Distinctively different predictors for long‐term outcomes between responders and nonresponders who underwent cardiac resynchronization therapy

Abstract Background It is common to develop heart failure (HF) events even in respondents to cardiac resynchronization therapy (CRT) during a long‐term observation period. We investigated the predictors for long‐term outcome in responders in comparison with nonresponders in patients diagnosed with HF along with implanted CRT. Methods We enrolled 133 consecutive patients (mean age, 70 ± 10 years; 72 males) implanted with CRT from April 2010 to July 2019. Accurate follow‐up information (mean follow‐up period, 983 ± 801 days) was obtained from 66 responders and 53 nonresponders. Results Kaplan‐Meier event‐free curves showed that major adverse cerebral and cardiovascular event (MACCE)‐free ratio was significantly lower as the stage of renal function progresses (log rank, 19.5; P < .0001). The baseline estimated glomerular filtration rate (e‐GFR) before CRT was not significantly different between nonresponders and responders. The e‐GFR after judgment of CRT response was lower in patients with MACCEs than those without. Cox proportional hazards regression analysis revealed that low baseline e‐GFR before CRT and after judgment of CRT response was closely related with MACCEs in responders, but not in nonresponders. The survival rate in responders without MACCEs assessed using Kaplan‐Meier analysis was significantly larger in the preserved e‐GFR (baseline value before CRT, >44 mL/min/1.73 m2) group than in the depressed group (log rank, 20.29; P < .0001). Conclusion We demonstrate that the factors for MACCEs during long follow‐up periods were distinctively different between responders and nonresponders. Patients with depressed e‐GFRs are suggested to have poor prognosis even if they are responders to CRT.


| INTRODUC TI ON
For patients with severe heart failure (HF), there is no doubt that cardiac resynchronization therapy (CRT) can be the beneficial method on improving cardiac function, subjective symptoms, exercise capacity, and outcome. [1][2][3][4] Responders to CRT demonstrating reverse left ventricular (LV) remodeling after 6 months show good clinical outcomes after CRT. 5 However, it is common to develop HF events even in responders through long life cycle. In a study, approximately about 30% or more of the responders became transient responders 2 years after CRT. The transient responders showed significantly greater hospitalization rates for HF than responders with persistent response. 6 Nevertheless, it was not very clear about whether the factors predicting cerebral and cardiovascular events would be different between responders and nonresponders after a long-term follow-up.
Cardiorenal syndrome comprises clinical conditions where both cardiac and renal dysfunctions coexist. It results in impaired cardiac systolic function and glomerular filtration, which lead to cardiac volume decompensation. 7 This study determines the factors predicting longterm prognosis (cerebral and cardiovascular events) in responders in comparison with those in nonresponders with HF who underwent CRT.

| Patient selection
We identified and considered 133 patients with HF who had a CRT pacemaker implanted for the therapy of severe HF from April 2010 to July 2019. In this study, the indications for CRT adhere to the Japanese Circulation Society guidelines. 8 Patients who lack long-term follow-up and those undergoing dialysis were omitted. A total of 119 patients were eligible and registered to the study ( Figure 1). CRT responders were defined as patients with over 15% reduction in the left ventricular end-systolic volume (LVESV). 9 at the least 6 months after undergoing CRT operation.
Brain natriuretic peptide (BNP) or pro-BNP, and estimated glomerular filtration rate (e-GFR) were evaluated before CRT as the baseline value. They were also evaluated just after judgment of CRT response.
Patients who had readmitted due to HF, who had worsening of the NYHA class on the final follow-up observation, who had continuous moderate-to-severe deterioration on the clinical composite score 10 at the final medical examination, who lastingly stopped biventricular pacing due to worsening HF, or who died within 6 months after CRT were assigned to nonresponders. 9 This research was approved by the ethics review board of Oita University, and we received informed consent from whole patients. Our study was directed consistent with the principles expressed in the Declaration of Helsinki of the World Medical Association.

| Echocardiographic assessment
We performed M-mode, 2D, and Doppler echocardiographic measurements by a 1.5-4.0 MHz transducer at a fitting depth on apical and parasternal views. LVESV, left ventricular end-diastolic volume (LVEDV), and LV ejection fraction (LVEF) were evaluated using echocardiography by modified biplane Simpson's rule.

| Follow-up
All 119 patients performed the complete examination about CRT, and the status of clinical function was checked every 4-6 months after operation. Exact follow-up data for 983 ± 801 days were obtained. The endpoint was prespecified as the occurrence of major adverse cerebral and cardiovascular events (MACCEs), which contained stroke, nonfatal myocardial infarction, cardiovascular death, percutaneous coronary intervention or coronary artery bypass graft, implantable cardioverter-defibrillator treatment for fatal arrhythmia, and hospitalization due to congestive HF. Among the aforementioned events, the statistical analysis only considered the first event.

| Statistical Analysis
Data were shown in mean ± SD. The analysis of variance was used for continuous variables, and the Chi-squared test was used for categorical variables. The Student's t test was used to analyze the difference between groups. Univariate Cox proportional hazard regression analyses were done to distinguish the factors for prediction of MACCEs. The risk model was applied and gender, age, heart rates, blood pressures, etiologies of cardiomyopathy, QRS durations, LVEF, LVESV, LVEDV, BNP, and e-GFR were contained in the risk factors. The results are given as hazard ratios with 95% confidence intervals. A P value <0.1 was considered statistically substantial. Multiple Cox regression analysis was performed using values which had significance in a univariate analysis. Kaplan-Meier survival analysis was performed to associate MACCE-free ratio among the preserved and depressed e-GFR groups. All analyses were calculated by JMP (version 13.2.0; SAS Institute, Cary, NC, USA).

| Clinical characteristics
Accurate follow-up information (mean follow-up duration, 983 ± 801 days) was obtained from 66 responders and 53 nonresponders. Table 1 showed baseline characteristics of the responders and the nonresponders. Significant difference was not obtained regarding gender, age, QRS duration, heart rate, blood pressures, echocardiographic measurements, and values of blood samplings between the responders and the nonresponders. Regarding primary HF etiology, the number of patients with dilated cardiomyopathy was larger in the responders compared to the nonresponders. There was no significant difference regarding the prevalence of atrial fibrillation between nonresponders and responders (7.6% vs 16.7%, P = .13), and between patients with MACCEs and without (15.8% vs 9.7%, P = .31), respectively.

| The chronological changes of e-GFR and BNP and the occurrence of MACCE
The rates of responder to CRT among the groups based on stages of CKD 11 before CRT implantation were not significant (P = .601), as follows. The rates of responder to CRT were 63.6% in the patients with e-GFR ≥ 60 mL/min/1.73 m 2 (n = 22), 52.1% in the patients with e-GFR <60 and ≥30 ml/min/1.73 m 2 (n = 73), and 58.3% in the patients with e-GFR < 30 mL/min/1.73 m 2 (n = 24).  Table 2 showed the values of e-GFR and BNP before CRT and immediately after the judgment of CRT response. BNP after the judgment of CRT response was greater in the nonresponders than in the responders. There was no significant change between the two groups. Table 3 showed the values of e-GFR before CRT and after the judgment of CRT response. The e-GFR before CRT and after the judgment of CRT response was lower, and BNP after judgment of .76 Abbreviations: before CRT, baseline value before CRT; BNP, brain natriuretic peptide; CRT, cardiac resynchronization therapy; DBP, diastolic blood pressure; e-GFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction; SBP, systolic blood pressure.

TA B L E 1 Baseline clinical characteristics
CRT response was greater in patients with MACCEs than those without in responders. The baseline BNP before CRT, and that after judgment of CRT response were greater in patients with MACCEs than those without in nonresponders. There was no significant change of the other values for the occurrence of MACCEs between the two groups. and both levels of e-GFR before CRT and after judgment of CRT response and BNP after judgment of CRT response were associated with MACCEs in responders, as shown in Table 4.

| Difference of outcome by response to CRT
Multivariate Cox proportional hazards regression analysis clarified that high baseline plasma BNP levels both before CRT and after judgment of CRT response were independently related with the incidence of MACCEs in the nonresponders (Table 5; Model 1, 2), whereas low e-GFR levels both before CRT and after judgment of CRT response had independent relationship to the occurrence of MACCEs in the responders (Table 5; Model 3, 4). High plasma BNP levels after judgment of CRT response were independently related to the incidence of MACCEs in the responders ( Abbreviations: after CRT, after judgment of CRT response; before CRT, baseline value before CRT; BNP, brain natriuretic peptide; CRT, cardiac resynchronization therapy; e-GFR, estimated glomerular filtration rate; Δ Before CRT -after CRT, change of value between before CRT and after CRT.

TA B L E 2
The chronological changes of e-GFR and BNP according to the CRT response

| Comparison of patients by e-GFR in responders
On the basis of the receiver-operating characteristic curve for predicting MACCEs, the baseline e-GFR before CRT could be set at ≤44 mL/min/1.73 in responders (Figure 4). In the responders, Table 6 showed the baseline clinical characteristics between the preserved e-GFR group and the depressed e-GFR group.
Patients in the depressed e-GFR group were older than those in the preserved e-GFR group (P < .05). The numbers of patients with cardiac sarcoidosis (P < .05) and baseline e-GFR before CRT (P < .0001) were lower in the depressed e-GFR group than the preserved e-GFR group. There were no significant differences in gender, prevalence of atrial fibrillation, QRS duration, blood pressures, echocardiographic findings, and blood sample findings, except for e-GFR.
Through the follow-up period (1001 ± 621 days in the de-

| Kaplan-Meier MACCE-free estimation between the groups by e-GFR in responders
In the responders, the MACCE-free rate estimated by Kaplan-Meier analysis was obtained that the patients in the preserved e-GFR group were significantly higher than in the depressed e-GFR group (log rank, 20.29; P < .0001 and Wilcoxon, 17.39; P < .0001) ( Figure 5).

| Main findings
In this study, MACCEs developed in 62% of the nonresponders and 36% of the responders. Approximately 1500 days after CRT, the Abbreviations: after CRT, after judgment of CRT response; before CRT, baseline value before CRT; BNP, brain natriuretic peptide; CRT, cardiac resynchronization therapy; e-GFR, estimated glomerular filtration rate; Δ Before CRT -after CRT, change of value between before CRT and after CRT.

| Difference of risk factors for MACCEs by response to CRT
In this study, the risk factors for MACCEs were distinctively different between the responders and nonresponders through the follow-  Abbreviations: before CRT, baseline value before CRT; BNP, brain natriuretic peptide; CRT, cardiac resynchronization therapy; DBP, diastolic blood pressure; e-GFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction; SBP, systolic blood pressure.
*P < .05, **P < .01 risk factors for MACCEs did not match between the responders and nonresponders may be elucidated by many different backgrounds of the patients in this study in each group.
CRT response rate in this study tended to be lower than that in the previous studies. 9,12 Baseline e-GFR before CRT implantation was relatively low in all patients in this study; average e-GFR: 45 mL/min/1.73 m 2 (nonresponder) and 50 mL/min/1.73 m 2 (responders), respectively. Impairment of renal function (e-GFR <60 mL/min/1.73 m 2 ) was directly associated with a significant increase in mortality. 13 One of the reasons for low CRT response rate might be due to relatively lower baseline e-GFR in our patients.

| Renal dysfunction as a predicting factor for MACCEs in the responders
A study showed that many patients with HF after CRT had cardiorenal syndrome, a condition characterized by volume overload due to cardiac and renal dysfunctions. Patients with cardiorenal syndrome were more likely to experience MACCEs and become nonresponders with poor cardiac function improvement. 14 Multiple risk factors cause the progression of cardiac and renal dysfunctions, and they lead to severe HF. Additionally, the interconnection between multiple organs permits multiple dysfunction in multiple interconnected organs, such as increasing sodium ion, holding plasma volume, and systemic inflammation, through the stimulation of the renin angiotensin aldosterone system and the neurohormonal system. 7,15 These multiple functional changes occur the cardiac remodeling and the glomerular filtration progression. As a result, both the kidneys and the heart are steadily impaired. 14 Our results suggest that even in CRT responders judged 6 months after implantation, renal dysfunction at implantation requires close attention because of the potential high risk of developing MACCEs.

| Long-term outcomes in the responders
In a study, about one-fifth of the patients after CRT became super responders with normalization of LVEF. 16 Dramatic reverse LV remodeling is needed to obtain clinical improvement after 6 months of CRT and better survival and less hospitalizations for HF even after that. 16 Alternatively, MADIT-CRT, in Multicenter Automatic Defibrillator Implantation Trial with CRT, deaths or ICD activation for fatal arrhythmia developed in 12% of the responders approximately 3 years after CRT. 17 Several responders revealed a response for only short term, and reverse remodeling effects of LV were transitory in part of the responders. 18 They were called brief responders, 18 transient responders, 6 or late nonresponders. 19 Brief responders demonstrate reverse LV remodeling in the short term.
However, the beneficial effect could not be sustained through long period, 18  We should consider some limitations in this study. First, we estimated a small number of patients in a single center that might make the data analysis hard. Thus, future larger size studies are needed to detect the accurately outcome of the responders to CRT. Second, our analysis in this study was retrospective in nature, thus sampling bias and incomplete data might occurred. Consequently, prospective studies are necessary to detect the prognosis of the responders to CRT in future.

| CON CLUS ION
Our results suggest that the factors for MACCEs during a long followup period were distinctively different between the responders and nonresponders who underwent CRT. We believe that patients with depressed e-GFR have poor prognoses, even if they are CRT responders.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest. This study was approved by the institutional review board (IRB) of Oita University, and the IRB approval number was B11-053. Hidekazu Kondo https://orcid.org/0000-0003-2451-5845