hg to 17±2 mm Hg, n=27) to that observed in young control subjects (10±2 mm Hg to 16±2 mm Hg, n=20).3 Impaired early diastolic relaxation does not influence LV stiffness4 and LV compliance is preserved in older subjects who exercise regularly. In 249 subjects without cardiovascular disease, the end-diastolic elastance increased by only 8% over 4 years.5

Pharmacological stress

In healthy subjects, dobutamine increases the LV intraventricular pressure gradient in early diastole (an indicator of suction) which doubles from an average of 2.8 mm Hg at rest to 5.0 mm Hg at peak dose (36±6 μg/kg/min).6 In 16 patients aged 57±8 years who had normal LVEF and normal coronary arteriography, mean LV diastolic pressure fell during dobutamine from 12±3 mm Hg to 9±2 mm Hg.7 A possible explanation for improved diastolic distensibility during β-adrenergic stimulation may relate to the large sarcomeric protein titin which forms an elastic scaffold linking myosin filaments to the z bands, that acts like an intracellular molecular spring; β-adrenergic stimulation through activation of protein kinase G increases phosphorylation of titin8 which makes it more elastic.

Isometric stress

Maximal isometric exercise can be sustained only for a short time but less effort can be held for long enough to allow its haemodynamic consequences to be measured. In 53 healthy subjects aged 45±17 years, 30% of maximal hand grip maintained for 6.6 min increased heart rate by 20%, blood pressure by 17% and cardiac output by 27%, while LV end-diastolic volume and stroke volume measured by MRI did not change.9 In 100 normal volunteers, the contribution of atrial contraction to LV filling during hand grip exercise increased with age.

Athletes who have trained during isometric exercise (by weightlifting) have greater increases in LV end-diastolic volume during exercise than have sedentary controls (+17 mL compared with +3 mL).10 Diastolic filling was greater in the athletes despite an increase in LV mass, and since end-systolic volumes were similar, their stroke volume also increased more on exercise.

Changes in loading

Afterload varies throughout systole. In health, aortic pressure reaches its maximum early in systole, LV wall stress then declines during ejection, and end-systolic wall stress is low because the end-systolic volume is small and the wall is thickened. Arterial stiffness increases non-linearly with increases in pressure so it is higher during systole than diastole; this pressure-dependency is the best predictor of LV mass,11 which implies that above-average increases in LV afterload on exercise may be harmful.

LV contraction is prolonged when afterload is increased by augmenting systolic pressure. In patients with preserved LV global systolic function who were studied before cardiopulmonary bypass, early systolic loading caused by constricting the ascending aorta delayed the onset of the fall of LV pressure without altering its rate of fall, and neither changed the rate of early diastolic filling nor increased filling pressures, whereas in patients with reduced EF and especially after additional volume loading, increasing afterload delayed the development of early diastolic suction and increased filling pressures.12 In contrast, in an experimental study, increasing LV afterload in late systole caused an earlier onset of the fall in LV pressure but at a slower rate (with prolongation of τ) which delayed and reduced early diastolic filling. Hypertensive subjects have higher LV end-systolic wall stress than normotensive controls.

LV function is also affected by acute changes in preload. Early diastolic filling is determined by lengthening load (equivalent to preload), LV relaxation (pressure decay) and restoring forces.13 The preload-dependence of mitral E velocity decreases with worsening relaxation. Mitral E and myocardial e velocities are sensitive to increases in preload, for example during straight leg raising, and to reductions during a Valsalva manoeuvre, after glyceryl trinitrate, or with application of lower body negative pressure. In 18 subjects with LVEF>45%, glyceryl trinitrate advanced the onset of untwisting by 22 ms, whereas rapid infusion of 750 mL saline delayed the time of peak untwisting from 37 ms before, to 9 ms after, mitral valve opening.14

Stress-induced diastolic dysfunction

In HFPEF, diastolic filling may be delayed, slowed, shortened, reduced or associated with elevated LV pressures. Patients may have a limited capacity to increase early diastolic filling and cardiac output during exercise without increasing filling pressures. Mechanisms that may underlie abnormal diastolic function at rest and influence functional reserve are listed in table 1.

Table 1

Possible mechanisms and diagnostic targets for diastolic dysfunction on exercise

Ventricular-arterial coupling

Arterial elastance is influenced by heart rate, and aortic stiffness correlates with enhanced sympathetic tone at rest and during exercise. In HFPEF, moderate exercise revealed an increase in arterial stiffness and afterload that was not detected at rest, such that acute adaptation of ventricular-arterial coupling was impaired.15 Abnormal coupling contributes to the development of HFPEF and to acute pulmonary oedema. Increased arterial stiffness during exercise may augment loading disproportionately in those subjects who have worse resting aortic function. Early diastolic lengthening of the LV is inversely related to conduit arterial stiffness.16

In patients with HFPEF and few symptoms, diastolic function at rest and during exercise is related to central arterial pressure but not to peripheral pressure.17 A study during isometric exercise (40% maximal hand grip) demonstrated that LV hypertrophy was better explained by dynamic than by resting large arterial properties; adding carotid-femoral pulse wave velocity increased the accuracy of predictive models from 48% to 68%.18

Early diastolic filling

In HFPEF, the LV cannot increase suction normally on exercise because peak untwisting is delayed and its amplitude is reduced. Suction can be estimated by studying the velocity of propagation of early diastolic flow through the mitral valve (Vp). During semisupine exercise Vp increased less in subjects with HFPEF (mean increment +23%) than in controls (+56%; p<0.001).19 Reduced early diastolic filling is reflected by lower mitral annular velocities (e′) (figure 2, lower panels). During dobutamine, suction increased less in patients with diastolic dysfunction in proportion to the severity of their disease. In obese subjects, the increase in diastolic filling rate was 38% compared with 70% in controls. In patients with HFPEF dobutamine reduced mitral annular velocity (e′).

Myocardial ischaemia and diastolic stunning

Most coronary arterial flow occurs during early diastole. Since blood is sucked into the coronary arteries when the myocardium relaxes, reduced elastic recoil and early diastolic relaxation in hypertensive hearts explain sluggish flow and poor perfusion even when the coronary arteries are enlarged. Early diastolic LV dysfunction during exercise might impair perfusion even more.

Subendocardial arterioles and capillaries are squeezed more than subepicardial vessels during systole and they take longer to resume their full diastolic dimensions. The refilling of intramyocardial vessels during diastole depends on the rate of ventricular relaxation and the transmyocardial pressure gradient. Extravascular compressive forces, whose effects are limited at rest, become amplified as preload increases, for example during exercise. In patients with hypertension, the coronary microcirculation is impaired by extravascular compression from fibrosis and hypertrophy, as well as by endothelial dysfunction and increased coronary arterial tone. In HFPEF, coronary microcirculation is also affected by inflammation.8

Diastolic dysfunction can persist after exercise in patients with coronary artery disease for 30 min, 1 h,20 2 days or longer; recurrent ischaemia may produce cumulative effects that can be described as diastolic stunning or ischaemic memory. The magnitude of diastolic impairment is determined by the severity of exercise-induced ischaemia. In ischaemic myocardium, there may be delayed onset and reduced amplitude20 of early diastolic relaxation and filling, and increased regional myocardial stiffness.

LV compliance and filling pressures

Cardiomyocyte stiffening and interstitial fibrosis reduce LV compliance and shift the end-diastolic pressure-volume relationship upwards, so that at any given end-diastolic volume, pressure is increased. Abnormal myocardial stiffening may not be a problem during rest, but with exercise the limited ability to distend the ventricle explains why there is an attenuated increase in stroke volume and a more marked elevation of LA pressure than in healthy hearts. Patients with HFPEF have a steeper slope of the diastolic pressure-volume relationship than age-matched controls,3 and elevated filling pressures.21

In 20 patients with newly diagnosed HFPEF, exercise increased LV diastolic chamber stiffness by 50%.22 High filling pressures may be needed to maintain adequate filling and stroke volume, but worsening diastolic dysfunction causes breathlessness. Interaction between reduced early diastolic filling on exercise and increased dependency on filling during atrial contraction is illustrated in figure 1 (lower panels).

A stiff LV will not increase its volume very much when diastolic pressure is increased during exercise, but there will be a rise in LA and pulmonary arterial (PA) pressures and then in RV diastolic and right atrial (RA) pressures. Increased LA, RA and RV volumes set the stage for ventricular interactions. In a person who is rather inactive, either a sudden or slow onset of exercise will mobilise blood from the venous compartment. High RA and RV diastolic pressures cause stronger interactions via the septum and they increase pericardial pressure with stretch and ‘stiffening’ of the pericardium that causes constraint.

Inability to increase end-diastolic volume and stroke volume during exercise may be an important limiting factor in patients with HFPEF.22–24


Patients with HFPEF often have systolic (61%) and diastolic (36%) dyssynchrony and both are independent predictors for the development of clinical HFPEF in hypertensive heart disease.25 Dyssynchrony may increase during stress and uncoordinated contraction and relaxation may impair early diastolic filling.

Pulmonary hypertension

Resting PA pressure increases with age, by 20% between the fifth and eighth decades but by only +8% in those without cardiopulmonary disease. During exercise, PA pressure increases linearly in proportion to the cardiac output; the normal slope of the relationship between mean PA pressure and flow ranges from 0.5 mm Hg min/L to 3 mm Hg min/L.26 In 70 healthy volunteers, peak systolic PA pressure estimated by echocardiography increased during exercise on a semisupine bicycle from 27 mm Hg to 51 mm Hg; at a low workload of 25 Watts it never exceeded 60 mm Hg, but at peak exercise, systolic PA pressure was >60 mm Hg in 36% of subjects aged >60 years. The upper limit of normal for mean PA pressure during exercise is 34 mm Hg when the cardiac output does not exceed 10 L/min, and 52 mm Hg at a cardiac output <30 L/min.

PA pressure can be used to infer elevated LV filling pressures or increased pulmonary vascular resistance in suspected heart failure only if there is a disproportionate increase. Measurements need to be taken during exercise because PA haemodynamics change rapidly after stopping exercise. Exercise-induced pulmonary hypertension is common in subjects with normal EF and is strongly associated with age, gender, systolic BP and LV filling pressures.27 PA systolic pressures ≥60 mm Hg at low workloads should be considered abnormal.27

Diastolic limits to peak functional capacity

Patients with HFPEF have similar reductions in exercise tolerance to patients with systolic heart failure, with oxygen consumption (VO2 max) lower by 29–38% and maximal workload lower by 31–50% compared with controls.27 The limiting factors may differ in patients and athletes, but in general VO2 max correlates more with diastolic than with systolic function. Athletic training leads to an increase in the volume of all cardiac chambers with a secondary increase in pericardial volume. Stroke volume reserve is enhanced because LV end-diastolic volume is increased while end-systolic volume is relatively unchanged. During exercise LV end-diastolic volume usually increases (as long as the increase in intracavitary pressure is greater than that surrounding the LV) to a point where further increases are limited by the pericardium. In one recent study, exercise capacity was best predicted by the increase in intraventricular diastolic pressure gradient on exercise (r=0.8, p<0.001).28

In patients with HFPEF, the intensity of the forwards-travelling expansion wave generated by the LV before aortic valve closure correlates with exercise capacity (R=0.54). In older people with HFPEF, decreased augmentation of LV early diastolic suction during dynamic exercise correlated strongly with aortic stiffening and with reduced exercise capacity.28 Exercise capacity is inversely related to LV stiffness and to estimated LV end-diastolic pressure in patients with hypertrophic cardiomyopathy.

Diagnostic targets for diastolic stress tests

The rationale for diastolic stress testing is clear, but there may be multiple mechanisms for impaired functional reserve and therefore multiple targets. The normal response to exercise includes an increase in myocardial contractility, diastolic suction and heart rate, so that stroke volume, cardiac output and blood pressure all increase despite systemic arteriolar vasodilation. Stroke volume reserve should be assessed in all symptomatic subjects, irrespective of the aetiology of their symptoms.

Echocardiographic tests have good accuracy for estimating many haemodynamic variables, at least at rest and in patients with reduced EF, but it is unclear which variables should be key diagnostic targets during stress testing for suspected HFPEF and there is little consistency in the literature.29 Targets that relate to early diastolic relaxation or to late diastolic compliance are listed in figure 3; all merit investigation. A reliable and simple protocol needs to be developed and validated in patients with suspected HFPEF compared with healthy age-matched normal subjects, against independent reference criteria such as brain natriuretic peptide (BNP) levels or VO2 max.29 Different protocols may be appropriate to test reserve of diastolic suction or changes in filling pressures related to compliance. Clinical decision points should be determined from their relationship to clinical outcomes.

Figure 3

Echocardiographic variables that can demonstrate early diastolic or late diastolic dysfunction in patients with heart failure with preserved EF (HFPEF). The drawing illustrates pressure traces from the LV (bold line) and the left atrium (LA; dotted line) during diastole. The lists include changes that may be observed in patients with HFPEF. Note that some variables (marked with*) show biphasic rather than unimodal changes with worsening disease.

Normal changes on exercise that might be detected using echocardiography include an increase in the peak untwisting rate, a shortening of the LV isovolumic relaxation time, an increase in the propagation velocity of mitral inflow (Vp), and increases in the early diastolic velocities of longitudinal myocardial lengthening (e′) and mitral inflow (E) without any significant change in their ratio (E/e′). The reproducibility of many of these echocardiographic measurements obtained during exercise is suboptimal. Variables such as the mitral E/A ratio, the isovolumic relaxation time and the deceleration time of mitral inflow, show biphasic relationships with disease which make isolated measurements ambiguous. Atrial deformation during contraction (sometimes called booster pump function) gives information about LV compliance and atrial afterload. Although classically shortened in patients with high LV filling pressures, the mitral A wave may reveal increased dependence on atrial function during exercise (see figure 1).

Patients with HFPEF exhibit chronotropic incompetence during maximal exercise with 40% less increase in heart rate than controls,24 and they may have abnormal heart rate recovery after exercise indicating impaired cardiac vagal tone and delayed vagal reactivation.30 Chronotropic incompetence may be an overlooked cause of exercise intolerance in HFPEF but it is easy to measure. Abnormal heart rate recovery is defined as a fall in heart rate after stopping exercise that is <10–12 bpm at 1 min30 or <22 bpm at 2 min.


It is artificial to consider LV systolic and diastolic functions independently. Patients with HFPEF have impaired longitudinal systolic function of the LV as well as global diastolic dysfunction. Nonetheless, consensus that diastolic dysfunction is an important cause of symptoms in patients with normal LVEF means that a reliable clinical test is needed. Patients should be assessed during stress and preferably during dynamic exercise, with possible mechanisms for symptoms being considered during detailed studies using echocardiography or other imaging modalities. Interpretation should allow for the effects of ageing and the relationships of LV systolic and diastolic function with cardiovascular risk factors including exercise history. The contribution of changes in the peripheral circulation and metabolism should also be investigated. Without a logical approach, many difficulties may be encountered while integrating diastolic stress testing into clinical practice.


View Abstract


  • Contributors The review was devised by AGF, and TE who wrote the first draft. All authors contributed to several versions of the manuscript with detailed comments and suggestions for revision. AGF edited the final version, which all the named authors have read and approved.

  • Funding This study is supported by a grant from the European Union for the MEDIA project (The Metabolic Road to Diastolic Heart Failure; EU FP7 project number: 261409). TE had a Research Fellowship from the Heart Failure Association of the European Society of Cardiology.

  • Competing interests None declared.

  • Ethics approval South East Wales Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.