Author + information
- Sern Lim, MD⁎ ()
- ↵⁎University Hospital Birmingham NHS Trust, Edgbaston, Birmingham B15 2TH, United Kingdom
Hsu et al. (1) reported on the predictors of “super-response” (defined by ejection fraction) to cardiac resynchronization therapy (CRT) and the association with morbidity and mortality. By design, the study required paired echocardiographic studies (baseline and 12 months), which inevitably led to the exclusion of 337 (31%) patients randomized to CRT-D in the MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy) study (2). The authors rightly acknowledged the potential for selection bias in favor of CRT, preferentially “selecting for a super-response group.” However, it is the exclusion of patients who may have perished within the first 12 months (perhaps due to adverse response to CRT) that is of more concern, as the adverse effects of CRT may be well underestimated.
There were 74 deaths in the 1,089 patients (6.8%) randomized to CRT in the MADIT-CRT study over an average follow-up of 2.4 years (29 months) (2). This is comparable to the 53 deaths in the 731 patients (7.3%) randomized to the implantable cardioverter-defibrillator (ICD)-only group—an annualized mortality rate of about 3%. There were 25 deaths in the subgroup of 752 patients in the current analysis by Hsu et al. (1). The majority of deaths (48%) occurred in the “hypo-responders” (12 deaths in 190 hypo-responders, or 6.3%) over a median follow-up of 15.2 months following the 12-month echocardiographic study (i.e., from 12 to 27 months of follow-up)—almost double the annualized mortality rate in the ICD-only group in the MADIT-CRT study (assuming linear mortality rates).
However, even this doubling of mortality rates may have underestimated the adverse effects of CRT in hypo-responders. By the process of deduction, of the 74 deaths in the MADIT-CRT study, only 25 deaths were accounted for in the current study (1). Hence, the other 49 deaths must have occurred either in the 337 patients who were not included in the current analysis, and/or in the last 2 months of follow-up (27 to 29 months) in the 752 patients included in the current study. That the majority of deaths (49 of the 74 deaths) may have occurred in the minority of patients (n = 337) would infer either a much higher mortality rate among hypo-responders (if the proportion of hypo-responders is similar in the 337 patients) and/or a much higher proportion of hypo-responders (if mortality rate among hypo-responders is similar to the current report). I would urge the authors to present more data on these deaths to clarify the risk of CRT among hypo-responders.
Finally, the data presented by Hsu et al. (1) indicate that hypo-responders are more likely to have right bundle branch block and shorter QRS duration (<150 ms), which is consistent with other published reports (3,4). As physicians compelled by the oath of primum non nocere, and facing mounting evidence suggesting limited benefit of CRT in patents with these characteristics (5), should we not be asking the next obvious question: if there is no evidence of benefit, is there evidence of harm?
- American College of Cardiology Foundation
- Hsu J.,
- Solomon S.D.,
- Bourgoun M.,
- et al.