MATERIALS AND METHODS
Medical records and imaging for adult patients undergoing implantation
of a primary CF-LVAD with six-month follow-up as of January 31, 2021
were reviewed. Patients with durable BiVADs, non-apical cannulation, or
lack of CTs were excluded. Data were collected prospectively in an
IRB-approved database (University of Maryland IRB Protocol
#HP-00058592).
Before April 2016, the default technique was CS. Subsequently, the LTHS
technique as described by Schmitto2, became the
preferred approach. Intraoperative IFC position was assessed with
transesophageal echocardiography. Hemodynamic and echo-guided LVAD
optimization studies were also performed in the OR.
Preoperative CT scans were obtained for surgical planning and to rule
out thoracic pathology. Postoperative scans were obtained to assess LVAD
position. Contrast was given if permitted by renal function. Serial CT
scans were not performed in stable patients. In patients with multiple
scans, the most recent contrast-enhanced CT was utilized.
Preoperative anatomic variables included LV end-diastolic dimension
(LVEDD) and body mass index (BMI). We hypothesized that a smaller LVEDD
would make optimal IFC placement more difficult. We hypothesized that
extremes of BMI would contribute to malposition either by offering less
intrathoracic space for the pump (low BMI) or by abdominal mass effect
on the diaphragm (high BMI).
Preoperative CT measurements were: midline- LV apex distance in the
axial plane (Figure 1A); and inclination of the LV apex-LV outflow tract
axis above the horizontal (LVOT angle) in the coronal plane (Figure 1B).
These measurements were derived from our prior work3,
but we hypothesized different effects on IFC position for the CF-LVAD
than observed for the axial-flow device. Since IFC position in the
CF-LVAD tends toward horizontal, we expected a higher, rather than
lower, LVOT angle to be associated with IFC malposition. For midline-LV
apex distance, we expected a larger value to be associated with less
space between apex and ribs, forcing the IFC downward and/or lateral.
Quantitative IFC position was calculated by measuring angular deviation
from an axis intersecting the LV apex and the center of the mitral valve
(MV). Angles in two orthogonal planes, anterior and lateral, aligned
with this axis were measured using Aquarius iNtuition 4.4 (TeraRecon;
Foster City, CA) (Figure 2). In the anterior plane, positive
displacement signifies angulation toward the superior wall, and negative
toward the inferior wall. In the lateral plane, positive displacement is
toward the lateral wall, and negative toward the septum. Total
malposition magnitude represents the sum of the magnitudes of the two
angles.
LV unloading was assessed using the change in LVEDD and MR compared to
the last pre-LVAD echocardiogram. MR was graded as none (0), mild (1),
moderate (2), or severe (3). Both early (1-3 months post-LVAD) and late
(6-12 months post-LVAD) measurements were used. LVAD FI was obtained
from estimated LVAD flow (L/min) divided by patient BSA
(m2) at one-, three-, and six-month timepoints. Our
institution does not routinely obtain invasive hemodynamic measurements
on stable LVAD patients.
Postoperative outcomes included PT/eCVA, HFRAs and survival free of
urgent transplant or explant for LVAD dysfunction, pump replacement, or
pump repositioning.
Pump thrombosis was either thrombus identified within the LVAD at
explant, or ≥2 of: increased lactate dehydrogenase (≥1.5× stable
baseline or ≥2.5× upper normal) or plasma hemoglobin (≥20 mg/dL);
sustained abnormal power consumption (≥±1.5 W from stable baseline, or
above manufacturer’s suggested upper normal4); and
new-onset clinical or hemodynamic signs of congestive heart failure.
Embolic CVA was a focal neurologic deficit persisting > 24
hours, or a new ischemic lesion on CT imaging.
HFRAs were any readmission for heart failure symptoms or
LVAD/hemodynamic derangements. This excluded infection and bleeding
events, INR derangements, and elective admissions for planned diagnostic
or surgical procedures. Prevalence of 30-day and overall HFRAs was
assessed.
Statistical analysis was performed with Statistica 13 (Dell, Inc; Tulsa,
OK). Normality of continuous variables was assessed with the
Shapiro-Wilk test. Normally distributed variables are reported as mean ±
standard deviation; non-normal as median (interquartile range).
Categorical variables were compared with Fisher’s exact test for 2x2
tables and the Pearson chi-squared test for higher dimensional tables.
Association between preoperative anatomy and IFC angulation was assessed
with linear regression. Association between cannula position and
continuous outcome variables was also assessed with linear regression,
while its relationship to dichotomous outcome variables was assessed
with logistic regression. Cox proportional hazards regression was used
to assess the effect of IFC position on survival free of need for
surgical intervention. For all tests, a two-sided p-value <
0.05 was considered significant.