Response to (Resynchronization) Therapy in Chronic Heart Failure: Time for a Different Approach

Although chronic heart failure (CHF) remains an incurable syndrome of exercise intolerance, cardiac dysfunction and reduced longevity, the last two decades have seen an unprecedented improvement in the quality and quantity of life for patients diagnosed with CHF caused by left ventricular systolic dysfunction (LVSD). One of the major advances has been cardiac resynchronization therapy (CRT), which can improve symptoms and prognosis in CHF patients with LVSD and conduction delay.1 Guidelines describing the criteria for selection of CHF patients for CRT are based upon large randomized, placebo-controlled studies demonstrating reduced hospitalization and mortality (over a finite time).3 4 In addition to these studies, there are hundreds of smaller, mostly observational, studies using multiple imaging techniques and various measures of response to try to identify subgroups of patients more or less likely to benefit from CRT. Thankfully, none of the proposed pre-assessment deselection techniques have been adopted into international guidelines, and the indications remain resolutely a broad QRS, left ventricular systolic dysfunction, and sinus rhythm. The data presented by Versteeg et al.2 contribute to the increasing recognition of the pointlessness of trying to predict response to CRT from baseline variables, with the only predictor of symptomatic response being a broad QRS complex. Furthermore their data show that changes in echocardiographic variables are unrelated to changes in symptoms following CRT. These unique data should not only stimulate discussion about the use of contemporary measures of response (often based upon arbitrary percentages of change from baseline) and other surrogate outcomes in chronic disease, but also question the use of cohort studies to deselect subgroups of patients previously included in prospective randomized placebo-controlled trials. The response of any individual to any intervention for a chronic disease is variable and CHF is no different. Cardiac resynchronization therapy is associated with a greater symptomatic response than medical therapy,5 yet many patients experience no improvement. Versteeg et al.2 show us that improvements in symptoms and improvements in cardiac function tend not to occur together (only 30% have a response in both, although 80% have a response in either domain), but should this concern us? In the race to select patients with greatest ‘benefit’, it is often forgotten that stability or a slowing of deterioration is a powerful effect of modern treatment for CHF but is impossible to observe from non-randomized longitudinal cohort studies. Cohort studies are also unable to address the ‘ceiling effect’ commonly seen in older patients with co-morbidities and in younger patients with mild disease. In both situations, patients with an excellent clinical outcome are labelled as ‘non-responders’. For example the data by Versteeg et al.2 demonstrate the ceiling effect in patients with less severe disease, where those with smaller heart dimensions are more frequently classed as non-responders. In a chronic disease of uncertain course, only a randomized placebo-controlled trial offers clarity in terms of predictors of outcome, as, rather than looking at the experiences of individuals, one can compare outcomes in groups of patients displaying similar features but exposed to different treatments. Without this one cannot model what would have happened to the patient had they received different treatment. Such an analysis from the CARE-HF (Cardiac Resynchronization in Heart Failure)6 study showed that not only do baseline symptoms not relate to mortality benefit, but powerful markers such as baseline natriuretic peptides, and the presence of mitral regurgitation also do not predict the benefit of CRT on hard outcomes.7 If we are to use surrogate measures in evaluating a treatment, we need to adopt a patient-orientated approach. Is an arbitrary degree of improvement in a surrogate endpoint such as cardiac function relevant to patients? On the other hand, although symptoms are important to patients, and their severity,8 9 but not nature10 are related to outcome, we cannot assume that a lack of change following a particular treatment implies that it has no effect on outcome. Equally, we must not assume that improving symptoms11 has a beneficial (or even neutral) effect on prognosis.12 13 Rather than using the data from Versteeg et al.2 to begin a costly and distracting search for more obscure predictors with which to deselect patients currently indicated for CRT, we should use them to support an open and frank discussion with patients and their carers, preferably before the implant, about what CRT might and might not do for them. Based upon large randomized studies, the procedure extends average longevity by a year whether symptoms or cardiac function improve or not, and although many patients do feel better as a result of CRT, it seems that at the very least, patients are likely to be less symptomatic at any given time-point in the future than they would have been without the device (Figure 1). Crucially, we must abandon the term ‘non-responder’ for all treatments for CHF including CRT. Beta-blockers and angiotensin-converting enzyme inhibitors improve outcomes in patients with CHF, but we do not question our prescription if symptoms or the echocardiogram fail to improve, yet we label such patients as ‘non-responders’ following CRT. Chronic heart failure remains a chronic incurable disease and failure to improve does not mean failure of treatment. If we explain this to patients at an early stage, we will avoid the common pitfall of setting ourselves, our patients, and their carers flawed and clinically irrelevant targets, which when not achieved can have a negative effect everyone. None.