CARDIOVASCULAR
ORIGINAL ARTICLES: CARDIOVASCULAR
Long-Term Neurocognitive Function After
Mechanical Aortic Valve Replacement
Daniel Zimpfer, MD, Martin Czerny, MD, Philipp Schuch, MD, Richard Fakin, MD,
Christian Madl, MD, Ernst Wolner, MD, PhD, and Michael Grimm, MD
Departments of Cardiothoracic Surgery and Internal Medicine, University of Vienna, Vienna, Austria
Background. Mechanical aortic valves are a possible
source of microemboli potentially causing cerebral in-
jury. Therefore, the long-term impact of mechanical aor-
tic valve replacement on neurocognitive function is
uncertain.
Methods. In this prospective, contemporary study, we
followed 32 consecutive patients (aged 51 8 years;
range, 38 to 70; EuroSCORE [European System for Car-
diac Operative Risk Evaluation] 4.4 1.7) undergoing
isolated aortic valve replacement with a mechanical pros-
thesis. A cohort of age- and sex-matched patients (n 28,
aged 50 7 years ) served as nonsurgical controls. After
aortic valve replacement, neurocognitive function was
serially reevaluated at 7-day (n 32), 4-month (n 31),
and 3-year (n 29) follow-up. Neurocognitive function
was measured by means of P300 auditory evoked
potentials.
Results. Before the operation, P300 peak latencies were
comparable between surgical patients (361 32 ms) and
nonsurgical controls (365 33 ms, p 0.783). In patients
P ostoperative and long-term neurocognitive deficit
after open-heart surgery with cardiopulmonary by-
pass has turned out as an adverse event in the past,
possibly limiting the merits of surgery [1]. Postoperative
as well as long-term neurocognitive deficit after coronary
artery bypass grafting is well documented throughout
the literature. In contrast, the long-term development of
neurocognitive function after aortic valve replacement is
uncertain.
Neurocognitive deficit after coronary artery bypass
grafting is believed to mainly depend on intraoperative
damage. In contrast, long-term neurocognitive deficit
after aortic valve replacement might turn out to be the
result of a combined process of intraoperative damage
and cumulative damage caused by microemboli originat-
ing from prosthetic cardiac valves.
The aim of the present paper was to objectively mea-
sure long-term neurocognitive function in patients after
mechanical aortic valve replacement and to compare the
findings with nonsurgical controls.
Accepted for publication June 10, 2005.
Address correspondence to Dr Zimpfer, Department of Cardiothoracic
Surgery, University of Vienna, Wahringer Guertel 18-20, Vienna A-1090,
Austria; e-mail: daniel.zimpfer@meduniwien.ac.at.
2006 by The Society of Thoracic Surgeons
Published by Elsevier Inc
undergoing aortic valve replacement, P300 peak latencies
were prolonged 7 days after surgery (380 32 ms) as
compared with before the operation (361 32 ms, p <
0.0001) and as compared with nonsurgical controls (364
34 ms, p 0.002). At 4-month (369 30 ms, p 0.752) and
3-year (370 31 ms, p 0.825) follow-up, P300 peak
latencies normalized as compared with before operation
and as compared with nonsurgical controls (4-month
follow-up 363 31 ms, p 0.832; 3-year follow-up 366
32 ms, p 0.432). We found no difference in patients with
different valve types.
Conclusions. Despite previous assumptions based on
the potential occurence of microemboli in patients with
mechanical valves, mechanical aortic valve replacement
has no adverse long-term impact on neurocognitive func-
tion. This finding is only valid for patients with a
comparable age range undergoing isolated aortic valve
replacement.
(Ann Thorac Surg 2006;81:29 33)
2006 by The Society of Thoracic Surgeons
Material and Methods
Patients
After approval was obtained by the Ethics Committee of
the University of Vienna, 32 patients who underwent
elective mechanical aortic valve replacement at our de-
partment between January and May 2000 gave their
written informed consent and were enrolled in this
prospective study. Exclusion criteria were a history of one
of the following medical conditions: (1) prior stroke with
residual deficit, (2) uncontrolled hypertension, (3) carotid
artery stenosis of 75% or greater, (4) psychiatric illness
requiring treatment, (5) alcoholism, (6) renal disease
(defined as a creatinine more than 2.0 mg/dL [177 mol/
L]), and (7) active liver disease.
Nonsurgical Controls
For nonsurgical controls, we screened patients admitted
to the department of internal medicine for routine med-
ical checkup. The same exclusion criteria used in patients
undergoing aortic valve replacement were applied on
control subjects. Patients were contacted by the study
coordinator. Patients were informed by the study coordi-
nator about the planned tests as well as the frequency of
reexamination. All patients serving as controls had to
give their written and informed consent. Tests were not
0003-4975/06/$32.00
doi:10.1016/j.athoracsur.2005.06.039
 30 ZIMPFER ET AL Ann Thorac Surg
CARDIOVASCULAR

NEUROCOGNITIVE FUNCTION AFTER AVR 2006;81:29 33

Table 1. Patient Characteristics (20%) target-tones of 2,000 Hz at 75 dB HL. Filter band-

pass was 0.01 to 30 Hz. Active electrodes were placed at
p
Variable AVR Controls Value Cz (vertex) and Fz (frontal), respectively, and referenced
to linked earlobe A1/2 electrodes (10/20 international
Number 32 28 system) [11]. During the paradigm, the subjects were
Age (mean SD) 51.2 8.1 50.5 7.2 0.864 instructed to keep a running mental count of the rare
Age range (years) 38 70 39 69 2,000 Hz target tones. To verify attention, P300 recordings
Sex (m, %) 68.2 67.8 0.769 with a discrepancy of greater than 10% between the
Ethnicity (white, %) 100 100 1.000 actual number of stimuli and the number counted by the
Married (%) 69.5 67.7 0.564 subjects were rejected and repeated. The P300 evoked
Mean education level (years) 12.4 12.6 0.787 potential recording resulted in a stable sequence of
Smoking (%) 30.2 33.6 0.436 positive and negative peaks. Latencies (ms) of the cogni-
Diabetes (%) 23.1 26.1 0.321 tive P300 peak were assessed. To confirm reproducibility,
Hypertension (%) 55.4 51.2 0.442 two sets of P300 measurements were recorded in all
Previous cerebrovascular 0 0 1.000 patients.
accident (%)
Ejection fraction (%, mean SD) 58 1 Follow-Up
EuroSCORE (points, mean SD) 4.4 1.7 In addition to neurocognitive testing, patients were stud-
ied by means of echocardiography, electrocardiography,
AVR aortic valve replacement; EuroSCORE European System for blood tests, and clinical investigations at all points of
Operative Risk Evaluation.
follow-up. Echocardiography was used to assess func-
tional state of heart valves. Persistent atrial fibrillation
performed as part of another study. Of the patients was defined as presence of atrial fibrillation at baseline
contacted, 28 gave their written and informed consent and 3-year follow-up.
and were enrolled.
Anesthesia and Surgical Procedure
Neurocognitive Testing Patients were premedicated with midazolam. Addition-
Neurocognitive testing and physical examinations were ally midazolam in 1-mg increments was administered
completed preoperatively, 7 days, 4 months, and 3 years intravenously as needed for general anesthesia with
after surgery. All examinations were performed individ- midazolam, ethmidate, fentanyl, and pancuronium. Pa-
ually by the same experienced investigator. Neurocogni- tients were ventilated with oxygen in air; ventilation was
tive testing consisted of cognitive P300 evoked potentials. set to a tidal volume of 8 mL/kg and a respiratory rate of
To avoid any influences due to biorhythm, all investiga- 12 breaths per minute, positive end-expiratory pressure
tions were performed in the afternoon under comparable 5. The transesophageal echocardiography (TEE) probe
conditions. Special care was taken to ensure that patients was placed after anesthetic induction in all patients. The
were free from narcotics and sedatives (in the perioper- TEE views used to assess regional wall motion abnormal-
ative tests) for at least 2 days before testing. ities included the transesophageal four- and two-
chamber views and the transgastric short- and long-axis
Cognitive P300 Evoked Potentials views.
Cognitive P300 evoked potentials have previously been Surgical access was gained through a median sternot-
used to measure neurocognitive function in various met- omy. All patients underwent mildly hypothermic cardio-
abolic disorders, patients undergoing heart transplanta- pulmonary bypass (CPB [35C]) with intermittent cold
tion, and patients undergoing open-heart surgery [2 6]. blood cardioplegia with a hot shot before opening the
Cognitive P300 evoked potential are a valid marker of cross clamp. The CPB circuit consisted of a hollow-fiber
central nervous system activity [22, 23]. Cognitive P300 oxygenator (Bard HF 5701; CR Bard, Havorhill, Massa-
auditory evoked potentials are the result of an activation chusetts) and a lining system primed with Ringer lactate,
of a widespread network of cortical structures, including mannitol, heparin, and apoprotein. Flow during CPB was
association areas in the parietal, temporal, and prefrontal maintained at 2.5 L min1 m2. Blood cardioplegia was
cortex, as well as the hippocampus [7]. As a result of the maintained at 4:1 ratio. Hematocrit was kept above 20%
involvement of these brain regions in the P300 genera- with packed red blood cells if necessary. Perfusion pres-
tion, P300 can be used as a general indicator for neuro-
cognitive function [8 10]. Cognitive P300 auditory
evoked potentials have been shown to be abnormal in Table 2. Aortic Valve Types
patients with magnetic resonance imagingproven cere-
bral lesions after cardiac surgery. Cognitive P300 evoked Mechanical Aortic Valve Prostheses Number
potentials were recorded with Ag/AgCl electrodes on a Carbomedics mechanical aortic valve 11
Nicolet 2000 (Nicolet, Madison, Wisconsin). The P300 Medtronic Hall mechanical aortic valve 5
evoked potentials were generated after a binaurally pre- Medtronic Mosaic mechanical aortic valve 6
sented tone discrimination paradigm (odd-ball para-
On-X mechanical aortic valve 10
digm) with frequent (80%) tones of 1,000 Hz and rare
Ann Thorac Surg
2006;81:29 33
Table 3. Operative Data and Clinical Outcome
Variable Mechanical Valve
Operative data
Operation time (min) 208 43
Cardiopulmonary bypass 101 30
Cross-clamp time (min) 72 29
Adverse events
Death (n) 0
Myocardial infarction (n) 0
Stroke (n) 0 Fig 2. Graph showing serial assessments of cognitive brain function
Adult respiratory distress syndrome 0 by P300 auditory evoked potentials in patients with mechanical aor-
Bleeding (n) 0 tic valves: Medtronic Hall mechanical aortic valve (black line), Car-
Sternum revision 1 bomedics mechanical aortic valve (dashed line), Edwards Mira me-
Persistent artrial fibrillation (%) 20
sure during CPB was kept above 50 mm Hg with phen-
ylephrine if necessary. Before opening of cross-clamp as
well as weaning from cardiopulmonary bypass, careful
deairing was performed through the apex of the heart
and the ascending aorta under continuous inflation of the
lungs, which was vigorously controlled by TEE monitor-
ing. Heparin was antagonized with protamin sulfate until
preoperative activated clotting time was achieved. Mean
arterial pressure after CPB was kept above 60 mm Hg
with volume and vasoactive drugs as appropriate. Inten-
sive care unit treatment was performed according to
institutional standards.
Anticoagulation Therapy
Anticoagulation therapy was perioperative 2 7,500 IE
daily low molecular weight heparin dalteparin-natrium
(Fragmin; Pharmacia & Upjohn GmbH, Vienna, Austria);
on day 5 start with phenoprocoumon (Marcumar; Roche
Austria GmbH, Vienna, Austria) life long (targeted inter-
national normalized ratio [INR] range: 2.5 to 3.5; targeted
INR: 3.0). No change in anticoagulation regime in pa-
tients with atrial fibrillation. The INR values were regu-
larly monitored by the patients general practitionerall
Fig 1. Serial assessments of cognitive brain function by cognitive
P300 evoked potentials. The black line represents patients undergo-
ing aortic valve replacement; the gray line represents age- and sex-
matched control subjects. *p 0.001 compared with preoperative
values. p 0.002 between the two groups. (FUP follow-up;
pre-OP preoperative.)
ZIMPFER ET AL 31
CARDIOVASCULAR
NEUROCOGNITIVE FUNCTION AFTER AVR
chanical aortic valve (dotted line), and On-X mechanical aortic
valve (gray line). (FUP follow-up; pre-OP preoperative.)
patients were within the therapeutic range throughout
the study period.
Statistical Analysis
Data are reported as mean SD. The time course of P300
auditory evoked potentials was analyzed by means of
two-way analysis of variance (ANOVA). Comparison of
P300 evoked potentials was performed using ANOVA
after testing for normality of distribution. All p values of
serial measurements were corrected (Bonfferoni-Holm).
Categorical variables were compared using the 2 test or
Fishers exact test as appropriate. All p values less than
0.05 were considered as significant, two sided. A power
analysis was performed before conducting the study. The
power analysis was based a power of 0.85 and an alpha of
0.05. The study was analyzed using SPSS, version 12.0
(SPSS, Chicago, Illinois).
Results
Thirty-two patients undergoing isolated aortic valve re-
placement at our institution were prospectively ob-
served. The baseline characteristics of patients as well as
controls are given in Table 1. Patients and controls were
comparable with regard to demographic variables. De-
tailed information about the mechanical valves used is
given in Table 2.
Clinical Outcome
We observed no death in the present study. Operative
data and clinical outcome are given in Table 3.
P300 Auditory Evoked Potentials
Before the operation, P300 peak latencies were compara-
ble between surgical patients (361 32 ms) and nonsur-
gical controls (365 33 ms, p 0.783). In patients
undergoing mechanical aortic valve replacement, P300
peak latencies were prolonged 7 days after surgery (380
32 ms) as compared with before the operation (361 32
ms, p 0.0001) and as compared with nonsurgical
controls (364 34 ms, p 0.002; Fig 1). At 4-month (369
30 ms, p 0.752) and 3-year follow-up (370 31 ms, p
 32 ZIMPFER ET AL
CARDIOVASCULAR
NEUROCOGNITIVE FUNCTION AFTER AVR
0.825), P300 peak latencies normalized as compared
with before operation and as compared with nonsurgical
controls (4-month follow-up 363 31 ms, p 0.832;
3-year follow-up 366 32 ms, p 0.432).
Comparing neurocognitive function in patients with
different types of mechanical aortic valves we found no
difference. Time course of P300 auditory evoked poten-
tials in patients with different types of mechanical aortic
valves is given in Figure 2.
Comment
As shown by means of objective testing, mechanical
aortic valve replacement has no long-term impact on
neurocognitive function.
Neurocognitive deficit, defined as combination of def-
icits in memory, learning, concentration, and visual mo-
tor response, is an adverse event of open-heart surgery,
with an incidence of as high as 80% perhaps the most
common adverse event [3, 4, 1214]. Roach and col-
leagues [14] reported on a multi-institutional prospective
study that that neurocognitive deficit is associated with
increased mortality (10%), a twofold increase in hospital
length of stay, and a sixfold likelihood of discharge to a
nursing home. These are associated with a tremendously
increased use of health care resources. From the patient;s
view, the impact of neurocognitive deficit is devastating,
as it has been shown to reduce subjective working
capacity, quality of life, job-related abilities, and produc-
tive working status, and to impair car-driving abilities
[15, 16]. Summarizing, neurocognitive deficit is a draw-
back of open-heart surgery, as it may reduce the merits of
surgical intervention. Considering the high number of
mechanical valves implanted worldwide each year and
that mainly younger patients receive mechanical cardiac
valves, neurocognitive impairment caused by mechanical
cardiac valves has important clinical and economic
implications.
By means of P300 auditory evoked potentials and in
comparison with age- and sex- matched patients, we
have shown that postoperative neurocognitive deficit
(7-day follow-up) is reversible in patients undergoing
mechanical aortic valve replacement. We have previ-
ously reported and discussed these findings in a study
comparing neurocognitive function in patients undergo-
ing mechanical and biological aortic valve replacement
[17]. Furthermore, we have shown that mechanical aortic
valve replacement has no long-term impact on neurocog-
nitive function. Mechanical heart valves have in the past
been shown to be the source of microeboli, detected as
microembolic signals, entering the cerebral blood circuit.
However, the clinical relevance of microemboli is a
matter of discussion. Results of previous studies address-
ing neurocognitive function in patients with mechanical
aortic valves are contradictive [18 20]. Studies address-
ing neurocognitive function in patients with mechanical
valves in the past concentrated on establishing a corre-
lation between number of microembolic signals and
neurocognitive function. The main limitation of these
studies is that patients were examined at single occasions
Ann Thorac Surg
2006;81:29 33
only, no follow-up data were provided, and no control
groups were included [18 20]. In contrast to previous
studies, we provide long-term data and include a control
group. Our findings significantly question the clinical
relevance of circulating microemboli, as we found no
long-term neurocognitive injury in patients with me-
chanical valve replacement.
Neurocognitive function was measured by means of
P300 auditory evoked potentials. In healthy persons, P300
peak latencies are increased with age; and P300 auditory
evoked potentials have been used by us and others to
detect neurocognitive disorders after cardiac surgery and
have been shown to correlate with magneatic resonance
imagingproven cerebral lesions after cardiac surgery
[35, 17, 24].The clinical relevance of cognitive P300
evoked potentials is based on their being shown to be
related to cognitive impairment rating, rapid evaluation
of cognitive function tests, orientation, stimulus evalua-
tion, selective attention, visual pattern recognition, and
digit span [2, 3,]. Therefore, P300 evoked potentials are a
valid marker of cognitive function [22, 23] We used P300
auditory evoked potentials for several reasons in the
present study. In the past, P300 auditory evoked poten-
tials have been shown to be much more sensitive and
accurate in detecting neurocognitive deficit than psycho-
metric tests or electroencephalograms [21]. The P300
technique lacks several limitations of psychometric test
batteries. Psychometric test batteries are affected by
biases such as long performance times (stressing atten-
tion), visual impairment, and influence of psychomotor
function as well as level of education and learning effects
[25, 26]. Moreover, the P300 technique has a very low
intraindiviual test-retest variability with a coefficient of
variation of below 2%, which further stresses its useful-
ness for cognitive follow-up studies [3]. All P300 record-
ings were taken repeatedly (double tracing) to confirm
reproducibility of measurements. The high standard de-
viations of mean P300 peak latencies in patients and age-
and sex-matched control subjects are the result of age
dependency of cognitive P300 measurements.
Limitations
The present study is limited in that we performed no
transcranial doppler measurements. The reason for this
is the inability of currently available transcranial doppler
systems to differentiate between the size and the nature
(particular and gaseous) of emboli. It seems plausible
that more severe damage is caused by particular emboli.
Therefore, data obtained by the currently available trans-
cranial doppler systems might be misleading. Further-
more, we did not perform magnetic resonance imaging
studies. The present data are valid only for elective-
surgery patients with a comparable age range undergo-
ing aortic valve replacement with mildly hypothermic
cardiopulmonary bypass and can not be extrapolated to
patients in different age ranges.
In summary, despite previous assumptions based on
the presence of microemboli in patients with mechanical
valves, mechanical aortic valve replacement has no long-
term impact on neurocognitive function.
Ann Thorac Surg
2006;81:29 33
We thank Daniela Dunkler, MS (Stat), for the statistical analysis
of the work.
References
1. Murkin JM, Baird DL, Martzke JS. Long-term neurological
and neuropsychological outcome 3 years after coronary
artery bypass surgery. Anesth Analg 1996;82:328.
2. De Feo P, Gallai V, Mazzotta G, et al. Modest decrements in
plasma glucose concentration cause early impairment in
cognitive function and later activation of glucose counter-
regulation in the absence of hypoglycemic symptoms in
normal man. J Clin Invest 1988;82:436 44.
3. Grimm M, Yeganehfar W, Laufer G, et al. Cyclosporine may
affect improvement of cognitive brain function after success-
ful cardiac transplantation. Circulation 1996;15:1339 45.
4. Zimpfer D, Czerny M, Kilo J, et al. Cognitive deficit after
aortic valve replacement. Ann Thorac Surg 2002;74:40712.
5. Engelhardt W, Dierks T, Pause M, Sold M, Hartung E, Silber
R. P300-mappinga neurophysiological tool to quantify
cerebral dysfunction after coronary artery bypass grafting.
Eur J Cardiothorac Surg 1995;9:127.
6. Kiehl KA, Laurens KR, Duty TL. Neuronal sources involved
in auditory target detection and novelty processing: an event
related fMRI study. Psychophysiology 2001;38:133 42.
7. Picton TW. The P300 wave of the human event-related
potential. J Clin Neurophysiol 1992;9:456 79.
8. Polich J, Ehlers CL, Otis S. P300 latency reflects the degree of
cognitive decline in dementing illness. Electroencephalogr
Clin Neurophysiol 1986;63:138 44.
9. Pozessere G, Valle E, de Crignis S. Abnormalities of cogni-
tive functions in IDDM revealed by P300 event-related
potential analysis. Comparison with short-latency evoked
potentials and psychometric tests. Diabetes 1991;40:952 8.
10. Polich J. P300 clinical utility and control of variability. J Clin
Neurophysiol 1998;15:14 33.
11. Jasper HH. The ten/twenty electrode system of the Interna-
tional Federation. Electroencephalogr Clin Neurophysiol
1958;10:3715.
12. Newman MF, Kirchner JL, Phillips-Bute B, et al. Longitudi-
nal assessment of neurocognitive function after coronary-
artery bypass surgery. N Engl J Med 2001;344:395 402.
13. Sotaniemi KA, Mononen H, Hokkanen TE. Long-term cere-
bral outcome after open-heart surgery: a five-year neuropsy-
chological follow-up study. Stroke 1986;17:410 6.
ZIMPFER ET AL 33
CARDIOVASCULAR
NEUROCOGNITIVE FUNCTION AFTER AVR
14. Roach GW, Kanchuger M, Mangano CM, et al. Adverse
cerebral outcome after coronary bypass surgery. Multicenter
Study of Perioperative Ischemia Research Group and the
Ischemia Research and Education Foundation Investigators.
N Engl J Med 1996;335:1857 63.
15. Newman MF, Grocott HP, Mathew JP, et al. for the Neuro-
logic Outcome Research Group and the Cardiothoracic An-
esthesia Research Endeavors (CARE) Investigators of the
Duke Heart Center. Report of the substudy assessing the
impact of neurocognitive function on quality of life 5 years
after cardiac surgery. Stroke 2001;32:2874 81.
16. Ahlgren E, Lundqvist A, Nordlund A. Neurocognitive im-
pairment and driving performance after coronary artery
bypass surgery. Eur J Cardiothorac Surg 2003;23:334 40.
17. Zimpfer D, Kilo J, Czerny M, et al. Neurocognitive deficit
following aortic valve replacement with biological/
mechanical prosthesis. Eur J Cardiothorac Surg 2003;23:544
51.
18. Notzold A, Droste DW, Hagedorn G, et al. Circulating
microemboli in patients after aortic valve replacement with
pulmonary autografts and mechanical valve prostheses. Cir-
culation 1997;96:1843 6.
19. Deklunder G, Roussel M, Lecroart JL, Prat A, Gautier C.
Microemboli in cerebral circulation and alteration of cogni-
tive abilities in patients with mechanical prosthetic heart
valves. Stroke 1998;29:1821 6.
20. Deklunder G, Prat A, Lecroart JL, Roussel M, Dauzat M. Can
cerebrovascular microemboli induce cognitive impairment
in patients with prosthetic heart valves? Eur J Ultrasound
1998;7:4751.
21. Pozessere G, Valle E, de Crignis S. Abnormalities of cogni-
tive functions in IDDM revealed by P300 event-related
potential analysis. Comparison with short-latency evoked
potentials and psychometric tests. Diabetes 1991;8:952 8.
22. Polich J. Meta-analysis of P300 normative aging studies.
Psychophysiology 1996;33:334 5.
23. Polich J, Herbst K. P300 as a clinical assay: rationale, evalu-
ation and findings. Int J Psychophysiol 2000;38:319.
24. Toner I, Hamid SK, Peden CJ, Taylor KM, Smith PL. Mag-
netic resonance imaging and P300 (event-related auditory
evoked potentials) in the assessment of postoperative cere-
bral injury following coronary artery bypass graft surgery.
Perfusion 1993;8:3219.
25. Rossini ED, Karl MA. The Trail Making Test A and B: a
technical note on structural nonequivalence. Percept Mot
Skills 1994;78:625 6.
26. Kempen JH, Krichevsky M, Feldman ST. Effect of visual
impairment on neuropsychological test performance. J Clin
Exp Neuropsych 1994;16:22331.