Aortic root enlargement to mitigate patient–prosthesis mismatch: do early adverse events justify reluctance?^†

Abstract

Open in new tabDownload slide

OBJECTIVES

Concomitant aortic root enlargement (ARE) at the time of surgical aortic valve replacement can be performed to avoid patient–prosthesis mismatch, an important predictor of adverse long-term outcome.

METHODS

We performed a single-centre, retrospective analysis of 4120 patients receiving isolated aortic valve replacement, of whom 171 (4%) had concomitant ARE between January 2005 and December 2015. The analysis of postoperative outcome and early mortality was performed. Owing to inequality of the groups, patients were matched 1:1.

RESULTS

The mean age of all 4120 patients was 68.8 ± 10.5 years, and comorbidities were equally balanced after matching. The mean aortic cross-clamp time, cardiopulmonary bypass time and total operative time were prolonged by 19, 20 and 27 min in the ARE group, respectively. Early mortality was not statistically significantly different with 1.4% in the surgical aortic valve replacement and 1.8% in the ARE group. Postoperative complications were <5% in all matched 338 patients: bleeding (3% vs 3%), pericardial effusion (3.0% vs 4.2%), sternal instability (1.8% vs 0%) and sternal wound infection (3.0% vs 1.2%). A significant higher number of patients had respiratory failure after ARE (unmatched: 17.1% vs 9.9%, P < 0.001; matched: 18.3% vs 9.5%, P = 0.028). Factors independently associated with overall mortality were age [hazard ratio (HR) 1.71], chronic obstructive pulmonary disease (HR 1.47), diabetes (HR 1.82), atrial fibrillation (HR 2.14) and postoperative respiratory failure (HR 2.84).

CONCLUSIONS

ARE can be performed safely in experienced centres with no significant increase in the risk of early postoperative surgical complications and early mortality. However, the surgeon and the intensive care unit team should be aware of an increased risk for postoperative respiratory failure in ARE patients.

INTRODUCTION

Aortic root enlargement (ARE) at the time of surgical aortic valve replacement (sAVR) has been proven to be a good option for patients with a small aortic annulus and the impending risk of patient–prosthesis mismatch (PPM) [1–3]. ARE, which was first described by Nicks in 1970, utilizes a single-patch technique to augment the hypoplastic aortic root [4].

Short- and long-term results of isolated sAVR are excellent with a mortality in patients <75 years of up to 1.9% [5] and a 30-day mortality of up to 9% in high-risk patients >80 years [6]. New developments focussing on improving haemodynamic profiles with low gradients and the maximal effective orifice area (EOA) have resulted in the optimization of biological valve prostheses [7].

The aim of this study was to determine the perioperative and postoperative risks and outcomes of concomitant ARE in sAVR.

METHODS

Our institutional database was retrospectively reviewed from January 2005 to December 2015 to gather contemporary data. After exclusion of patients with previous cardiac surgery, with missing data (i.e. relevant comorbidities, incomplete echo data, etc.), emergency cases and other major concomitant procedures [e.g. coronary artery bypass surgery (CABG), repair/replacement of another heart valve and ascending/arch replacement], 4120 consecutive patients were included. A total of 3949 (95.8%) patients underwent isolated sAVR group and 171 (4.2%) received sAVR with concomitant ARE (ARE group).

Patient demographics and indication for surgery

A total of 4120 patients (mean age 68.8 ± 9.5 years, range 18–92 years) underwent sAVR with and without ARE. The major pathology was aortic valve stenosis in 89.2% and aortic valve regurgitation in 7.8%. Patient characteristics and preoperative data are listed in Table 1 (before matching) and Table 2 (after matching).

Surgical technique

Conventional sAVR was performed through full median sternotomy (67%) or minimally invasive via partial upper sternotomy (33%). The cardiopulmonary bypass (CPB) was established via direct aortic and right atrial cannulation in most instances. Myocardial protection consisted of antegrade (rarely retrograde) administration of blood or crystalloid. Standard techniques were used to excise the native aortic valve and the surrounding calcified tissue followed by insertion of a biological or mechanical prosthesis. The need for ARE was evaluated before surgery based on the echocardiographically measured aortic annulus and body surface area (BSA) of the patient. However, the aortic annulus was intraoperatively in vivo measured using standard prosthetic valve sizers and the decision for root enlargement was re-evaluated by the surgeon. In rare cases, the decision for ARE was made in the operating room (OR) because of the smaller annulus than expected or inability to close the aortotomy. In case of ARE, the Nick single-patch [4] technique was utilized. Concomitant procedures were the surgical resection of secondary septal myocardial hypertrophy (11%), and bilateral pulmonary vein ablation by radiofrequency for atrial fibrillation (10%). Operative data are listed in Table 3. The prosthesis types used can be found in the Supplementary Material, Table S1.

Echocardiographic data based patient–prosthesis mismatch calculation

Data sets of preoperative and postoperative transthoracic and transoesophageal echocardiography were reviewed. For calculation of the PPM, the EOA of the prosthesis (as given by the manufacturer) was divided by the BSA. The definition of PPM was utilized according to previously published data with moderate PPM defined as an EOA index between 0.66 and 0.85 cm²/m² and severe PPM ≤0.65 cm²/m² [1].

Outcomes

The primary outcome was early mortality, defined as postoperative death within 30 days of surgery or within the primary hospitalization period. Secondary end points were length of aortic cross-clamp time, CPB time, length of surgery, re-exploration for bleeding, pericardial effusion, intensive care unit (ICU) stay, postoperative respiratory failure (RF)–defined as non-invasive ventilation, reintubation or tracheostomy- and time to discharge. Overall mortality was defined as death from any cause during follow-up.

Follow-up

Follow-up was performed according to the institutional database supplemented by individual patient records. Follow-up data for long-term survival or adverse outcomes were routinely recorded via direct telephone interviews with the patient, a close relative or the referring physician, and it was closed on 21 July 2017. The median follow-up time was 3.12 years (range 0–12.9 years, total patient-years = 13 797).

Statistical analysis

Data were transferred to SPSS (IBM SPSS Statistics, version 24.0) for statistical analysis. Study groups are characterized by frequency for categorical variables and mean ± standard deviation for continuous variables.

To overcome the inequality of the groups, we matched sAVR and ARE patients 1:1 following the concept of full matching [8]. Owing to missing values, we decided to match patients exactly by sex, pulmonary hypertension >60 mmHg, urgency and aortic valve dysfunction, and age and left ventricular ejection fraction (LVEF) with caliper 1. We performed matching by means of the R package ‘Matching’. For balance checking, standardized mean differences and ORs were calculated with a 95% confidence interval (CI).

In the unmatched cohort, frequencies were compared by the χ² test and skew distributed continuous variables by the Mann–Whitney U-test. Survival data were depicted by the Kaplan–Meier method and analysed with Cox regression. Because analysis methods must take into account the dependency structure after matching, the McNemar test was applied for pairs of binary characteristics. Skew distributions were compared by the Wilcoxon’s signed-rank test, and conditional logistic and Cox regression was performed. Validity of the proportional-hazards assumption was checked by means of Schoenfeld residuals. Because patients with RF have increased mortality risk in the first 3 months, a Cox regression model with time-dependent covariates was built.

Looking for multiple covariates associated with mortality, we selected variables by backwards exclusion following the Akaike information criterion and reduced the resulting model further to get clinically sensible models. We estimated effects as hazard ratios (HRs) or ORs with 99% CI.

All tests were performed 2-tailed with a predefined significance level of 5%.

RESULTS

Preoperative patient characteristics

The mean age of all 4120 patients was 68.8 ± 11 years with only marginal differences between patients receiving sAVR with and without ARE (68.9 ± 11 vs 67.4 ± 10). The ARE group included a total of 67% of female patients, whereas in the sAVR group only 42% of patients were women [OR 2.71 (1.95–3.82)]. In both groups, the comorbidities, such as diabetes mellitus (32% vs 32%, OR 0.99), arterial hypertension (89% vs 86%, OR 1.29), smoking (26% vs 30%, OR 0.82), hyperlipidaemia (57% vs 52%, OR 1.22), peripheral vascular disease (14% vs.15%, OR 0.92) and preoperative renal dysfunction requiring dialysis (2% vs.1%, OR 1.39) were present. Pulmonary hypertension was significantly higher in the ARE group (24% vs 15%, OR 1.86). No relevant differences in body mass index (29 kg/m² vs 29 kg/m²) and preoperative LVEF (60% vs 59%) were detected. The matched cohort consisted of 169 patients each (total n = 338), see Table 2.

Intraoperative data

Intraoperative data are listed in Tables 3 and 4. In the unmatched cohort, two-thirds of patients were operated on through full median sternotomy and one-third of patients through partial sternotomy. After matching 62% of sAVR patients and 70% of ARE patients, received a full median sternotomy (P = 0.14). The aortic cross-clamp time was in mean by 19 (95% CI 16–25) min prolonged in the ARE group (77.7 ± 20 vs 58.3 ± 17, P < 0.001), CPB time was in mean 20 (95% CI 16–24) min longer (101.2 ± 27 vs 81.3 ± 24, P < 0.001) and consequently the total length of surgery was in mean 27 (95% CI 21–33) min longer (176.8 ± 38 vs 149.8 ± 36, P < 0.001), see Table 3. One of 4 patients in the ARE group underwent septal myectomy compared to 1 of 10 sAVR patients (P < 0.001). These differences were maintained after matching. The need for blood products (red blood cells (RBCs) and platelets) was significantly higher in the ARE group (P < 0.001 and P = 0.034).

Prior to matching, ‘postoperative rates’ of moderate PPM were similar between the groups (27% vs 25%, P = 0.12), see Table 4. However, the remaining incidence of severe PPM was higher in sAVR patients compared to ARE (3% vs 0.6%, P = 0.12) patients. This difference was still present after 1:1 matching (2% vs 0.6%, P = 0.23). More patients with sAVR had remaining moderate PPM after matching (30% vs 25%, P = 0.23), but this was not statistically significant.

Postoperative outcome and early mortality

Early mortality in the entire cohort was low with 1.4%, with only a marginal difference between the groups (1.4% vs 1.8%, P = 0.69). After matching, this parity remained. Postoperative complications were <5% in all matched 338 patients: bleeding (3% vs 3%), pericardial effusion (3% vs 4%), sternal instability (2% vs 0%) and sternal wound infection (3% vs 1%), however, with only a small number of events, as shown in Table 5. A significantly higher number of patients experiencing RF after sAVR with ARE was observed for the unmatched (17% vs 10%, P < 0.001) cohort and the matched (18% vs 10%, P = 0.028) cohort.

Risk factors for overall mortality and respiratory failure

Overall mortality was not significantly different—neither for all patients (sAVR: 13% vs ARE: 9%, P = 0.71) nor for the matched cohort (sAVR: 8% vs ARE: 20%, P = 0.37, Fig. 1). Factors that were not modifiable by the surgeon identified by the multivariable analysis independently associated with overall mortality were age [HR 1.68 (1.49–1.89)], chronic obstructive pulmonary disease (COPD) [HR 1.69 (1.30–2.18)], diabetes [HR 1.80 (1.51–2.15)] and atrial fibrillation [HR 1.87 (1.55–2.26)], as displayed in Table 6. The only variable that the surgeon and the ICU team need to be aware of with a significant association to survival was postoperative RF with an HR of 2.84 (2.17–3.73). However, mortality risk for patients with RF was especially increased during the first 3 months (Supplementary Material, Fig. S1) with an HR of 2.84 × 3.52 = 10.

In a second model, taking PPM into account the overall mortality was also significantly negatively influenced by severe PPM [HR 1.69 (1.18–2.42)], see the Supplementary Material, Table S2.

However, a multivariable analysis for RF (Table 6) was performed showing a significant increase in risk in patients with ARE [HR 1.81 (1.10–2.96)], diabetes [HR 1.45 (1.10–1.90)], New York Heart Association (NYHA) class III/IV at the time of surgery [HR 1.64 (1.24–2.16)] accompanied by a risk increase due to the use of blood products [per unit: RBC concentrate HR 1.22 (1.17–1.27) and fresh frozen plasma HR 1.11 (1.07–1.16)]. Expectably, pulmonary hypertension [HR 1.57 (1.15–2.15)] and COPD [HR 1.85 (1.16–2.96)] had a negative effect on the outcome. Blood cardioplegia seemed to have a protective effect compared to crystalloid cardioplegia [HR 0.65 (0.43–0.98)].

One additional model was calculated to exclude differences due to the minimally invasive surgical approach or septal myectomy, see the Supplementary Material, Table S3.

DISCUSSION

The presented contemporary data from a high-volume centre support sAVR as a safe and effective benchmark procedure that offers excellent operative outcomes with an early mortality as low as 1.4%. This was slightly superior to the reported in-hospital mortality of 1.7% among all patients of the German aortic valve registry receiving isolated sAVR [9].

However, several clinical studies suggested that PPM has a negative impact on short- and long-term patient survival [1–3] and should be avoided by using the largest prosthetic valve size possible for the individual patient or utilizing ARE resulting in an increased EOA. Our aim was to ascertain intraoperative and postoperative complications associated with concomitant ARE and independent risk factors for early adverse events in a matched cohort of patients to eventually conclude whether the operative risk of a more complex procedure outweighs the benefits of reducing the PPM.

Patient–prosthesis mismatch

Over the past decades, patient age and frailty are progressively increasing. Aortic valve patients frequently exhibit numerous severe comorbidities with a significant impact on early adverse events and early mortality. Therefore, the extent of surgery needs to be chosen wisely for each patient by considering all options available (e.g. sAVR versus transcatheter aortic valve replacement (TAVR)). A low transvalvular gradient and implanting the largest prosthesis possible are essential in avoiding PPM. The impact of PPM has been a matter of debate [10, 11]—however, there is growing evidence showing that PPM has a negative influence on the short- and long-term outcomes [2, 12], and reoperation rates due to more rapid progression of valve deterioration in patients with an EOA <0.70 cm²/m² [13].

In accordance with several studies, we found severe PPM to be an independent risk factor for overall mortality. In line with data presented by Walther et al. [1] and Blackstone et al. [14] only a small group of our patients had severe PPM. About one-quarter of patients presented with moderate PPM; however, numerous studies have reported an incidence of 41–54% [15–17]. The largest study with 59 779 patients from the STS database (54% moderate, 11% severe PPM) concluded that any degree of PPM has a negative effect on long-term survival and increases readmission for heart failure and redo-operation [17]. It is, therefore, crucial to determine whether the benefit of a larger EOA outweighs the potential risks associated with concomitant ARE at the time of sAVR.

Outcome

In the present study, early and overall mortality were not different between the groups—neither in the unmatched nor the matched cohort. Also, the number of re-sternotomy due to bleeding or pericardial effusion showed no difference between the groups, which may indicate a hidden selection bias due to higher surgical experience among the surgeons routinely performing ARE. The results are in line with a clinical study of Coutinho et al. [18] in 2011 demonstrating—in an unmatched cohort—that ARE is safe and does not increase the risk of short-term mortality. Excellent long-term survival, comparable to an age- and gender-matched general population, was demonstrated in 239 patients with ARE [11]. Kulik et al. [19] also found ARE to be safe; however, 50% of the patients in this analysis had CABG or mitral valve surgery as a concomitant procedure and over 70% received ARE with the Manouguian technique. We were not able to confirm these findings with follow-up data from this contemporary patient group.

Respiratory failure

Even with no difference in mortality, RF occurred significantly more often in the ARE group. After matching for risk factors of RF (i.e. COPD and pulmonary hypertension), the rate of RF was still twice as high in these patients and was identified as an independent risk factor for overall mortality. Several studies have reported an incidence of up to 9% [20–22] of postoperative pulmonary complications after cardiac surgery. Filsoufi et al. [22] reported the highest incidence in patients receiving combined CABG and valve surgery or aortic procedure with a mortality rate of 15.5% among patients with RF. Age over 60 years, longer CPB time, preoperative pulmonary hypertension and intraoperative phrenic nerve injury were identified as independent risk factors [20, 21]. In our cohort, patients with ARE had significantly longer CPB times, which could be an explanation for the increase in postoperative RF.

ARE itself was identified as an independent risk factor for RF in the multivariable analysis. Other key risk factors were blood products per unit, advanced NYHA class and crystalloid cardioplegia.

In a large randomized controlled trial comparing restrictive and liberal transfusion of RBCs, no difference in 30-day mortality and respiratory dysfunction could be detected [23]. In non-cardiac surgery patients after pneumonectomy, the perioperative transfusion of RBCs has been shown to be a dose-dependent independent risk factor for RF [24]. In our analysis, RBC transfusion per unit had an OR of 1.22 (P < 0.001) for RF. In the general surgical population, Blum et al. [25] found that development of acute respiratory distress syndrome (ARDS) was also associated with RBC transfusion and crystalloid administration. As ARE patients in our analysis had more septal myectomy, it could be hypothesized that these patients had a higher level of general left ventricular hypertrophy due to prolonged left ventricular load as a result of tedious stenosis of the aortic valve requiring postoperative volume administration resulting in pulmonary fluid overload or even oedema more frequently.

This monocentric study confirms that early mortality and severe postoperative complications after ARE might not be any different from a standard sAVR. As long-term studies have clearly demonstrated the positive effects of avoiding severe PPM, ARE should be the considered standard for patients with a high risk of severe PPM (and no realistic chance of mitigating the effect of PPM by conservative weight loss). Because obesity is a progressive problem in our society—particularly in patients with the impending risk of future severe PPM—the benefits of preemptive ARE should be considered even in patients with only a moderate risk of PPM.

Limitations

There were several limitations associated with the study design, the small number of patients and the way patients were recruited. The presented data were analysed retrospectively. Data could only be analysed as documented. Moreover, we decided to analyse isolated sAVR to exclude as many influencing factors as possible, leaving us with 169 patients within each group. Finally, we analysed patients from 2004, at which time the fast track concept was not established, leading to a faster extubation of patients during the more recent years. However, we believe that because of the recruitment of patients within both groups at those earlier years, these should be well balanced.

CONCLUSION

In a high-volume centre, ARE at the time of sAVR is a safe and effective procedure with no significant increase in early mortality and with excellent short-term outcome. Severe PPM can be reliably eradicated by ARE. However, ARE is associated with an increase in postoperative RF. It is, therefore, paramount to balance the potential benefit of a larger EOA and the potential risks for the individual patient.

Conflict of interest: none declared.

Footnotes

REFERENCES

Author notes