Secondary Logo

Journal Logo

Clinical Cardiovascular

Intraventricular Flow Patterns in Patients Treated with Left Ventricular Assist Devices

Rossini, Lorenzo*; Braun, Oscar Ö.†,‡; Brambatti, Michela; Benito, Yolanda§; Mizeracki, Adam; Miramontes, Marissa*; Nguyen, Cathleen*; Martinez-Legazpi, Pablo§; Almeida, Shone; Kraushaar, Megan; Vu, Vi; May-Newman, Karen; Bermejo, Javier§; Adler, Eric D.; Kahn, Andrew M.; Del Alamo, Juan C.*,‖

Author Information
doi: 10.1097/MAT.0000000000001158


Left ventricular assist device (LVAD) support is a life-saving therapy for patients with advanced heart failure (HF) refractory to optimal medical treatment. The use of LVADs has increased significantly, and the treatment is now widely used both as a bridge to heart transplantation and as destination therapy.1,2 Despite improved survival, major complications such as thrombosis, bleeding, and stroke are still common, occurring with an incidence of 8–29%.3–5 Subclinical hemolysis and systemic inflammation are also strongly associated with LVAD thrombosis.6 The causes of the increased risk of thrombosis in LVAD-implanted patients are multifactorial and not fully understood. Nonetheless, intraventricular flow disturbances leading to abnormal shear stress and blood stasis are recognized as major risk factors7,8 and have been associated with thromboembolic strokes.9 Of note, recent studies of patients implanted with new generation LVADs have shown stroke rates comparable to the older generation of devices despite the absence of intrapump thrombosis,5 suggesting that the left ventricle (LV) may be a relevant site of local thrombosis and cardioembolism.

The normal LV flow pattern is characterized by a diastolic vortex that drives the transit of blood towards the aorta,10–12 contributing to diastolic transport and reducing kinetic energy losses and cardiac work.13,14 Moreover, it allows for washing the LV entirely in about 2 to 4 beats without inducing shear values high enough to activate platelets or to cause von Willebrand factor degradation.15–18 Devices such as LVADs drastically disrupt the LV blood flow patterns and may lead to blood stasis or abnormally large shear stresses.19,20

The assessment of intraventricular flow patterns during LVAD treatment has been limited mainly to in vitro,21,22in silico,23–25 and ex vivo26 models. These studies have suggested that pump speed, aortic valve opening, cannula location, and orientation are important determinants of intraventricular flow. However, modeling the flow inside the LVAD-assisted ventricle is challenging due to the complex interplays among the pulsatile function of the native myocardium, the continuous LVAD support, and the valves. Consequently, there is a need for in vivo data to evaluate intraventricular flow, alterations in stasis, and hemodynamic shear in LVAD-implanted patients.

We have previously implemented a method to quantify LV blood stasis and cumulated shear based on clinically applicable echocardiographic color Doppler velocimetry (echo-CDV).17,27–29 We have reported an anecdotal application of this method to an LVAD-implanted patient.20 We hypothesized that echo-CDV could be used to noninvasively characterize the effect of LVAD support in LV hemodynamics and help to understand the ventricular LVAD interplay. Therefore, we designed the current study to characterize the intraventricular flow patterns and to quantify the rates of LV blood washout and shear in a small sample of patients with Heartmate II LVADs. We compared flow patterns with data from nonimplanted subjects with either normal or dilated LVs. Finally, we assessed the effects of different LVAD pump speeds during ramp tests on intraventricular flow.


Study Population

Seven subjects undergoing LVAD treatment were prospectively selected from the Advanced Heart Failure Clinic at the UC San Diego Sulpizio Cardiovascular Center in La Jolla, CA. Inclusion criteria for study participants were: 1) presence of sinus rhythm; 2) suitable apical ultrasonic window; 3) clinical stability enabling a ramp study; and 4) Doppler signal-to-noise ratio allowing reliable postprocessing. All LVAD patients had ongoing treatment with a HeartMate II (Thoratec Corp, Pleasanton, CA) implanted between 2011 and 2017 and were examined at a median of 16 (range, 3–71) months after implantation. To show proof of concept of our methodology to the new generation Heartmate 3 LVAD, we enrolled one additional patient treated with this device.

Twenty healthy subjects and 20 patients with nonischemic dilated cardiomyopathy (DCM) were used as controls. These patients were previously recruited from the Hospital General Universitario Gregorio Marañón in Madrid, Spain, for another study.30 The corresponding Institutional Review Boards approved the studies of the two institutions, and all participants were provided written informed consent.

Image Acquisition and Analysis

Comprehensive 2D B-mode and color Doppler echocardiographic examinations were performed using Vivid series scanners and 2–4 MHz phased-array transducers (General Electric Healthcare, Milwaukee, WI). Standard 3-chamber view color Doppler sequences were acquired at each patient’s baseline (clinically determined optimal) pump speed, and at 200 rpm increments spanning the range: (baseline − 400 rpm) to (baseline + 400 rpm). The total number of acquisitions in the seven patients was 32 (three acquisitions were discarded due to poor Doppler signal-to-noise ratio). Aortic insufficiency (AI) was classified by an echocardiographer, and each LVAD patient/speed case was grouped accordingly (absent/mild vs. moderate/severe). Time-resolved vector blood velocity maps in a 3-chamber apical long-axis view were obtained with the echo-CDV algorithm using the endocardial boundary from EchoPac software. Further details are given in the Supplemental Material, Supplemental Digital Content,

From the reconstructed velocity field, the anteroseptal (clockwise [CW]) and inferolateral (counterclockwise [CCW]) sections of the LV vortex ring were identified.13,30 For each vortex ring section (henceforth referred to as vortex), we measured the circulation, Γ (representing the swirling strength), the location along the LV normalized long axis, X (X = 0 corresponding to the base and X = 1 to the apex), the radius, R,30 and the ratio between circulation of the CW and CCW vortices. These time-dependent quantities were averaged through the cardiac cycle.

The computation of the velocity pulsatility (VP), the pulsatility index (VPI), the time spent by blood inside the LV (residence time, [s]), and the cumulated shear index (CSI) are described in the Supplemental Material, Supplemental Digital Content,,27–34 To report and compare data between groups, we obtained instantaneous and CSI maps at the R-wave instant after 5 seconds of integration . Blood domains with increased residence time (> 2 seconds) were identified, and their area () was computed as a percentage of the LV chamber. Likewise, domains with elevated exposure to shear (CSI > 200/s) were identified, and their areas () computed. Global chamber indices of blood stasis and shear exposure were quantified by the spatial maxima and averages of the TR and CSI maps across the whole LV. Variables were reported as median and interquartile range (IQR). Details about the statistical analysis can be found in the Supplemental Material, Supplemental Digital Content,


The median age of the LVAD group was 74 (IQR, 64–78) years. All subjects were males, and six of them (86%) had nonischemic cardiomyopathy. The median age of the DCM group was 62 (IQR, 52–72), and all subjects were diagnosed with nonischemic cardiomyopathy. The healthy control group had a median age of 56 (IQR, 53–66). All subjects in the DCM and healthy control groups were males to match the LVAD group. Age-matching among groups was not performed due to the advanced age of the LVAD group. Table 1 reports echocardiographic and demographic data.

Table 1. - Demographic and Echocardiographic Data
Pre-LVAD LVAD p Value
N 7 7 20 20
Age at examination 74 (62–76) 74 (64–78) 62 (52–72) 56 (53–66) 0.292
Gender (males) 7 (100%) 20 (100%) 20 (100%)
Nonischemic cardiomyopathy 6 (86%) 20 (100%) 0 (0%)
Planned destination therapy 3 (43%) - -
Device type (HMII) - 7 (100%) - -
Baseline pump speed (rpm) - 9,000 (8,400–9,100) - -
LV EDV (ml) 284 (220–350) 104 (90–226) 129 (92–187) 75 (65–86)Table 1. 0.0025
LV ESV (ml) 240 (183–300) 77 (61–160) 89 (65–131) 27 (26–32)Table 1. <0.001
LV EF 17 (13–22) 28 (24–33) 28 (23–33) 63 (59–65)Table 1.Table 1. <0.001
HR 105 (76–117) 68 (61–84) 61 (58–65) 60 (56–66) 0.233
Data presented as median (interquartile range).
*p < 0.05 vs. LVAD group.
p < 0.05 between DCM and control groups.
DCM, dilated cardiomyopathy; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; HMII, HeartMate II; HR, heart rate; LV, left ventricle; LVAD, left ventricular assist device.

Blood Flow Patterns

Figure 1A and Video 1, Supplemental Digital Content 1, show a sequence of instantaneous vector maps of LV blood velocity for representative subjects of the LVAD, DCM, and healthy groups. Continuous suction from the LVAD cannula drives a strong mitral jet that extends all the way from the LV base to the apex and persists throughout the whole cardiac cycle (Figure 1A, left column). By comparison, in nontreated LVs, the filling jet rolls up into a CW swirling pattern that redirects blood towards the aortic valve during late diastole and early systole (Figure 1A, center and right columns).

Figure 1.
Figure 1.:
Velocity maps and vortex properties. A, Velocity maps with vortices highlighted for an left ventricular assist device (LVAD) case (first column), a dilated cardiomyopathy (DCM) case (second column), and a healthy control case (third column) at different time instants within the cardiac cycle (rows as % of RR interval). The blue-green colors and black arrows indicate the instantaneous velocity magnitude and direction. The red (and white) ellipses represent the main clockwise (CW; and secondary counter-CW) vortices. B, Boxplots and scatter plots of vortex properties (circulation Γ CW, radius R CW, and longitudinal position X CW of the CW main vortex) for patients with LVAD support (blue), DCM patients (orange), and healthy controls (green). $p < 0.05 vs. LVAD group; #p < 0.05 between DCM and healthy groups. $p < 0.05 vs. LVAD group. C, Velocity pulsatility (VP) maps for the same cases of A; white arrows represent the time-averaged velocity field. D, Boxplots and scatter plots of velocity pulsatility index (VPI) for the same groups of B.

Quantification of vortex properties (Figure 1B and Table 2) showed that the main CW vortex was significantly stronger in the LVAD and DCM groups than in the normal group with values of ΓCW = 111 (76–148), 99 (73–133), and 60 (38–76) cm2/s, for the LVAD, DCM, and normal groups, respectively (p = 0.007). The main vortex was also larger and located closer to the apex in DCM and LVAD subjects than in normals (RCW = 1.0 [0.9–1.5], 1 [0.7–1.2], and 0.7 [0.5–0.8] cm, p = 0.01; XCW = 0.4 [0.4–0.4], 0.4 [0.3–0.5], and 0.3 [0.2–0.3], p < 0.001). Due to the more symmetric flow channel created by LVAD support, the secondary CCW vortex was stronger in LVAD than in DCM and normal groups (ΓCCW = 43 [25–71], 14 [8–17], and 8 [5–16] cm2/s, p = 0.05).

Table 2. - Flow Properties in the Three Groups
p Value
Vortex circulation (cm2/s) CCW 43 (25–71) 14 (8–17) 8 (5–16)
CW 111 (76–148) 99 (73–133) 60 (38–76)Table 2. 0.007
Vortices circulation ratio (cm2/s) CCW/CW 0.4 (0.3–0.6) 0.1 (0.1–0.2) 0.1 (0.1–0.3) 0.02
Vortex radius (cm) CCW 0.4 (0.3–0.8) 0.3 (0.2–0.4) 0.2 (0.1–0.3)
CW 1 (0.9–1.4) 1 (0.7–1.2) 0.7 (0.5–0.8) 0.01
Vortex centroid location (nd) CCW 0.3 (0.3–0.5) 0.3 (0.2–0.4) 0.3 (0.3–0.4)
CW 0.4 (0.4–0.4) 0.4 (0.3–0.5) 0.3 (0.2–0.3)Table 2.Table 2. <0.001
Pulsatility index (nd) 1.1 (1.1–1.4) 2.5 (2.2–2.8)Table 2. 2.3 (2.1–2.7)Table 2. <0.001
Residence time (seconds) Avg 0.4 (0.3–0.6) 1.9 (1.4–2.3)Table 2. 1.4 (1.1–1.6)Table 2. <0.001
Max 2.2 (1.4–2.8) 5.2 (4.7–5.9)Table 2. 5.4 (4.6–6.1)Table 2. 0.001
Area of regions with T R > 2 seconds (%) 0.1 (0–5.9) 40.7 (24.6–54.4)Table 2. 27.1 (18.1–31.4) 0.003
CSI (100/s) Avg 1.7 (0.9–1.8) 1.9 (1.5–2.3) 1.5 (1.4–2.2)
Max 6.4 (5–7) 6 (4.6–7.1) 5 (3.9–6.7)
Area of regions with CSI > 200/s 27.5 (10.6–41.4) 37.6 (24–49.8) 32.9 (22.6–39.3)
Data presented as median (interquartile range).
*p < 0.05 between DCM and control groups.
p < 0.05 vs. LVAD group.
CCW, counterclockwise; CSI, cumulated shear index; CW, clockwise; DCM, dilated cardiomyopathy; LVAD, left ventricular assist device; TR, residence time.

LV blood flow in LVAD patients was less pulsatile over the cardiac cycle than in the two control groups (Figure 1D) with values of VPI = 1.1 (1.1–1.4), 2.5 (2.2–2.8), and 2.3 (2.1–2.7), p < 0.001. Notably, their VP maps (Figure 1C) displayed a region of low pulsatility that extended from the mitral valve to the pump cannula and colocalized with the jet induced by cannula suction. In contrast, pulsatility was higher and more uniformly distributed in DCM and healthy subjects.

Intraventricular Blood Transit and Shear Exposure

Figure 2A and Video 2, Supplemental Digital Content 2, display instantaneous maps of residence time, , for the same representative cases shown in Figure 1. These maps suggest that suction from the cannula cleared the apical portion of the LV cavity in LVAD subjects. In contrast, nonimplanted DCM patients showed large regions of increased residence time that colocalized with the persistent CW diastolic vortex typically found in these patients.16 Consistent with these results, the average LV residence time was lowest in the LVAD group (Average = 0.4 [0.3–0.6] s, p < 0.001), followed by the DCM (1.9 [1.4–2.3] s) and the normal (1.4 [1.1–1.6] s) groups (Figure 2B and Table 2). Furthermore, LVAD patients had significantly smaller regions with seconds than DCM patients and normal controls (SR,2s = 0.1 [0–5.9], 40.7 [24.6–54.4], and 27.1 [18.1–31.4]%, p = 0.003).

Figure 2.
Figure 2.:
Residence time. A, Residence time maps at peak R-wave after 5 seconds of integration for the same cases of Figure 1A; black arrows indicate the velocity instantaneous magnitude and direction. B, Boxplots and scatter plot of the instantaneous space-averaged residence time at peak R-wave after 5 seconds of integration for the same groups of Figure 1B. $p < 0.05 vs. left ventricular assist device (LVAD) group. DCM, dilated cardiomyopathy.

Maps of the cumulative shear index CSI(x,y,t) are shown in Figure 3A and Video 3, Supplemental Digital Content 3, In LVAD patients, the basal-to-apical jet driven by cannula suction created continuous shear exposure along the thin edges of the jet (see Figures 1 and 3A, Supplemental Digital Content, This pattern led to higher cumulative shear in those regions and to lower cumulative shear in the core of the jet (Figure 3B). In non-LVAD subjects, the shear exposure had more complex dynamics: it was mostly localized at the edges of the E- and A-waves’ filling jets and rolled up driven by the main CW LV vortex (see Figure 1, Supplemental Digital Content, Subsequently, the differences in LV washout between the normal and the DCM patients transport had important consequences in terms of cumulated shear. While in normals most of the shear-exposed blood was ejected during systole, a substantial amount of shear-exposed blood could remain trapped in the larger, more persistent vortex of the DCMs (compare central and right panels of Figure 3A). Overall, LVAD patients had slightly higher maximum values of CSI (CSImax = 6.4 [5–7], 6 [4.6–7.1], and 5 [3.9–6.7]/s) (Figure 3B and Table 2). The IQRs in Table 2 suggest that the average value cumulative shear in the three groups should range between 90 and 230/s, and these data were used to establish CSI > 200/s as a threshold for elevated cumulative shear in our study. Based on this threshold, we found that the fraction of LV chamber size occupied by blood with elevated CSI was highest in the DCM cohort and smallest in LVAD patients (Table 2).

Figure 3.
Figure 3.:
Cumulative shear. A, Cumulative shear index (CSI) maps at peak R-wave after 5 seconds of integration for the same cases of Figure 1A. B, Boxplots and scatter plot of the instantaneous maximum CSI at peak R-wave after 5 seconds of integration for the same groups of Figure 1B. DCM, dilated cardiomyopathy; LVAD, left ventricular assist device.

To illustrate the application of our blood flow imaging tools to patients implanted with third-generation LVADs, we imaged one patient treated with a Heartmate 3 device. The blood flow patterns in this patient (see Figure 2, Supplemental Digital Content, were qualitatively similar to those observed in the Heartmate II cohort. Of note, a persistent CCW vortex caused relatively high values of TR and cumulative shear near the inferolateral LV wall. However, this pattern was not uncommon in the Heartmate II patients, particularly at low pump speeds (Figure 5).

Figure 4.
Figure 4.:
TR and cumulative shear index (CSI) in left ventricular assist device (LVAD) support with aortic insufficiency. A–B, Residence time maps (A) and CSI maps (B) for a representative LVAD-implanted patient with severe aortic insufficiency at different time instants within the cardiac cycle (as % of RR interval). The checkered region highlights regurgitant blood entering the LV through the aortic valve. C–D, Boxplots and scatter plots of the space-averaged residence time (C) and size of the high shear region (CSI > 200/s; D) at peak R-wave after 5 seconds of integration grouped according to the degree of aortic insufficiency.
Figure 5.
Figure 5.:
Ramp study. Velocity (A), residence time (B), cumulative shear index (CSI; C), and velocity pulsatility (VP; D) maps for a sample left ventricular assist device implanted patient at different pump speeds (columns) at peak R-wave after 5 seconds of integration.

The Effect of Aortic Insufficiency

The relatively widespread of the data in the LVAD group (Figures 2B and 3B) motivated a more detailed analysis of this group based on the presence of AI at baseline pump speed (i.e., AI cohort vs. no-AI cohort). Figure 4A displays a sequence of maps of spanning the cardiac cycle for a representative subject of the AI cohort. The plots illustrate how a substantial volume of blood returns to the LV through the aortic valve forming a backflow jet that alternates with the filling jet flowing through the mitral valve. The interaction between these two jets forces intraventricular blood to oscillate back and forth in the chamber, thus impairing LV washout and increasing residence time. This interaction could be enhanced by LVAD support given that cannula suction drives the backflow blood region all the way to the LV apex, as shown in the example case of Figure 4A. In contrast, AI backflow is observed to remain confined near the LV base in nonimplanted patients (see Figure 2, Supplemental Digital Content,

Consistent with these observed changes in LV flow patterns, LVAD patients in the AI cohort showed an increase in LV blood average residence time (average = 0.8 [0.6–1.3] vs. 0.3 [0.2–0.3] s; Figure 4C) and size of the region with TR > 2 seconds (SR,2s = 11.1 [5.6–30.6]% vs. 0.1 [0.0–0.2]%) compared to the non-AI cohort. Notably, these differences were observed even if blood reentering the LV from the aortic root was tagged with a seconds boundary condition, an approximation that likely underestimates the true residence time in the AI cohort. The deficiency in blood clearing observed in the AI cohort also caused shear-exposed blood to remain inside the LV for a longer time, allowing for larger regions of elevated cumulative shear to form (e.g. compare the CSI(x,y) map of the AI case in Figure 4B with the non-AI case of Figure 3B). Thus, the regions of elevated shear were found to occupy more space in the AI cohort than in the non-AI cohort with values of = 42.6 (41.4–50.5)% and 10.6 (7.2–15.8)% (Figure 4D), even if the maximum values of CSI were comparable for the two cohorts. See Table 3 for a summary of residence time and cumulative shear in the AI and non-AI cohorts.

Table 3. - Aortic Insufficiency
None-to-Low Moderate-to-Severe
N 4 3
Residence time (second) Avg 0.3 (0.2–0.3) 0.8 (0.6–1.3)
Max 1.6 (1.0–2.2) 3.4 (2.5–4.2)
Size of regions with T R > 2 seconds (%) 0.1 (0.0–0.2) 11.1 (5.6–30.6)
CSI (100/s) Avg 0.9 (0.8–1.1) 2.0 (1.9–2.3)
Max 5.6 (4.0–7.2) 6.4 (5.9–6.9)
Size of regions with CSI > 200/s 10.6 (7.2–15.8) 42.6 (41.4–50.5)
Residence time (TR) and cumulative shear indices (CSIs) for the left ventricular assist device patients split into two subsets according to the degree of aortic regurgitation. Data presented as median (interquartile range).

Small Variations of LVAD Speed Around Nominal Speed Scarcely Affect LV Flow Patterns

We mapped blood flow velocity, pulsatility, residence time, and cumulative shear during ramp studies (Figure 5) with small LVAD speed changes (each patient’s baseline speed ± 400 rpm). These data suggest that, excepting flow pulsatility, the main features of intraventricular flow in LVAD patients remain almost constant with these pump speed variations (see Table 1, Supplemental Digital Content, Overall, with increased pump speed, a more continuous direct jet between the mitral valve and the LVAD cannula tends to be established, leading to a decrease in pulsatility, but this does not significantly impact global indices of residence time or shear stresses.


This study has quantified how LVAD support affects intraventricular flow patterns, shear stresses, and blood transport in a small cohort of axial flow LVAD patients compared to DCM patients and healthy controls. Despite the increasingly widespread use of LVADs in advanced HF, the characterization of blood flow inside the LVAD-assisted ventricle has not been well studied and may provide useful information to decrease adverse events. Our results reveal that, while substantially altering the normal LV flow pattern, LVAD support largely reverses the negative impact of DCM on blood transit through the ventricle. LVAD patients were found to have significantly lower values of intraventricular residence time than DCM patients and even normals. However, this reduction in residence time did not correspond with a reduction in blood shear exposure. We show that these results are relatively independent of LVAD pump speed for small changes around the clinically indicated baseline values, whereas blood transit worsened when LVAD support caused moderate or severe AI. These findings provide new insight into blood flow dynamics in the LVAD-supported ventricle and may have important implications for device design and programming.

Left Ventricular Flow Patterns Under LVAD Support

The hemodynamics of the native LV are dominated by vortices that form during early filling and atrial contraction and evolve into a large CW swirling cell that follows the chiral arrangement of the LV inflow tract, the main chamber, and the aortic LV outflow tract.10,11,35 LVAD treatment significantly affects these dynamics due to the suction forces created by the pump.36,37 We found that LVAD treatment reroutes the transit of blood through the LV so that it forms a straight channel between the mitral valve and the LVAD cannula, instead of following a chiral path. The vorticity associated with the boundaries of the jet in LVAD-supported ventricles still results in a vortex ring. However, consistent with existing in vitro data,21 the net CW rotational motion of blood found in the native LV is reduced by LVAD treatment. In cases with AI, backflow from the aortic tract formed a CCW swirling blood “compartment” that interacted with the natural CW swirling region, creating two separate pockets of blood that rotated inside the LV in alternating directions, and affecting blood residence time and its exposure to shear. Suction from the cannula may accentuate the effects of AI in LVAD patients by driving the aortic backflow close to the LV apex. Larger studies are required to correlate these findings with clinical outcomes, such as thromboembolic events, bleeding, and HF.

The Efficiency of Rerouting Blood Transit Through the LV

While the native ventricle alternates between reservoir (diastole) and booster (systole) function, LVAD support forces the ventricle to operate as a conduit. The potential implications of blood rerouting in the total work exerted by the native heart remain debatable after two decades of investigation.11,14,15,35,38 The implications for platelet activation and thrombosis, which are particularly relevant in LVAD patients, have received less attention. Simulation studies14 suggest that the chiral arrangement of the inflow tract, the main LV chamber, and the outflow tract minimize the shear between the filling jet and the intraventricular blood. In contrast, we found that LVAD patients experienced more exposure to shear in the thin layers surrounding the inflow jets than DCM patients and normals. The maximum values of shear found in LVAD-treated patients (Figure 3B, note that a CSI of 1000/s corresponds to 4 Pa assuming blood viscosity equals ) are significantly lower than the reported threshold for platelet activation (50 Pa). Still, they are comparable to the threshold for von Willebrand factor activation/degradation (9 Pa).39 One could argue that the instantaneous shear stresses inside the ventricle are orders of magnitude lower than inside the LVAD pump. However, it is challenging to measure these stresses or image flow inside the implanted pump. Thus, the intraventricular CSI values proposed in this study could be a useful metric to evaluate hemodynamic disturbances caused by LVAD treatment.

Our flow measurements in LVAD and healthy subjects suggest that, by establishing a shorter, straighter route for blood transit inside the ventricle, LVAD treatment significantly decreases the LV residence time of blood, but the same trend did not apply to cumulative exposure to shear. We found that strong, persistent vortices in DCM patients can trap the blood inside the LV for long times, during which blood is continuously exposed to shear. LVAD treatment causes high instantaneous shear stresses along the edges of the longitudinal jet created by suction at the cannula. However, by improving LV blood transit, it prevents blood from being exposed to increased shear for prolonged periods. When LVAD suction was strong enough to trigger AI, the rerouting of blood transit by LVAD treatment was disturbed, becoming less efficient in balancing residence time with blood exposure to shear. These results highlight the need to consider the efficiency of rerouting LV blood transit by jointly assessing blood stasis and cumulative shear, as they are tightly interrelated, but do not necessarily vary in the same direction after an intervention.

Clinical Implications

LVAD therapy is associated with “hemocompatibility events” such as cerebrovascular accidents, pump thrombosis, and hemolysis potentially related to thrombosis inside the ventricle. These complications occur at similar rates in both axial and centrifugal pumps. Currently, there is a lack of clinical tools to guide optimal LVAD settings and cannula placement. Echocardiographic ramp studies are sometimes used to diagnose LVAD thrombosis, test for ventricular recovery, facilitate device weaning, and choose pump speeds, but there is no evidence these studies decrease rates of thromboembolism. Thrombosis is associated with platelet activation and stasis, and hemolysis is known to be associated with high shear stress. Still, currently, these quantities are difficult to estimate in clinical practice; therefore, the risks of thrombosis and hemolysis are difficult to mitigate. In this study, we have demonstrated the utility of echo-CDV to evaluate LV hemodynamics in patients with LVADs, and to quantify both LV stasis and shear exposure. We observed that LV stasis was reduced in the group with LVADs compared to the DCM group, while shear was not. Besides, pump settings that resulted in significant AI were associated with a relative increase in stasis and shear exposure. Though AI is a relatively frequent complication of LVAD use,40 an association between AI and thromboembolic events has not been established. Nonetheless, our results suggest a potential connection between these two phenomena. In addition, echo-CDV may be useful in guiding LVAD cannula placement and optimal pump settings and therefore decrease the rate of complications and improve outcomes.

Strengths and Limitations

The development of in silico and in vitro analyses of intraventricular flow in the LVAD-supported ventricle21,23–26 has not been paralleled by a similar surge in preclinical or clinical experiments. This lag may be due partly to the inability to perform magnetic resonance imaging (MRI) on LVAD patients. In the current study, we have demonstrated the utility of measuring intraventricular flows with ultrasound. We expect that these new in vivo analyses should facilitate further work to overcome limitations of in vitro and in silico models such as the difficulty of modeling myocardial motion, valve dynamics, and the physiologic response to changes in LVAD support.

The noninvasiveness and portability of echocardiography make this modality well suited for the assessment of intraventricular flow in LVAD patients. In the last decade, several echocardiographic methods have been developed to visualize LV blood flow.41 Particle image velocimetry (PIV) applied to contrast ultrasound sequences (echo-PIV) has proven useful and, given that contrast agents seem to be safe in patients implanted with third-generation LVADs,42 it is a promising modality to quantify LV flow in these patients. However, echo-PIV requires fine-tuning of the contrast agent infusion,43 which may be particularly challenging in LVAD patients whose pumps destroy contrast agent bubbles.44

When used in the LV apical long-axis view, echo-CDV has good agreement with in silico,45in vitro,27 and in vivo reference methods.30 However, it is limited to a 2D plane of the LV and has not been quantitatively validated in LVAD patients yet, partly due to the difficulty of performing MRI in these patients. Because it imposes free slip boundary conditions at the LV endocardium, echo-CDV likely underestimates endocardial blood shear. Nevertheless, the cumulative shear values obtained here are in good agreement with values measured in vitro with PIV.33

The LVAD cohort in this pilot study is small and heterogeneous, although most patients had nonischemic cardiomyopathy (86%) and were >50 years of age (86%). Considering this limitation, the DCM cohort was included to control for age and nonischemic cardiomyopathy. The study and control cohorts were imaged at two different institutions which could introduce intersite variability. It would have been desirable to follow the patients’ clinical history and complication events, but unfortunately, such information is not available. The study was performed in axial flow devices; further studies shall address differences with centrifugal flow devices as well as next-generation devices. Larger studies with different pumps are needed to understand whether complication events are caused by the LV flow pattern, by the pump design, or by a combination of the two. Echo-CDV allows observing only intraventricular flow patterns once an LVAD is implanted, although a comprehensive preoperative analysis of the LV-LVAD system would require significantly involved modeling of multiscale biomechanical phenomena.


In this pilot study, we demonstrated the potential utility of echo-CDV to characterize intracardiac flows in patients treated with LVADs. Compared with patients with DCM, those with LVADs had lower residence times and similar cumulative shear. Aortic regurgitation significantly affected flow patterns, resulting in increased stasis and shear. Future studies are warranted to evaluate the utility of this technique to help optimize the treatment of patients with LVADs and decrease adverse events.


1. Kirklin JK, Pagani FD, Kormos RL, et al. Eighth annual INTERMACS report: Special focus on framing the impact of adverse events. J Hear Lung Transplant. 2017; 36:1080–1086
2. Yancy CW, Jessup M, Bozkurt B, et al. 2017 ACC/AHA/HFSA focused update of the 2013 ACCF/AHA guideline for the management of heart failure: A report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines and the heart failure society of America. J Am Coll Cardiol. 2017; 70:776–803
3. Slaughter MS, Rogers JG, Milano CA, et al.; HeartMate II Investigators. Advanced heart failure treated with continuous-flow left ventricular assist device. N Engl J Med. 2009; 361:2241–2251
4. Rogers JG, Pagani FD, Tatooles AJ, et al. Intrapericardial left ventricular assist device for advanced heart failure. N Engl J Med. 2017; 376:451–460
5. Mehra MR, Naka Y, Uriel N, et al. A fully magnetically levitated circulatory pump for advanced heart failure. N Engl J Med. 2017; 376:440–450
6. Bartoli CR, Zhang D, Kang J, et al. Clinical and in vitro evidence that subclinical hemolysis contributes to LVAD thrombosis. Ann Thorac Surg. 2018; 105:807–814
7. Lowe GDO. Virchow’s triad revisited: Abnormal flow. Pathophysiol Haemost Thromb. 2003; 33:455–457
8. Shah P, Birk S, Maltais S, et al. Left ventricular assist device outcomes based on flow configuration and pre-operative left ventricular dimension: An interagency registry for mechanically assisted circulatory support analysis. J Hear Lung Transplant. 2017; 36:640–649
9. Lip GY, Piotrponikowski P, Andreotti F, et al.; Heart Failure Association (EHFA) of the European Society of Cardiology (ESC) and the ESC Working Group on Thrombosis. Thromboembolism and antithrombotic therapy for heart failure in sinus rhythm: An executive summary of a joint consensus document from the ESC Heart Failure Association and the ESC Working Group on Thrombosis. Thromb Haemost. 2012; 108:1009–1022
10. Rodevand O, Bjornerheim R, Edvardsen T, Smiseth OA, Ihlen H. Diastolic flow pattern in the normal left ventricle. J Am Soc Echocardiogr. 1999; 12:500–507
11. Kilner PJ, Yang GZ, Wilkes AJ, Mohiaddin RH, Firmin DN, Yacoub MH. Asymmetric redirection of flow through the heart. Nature. 2000; 404:759–761
12. Bolger AF, Heiberg E, Karlsson M, et al. Transit of blood flow through the human left ventricle mapped by cardiovascular magnetic resonance. J Cardiovasc Magn Reson. 2007; 9:741–747
13. Martínez-Legazpi P, Bermejo J, Benito Y, et al. Contribution of the diastolic vortex ring to left ventricular filling. J Am Coll Cardiol. 2014; 64:1711–1721
14. Pedrizzetti G, Domenichini F. Nature optimizes the swirling flow in the human left ventricle. Phys Rev Lett. 2005; 95:108101
15. Watanabe H, Sugiura S, Hisada T. The looped heart does not save energy by maintaining the momentum of blood flowing in the ventricle. Am J Physiol Heart Circ Physiol. 2008; 294:H2191–H2196
16. Hendabadi S, Bermejo J, Benito Y, et al. Topology of blood transport in the human left ventricle by novel processing of Doppler echocardiography. Ann Biomed Eng. 2013; 41:2603–2616
17. Benito Y, Martinez-Legazpi P, Rossini L, et al. Age-dependence of flow homeostasis in the left ventricle. Front Physiol. 2019; 10:485
18. Thamsen B, Blümel B, Schaller J, et al. Numerical analysis of blood damage potential of the HeartMate II and HeartWare HVAD rotary blood pumps. Artif Organs. 2015; 39:651–659
19. Faludi R, Szulik M, D’hooge J, et al. Left ventricular flow patterns in healthy subjects and patients with prosthetic mitral valves: An in vivo study using echocardiographic particle image velocimetry. J Thorac Cardiovasc Surg. 2010; 139:1501–1510
20. Rossini L, Martinez-Legazpi P, Vu V, et al. A clinical method for mapping and quantifying blood stasis in the left ventricle. J Biomech. 2016; 49:2152–2161
21. Wong K, Samaroo G, Ling I, et al. Intraventricular flow patterns and stasis in the LVAD-assisted heart. J Biomech. 2014; 47:1485–1494
22. Reider C, Moon J, Ramesh V, et al. Intraventricular thrombus formation in the LVAD-assisted heart studied in a mock circulatory loop. Meccanica. 2017; 52:515–528
23. Liao S, Neidlin M, Li Z, Simpson B, Gregory SD. Ventricular flow dynamics with varying LVAD inflow cannula lengths: In-silico evaluation in a multiscale model. J Biomech. 2018; 72:106–115
24. Prisco AR, Aliseda A, Beckman JA, Mokadam NA, Mahr C, Garcia GJM. Impact of LVAD implantation site on ventricular blood stagnation. ASAIO J. 2017; 63:392–400
25. Chivukula VK, Beckman JA, Prisco AR, et al. Left ventricular assist device inflow cannula angle and thrombosis risk. Circ Hear Fail. 2018; 11:e004325
26. Klotz S, Meyer-Saraei R, Frydrychowicz A, et al. Proposing a novel technique to exclude the left ventricle with an assist device: Insights from 4-dimensional flow magnetic resonance imaging. Eur J Cardio-thoracic Surg. 2016; 50:439–445
27. Garcia D, Del Alamo JC, Tanne D, et al. Two-dimensional intraventricular flow mapping by digital processing conventional color-Doppler echocardiography images. IEEE Trans Med Imaging. 2010; 29:1701–1713
28. Martinez-Legazpi P, Rossini L, Pérez Del Villar C, et al. Stasis mapping using ultrasound: A prospective study in acute myocardial infarction. JACC Cardiovasc Imaging. 2018; 11:514–515
29. Rossini L, Martinez-Legazpi P, Benito Y, et al. Clinical assessment of intraventricular blood transport in patients undergoing cardiac resynchronization therapy. Meccanica. 2017; 52:563–576
30. Bermejo J, Benito Y, Alhama M, et al. Intraventricular vortex properties in nonischemic dilated cardiomyopathy. AJP Hear Circ Physiol. 2014; 306:H718–H729
31. Hellums JD. 1993 Whitaker lecture: Biorheology in thrombosis research. Ann Biomed Eng. 1994; 22:445–455
32. Ramstack JM, Zuckerman L, Mockros LF. Shear-induced activation of platelets. J Biomech. 1979; 12:113–125
33. Vu V, Rossini L, Montes R, et al. Mitral valve prosthesis design affects hemodynamic stasis and shear in the dilated left ventricle. Ann Biomed Eng. 2019; 47:1265–1280
34. Fraser KH, Zhang T, Taskin ME, Griffith BP, Wu ZJ. A quantitative comparison of mechanical blood damage parameters in rotary ventricular assist devices: Shear stress, exposure time and hemolysis index. J Biomech Eng. 2012; 134:081002
35. Bermejo J, Martínez-Legazpi P, del Álamo JC. The clinical assessment of intraventricular flows. Annu Rev Fluid Mech. 2015; 47:315–342
36. Goodwin M, Nemeh HW, Borgi J, Paone G, Morgan JA. Resolution of mitral regurgitation with left ventricular assist device support. Ann Thorac Surg. 2017; 104:811–818
37. Cowger J, Pagani FD, Haft JW, Romano MA, Aaronson KD, Kolias TJ. The development of aortic insufficiency in LVAD supported patients. Circ Hear Fail. 2015; 3:668–674
38. Seo JH, Mittal R. Effect of diastolic flow patterns on the function of the left ventricle. Phys Fluids. 2013; 25:110801
39. Thamsen B, Mevert R, Lommel M, et al. A two-stage rotary blood pump design with potentially lower blood trauma: A computational study. Int J Artif Organs. 2016; 39:178–183
40. Jorde UP, Uriel N, Nahumi N, et al. Prevalence, significance, and management of aortic insufficiency in continuous flow left ventricular assist device recipients. Circ Hear Fail. 2014; 7:310–319
41. Sengupta PP, Pedrizzetti G, Kilner PJ, et al. Emerging trends in CV flow visualization. JACC Cardiovasc Imaging. 2012; 5:305–316
42. Schinkel AFL, Akin S, Strachinaru M, et al. Safety and feasibility of contrast echocardiography for the evaluation of patients with HeartMate 3 left ventricular assist devices. Eur Heart J Cardiovasc Imaging. 2018; 19:690–693
43. Gao H, Claus P, Amzulescu MS, Stankovic I, D’hooge J, Voigt JU. How to optimize intracardiac blood flow tracking by echocardiographic particle image velocimetry? Exploring the influence of data acquisition using computer-generated data sets. Eur Heart J Cardiovasc Imaging. 2012; 13:490–499
44. Platts DG, Bartnikowski N, Gregory SD, Scalia GM, Fraser JF. Contrast microsphere destruction by a continuous flow ventricular assist device: An In Vitro evaluation using a mock circulation loop. Biomed Res Int. 2017; 2017:4907898
45. Uejima T, Koike A, Sawada H, et al. A new echocardiographic method for identifying vortex flow in the left ventricle: Numerical validation. Ultrasound Med Biol. 2010; 36:772–788

left ventricular assist device; echocardiography; heart failure

Supplemental Digital Content

Copyright © 2020 by the ASAIO