Reverse left ventricular remodelling after aortic valve replacement for severe aortic insufficiency

Abstract

 

OBJECTIVES

The goal of this study was to investigate the long-term outcome of aortic valve replacement (AVR) for severe aortic insufficiency with a focus on pre- and postoperative left ventricular (LV) function to explore predictive factors that influence the recovery of LV function and clinical outcome.

METHODS

A total of 478 patients who underwent AVR for pure severe aortic insufficiency were grouped according to the preoperative echocardiographical LV ejection fraction (EF): low (LO) EF <35% (n = 43), moderate EF 35–50% (n = 150) or normal EF >50% (n = 285).

RESULTS

Actuarial survival at 10 years post-AVR was 64% with a LO EF, 92% with a moderate EF and 93% with a normal EF (P = 0.016), whereas 10-year rates of freedom from major adverse cerebral and cardiovascular events were 47%, 79% and 84%, respectively (P < 0.0001). Echocardiography at 1 year post-AVR demonstrated that EF substantially improved in all groups. We noted a significant difference in survival (P = 0.0086) and in freedom from major adverse cerebral and cardiovascular events (P = 0.024) between patients with an EF ≥35% and those with an EF <35% in the LO EF group. The multivariable logistic regression model showed that predictive factors for lack of improvement in EF 1 year post-AVR in the LO EF group included plasma brain natriuretic peptide >365 pg/mL (P = 0.0022) and echocardiographic LV mass index) >193 g/m² (P = 0.0018).

CONCLUSIONS

Long-term outcome post-AVR for severe aortic insufficiency was largely influenced by preoperative LV function. Predictive factors of failure to recover ventricular function post-AVR included EF <25%, pre-brain natriuretic peptide >365 pg/mL or LV mass index >193 g/m².

INTRODUCTION

Although severe aortic insufficiency (AI) with left ventricular (LV) dysfunction is the established surgical indication for aortic valve replacement (AVR), the surgical effects on LV function associated with long-term outcomes are not fully understood [1, 2]. Besides, the presence of LV dysfunction is reported to be associated with poor long-term outcomes of AVR [3–5]. AI produces volume overload in the left ventricle, which leads to increased wall stress and LV mass, whereas reduced diastolic blood pressure due to AI impairs coronary blood flow inducing global ischaemia of the LV tissue. As a result, severe AI leads to pathological hypertrophy of LV myocytes, accumulation of the fibrous components in the LV tissue and subsequent deterioration of the contractile function of the left ventricle with dilatation of the LV cavity [6–8]. Importantly, the ‘point of no return’ in severe AI is not fully understood. Moreover, the effect of AVR on the long-term prognosis possibly related to the recovery of the LV function is not fully understood [9, 10]. Therefore, the goal of this study was to review the long-term outcome of AVR for severe AI with a focus on pre- and postoperative LV function to explore predictive factors that influence the recovery of LV function and clinical outcome.

MATERIALS AND METHODS

Ethical statement

This study was approved by the National Cerebral and Cardiovascular Center (NCVC) institutional review board (approval number: M300263; 18 July 2018). The need for individual patient consent was waived by the review board.

Study cohort, data collection and study end points

The prospective institutional cardiac surgical database contained 478 consecutive patients who underwent AVR for severe AI with a prosthetic valve in the NCVC Hospital in Japan between January 2001 and April 2018. Patients having coronary artery disease, active infective endocarditis, other valve diseases or myocardial complications were excluded. Medical charts, operation reports and referral letters were reviewed and were further supplemented by telephone interviews for patients under the care of distant physicians. The background and characteristics of the cohort, which were retrieved from the electronic medical charts, were documented according to the guidelines (Table 1) [11]. Data collection was performed in June 2018. Death of a cardiac cause or major adverse cerebral and cardiovascular events (MACCE) were retrieved according to the guidelines [11]. Renal function was classified according to the National Kidney Foundation Kidney Disease Outcomes Quality Initiative guidelines for the evaluation, classification and stratification of chronic kidney disease [12]. Primary end points included death of any cause and MACCE, whereas secondary end points included echocardiographic structure and function post-AVR. The follow-up rate for survival at the end of April 2018 was 96%, and the mean follow-up period was 7.8 ± 0.5 years. Informed consent for this study was waived because of the retrospective, observational study design.

Table 1:

Baseline characteristics

Variable [mean ± SD or n (%)]	LO EF < 35% (n = 43)	MED 35–50% (n = 150)	Nl EF ≥ 50% (n = 285)	P-value
Male sex	32 (74)	105 (70)	202 (71)	0.21
Age (years)	64.1 ± 2.5	60.0 ± 1.9	59.3 ± 1.0	0.73
BSA (m2)	1.55 ± 0.2	1.53 ± 0.4	1.54 ± 0.1	0.85
Hypertension	14 (33)	47 (31)	91 (32)	0.31
Hyperlipidaemia	14 (33)	45 (30)	80 (28)	0.044
Diabetes mellitus	9 (21)	18 (12)	37 (13)	0.028
CKD (GFR, ml/min/1.73 m2)
G3a, b (30–59)	10 (23)	38 (25)	66 (23)	0.51
G4 (15–29)	3 (7.0)	6 (4.0)	9 (3.2)	0.045
G5 (<15)	2 (4.7)	1 (0.76)	0 (0)	0.019
Atrial fibrillation	9 (21)	25 (17)	40 (14)	0.028
Chronic obstructive pulmonary disease	5 (12)	11 (7.3)	26 (9.1)	0.021
Smoker	14 (33)	47 (31)	88 (31)	0.068
NYHA functional class
II	14 (33)	76 (51)	202 (71)	0.023
III	18 (42)	45 (30)	75 (26)	0.013
IV	11 (26)	29 (19)	8 (2.8)	0.019
BNP (pg/ml)	564 ± 86	187 ± 40	128 ± 21	0.0015
STS Score (%)	5.6 ± 3.7	4.7 ± 1.4	2.0 ± 0.5	0.022
EuroSCORE II (%)	4.9 ± 1.3	3.2 ± 0.7	1.8 ± 0.4	0.038
Aetiology
Annuloaortic ectasia	2 (4.6)	6 (4.0)	14 (4.9)	0.21
Bicuspid aortic valve	4 (9.3)	19 (13)	41 (14)	0.031

BNP: brain natriuretic peptide; BSA: body surface area; CKD: chronic kidney disease; EF: ejection fraction; GFR: glomerular filtration rate; LO: low; MED: moderate; Nl: normal; NYHA: New York Heart Association; SD: standard deviation; STS: Society of Thoracic Surgery.

Surgical indications and procedures

Surgical indications of the study cohort were determined preoperatively by the institutional heart team, essentially according to the guidelines [11]. All members of the study cohort were examined preoperatively by transthoracic and/or transoesophageal echocardiography, fluoroscopy and/or computed tomography-based coronary angiography. AVR was performed in all cohorts under induced cardiac arrest by intermittent tepid blood cardioplegia infusion. Tepid (32°C) blood cardioplegia has been the standard cardioplegia used in the authors’ institute during the last 30 years to facilitate the protection of the myocytes and the recovery of myocyte function after reperfusion. Although tepid versus cold cardioplegia was not compared randomly, the authors consider that tepid cardioplegia enhances postoperative recovery compared to cold cardioplegia. In the setting of AVR for severe AI, retrograde intermittent cardioplegia infusion was used predominantly.

The type of prosthesis, such as a biological or a mechanical prosthesis, was determined according to the guidelines and by a discussion with the patients, whereas the prosthesis size was determined by the surgeon intraoperatively so that the most appropriately sized prosthesis was implanted.

Transthoracic echocardiography

All cohorts were examined by standard transthoracic and/or transoesophageal echocardiography within 14 days pre/postoperatively. Patients who visited the outpatient clinic of the NCVC Hospital annually after their operation were examined annually by transthoracic echocardiography. Data were retrieved from the official echocardiographic report, including the degree of AI, which was assessed by an expert engineer and expert doctor separately according to the guidelines [11]. Although some patients who were treated only by distant physicians were not examined by institutional echocardiography, their clinical progress was regularly reported to our institute.

The LV mass index (LVMI) to body surface area (BSA) (g/m²) is estimated by the LV cavity dimension and wall thickness at end-diastole by transthoracic echocardiography.

LVMI is calculated from LV mass and BSA as follows:

$$\text{LV} \text{mass} \left(\mathrm{g}\right)=0.8\left\{1.04[([\text{LVEDD}+\text{IVSd}+\text{PWd}{]}^3-{ \text{LVEDD}}^3)]\right\}+0.6$$

$$\text{LVMI} \left(\mathrm{g}/{\mathrm{m}}^2\right)=\text{LV} \text{mass}/\text{BSA}$$

where LVEDD is the LV end-diastolic diameter (mm), IVSd is the interventricular septal thickness at end-diastole (mm) and PWd is the posterior wall thickness at end-diastole (mm) [13].

Grouping

The study cohort was grouped according to the degree of ejection fraction (EF) in the preoperative echocardiography study to review clinical progress and echocardiographic changes in each group. The patients with preoperative EFs <35% were categorized as the low (LO) EF (LO EF) group (n = 43); those with EF between 35% and 50% were categorized as the moderate (MED) EF (MED EF) group (n = 150); and those with an EF 50% or more were categorized as the normal (Nl) EF (Nl EF) group (n = 285), as reported by Kamath et al. [1].

Statistical analyses

Statistical analyses were performed using JMP statistical software (version 14.0, SAS Institute, Cary, NC, USA). Continuous data are presented as mean ± standard deviation unless otherwise specified. Categorical variables are presented as numbers and percentages. Baseline continuous variables were compared among the groups (LO EF vs MED EF vs Nl EF) using a one-way analysis of variance. Categorical variables were compared using the χ² test and the Fisher’s exact test. Preoperative and postoperative echocardiographic values were compared using the Wilcoxon signed-rank test. P < 0.05 was considered significant. Actuarial survival and freedom from MACCE were compared among the 3 groups (LO EF vs MED EF vs Nl EF) using Holm’s method. We adjusted for multiplicity using Holm's step-down procedure. To identify predictors of reverse LV remodelling, all clinical parameters relating to the risk factors for non-reverse remodelling in EF < 35% were proposed for inclusion in a univariable logistic regression model. Univariable testing was used as a prescreening method for variable inclusion in a multivariable model. Variables with a univariable P-value of less than 0.25 were examined. The most significant predictors per domain were then entered into a stepwise multivariable logistic regression model to identify significant independent predictors. A P-value of <0.05 was considered significant. Odds ratios and 95% confidence intervals were reported. Overall survival and freedom from MACCE after AVR were estimated by the Kaplan–Meier method.

The area under curve was calculated by obtaining the receiver operating characteristic curve from the logistic regression model for the risk factors for non-reverse remodelling in EF <35%. The cut-off value with the most favourable sensitivities and specificities was determined using the Youden index from the receiver operating characteristic curve.

RESULTS

No inter-group difference in the surgical procedures

The size and type of the implanted prosthetic valves are summarized in Table 2. Biological prostheses were used predominantly in all groups. Of the biological prostheses, a stented bovine pericardial valve such as the Carpentier-Edwards Perimount Magna Ease (Edwards Lifesciences Corp., Irvine, CA, USA) was most commonly used followed by a stented porcine aortic valve, such as the Medtronic Mosaic Ultra (Medtronic, Inc., St Paul, MN, USA). Of the mechanical prostheses, a bileaflet titanium valve, such as the Advancing the Standard prosthesis (Medtronic) or On-X prosthesis (On-X Life Technologies Inc., CryoLife, Kennesaw, GA, USA), was used. No significant inter-group difference was found in the type or size of the prosthesis.

Table 2:

Surgical procedures

Variable [mean ± SD, or n (%)]	LO EF < 35% (n = 43)	MED EF 35–50% (n = 150)	Nl EF ≥ 50% (n = 285)	P-value
Biological prosthesis	29 (67)	100 (67)	188 (66)	0.75
CEP Magna EASE	14 (33)	51 (34)	99 (35)	0.92
Mosaic ultra	9 (21)	33 (22)	60 (21)	0.63
Trifecta	2 (5.2)	5 (3.3)	14 (4.8)	0.58
Epic supra	0 (0)	1 (0.7)	2 (0.7)	0.073
Medtronic freestyle	3 (7.0)	8 (5.3)	2 (0.7)	0.0068
Others	1 (2.3)	2 (1.3)	11 (3.9)	0.27
Mechanical prosthesis	14 (33)	50 (33)	97 (34)	0.75
ATS standard	7 (16)	27 (18)	48 (17)	0.99
ATS AP360	1 (2.3)	5 (3.3)	8 (2.8)	0.23
ON-X	1 (2.3)	4 (2.7)	7 (2.5)	0.41
SJM regent	2 (5.2)	7 (4.7)	15 (5.3)	0.37
Carbomedics	1 (2.3)	3 (2.0)	7 (2.5)	0.54
Others	2 (5.2)	4 (2.7)	12 (4.2)	0.46
Concomitant procedure
Bentall	5 (12)	18 (12)	31 (11)	0.47
Ascending aortic replacement	3 (7.0)	11 (7.3)	20 (7.0)	0.32
Maze	2 (4.7)	11 (7.3)	18 (6.4)	0.041
Pacemaker implant	2 (4.6)	6 (4.0)	12 (4.2)	0.145

EF: ejection fraction; LO: low; MED: moderate; Nl: normal; SD: standard deviation.

Operation time (P = 0.031) and cardiopulmonary bypass time (P = 0.019) were significantly longer in the LO EF group versus the other 2 groups, though aortic cross-clamp time was not significantly different among the 3 groups (Table 3).

Table 3:

Operative outcomes

Variable [mean ± SD, or n (%)]	LO EF < 35% (n = 43)	MED 35–50% (n = 150)	Nl EF ≥ 50% (n = 285)	P-value
Operation time (min)	296 ± 57	242 ± 39	235 ± 33	0.031
Cardiopulmonary bypass time (min)	132 ± 18.3	120 ± 15.9	116 ± 27.4	0.019
Aortic cross-clamp time (min)	82.2 ± 14.5	79.6 ± 16.4	77.3 ± 5.8	0.221
Intubation time (h)	14.3 ± 2.5	11.7 ± 5.9	8.2 ± 0.8	0.018
Intubation >24 h	11 (26)	4 (2.7)	3 (1.1)	0.023
Permanent pacemaker	3 (7.0)	1 (0.7)	1 (0.4)	0.032
Atrial fibrillation	11 (26)	22 (17)	42 (15)	0.018
Renal failure requiring dialysis	6 (14)	2 (1.3)	0	0.012
Need for IABP	6 (14)	0	0	0.008
Need for VA-ECMO	4 (9.3)	0	0	0.011
Postoperative ICU stay (days)	5.2 ± 6.1	4.4 ± 2.7	2.1 ± 0.3	0.027
In-hospital deaths	5 (12)	0	0	0.020
Cause of death
Aspiration pneumonia	3 (7.0)	0	0	0.008
Heart failure	2 (4.7)	0	0	0.011
Cerebrovascular accident	0	0	0
Para-/hemiplegia	0	0	0
Late mortality rate	4 (9.3)	9 (6.0)	8 (2.8)	0.045
Cause of death
Aspiration pneumonia	1 (2.3)	3 (2.0)	0	0.026
Heart failure	0	3 (2.0)	3 (1.1)	0.65
Postoperative myocardial infarction	2 (4.7)	0	0	0.041
Cerebral infarction	2 (4.7)	1 (0.7)	0	0.027
Aortic dissection	0	0	1 (0.4)	0.071
Bowel ischaemia	0	0	1 (0.4)	0.071
Aortic aneurysm	1 (2.3)	1 (0.67)	0	0.057
Cancer	0	1 (0.67)	2 (0.8)	0.066
Other	0	0	1 (0.4)	0.071
Postoperative echocardiography in-hospital
Ejection fraction (%)	27 ± 4.3	39 ± 1.4	48 ± 1.2	0.027
Paravalvular AI (%)	3 (7.0)	10 (6.7)	18 (6.3)	0.22
Peak velocity (m/s)	2.3 ± 0.073	2.3 ± 0.085	2.4 ± 0.11	0.13
Peak PG (mmHg)	21.9 ± 1.7	21.1 ± 1.6	23.2 ± 2.2	0.26
Mean PG (mmHg)	12.7 ± 0.94	12.1 ± 0.96	11.9 ± 0.91	0.071
EOA (cm2)	1.45 ± 0.15	1.51 ± 0.092	1.78 ± 0.13	0.055
Indexed EOA (cm2/m2)	0.89 ± 0.078	0.91 ± 0.054	1.13 ± 0.083	0.089

AI: arterial insufficiency; EF: ejection fraction; EOA: effective orifice area; IABP: intra-aortic balloon pump; ICU: intensive care unit; LO: low; MED: moderate; Nl: normal; PG: pressure gradient; SD: standard deviation; VA-ECMO: venoarterial extracorporeal membrane oxygenation.

In-hospital outcome post-aortic valve replacement for severe aortic insufficiency

No in-hospital deaths were reported in the MED EF or Nl EF groups. In contrast, 5 in-hospital deaths (12%) post-AVR occurred in the LO EF group; the deaths were associated with aspiration pneumonia in 3 patients and with low cardiac output syndrome in 2 patients (Table 3). Venoarterial extracorporeal membrane oxygenation (VA-ECMO) was used in 4 of the 5 patients who died in-hospital.VA-ECMO was not used in the remaining patient who had pneumonia: This patient was an 89-year-old man whose family refused invasive treatments such as VA-ECMO. In addition, the incidences of the need for a postoperative permanent pacemaker, of atrial fibrillation and of renal failure requiring dialysis were significantly higher in the LO EF group than in the other groups. Furthermore, intubation time, intubation >24 h, need for an intra-aortic balloon pump, VA-ECMO and length of postoperative stay in the intensive care unit were significantly worse in the LO EF group compared to the other groups. Other postoperative complications, such as a cerebrovascular accident, were not significantly different among the 3 groups. No patients showed a patient-prosthesis mismatch in any group.

Preoperative ejection fraction-dependent long-term outcome

During the follow-up of 7.8 ± 0.5 years post-AVR for severe AI, 26 patients died, including 9 patients (21%) in the LO EF group, 9 patients (6.0%) in the MED group and 8 patients (2.8%) in the Nl EF group. Although 5 patients died in-hospital (overall incidence, 1.0%), 21 patients died after discharge from the hospital (Table 3).

Actuarial survival and MACCE were estimated statistically and compared among the 3 groups using the Kaplan–Meier method; multiple comparisons were made using Holm’s method (Fig. 1). The 10-year survival rates were 64% in the LO EF group, 92% in the MED EF group and 93% in the Nl EF group, with a significant inter-group difference (P = 0.016). Besides, the 10-year freedom from MACCE was 47% in the LO EF group, 79% in the MED EF group and 84% in the Nl EF group, with a significant inter-group difference (P < 0.0001).

Figure 1:

Long-term outcomes post-aortic valve replacement for aortic insufficiency. (A) Actuarial survival late after aortic valve replacement for each group (LO EF, MED EF and Nl EF), compared with multiple comparisons by Holm’s method (P = 0.016). (B) Freedom from MACCE late after aortic valve replacement for each group (LO EF, MED EF and Nl EF) compared with multiple comparisons by Holm’s method (P < 0.0001). LO EF: low ejection fraction; MACCE: major adverse cerebral and cardiovascular events; MED EF: moderate ejection fraction; Nl EF: normal ejection fraction.

Variable degree of recovery of left ventricular function depending upon the preoperative function

The in-hospital postoperative EF was slightly lower than that obtained from the preoperative echocardiographic scans in all 3 groups (Fig. 2). The postoperative EF value was highly variable in the LO EF group, whereas the postoperative EF in the MED EF group was relatively consistent and was not significantly different from that in the Nl EF group. Both the LV end-diastolic diameter (LVEDD) and the LV end-systolic diameter of the showed a marked reduction postoperatively in the 3 groups (Fig. 2). Of note, postoperative LVEDD and LV end-systolic diameter were not significantly different between the MED EF and Nl EF groups.

Figure 2:

Changes in cardiac dimensions at 1 year after aortic valve replacement. Graph shows changes in EF and end-diastolic diameter and end-systolic diameter of the left ventricle with aortic valve replacement in groups. Postoperative EF was significantly increased in all groups, and both LVEDD and LVESD are slightly increased in all groups. EF: ejection fraction; LVEDD: left ventricular end-diastolic diameter; LVESD: left ventricular end-systolic diameter; LO: low; MED: moderate; Nl: normal.

Exploration of factors associated with reverse left ventricular remodelling post-aortic valve replacement

Because the LO EF group showed a variable degree of recovery in the EF postoperatively, its effect on long-term outcomes was studied by dividing the patients into the following 2 groups: a reverse remodelling group comprising 31 patients (6.5%) who had postoperative EFs ≥35% and a non-reverse remodelling group comprising the remaining 12 (2.5%) patients, who had postoperative EFs <35%.

Four in-hospital deaths were reported in the non-reverse remodelling group; these were associated with aspiration pneumonia in 3 patients and with heart failure in 1 patient. In contrast, 1 in-hospital death was reported in the reverse remodelling group that was associated with heart failure. Actuarial survival and freedom from MACCE in the 2 groups, assessed by the Kaplan–Meier method, showed that the 5-year survival rate was 96% in the reverse remodelling group and 63% in the non-reverse remodelling group, with a significant inter-group difference (P = 0.0086) (Fig. 3A). Of note, freedom from MACCE was significantly and markedly greater in the reverse remodelling group versus the non-reverse remodelling group (P = 0.024) (Fig. 3B). MACCE included death of any cause in 9 patients, myocardial infarction in 2 patients, cerebral infarction in 2 patients and aortic dissection in 1 patient. The multivariable logistic regression model showed that predictive factors of non-reverse remodelling post-AVR in the LO EF group included a plasma brain natriuretic peptide (BNP) level >365 pg/ml (P = 0.0022) and preoperative echocardiographic LVMI >193 g/m² (P = 0.0018) (Fig. 4, Table 4).

Figure 3:

Long-term outcomes of the LO EF group. (A) The long-term outcome as actuarial survival between the 2 groups was assessed by the Kaplan–Meier method. As a result, the 5-year survival rate was 96.2% in the RR group. In contrast, the 5-year survival rate was 63.3% in the non-RR group. Of note, there was a significant inter-group difference in survival (P = 0.0086). (B) Long-term outcome in terms of freedom from MACCE between the 2 groups was assessed by the Kaplan–Meier method. Freedom from MACCE was significantly and markedly greater in the RR group than in the non-RR group (P = 0.024). EF: ejection fraction; MACCE: major adverse cerebral and cardiovascular events; RR: reverse remodelling.

Figure 4:

Exploring cut-off values of irreversibility of left ventricular function. Receiver operating characteristic analysis of preoperative BNP level and LVMI is presented. Best cut-off values with preoperative BNP and LVMI: <365 pg/ml for pre-BNP and <193 g/m² for LVMI. The AUCs were 0.73 and 0.80, respectively. AUC: area under the curve; BNP: brain natriuretic peptide; CI: confidence interval; LVMI: left ventricular mass index.

Table 4:

Predictive risk factors for non-reverse remodelling with ejection fraction <35%

	Univariable			Multivariable
	OR	95% CI	P-value	OR	95% CI	P-value
Age	1.1	0.9–1.1	0.15
Male	1.2	0.4–2.2	0.19
NYHA ≤ III	2.9	2.0–3.9	0.001
COPD	3.1	1.1–8.2	0.043
LVEDD (mm)	1.3	0.4–1.8	0.041
DcT (ms)	1.8	0.8–3.4	0.2
EF < 25%	3.2	0.6–9.2	0.033	8.9	0.7–10.0	0.0048
BNP > 365 (pg/ml)	9.7	8.3–70.3	0.022	31	89–362	0.0022
LVMI > 193 (g/m2)	4.8	1.8–5.6	0.029	4.2	0.9–13.1	0.0018
CKD ≤ G4	1.3	0.7–6.1	0.023
β-blocker	2.1	0.8–4.5	0.04
Loop diuretic	1.4	0.5–2.7	0.07

BNP: brain natriuretic peptide; CI: confidence interval; CKD: chronic kidney disease; COPD: chronic obstructive pulmonary disease; DcT: deceleration time; EF: ejection fraction; LVEDD: left ventricular end-diastolic diameter; LVMI: left ventricular mass index; NYHA: New York Heart Association; OR: odds ratio.

DISCUSSION

We reviewed the long-term outcomes of AVR for 478 patients who had severe AI by dividing the cohort into groups, depending on the preoperative EF. The LO EF group had extracardiac comorbidities such as diabetes or chronic kidney disease that produced a markedly higher predictive risk of surgery than those of the other groups. Although surgical procedures including size or type of prosthesis were not significantly different among the 3 groups, 5 in-hospital deaths and significant postoperative complications were recorded in the LO EF group. In the long term, the LO EF group showed a significantly higher mortality rate and a greater incidence of MACCE than the other groups. In an average of 2 years postoperatively, echocardiographic EF information was recovered consistently in the MED EF group but its availability was highly variable in the LO EF group. In the LO EF group, there was a marked and significant difference between the cohorts with or without reverse LV remodelling. The preoperative plasma BNP level and the echocardiographic LVMI were the determinants of postoperative reverse remodelling in the LO EF group.

Two articles with results similar to those of the present study were published by Chaliki et al. in 2002 [14] and by Kaneko et al. in 2016 [15]; both groups documented the long-term outcome of AVR for patients having severe AI with a focus on preoperative EF <35%, 35–50% or >50%. These 3 studies, together with the present study, document the following consistent and inconsistent results: First, Chaliki et al. reported that postoperative outcomes, either in-hospital or long-term outcomes, were dependent on the preoperative EF, as presented in this study. In contrast, Kaneko et al. reported that in-hospital and long-term outcomes were not associated with preoperative EF. Kaneko et al. concluded that reduced EF is not predictive of long-term outcome, because LV remodelling is reversed after AVR. Kaneko et al. also commented that the preoperative patient selection process might have excluded patients unlikely to show reverse LV remodelling after AVR. In fact, the average EF of the LO EF group in the present study was 27 ± 1%, whereas that in Kaneko’s study was 30 ± 5%. There would be a ‘point of no return’ for reverse LV remodelling in patients with an EF <30%. However, the authors consider that measurement of EF is not always accurate and that multiple modalities would more accurately reflect the viability of the myocardium. The authors therefore used a logistic regression model to explore risk factors of irreversible LV remodelling, which is the second conclusion of this manuscript.

The present study statistically determined that the ‘point of no return’ would be a preoperative level of plasma BNP of >365 pg/ml and echocardiographic LVMI >193 g/m².

Second, the MED EF group, in which the preoperative EF was 35–50%, showed an identical long-term outcome to the Nl EF group with a consistent recovery of EF postoperatively in the present study, whereas Chaliki’s study showed significantly less survival in the MED EF group versus the Nl EF group. In addition, the overall survival in Chaliki’s report appears to be worse than that in the present study and in Kaneko’s report. The development of postoperative conservative therapies for the heart and/or extracardiac disease conditions, such as diabetes or chronic kidney disease, in the last decade might have contributed to the outcome of AVR for severe AI. In particular, the MED EF group showed nearly full recovery of the EF and the end-systolic diameter/end-diastolic diameter of the left ventricle postoperatively in this study, indicating that heart function and structure with an EF of 35–50% associated with severe AI are reversible by AVR and subsequent conservative treatments. AVR for EF 35–50% is therefore justified. The surgical indications for cases of EF <35% have not been fully established; however, patients considered for AVR in this setting should likely also be considered for advanced support such as VA-ECMO or a ventricular assist device in advance of the AVR.

Finally, most importantly, the present study revealed (i) variability in the degree of recovery of the EF and LVEDD/LV end-systolic diameter post-AVR for severe AI with poor LV function; (ii) poor long-term outcomes associated with irreversible poor LV function; and (iii) predictive risk factors of irreversible poor LV function, such as EF <35%, plasma BNP >365 pg/ml and/or LVMI >193 g/m². However, AVR is not contraindicated in patients for whom irreversible poor LV function is predicted [16]. A comparison of AVR with optimized medical treatment might be warranted for this patient population. Moreover, transcatheter AVR might be indicated for them [17, 18].

This study is limited by the retrospective study design. However, the study cohort is a consecutive series in the single institution where surgical indications and procedures were consistent throughout the study period. Also, follow-up was completed in 96% of the total cohort because most of the patients routinely visited the outpatient clinic of the institute annually or more frequently.

CONCLUSION

Long-term outcome post-AVR for AI was largely influenced by preoperative LV function. The patients with preoperative EF <35%, in general, showed a substantial recovery of LV function post-AVR, whereas the population that failed to show recovery exhibited poor prognosis in the long term. Predictive factors of failure in the functional recovery post-AVR included EF <25%, pre-BNP >365 pg/ml or LVMI >193 g/m².

Conflict of interest: none declared.

ABBREVIATIONS

AI

Aortic insufficiency

AVR

Aortic valve replacement

BSA

Body surface area

BNP

Brain natriuretic peptide

EF

Ejection fraction

LV

Left ventricular

LVEDD

Left ventricular end-diastolic diameter

LVESD

Left ventricular end-diastolic diameter

LVMI

Left ventricular mass index

LO EF

Low ejection fraction

MACCE

Major adverse cerebral and cardiovascular events

MED EF

Moderate EF

NCVC

National Cerebral and Cardiovascular Center

Nl EF

Normal EF

VA-ECMO

Venoarterial extracorporeal membrane oxygenation

Author contributions

Teppei Toya: Writing—original draft. Satsuki Fukushima: Writing—review & editing. Yusuke Shimahara: Other. Shingo Kasahara: Supervision. Junjiro Kobayashi: Supervision. Tomoyuki Fujita: Supervision.

Reviewer information

Interactive CardioVascular and Thoracic Surgery thanks Manuel J. Antunes, Giovanni Alfonso Chiariello and the other, anonymous reviewer(s) for their contribution to the peer review process of this article.

 

Presented at the 98th Annual Meeting of the 2018 AATS, San Diego, CA, 28 April to 1 May 2018.