Background: Left ventricular assist devices (LVAD) often have an integrated algorithm estimating
pump flow rate from motor current, rotational speed and pre-set viscosity. In few
paracorporeal devices, the flow rate has recently been additionally determined by
ultrasonic sensors, which are attached to the outflow cannula. We aimed to compare
estimated and measured flow rate for an intracorporeal device in vivo.
Methods: A modified centrifugal-flow LVAD was implanted into two German Landrace pigs (both
female, weights 115 and 111 kg). Between pump outlet and outflow graft, a short piece
of PVC tubing was equipped with an ultrasonic flow sensor. Applying different rotational
pump speeds (3‒8 krpm), LVAD flow estimate (Q_LVAD) and ultrasound-measured flow (Q_US)
were recorded. Q_LVAD was related to Q_US, pump speed [krpm], pulsatility index (PI),
power [W], hemodynamic parameters (mean arterial and pulmonary pressure (MAP and MPP
[mmHg]) and hematocrit (Hct [%]) using linear regression.
Results: Overall, Q_LVAD was greater than Q_US (3.55 ± 0.73 L/min vs. 2.21 ± 0.69 L/min, p < 0.001) (Table). There was a high positive correlation between both (R = 0.86, Q_LVAD = 0.90
× Q_US + 1.55 L/min) and a moderate one between Q_LVAD and pump parameters. Referring
to Q_US as reference, there was a flow overestimation by 1.05 L/min or 21% at a physiological
flow rate of 5 L/min (Q_LVAD vs. Q_US = 6.05 vs. 5 L/min). Estimating LVAD flow from
pump, hemodynamic and laboratory parameters (Q_LVAD = 0.41 × speed + 0.25 × PI ‒ 0.09
× power + 0.01 × MAP + 0.01 × MPP ‒ 0.04 × Hct ‒ 1.63 L/min) was more accurate than
from pump parameters only (adj. R2 = 0.75 vs. 0.67).
|
Comparison of estimated and measured LVAD flow rate
|
|
Q_US (L/min)
|
2
|
3
|
4
|
5
|
6
|
7
|
8
|
|
Q_LVAD (L/min)
|
3.35
|
4.25
|
5.15
|
6.05
|
6.95
|
7.85
|
8.75
|
|
Mean difference (L/min)
|
1.35
|
1.25
|
1.15
|
1.05
|
0.95
|
0.85
|
0.75
|
|
Relative difference (%)
|
68
|
42
|
29
|
21
|
16
|
12
|
9
|
Conclusion: In vivo there is a strong correlation between estimated and measured LVAD flow rate.
However, we found a relevant flow overestimation by the LVAD at physiological flow
rates, of which users should be aware. To increase the accuracy of flow estimation,
hemodynamic and laboratory parameters should be considered in addition to pump parameters.
