OUP user menu

Is central nervous system processing altered in patients with heart failure?

Stuart D. Rosen , Kevin Murphy , Alexander P. Leff , Vincent Cunningham , Richard J.S. Wise , Lewis Adams , Andrew J.S. Coats , Paolo G. Camici
DOI: http://dx.doi.org/10.1016/j.ehj.2004.03.025 952-962 First published online: 1 June 2004


Aims Breathlessness is a cardinal symptom of heart failure and the altered regulation of breathing is common. The contribution of abnormal central nervous system activity has not previously been investigated directly, although abnormal autonomic responses have been described. Our aim was to assess whether heart failure patients exhibit different patterns of regional brain activation after exercise stress.

Methods We used positron emission tomography with H215O, to measure changes in regional cerebral blood flow (rCBF) and absolute global cerebral blood flow (gCBF) in 6 male class II/III heart failure patients and 6 normal controls. Breathlessness (0–5 visual analogue scale) and respiratory parameters were measured at rest, after horizontal bicycle exercise and during isocapnic hyperventilation. CBF was measured in each condition in all subjects.

Results Both groups were similarly breathless after exercise and the respiratory parameters were comparable. rCBF differences for the main comparison (exercise vs hyperventilation) were: activation of the right frontal medial gyrus (Math) and left precentral gyrus (Math) in controls but not in patients. Both groups had rCBF increases in the left anterior cingulate (Math) and right dorsal cingulate cortex (Math). The gCBF did not differ between exercise, isocapnic hyperventilation and rest in patients but, in controls, gCBF was greater after exercise compared to either isocapnic hyperventilation or rest.

Conclusion Heart failure patients had a distinct pattern of regional cortical activity with exercise-induced breathlessness but unvarying CBF values between conditions. These central neural differences in activity may contribute to some features of heart failure, such as variability in symptoms and autonomic dysregulation.

  • Heart failure
  • Breathlessness
  • Cerebral blood flow
  • Exercise
  • Brain
  • Autonomic nervous system
  • Positron emission tomography
  • ANOVA analysis of variance
  • ANS autonomic nervous system
  • CHF chronic heart failure
  • CNS central nervous system
  • ECG electrocardiogram
  • fR respiratory frequency
  • gCBF global cerebral blood flow
  • MathCO2 partial pressure of carbon dioxide
  • PET positron emission tomography
  • rCBF regional cerebral blood flow
  • SaO2 arterial oxygen saturation
  • SPM statistical parametric mapping
  • Math volume of CO2 exhaled
  • VE minute ventilation
  • VEMath slope ventilatory equivalent for CO2
  • vs versus
  • VT tidal volume


In health, the regulation of breathing and cardiac output is very closely co-ordinated, maximising the efficiency of oxygen transfer to the body and delivery of oxygen at the tissue level. In cardiac or pulmonary disease, when cardiac output or oxygenation is abnormal, overall cardiopulmonary efficiency declines and the work of breathing is increased for the required oxygen delivery required. Both diseases are associated with breathlessness and fatigue, although the precise mechanisms involved probably differ somewhat between them.1,2

In heart failure, even when objective signs of pulmonary disease are absent, some patients display lower values of MathCO2 and higher respiratory frequency (fR).3 An altered ventilatory response to exercise (VEMath slope) has also been demonstrated and is an independent marker of prognosis.4–6 Dysregulation of breathing in chronic heart failure (CHF) might involve changes of control at several levels, ranging from peripheral ergoreflex activation7 and peripheral chemosensitivity,8 through abnormal autonomic reflexes9–11 to an altered central command.12 Furthermore, the relationship of these variables to the subjective sensation of breathlessness is elusive.2,13,14

The role of the central nervous system (CNS) in the regulation of breathing has been investigated in normal individuals using several technologies,15–24 including positron emission tomography (PET) with H215O, one of the most direct means of exploring neural function in vivo in man.25–27 Regional cerebral blood flow (rCBF) is measured as an index of regional synaptic activity during particular tasks or conditions.28

Although several independent methods (e.g., analysis of heart rate variability10–11) have pointed to abnormalities of automatic nervous system (ANS) function in chronic heart failure (CHF), altered activity of the CNS has not been systematically investigated. Because known abnormalities of breathing regulation have been demonstrated by other techniques, we predicted that important functional abnormalities of CNS activity might occur in CHF. We also sought to clarify whether the wide variation in the experience of breathlessness, known to correlate poorly with objective measures of impairment of cardiac function,14 might be explicable in terms of differences in cerebral cortical activation. (We have previously demonstrated this comparing painful and silent myocardial ischaemia.29) The specific hypothesis that we tested in this study was that the pattern of CNS activation, during physical stress is different in patients with CHF from that in age-matched normal controls.


Study population

Selection of heart failure patients

Six dextral male patients [age 62 (11) years], mean (SD) were recruited from consecutive out-patients of Hammersmith and Charing Cross Hospitals over a period of 1.5 years. All were patients with symptomatic systolic CHF, New York Health Association (NYHA) class II or III, controlled on medication. The aetiology of the CHF was coronary artery disease with previous myocardial infarction in 5 cases and idiopathic dilated cardiomyopathy in 1 case. During exercise testing, 5/6 patients and all controls had ECGs that were amenable to the detection of ischaemia; one patient was in left bundle branch block. There was no detectable inducible ischaemia during exercise in the study population. Echocardiography was performed according to standard protocols to assess left ventricular function. Lung function tests were also performed to exclude asthma or chronic obstructive pulmonary disease. Diabetes and autonomic neuropathy were also excluded, the latter by standard bedside tests. In addition, subjects with unstable cardiovascular disease, drug or alcohol addiction and those who had already undergone a PET scan or any other study involving ionising radiation within the last two years were excluded. Patients with clinical evidence of cerebrovascular disease were also excluded. The 6 patients who satisfied the above criteria were the result of investigating over 40 patients with CHF during an 18-month period. The characteristics of the patient group are shown in Table 1.

View this table:
Table 1

Characteristics of the patients and controls

16716462AMI 5 yearsS,A,D,ASFS 26%; anteroseptal akin; inf hypokinNSLBBB Negative maximal DSE90% prox LAD; Cx and RCAII
26817190AMI 6 yearsN,D,B,S,AS,F,ALFS 27% apical and inf infarctsEx 20 yearsQ V1-5; L-axis Embedded Image ST I, aVL,V3-6Occ LAD; 70%RCA; unobst Cx; Eff PCI-LADII
35416863AMI 1.5 yearsAS,A,D, LithiumFS 15% Ant and inf infarcts; dil LA and LVNSPoor R V1-3; nonspec Embedded Image ST V5,6Small unobst arteriesII
47016761AMI 8 yearsAS,D,N,AFS 16%; ant, apical and infpost hypokin, apex akinNSIncomplete LBBB; Embedded Image LA; Inferolat ST/T; VE's80% mid LAD; Occ Cx after OM1; Occ RCA; 3 SVG's and LIMAIII
571166111AMI 4 yearsA,Warfarin,N,DFS 3%; septum akin; hypo through-outEx 28 yearsPoor R V1-3; q in V4; Embedded Image T V1-6, I, aVL90% LAD; 90% Cx; Occ RCA; LIMA to LAD; SVG's: OM1 & RCAIII
64518584DCM 4 yearsDig,amio,D, Warfarin,S,AFS 23%; apical hypo10 g/wkEmbedded Image T inferolat; Q V1-3; VE'sNot performed (Patient refused)II
16415665No WMA; FS 33%NSWithin normal limits
26016766No WMA; FS 47%NSWithin normal limits
36618084No WMA; FS 29%Ex 25 yearsWithin normal limits
47216370No WMA; FS 40%20/day for 56 yearsWithin normal limits
57117859No WMA; FS 30%NSWithin normal limits
67217177No WMA; FS 29%NSWithin normal limits
  • Cause: AMI, acute myocardial infarction, number of years previously; DCM, dilated cardiomyopathy. Medication: S, statin; A, ACE-inhibitor; D, diuretic; AS, aspirin; N, nitrate; B, β-blocker; F, fibrate; AL, α-blocker; Dig, digoxin; Amio, amiodarone. Echocardiogram ; FS, fractional shortening; ant, anterior; inf, inferior; infpost, inferoposterior; akin, akinesis; hypokin, hypokinesis; dil, dilated; LA, left atrium; LV, left ventricle; WMA, wall motion abnormality. Smoke: smoker status; Ex, ex-smoker; NS, non-smoker. ECG: LBBB, left bundle branch block; DSE, dobutamine stress echocardiogram; L-axis, left axis deviation; Embedded Image ST, ST segment depression; Embedded Image LA, left atrial enlargement; Nonspec, non-specific changes; Poor R, poor R wave development; ST/T, ST and T wave changes; VE's, ventricular ectopic beats. Angiogram: LAD, left anterior descending coronary artery; Cx, left circumflex artery; RCA, right coronary artery; prox, proximal stenosis; Occ, occluded; Unobst, unobstructed; Ef, effective; PCI, percutaneous coronary intervention; OM1, first oblique marginal branch of circumflex; SVG, saphenous vein graft; LIMA, left internal mammary artery graft. NYHA: New York Heart Association heart failure class.

Control subjects

Six dextral normal sedentary male controls [age 67 (5) years, Math= NS vs the patients] were also studied. These were selected on the basis that they were both age- and sex-matched with respect to the patients. They were recruited randomly from hospital staff, visitors and relatives of patients on the basis of reply to a general call for study volunteers. They all had clear past medical histories with no background of cardiac or pulmonary disease, or risk factors for coronary artery disease. They underwent the same detailed cardiopulmonary assessment as the patients and demonstrated normal results, i.e., normal resting and exercise ECGs, normal echocardiography and normal lung function.

Population size

The sample size (6 patients and 6 controls) was chosen on the basis of several previous brain PET studies of this nature, in which this sample size was adequate to demonstrate significant differences in regional cerebral blood flow.

Pre-scanning assessment

Prior to the PET scanning session, all subjects underwent clinical rehearsals at the Respiratory Physiology Laboratory, Charing Cross Hospital to learn how to use the horizontal exercise bicycle. They had a symptom-limited test during which, in addition to continuous ECG, their blood pressure, heart rate, ventilation (ultrasonic respiratory flowmeter), MathCO2 (capnograph) and oxygen saturation (finger oximeter) was monitored. They also gave an estimate of their perceived breathlessness upon exertion using a modified Borg scale;30 in this case, a 0–5 scale (0=no sense of breathlessness, 5=intolerably severe breathlessness) with increments of 0.5.

On a separate occasion, subjects were taught to copy the rate and depth of breathing that they had displayed during the practice horizontal bicycle test. This was to provide a control condition for the physical respiratory efforts associated with post-exercise breathlessness. However, to avoid hyperventilation-induced hypocapnia, we adjusted the amount of CO2 in the inspired gas mixture, to keep the MathCO2 in the normal range. The condition was therefore termed `isocapnic hyperventilation'. From pilot work, we observed that isocapnic ventilation alone generated very little sensation of breathlessness. The idea was therefore that in the analysis of the scan data (described below), the exercise run, minus isocapnic ventilation, would equate to (physical effort of respiration+sensation of dyspnoea)−(physical effort of respiration), i.e., as close as possible to a true representation of the sensation of dyspnoea.

PET scanning protocol

On a different day, patients and controls attended the MRC Clinical Sciences Centre/IRSL, Hammersmith Hospital for PET scanning. This was carried out using an ECAT EXACT3D tomograph (model 966, CTI, Knoxville, TN, USA).31,32 The acquisition system of this scanner has a flexible design which can record data in both frame and list mode. List mode acquisition was used in the present study, thus providing efficiency of data storage and high temporal sampling with flexible post-hoc frame re-binning. Emission scanning was performed with an energy window of 350–650 keV. Transmission scanning was performed with a single photon point source (150 MBq of 137Cs, Math MeV, Math years), contained in a small pellet which was driven in a fluid-filled steel tube wound into a helix and positioned just inside the detector ring.

A series of measurements of rCBF were carried out, using H215O as the flow tracer. For each CBF measurement, 6 mCi activity of H215O were administered as a bolus over 160 s, (build-up period of 120 s, infusion for 20 s and flush for 20 s).33 After cannulation of the radial artery and an antecubital vein, patients and controls underwent a series of 12 scans. These comprised 3 different conditions of rest, post-exercise breathlessness and isocapnic hyperventilation each repeated 4 times in a randomised sequence. The duration of each scan including the delay between scans was 8 min:

Run 1Isocapnic hyperventilation;
Run 2Rest 1;
Run 3Horizontal bicycle exercise;
Run 4Isocapnic hyperventilation;
Run 5Rest 2;
Run 6Horizontal bicycle exercise;
[4 min break to download data from PET camera to computer]
Run 7Horizontal bicycle exercise;
Run 8Isocapnic hyperventilation;
Run 9Rest 3;
Run 10Rest 4;
Run 11Horizontal bicycle exercise;
Run 12Isocapnic hyperventilation.

For each 8-min cycle, when the run was bicycle exercise, there was unloaded exercise between t0 and 1.00 min, then 50W exercise between 1.00 and 4.00 min of the 8 min cycle. When the run was isocapnic ventilation, each subject breathed in time with a metronome at the same fR and VT as at the end of the bicycle exercise condition, between 3.00 and 7.30 min of the 8 min cycle. In all conditions, the H215O build-up was between Math and Math min, the infusion between 3.30 and 4.30 min, with the rise noted on the PET camera between 4.20 and 4.25 min, peaking at 4.55 min (Fig. 1). The list mode acquisition was between 2.30 and 7.30 min. Scan acquisition was performed immediately on cessation of cycling because movement artefacts during cycling prevented the acquisition of useful data.

Fig. 1

Acquisition schedule – post-exercise breathlessness.

During the scanning sequence, ∼200 mL of blood (5 mL/min between 3.50 and 7.20 min of each cycle plus a discrete 5 mL sample at 6.50 min) was taken from each subject. During each PET scan, blood was sampled continuously from a radial arterial line for scintillation counting by an on-line bismuth germanium oxide system. This allowed the rCBF measurements to be quantified in absolute units (described below).

In total, subjects had a radiation exposure of 3.54 mSv (12 runs, each equivalent to an exposure of 0.27 mSv, plus 0.3 mSv for the transmission scan).

Monitoring and measurements

During the PET study, patients were monitored repeatedly using: a 12 lead ECG (Marquette), continuous ECG for rhythm (Siracust), blood pressure (Dinamap), respiratory flow rate (ultrasonic flowmeter), end-tidal MathCO2 (Capnograph), arterial oxygen saturation – SaO2 (finger oximeter), a modified Borg scale of perceived breathlessness and distress.

Analysis of PET images

PET images were transformed into a standard stereotactic space. Regional blood flow measurements were corrected for global changes in blood flow and comparisons of rCBF across conditions were performed with the t statistic (more precisely a block design ANCOVA) on a voxel by voxel basis by statistical parametric mapping (SPM96) software (Wellcome Department of Cognitive Neurology, Queen Square).34–37 rCBF changes related to the post-exercise breathlessness runs were compared with isocapnic hyperventilation runs as well as with baseline conditions. These analyses permitted the construction of statistical parametric maps for the description of significant changes in rCBF between the different test conditions. Significant changes were identified by applying a statistical threshold of 0.05, corrected for multiple comparisons.

For the computation of global cerebral blood flow (gCBF), arterial blood was sampled throughout the scanning procedure from the radial arterial line. gCBF measurements were obtained (mL blood/min/mL tissue) by means of least squares fits of total tissue radioactivity using the Kety model.38

Statistical evaluation

Besides the use of SPM for the analysis of the rCBF data, the intra-group respiratory variables, between different conditions, were analysed with a 2 factor ANOVA (two-sided). However, due to concerns over possible correlations between the post-exercise data and the isocapnic hyperventilation data, we compared the respiratory variables between groups for the different conditions using two-tailed paired t tests. The t test was also used to compare age and echocardiographic fractional shortening between the study groups. The statistical comparisons were performed using Statview SE+ Graphics® 4.0 software. Statistical significance was defined as Math.

Ethical considerations

This study was approved by the Research Ethics Committee, Hammersmith Hospital and by the UK Administration of Radioactive Substances Advisory Committee (ARSAC). The investigation conformed with the principles outlined in the Declaration of Helsinki (Cardiovascular Research 1997;35:2–4). All subjects gave written informed consent for their participation in the study.


Patient and control characteristics are presented in Table 1. The only co-morbid illnesses were two previous deep vein thromboses in patient 2 and diverticulosis in patient 5. No control had any intercurrent illness, nor was any taking medication. Lung function tests were within normal limits in all cases. The details of the respiratory parameters are in Tables 2–4.

View this table:
Table 2

Respiratory parameters under resting conditions

PatientEmbedded ImageEmbedded ImageEmbedded ImageCO2HRfREmbedded Image
P (vs Con)0.7690.7810.6220.250.8870.988
ControlEmbedded ImageEmbedded ImageEmbedded ImageCO2HRfREmbedded Image
  • Embedded Image, inspiratory time (s); Embedded Image, expiratory time (s); Embedded ImageCO2, end tidal partial pressure of CO2 (mmHg); HR (beats min−1); fR respiratory frequency (min−1); Embedded Image, ventilation, i.e. expiratory volume×fR (L min−1); Con, Control subjects.

View this table:
Table 3

Respiratory parameters during isocapnic hyperventilation

PatientEmbedded ImageEmbedded ImageEmbedded ImageCO2HRfREmbedded Image
Embedded Image (vs Rest)0.1980.0680.2040.6570.0060.004
Embedded Image (vs Con)0.5150.3530.1900.1650.2980.939
ControlEmbedded ImageEmbedded ImageEmbedded ImageCO2HRfREmbedded Image
Embedded Image (vs Rest)0.0440.2370.0190.9100.0100.021
  • For legend, see Table 2, plus: IH, isocapnic hyperventilation.

View this table:
Table 4

Respiratory parameters during post-exercise breathlessness

PatientEmbedded ImageEmbedded ImageEmbedded ImageCO2HRfREmbedded Image
Embedded Image (vs Rest)0.2570.3600.2260.1940.1830.002
Embedded Image (vs IH)0.5730.1330.5930.1400.2800.042
Embedded Image (vs Con)0.5270.2360.1140.6860.2810.719
ControlEmbedded ImageEmbedded ImageEmbedded ImageCO2HRfREmbedded Image
Embedded Image (vs Rest)0.1830.8720.0360.0030.4450.002
Embedded Image (vs IH)0.3140.2590.5310.0020.2020.079
  • For legend, see Table 2.

Respiratory parameters

NB: Values of SaO2 were above 96% in all subjects at all stages of testing and will therefore not be presented in greater detail.

1. At rest. There were no significant differences between the patients and controls for inspiratory and expiratory times and volumes, MathCO2, heart rate, fR or Math at rest. (Table 2)

2. Isocapnic hyperventilation. The differences between isocapnic hyperventilation and rest for the two groups are detailed in the Table 3. As can be seen, there were no significant differences between the patients and controls for the respiratory variables during isocapnic hyperventilation.

3. Post-exercise breathlessness. There was a non-significant trend for MathCO2 to be lower after exercise in patients (33.4±vs 36.8±mmHg; Math, two-tailed). Otherwise, there were no significant differences between the groups for the respiratory parameters (see Table 4).

Breathing is increased during exercise or hyperventilation and, as expected, the comparisons in Tables 2–4 reflect this.

Comparing post-exercise breathlessness to isocapnic hyperventilation, no differences between groups for the respiratory variables and heart rate were apparent (2-tailed, paired t test), although there were differences in these variables between the conditions (Math for patients for Math, and 0.002 for HR for controls]. No significant differences were found when testing for an interaction between these comparisons (CHF vs Controls and Post-exercise breathlessness vs isocapnic hyperventilation). For details, see Table 4.

Perception of breathlessness

There were no significant differences in the perception of breathlessness between the patients and controls within the test conditions. Both groups felt more breathless during exercise compared to resting conditions (1.99±0.48 vs 0.13±0.16 Math, for patients and 1.29±1.14 vs 0.25±0.61, Math for controls). Both groups also felt more breathless after exercise compared to during isocapnic hyperventilation (1.99±0.48 vs 0.33±0.35, Math, for patients and 1.29±1.14 vs 0.29±0.48, Math for controls). For both groups on 2 way ANOVA, post-exercise breathlessness vs isocapnic hyperventilation, Math.

PET findings

1. Post-exercise breathlessness vs rest. NB: This equates to: {physical effort of respiration+sensation of dyspnoea} compared to rest. Areas of the brain activated in this comparison, in both patients and controls, were the right inferior temporal gyrus (BA 20; 60, −28, −22; Math) and the right anterior insula (BA 45; 26, 14, 2; Math). There were no significant areas of activation found in patients that were not found in controls, nor vice versa (see Table 5).

View this table:
Table 5

Areas of increase in rCBF in controls only for the comparison (post exercise breathlessness vs rest)

AreaxyzZEmbedded ImagexYzZEmbedded Image
Insula/frontal op (BA 38)−2814125.400.001
Cerebellum (vermis)−8−56−245.200.003
Pre-central gyrus (BA 4)−4−26784.950.00820−24845.130.004
Post-central gyrus64−20−244.870.011
Fusiform gyrus34−64−284.750.019
  • Table 5 reports the co-ordinates in the x, y and z axes of the significant rCBF increases for the controls only, for the Comparison {post-exercise breathlessness−rest} with reference to the stereotactic space defined by the atlas of Talairach and Tournoux.36 Statistical magnitudes are expressed as Embedded Image scores (Embedded Image) and Embedded Image values. BA, Brodmann Area, frontal op, frontal operculum.

2. Isocapnic hyperventilation vs rest. NB: This equates to the physical effort of respiration compared to rest. There were significant rCBF increases in common for the patient and control groups with respect to the right lentiform nucleus (28, 12, 4; Math and 38, −6, 6; Math) and left cerebellum (−32, −54, −40; Math) (see Table 6).

View this table:
Table 6

Areas of increase in rCBF, in controls only for the comparison: (isocapnic hyperventilation vs rest)

AreaxyzZEmbedded ImagexyzZEmbedded Image
Pre-central gyrus (BA 6)−600186.300.000682125.220.002
Lentiform nucleus−26−265.550.000
Inf parietal lobule (BA 40)66−42465.430.001
  • Table 6 reports the co-ordinates in the x, y and z axes, of the significant rCBF increases for the controls only, for the comparison {isocapnic hyperventilation−rest} with reference to the stereotactic space defined by the atlas of Talairach and Tournoux.36 Statistical magnitudes are expressed as Embedded Image scores (Z) and Embedded Image values. BA, Brodmann Area; Inf, inferior.

3. Post-exercise breathlessness vs isocapnic hyperventilation. NB: This equates to {physical effort of respiration+sensation of dyspnoea} compared to the physical effort of respiration alone. This main comparison identified the following areas of brain activation to be common to both patients and controls: the left anterior cingulate gyrus (−28, 32, 10; Math), and the right dorsal cingulate (18, −54, 26; Math).

However, when patients and controls were compared directly, the control group exhibited increased rCBF in the right frontal medial gyrus (BA6; 2, −24, 64; Math) and the left pre-central gyrus (BA4; −18, 26, 62; Math), activations which were not found in the patient group. See Fig. 2.

Fig. 2

Regional cerebral blood flow increases for the comparison {Post exercise breathlessness−isocapnic hyperventilation} present in the controls but not in the CHF patients. The activation of the medial post-central gyrus (I) and left superior frontal gyrus (II) in the controls is evident. R, right; L, left; Ant, anterior; Post, posterior. For the stereotactic co-ordinates, see text.

4. Absolute cerebral blood flow in the test conditions. In the patients, there were no significant differences in gCBF among the study conditions (0.51±0.07 mL/min/mL after exercise; 0.51±0.10 mL/min/mL during isocapnic hyperventilation and 0.49±0.09 mL/min/mL at rest, Math). In the controls, gCBF was greater after exercise than after isocapnic hyperventilation (0.40±0.24 mL/min/mL vs 0.35±0.03 mL/min/mL; Math) and greater after exercise than at rest (0.40±0.24 mL/min/mL vs 0.35±0.04 mL/min/mL; Math). There were no differences between the isocapnic hyperventilation condition and at rest. Between patients and controls, there were significant differences (Math) for each of the 3 conditions.


Principal findings of the study

We examined the central neural correlates of control of breathing in CHF and found:

  1. The respiratory parameters were mostly comparable between the 2 groups for the different study conditions. However, during isocapnic hyperventilation and after exercise the patients tended to shorter breathing times, greater respiratory frequency and lower MathCO2;

  2. There was little subjective difference between the group of 6 Class II/III CHF patients and 6 age-matched sedentary controls;

  3. There were areas of brain activation common to both groups – the right lentiform nucleus and left cerebellum in the isocapnic hyperventilation condition and the right inferior temporal gyrus and right anterior insula after exercise. For the main effect (exercise vs hyperventilation), a comparison which equates to {physical effort of respiration+sensation of dyspnoea} compared to physical effort of respiration alone, the controls showed activations not present in the patients, chiefly the medial post-central gyrus and left superior frontal gyrus;

  4. Absolute gCBF did not differ significantly between the study conditions in the patient group but in controls it was significantly greater during post-exercise breathlessness than during either isocapnic hyperventilation or the resting condition.

Abnormalities of respiratory function in heart failure

Impaired ventilatory efficiency (altered VEMath slope)4–6 is an acknowledged feature of CHF patients and may have a number of causes including central factors (e.g., chemoreceptor stimulation by hypoxia, hypercapnia or acidosis)8 and peripheral ones.7 Our CHF patients showed a trend towards greater fR and lower MathCO2 than the controls, although there were no statistically significant differences in the respiratory variables compared to controls.

Perception of breathlessness in heart failure

Breathlessness is a normal experience after excessive physical exertion and is more pronounced in the physically deconditioned. It is also a cardinal symptom of CHF, in which several mechanisms probably contribute to its generation.1,2 Although counter-intuitive, there is little or no relationship between symptoms of CHF and objective indices of function1,2,14 The issue is of obvious clinical importance, with a spectrum of symptoms ranging from excessive, causing debility in patients with well-preserved organ function on the one hand, to lack of an important `early warning system' in patients with significant pathology on the other. However, to date, the role of the CNS in generating such variability of symptom perception has not been studied directly. One of the motives to perform the present study was to address this. We attempted to identify if rCBF, in a particular cerebral area, co-varied with the subjective sensation of breathlessness in the CHF patients. No such area was identified. In fact there was less cerebral activation in the patients than in the controls, although both groups had similar subjective sensations of breathlessness. This subjective similarity may itself be considered rather surprising.

Another situation in which reduced cortical activation is observed is in habituation of the response to an aversive stimulus.39,40 It could be conjectured that in CHF, there is habituation to afferent signals at some level within the neuraxis, which may be a factor in the complex relationship between symptoms and cardiac function in CHF. Concerning the sensation of post-exercise breathlessness, the lack of a difference between CHF patients and controls was rather unexpected. It is possible that the patients lacked a stimulus of sufficient intensity to generate additional foci of activation. However, even if this were the case, the subjective rating of breathlessness of the patients was not less than that of the controls and, furthermore, the haemodynamic responses were equivalent. Thus the principle negative finding of the study (the areas of brain activation found in controls but not patients) still appears to be a significant observation and one which requires explanation.

Functional imaging of the brain in the study of respiratory control

Functional imaging (mainly PET with H215O) has previously been used in studies of breathing regulation in normal subjects during volitional inspiration15 and expiration.16 Subsequently, CO2-stimulated breathing has also been studied,18 with a control condition in the form of passive isocapnic respiration at equivalent fR and VT. Neuronal activation was identified in the upper brainstem, midbrain, hypothalamus, hippocampus, parahippocampus, fusiform gyrus, cingulate area, insula and frontal, temporo-occipital and parietal cortices. Although the main focus of that study was the motor control of breathing, the finding of substantial limbic system activation, is significant in the context of perception of breathlessness, because the CO2 inhalation produced a conscious urge to breathe that was often severe enough to be described as breathlessness.

In a further investigation,17 the increase in breathing during and after right leg bicycle was explored. As well as demonstrating increases in rCBF in the `leg' areas, there were also increases in the superolateral cortical areas bilaterally, previously noted to be activated during volitional breathing. After exercise, only the superolateral areas continued to show increased rCBF; in this study many of the subjects were feeling breathless during the image acquisition because of the high exercise workload. It should be emphasised however that, unlike the controls of our study, the subjects in these studies were mainly young, fit males with an understanding of respiratory physiology.

More recently, Critchley et al., also using PET with H215O, identified the central neural correlates of exercise and mental stress,41 demonstrating rCBF increases in the cerebellar vermis, right anterior cingulate and right insula which covaried with mean arterial pressure. rCBF increases were also found in the pons, cerebellum and right insula which co-varied with heart rate. Decreases in rCBF were reported for the pre-frontal and medial temporal regions. The areas identified were considered representative of the regions involved in integrated cardiovascular response patterns associated with volitional and emotional behaviours.

With respect to the current study's normal subjects, our principal findings are compatible with the above. We found left insular activation during post-exercise breathlessness; activation of this cerebral region was also a feature of Banzett et al., study of air hunger,20 it occurred in Corfield et al., study of CO2-stimulated breathing18 and in Williamson's study of hypnotic sense of effort.23 The cerebellar activation observed in our study corresponds to similar activations in studies of volitional inspiration by Ramsay et al.,16 and Colebatch et al.,15 the vermis activation also features in the studies of Pfeiffer on breathing against a resistive load21 and that of Isaev24 as well as in the Corfield et al., CO2-stimulated breathing study.18 The latter study and Williamson's hypnotic sense of effort study also feature right anterior insular activation as found in our patients and controls.23

With respect to the isocapnic hyperventilation condition in our controls, the activations of the cerebellum and right superior frontal gyrus were also found in the studies of Ramsay,16 Corfield,18 Fink17 and colleagues. For our key comparison (post-exercise breathlessness vs isocapnic hyperventilation) the left superior frontal activation in our controls corresponds to that found in Corfield's CO2 study,18 whilst the left anterior cingulate activation, common to our patients and controls features in the studies of Fink,17 Williamson19,23 and Corfield.18 The most fascinating contrast in the present study, however, is the absence of distinguishing activations among the patients. The observation of no subjective difference in the perception of breathlessness between CHF patients and the controls suggests that the sensation of breathlessness may depend upon different brain mechanisms in CHF from those found in health. Alternatively, or additionally, there may be differences in central command in relation to exercise in CHF.

Autonomic dysfunction in patients with heart failure

Enhanced sympathetic activity is widely recognised as a pathophysiological feature of CHF. Several independent investigative techniques have pointed to abnormal neural regulation in CHF.10,11,42–47 In particular, an assortment of heart rate variability studies have indicated that at different stages of CHF, there are differences in the degree of alteration of neural regulation and the heart's responsiveness to it. However, the precise neurophysiological substrate of such abnormalities, including any potential contribution of the higher centres of the CNS, remains to be elucidated.

The present study is open and observational and its principal value lies in the proposition of hypotheses. It is possible that the apparent absence of cerebral activations in the different conditions in the CHF patients might be related to the reduced heart rate variability and baroreflex sensitivity known to occur in this disease. Unfortunately, we do not have specific data on heart rate variability and baroreflex sensitivity in these particular subjects, so any association remains speculative. It is tempting to hypothesise that one mechanism maintaining the degree of variability of these may be additional intermittent inputs from the cerebral cortex to the brainstem. A further prospective study with detailed autonomic functional assessment is necessary to clarify this.

Cerebral vascular reactivity in heart failure

Cerebral autoregulation is a fundamental physiological response to changes in systemic haemodynamic conditions and is well preserved in a wide range of conditions. However, a number of studies have demonstrated that cerebrovascular reactivity is attenuated in CHF. Paulson and colleagues, measuring cerebral blood flow (CBF) by the intracarotid xenon-133 (133Xe) injection technique48 found mean CBF to be lower in patients with CHF. They also found that CBF was not reduced further by administration of captopril, despite a marked reduction in blood pressure. This effect was interpreted as a shift in the limits of cerebral autoregulation, probably mediated by larger cerebral arteries.49 More recently, Kamishirado et al., measuring CBF by analysing the Patlak-Plot curve obtained from radionuclide angiography, reported an increase in CBF in patients with CHF treated with enalapril, independent of any effect on cardiac output.50 Consistent with standard therapy for CHF, all but one of our patients in the present study were treated with ACE-inhibitors and therefore the relatively greater values of gCBF that they had in all test conditions might be attributable to medical therapy. Other than treating all controls with an ACE-inhibitor, it is difficult to envisage how to remove this potentially confounding factor. However, this does not necessarily account for the lack of variability in gCBF in patients between conditions.

With regard to rCBF in CHF, in a recent paper, a swine model of pacing-induced heart failure was described by Caparas et al.51 CBF was found to be reduced compared to controls both at rest and during treadmill exercise, although there was a significant increase in CBF between rest and exercise in the heart failure swine. Specific regions of blunted increase in perfusion were the parietal and occipital cortex and the supra-pyramidal medulla. To date, however, there have been no direct studies of rCBF in vivo in man in cardiac disease. We employed the technique of least squares fits of total tissue radioactivity using the Kety model.38 This was necessary because the SPM analysis treats gCBF as a co-variate of no interest, so only relative increases in rCBF between conditions can be identified.

Limitations of the study

  1. We did not find a significant difference in subjective breathlessness during exercise. This might be explained, at least in part, by the controls being (consistent with the demographics of heart failure) middle-aged, deconditioned males rather than the more commonly studied young and fit individuals. The workload was also not very demanding. From the perspective of the participants however, the workload was considered `reasonable' in terms of their activities of daily life; their indications were that they would not generally exert themselves beyond this level. This relatively low level of physical stress might have a bearing on the lack of significant differences among the respiratory variables. Furthermore, the number of patients studied is small, all were in NYHA classes II and III and all were probably `low perceivers' in terms of their sensation of breathlessness. On the positive side, the eventual population was tightly characterised and matching was close between the controls and the patients.

  2. It is certainly possible that widening of the age range and inclusion of female subjects would yield different results (we know, for example, that autonomic responses vary between sexes and age groups). A larger study would be desirable, but the study protocol was exacting.

  3. In our quest for a `pure' population of patients with breathlessness due to CHF, those with respiratory disease were excluded. Among these may have been some patients in whom impairment of lung function was a direct consequence of their cardiac disease, e.g., with bronchoconstriction provoked by an exercise-induced increase in end diastolic pressure.

  4. It has to be acknowledged that there has been multiplicity of statistical testing, particularly of the respiratory data; in the presence of small sample sizes, there is therefore a disproportionate risk of false positive results. However, the respiratory data are only a secondary focus of the paper and the brain imaging data are more significant and less susceptible to eccentric outcomes.

  5. It is possible that certain cerebral regions identified were only tangentially related to the conditions studied. Thus, we cannot exclude effects of mental counting during the isocapnic hyperventilation runs, in which subjects set their respiratory frequency to one initially derived from a metronome beat. The pre-central activations may be associated with this.

  6. The lack of a real-time autonomic marker of cardiac dysfunction has caused difficulties in the analysis of data for this study. It is possible that the study was simply too focussed on the respiratory and neuroimaging aspects. No continuous ECG data, from which a useful measure of autonomic activity might have been made, were obtained. The authors hope to rectify this in subsequent work.

  7. It must also be acknowledged that, certainly with respect to deductions made from the gCBF results, there is reliance in the present study on intra-group evaluations. Thus, in the absence of statistically significant differences, the implications of the observation must remain speculative.


We have investigated central neural activity in patients with CHF and in matched controls in conditions of post-exercise breathlessness, rest and a control condition, isocapnic hyperventilation. Our principle finding is the absence of specific cerebral regional activations in heart failure patients, which are demonstrable in controls. As a consequence of this, it is suggested that, in heart failure, the perception of breathlessness and respiratory control may depend upon different brain mechanisms from those in health. We further noted a lack of significant change in absolute blood flow between conditions in patients and speculate that this might suggest a reduction in reactivity of the cerebral circulation. Such altered reactivity might contribute to the generation of symptoms and/or the autonomic dysfunction found in CHF.


This work was supported, in a large part, by the British Heart Foundation, through Dr. Rosen's Intermediate Research Fellowship.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
  30. [30]
  31. [31]
  32. [32]
  33. [33]
  34. [34]
  35. [35]
  36. [36]
  37. [37]
  38. [38]
  39. [39]
  40. [40]
  41. [41]
  42. [42]
  43. [43]
  44. [44]
  45. [45]
  46. [46]
  47. [47]
  48. [48]
  49. [49]
  50. [50]
  51. [51]
View Abstract