Patient characteristics and clinical outcomes

A total of 125 CF-LVAD patients with none or trivial preoperative AI were enrolled, and their characteristics are summarized in Table 1. Age was 43 ± 14 years, 87 (70%) were male, and BSA was 1.61 ± 0.20 m2. During the study period, 16 patients died, 40 underwent heart transplantation after 33 ± 11 months of waiting, and 7 underwent CF-LVAD explantation. During the 30 ± 16 months of CF-LVAD support, survival rate was 93% at 1 year and 89% at 2 years (Supplemental Fig. 1).

Table 1 Patient characteristicsPreoperative risk factors for de novo AI

Preoperative AI grade was none in 86 (69%) and trivial in 39 (31%) patients. During 20 ± 14 months of echocardiographic follow-up, 32 patients developed de novo moderate or severe AI after CF-LVAD implantation. The rate of freedom from de novo AI was 86% at 1 year and 67% at 2 years.

Multivariable analysis showed that higher preoperative RVSWI [hazard ratio (HR), 1.12 /g/m2/beat; 95% confidence interval (CI), 1.00–1.20 /g/m2/beat; p = 0.047] and preoperative trivial grade AI (HR, 2.8; 95% CI, 1.2–6.4; p = 0.020) were independent risk factors for de novo AI (Table 2). In the present cohort, the median RVSWI value was 6.8 g/m2/beat and rate for freedom from de novo AI was significantly higher in patients with preoperative RVSWI < 7 g/m2/beat (p = 0.03) (Fig. 2).

Table 2 Preoperative risk factors for de novo AIFig. 2

Kaplan–Meier estimates of freedom from de novo AI in lower preoperative RVSWI (< 7 g/m2/beat) patients versus higher RVSWI (≥ 7 g/m2/beat) patients

On the other hand, other risk factors for de novo AI, including age, female sex, and BSA, were not statistically significant [1, 8]. Other preoperative hemodynamic parameters, including RAP/PCWP, PAPi, mPAP, CVP, and CI were not also associated with de novo AI (Table 2). In our study cohort, etiology (p = 0.02) and preoperative intra-aortic balloon pumping (p = 0.02) was associated with preoperative RVSWI (Supplementary Table 1), but those factors were not risk factors of de novo AI development (Table2). Inotropic support (p = 0.32) was not associated with RVSWI and left ventricular ejection fraction (r = 0.21, p = 0.02) was not correlated with RVSWI.

Outflow graft anastomosis design

Postoperative computed tomographic images were available for 112 (90%) patients. The mean distance between the sino-tubular junction and graft was 24 ± 5 mm, while inclination angle was 58 ± 17° and azimuth angle was 105 ± 20°. These geometrical parameters related to outflow graft anastomosis were not associated with de novo AI development (Table 3).

Table 3 Outflow graft anastomosis design and association with de novo AIPump speed settings

CF-LVAD pump speed settings at discharge are shown according to the device in Table 4. In all devices, pump speed settings seemed to be similar or lower as compared with previous reports probably because of small BSA of our patients [13, 16, 22, 23]. Associations between pump speed settings and de novo AI development were difficult to be statistically analyzed because of the low number of patients with each device. Among the devices, HeartMate II was implanted in a relatively large number of patients (n = 48), but hazard ratio per 100 rpm was approximately 1.0 (Supplementary Table 2).

AV opening status

In the present cohort, the rate of continuous AV closure at 1 month was 78%, while those at 3, 6, 12, and 24 months were approximately 65%. The rate of AI development was compared between patients with a continuously closed AV in all echocardiographic follow-up (n = 65) and those showing AV opening at least once (n = 59). Cox proportional-hazards model showed that continuously closed AV was associated with de novo AI (HR, 2.2; 95% CI, 1.2–6.4; p = 0.043).

Impact of preoperative RVSWI on AV opening status

Since higher RVSWI was the significant risk factor for de novo AI, we examined the influence of preoperative RVSWI on AV opening status under LVAD support. The longitudinal analysis using generalized mixed effects model showed that higher preoperative RVSWI was associated with continuous AV closure after LVAD implantation (Odd ratio, 1.20/g/m2/beat; 95% CI, 1.00–1.43/g/m2/beat; p = 0.047) (Supplementary Table 3).

Relationship between preoperative RVSWI and postoperative pump flow rate

De novo AI development in patients with higher preoperative RVSWI (Table 2, Fig. 2) may be explained by greater pump flow rate, which could impose a higher hemodynamic shear stress on AV leaflets. However, pump flow rate cannot be accurately estimated by the pump power consumption [24]. Furthermore, even if determined with right heart catheterization, accurate measurement of pump flow rate is impossible in patients with AV opening or significant AI because of antegrade or retrograde flow through the AV. Therefore, the relationship between preoperative RVSWI and postoperative pump flow rate was examined in patients with a continuously closed AV and without significant AI, in whom pump flow rate was equivalent to cardiac output measured with right heart catheterization.

Postoperative right heart catheterization was performed in 22 patients (18%) who had a continuously closed AV and no significant AI. At 88 ± 52 days after CF-LVAD implantation, pump flow index, which is equivalent to CI, was 2.8 ± 0.7 L/min/m2 in patients with preoperative RVSWI of 6.7 ± 3.2 g/m2/beat. Pearson’s correlation analysis revealed that pump flow index was positively correlated with preoperative RVSWI (r = 0.44, p = 0.04) (Fig. 3).

Fig. 3

Relationship between preoperative RVSWI and CF-LVAD pump flow index in patients with continuously closed AV