Transcatheter Aortic Valve Replacement for Degenerative Bioprosthetic Surgical Valves: Results From the Global Valve-in-Valve Registry

The Global Valve-in-Valve Registry was initiated in December 2010. The registry is not supported by any external funding and was designed to collect data from centers across the world that had TAVR experience with the use of either CoreValve or Edwards SAPIEN devices. There was retrospective collection of data from cases performed before registry initiation and prospective data collection after that time.

A total of 38 centers from Europe, North America, Australia, New Zealand, and the Middle East contributed all of their VIV experience to the registry (Appendix I in the online-only Data Supplement). Continued communication with involved centers (D.D.) regarding clinical events was initiated. Data were collected with the use of a dedicated case report form. All inconsistencies were resolved directly with local investigators and on-site data monitoring. For each bioprosthesis, we collected the operation year, valve type, label size, and internal diameter. Baseline bioprosthetic valve area, gradients, and degree of regurgitation were gathered from the last echocardiographic study before the procedure. Early postimplantation hemodynamic analysis included data from intraprocedural echocardiographic analysis or from the first echocardiographic evaluation after the procedure. Images from these procedures were collected into a library of VIV procedures. The inclusion of patients included in this registry was approved in each center by a local ethical committee.

Bioprosthetic valve mechanism of failure (eg, stenosis or regurgitation) was evaluated according to the criteria of the American Society of Echocardiography.³⁰ Patients with at least a moderate degree of both stenosis and regurgitation were included in the “combined stenosis and regurgitation” group. Other patients were categorized, according to the primary mechanism of failure, in either the “stenosis” group or the regurgitation group.

Successful VIV implantation was defined as a procedure having all of the following: successful vascular access, delivery, and deployment of a device; successful retrieval of the delivery system; intended performance of the device with neither severe stenosis (mean aortic gradient >40 mm Hg or peak velocity >4 m/s) nor moderate or severe regurgitation; and the patient being transferred alive out of the catheterization suite. Major clinical end points were assessed according to the Valve Academic Research Consortium (VARC) criteria.³¹ Follow-up data were collected for patients according to the time frame elapsed from the index VIV procedure to data lock for present analysis.

Statistical analysis was performed with the use of SAS version 9.1 (SAS Institute, Cary, NC). Results are presented as mean±SD for continuous variables with normal distribution, as median and interquartile range (25th, 75th percentiles) for continuous variables without normal distribution, and as percentages for categorical data. Student t test was used to compare normally distributed continuous variables between the CoreValve and Edwards SAPIEN groups, and the Wilcoxon rank sum test was used for variables not normally distributed. One-way ANOVA was used to compare the stenosis, regurgitation, and combined groups for normally distributed continuous variables, and the Kruskal-Wallis test was used for non–normally distributed data. The χ² and Fisher exact tests were used to compare categorical variables. The Kaplan–Meier method and comparison between the CoreValve and Edwards SAPIEN groups were performed with the use of the log-rank statistic. High postprocedural gradients were defined according to the VARC criteria as those having mean gradients ≥20 mm Hg.³¹ Changes in valve gradients were evaluated with repeated measures analysis. For multivariable analysis of predictors for high postprocedural gradients, a logistic regression was used. The initial selection of variables entered into the univariate model included the following: sex, age, functional class, baseline echocardiographic parameters (aortic valve area, aortic regurgitation degree, and left ventricular ejection fraction), bioprosthetic type (stented versus stentless), internal diameter, the device used during the VIV procedure (CoreValve versus Edwards SAPIEN), device size, preimplantation valvuloplasty, and postimplantation valvuloplasty. Variables with P<0.10 in the univariate analysis, as well as other critical parameters possibly affecting the analysis (eg, bioprosthesis size or performance of postimplantation valvuloplasty), were further examined in a stepwise model. The results of the multivariate analysis are presented as odds ratio with 95% confidence interval (CI). A 2-sided P value <0.05 was considered statistically significant.

The authors are solely responsible for the design and conduct of this study, all analyses, drafting and editing of the manuscript, and its final contents.

Table 1shows the clinical characteristics of 202 patients included in the registry; 124 patients underwent VIV procedure with the use of the CoreValve device and 78 patients with the use of the Edwards SAPIEN device. Patients' mean age was 77.7±10.4 years (distribution, 25–91), and 52.5% were men. The CoreValve group had lower left ventricular ejection fraction and a higher rate of diabetes mellitus than did the Edwards SAPIEN group (49±13% versus 52.8±10.4%, P=0.03; 34.7% versus 20.5%, P=0.03, respectively).

The patients included in the trial had between 1 and 4 previous SAVRs (85.6% had 1 previous surgery). The median time from the last SAVR to the VIV procedure was 9 years (interquartile range, 6 to 13 years). Implanted bioprostheses were stented in 76.7% of cases (n=155) and stentless in 23.3% (n=47). Edwards SAPIEN was used to treat a higher proportion of stented bioprosthetic valves than CoreValve (87.2% versus 70.2%; P=0.005). There was no significant difference between the CoreValve and Edwards SAPIEN groups in the internal diameter size of bioprostheses. Fifty-two patients had small bioprostheses with internal diameter <20 mm (28.2% of CoreValve cases versus 21.8% of Edwards-SAPIEN; P=0.32). Appendix II in the online-only Data Supplement includes data on the degenerative bioprostheses included in the registry.

The mechanism of failure was stenosis (n=85; 42.1%), regurgitation (n=68; 33.7%), and combined stenosis and regurgitation (n=49; 24.3%). The distribution of mechanism of failure did not differ between the CoreValve and Edwards SAPIEN groups. The stenosis group had lower aortic valve orifice area and higher maximum/mean valve gradients than the regurgitation and combined groups (0.69±0.3 versus 1.63±0.53 versus 0.91±0.2 cm², respectively; 83±24.2/50.5±16.2 versus 34.6±14.1/17.8±7.8 versus 67.7±16.8/38.8±11.1 mm Hg, respectively).

Table 2shows the procedural characteristics and early results of VIV procedures in the total cohort and is further separated into the CoreValve and Edwards SAPIEN groups. In both groups, the majority of devices were of the smallest available size: 26-mm CoreValve (79.8%) and 23-mm Edwards SAPIEN (79.5%). The main access route in the CoreValve group was femoral (91.1%) versus apical in the Edwards SAPIEN group (69.2%). General anesthesia, transesophageal echocardiogram, and preimplantation valvuloplasty use were all more common in the Edwards SAPIEN group.

Table 2 depicts procedural results and 30-day outcome. There were 14 procedural failures (6.9%) according to the registry definition and 83 device failures according to the VARC definition (41.1%). Most of the VARC device failures (62.7%) were secondary to elevated postprocedural gradients of moderate degree (mean gradients, 20–39 mm Hg). Procedural success, according to the registry definition, and device success according to the VARC definition were both higher in the CoreValve group (96.8% versus 87.2%, P=0.009 and 64.5% versus 50%, P=0.04, respectively).

Thirty-one cases had initial device malposition (15.3%; 16.9% CoreValve versus 12.8% Edwards SAPIEN; P=0.41) (Figure 1A and 1B). Additional maneuvers during the procedures included 11 attempts to retrieve the device during CoreValve procedures (8.9%) and implantation of a second TAVR device in 17 cases (8.1% CoreValve versus 9% Edwards SAPIEN; P=0.82). Postimplantation valvuloplasty was used in 25 cases (16.9% CoreValve versus 5.1% Edwards SAPIEN; P=0.01). Ostial left main coronary obstruction occurred in 7 cases (3.2% of CoreValve cases versus 3.8% of Edwards SAPIEN cases; P=1.0) (Figure 1C and1D). In-hospital mortality in the patients with ostial coronary obstruction was 57.1%. Three cases with ostial coronary obstruction appeared during VIV procedures performed inside Mitroflow stented valves (Sorin Group Inc, Vancouver, British Columbia, Canada), 2 cases in Freedom stentless valves (Sorin), 1 case in a stentless CryoLife O'Brien valve (CryoLife International, Atlanta, GA), and 1 in a stented Mosaic valve (Medtronic). The rate of ostial coronary obstruction was 0.9% for stented bioprostheses (other than Mitroflow) and 2.2% for stentless bioprostheses (other than Freedom), without a statistically significant difference (P=0.47). There was an ostial coronary obstruction event in 7.7% of the Mitroflow cases (more than in other stented valves; P=0.049) and in 50% of the Freedom stentless (more than in other stentless valves; P=0.02).

Median duration of hospital stay was 8 days (interquartile range, 6–12 days). At 30 days, 185 patients survived, 17 patients died (8.4%), and no patient was lost to follow-up. There were no differences between the CoreValve and Edwards SAPIEN groups in mortality, major vascular complication, or stroke. Overall, postprocedureal aortic valve gradients decreased to 28.4±14.1 mm Hg (maximum) and 15.9±8.6 mm Hg (mean). Mean postprocedural gradients were 5 mm Hg higher in the Edwards SAPIEN group than in the CoreValve group (P<0.0001). After the procedure, significant aortic regurgitation (≥+2) appeared in 10 cases (5%), and left ventricular ejection fraction was 51.3±11.8%, which is similar to that measured before the procedure at 50.5±12.2% (P=0.62). There was no difference between the CoreValve and Edwards SAPIEN groups in the need for a permanent pacemaker implantation (8.9% versus 5.1%, respectively; P=0.31).

Postprocedural gradients were assessed in 197 patients (5 patients died during the index procedure before the analysis). High postprocedural gradients (mean gradients ≥20 mm Hg) were recorded in 56 patients (28.4%): 26 in the CoreValve group versus 30 in the Edwards SAPIEN group (Figure 2). The rate of elevated postprocedural gradients was higher after Edwards SAPIEN than after CoreValve implantations (40% versus 21.3%, respectively; P<0.0001). There was a significant difference in the rate of high postprocedural gradients between the Edwards SAPIEN and the CoreValve groups for VIV procedures performed inside small surgical bioprostheses (internal diameter <20 mm): 58.8% versus 20%, respectively (P=0.005);Figure 3 shows representative case examples.Table 3shows the results of analysis for predictors for high postprocedural gradients. The only independent predictors were baseline surgical bioprosthetic valve area (odds ratio, 0.87 per 0.1-cm² increment; CI, 0.79–0.94; P=0.001) and implantation of the Edwards SAPIEN device versus CoreValve (odds ratio, 2.28; CI, 1.17–4.43; P=0.02). Preimplantation valvuloplasty and TAVR implantation inside a stented bioprosthesis were not independent predictors for elevated postprocedural gradients in multivariate analysis.

The overall patients' Kaplan–Meier survival curve is depicted inFigure 4. Median follow-up time was 285 days (interquartile range, 114–509 days). There were 23 deaths within the first postprocedural year. At 1 year, the total number of patients at risk was 87 (CoreValve, n=48; Edwards SAPIEN, n=39), and the follow-up rate was 54.4%. The calculated 1-year survival was 85.8% (CI, 79.9–91.6%). There was no significant difference in 1-year survival between patients undergoing CoreValve VIV procedures (89%; CI, 82.1–95.8%) and those undergoing Edwards SAPIEN VIV procedures (81.8%; CI, 71.9–91.6%; log rank P=0.25).

Figure 5shows hemodynamic and clinical results up to 1 year of follow-up. Thirty-day mean gradients, functional class, and aortic regurgitation results were maintained at 1-year follow-up (P=0.13, P=0.33, P=0.49, respectively).

The VIV approach results in considerable hemodynamic improvement, including a decrease in valve gradients and aortic regurgitation level. Clinical results were maintained at 1-year follow-up among analyzed patients and are comparable with other TAVR cohorts.^8–10 Nevertheless, even though implantation success according to the registry definition was high (93.1%), the rate of VARC-defined “device success” was relatively low (58.9%), which was primarily due to elevated postprocedural gradients of moderate degree. It should be emphasized that the VARC definition was created and originally intended for TAVR procedures performed in native aortic valves and not for VIV procedures.

Significantly elevated postprocedural gradients are common after VIV. In the present registry, the rate of high postprocedural gradients (mean, ≥20 mm Hg) was relatively high (28.4%), and the mean postprocedural gradient (15.9 mm Hg) was higher than after procedures performed inside native aortic valves (≈10 mm Hg).^8–10 This is probably due to reduced area available for the functioning valve when implanted inside a surgical bioprosthesis (ie, reduced potential orifice area) or patient-prosthetic mismatch. In an in vitro model of hemodynamic performance of a 23-mm Edwards SAPIEN device within a degenerated surgical bioprosthesis, incomplete stent expansion resulted in leaflet distortion when implanted in 19- and 21-mm Carpentier Edwards bioprostheses.²⁸

In the present analysis, the 2 independent predictors for high postprocedural gradients (severity of bioprosthetic stenosis and use of an Edwards SAPIEN device) are both associated with reduced potential orifice area. The difference between Edwards SAPIEN and CoreValve results is probably secondary to the fundamental dissimilarity between the devices: The functioning part of the Edwards SAPIEN valve is located inside the native valve annulus (intra-annular), and that of the CoreValve is usually located above that plane (supra-annular). Consequently, the CoreValve device depends much less on surgical bioprosthesis dimensions, and its functioning part may have larger potential orifice area. In support of this theory, in an in vitro evaluation, a custom-designed supravalvular TAVR device has achieved more favorable hemodynamic measurements.³² In the present registry, at 1-year follow-up there was no statistically significant increase in valve gradients in comparison to early postprocedural gradients. However, the possible impact of elevated gradients should be examined with longer follow-up.

Thirty-day mortality and stroke rates after VIV procedures (8.4% and 2%, respectively) are comparable to those in other TAVR cohorts.^8–10 Nevertheless, there are 2 major safety concerns when VIV is performed: device malposition and ostial coronary obstruction. The relatively high malposition rate during VIV cases (15.3%) could be secondary to the relative lack of valvular calcification and, in some cases, difficulty in defining the optimal target for implantation during the procedure, particularly in stentless bioprostheses, in which no anatomic markers are available.

Ostial left-main obstruction, which is reported only rarely during native valve TAVR, seems to be more common during VIV procedures (3.5%). This complication has a dreadful prognosis. Most of the patients having ostial left-main obstruction died during their hospital stay. The propensity for this complication is clearly related to the spatial geometry of the surgical valve leaflets inside the aortic sinuses and seems to be more common within the Mitroflow and Freedom valves. The Mitroflow leaflets are unique, being relatively long (≈13 mm) and mounted externally over the stent rather than internally, as in most other stented bioprostheses.²⁹

The VIV procedure is technically demanding and should be reserved for highly experienced centers. Operators should be skilled in the handling of device malposition, retrieval techniques, and implantation of a second TAVR device, if needed. During screening, the Heart Team must have all of the information about and be familiar with the particular characteristics of the surgical bioprosthesis: the mode of degeneration and echocardiographic parameters (isolated paravalvular regurgitation should be excluded); valve size (most importantly, bioprosthesis internal diameter should be used for TAVR device and size selection; because of better hemodynamic results, CoreValve may be preferred in cases of small bioprosthesis with internal diameter <20 mm); valve position (intra-annular versus supra-annular); valve type (stented versus stentless); the relation of the bioprosthetic valve to its radio-opaque markers; and the risk for coronary obstruction when VIV is performed inside that specific bioprosthesis.

Operators should be aware of the risk of device malposition, and every effort should be taken to ensure correct localization and overlap with the bioprosthesis sewing ring. One cannot underestimate the importance of having coaxial positioning during the final stage of device deployment. The use of transesophageal echocardiography during these procedures is encouraged, especially during procedures performed inside stentless valves, where no radiopaque markers can assist in positioning. The use of software that will possibly enable improved device localization, such as the C-THV (Paieon Inc, Israel) or the Heart Navigator (Philips, Eindhoven, Netherlands), may be helpful and should be further evaluated.³³

Although no case of acute hemodynamic instability after preimplantation valvuloplasty was encountered in the registry, the practice of balloon dilatation inside degenerative bioprosthesis before valve implantation should be reserved for critically stenotic bioprostheses and probably avoided in patients with severe regurgitation. In comparison with native aortic valve leaflets, surgical valves seem to be susceptible to tearing after balloon valvuloplasty.³⁴,³⁵ That could lead to acute severe valve regurgitation, debris embolization, and stroke. It must be emphasized, however, that avoiding preimplantation valvuloplasty could result in a need for postimplantation balloon inflation, as occurred in the registry CoreValve group.

When preimplantation valvuloplasty is utilized, the maneuver may be used to examine the risk for ostial coronary obstruction by injecting contrast above the inflated balloon and evaluating the flow into the coronary arteries. Patients at high risk of ostial coronary occlusion should probably be managed conservatively, or, in selected Edwards SAPIEN cases, a preventive wire may be inserted into the left main coronary artery before device implantation.

Interestingly, the rate of pacemaker implantation after the CoreValve VIV procedures (9%) was relatively low compared with that after native aortic valve CoreValve procedures. This is most likely due to some degree of protection created by the surgical bioprosthesis frame, resulting in diminished compression of the TAVR device on the interventricular basal septum.³⁶ Obviously, strict ECG monitoring should be the standard of care after all TAVR procedures; however, it seems that after CoreValve VIV implantations, prolonged preventive use of a temporary pacemaker is not warranted when no conduction abnormality appears.

In the future, dependent on expanded clinical data, the VIV approach may affect SAVR practice in several respects: transcatheter treatment of patients with failed bioprosthetic valves with markedly elevated risk for reoperation, selection of valve class (biological much more than mechanical) and valve type (with minimal risk for ostial coronary obstruction) during surgery, and performance of surgical annular dilatation in small aortic roots.

A randomized controlled trial comparing reoperative SAVR and VIV in patients with failed bioprostheses has never been executed, and because VIV treatment is still infrequent, it will be quite difficult to conduct such a trial. As a result, there are not enough data to justify VIV instead of reoperation in most high-risk patients with failed aortic bioprostheses. Nevertheless, VIV could be an acceptable approach in carefully selected high-risk patients and in those considered as having no option (ie, those with no other effective treatment option for their illness).

VIV practice may also affect the selection of valve type during SAVR. A less invasive approach for failed bioprostheses could definitely be an argument in favor of using bioprosthesis in younger patients undergoing SAVR. In addition, surgical valves with increased risk for ostial coronary obstruction after VIV should probably be used less frequently during SAVR because the risk of later bioprosthetic failure and the possible need for a subsequent VIV procedure should be taken into account.

Because the VIV approach is novel, 1-year follow-up was reached in only 87 patients, less than half of the original group evaluated. Therefore, long-term results of the registry data should be analyzed with caution. Registry results reveal that elevated gradients are relatively common after VIV procedures. However, the study was not powered to correlate these echocardiographic results and clinical outcomes. For that evaluation, another trial including a larger patient group followed for a longer time period might be beneficial.

The VIV procedure is clinically effective in the vast majority of patients with degenerated stenotic or regurgitant bioprosthetic valves. Short- and intermediate-term results after these procedures are favorable. Safety and efficacy concerns include device malposition, ostial coronary obstruction, and high postprocedural gradients.