CRT-D versus CRT-P
CRT, as a treatment of HF, has been demonstrated the efficacy and safety in patients with impaired LV function and a wide QRS duration [4,28]. In clinical practice, a high proportion of death among HF patients are due to electrical disturbances, which occur suddenly and unexpectedly. ICD therapy is also recommended in those HF patients for primary prevention to reduce the risk of sudden cardiac death and all-cause mortality [7]. Due to the overlapped indication between two therapies, combined treatment (CRT-D) are going to mainstream gradually. Although no specific recommendation was given to clinicians in choice of CRT-D or CRT-P, the majority patients of US underwent implantation of CRT-D [29]. However, the only randomized trial (COMPANION) did not show a significant difference in mortality between CRT-D and CRT-P directly [4]. A recent meta-analysis, which pooled unadjusted relative risk, revealed an association of additional ICD and lower risk of all-cause mortality in CRT patients [5]. Nevertheless, it is not confident to draw this conclusion unless significant differences in clinical characteristics have been fully considered. Moreover, rate of SCD declined substantially among HF patient with reduced ejection fraction due to increasing use of evidence-based pharmacotherapies [30]. It remains conflicting whether additional ICD can still benefit HF patient receiving CRT.
Due to the absence of strong randomized evidences and recent controversial results, we performed this meta-analysis using adjusted HR to minimize the influence of patient selection bias. In included studies, patients undergoing CRT-P are generally older, less often male gender and less ICM, and these differences are all be corrected by Cox multivariate analysis. Our work indicated a similar result which approve the advantage of additional ICD on CRT patients, but the extent of benefit is decreased compared with previous meta-analysis. Due to the lack of guidance on device selection, clinical practice varies widely around the countries and centers. A cause-of-death analysis of patients with CRT by Marijon et al reported that no statistically difference in SCD incidence between CRT-P and CRT-D, where only 13% patients implanted CRT-D in secondary prevention. In our subgroup analyses, no benefit was observed when we restricted CRT-D indication to primary prevention [20]. A large observational study by Barra et al suggested a low absolute risk of SCD among CRT patients [11], and Ruwald et al reported a lower risk of ventricular arrhythmias in responders to CRT [31]. Hence, as the cost add-on and inappropriate shock therapy of ICD was brought to consideration [32,33], it may suggest the unnecessity of ICD for CRT patients who did not experience a ventricular arrhythmia causing haemodynamic instability, especially in super-responders. In all included studies with less than 5 years follow-up duration, CRT-D keeps benefits in mortality and there is small amount of heterogeneity across the studies. Limited by few studies reported long-term outcome and nonnegligible heterogeneity, it is still unclear about the performance of CRT-D on long-term (≥5 years). More studies are needed to illustrate it, which can provide clinical implications to patients with long life expectancy. Although CRT-D did not present an advantage in all-cause mortality in non-European countries, we are more inclined to attribute this difference to inconsistence of indication worldwide rather than racial difference [34].
In recent studies, propensity score matching technique has been used in observational studies to adjust for the influence of differences in the clinical characteristics. With more consistent baseline characteristics between CRT-D and CRT-P group, the pooled effect from four included studies did not indicate a lower mortality in patients with CRT-D. However, only part of studies reported HR after propensity matching, and those studies also concluded insignificant results by using adjusted HR. There may be a certain degree of publication bias at all. In addition, another study including 11 925 CRT-P recipients and 59 534 CRT-D recipients suggested no difference in survival between CRT-P and CRT-D recipients, which used instrumental variable analysis to eliminate the influences of confounders [35]. Different results from different statistic methods offered us an assumption that additional ICD therapy may not benefit CRT patients. As those methods can only eliminate confounders we have known, strong evidence from randomized controlled trials are urgently needed. We are anticipating the result of RESET-CRT clinical trial (ClinicalTrials.gov number NCT03494933), which randomized heart failure patients with LVEF≤35% and a CRT indication to either CRT-D or CRT-P implantation.