One-year Outcomes and Predictors of Mortality After MitraClip Therapy in Contemporary Clinical Practice: Results From the German Transcatheter Mitral Valve Interventions Registry

Miriam Puls; Edith Lubos; Peter Boekstegers; Ralph Stephan von Bardeleben; Taoufik Ouarrak; Christian Butter; Christine S. Zuern; Raffi Bekeredjian; Horst Sievert; Georg Nickenig; Holger Eggebrecht; Jochen Senges; Wolfgang Schillinger


Eur Heart J. 2016;37(8):703-712. 

In This Article


Baseline Characteristics

Between August 2010 and July 2013, 828 patients were enrolled prospectively into the TRAMI registry in 21 German sites (a complete list of participating sites is provided in Supplementary material online, Table S1 One-year follow-up could be performed in 749 patients (90.5%) at a median of 386 days after MitraClip implantation. Only patients with available 1-year follow-up were considered in the following analyses (with the exception of the Cox regression model). Concerning the missing individuals, 16/79 withdrew their consent, and 63/79 were lost to follow-up.

Baseline characteristics of the 749 remaining participants are displayed in Table 1. Patients enrolled in TRAMI were predominantly male (61.4%) and characterized by advanced age (median, 76.0 years), high estimated surgical risk (median log EuroSCORE 20.0%, median STS score 6.0%), and a high burden of comorbidities (i.e. coronary artery disease 78.1%, renal failure 65.5%, severely reduced LVEF 33.7%). Furthermore, patients were in advanced stages of heart failure (NYHA class III in 70.5%, NYHA class IV in 18.5%). Consecutively, they exhibited reduced functional capacities (median 6 min walk test distance: 200.0 [IQR: 120.0–320.0] m) and had elevated levels of NT-pro BNP and BNP (median NT-pro BNP: 3497.0 [IQR: 1559.0–6880.5] pg/mL; median BNP: 692.0 [IQR: 244.0–1380.0] pg/mL). The prevalent valvular pathogenesis was secondary MR (71.3%).

Compared with our study cohort, the 79 patients who were excluded due to missing 1-year follow-up exhibited a slightly higher estimated surgical risk at baseline documented by significantly higher values for log. EuroSCORE (median, 22.5 vs. 20%, P = 0.03) and systolic pulmonary artery pressure (P < 0.01), and by a tendency towards a higher age (P = 0.06) and a higher prevalence of COPD (P = 0.07) (see Supplementary material online, Table S2

Patient Allocation

The decision for patient allocation to percutaneous therapy was left to the discretion of the participating centres and was made by a heart team in 60.9%, by a cardiologist alone in 37.8%, and by a cardiac surgeon in only 1.3% of patients. The most common reason for denying surgery was estimated surgical high-risk status (58.0% of cases, log. EuroSCORE ≥20% in 50% of patients), followed by age (48.3%), patient's preference (25.0%), frailty (21.2%), limited prognosis due to non-cardiac (mostly malignant) comorbidity (22.3%), and inoperability (11.4%) (entry of more than one reason was possible).

In-hospital and 30-day Outcome

In TRAMI, MitraClip implantation was successful in most cases. On average, procedural time was 102.8 ± 54.1 min, radiation time 28.8 ± 57.9 min, and 1.4 ± 0.6 clips were implanted. Procedural failure (defined as severe residual MR, abortion of MitraClip procedure, conversion to open heart surgery, or failure as assessed by the interventional team) was observed in only 3.2% of patients. Also, the intervention proved to be safe: Intra-procedural death occurred in only one patient (0.1%), in-hospital mortality was 2.4% (n = 18), and MACCE rate was 3.1% (n = 6 cases of stroke, n = 0 of myocardial infarction).

Specific procedural complications are displayed in Table 2. The most common adverse events were respiratory failure with consecutive invasive or non-invasive ventilation (6.5%), and severe bleeding necessitating transfusion (7.0%). No case of clip embolization occurred, but single-leaflet clip attachment was observed in five cases (0.7%). Consecutively, an additional mitral valve procedure became necessary during the index hospitalization in 11 patients (surgery in six and percutaneous interventions in five cases). The median length of hospital stay was 9 days [IQR 6.0–15.0]. The vast majority of patients (89.3%) were regularly discharged to their normal social environment, 6.3% to cardiac rehabilitation facilities, and the remaining to nursing homes. Echocardiography at time of discharge revealed the presence of severe MR in only 2.3% of cases, whereas 85.2% had none or mild MR. At 30 days, exact cumulative mortality was 4.5%.

Compared with our study cohort, the 79 patients who were excluded due to missing 1-year follow-up had a significantly lower rate of procedural success (92.2 vs. 96.8%, P = 0.03) and consecutively a higher rate of additional mitral valve procedures at 30 days (P = 0.02, see Supplementary material online, Table S3 However, in-hospital mortality in this group was (per definition) 0%, and the incidence of all other complications did not differ.

One-year Outcomes: Safety, Efficacy, Heart Failure, and Quality of Life

Adverse events and measures of treatment efficacy at 12 months are listed in Table 3. Exact cumulative mortality at 1 year was 20.3% (152/749 patients) (see also Figure 1). Causes of death were the following: sudden unexpected death in 23/152 cases (15.1%), other cardiovascular causes in 56/152 patients (36.8%), non-cardiovascular reasons in 19/152 cases (12.5%), and unknown/unreported reasons in 54/152 patients (35.5%). The cumulative incidences of TIA, stroke and myocardial infarction at 1 year were 3.8%, 2.1% and 0.9%, respectively. During the first year of follow-up, 14.1% of patients were re-hospitalized due to cardiac decompensation and 17.8% due to other cardiovascular reasons. An additional mitral valve procedure became necessary in cumulatively 8.5% of patients (surgery in 2.3% and second MitraClip implantation in 5.2%).

Figure 1.

Kaplan–Meier curve displaying overall survival after MitraClip implantation in transcatheter mitral valve interventions registry patients.

At 1 year, 63.3% of patients had no or few symptoms of heart failure and pertained to NYHA functional classes I or II (in contrast to 11.0% at baseline). Health-related quality of life was measured by the EQ-5 dimensions descriptive system and by the EQ VAS. Compared with baseline, significantly more patients were completely independent concerning the dimension 'self-care' (74.0 vs. 58.6%, P = 0.005), and significantly more patients reported no problems concerning the dimension 'anxiety/depression' (66.7 vs. 48.9%, P < 0.0001). Regarding the dimensions 'mobility', 'usual activities,' and 'pain/discomfort', no significant changes could be observed. However, patients' self-rated health status on the EQ VAS improved significantly from 50.0 [IQR 40.0–60.0] at baseline to 60.0 [IQR 50.0–70.0] at 1 year (P < 0.0001).

Predictors of 1-year Mortality

In order to identify risk factors for long-term mortality (cumulative hospital and post-hospital mortality), baseline and procedural characteristics were compared between survivors and non-survivors at 1 year. Effects that proved to be statistically significant in univariable analysis were further subjected to multivariable Cox regression analysis, as well as the covariates 'gender' and 'age>75 years'. Due to high numbers of missing values, the covariates 'total procedure time', 'fluoroscopy time,' and 'PAP sys >45 mmHg' were not used for the Cox regression model. The discharge medication was omitted in order to keep the 18 cases of in-hospital mortality for calculation. Due to redundancies, the surgical risk scores (which were predictive for mortality in univariable analysis) were not inserted into the model. Finally, the covariates 'sinusrhythm, prior stroke, 'number of implanted clips ≥2', 'COPD', and 'prior cardiac decompensation' were excluded by the forward selection procedure (Table 4).

According to our multivariable analysis, significant predictors of 1-year mortality were NYHA class IV (HR 1.62, P = 0.02), anaemia (HR 2.44, P = 0.02), previous aortic valve intervention (HR 2.12, P = 0.002), serum creatinine ≥ 1.5 mg/dL (HR 1.77, P = 0.002), peripheral artery disease (HR 2.12, P = 0.0003), left ventricular ejection fraction <30% (HR 1.59, P = 0.01), severe tricuspid regurgitation (HR 1.84, P = 0.003), and procedural failure (HR 4.36, P < 0.0001).

In order to test the prognostic performance of the developed risk model, we calculated the area (AUC) under the receiver operating characteristic curve: The quality of the regression model could be confirmed by a high c-value (0.75). In addition, we internally validated the model by using a five-fold cross-validation. Therefore, the study population was randomly divided into five equal groups. For each run, the AUC was calculated in the training group and then applied to the test group. Combining all the five test groups, we obtained an AUC average of 0.685, indicating the validity of the model.