Long-term results of the Mitroflow aortic pericardial bioprosthesis in over 800 patients: limited durability and mechanisms of dysfunction†

Abstract

INTRODUCTION

The Mitroflow aortic pericardial bioprosthesis (LivaNova) has been frequently implanted for its outstanding haemodynamic properties. Two prosthesis models have been implanted during the last 10 years. The LA/LXA was implanted until 2011, the DLA model thereafter. The most important difference between these 2 models is the presence of a phospholipid reduction treatment with octanediol in the DLA model.

Most of the available case series reported results of the LA/LXA model, the follow-up of the DLA model being still too short [1–12]. The LXA/LA model showed optimal long-term durability results in many reports [1–6].

However, some authors have recently reported unsatisfactory early durability results of the LA/LXA model, with early calcification [7–13]. Risk factors for SVD were the presence of a patient prosthesis mismatch and preoperative dyslipidaemia. SVD was a risk factor for mortality [7].

For this reason, we retrospectively reviewed our 10-year experience with the Mitroflow aortic prosthesis. The aims of this study were to present the long-term results of the Mitroflow bioprosthesis and to assess the risk factors for SVD and mortality.

METHODS

Patients

All patients who underwent aortic valve replacement (AVR) with a Mitroflow aortic bioprosthesis at our institution between November 2005 and January 2015 with a complete clinical follow-up were included in the study. Patients who were lost to follow-up after their postoperative course were excluded from the study.

Follow-up ended on 1 April 2016 and amounted to a mean of 45 (SD = 32) months (range 0, 120 months; 3142.7 patient-year) and a median of 42 [interquartile range (IQR) = 21, 70] months.

Follow-up was performed in order to reduce the number of patients lost to follow-up as much as possible. Patients and their referring physicians or cardiologists were first contacted by phone. Then, a questionnaire, which contained questions about the actual clinical status, cardiac reoperations and possible valve-related events, was sent to the patients. The referring physicians or cardiologists were asked to send us reports of ambulatory visits and echocardiographies. If it was not possible to reach the patients and their physicians, we contacted the health insurances and the general register office in order to obtain the missing information.

Descriptions of morbidity and mortality were based on the guidelines for reporting morbidity and mortality after cardiac valve surgery [14]. Evidence of SVD was based on echocardiographic data, such as the presence of a mean transprosthetic gradient >40 mmHg or an aortic regurgitation/stenosis more than moderate, and on intraoperative evidence in those patients who underwent redo AVR for SVD. The first and last echocardiographic reports were usually reviewed. The quality of the reports varied greatly from reports including complete echocardiographic assessment of the prosthesis with pressure gradients and prosthesis valve areas, to reports including only a qualitative assessment of the prosthesis function. In case of SVD evidence, all available previous echocardiographic reports were carefully reviewed in order to determine the earliest reported occurrence of SVD.

Patients consented handling their data for retrospective descriptive studies at the time of signing the informed consent to AVR. Therefore, the institutional ethics committee waived the need for further patient consenting to this study.

The first Mitroflow prosthesis was implanted in November 2005, the LA/LXA model until November 2011 and the DLA model since December 2011. The technique for AVR and Mitroflow implant, the postoperative patient management and the anticoagulation protocol did not change during the study period [15].

Data analysis

Data analysis was performed using SPSS 24.0 (IBM, Armonk, NY, USA). Primary end-points were SVD and mortality. SVD was further analysed by reporting separately the freedom from AVR due to SVD and, additionally, the freedom from SVD only in those patients who had complete follow-up echocardiographic reports that included quantitative measurements such as pressure gradients and prosthesis valve area. Secondary end-points were cardiac redo of any type, cerebrovascular and bleeding events, endocarditis and non-SVD [14]. Categorical and continuous variables were summarized as percentages, and mean (standard deviation) or median with 25–75% IQR, respectively. The independent-samples Student’s t-test and the χ² test were used for comparisons of continuous and categorical variables among included and excluded patients, respectively. The paired Student’s t-test was used for inpatient comparison of mean and maximal transvalvular gradients at discharge and at last echocardiographic controls.

Survival estimates along with freedom from other end-points were calculated by the product-limit method of Kaplan–Meier. Differences among groups were quantified using the log-rank test.

At the multivariable analysis, SVD and mortality were considered as time-to-event outcomes. The variables included in the univariable and multivariable analyses are reported in Tables 1, 2 and 5. The Cox regression analysis was performed to identify independent risk factors for SVD and mortality. The model was constructed including risk factors with univariable P-values <0.1. Results were reported as hazard ratios with 95% confidence interval and corresponding P-value. The proportional hazards assumption was tested using the complementary log–log Kaplan–Meier plots and including the time-dependent coefficients into the regression model.

As we faced competing risks of SVD and death in the described analyses, we also performed a multistate analysis with R to find out the rates of patients experiencing death after valve degeneration and of those being alive after valve degeneration. The previously described statistics were calculated to permit comparisons with performance descriptions from the literature.

Two-tailed P-values ≤0.05 were considered significant.

RESULTS

Patient characteristics

Between November 2005 and January 2015, among the 916 patients undergoing AVR with the Mitroflow bioprosthesis at our Institution, 832 (90.8%) had a complete clinical and echocardiographic follow-up and were included into the study. The remaining 84 (9.2%) patients were lost to follow-up and excluded from the study.

There was no difference between included and excluded patients regarding preoperative variables, such as male gender (P = 0.24), age (P = 0.077), EuroSCORE II (P = 0.38), and intraoperative variables such as prosthesis size [median 21 (IQR = 21, 23) mm vs median 21 (IQR = 21, 23) mm, P = 0.68], concomitant coronary artery bypass grafting (P = 0.32), cardiopulmonary bypass (P = 0.85) and cross-clamp times (P = 0.43).

Pre-, intra- and postoperative characteristics of included patients are reported in Tables 1 and 2, respectively. Mean and maximal transvalvular gradients at discharge are reported in Table 3.

Structural and non-structural valve deterioration

At follow-up, 52 (6.2%) patients developed SVD. Of these patients, at transthoracic echocardiography, 38 (73.0%) patients showed prosthesis stenosis, 7 (13.5%) patients prosthesis regurgitation and 7 (13.5%) patients combined prosthesis stenosis and regurgitation. Risk factors for SVD at the univariable and multivariable analyses are reported in Table 4. Increasing age was a protective factor against SVD. Freedom from SVD is reported in Table 5 and Fig. 1A. Patients younger than 60 years had a higher incidence of SVD than patients older than 70 years. Instead, there was no difference in SVD incidence after stratification according to prosthesis size and model. In this last case, follow-up of the LA/LXA model was longer than the follow-up of the DLA model (54, SD = 35 months for the LA/LXA model vs 29, SD = 14 months for the DLA model). The overall freedom from AVR due to SVD and the freedom from SVD in those 440 patients with complete follow-up echocardiographic reports are reported in Table 5 and Fig. 2A. There was no difference among the 440 patients and the whole study population regarding freedom from SVD.

Figure 1

The overall freedom from SVD (A), a competing risk analysis for SVD and mortality in all patients (B) and in those who were younger than 70 years (C) or older than 70 years (D) at time of Mitroflow implant are shown. The height of each coloured stripe at a certain time after the operation indicates the percentage of patients being in this state. (C and D) The prostheses of younger patients degenerate earlier, but this is not associated with a higher mortality rate. Patients who underwent a reoperation on their Mitroflow valve are not differentiated by this depiction; they can be either in the ‘alive after degeneration’ or in the ‘death after degeneration’ stratum.

Open in new tabDownload slide

Figure 2

Freedom from aortic valve replacement due to structural valve deterioration (A), overall survival (B), freedom from endocarditis (C) and freedom from cerebrovascular events (D). Patients at risk are reported above the x-axis.

Open in new tabDownload slide

Multistate analysis showed that 7.1% of patients died after SVD 9 years after implantation and 10.0% were alive after SVD at this time point (Fig. 1B–D). Thus, at 9 years, 82.9% did not experience SVD.

At follow-up, 25 (3.0%) patients showed non-SVD, of these, 16 (64.0%) patients showed endocarditis involving the Mitroflow prosthesis and the remaining 9 (36.0%) patients a paravalvular leakage. Freedom from non-SVD is reported in Table 6.

Cardiac reoperations

Freedom from cardiac redo for any reason is reported in Table 6. Among the 42 (5.0%) patients who underwent a cardiac redo, 5 (11.9%) patients did not require redo AVR (mitral valve surgery, n = 3; coronary artery bypass grafting, n = 2). The remaining 37 (88.1%) patients required redo AVR, for SVD in 20 (54.0%) patients and for non-SVD in 17 (46.0%) patients (endocarditis, n = 16; paravalvular leakage, n = 1). The Mitroflow prosthesis was replaced with a mechanical aortic prosthesis in 10 (27.0%) patients and with a biological prosthesis in 26 (70.3%) patients (median prosthetic size 21 mm, IQR 21–23 mm). In the single patient undergoing redo AVR for paravalvular leakage, the leak was sealed with a suture and the Mitroflow prosthesis was left in place.

Among the 20 redo AVRs for SVD (LA/LXA model, n = 18; DLA model, n = 2), 15 (75.0%) patients underwent surgical AVR (SAVR) and 5 (25.0%) patients a transcatheter aortic valve-in-valve implantation (TAVI V-i-V). Median prosthetic size was 23 mm (IQR 21–23 mm). In SAVR patients, 11 (73.3%) patients received a biologic prosthesis. Macroscopic description of the degenerated prosthesis was available in 12 (80.0%) patients (calcification, n = 10; cusp tear, n = 2; pannus, n = 1). Six prostheses were analysed microscopically and showed various degrees of calcification (n = 3), fibrosis (n = 5) and macrophagic inflammatory infiltrate (n = 4). Postoperatively, 2 of the redo SAVR patients required pacemaker implantation. There was no in-hospital death. In TAVI V-i-V patients, 4 patients underwent transfemoral TAVI and 1 patient transapical TAVI. A CoreValve prosthesis (Medtronic, Inc.) was implanted in 4 patients (Mitroflow size 21 mm, n = 2; 23 mm, n = 2) and a Sapien XT (Edwards LifeSciences) in 1 patient (Mitroflow size 21 mm), without any procedural complication or coronary flow obstruction. There was no in-hospital death.

Mortality

Overall survival, as well as survival after hospital discharge, is reported in Table 6 and Fig. 2B. During the study period, 311 (37.3%) patients died, 69 (8.3%) in hospital and the remaining 242 (29.0%) after hospital discharge. Among the patients who died after hospital discharge, 127 (52.4%) patients died of cardiac-related causes. Among these patients, 16 (12.6%) died of valve-related events. Of the cases with late valve-related mortality, 8 patients had concomitant prosthesis endocarditis and another 8 patients concomitant SVD (n = 7) or non-SVD (n = 1). Multivariable analysis for overall mortality is reported in Table 4. SVD did not emerge as an independent risk factor for mortality.

Other end-points and echocardiography

During the study period, 27 (3.2%) patients developed endocarditis, which involved the Mitroflow prosthesis in 19 (70.3%) patients and required redo AVR in 16 (59.2%) patients. Four (14.8%) patients had already presented endocarditis as indication for native AVR. Freedom from endocarditis is reported in Table 6 and Fig. 2C.

At follow-up, 9 (1.1%) patients had major bleedings (intracranial bleeding, n = 4; gastrointestinal bleeding, n = 2; haemoptysis, n = 1; nasal bleeding, n = 1; other, n = 1) and 50 (6.0%) patients had cerebrovascular events (stroke, n = 20; transitory ischaemic attack, n = 30). Freedom from major bleeding and cerebrovascular events is reported in Table 6 and Fig. 2D.

In those 440 patients where gradients and prosthesis valve areas were reported, after 38 (SD = 26) months, mean and maximal transvalvular gradients increased significantly (P < 0.001 in both cases, respectively) in comparison with hospital discharge (Table 3).

DISCUSSION

This retrospective study showed that the durability of the LA/LXA Mitroflow model, which was still very good at 5-year follow-up, dropped to 67.9% at 9-year follow-up according to the Kaplan–Meier analysis. The 440 patients with complete echocardiographic reports at follow-up showed also a similar low 9-year freedom from SVD. However, according to the multistate analysis, at 9 years, 82.9% of patients remained free from SVD. The DLA model, which has replaced the LA/LXA model on the market since 2011, showed optimal early durability.

The long-term durability of the Mitroflow still remains controversial. In several past publications, the LA/LXA Mitroflow model showed outstanding haemodynamic performance due to its peculiar design as well as satisfactory long-term durability. Several studies reported 5-year, 10-year, 18-year and 20-year freedom from SVD of over 90%, 80%, 65.5% and 62.3%, respectively [1–6].

On the contrary, Piccardo et al. reported an actuarial freedom from echocardiographic signs of SVD at 10 and 15 years of 77% and 56% [10]. Ten-year freedom from valve explant due to SVD was worse for the Mitroflow than for the Carpentier-Edwards (Edwards LifeSciences) pericardial prosthesis, as reported by Nielsen et al. [9]. De Paulis et al. have recently reported a 9-year freedom from SVD of only 77.4% even in patients older than 70 years and with small aortic annulus [11]. Sénage et al. also reported an earlier onset of SVD, especially in patients with small prosthesis size [7]. However, some authors have recently questioned the definition of SVD used by Sénage et al. [16]. Risk factors for Mitroflow SVD were the presence of patient–prosthesis mismatch, hyperlipidaemia, younger age at implant, absence of anti-mineralization treatment and small (19 mm and 21 mm) prosthesis size [7–10, 17].

In the present study, the Kaplan–Meier and multistate analyses both show a limited 9-year durability of the LA/LXA model, which was worse than the 84% and the 98% 10-year freedom from SVD reported for other aortic prostheses, such as the Carpentier-Edwards Perimount [18] and the Medtronic Mosaic (Medtronic, Inc.) [19] prostheses, respectively. Moreover, our 9-year freedom from AVR due to SVD was also worse than the one reported by other authors [5, 9–11]. The presence of SVD was not a risk factor for mortality. We identified only younger patient age as a risk factor for SVD. Small prosthesis size was not identified as a risk factor, since only 3% of patients received a 19-mm prosthesis. In contrast to a previous case series [6], the DLA Mitroflow model with its anticalcification treatment did not emerge as protective factor against SVD in our study. The short follow-up of the DLA model explains this result. The role of younger patient age as risk factor for SVD has been well documented, it is independent of prosthesis design, and it has been associated with a higher immunologic response against the bioprosthesis tissue in younger than older patients [8, 9]. This should be kept in mind especially in the upcoming TAVI era, when indication for TAVI will move from high- to low-intermediate surgical risk patients and from older to younger patients.

Concerning the pathologic mechanisms of SVD, the degenerated LA/LXA Mitroflow prostheses showed more often calcification than cusp tear, in accordance with the results reported by Luk et al. [13]. Consequently, patients with a degenerated Mitroflow prosthesis developed prosthesis stenosis more often than regurgitation. This fact has important clinical consequences since patients with SVD withstand more easily a pressure than a volume overload. In contrast, degenerated stentless prostheses usually develop more often cusp tears and abrupt aortic regurgitation, which may lead to pulmonary oedema [20]. In our case series, no patient underwent emergent redo AVR for SVD. Since calcification has an important role in prosthesis degeneration, anticalcification treatments have been recently applied in most of the available bioprosthesis, including the DLA Mitroflow model [17]. Longer follow-up is required to evaluate the impact of anticalcification on DLA model durability.

Our study showed that TAVI V-i-V can be an alternative to SAVR for redo AVR due to SVD. However, some authors reported a higher risk of coronary artery obstruction in patients undergoing TAVI V-i-V for degenerated Mitroflow prosthesis, due to the peculiar prosthesis design [21, 22]. Patients with a 19-mm degenerated Mitroflow were especially at higher risk. Other authors reported that even a 21-mm degenerated Mitroflow prosthesis may be too small to host a 23-mm Sapien XT prosthesis [23]. In our study, TAVI patients did not show any degree of coronary obstruction, probably because they had larger prosthesis sizes (21 mm or 23 mm). Proper procedural planning using multi-slice computed tomography is of paramount importance. The position of the coronary ostia, the distance between them and the prosthesis ring and the dimension of the sinuses of Valsalva must be carefully evaluated [24].

Thus, waiting for the long-term durability results of the DLA model with its anticalcification treatment, we suggest implanting the Mitroflow prosthesis only in patients older than 70 years. The competing risk analyses show that in patients older than 70 years more than 85% remain free from SVD within nearly 10 years after implantation. Moreover, according to the available literature [7, 10, 11], caution must be taken when implanting the Mitroflow prosthesis in patients with smaller annulus diameter.

Finally, in-hospital mortality was 8% and mean cross-clamp time amounted to 67 min in our study. However, the prevalence of preoperative comorbidities such as endocarditis (Table 1) and the prevalence of associated cardiac procedures (Table 2) were higher in our study population than elsewhere [5, 9].

Limitations

The retrospective nature of this study implies several limitations. Due to lack of follow-up, 84 patients were excluded from the study. However, we demonstrated that there was no difference in pre-, intra- and postoperative variables between included and excluded patients.

Echocardiographic examinations were performed by different cardiologists and at non-standardized time points. Although the guidelines are quite explicit in recommending the frequency of echocardiographic controls in patients with an aortic prosthesis, many patients still do not undergo any control at all. Similarly, aortic valve area and trans-prosthetic gradients were not described in all echocardiographic reports, and thus were available for 580 and 440 patients at discharge and at last follow-up control, respectively (Table 3).

CONCLUSIONS

The LA/LXA Mitroflow model showed limited long-term durability, but SVD was not associated with worse survival. Degenerated prostheses showed more stenosis than regurgitation. Patient age played an important role in the development of SVD. While waiting for longer follow-up, the new DLA model should be implanted preferably in patients older than 70 years.

Conflict of interest: Axel Haverich is consultant for Edwards.

REFERENCES

Author notes