Fatigue and dyspnoea in heart failure: insights from two large randomised clinical trials

Heart failure is a complex clinical syndrome characterised by typical symptoms (like dyspnoea, fatigue, palpitations or chest pain) and signs (like oedema, pulmonary crackles, displaced apex beat and increased jugular venous pressure). The possible importance of symptoms as predictors of subsequent outcomes has received little attention in the medical literature yet is clearly of great potential clinical importance (for example in identification, monitoring and treatment of high risk patients). Fatigue and dyspnoea are the two most prevalent symptoms in patients with heart failure ranging from 50-91% for fatigue and similar (53-89%) for dyspnoea. However, the underlying pathophysiological mechanisms of dyspnoea and fatigue in heart failure remain unclear. It has been proposed that decreased oxygen delivery to muscle due to an impaired pump function of the failing heart leads to a build-up of anaerobic metabolic products which may account for both symptoms. Some hypotheses attribute the impaired oxygen delivery to muscle to reduced blood flow due to persistent vasoconstriction and endothelial dysfunction, rather than just to a limited cardiac output. Other potential mechanisms include abnormalities in muscle metabolism, possibly due to changes in cellular subtype, which limit the ability to utilise oxygen and a miss-match between energy requirement and energy production. It has now been recognised that disturbances of central hemodynamic function are no longer the major determinants of exercise capacity in patients with heart failure. If central hemodynamic parameters are improved, there is no immediate change in symptoms, which points to an impaired ability of the muscle to extract oxygen, leading to dyspnoea. The lack of consensus and understanding of the pathophysiological mechanisms of heart failure symptoms, together with poor and subjective tools for their measurement has led to a delay in the development of effective symptomatic treatment. This in turn may have important prognostic implications such as decreased quality of life, increased hospital admissions and even increased mortality. The aim of this work was to examine the correlates of symptoms and change in symptoms. Additionally I set out to examine the association between symptom severity (at baseline and the change in symptom severity over 6 months) and clinical outcomes (namely heart failure hospitalisation, cardiovascular death and all-cause mortality) after adjustment for a series of other known prognostic factors. A cohort of 3830 men and women with LVEF (left ventricular ejection fraction) ≤35% who participated in the Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA) was examined. This population was chosen because the trial medication (rosuvastatin) had no effect on the primary outcome (composite of death form cardiovascular causes, nonfatal myocardial infarction and nonfatal stroke) (HR 0.92; 95% CI 0.83 to 1.02; P = 0.12) or death from any cause (HR 0.95 in the rosuvastatin group 95% CI, 0.86 to 1.05; P=0.31) compared to placebo, meaning that any result I obtain is unlikely to be due to an effect of the trial drug and because this cohort would be representative of a population with heart failure who were well treated with contemporary evidence-based medicine CORONA was a multicentre, randomised, double-blind, placebo-controlled study which enrolled a total of 5011 patients aged ≥ 60 years with symptomatic (NYHA class II-IV), systolic (LVEF ≤ 40% but no more than 35% in patients with NYHA class II) heart failure. Patients were randomised to receive 10 mg of rosuvastatin or matching placebo once daily. Symptoms were measured at baseline (randomisation visit), 6 weeks after randomisation, and 3 monthly thereafter in this trial population. (1) Investigators were asked to evaluate symptoms using the following statement: “State symptoms during the past few days: Tick lowest level of physical activity causing symptoms”. Fatigue “during the past few days” was measured using a five-point exertion scale (0 none, 1 heavy exertion, 2 moderate exertion, 3 slight exertion, 4 rest), recorded by the investigator. Dyspnoea “during the past few days” was measured using a four-point exertion scale (1 heavy exertion, 2 moderate exertion, 3 slight exertion, 4 rest); a four- rather than five-point scale was used for dyspnoea because the presence of dyspnoea at baseline was an inclusion criterion for CORONA. Data were analysed in several ways to comply with the objectives of this thesis. I examined prevalence and severity of fatigue and dyspnoea by using descriptive statistics. I also analysed baseline characteristics (at visit prior to randomisation and randomisation visit) according to fatigue and dyspnoea severity, reporting means and standard deviations for continuous variables (medians and interquartile ranges for variables that were not normally distributed) and percentages for categorical variables and comparing across symptom groups by running appropriate tests. Ordered logistic regression was used to examine which baseline characteristics were independently associated with symptom severity at baseline, while Cox proportional hazards regression was used to examine how symptoms were related to the risk of clinical events. I used multinomial logistic regression to identify independent predictors of change in symptom severity from baseline to the 6 month visit (chi2 was used to obtain p values), classifying patients as showing a decrease (reduction in score), an increase (an increase in score) or no change (unchanged score) in symptoms and analysed the relationship between change in symptoms and subsequent clinical outcomes using Cox regression. Finally, I examined the effect of rosuvastatin treatment for six months on symptom severity using Cox regression survival analysis. Additionally, a cohort of 8399 patients with chronic symptomatic heart failure with reduced ejection fraction from PARADIGM-HF was examined. Dyspnoea and fatigue on effort in PARADIGM-HF were recorded in every visit as “present” or “absent”. I found that at baseline 95% of CORONA trial participants reported some level of fatigue on exertion and most of them (85%) reported high symptom severity (from moderate exertion to symptoms at rest). In PARADIGM-HF 52% reported fatigue on effort. Dyspnoea showed a similar pattern, although some level of dyspnoea was an inclusion criterion for CORONA where 91% reported dyspnoea from moderate exertion to dyspnoea at rest, while 86% reported dyspnoea on effort in PARADIGM-HF. I found that a limited number of variables (history of hypertension and coronary heart disease; NYHA functional class; and use of mineralocorticoid receptor antagonists) were independently associated with both fatigue and dyspnoea (only with fatigue for PARADIGM-HF), with no variables clearly associated with only one of these symptoms. This similarity in variables associated with each symptom and the lack of association of dyspnoea with ejection fraction or NT-proBNP suggests that “peripheral” (i.e. changes in muscle bulk and metabolism), rather than “central” mechanisms may explain the origin of both symptoms. I also found that worst baseline symptom severity is strongly associated with adverse clinical outcomes, but this association is lost after adjustment for other well-known cardiovascular prognostic variables like NT-proBNP, LVEF and NYHA class, in both cohorts. However in CORONA, change in symptom severity after 6 months was strongly associated with clinical outcomes, even after adjustment for the previously mentioned prognostic factors; with a decrease in symptom severity proving to be protective while an increase over six months being associated with a higher risk of CV death, HF hospitalisation of all-cause mortality. Statin treatment had no convincing effect on symptom severity. In conclusion, I found that both fatigue and dyspnoea were highly prevalent in both cohorts and that they seem to have the same baseline correlates. This supports the theory that both symptoms might be different expressions of the same pathophysiological process. Change in symptom severity after 6 months seems to be strongly associated with outcomes independent of other known prognostic factors, which shines a light on the importance of prompt and targeted interventions to improve symptom severity, or at the very least to prevent deterioration