Author + information
- Received September 24, 2013
- Revision received December 9, 2013
- Accepted December 20, 2013
- Published online April 1, 2014.
- Edith Lubos, MD∗,
- Michael Schlüter, PhD∗,
- Eik Vettorazzi, MSc†,
- Britta Goldmann, MD∗,
- Daniel Lubs, MD∗,
- Johannes Schirmer, MD‡,
- Hendrik Treede, MD‡,
- Hermann Reichenspurner, MD, PhD‡,
- Stefan Blankenberg, MD∗,
- Stephan Baldus, MD∗ and
- Volker Rudolph, MD∗∗ ()
- ∗Department of General and Interventional Cardiology, University Heart Center, Hamburg, Germany
- †Department of Medical Biometry and Epidemiology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany
- ‡Department of Cardiovascular Surgery, University Heart Center, Hamburg, Germany
- ↵∗Reprint requests and correspondence:
Dr. Volker Rudolph, Uniklinik Köln, Herzzentrum, Klinik III für Innere Medizin, Kerpener Str. 62, 50924 Köln, Germany.
Objectives The aim of the study was to assess predictors of acute procedural failure in surgical high-risk patients undergoing MitraClip (Abbott Vascular, Abbott Park, Illinois) therapy.
Background MitraClip implantation is a novel percutaneous option to treat significant mitral regurgitation (MR).
Methods In 300 patients (75 ± 9 years of age, 190 [63%] men), of whom 32 (10.7%) had been unsuccessfully treated (discharge MR grade of >2+), baseline clinical and echocardiographic variables were evaluated by exact logistic regression and classification tree analyses to assess their impact on acute procedural failure. Acute procedural failure was differentiated into aborted procedure (no MitraClip implanted; n = 11) and “clip failure” (inadequate MR reduction despite MitraClip implantation; n = 21).
Results Multivariate logistic regression identified effective regurgitant orifice area (EROA), mitral valve orifice area (MVOA), and mean transmitral pressure gradient (TMPG) as independent predictors of overall acute procedural failure. Classification tree analysis revealed that an EROA >70.8 mm2 (n = 28) was associated with a high rate (25%) of clip failures, whereas the combination of an MVOA ≤3.0 cm2 and a TMPG ≥4 mm Hg (n = 16) was associated with a high rate (37.5%) of aborted procedures. Failure rates of ≤10% were observed in all patients with an EROA ≤70.8 mm2 and either an MVOA >3.0 cm2 (n = 217) or an MVOA ≤3.0 cm2 in concert with a TMPG ≤3 mm Hg (n = 39). Multinomial logistic regression identified an EROA >70.8 mm2 and a TMPG ≥4 mm Hg as independently predictive of clip failure, but an MVOA ≤3.0 cm2 and a TMPG ≥4 mm Hg as independently predictive of procedure abortion.
Conclusions In surgical high-risk patients undergoing MitraClip therapy, a TMPG ≥4 mm Hg, an EROA ≥70.8 mm2, and an MVOA ≤3.0 cm2 carry an increased risk of procedural failure.
Percutaneous mitral valve repair using the MitraClip system (Abbott Vascular, Abbott Park, Illinois) has become an accepted therapeutic option in patients with significant mitral regurgitation (MR) deemed at high surgical risk. The feasibility and safety of the percutaneous approach have been demonstrated, and its efficacy is currently further tested in ongoing clinical trials.
Despite reported primary procedure failure rates between 9% and 23% (1,2), predictors of the acute procedural outcome have not been assessed to date. To further improve the technique and possibly allow appropriate patient selection, identification of such predictors is of paramount importance. We sought to assess the acute procedural outcomes of 300 consecutive patients who underwent MitraClip implantation at our institution and determine variables affecting procedural failure.
Between September 2008 and April 2012, 300 consecutive patients with an MR of grade 3+ (moderate to severe) or 4+ (severe) were adjudicated as not amenable to surgery by heart team consensus and underwent MitraClip therapy at our center. Baseline characteristics of the study population are given in Table 1. These were elderly patients with a mean age of 75 years; they had high median logistic EuroSCORE (European System for Cardiac Operative Risk Evaluation) of 24% and an increased prevalence of comorbidities such as renal failure (55%), atrial fibrillation (65%), and coronary artery disease (66%). The majority of patients (68%) had mitral regurgitation of functional etiology (FMR). Anatomic exclusion criteria of the EVEREST II (Endovascular Valve Edge-to-Edge Repair Study II), (3) such as a left ventricular end-systolic diameter >55 mm and a mitral valve orifice area <4.0 cm2, were met by 39% and 57% of the patients, respectively. All patients had provided written informed consent before the intervention.
Echocardiographic measurements were performed according to American Society of Echocardiography and European Association of Echocardiography guidelines (4,5). In particular, the effective regurgitant orifice area (EROA) was determined by the proximal isovelocity surface area method, and the mitral valve orifice area (MVOA) was calculated with the pressure half-time method using the continuous-wave Doppler inflow profile; the mean transmitral pressure gradient (TMPG) was also assessed from the continuous-wave Doppler inflow profile. MR severity after the intervention was assessed as described previously (6).
Percutaneous mitral valve repair
Percutaneous mitral valve repair with the MitraClip system was described in detail previously (2). Acute procedural success was defined as the placement of ≥1 clips resulting in a discharge MR grade of 2+ or lower. An aborted procedure was defined as a procedure in which no clip could be implanted. Intraprocedurally, a TMPG of no more than 5 mm Hg was accepted after clip implantation. However, 4 patients with a higher TMPG already at baseline were accepted for MitraClip implantation because of the lack of alternative therapeutic options.
Continuous variables are presented as mean ± SD or as median and interquartile range (IQR), where appropriate; categorical variables are presented as absolute numbers and percentages. Comparisons of continuous variables were performed using the Mann-Whitney U test. Comparisons of categorical variables were performed by the Fisher exact test (for 2 × 2 tables) or chi-square test (for [n > 2] × 2 tables). Exact univariate and multivariate logistic regression analyses comprising pertinent clinical and echocardiographic variables, including all EVEREST trial (Endovascular Valve Edge-to-Edge Repair Study II) exclusion criteria (3), were used to determine variables potentially predictive of acute procedural failure. Regression analyses could not be performed separately for degenerative/mixed origin mitral regurgitation (DMR) and FMR because of the low number of failures encountered. Accordingly, these analyses only included variables applicable to both MR etiologies. Variables entered into the logistic regression analyses are listed in the Online Appendix. Exact multivariate logistic regression analysis used Firth's penalized likelihood approach (7).
To distinguish between procedures that had to be aborted and procedures that did not result in adequate MR reduction despite clip implantation (“clip failure”) and to establish the interrelationship of variables affecting either type of procedural failure, all variables used in the exact logistic regression were also entered into a classification tree analysis. This nonparametric method constructs a binary tree in which the dataset is split at each node into 2 subsets using the covariate that has the strongest association with the procedural outcome. Missing data in split variables are handled by a “majority vote” for the split (i.e., for a missing value, it is assumed that it falls in the larger group of observed values for the corresponding split). The final result is a tree that depicts the best variables to split the data and provides information about relevant variables for each class of objects (i.e., success, aborted procedure, and clip failure). Bonferroni corrections for multiple testing were used when “growing” the tree. In a final step, an exact multinomial logistic regression, with cutoffs determined from the classification tree analysis, was performed. Comparisons between terciles for assessment of the procedural learning curve were calculated by chi-square test for trend in proportions. A 2-sided p value <0.05 was considered statistically significant. R software, version 2.15.2 (R Foundation for Statistical Computing, Vienna, Austria), was used for statistical analyses. For logistic regressions, the exact methods implemented in the R packages logistf and pmlr were used; classification tree analysis used the R party package (8).
Acute procedural success was achieved in 268 patients (89.3%), with a single clip implanted in 177 (66%) and multiple clips implanted in 91 patients. Median device time (i.e., time from transseptal puncture to withdrawal of the clip delivery system from the left atrium) amounted to 63 min (IQR: 41 to 101 min) in successfully-treated patients. Procedural failure in 32 patients was due to failure to implant a clip (i.e., aborted procedures) in 11 patients and failure to reduce MR severity to grade ≤2+ despite clip implantation (clip failure) in the other 21 patients (1 clip in 10, ≥2 clips in 11 patients). Median device time in these 32 patients was 75 min (IQR: 61 to 131 min) (p = 0.0142 vs. successfully-treated patients). Online Figure 1 shows the impact of MitraClip implantation on MR severity.
Baseline characteristics of patients according to procedural outcome (success vs. overall failure) are given in Table 1. Compared with successfully-treated patients, patients in whom MitraClip therapy failed more often had DMR (Fig. 1), and the severity of MR was more often grade 4+. Moreover, they had a significantly larger median EROA (56 mm2 vs. 41 mm2), a significantly smaller MVOA (3.5 cm2 vs. 3.9 cm2), and a significantly higher TMPG (3.6 mm Hg vs. 2.4 mm Hg). Pertinent baseline characteristics of the 32 patients with procedural failure are given in Table 2.
Baseline patient characteristics and procedural outcomes according to MR etiology
Because DMR (n = 95) and FMR (n = 205) represent highly distinct morphological and clinical entities, it is prudent to assess potential differences between the etiologies in baseline variables that might affect procedural outcome. DMR patients differed from FMR patients in that they had a higher prevalence at baseline of grade 4+ MR (n = 55 [58%] vs. n = 81 [40%]; p = 0.0040) and, accordingly, a larger median EROA (52 mm2 [IQR: 35 to 71 mm2] vs. 39 mm2 [IQR: 32 to 49 mm2]; p < 0.0001). Furthermore, TMPG was significantly increased in DMR patients (3.4 ± 2.3 mm Hg vs. 2.1 ± 1.3 mm Hg; p < 0.0001). The difference in MVOA (DMR: 3.9 ± 1.1 cm2 vs. FMR: 3.7 ± 0.9 cm2) did not reach statistical significance (p = 0.085).
Procedural failures were encountered significantly more often in DMR than FMR patients (Fig. 1). No difference between FMR and DMR was encountered with regard to the distribution of the type of procedural failure (aborted procedure and clip failure) (DMR, 5 [31%] and 11 [69%] vs. FMR, 6 [38%] and 10 [62%]; p = 0.913).
Predictors of overall procedural failure
When all procedural failures (n = 32) were compared with the 268 successfully-treated patients, MR etiology, EROA, TMPG, and MVOA were identified as univariate predictors of procedural failure by exact binary logistic regression analysis (Table 3). On exact multivariate logistic regression, only EROA and TMPG remained independently predictive of overall procedural failure, with the odds of failure increasing by a factor of 1.21 (95% confidence interval [CI]: 1.09 to 1.39, i.e., by 21% [95% CI: 9% to 39%] per 10-mm2 increase in EROA) and a factor of 1.26 (95% CI: 1.03 to 1.54, i.e., by 26% [95% CI: 3% to 54%] per 1-mm Hg increase in TMPG) (Table 3).
Predictors of differential procedural outcome
When EROA, MVOA, and TMPG were entered into a classification tree analysis, the interrelationship between these variables with respect to their impact on abortion of the procedure or clip failure on the one hand and procedural success on the other was revealed in 4 different outcome groups (Fig. 2). The first split at the root of the tree occurred at an EROA of 70.8 mm2, because the highest rate of clip failure (7 patients, 25%) was encountered in the 28 patients with an EROA >70.8 mm2. Of the 7 clip failure patients, 5 (71%) had DMR, with a flail leaflet being responsible for the high EROA in 4 of the 5 patients, and 2 patients had FMR, with a complete loss of leaflet coaptation present in 1 of them. A similarly high rate of clip failure was not observed in any of the subgroups of patients with an EROA ≤70.8 mm2.
The next split at an MVOA of 3.0 cm2 identified the 217 patients with an EROA ≤70.8 mm2 and an MVOA >3.0 cm2 in whom the highest success rate (204 of 217, 94.0%) of all patient subgroups was seen. TMPG had no impact on procedural outcome in these patients.
The overall success rate in the 55 patients with an EROA ≤70.8 mm2 but an MVOA ≤3.0 cm2 was 80% (n = 44). However, a TMPG ≥4 mm Hg in concert with an MVOA ≤3.0 cm2, present in 16 patients, yielded the lowest success rate (56.3%, 9 patients) of all subgroups and the highest rate of aborted procedures (37.5%, n = 6). DMR was present in 4 of the 6 patients; the causes of MR were leaflet prolapse in 3 patients and pre-existing mitral stenosis in an annuloplasty ring in 1 patient. When TMPG was ≤3 mm Hg, a success rate of 89.7% (35 of 39 patients) was encountered despite the presence of an MVOA ≤3.0 cm2.
Exact multinomial logistic regression analysis (Table 4, Fig. 3) validated the classification tree analysis by showing that an EROA >70.8 mm2 (odds ratio [OR]: 5.54; 95% CI: 1.80 to 16.3) and a TMPG ≥4 mm Hg (OR: 3.08; 95% CI: 1.04 to 8.75) are independently predictive of clip failure, whereas an MVOA ≤3.0 cm2 (OR: 0.19; 95% CI: 0.04 to 0.82) and a TMPG ≥4 mm Hg (OR: 11.0; 95% CI: 2.7 to 61.8) are independently predictive of an aborted procedure.
Because no formal criteria for patient selection are in effect at our institution, a potential learning curve must be appreciated when interpreting our results. There was a nonsignificant trend (p = 0.067) toward an increase of the procedural failure rate throughout our experience, with an overall failure rate of 7% in the first 100 patients, 10% in the second tercile, and 15% in the third tercile (Online Fig. 2). Interestingly, the prevalence of patients with an MVOA <3.0 cm2 also increased from the first to the last tercile (17% in the first, 22.5% and 28.7% in the second and third terciles, respectively). None of the other variables identified as predictors of procedural failure revealed such a trend (Online Table 2).
In this single-center study of 300 elderly high-risk patients with predominantly FMR, most of whom would not have been enrolled in the EVEREST II trial, a low overall procedural failure rate of 10.7% was encountered (9). Procedural failure was due to either failure to implant a MitraClip (aborted procedure in 3.7% of patients) or failure to reduce MR severity to grade ≤2+ despite clip implantation (clip failure, 7.0%). The main finding of this study was that variables reflecting either the degree of mitral regurgitation (EROA) or the degree of mitral stenosis (MVOA and TMPG) emerged as predictors of overall procedural failure. For clip failure, an EROA ≥70.8 mm2 and a TMPG ≥4 mm Hg were independently predictive, whereas TMPG ≥4 mm Hg and MVOA ≤3.0 cm2 were independently predictive of an aborted procedure.
Effective regurgitant orifice area
Interestingly, EROA was identified as the most significant predictor of overall acute procedural outcome; as shown in the classification tree analysis, an EROA >70.8 mm2 revealed a subgroup of patients at particularly high risk of clip failure (i.e., for inadequate MR reduction despite clip implantation independently of MR etiology). Such a high EROA is commonly the manifestation of severe pathology of the mitral valve. In fact, of the 7 patients with clip failure and an EROA >70.8 mm2, 5 patients with DMR had a flail mitral leaflet, 1 patient with FMR had a complete loss of leaflet coaptation, and 1 other patient had an excellent intraprocedural result after implantation of 1 clip even after norepinephrine challenge, but showed severe MR on discharge without evidence of clip detachment.
MVOA and TMPG
Patients with an EROA ≤70.8 mm2 and an MVOA >3.0 cm2 exhibited an excellent procedural success rate of 94.0%. It is therefore reasonable to assume that the simultaneous presence of these 2 features provides strong evidence in favor of MitraClip therapy. In contrast, in patients with an MVOA <3.0 cm2, the success rate decreased to 80% and further to only 56% if the reduced MVOA was combined with a TMPG ≥4 mm Hg. Not surprisingly, in 4 of 6 patients in the latter subgroup, the procedure was aborted because clip implantation resulted in an unacceptably high TMPG. In the other 2 patients of this group, the procedure was aborted because of a short length of the posterior mitral leaflet in 1 patient and the inability to advance the guide catheter in the other.
MR etiology and morphological aspects of the mitral valve
In our population, procedural failures occurred more often in DMR than in FMR, which, despite the low number of events, is in line with the opinion of most interventionists that DMR poses a relatively greater challenge to MitraClip therapy. Nevertheless, according to our analyses, MR etiology was not identified as an independent predictor of procedural failure, whereas the predictive value of EROA, MVOA, and TMPG was independent of MR etiology. Thus, although, for example, an EROA >70.8 mm2 is likely more prevalent in DMR, and indeed was in our population, the likelihood of failure is equally high in FMR and DMR once an EROA >70.8 mm2 is encountered.
Our data thus do not indicate that the higher risk of procedural failure in DMR patients should be taken as an argument against MitraClip therapy in these patients. Instead, our data advocate that variables predictive of procedural failure, such as those presented in this study, should be assessed regardless of MR etiology to possibly reduce procedural failure rates further and obtain durable MR reduction.
Pronounced elevation of EROA as well as increased TMPG and decreased MVOA must necessarily be a result of morphological or geometric alterations of the mitral valve apparatus (10–12). The complexity of the pathomorphology of MR poses several challenges when trying to formally assess which morphological aspects of the mitral valve predict procedural failure: some variables commonly thought to affect procedural outcome are specific to MR etiology (e.g., coaptation depth in FMR, location of flail leaflet in DMR) and may require further subclassification (e.g., flail gap). However, the low number of procedural failures in our study defied analyses according to MR etiology. Furthermore, some morphological variables may interact with each other (e.g., leaflet calcification with short leaflet length), which again would necessitate large numbers of patients with that feature to obtain robust regression analysis results. Finally, predictor variable analyses might be obscured by factors that are difficult to account for, such as the unexpected difficulty to advance the guide catheter through the inferior vena cava because of previous abdominal trauma.
Accordingly, EROA, MVOA, and TMPG emerge as clinically useful variables when evaluating patients for MitraClip implantation regardless of the etiology of MR. These variables are routinely assessed in most echocardiographic facilities, and their assessment is well defined by current echocardiographic guidelines (5,13). They could be applied in 3 ways: 1) the combination of an EROA ≤70.8 mm2 and an MVOA >3.0 cm2 strongly supports the technical feasibility of MitraClip implantation; 2) in contrast, pathological values of the these variables mandate critical morphological evaluation of the suitability for MitraClip therapy, and every effort should be taken to identify such a pathology; and 3) if an untoward morphology for MitraClip implantation is encountered, these parameters might give additional guidance in separating morphologies still amenable for MitraClip from those with a high risk of procedural failure.
Figure 4 illustrates a potential decision tree for guidance in patient selection. Once commonly accepted contraindications (14) are ruled out, selection could be guided by the echocardiographic variables evaluated in the present study. In cases of an EROA ≤70.8 mm2 and an MVOA >3.0 cm2, a high rate of success can be expected; however, in cases of an MVOA ≤3.0 cm2 and a TMPG ≥4 mm Hg, MitraClip therapy should only be performed exceptionally. In patients with an EROA >70.8 mm2, further evaluation is required. Issues that appear to be of importance in this context are annular diameter (particularly bicommissural) and annular calcification, to evaluate whether implantation of multiple clips is feasible, and assessment of achievability of leaflet grasping (e.g., absent coaptation in FMR or late flail gap in DMR). However, the significance of these parameters in this setting will have to be evaluated in future studies.
Despite the relatively high total number of patients who underwent MitraClip therapy in our study, it must be emphasized that the number of patients reaching the endpoint of procedural failure is low. Therefore, analyses separated by etiology could not be performed and variables specific to either etiology were not assessed. For the same reason, it cannot be ruled out that procedural failure of MitraClip therapy might be influenced by additional parameters, which, although included in our analyses, might require larger numbers of events to be identified as relevant for procedural outcome. Thus, even though it cannot be claimed that our analyses are comprehensive, they provide a solid base for the assumption that EROA, TMPG, and MVOA are important predictors for procedural failure in MitraClip therapy. Additionally, it must be mentioned that echocardiographic variables were not assessed by an independent core laboratory.
Furthermore, our data reflect the experience of a single center, and thus care must be taken when extrapolating these results to other centers. Nevertheless, it is reasonable to assume that the predictors identified herein are of value for patient selection as their assessment has been well defined over the years, and it is reasonable to assume that the identified parameters reflect true technical limitations of the procedure, although cutoff values may shift (e.g., depending on the experience with the procedure). Moreover, a learning curve with regard to all procedural aspects, including patient selection, interventional approach, and procedural outcomes, must be appreciated. Indeed, procedural failure rates showed a nonsignificant trend toward an increase throughout the period of observation, along with a significant increase in the prevalence of patients with an MVOA <3.0 cm2, which might point toward a more liberal patient selection with growing experience. In this context, it is important to point out that no formal exclusion criteria (e.g., EVEREST criteria) were pre-defined for patient selection. On the one hand, this might have resulted in variation of these criteria throughout the observation period, but, on the other hand, it allowed us to demonstrate that MitraClip therapy is feasible with a high procedural success rate in patients not fulfilling the anatomic exclusion criteria defined in EVEREST.
Our study shows that, regardless of MR etiology, EROA, MVOA, and TMPG are independently predictive of acute procedural failure in surgical high-risk patients undergoing MitraClip implantation. These variables thus evolve as readily assessable, well-validated variables to guide patient selection for MitraClip therapy.
Dr. Rudolph has received an unrestricted research grant and lecture fees from Abbott Vascular. Dr. Baldus has received research grants and honoraria from Abbott. Dr. Blankenberg has received research grants from Boehringer Ingelheim, Bayer, Abbott Diagnostics, Siemens, and Thermo Fisher; is an advisory board member for Novartis, Boehringer Ingelheim, Bayer, Thermo Fisher, and Roche; and has received lecture fees from AstraZeneca, Bayer, Boehringer Ingelheim, Siemens, Abbott Diagnostics, Abbott, Medtronic, Pfizer, Roche, Siemens Diagnostics, and Thermo Fisher. Dr. Goldman has received a research grant from Abbott Vascular. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Bakhus and Rudolph are currently affiliated with Uniklinik Köln, Herzzentrum, Klinik III für Innere Medizin, Cologne, Germany.
- Abbreviations and Acronyms
- confidence interval
- mitral regurgitation of degenerative/mixed origin
- effective regurgitant orifice area
- mitral regurgitation of functional origin
- interquartile range
- mitral regurgitation
- mitral valve orifice area
- odds ratio
- mean transmitral pressure gradient
- Received September 24, 2013.
- Revision received December 9, 2013.
- Accepted December 20, 2013.
- American College of Cardiology Foundation
- Rudolph V.,
- Knap M.,
- Franzen O.,
- et al.
- Lancellotti P.,
- Tribouilloy C.,
- Hagendorff A.,
- et al.
- Firth D.
- Agricola E.,
- Oppizzi M.,
- Pisani M.,
- Meris A.,
- Maisano F.,
- Margonato A.