Etiologic impact on difference on clinical outcomes of patients with heart failure after cardiac resynchronization therapy: A systematic review and meta-analysis.

OBJECTIVE: To compare long-term clinical outcomes between patients with heart failure due to non-ischemic cardiomyopathy (NICM) and those due to ischemic cardiomyopathy (ICM) after cardiac resynchronization therapy (CRT). METHODS AND
RESULTS: EMbase, PubMed, and Cochrane Library were searched for published studies up to December 2017. Twenty-one observational studies with 12,331 patients were enrolled in the present meta-analysis. The results demonstrated that the all-cause mortality in NICM patients was significantly lower than that in ICM patients (RR 1.37, 95% CI 1.16-1.61). In terms of echocardiographic parameters, NICM patients exhibited statistically significant improvement in left ventricular ejection fraction (LVEF) (MD 2.70, 95%CI -4.13 to -1.28), and a significant decrement in left ventricular end-systolic volume (LVESV) (MD 10.41,95% CI 2.10-18.73) and left ventricular end diastolic diameter (LVEDD) (MD 7.63, 95% CI 2.59-12.68) as compared with ICM patients. No significant difference was observed in the improvement of New York Heart Association Functional Classification (MD 0.05, 95% CI -0.05 to 0.15), pulmonary arterial systolic pressure (PASP) (MD -0.61, 95% CI -4.36 to 3.14), and severity of mitral regurgitation (MD 0.00, 95% CI -0.08 to 0.07) between the 2 groups.
CONCLUSIONS: Our meta-analysis illustrated that patients with HF due to NICM tended to have better clinical outcomes and LV reverse remodeling as compared with those due to ICM. This finding may help clinicians select patients who respond favorably to CRT, though further research is required to clarify the potential confounding factors and underlying mechanisms for this phenomenon.

Introduction

Heart failure (HF) is the final stage of the most common cardiovascular syndrome around the world.[ It is enumerated that about 25% HF patients experienced varying degrees of asynchronous cardiac contraction.[ Cardiac resynchronization therapy (CRT) that aims to correct impaired ventricular electromechanical coupling, reverse structural remolding and create a more uniform distribution of myocardial blood has been established as a cornerstone for drug-refractory HF.[

The American College Cardiology and European Society Cardiology (ACC/ESC) HF guidelines recommend prophylactic implantation and CRT for symptomatic patients with HF in sinus rhythm and a pronged QRS interval despite optimal medical therapy.[ However, there are still about 30% HF patients who failed to respond to CRT.[ Some previous studies suggested that different etiologies of HF might affect the responsiveness to CRT.

There are two relevant meta-analyses reported by Chen et al and Makki et al.[ Chen et al searched Medline, Embase, and Cochrane Library from inception to 2012 and included 14 observational studies with 3463 patients. Makki et al searched several databases up to 2013 and included 6 studies. However, the definition of primary endpoints in these 2 studies is ambiguous. In addition, the statistical results in some previous studies[ are not sufficiently reliable and even controversial due to the lack of rigorous research types, incomplete database indexes, and small sample sizes. The aim of the present meta-analysis is intended to make a more comprehensive assessment on the effectiveness of CRT on HF due to ischemic cardiomyopathy (NICM) and ischemic cardiomyopathy (ICM) by summarizing studies published in related databases, hoping to draw a more reliable conclusion.

Methods

Data source and search strategy

We searched PubMed, Cochrane Library, and EMBASE databases up to December 2017 for evaluating the effect of CRT on clinical outcomes and long-term prognosis between patients with HF due to ICM and those due to NICM. The following medical subject heading terms were used:

HF;

cardiomyopathy; and

CRT.

This search was then supplemented with careful examination of reference lists of identified reports for any relevant studies missed initially. There were no language restrictions. The detailed search strategies are displayed in Figure 1.

Figure 1

Flow diagram showing the study selection process.

Inclusion and exclusion criteria

Studies were considered for inclusion if they met the following qualified criteria:

performed a contemporaneous comparison between ICM and NICM groups in response to CRT (including CRT alone or CRT-defibrillator [CRT-D], but not including implantable cardioverter defibrillator [ICD]alone);

originally reported the primary and/or secondary outcomes;

had more than 30 participants;

had a minimum follow-up period of 6 months; and

reported relative risk (RR) with 95% confidence interval (CI), or provided base-line data that could be calculated.

Studies were excluded if they were

animal experiments, non-original literature, reviews, editorials or case reports; and

data that could not be extracted, calculated, or were not associated with CRT intervention.

Outcome definition

The all-cause mortality rate was considered to be the main clinical outcome during the follow-up period. In addition, the New York Heart Association (NYHA) functional classification was also used as an indicator of clinical outcomes. We assessed the left ventricular (LV) function and size measured by echocardiography, including LV ejection fraction (LVEF), LV end-systolic volume (LVESV), LV end-diastolic volume (LVEDV), mitral regurgitation (MR) severity, and pulmonary arterial systolic pressure (PASP).

Data extraction and quality assessment

Two review authors (Chen JS and Wang J) independently extracted information from included trials using the proforma process piloted on a random sample of papers. Disagreements between the reviewers concerning the decision were resolved by consultation with the third reviewer (Niu XW). We reported details of study design, participants, interventions, mean follow-up time, QRS duration, NYHA functional classification and efficacy outcomes. We also recorded details of relevant therapies provided to the patients. When a trial was presented in an abstract form, we further searched for information on the Internet and checked for the best available resources or publication. Full-text articles were included if they met the study criteria and provided pertinent information on outcomes. Quality assessment was performed by using the Newcastle-Ottawa Scale (NOS). Publication bias was quantified by the Egger's regression for which data from ten or more studies were available.

Statistical analysis

Continuous variables were analyzed using the mean difference (MD) with 95% CI. RR was used for dichotomous outcomes as the confirmatory effect size estimate. A random-effects meta-analysis of the study outcomes was performed with the pooled effect size. The between-study heterogeneity was assessed by the I2 measure. With I2 values of 50% or less, heterogeneity was acceptable referring to Cochrane handbook and in the case of a high level of heterogeneity with an I2 value of 50% or larger. We performed a sensitivity analysis by comparing the results of meta-analysis of included studies with the results of the remaining studies after elimination of low-quality studies. All analyses were made using the R software. A P value <.05 was pre-specified to indicate statistical significance.

Results

Baseline characteristics of the included studies

Twenty-one studies [ involving 12,331 patients (5736 ICM and 6595 NCM) met the inclusion criteria for meta-analysis. Most of these studies were prospective trials in nature, and 6 were retrospective trials. The length of follow-up ranged from 6 to 48 months (median 17.7 months). Age distributions of both groups were the same. Male patients in ICM group accounted for 83% and 66% NICM group. Most patients recruited to the identified studies were in NYHA class III and IV with LVEF <35%. The average QRS interval of the two sets was greater than 150 ms. The application rate of diuretics fluctuated from 74% to 100%. The use rate of angiotensin converting enzyme inhibitor (ACEI) was 70% to 96%. The median of the clinical application rate of β-adrenergic blockade and aldosterone receptor antagonists was 76.6% and 80% respectively. The characteristics of the included studies and the associated patient characteristics are summarized in Tables 1 and 2. Details of study quality assessment are shown in Table 3. The median of NOS scores was 8. The detailed scoring processes are reported in Supplementary 4..

Table 1

Characteristics of the studies included in this Meta-analysis.

Table 2

Characteristics of the patients enrolled in this Meta-analysis.

Impact of etiologic differences on the clinical outcome

All-cause mortality

After exclusion of 1426 patients whose primary endpoints were not available, 10905 patients were analyzed for the endpoint of all-cause mortality. During a 12-month follow-up period, the pooled analysis of observational studies showed that patients in ICM group had a greater risk for all-cause mortality than patients in NICM group (pooled RR = 1.37, 95% CI = 1.16–1.61) (Fig. 2). Test of heterogeneity (I2 = 38%, P = .01) with random-effect model was acceptable.

Figure 2

A forest plot for all-cause mortality.

NYHA classification

We extracted data from 6 trials, totaling 1234 patients with ICM and 1248 patients with NICM. Comprehensive results of 6 observational studies showed no significant difference between the 2 groups when the NYHA classification was used (MD 0.05, 95%CI -0.05 to 0.15)(Fig. 3).

Figure 3

A forest plot for New York Heart Association. SD = standard deviation.

Impact of etiologic differences on echocardiographic outcomes

LVEF

Thirteen studies comprising 3925 patients performed echocardiography 6 months after CRT to ascertain whether the efficacy and effectiveness of CRT was affected by the underlying HF etiology. The LVEF improvement in NICM group was better than that in ICM group (MD -2.70, 95% CI -4.13 to -1.28). There existed heterogeneity (I2 75%, P <.01) with random-effect model (Fig. 4).

Figure 4

A forest plot for left ventricular ejection fraction. SD = standard deviation.

LVESV

In 9 studies, CRT was administered for more than 6 months, totaling 2998 patients with echocardiographic changes in a reduction in LVESV. The risk of prolonged (> 6-month) administration of CRT was higher in ICM patients than that in NICM patients (MD 10.41, 95% CI 2.10–18.73). Heterogeneity across trials was acceptable (I2 36%, P = .13) (Supplementary Fig. 5).

Figure 5

A forest plot for left ventricular end-systolic volume. SD = standard deviation.

LVEDV

Nine studies involved the research on LVEDV. They reported that ventricular function in NICM group was better than that in ICM group (MD 10.41, 95% CI 2.10to 18.73) (Supplementary Fig. 6).

Figure 6

A forest plot for left ventricular end-diastolic volume. SD = standard deviation.

MR severity and PASP

As shown in Figures 7 and 8, there was no significant difference in MR severity and PASP between the 2 groups (MD 0.00, 95%CI -0.08 to 0.07 vs MD -0.61, 95% CI-4.36 to 3.14).

Figure 7

A forest plot for MR severity. SD = standard deviation.

Figure 8

A forest plot for PASP. SD = standard deviation.

Publication bias and sensitivity analysis

We did not observe significant bias based on the Egger regression (P = .69). We also conducted a sensitivity analysis of the results of significant heterogeneity (LVEF) to investigate their latent sources and evaluate the robustness of these outcomes. After eliminating each of the included studies 1 by 1 to each outcome, we found that Gasparini et al might be the sources of heterogeneity for LVEF, and heterogeneity of the pooled data analysis decreased significantly after excluding that study (I2 = 45, P = .05). In addition, study exclusion may also affect the pooled analysis (pooled MD -2.18, 95%CI -3.23 to -1.13).

Discussion

The results of the present meta-analysis suggest that NICM patients are associated with a greater reduction in the primary clinical endpoint of all-cause mortality but are comparable to the secondary clinical endpoints including NYHA symptomatic class as compared with ICM patients. With respect to echocardiographic outcomes, NICM patients tended to obtain significant reverse LV remodeling compared with ICM patients treated with CRT.

CRT has been shown to improve prognosis (all-cause mortality) and cardiac function in HF patients. However, there is a significant discrepancy in the utilization of CRT between ICM and NICM case, indicating that the impact of CRT on symptoms, quality of life, morbidity, and mortality is similar between patients with and without ICM such as MIRACLE, Zweerink, and CARE-HF.[ This disparity can also be found in our study (47% vs 53%). It was found in our study that NICM patients obtained a significant reduction in all-cause mortality compared with ICM patients. Pooled analysis by Chen et al who assessed etiologic differences in response to CRT showed that NICM acquired a statistically significant greater reduction in the risk of mortality or HF hospitalization.[The MADIT-CRT study involving symptomatic ICM and NICM patients showed a significant difference in response to CRT-D, suggesting that risk of assessment for CRT-D should be etiology-specific.[

The reasons behind these differentials remain unclear, though potential explanations have been presented. First, the present study showed that the presence of myocardial scar tissues is a predictor of poor responsiveness,[ which might affect the results of our meta-analysis. However, no study reported data regarding the location and the size of the infracted myocardium (total scar burden) which is important for response to CRT, so we were unable to perform subgroup analysis. Second, the incidence of metabolic syndrome, cerebrovascular disease and renal insufficiency in ICM patients is high. These factors may indirectly affect the long-term prognosis of patients after CRT. It cannot improve the hemodynamic state of patients with HF patients.[ Data from our study also support this interpretation. Our study also showed a significant difference in the occurrence of diabetes mellitus between ICM and NICM patients (31% vs 17%).

This study also demonstrated that NICM obtained greater benefits from CRT in the secondary endpoint in LVESV and LVESV, most probably due to inexorable progression of ischemic disease. However, no significant difference was observed between NICM and ICM patients in the other echocardiographic outcomes such as MR severity and PASP. On the one hand, we only discussed the improvement of PASP 6 months after CRT in HF patients due to ICM and NICM. The REVERSE study showed that LV remolding and symptom benefits from CRT sustained 12 months in HF patients.[ There are insufficient data to explore the improvement in PSBP after longer follow-up periods. On the other hand, studies have shown that the effect of CRT in improving the degree of MR is limited.[ Severe LV dilatation, irreversible MR and extremely severe regurgitation may be the reasons why CRT was unresponsive in these studies. Hence, longer follow-up observations to obtain more accurate ultrasonic parameters are required to see whether NICM patients could also benefit more from CRT in terms of the MR severity and PASP in the long run.

Other clinical studies have tried to elucidate the mechanisms underlying the advantages of NICM patients in response to CRT during the follow-up period.[ Some researchers found that NICM patients seemed more likely to experience death from pump failure, while ICM patients were more likely to experience sudden cardiac death.[ This provides a potential explanation that NICM patients might derive more benefits from CRT, and male patients might probably obtain more survival benefits from the use of CRT-D. In addition, a recommended dose of ACEI and β-adrenergic blockade after CRT is the decisive factor in improving the mortality and hospitalization rate of HF patients.[ In our study, the application rate of ACEI was different (NICM 87% vs ICM 83%), which may also be a potential factor affecting the prognosis of patients.

This meta-analysis provides new clues to support the hypothesis that NICM patients could obtain better clinical benefits from CRT than ICM patients, suggesting that different etiologies of HF may affect the response to CRT. To improve the symptoms and reduce the morbidity of cardiomyopathies including HF, it is reasonable to recommend that CRT should be considered as a priority in NICM patients with sinus rhythms, an extended QRS duration, LBBB QRS morphology, and left bundle branch block with LVEF≤35% despite optimal medical therapy. In addition, appropriate amendments in the currently available guidelines about the use of CRT seem necessary by considering the impact of etiologic differences on CRT performance in selected patients.

Highlights and limitations

This meta-analysis is a summary of evidence from cohort studies published until 2017 with regard to response to CRT between ICM and NICM patients by setting up explicit inclusion and exclusion criteria during the integration of the literature to improve the stability of the results of the study. Meanwhile, data were extracted by two investigators independently and closely, and any disagreement was resolved by discussion with a third opinion so as to reduce the occurrence of migration. The number of participants included in the study was three times that of the previous ones. Meanwhile, there are a few methodological shortcomings. First, some observational studies included in this meta-analysis treated the patients in a non-random way, which may confound the comparison between primary and secondary outcomes. In addition, different loss to follow-up is also a concern in the meta-analyzed cohorts, knowing that dropouts are more likely to occur in patients at higher risk of ICM, which may induce a selection bias in comparison of changes in LVEF and LVESV because of information censoring.

Conclusion

Overall, NICM patients may obtain more beneficial effects from CRT than ICM patients with respect to the clinical and echocardiographic outcomes. Larger randomized controlled trials and long-term follow-up observations are necessary to clarify the potential association between the etiology of HF and reactivity after CRT.

Author contributions

Conceptualization: jianshu chen.

Data curation: fengmei Chen.

Formal analysis: jianshu chen, fengmei Chen.

Investigation: jianshu chen.

Methodology: jianshu chen, Xiao-Wei Niu.

Project administration: jianshu chen.

Resources: jianshu chen.

Software: jianshu chen, Xiao-Wei Niu.

Writing – original draft: Ya-Li Yao, jianshu chen.

Writing – review & editing: Ya-Li Yao, jianshu chen.