Cardiac resynchronisation therapy for chronic heart failure: predicting and measuring ‘response’
- Correspondence to Dr Klaus K A Witte, Division of Cardiovascular and Diabetes Research, Leeds Institute of Genetics, Health and Therapeutics, Multidisciplinary Cardiovascular Research Centre, University of Leeds, Clarendon Way, Leeds LS2 9JT, UK;
- Received 7 December 2012
- Accepted 10 December 2012
- Published Online First 24 January 2013
Chronic heart failure (CHF) is a complex syndrome involving all organ systems, and frequently affects patients with multiple co-morbidities. The condition is characterised by an inexorable, although variable and unpredictable decline. Cardiac resynchronisation therapy (CRT) can improve symptoms and prognosis in CHF patients with left ventricular (LV) systolic dysfunction (LVSD) and conduction delay.1 National and international guidelines describe simple criteria based on data from randomised placebo-controlled studies to help physicians decide which patients should receive CRT.2 Recently the criteria have expanded to include patients with less severe symptoms.
Despite proven benefits on prognosis, CRT is perceived as an irreversible and expensive treatment (mainly because of reluctance to provide CRT without an implantable defibrillator), not guaranteed to improve symptoms. Patients without symptomatic improvement are classed as ‘non-responders’. This term has undoubtedly slowed uptake of CRT and should now be abandoned. In fact CRT is associated with a greater response rate in terms of symptoms than medical therapy,3 and the vast body of literature, mostly obtained from longitudinal datasets describing complex imaging techniques, has merely proven that in a patient with heart failure symptoms and LVSD, the most reliable predictor of ‘response’ is an ECG showing left bundle branch block.4
Determining what constitutes ‘response’ is equally as difficult, as it is in any chronic condition.3 When faced with a patient in whom LV function and symptoms do not improve following CRT and who dies 12 months after the procedure, one might feel that the treatment has failed. On the other hand, this same patient may have had a much more aggressive course without the device, dying much sooner. What is clear is that deactivating the device in patients labelled as ‘non-responders’ leads to an acute deterioration.5
Despite the huge effort in identifying cardiac predictors of clinical response, and measuring it, few datasets have stratified patients into groups based on other features that determine outcome such as diabetes mellitus,6–8 renal function or pre-implant disease duration. This makes the data presented by Verbrugge et al particularly interesting.9 No previous randomised or observational studies of CRT document the duration of symptoms prior to implantation, despite the fact that it seems an obvious risk factor for subsequent mortality. The studies focus instead on markers of severity, such as symptoms and LV function, assuming that the two are closely related. Although the data presented by Verbrugge et al have failings that cannot easily be adjusted for (renal function was worse and there was a higher prevalence of diabetes mellitus in those with longer disease), there does seem to be an inverse relationship between the duration of symptomatic heart failure prior to CRT and subsequent survival. This seems intuitive; if the mean survival of a patient diagnosed with heart failure is 3 years, patients will live longer with CRT if they receive the device at 1 year than if they receive it 2 years after their first presentation. However, the data also give a glimpse of the possible effects of co-morbidities, in this case renal function. The inverse relationship between disease duration and outcome following CRT was not seen in patients with normal renal function. Perhaps the patients with most to gain from early CRT are those in whom the syndrome has progressed as evidenced by deteriorating renal function, while those with severe LV dysfunction but no evidence of collateral organ involvement might have less short-term prognostic benefit. Further research on existing and future datasets is needed in patient groups stratified not only by cardiac function and symptoms but also by co-morbidities such as renal dysfunction, diabetes mellitus, airways disease and pre-existing pacemakers. This might identify not only patients with less severe cardiac dysfunction who could benefit from CRT, but also patients within our current criteria who do not, thereby refining cost and clinical effectiveness.
This unique dataset also serves to stimulate discussion about endpoints. Although event-free survival was worse, and ‘pump failure’ more common, in people receiving CRT later after presentation, this was unrelated to the degree of LV remodelling, rather the presence of renal dysfunction, throwing into question whether LV function is an appropriate endpoint for studies of CRT and possibly CHF as a whole. The data serve to remind us that outcomes in CHF patients have little to do with how the heart looks, but rather the severity of the syndrome as a whole (of which renal function is a powerful part),10 and that determining the mortality benefit of treatment based on arbitrary measures of severity (such as resting LV function, haemodynamic variables and even symptoms) is unreliable. In patients with LV dysfunction and a broad QRS, we should perhaps ignore the lure of complex imaging techniques obtained from longitudinal studies looking at surrogate endpoints, and focus our resources firmly on achieving greater penetration of optimal therapy applied within an integrated care service and implant CRT in patients displaying the simple features that identify a population proven to benefit, while avoiding the cost and distraction of unreliable surrogate endpoints.
Finally, physicians looking after CHF patients need to recognise that CRT has not failed in those individuals who do not feel better following the procedure and they have a duty to make sure that patients and their carers understand this, preferably prior to the procedure. β-Blockers and ACE inhibitors improve outcomes in patients with CHF, yet we do not question our prescription if symptoms or the echocardiogram fail to improve. Why should CRT be any different?
Competing interests None.
Provenance and peer review Commissioned; internally peer reviewed.