Primary outcome of this study was mortality of all causes. The secondary outcome comprised dyspnoea reduction classified by New York Heart Association (NYHA) class.

In accordance to the recommendations of the Mitral Valve Academic Research Consortium/MVARC [12]), technical success was defined as ability to deploy the device as intended and successful retrieval of the delivery system without peri-procedural mortality or need of emergency surgery or intervention. Due to the study design as a retrospective registry, device success and procedural success were adjusted from MVARC-recommendations to discharge conditions: device success at discharge was defined as successful placement of the device without procedural mortality or stroke, missing evidence of functional failure of the device or device-related complications until discharge and post-interventional reduction of MR to optimal or acceptable levels without significant mitral valve stenosis. The definition of procedural success at discharge included device success achieved in the absence of major clinical complications (according to MVARC [12]). MR and TR were graded by experienced echocardiographers according to society recommendations [13, 14]; TR severity was denominated in 4 grades: no/trace, mild, moderate, and severe (comprising the subgrades “severe”, “massive” and “torrential”, as defined by the most recent TR-classification scheme [15]). Renal insufficiency was defined by a glomerular filtration rate < 60 ml/min*kg. Pulmonary hypertension was determined by invasive measurements (if available at baseline) or echocardiographic high probability, according to guidelines [16]. For statistical evaluation of sPAP (systolic pulmonary arterial pressure), only values derived by non-invasive means (echocardiographic assessment of right ventricular systolic pressure derived from RV/RA-gradient plus estimated central venous pressure) were used. Obesity was defined as BMI ≥ 30 kg/m². Echocardiographic left and right ventricular analyses and quantification were based on transthoracic echocardiography measurements in accordance to ASE/EACVI recommendations; RV dysfunction was predominantly defined by a reduced TAPSE (tricuspid annular plane systolic excursion) of < 17 mm [17].

Continuous parameters are presented as median and interquartile range (IQR) when non-normally distributed (as tested by Kolmogorov–Smirnov and Shapiro–Wilk tests), and otherwise absolute numbers and percentages. Continuous variables were compared using the Wilcoxon–Whitney U test or Wilcoxon signed rank test and categorical variables with Fisher’s exact or chi² test, as appropriate. We compared TMVR patients with severe vs. non-severe-grade TR as well as patients with post-interventional TR reduction to those without, including Kaplan–Meier Curves. Cox regression analyses were computed to examine the impact of TR at baseline as well as post-procedural changes of TR grade on short- and long-term mortality. Results were presented as Hazard Ratios (HR) with 95% confidence interval (CI) (i) univariate/unadjusted and (ii) multivariate/adjusted for factors which had been identified as having relevant impact on patients’ long-term prognosis in our cohort [18]: patients’ age at the time of procedure, gender, NYHA class before intervention, left ventricular ejection fraction (LVEF) at baseline, coronary artery disease, chronic obstructive pulmonary disease (COPD), renal function/baseline creatinine values, peri-interventional reduction of MR-grade, as well as existence of a pacemaker, in a multivariate fashion. A propensity match model was calculated for differences in the distribution of no/trace and mild vs. moderate and severe MR grades at discharge.

The software SPSS^® (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp.) was used for computerised analysis. P values of < 0.05 (two-sided) were considered to be statistically significant.

Between 09 June 2010 and 08 March 2018, 725 consecutive patients underwent percutaneous edge-to-edge-therapy at our center. Of those, 119 were excluded: 90 (12.4%) had primarily undergone a simultaneous combination of edge-to-edge repair and other forms of TMVR (e.g., interventional annuloplasty or chordal reconstruction), technical failure occurred in 8 patients (1.3%/technical success 98.7%; resulting mitral stenosis, leading to abortion of clip implantation in 6 patients; hemodynamic instability before insertion of the transseptal sheath in 1 patient, pericardial tamponade before introduction of the implantation with peri-interventional death in 1 patient), and 21 subjects (2.9%) had additionally undergone interventional tricuspid valve repair until the end of the follow-up period (mostly, due to severe symptomatic TR; TR repair was performed in various time intervals after TMVR). In total, 606 patients were included (Fig. 1).

At the time of the index procedure, mean age was 78.5 ± 7.3 years [median 79.1 (IQR 74.1/83.7); 88.1% aged over 70 years] and sex distribution was nearly balanced (46.5% females). Leading etiology was FMR in 56.4%, DMR was present in 31.2% and mixed pathomechanism in 12.4%. The mean logistic Euroscore was 29.7% ± 17.0 [median 26.0 (18.4/38.7)]. Mean LVEF was moderately impaired [41.6% ± 13.4; median 42.0% (30.0/55.0)]. MR at baseline was classified severe in 91.9%, and moderate in 8.1% of the treated patients. Regarding dyspnoea, 99.4% had been symptomatic and 89.2% were graded as NYHA classes III/IV. Mean BNP level was elevated (961 ± 1116 pg/ml). Adverse events during the index visit were recorded in 8 subjects (1.3%:2/0.3% immediate surgical treatment during intervention, 2/0.3% peri-interventional myocardial infarction, 3/0.5% stroke, 2/0.3% hemodynamically relevant pericardial effusion). MR reduction was achieved in 94.0%. At discharge, device success was observed in 92.7% and procedural success in 91.7%.

After a median follow-up of 511 [IQR161/981] days (mean 674 ± 619), survival status was available in 96.9%. Whereas in-hospital mortality was 2.5%, survival was 94.5% after one month, 74.5% after 1 year, 54.5% after 3 years, 37.6% after 5 years and 21.7% after 7 years. At baseline, echocardiographic assessment of TR was available in 92.4%; TR was categorized severe in 23.2%, medium in 34.3%, mild in 36.3% and no/trace in 6.3%. At 1-month [median time 43 (IQR37/55) days], echocardiographic examination results were available in 75.0% of the living patients (70.3% of the whole group) with TR graded as severe in 16.0%, medium in 29.1% and mild in 47.2%, whereas no/trace TR was found in 7.7% of the patients. At 1-year follow-up [median 366 (IQR350/377) days, echocardiography allowing TR quantification available in 63.2% of the surviving patients/42.1% of the whole group], 17.0% were graded as severe TR, 29.2% moderate, 46.6% mild and 7.2% no/trace. While changes in TR severity at both follow-ups versus baseline findings were significant (p < 0.001 at 30 days, p = 0.024 at 1 year), no significant differences were found regarding changes between 1-month and 12-month assessments (p = 0.435) (Fig. 2).

For further statistical analyses regarding the impact of post-interventional TR reduction, only subjects with detectable (i.e., mild, medium or severe) TR at baseline (n = 525, 93.7% of the individuals with echocardiographic assessment), were included. At 1-month, improvement of at least 1 TR grade was documented in 34.9% (of  n = 378) of these patients; at 1 year, TR reduction compared to baseline was found in 35.3% (of  n = 235) of the patients.

The groups of patients with severe vs. non-severe grade TR before intervention were comparable regarding most parameters, including gender, cardiovascular risk factors and etiology (Table 1). In median, patients with severe TR were older [81.1 (76.6/84.9) vs. 78.5 (73.5/83.3) years, p < 0.001] and atrial fibrillation was more common (83.8 vs. 66.7%, p < 0.001) in patients with severe TR. While there was no difference regarding distribution of baseline MR grades, MR grades at discharge were lower in patients with severe TR (remaining moderate or severe MR in 20.7% vs. 30.5%, p = 0.030). Patients with severe TR presented with slightly higher Creatinine values at 30 days [1.39 (1.05/1.91) vs. 1.23 (0.99/1.60) mg/dl, p = 0.037].

When comparing groups with post-interventional TR reduction to those without, both groups were nearly balanced regarding most baseline parameters (Table 2). As expected, higher baseline grades of TR were more often found in patients with a post-interventional decrease in TR severity (moderate or severe TR in 90.9 vs. 35.2%, p < 0.001). Furthermore, moderate or severe MR at discharge was less common in patients with TR reduction (17.7 vs. 33.8%, p = 0.002). Logistic Euroscore I was predominantly indicating high surgical risk in both groups, yet with a lower median value for patients with post-interventional TR reduction [22.0% (16.1/30.5) vs. 26.5% (18.5/37.9), p = 0.015]. Regarding medical therapy, the frequency of intake of RAS-blockers was higher (88.6 vs. 79.7%, p = 0.032) in patients with TR-grade reduction.

With regard to long-term survival of TMVR patients with severe TR in comparison to those with lower grades of TR at baseline, relevant differences were found (Fig. 3). While in-hospital (3.8% vs. 2.3%, p = 0.356) and 30-day mortality (7.6 vs. 5.3%, p = 0.370) were similar, 1-year survival was significantly reduced in patients with severe TR (65.2% vs. 77.0%, p = 0.030) with still relevant differences at 3 (45.8% vs. 56.7%, p = 0.193) and at 5 years (25.0% vs. 41.1%, p = 0.224). This finding was even more pronounced in FMR patients (1-year survival 56.9% vs.77.0%, p = 0.007), in contrast to DMR patients (1-year survival 80.6% vs. 78.7%, p = 1.000). Thus, severe baseline TR was identified as predictor for a higher 1-year mortality in the total cohort [crude HR 1.68 (95% CI 1.12–2.54), p = 0.013], even after adjustment for other factors influencing the long-term survival [adjusted HR 1.65 (95% CI 1.01–2.68), p = 0.044; HR after propensity matching for MR-grade at discharge 1.71 (95% CI 1.03–2.86), p = 0.040]. Regarding long-term survival until ultimate follow-up, HR for severe TR indicated a trend towards impaired survival [crude HR 1.34 (95% CI 0.99–1.81), p = 0.063, adjusted HR 1.36 (95% CI 0.96–1.92), p = 0.085]. For FMR patients, severe TR was predictive for impaired 1-year survival [crude HR 2.31 (95% CI 1.34–3.72), p = 0.002 as well as adj. HR 1.97 (95% CI 1.06–3.64), p = 0.031; HR after propensity matching for MR-grade at discharge 2.54 (95% CI 1.31–4.91), p = 0.006] with a borderline significance for the total follow-up period [crude HR 1.46 (95% CI 1.00–2.15), p = 0.052 as well as adj. HR 1.57 (95% CI 1.00–2.45), p = 0.049; HR after propensity matching for MR grade at discharge 1.63 (95% CI 1.00–2.66), p = 0.050].

Patients with post-interventional TR reduction had a better long-term survival in comparison to those without: 88.5% vs. 81.0% at 1-year (p = 0.102) with persisting non-significant differences for later follow-ups. When focusing on FMR-patients, prognosis was significantly better in subjects with post-interventional TR reduction at 1-year follow-up (92.9% vs. 78.3%, p = 0.025), while later differences were still recognizable, but not of statistical significance (Fig. 4). For the whole cohort, HR indicated for increased mortality at 1 year for in patients lacking post-interventional TR reduction when adjusted for risk factors [crude HR 1.71 (95% CI 0.89–3.29), p = 0.107, adj. HR 2.18 (95% CI 1.02–4.68), p = 0.046, HR after propensity matching for MR grade at discharge 2.39 (95% CI 1.03–5.58), p = 0.043]. In FMR patients, missing post-interventional TR reduction could be identified as predictor for significantly impaired survival at 1 year [HR 3.31 (95% CI 1.15–9.58), p = 0.027, adj. HR 6.82 (95% CI 1.88–24.71), p = 0.003, HR after propensity matching for MR-grade at discharge 6.31 (95% CI 1.46–27.32), p = 0.014] while long-term HR was not significant [adj. HR 1.27 (95% CI 0.74–2.18), p = 0.381]. Reasons why the benefit in survival by TR reduction beyond one year is gradually losing significance, e. g. being be caused by statistical effects (e. g. due to the smaller numbers at risk) or the survival difference decreases gradually, remains subject to future studies.

Patients with post-procedural TR reduction had a better baseline systolic right ventricular function as assessed by TAPSE [tricuspid annular plane systolic excursion, 1.8 (1.4/2.2) vs. 1.6 (1.3/1.9) cm, p = 0.016]. Post-procedural changes in TAPSE compared to baseline were not significant, neither for the whole cohort at one month (p = 0.069) and 1 year (p = 0.920), nor for the subgroup of patients presenting with reduction of TR severity at follow-up (p = 0.856 at 30 days, p = 0.475 at 1 year). When comparing the severe vs. non-severe baseline TR, no significant differences were observed in baseline TAPSE [1.6 (1.4/2.1) vs. 1.7 (1.4/2.1) cm, p = 0.316]. Furthermore, no relevant changes in TAPSE were found at 1 month (p = 0.151) and 12 months (p = 0.934) to baseline, even for the subgroup of patients with severe TR (p = 0.513 at 30 days, p = 0.228 at 1 year).

Systolic pulmonary arterial pressure (sPAP), as assessed by echocardiography, was significantly reduced by the procedure in the whole group, starting from 51 [45/60] mmHg reduced to 46 [39/54] mmHg at 30-day follow-up (p < 0.001) and 46 [0/53] mmHg at 1 year (p < 0.001 compared to baseline). Whereas baseline differences for the subgroups of severe vs. non-severe TR were not significant, patients with later TR reduction had higher assessed sPAP values before the procedure [53 (47/60) vs. 50 (43/60) mmHg, p = 0.035].

New York Heart Association class reduction by at least one grade at one-month follow-up was found nearly equally distributed in groups with or without severe baseline TR (70.8 vs. 69.9%, p = 0.887). In contrast, patients with post-interventional TR decrease were significantly overrepresented in the group of patients reporting dyspnoea reduction (78.6 vs. 65.9%, p = 0.021). BNP values decreased in the whole cohort [median 588 (275/1283) at baseline vs. 503 (272/955) pg/ml at 30 days, p = 0.008], accordingly. Absolute BNP levels at 1 month were lower in the group with TR reduction [417 (236/646) vs. 515 (278/1007) pg/ml, p = 0.007]. Yet, the finding was not accompanied by a higher proportion of patients with declining BNP values [54.4% vs. 53.1%, p = 0.907], or significant differences in relative changes in BNP levels (in mean −  11.1 vs. − 28.3%, p = 0.214) between baseline and 30-day follow-ups.

Due to the retrospective observational monocentric design on an all-comer population of consecutive patients undergoing TMVR lacking a control group, a potential selection bias cannot be excluded and should be taken into account. While follow-up regarding survival status is nearly complete, echocardiographic assessment of TR was available in 92.4% at baseline, in 75% at 1-month follow-up and in 63% at 1 year in this retrospective real-world cohort. Although most baseline parameters with individual impact on survival were distributed equally and also multivariate analyses pointed to a better survival of patients with post-interventional TR reduction, a slightly but significantly different Euroscore of both groups could have influenced results. Furthermore, a significantly higher mean age, prevalence of atrial fibrillation and RV dysfunction as well as non-significant differences in renal function and baseline medication should be taken into account as potential competing risks for the observed higher mortality of patients with severe baseline TR. According to MVARC recommendations, and due to the study design, all-cause mortality was defined as primary endpoint without further sub-stratification [12]. Regarding accuracy of the parameters presented, grading of TR was performed by experienced echocardiographers from our center in a semi-quantitative way according to current guidelines [13‐15] but not confirmed by an external core laboratory. sPAP values as provided were not taken from invasive measurement, but solely estimated by non-invasive means, thus, etiology of PH (pre- vs. post-capillary) could not be discriminated. Furthermore, TR severity could also be influenced by volume load (regredient BNP values at follow-ups).