JACC: CARDIOVASCULAR INTERVENTIONS
-, 2016
ª 2016 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION
PUBLISHED BY ELSEVIER
EDITORIAL COMMENT
Comparing Hemodynamics of
Contemporary Mechanical
Circulatory Support
Moving from In Silico to In Vivo Results*
Morton J. Kern, MD, MSCAI, Arnold H. Seto, MD, MPA
T he degree of myocardial recovery determines
the function and ultimately the survival of a
patient in cardiogenic shock. Therapeutic
efforts to improve myocardial function include inten-
sive pharmacotherapy (inotropes, vasopressors), and
use of mechanical circulatory support (MCS) devices,
including the intra-aortic balloon pump, the left ven-
tricular (LV)-to-aorta axial pump catheter (Impella CP
[IMP]) and the left atrial (LA)-to-femoral artery pump
(TandemHeart [TH]). All act through different mecha-
nisms, and to different degrees, to unload the heart,
increase blood pressure, cardiac output and systemic
perfusion, while reducing myocardial work and oxy-
gen consumption.
The hemodynamics of MCS devices are known to
us predominately from computer simulations (“in
silico” models) (1,2), in vivo animal studies (3,4), and
series of cardiogenic shock patients (5,6). In practice,
a specific device is selected based both on its pre-
sumed hemodynamic power as well as the patient’s
clinical features (e.g., status of the aortic valve, pe-
ripheral vascular disease, etc.) and technical chal-
lenges of device implantation (e.g., cannula size, ease
and speed of insertion, and availability of a trained
team on call).
Because of the different mechanisms of action, we
can expect unique hemodynamic profiles, especially
*Editorials published in JACC: Cardiovascular Interventions reflect the
views of the authors and do not necessarily represent the views of JACC:
Cardiovascular Interventions or the American College of Cardiology.
From the Veterans Administration Long Beach Health Care System,
University of California, Long Beach, California. Dr. Kern is a speaker for
Abiomed Inc., St. Jude Medical, Volcano Therapeutics, Acist Medical,
Opsens Inc., and Heartflow Inc. Dr. Seto has reported that he has no
relationships relevant to the contents of this article to disclose.
VOL. -, NO.
ISSN 1936-8798/$36.00
http://dx.doi.org/10.1016/j.jcin.2016.09.012
involving LV unloading and other performance
characteristics. To review, the intra-aortic balloon
pump inflates in diastole, generating an augmented
pressure pulse and deflates abruptly in systole
reducing LV afterload. The net hemodynamic effects
of intra-aortic balloon pump produce (to variable
degrees) an increase in mean arterial pressure, coro-
nary perfusion, and mild decrease in myocardial
work. The magnitude of these effects is smaller than
those of either the IMP or TH and requires a contracting
heart to function effectively. The IMP works through a
catheter-mounted axial pump, drawing blood from the
LV and expelling it into the ascending aorta. The TH
pulls blood from the left atrium through a trans-septal
inserted large (22-F) inflow cannula to an external
centrifugal pump, returning the blood to the arterial
circulation through the femoral artery. Both IMP and
TH substantially reduce LV stroke volume, LV loading,
and myocardial work, and increase cardiac output and
systemic pressure. Although the hemodynamics of
IMP and TH have been explored extensively by com-
puter simulations (1,2), it remains largely unknown to
what degree the MCS devices perform under in vivo
conditions, in the same subject with the same flow
rates. Moreover, hemodynamic differences from
computer simulations would be anticipated since any
in silico model cannot not completely replicate the
more complex behavior of the intact, in vivo animal
cardiovascular system (7).
Moving from in silico to in vivo, Weil et al. (8)
compared the hemodynamics of the IMP and TH
from a closed chest animal model of myocardial
infarction. Using a LV pressure–volume (PV)
catheter, PV loops were obtained before and after a
moderate-sized myocardial infarction was induced
with a 2-h circumflex coronary artery occlusion.
2 Kern and Seto
Hemodynamics of Contemporary Mechanical Circulatory Support
T A B L E 1 Comparative Hemodynamic Effects of Percutaneous Ventricular
Assist Devices
IMP TH
Heart rate (beats/min) ¼ ¼
Mean aortic pressure (mm Hg) ¼ ¼
Aortic pulse pressure (mm Hg), 34  2 to 27  4 34  2 to 14  1*†
control to device activation
LV end-diastolic pressure (mm Hg) 15  1 to 11  1* 15  2 to 7  4*†
Pulmonary capillary wedge pressure 15  2 to 13  2 15  2 to 9  1*
(mm Hg)
End-systolic elastance (mm Hg/ml) ¼ ¼
LV stiffness constant (b) þ þ
LV dP/dtmax reduction - þþ
LV stroke work reduction þ þþþ
LV pressure volume area reduction þ þþþ
LV preload recruitable stroke work - þ
reduction
Values are mean  SEM. *p < 0.05 vs. post-MI; †p < 0.05 vs. Impella CP (IMP). Adapted from
Weil et al. (8).
þ ¼ improved compared with control MI state; ¼ ¼ equivalent change between devices; - ¼ no
improvement compared to control; LV ¼ left ventricular; TH ¼ TandemHeart.
Although frank cardiogenic shock was not produced,
myocardial infarction shifted the PV loop to the right
due to an increased LV end-diastolic pressure
accompanied by reduced systolic pressure with
mildly reduced stroke volume. Both MCS devices
were then sequentially placed, with the order ran-
domized and the goal flow rates matched.
Both MCS devices maintained the aortic pressure
but, compared to IMP, the TH had a greater reduction
in LV end-diastolic pressure (but not volume), native
LV stroke volume, dP/dt max , stroke work, PV area,
and preload stroke work slope (Table 1). In short and
at odds with some prior reports (2–5), TH unloaded
and rested the heart to a greater degree than IMP. Of
particular interest and unique to this study was that
the hemodynamics and PV loops of the 2 MCS devices
were not only different from each other but also diff-
erent from the computer model predictions (Figure 1).
The investigators deserve our compliments for their
rigorous experimental methods. The greater decrease
in native stroke work by TH suggests a more powerful
volume unloading of the heart occurs despite the
relative increase in native end-systolic volume
compared with IMP as demonstrated by the different
shapes of the in vivo PV loops (Figure 1). The IMP
maintained end-systolic pressure, reduced stroke
volume and shifted the PV loop leftward, whereas the
TH increased end-systolic pressure with a pronounced
reduction in stroke volume and preload. As a result,
the TH produced a greater decrease in PV area (a well-
validated index of myocardial oxygen consumption)
than the IMP in the animal model, a contrasting
observation from the computer models. Conceivably,
JACC: CARDIOVASCULAR INTERVENTIONS VOL. -, NO. -, 2016
- 2016:-–-
more effective decreases in myocardial oxygen de-
mand would lead to greater myocardial recovery after
infarction, or more support during cardiogenic shock.
This well-performed study, comparing 2 commonly
used MCS devices head to head for the first time in
this way, suggests that, at comparable device flow
rates, the degree of LV unloading depends on
whether blood is withdrawn directly from the LA or
the LV. The IMP draws blood from the LV and
decreased LV end-diastolic volume and LV end-
diastolic pressure and maintained arterial pressure,
but in this study did not affect PV loop-derived
indices of myocardial work significantly. In contrast,
TH withdrawal of blood from the LA also reduced LV
end-diastolic volume and LV end-diastolic pressure,
and maintained arterial pressure while also reducing
native LV stroke volume, stroke work, dP/dtmax, PV
area, and preload-recruitable stroke work consistent
with enhanced LV unloading.
Why should the findings of Weil et al. (8) differ
from prior studies? Although neither a computer
model nor the induced animal infarction model
exactly duplicates the human cardiogenic shock
state, the in vivo conditions incorporate a number of
influences on myocardial function, arterial imped-
ance and afterload, and peripheral resistance that can
only be estimated in a more rudimentary fashion in
computer simulations. In contrast, the present study
did not truly induce a state of cardiogenic shock.
Because unloading with IMP occurs most effectively
when cardiac output is low and LV end-diastolic
pressure and volume are increased (and poorly
when these states are not present), the present model
may have unfairly tilted the scale against IMP.
Why aspiration of similar flow rates from the left
atrium compared with the LV would have such
different hemodynamic effects remains unclear.
Potentially, this result reflects the greater capacitance
of the LA (compared with the LV) to release volume to
the MCS device, particularly in euvolemic, nonshock
states, but this would presumably have been evident
from the flow rates measured. Alternatively, the au-
thors suggest that, by decreasing the LV end-diastolic
pressure to a greater extent, LA withdrawal may
improve coronary blood flow. However, coronary
blood flow was not measured in this study and would
not explain how LV end-diastolic pressure is reduced
more effectively.
The IMP and TH are powerful MCS devices that are
frequently used for high-risk percutaneous coronary
interventions and cardiogenic shock. In clinical prac-
tice, the selection of a percutaneous MCS device de-
pends not only on the hemodynamic power of the
device, but also and more important on the specific
JACC: CARDIOVASCULAR INTERVENTIONS VOL. -, NO. -, 2016 Kern and Seto 3
- 2016:-–- Hemodynamics of Contemporary Mechanical Circulatory Support

F I G U R E 1 Ventricular Effects of LA-to-Arterial MCS and LV-to-Aorta MCS

(Top) Computer simulation of ventricular effects of left atrial (LA)-to-arterial mechanical circulatory support (MCS; TH; left) and left
ventricular (LV)-to-aorta MCS (IMP; right) showing flow-dependent changes in pressure–volume loops, reductions in end-diastolic pressures,
increased end-systolic volume, and decreased LV stroke volume (from Burkhoff et al. [9]). (Bottom) PV loops for TH and IMP from the in vivo
animal model of Weil BR et al. (8) (IMP in blue and TH in green) compared with the corresponding unassisted post-myocardial infarction
baseline time-point (red). Both devices shifted the PV loop to the left and decreased LV end-diastolic pressure, but TH produced a greater
reduction in native LV stroke volume. Ao ¼ aorta; CGS ¼ cardiogenic shock; LVD ¼ left ventricular assist device.

clinical scenario. The TH takes an average of 45 to 60
min to insert, compared with <10 min for the IMP,
making the TH less useful in patients with unstable
cardiogenic shock (9). The risk of major bleeding with
the TH is also higher than IMP (25% vs. 6%) (4). The
carefully measured hemodynamic in vivo differences
of LV performance by Weil et al. (8) shift conventional
wisdom about the MCS device power order and should
bring new attention to the TH. Nonetheless, for each
patient in need of LV mechanical support, the operator
must choose the device that is most clinically and he-
modynamically appropriate.
REPRINT REQUESTS AND CORRESPONDENCE: Dr.
Morton J. Kern, Department of Medicine, Veterans
Administration Long Beach Health Care System,
Building 1, 5901 East 7th Street, Long Beach,
California 90822. E-mail: Morton.kern@va.gov.

REFERENCES

1. Rihal CS, Naidu SS, Givertz MM, et al. 2015 SCAI/
ACC/HFSA/STS clinical expert consensus state-
ment on the use of percutaneous mechanical cir-
culatory support devices in cardiovascular care.
J Am Coll Cardiol 2015;65:e7–26.
2. Burkhoff D, Cohen H, Brunckhorst C, et al., therapy with intraaortic balloon pumping for
TandemHeart Investigators Group. A randomized treatment of cardiogenic shock. Am Heart J 2006;
multicenter clinical study to evaluate the safety 152:469.e1–8.
and efficacy of the TandemHeart percutaneous 3. Burkhoff D, Naidu SS. The science
ventricular assist device versus conventional behind percutaneous hemodynamic support:
4 Kern and Seto
Hemodynamics of Contemporary Mechanical Circulatory Support
a review and comparison of support strategies.
Catheter Cardiovasc Interv 2012;80:816–29.
4. Naidu SS. Novel percutaneous cardiac assist
devices: the science of and indications for hemo-
dynamic support. Circulation 2011;123:533–43.
5. Burkhoff D. Hemodynamic support: science and
evaluation of the assisted circulation with percu-
taneous assist devices. Interventional Cardiology
Clinics 2013;2:407–16.
6. Kapur NK, Paruchuri V, Pham DT, et al. Hemo-
dynamic effects of left atrial or left ventricular
JACC: CARDIOVASCULAR INTERVENTIONS VOL.
cannulation for acute circulatory support in a
bovine model of left heart injury. ASAIO J 2015;61:
301–6.
7. Atkinson TM, Ohman EM, O’Neill WW, Rab T,
Cigarroa JE. A practical approach to mechanical
circulatory support in patients undergoing percu-
taneous coronary intervention: an interventional
perspective. J Am Coll Cardiol Intv 2016;9:871–83.
8. Weil BR, Konecny F, Suzuki G, Iyer V,
Canty JM. Comparative hemodynamic effects
of contemporary percutaneous mechanical
-, NO. -, 2016
- 2016:-–-
circulatory support devices in a porcine model
of acute myocardial infarction. J Am Coll Car-
diol Intv 2016;xx:xxx–xxx.
9. Kar B, Adkins LE, Civitello AB, et al. Clinical
experience with the TandemHeart percutaneous
ventricular assist device. Tex Heart Inst J 2006;33:
111–5.
KEY WORDS cardiac output, hemodynamics,
ventricular assist device