Cardiac resynchronization therapy (CRT) is an important device-based, non-pharmacological treatment for chronic heart failure (HF). The multiple benefits of CRT in selected HF patients under optimized pharmacologic therapy include improvement in symptoms and quality of life, left ventricular (LV) remodeling and decreased mortality and hospital admissions for HF, and have been established by multiple large trials,1–5 leading to its recommendation in current guidelines.6 An important feature in HF is the presence of atrial fibrillation (AF), the arrhythmia most frequently associated with HF, which affects up to 45%-50% of patients, depending on the severity of HF.7,8 For HF patients still in sinus rhythm (SR), the annual incidence of AF is approximately 5%.9 AF is negatively related to prognosis, although some authors do not consider it an independent prognostic predictive factor after correction for age and comorbidities.10 Atrial arrhythmias, if not appropriately managed, may have a negative impact on the clinical benefits of CRT,11 since, in AF patients, CRT can only correct intra- and interventricular dyssynchrony. CRT is also hampered by high intrinsic ventricular rates and irregularity, leading to reduced capture, fusion and pseudo-fusion, and hence less effective biventricular pacing.12

Although the evidence from large randomized controlled trials is weak,13,14 and some authors have argued that HF patients in AF may respond less well to CRT,15–20 the European Society of Cardiology (ESC) guidelines recommend that this therapy should also be used for AF patients, as long as atrioventricular (AV) junction ablation is added in patients in whom continuous biventricular pacing is lost.6 Recently, the CERTIFY study21 showed that long-term survival after CRT among patients with AF and AV junction ablation is similar to that observed among patients in SR, and that mortality in AF patients treated with rate-reducing drugs is higher. Whether CRT is effective in the context of AF is still an important question to be addressed. The purpose of the present study was to prospectively assess the response to CRT in HF patients with permanent AF compared to those in SR.

In the present study, patients with permanent AF, one third of the study population, showed good response to CRT in the majority of cases without the need for AV junction ablation, in contrast to several previous studies.19

To date, randomized studies on CRT have been mainly restricted to patients in SR, excluding patients in permanent AF. This contrasts with the high prevalence of CRT use in AF patients in routine clinical practice, as observed in our data (35%) and as previously indicated by ESC surveys,23,24 which indicate a prevalence of 23%.23

It is well known that the likelihood of coexistent AF and HF is strongly related to the severity of the disease represented by NYHA functional class: prevalence of 5% for NYHA class I, 10%-25% for class II/III, and 50% for class IV.10 The probable reason for the higher prevalence of AF observed in our study population is their greater clinical severity (95% NYHA class III/IV, 5% class II) compared to patients in the European cardiac resynchronization registry (78% class III/IV, 22% class I/II).23 In our sample, AF patients were older, with worse functional capacity, as in previous studies, which may have influenced the effects of CRT in AF compared to SR patients.25,26 Our data are also in accordance with the greater proportion of men among AF patients, as previously demonstrated,27 and additionally showed a higher percentage of nonischemic cardiomyopathy.

Regarding the management of our patients, which followed the consensus for mandatory continuous biventricular capture, it assured rate control and rhythm regularization, in order to maximize the clinical benefit and improve prognosis of patients with permanent AF.28 This requires AV junction ablation in some cases, since pharmacological treatment may be inadequate to control ventricular rate at rest and during exercise. In the second ESC CRT survey,24 74% of European centers implanting CRT devices scheduled AV junction ablation only in the presence of uncontrolled HR, and only a minority of centers (11%) proceeded directly to AV junction ablation, regardless of HR. Frequent biventricular pacing, as previously defined, was taken as ≥95% of biventricular capture.29 In our study, only 5.7% of AF patients underwent AV junction ablation during the first six months after implantation, the other 94.3% being treated with pharmacological therapy for HR control. The low proportion of patients needing AV junction ablation could be explained by good pharmacologic HR control with adequate dosages of beta-blockers, digoxin and other drugs with bradycardia effects, carefully titrated and increased whenever needed. Furthermore, some of these patients underwent CRT implantation due to left ventricular dysfunction, heart failure and indication for pacing due to bradyarrhythmia. However, there was no control group to compare these effects, since the other group was made up of SR patients, who might not need the same drugs or the same dosages. At this point, it is important to note that pre- and post-CRT baseline HR and maximum HR during CPET were not significantly different between SR and AF patients.

Concerning the effects of CRT in this real-life study, several clinical, echocardiographic and functional variables of exercise testing changed significantly after CRT in both SR and AF patients: NYHA class, LVEF, LVESV and VE/VCO2 slope (at three months), and LVEDV and VE/VCO2 (at six months). Similarly, some authors have shown that CRT in AF patients improves symptoms,14,30 while others suggest that CRT is only effective after AV junction ablation, which was certainly not the case in our patients. In contrast to our results, some previous studies in AF patients demonstrated that, despite similar changes in LVEF, there was less improvement in NYHA functional class.25,26,31

An important issue is the CRT response rate, bearing in mind that definitions of CRT response in the literature differ widely.32 It is interesting to observe that neither SR nor permanent AF significantly influenced the percentage of responders in this sample, either clinical (SR 78.6% vs. AF 80.7%) or echocardiographic (SR 77.4% vs. AF 82.1%). Some studies, however, have shown different results. In a meta-analysis by Wilton et al., which included 33 studies (7495 patients), a lower rate of CRT response was observed in AF patients than in SR, with no response in 34.5% vs. 26.7%, respectively (p=0.01).26 Also, a more recent study33 confirmed the benefit of CRT in HF patients with AF, although inferior to that of SR patients, with more frequent non-response. In contrast to the results of these two publications and to our own data, single-center randomized studies have demonstrated little evidence regarding CRT effectiveness in AF.34 The RAFT study35 included more patients with permanent AF than all other published studies combined. RAFT failed, however, to demonstrate a clear improvement in any clinical or surrogate outcome with CRT in patients with permanent AF, despite a trend for fewer HF hospitalizations. This poor outcome might be attributed to suboptimal delivery of CRT, because only one third of patients received ≥95% ventricular pacing.35 It should be noted that in this study many patients were in NYHA class II. To increase the percentage of pacing in AF, AV junction ablation was applied in our study, increasing CRT response.26

In the present study, there was a significant mean increase in LVEF at three and six months in both groups, which was not statistically greater in SR patients. Despite the significant decrease in both groups in mean LVESV and LVEDV (only at six months), other authors found less improvement in LVESV in AF patients.26 Mean LVM and LAV decreased significantly at six months, but only in the SR group. It is known that HF facilitates atrial remodeling, which promotes the development and maintenance of AF,36 explaining the larger LAV in AF patients. LAV was smaller in SR patients, and consequently the changes were less marked, which facilitated reverse remodeling after CRT. We may hypothesize that LA reverse remodeling and more profound LV reverse remodeling take longer (more than three months) until a significant change is achieved after CRT, and are probably positively influenced by the presence of SR.

On the other hand, regarding the functional effects as assessed by CPET, although VE/VCO2 slope decreased significantly in both groups, only patients with AF had a significant increase in CPET dur (an absolute increase three times that of SR) and also in VO2max three months after CRT. Interestingly, after adjusting for age and baseline VO2max (which was higher in SR and lower in AF patients), there was no longer a statistically significant improvement in VO2max in AF patients (multivariate linear regression, p=0.083). This shows the importance of baseline VO2max and age in the VO2max response observed after CRT in AF.

A previous study37 demonstrated that treatment of HF patients with CRT improves exercise capacity and that this increase is most substantial among patients with a lower baseline VO2max (percentage predicted for age), the authors concluding that baseline CPET can therefore be used to identify patients who are more likely to exhibit functional improvement after CRT. Those with predicted VO2max <40% had much greater improvement in VO2max. In agreement with this, our AF patients had lower mean pre-CRT VO2max (probably related to older age, and also predicted VO2max) and more improved VO2max after CRT than SR. Also, HF patients with mean baseline VO2max <14 ml/kg/min, which was the case in our AF group, benefited most from the implantation of a CRT device.38 The percentage of functional CPET responders was significantly higher in AF patients than in SR patients, for the reasons mentioned above. Despite these better values, post-CRT mean VO2max in AF did not exceed that in SR, and values at six months were similar in the two study groups.

In patients with advanced HF, variation in VO2max is an important predictor of outcomes, including clinical deterioration or death, especially in patients with ischemic cardiomyopathy or not receiving beta-blockers.39 Exercise capacity is objectively quantified by measurement of VO2max, carbon dioxide production (VCO2), and minute ventilation.40 Not surprisingly, VO2max has a strong linear correlation with both cardiac output and skeletal muscle blood flow.41 Peak exercise capacity is defined as the maximum ability of the cardiovascular system to deliver oxygen to exercising skeletal muscle and of the exercising muscle to extract oxygen from the blood.42 As a result, exercise capacity is determined by three factors: pulmonary gas exchange; cardiovascular performance, including the peripheral vascular tree; and skeletal muscle metabolism. It has been demonstrated that CRT significantly improves all ventilation and metabolic parameters of patients with HF and ventricular conduction delay. Patients with more depressed metabolic and ventilation parameters and higher HR at baseline seem to benefit most from this therapeutic approach.38 These results are in agreement with those observed in our study. AF patients were more deconditioned, with worse physical condition and lower exercise capacity, related to severe HF and aging, as demonstrated, and after CRT they had greater improvement in cardiovascular performance. Mean CPET dur increased in both groups, early at three months, with a more than three-fold change in AF patients, in whom the change was statistically significant (unlike in SR patients), attaining similar values for CPET dur after CRT. CRT did not significantly alter exercise capacity in SR patients, but this finding is not surprising in patients whose exercise capacity was not so severely impaired at baseline. This observation is consistent with the study referred to above, which demonstrated that HF patients with relatively preserved exercise capacity at baseline achieve only minor improvement in exercise capacity during CRT.38

As mentioned above, the mean decrease in VE/VCO2 slope was significant at three months in both groups, which is also an important beneficial effect of CRT, since a lower VE/VCO2 slope in HF is associated with better prognosis.43

In conclusion, beneficial effects of CRT were demonstrated in HF patients, both in permanent AF and in SR, with similar proportions of clinical and echocardiographic responders.

Both groups showed LV reverse remodeling independently of cardiac rhythm, to a larger extent in SR patients, who also showed LV mass reduction and LA reverse remodeling, which were not present in AF patients. Additionally, AF patients, initially with less exercise functional capacity, had a greater improvement, with more functional responders. According to our results, permanent AF should not by itself be considered a factor against deciding to treat selected HF patients with CRT.

This work, analyzing the use of CRT in HF patients in permanent AF, has the inherent limitations of an observational study. It involves a medium-sized population sample, so the present results need to be treated with caution and should be reproduced in a larger permanent AF population, preferably in a prospective controlled clinical trial on CRT in AF, to confirm its results. Longer follow-up studies are needed.

This work was supported by an FCT grant (PTDC/DES/120249/2010).

The authors have no conflicts of interest to declare.