Highlighting the R in CRT

Abstract

Treatment of congestive heart failure by artificial pacing first attracted attention in the early 1990s when reports showed that standard right ventricular apical pacing with a shortened atrioventricular delay improved heart function in a subgroup of patients.1 However, few seemed to benefit from this approach,2 whereas by altering the pacing site to the left ventricle and using biventricular stimulation in patients with intramyocardial conduction delay, one achieved more consistent functional improvement.3,4 The short-term response was greater in patients with a wider QRS complex, typically a left bundle branch pattern,5 and subsequent clinical trials that showed biventricular pacing improved morbidity6,7 and mortality8 principally selected only those patients with a prolonged QRS duration. Article p 1440 However, the story was never that straightforward, as investigators found that the correlation between basal QRS duration and immediate mechanical responses to biventricular pacing was poor5 and not predictive of long-term outcome.9 QRS narrowing after cardiac resynchronization therapy (CRT) was not useful either.10,11 Furthermore, patients with a narrow QRS but mechanical dyssynchrony also benefited from CRT. This situation posed a problem because it was clear from the outset that identification of the right patients was important given the invasive nature and expense of CRT. In an early attempt to predict short-term response, we found mechanical dyssynchrony (assessed with magnetic resonance imaging) or the combination of QRS duration with functional data (basal dP/dtmax) better predicted short-term responders.5 Shortly after, echo-Doppler imaging methods were developed to assess dyssynchrony, and multiple groups found that these methods better predicted long-term CRT response.9,12–14 Just what constituted a CRT response had to be defined and varied from an end-systolic volume (ESV)13 reduction or increased ejection fraction12 to clinical improvement.14 One might think the plot would end here; …