Aprotinin and Tranexamic Acid for High
Transfusion Risk Cardiac Surgery
Bill I. Wong, MD, Richard F. McLean, MD, Stephen E. Fremes, MD,
Katherine A. Deemar, CPC, Ellen M. Harrington, BA, George T. Christakis, MD,
and Bernard S. Goldman, MD
Department of Anesthesia, Division of Cardiovascular Surgery, and Department of Clinical Perfusion, Sunnybrook and Women’s
College Health Science Center, University of Toronto, Toronto, Ontario, Canada
Background. Studies have shown that aprotinin and
tranexamic acid can reduce postoperative blood loss after
cardiac operation. However, which drug is more effica-
cious in a higher risk surgical group of patients, has yet
to be defined in a randomized study.
Methods. With informed consent, 80 patients undergo-
ing elective high transfusion risk cardiac procedures
(repeat sternotomy, multiple valve, combined proce-
dures, or aortic arch operation) were randomized in a
double-blind fashion, to receive either high dose aproti-
nin or tranexamic acid. Patient and operative character-
istics, chest tube drainage and transfusion requirements
were recorded.
Results. There was no significant difference between
the 2 treatment groups with respect to age, cardiopulmo-
nary bypass time, complications (myocardial infarction,
stroke, death), chest tube drainage (6, 12, or 24 hours),
C ardiovascular surgery is associated with a significant
consumption of allogeneic blood products, often as
a result of acquired hemostatic defects and/or incomplete
surgical hemostasis. Management of the abnormal bleed-
ing exposes the patient to the morbidity of reoperation
and/or excessive, and sometimes inappropriate, blood
product transfusions. The resulting increased risk of
transfusion-related complications, especially the trans-
mission of the hepatitis C virus and human immunode-
ficiency virus, has caused a renewed interest in dimin-
ishing the need for allogeneic blood transfusions by the
pharmacological reduction of bleeding.
The nonendothelial surfaces of the cardiopulmonary
bypass (CPB) circuit cause contact activation of the coag-
ulation cascades and of platelets, despite the use of
heparin. Thus, CPB is associated with several alterations
in the hemostatic mechanisms including: (1) increased
fibrinolysis; (2) decreased platelet numbers and function;
(3) dilution of clotting factors; and (4) the residual effects
of heparin or excess protamine [1]. All patients will lose
some blood during and after CPB, and many patients will
not require blood transfusions with presently employed
Accepted for publication Aug 31, 1999.
Address reprint requests to Dr Wong, M3–200, Department of Anesthesia,
Sunnybrook and Women’s College HSC, 2075 Bayview Ave, Toronto, ON,
Canada M4N 3M5; e-mail: bill.wong@utoronto.ca.
© 2000 by The Society of Thoracic Surgeons
Published by Elsevier Science Inc
Downloaded from ats.ctsnetjournals.org by on June 8, 2013
blood transfusions up to 24 hours postoperatively, total
allogeneic blood transfusions for entire hospital stay, or
induction/postoperative hemoglobin levels. However,
multiple regression analysis revealed a positive relation-
ship between cardiopulmonary bypass time and 24 hour
blood loss in the tranexamic acid group (p 5 0.001),
unlike the aprotinin group where 24 hour blood loss is
independent of cardiopulmonary bypass time (p 5 0.423).
Conclusions. Overall, there was no significant differ-
ence in blood loss, or transfusion requirements, when
patients received either aprotinin or tranexamic acid for
high transfusion risk cardiac operation. Aprotinin, when
given as an infusion in a high-dose regimen, was able to
negate the usual positive effect of cardiopulmonary by-
pass time on chest tube blood loss.
(Ann Thorac Surg 2000;69:808 –16)
© 2000 by The Society of Thoracic Surgeons
blood conservation techniques. However, some patients
or operations are at an increased risk for allogeneic
transfusions, because of excessive bleeding periopera-
tively. The risk factors include: (1) repeat cardiac opera-
tion; (2) complex procedures, such as multiple valve
replacements or aortic arch repairs; and (3) procedures
requiring long CPB times, such as combined procedures
(valve replacement plus myocardial revascularization)
[2]. These patients will benefit the most from the use of
pharmacological agents to reduce bleeding.
Pharmacological agents to reduce bleeding have re-
cently gained much interest, since they are readily avail-
able, easy to administer, can be used prophylactically, do
not require the use of costly equipment, and appear to be
very safe and efficacious. The perioperative use of tran-
examic acid, «-aminocaproic acid, and aprotinin have
gained acceptance around the world for the prophylactic
reduction of allogeneic blood transfusions in cardiac
operation patients. Tranexamic acid and «-aminocaproic
acid are synthetic lysine analogue antifibrinolytics. Tran-
examic acid was chosen for this study because it has been
more extensively studied in the cardiac operation popu-
lation. It competitively inhibits the activation of plasmin-
ogen to plasmin, and is also a weak noncompetitive
inhibitor of plasmin, blocking its action on fibrin. Thus,
tranexamic acid is thought to act by preventing the
0003-4975/00/$20.00
PII S0003-4975(99)01419-8
Ann Thorac Surg
2000;69:808 –16
premature dissolution of the normal fibrin clot. It is also
possible that tranexamic acid may help to preserve plate-
let function, by reducing the effect of plasmin on platelet
membrane receptors [3]. Aprotinin forms reversible
dose-dependent enzyme-inhibitor complexes of proteo-
lytic enzymes, including human trypsin, plasmin, and
kallikrein. Thus, it also has an antiplasmin effect, like the
antifibrinolytics, inhibiting fibrinolysis and preserving
platelet function. However, there may be other beneficial
effects such as an antikallikrein effect, which may inhibit
the contact phase of coagulation, diminishing the gener-
ation of thrombin [3]. Because of its varying actions on
different enzyme systems, its exact mechanism(s) of ac-
tion are still not entirely clear.
Both aprotinin and tranexamic acid appear to be effi-
cacious in reducing blood loss after cardiac operations
[4]. Aprotinin is more extensively reported in the litera-
ture for all patient subsets. It appears more efficacious in
attaining the goal of reducing allogeneic transfusion
requirements in the population at risk, while the efficacy
of tranexamic acid in high and low-risk populations has
more recently been confirmed in the literature [5]. For-
tunately, the incidence of complications, such as graft
thrombosis leading to myocardial infarction, stroke, renal
dysfunction, and allergic reactions is low [6]. There is
evidence that the hypercoagulable effects of aprotinin or
tranexamic acid may be related to patient risk factors,
inadequate anticoagulation, protamine reversal syn-
drome, and the use of these drugs in the postoperative
time frame. Cost is another factor to consider; aprotinin is
over three times more expensive than tranexamic acid
(CAN$930.00 vs CAN$277.50 per patient). This study will
try to determine which drug is the most efficacious. If
tranexamic acid can reduce bleeding to the same extent
as aprotinin, then it would it would make economic sense
to use the less costly drug therapy.
Patients and Methods
With the approval of the Sunnybrook HSC Research
Ethics Board (1993), 80 patients undergoing elective, high
transfusion risk cardiac procedures were studied after
giving informed consent. Preliminary sample size (esti-
mates based on retrospective data) suggested that 30
patients per group would have enough power (1 2 b .
0.80, a 5 0.05 two-sided test) to detect clinically signifi-
cant endpoints (200 mL difference in postoperative chest
tube blood loss or the transfusion of 2 units of packed red
blood cells). A total sample of 80 patients was chosen, to
ensure enough patients complete the protocol. The pro-
cedures included repeat cardiac operations (aortocoro-
nary bypass [ACB] or valvular operations), combined
procedures (valvular operation plus ACB), and other
complex procedures (multiple valve replacement, as-
cending aortic graft). Subjects were excluded if they had
recent (, 5 days) acetylsalicylic acid ingestion, thrombo-
lytic therapy (streptokinase, urokinase, or tissue plasmin-
ogen activator , 1 day) or anticoagulant therapy (hepa-
rin , 4 hours preoperative or warfarin , 3 days
preoperatively). As well, subjects with preexisting coag-
Downloaded from ats.ctsnetjournals.org by on June 8, 2013
WONG ET AL 809
APROTININ AND TRANEXAMIC ACID
ulation defects (including abnormal preoperative coagu-
logram [prothrombin time (PT) . 18 seconds or partial
prothrombin time (PTT) . 50 seconds] or platelet
count , 100 3 109/L), preexisting renal dysfunction
(serum creatinine . 200 mmol/L), or had autologous
predonation of blood, were excluded from the study.
Consenting subjects were randomized (double-blinded)
to receive either high-dose aprotinin (AP group) or
high-dose tranexamic acid (TA group).
Drug Administration
The randomization and preparation of the study drugs
was performed by the hospital’s department of phar-
macy. There was no attempt to stratify the randomization
process. The high-dose aprotinin regimen [7] included
intravenous infusion of 2 3 106 kallikrein inhibitory units
(KIU) of aprotinin-Trasylol ([Bayer AG; Leverkusen, Ger-
many], 10,000 KIU/mL of pure aprotinin in a preserva-
tive-free isotonic solution) infused over 20 minutes, after
induction of anesthesia and before skin incision. Subse-
quently, 5 3 105 KIU/h of aprotinin were administered
continuously throughout the operation, until skin clo-
sure. Additionally, aprotinin (2 3 106 KIU) was added to
the priming solution of the CPB circuit. Tranexamic acid
was administered as an intravenous bolus dose of 10 g
tranexamic acid-Cyklokapron ([Pharmacia & UpJohn Inc,
Missisgauga, Ontario, Canada], 0.1 g/mL tranexamic
acid) over 20 minutes, after induction of anesthesia and
before skin incision [8]. No further tranexamic acid was
used, but placebos consisting of 0.9% normal saline
solution were used for the continuous infusion, and the
addition to the CPB prime solution to mimic aprotinin
administration. A test dose of the study solution (1 mL)
was given to the patient to help detect any allergic
responses before the initial loading dose was given. The
prime dose was not added to the CPB prime solution
until after the patient safely received the initial loading
dose of the study drug.
Anesthetic, Surgical, and CPB Management
The anesthetic protocol consisted of intravenous fentanyl
citrate, midazolam, and pancuronium bromide. Follow-
ing tracheal intubation, anesthesia was maintained with
halothane or isoflurane. The subjects were ventilated to
normocapnia with a 50% oxygen/air mixture. Conduct of
the operation was carried out in accordance with the
usual protocols of the participating surgeons. Exposure
was provided by a median sternotomy, and extracorpo-
real circulation lines consisted of aortic arch and atrial
cannulas. A Cobe heart-lung machine (Cobe Cardiovas-
cular Inc, Arvada, CO) with Bentley (Baxter Healthcare
Corp, Bentley Laboratories Division, Irvine, CA) tubing
and reservoirs were used for CPB, including membrane
oxygenators (Maxima Hollow Fiber Oxygenator;
Medtronic Blood Systems, Inc, Anaheim, CA) and in-line
arterial line filters (Bard H-645 with Biothyl; C. R. Bard,
Inc, Murray Hill, NJ). Circuit prime consisted of 1500 mL
lactated Ringer’s solution, 100 mL 25% albumin, 25 g
mannitol, 50 mEq sodium bicarbonate, and 5000 units
heparin sodium. Priming solution was precirculated
810 WONG ET AL
APROTININ AND TRANEXAMIC ACID
through a 5 mm prebypass filter (Bentley) before cannu-
lation. CPB was employed with core temperatures of 27°
C to 33° C. Nonpulsatile flows of 2.4 L/min/m2 were used
at normothermia, and were no lower than 1.8 L/min/m2
at 28° C. Mean arterial pressure was maintained between
50 and 80 mm Hg. Oxygen inflow of 2 to 5 L/min was
adjusted for normal oxygenation and alpha-stat acid-
base balance.
The study did not intend for the inclusion of cases with
deep hypothermic circulatory arrest (DHCA). The proce-
dures did include repairs of the ascending aorta, with or
without aortic valve replacement. These cases generally
do not employ DHCA at our institution. However, one
mitral valve replacement (MVR) with ACB procedure
required DHCA, because of an aortic tear secondary to
aortic cannulation. This patient (TA group) required
DHCA to urgently replace the ascending aorta with a
graft. Circulatory arrest lasted 8 minutes at a temperature
of 26 °C. This patient was classified as an aortic operation
case. Another aortic valve replacement-ascending aortic
graft (AP group) required a short (36 second) period of
circulatory arrest at 20° C. The other 4 cases, involving
the ascending aorta, were performed under moderate
hypothermic CPB without circulatory arrest.
Patients were anticoagulated pre-CPB with 300 IU/kg
heparin sodium ([Hepalean; Organon Teknika, Toronto,
Canada], 1,000 USP U/mL of heparin sodium from por-
cine intestinal mucosa in 1% benzyl alcohol as preserva-
tive in an isotonic aqueous solution), supplemented with
50 to 100 IU/kg as needed, to maintain an activated
clotting time (ACT) of greater than 400 seconds during
CPB (HemoTec kaolin activator [HemoTec, Inc, Engle-
wood, CO]; note: aprotinin is reported to prolong the
ACT, but this effect is felt to result from the use of
celite-activated assays and not kaolin-activated test
tubes) [9]. Heparin neutralization after termination of
CPB was by slow intravenous injection of protamine
sulfate ([Lyphomed Canada Inc, Markham, Canada],
10 mg/mL protamine sulfate in preservative free isotonic
solution) at a dose of 1 mg/100 IU of estimated active
heparin. ACT after protamine administration was moni-
tored with the goal to return to near preheparinization
levels (not greater than 10% above baseline).
Blood Transfusion Protocol
Blood conservation methods used include reinfusion of
shed cardiotomy blood during CPB, and autotransfusion
of salvaged mediastinal-shed blood for the first 6 hours
postoperatively. Blood product transfusion guidelines
were used to standardize transfusion practice. Some
latitude in the decision to give transfusion products was
allowed, since it is often not clinically practical to wait for
laboratory tests when a patient is hemorrhaging. During
the operation, packed red blood cell (PRBC) concentrates
were transfused when the hematocrit/hemoglobin (Hct/
Hb) value was less than 0.20/70 g/L. In the postoperative
intensive care unit (ICU), the threshold for PRBC trans-
fusion was a Hct/Hb less than 0.25/80. The indication for
perioperative random donor platelets, fresh frozen
plasma, or cryoprecipitate transfusion was the presence
Downloaded from ats.ctsnetjournals.org by on June 8, 2013
Ann Thorac Surg
2000;69:808 –16
of excessive active bleeding (. 200 mL/h), and a labora-
tory demonstrated coagulation defect (platelet count ,
100 3 109/L, PT or PTT . 1.5 3 control value, or
fibrinogen level , 1.0 g/L).
Postoperative Intensive Care
With the exception of protamine to reverse the action of
heparin, the nonstudy postoperative use of additional
pharmacologic agents to reduce bleeding (antifibrinolyt-
ics: aprotinin, tranexamic acid, «-aminocaproic acid; co-
agulation factor enhancers: desmopressin acetate; or
agents to preserve platelet function: prostaglandins
[PGE1 or PGI2]) were avoided for the first 24 hours in the
study. The routine immediate postoperative use of low-
dose acetylsalicylic acid (325 mg orally per day) for the
protection of ACB graft patency, and low dose intrave-
nous heparin (PTT , 40) for antithrombotic protection of
valve prostheses continued, as per usual protocol, when
chest tube bleeding had diminished. Full therapeutic
anticoagulation (heparin and/or warfarin) for mechanical
prosthetic valves was instituted after chest tube removal
at 24 to 36 hours postoperatively. In addition, the use of
low dose (6 mg/h) intravenous dipyridamole (Persantine;
Boehringer Ingelheim (Canada) Ltd, Burlington, Ontario,
Canada) to promote arterial patency after coronary end-
arterectomy was instituted during the operation when
indicated.
Data Collection
Postoperatively, blood loss from the mediastinal chest
tubes was reported at 6, 12, and 24 hours from the time
the patient arrived in the ICU. In addition, the volume of
autotransfused blood in the first 6 hours was recorded.
Blood product transfusion requirements were docu-
mented (by type and donor unit exposure) for the entire
hospital stay. Hematological and biochemical studies
were monitored in the perioperative period. Periopera-
tive events that were noted included perioperative myo-
cardial infarction (new Q-waves on electrocardiogram
and elevation of creatinine kinase MB isoenzyme/total
creatinine kinase (CK) ratio . 5%), strokes (neurologic
examination), reoperation for bleeding, and the surgical
diagnosis discovered at reoperation (surgical bleeding,
leaking anastomoses; medical bleeding, coagulopathy).
Statistics
Data was analyzed with commercially available software
(JMP, SAS Institute Inc; SPSS, SPSS Inc, Chicago, IL). A
two-sided t-test was used to test normally distributed
variables. A nonparametric test (Wilcoxon two-sample
test) was used on blood transfusion data. The frequency
of transfusion was compared with a x2 test or Fisher’s
exact test for small sample sizes. All results are expressed
as mean and standard error of the mean, unless other-
wise stated. Multiple regression analysis (backwards
elimination) was performed to determine if any patient
or surgical variables (age, drug group, CPB time, CPB
temperature, procedure, and surgeon) impacted on the
primary endpoints (blood loss, blood transfusions) of the
study.
Ann Thorac Surg
2000;69:808 –16
Table 1. Surgical Procedure
Procedure Aprotinin Tranexamic Acid
ACB 1 valve 15 22
Repeat sternotomy
Redo ACB 3 1
Redo valve 7 7
Multiple valve 10 6 surgical procedure groups (repeat sternotomy, multiple
Ascending aortic graft 4 2
Total 39 38
ACB 5 aortocoronary bypass surgery; Redo 5 repeat sternotomy;
Valve 5 valve replacement surgery.
Results
Eighty patients were enrolled and randomized to the two
treatment groups. One patient (TA group) was excluded
when the operation was cancelled, because of the inabil-
ity to safely cannulate a heavily calcified aorta. Two
patients (1 AP, 1 TA) underwent simpler procedures, and
were not given the study drug at all. Of the remaining 77
patients (39 AP, 38 TA) who entered the protocol, analysis
of the whole group, or of surgical subsets revealed no
significant differences between the two drug treatment
groups with respect to age, weight, surgical procedure,
CBP time, and CPB temperature (Tables 1, 2).
There were 4 in-hospital deaths (2 AP, 2 TA). This
included a patient (TA group) undergoing repeat MVR,
who died intraoperatively from a myocardial infarction.
This subject is included in the demographic and surgical
data, but blood loss and transfusion data were not
complete. Thus, blood loss, transfusion data, and postop-
erative complications were analyzed with 76 patients.
Another patient underwent repeat ACB (AP group) and
died from a perioperative myocardial infarction 2 days
after operation. A third patient died after a week in the
ICU, after a combined MVR and ACB (AP group). This
patient developed severe right heart failure, low output
syndrome, ventricular arrhythmias, and renal failure.
The fourth patient who also underwent a combined MVR
and ACB (TA group) had a long and complicated post-
operative course, including low output syndrome, renal
failure, and wound infection. This patient died 4 months
postoperatively from hemodialysis complications.
Only one patient (AP group), who underwent a com-
bined ACB and valve replacement procedure, required
reoperation for surgical bleeding. There was no statisti-
cally significant difference between the two treatment
groups with respect to complications (PMI, stroke,
death), 6, 12, or 24 hour chest tube blood loss, blood
transfusions up to 24 hours postoperatively, total alloge-
Table 2. Baseline Characteristics and Clinical Outcomesa
Drug N Age (y) Weight (kg) CPB Time (min)
Aprotinin 39 65.4 (1.37) 75.7 (2.81)
Tranexamic acid 38 66.0 (1.72) 78.8 (3.35)
a
Values are absolute or mean (standard error of the mean).
CPB 5 cardiopulmonary bypass; MI 5 myocardial infarction.
Downloaded from ats.ctsnetjournals.org by on June 8, 2013
WONG ET AL 811
APROTININ AND TRANEXAMIC ACID
neic blood transfusions for entire hospital stay, or induc-
tion/postoperative hemoglobin levels (Tables 2– 4).
The largest group of surgical procedures included
those undergoing combined ACB and valve replacement
operation. Post hoc analysis of this more homogenous
subgroup also revealed no significant differences in the
outcomes described above. Further subanalysis of other
valve operations) did not reveal any significant differ-
ences because of the lack of power. For instance, when
the patients who underwent repeat sternotomy (redo)
were analyzed post hoc, there appeared to be a signifi-
cant difference, favoring aprotinin, in the use of PRBC’s,
1.50 6 0.54 versus. 3.57 6 0.81 units (p 5 0.046, Wilcoxon
2-sample test), in the intraoperative plus 24 hours post-
operative period. However, we are wary of potential
imbalances in randomization caused by the small size of
the subgroup. Also, given the risks of a type 1 error from
multiplicity, a Bonferroni correction of the p-value would
not allow this result to be characterized as statistically
significant.
Multiple regression analysis also revealed that CPB
time had a significant effect on 12 and 24 hour blood loss
in patients who received tranexamic acid (p 5 0.002, p 5
0.001 respectively), but not in those who received apro-
tinin (p 5 0.441, p 5 0.423 respectively). The p value for
the drug-by-CPB time interaction for the 12 and 24 hour
blood loss was 0.033 and 0.026, respectively. Linear re-
gression analysis revealed a positive relationship be-
tween the 24 hour blood loss and CPB time in the
tranexamic acid group (p 5 0.0037, analysis of variance),
unlike the aprotinin group where 24 hour blood loss is
independent of CPB time (p 5 0.33, analysis of variance).
Further multiple regression analysis did not find any
significant interaction between blood loss/transfusions
and other variables such as age, gender, surgeon, surgical
procedure, preoperative hemoglobin or CPB temperature.
Comment
Aprotinin, tranexamic acid, and «-aminocaproic acid
[4 – 8, 10] have been shown to reduce blood loss after
cardiac operations, when compared to placebo control
groups. The use of tranexamic acid in cardiac operations
has been reported in more randomized trials than «-ami-
nocaproic acid, thus tranexamic acid was chosen as the
synthetic antifibrinolytic to compare against aprotinin.
Also, a placebo control group was not used for this study
since the efficacy of the drugs has been established in the
literature. The cost of prophylactic therapy and the in-
complete safety data for these drugs [11], has reserved
CPB Temperature (° C) MI Stroke Hospital Death
159 (6.27) 27.8 (0.49) 4 1 2
169 (8.06) 28.4 (0.51) 5 0 2
812 WONG ET AL Ann Thorac Surg
APROTININ AND TRANEXAMIC ACID 2000;69:808 –16

Table 3. Blood Lossa

6 Hour 12 Hour 24 Hour Hb in
Blood Loss Blood Loss Blood Loss Initial Hb 24 Hours
Drug N (mL) (mL) (mL) (g/dL) (g/dL) Reoperation

Aprotinin 39 302 (44.2) 429 (50.5) 682 (61.3) 11.7 (0.24) 9.0 (0.14) 1
Tranexamic acid 37 310 (40.3) 462 (53.9) 746 (71.8) 11.9 (0.24) 9.0 (0.15) 0
Mean TA—AP difference 33 (74) 64 (94)
95% CI of difference 2114,180 2123,252
a
Values are absolute or mean (standard error of the mean).
AP 5 aprotinin group; CI 5 confidence interval; Hb 5 hemoglobin; TA 5 tranexamic acid group.

the use of these drugs for high-risk procedures/patients
in our institution. The high dose regimens described for
these drugs [7, 8] were used to affect a maximum benefit,
since these patients are at a higher risk for blood loss.
The goal of this study was to determine if either aprotinin
or tranexamic acid (in high doses) was more efficacious in
reducing blood loss and allogeneic transfusion require-
ments in high transfusion risk cardiac operation patients.
In the total patient group analysis, the amount of shed
mediastinal blood compares favorably with our meta-
analysis of previously published studies [4]. The 12 hour
chest tube blood loss (mean 6 standard error of the
mean) in all patients of our study was only 443 6 36.3 mL,
compared to means of 698 to 770 mL in the placebo
groups of the meta-analysis. The aprotinin and antifi-
brinolytic (tranexamic acid or «-aminocaproic acid) treat-
ment groups in the meta-analysis did have this amount
of blood loss (446 mL, 526 mL, respectively). When one
also considers that most of the studies in the meta-
analysis were not in the high-risk complex operation
category, our results certainly appear to be in line with
the expected results with prophylactic antifibrinolytics in
cardiac operation. The average duration of CPB in this
study (162 6 5.2 minutes) is much longer that that for
routine ACBs (80 to 100 minutes), and this length of CPB
contributes to the patient’s risk of postoperative bleeding
[2]. Remarkably, only 1 patient (1 of 76, 1.3%) returned to
the operating room for postoperative hemorrhaging, sim-
ilar to the reoperation event rate revealed in the meta-
analysis (0.44% to 1.47% in the drug treatment groups).
Multiple regression analysis of chest tube blood loss
versus CPB time, reveals a positive relationship between
blood loss and CPB time in patients who received tran-
examic acid. This trend is not seen in the patients who
received aprotinin. Also, the linear regression models for
the tranexamic acid is significantly different from a hor-
izontal line at the mean (p , 0.01; analysis of variance).
The fact that this trend is only seen in patients who
received tranexamic acid suggests that the overall effec-
tiveness of the medication (when only given as a bolus at
the beginning of the procedure) may be decreased when
CPB is prolonged. Aprotinin is given as a bolus and an
infusion, thus compensating for long CPB times. Perhaps,
if the tranexamic acid were also given as an infusion, this
would help maintain its efficacy for long CPB times. Is it
possible that aprotinin given in a high dose regimen is
more protective for long CPB times than a single bolus of
tranexamic acid? In addition, patients undergoing repeat
sternotomy may have a longer period of time between
the delivery of the drug and the start of CPB, since the
opening of the chest is more tedious secondary to adhe-
sions from the previous sternotomy. Further studies of
the pharmacokinetics of these drugs will enable more
accurate dosing of the medication to allow for long CPB
times.
A power analysis was performed to determine if the
study sample had enough power to detect differences in
clinically important outcomes. For postoperative blood
loss, a difference of at least 200 to 300 mL would be more
clinically relevant than a 100 mL difference, since the
larger amount may influence decisions to give transfu-
sions or drug therapy. Also, a reduction in the use of
PRBCs in the first 24 hours after operation by two units
would certainly be clinically important. The patient’s
preoperative hemoglobin level may influence the need
for one unit of blood perioperatively. Our power calcu-

Table 4. Transfusionsa
PRBC in Platelets in FFP in Cryo in Total Units
24 Hours 24 Hours 24 Hours 24 Hours in Hospital No Bld in No Bld in Reinf
Drug N (units) (units) (units) (units) (units) 24 Hours Hospital Bld

Aprotinin 39 2.03 (0.34) 1.74 (0.55) 0.31 (0.14) 0.51 (0.37) 5.90 (1.66) 12 (30.8%) 9 (23.1%) 7
Tranexamic acid 37 2.49 (0.37) 1.81 (0.57) 0.62 (0.30) 0.86 (0.52) 7.19 (1.71) 8 (21.6%) 6 (16.2%) 4
Mean TA—AP difference 0.46 (0.50)
95% CI of difference 20.55, 1.47
a
Values are absolute or mean (standard error of the mean).
AP 5 aprotinin group; Bld 5 blood products; CI 5 confidence interval; Cryo 5 cryoprecipitate; FFP 5 fresh frozen plasma; PRBC 5
packed red blood cells; Reinf bld 5 reinfusion of chest tube blood loss; TA 5 tranexamic acid group.

Downloaded from ats.ctsnetjournals.org by on June 8, 2013

Ann Thorac Surg
2000;69:808 –16
Table 5. Power Calculation
AP
Variable (observed SD)
12 Hour blood loss (mL) 315
24 Hour blood loss (mL) 383
PRBC in 24 hours (units) 2.12
AP 5 aprotinin group; PRBC 5 packed red blood cells; SD 5 standard deviation;
lations (Table 5) demonstrate that there is sufficient
power [(1 2 b) . 0.80] to detect a 209 mL blood loss
difference at 12 hours and 267 mL blood loss difference at
24 hours in the total group (significance level of a 5 0.05,
two-sided test). Also, there is sufficient power to detect a
1.43 unit difference in the transfusion of PRBCs in the
total study. Since our study did not find any significant
differences in blood loss or transfusion needs between
the two drug groups (small deltas), the power analysis
reassures us that clinically important outcomes were not
missed by insufficient sample size.
There have been 14 cardiac operation studies [18 –31]
directly comparing the efficacy of aprotinin against a
synthetic lysine analogue antifibrinolytic (tranexamic
acid and/or «-aminocaproic acid) (Table 6). The variation
in patient population and drug dosages complicates
direct comparisons between studies. As expected, most
studies did find that aprotinin (11 out of 11), tranexamic
acid (4 out of 7), or «-aminocaproic acid (2 out of 2)
reduced postoperative blood loss when compared to
available placebo controls. Only about a third of the
studies (5 out of 14) [18, 21, 25, 26, 28] found aprotinin to
be superior to tranexamic acid or «-aminocaproic acid in
reducing postoperative blood loss. Interestingly, none of
the studies were able to demonstrate a clear superiority
of any of these drugs in reducing the average transfusion
requirements. The study by Landymore and colleagues
[25] did find that the aprotinin and «-aminocaproic acid
groups had significantly lower blood transfusions than
the tranexamic acid group. Interestingly, only 2 out of 6
studies [19, 26], which reported the data, found aprotinin
to be significantly superior to tranexamic acid or «-ami-
nocaproic acid in reducing the proportion of patients
who required transfusions.
Relatively small sample size and the low risk popula-
tion studied have limited these comparative studies.
Most of these studies utilized primary ACB or valve
operation patients and only 4 studies had repeat sternot-
omy or combined procedures (ACB and valve replace-
ment). By utilizing a high-risk group of patients to
compare the efficacy of two blood-sparing drugs (aproti-
nin and tranexamic acid), one may be able to show a
greater effect with the prophylactic drug treatment. Pa-
tients who are at risk of greater blood loss after cardiac
operation have a greater potential to benefit from pro-
phylactic pharmacologic treatment. It would appear that
aprotinin and tranexamic acid can reduce blood loss and
blood transfusions in comparison to placebo controls.
However, there are not enough data to tell if aprotinin is
Downloaded from ats.ctsnetjournals.org by on June 8, 2013
WONG ET AL 813
APROTININ AND TRANEXAMIC ACID
Detectable Delta
TA SD Used in (5% level, 80%
(observed SD) Power Analysis power, 38 per group)
327 321 209
437 410 267
2.28 2.2 1.43
TA 5 tranexamic acid group.
superior to the synthetic antifibrinolytics. Also, the blood
saving advantage of aprotinin over other antifibrinolytics
may be small, especially in low risk populations. Aproti-
nin does appear to be the most beneficial of these
blood-sparing drugs when one examines the meta-
analyses of randomized trials [4, 5]. However, direct
comparison trials suggest that the reduction of allogeneic
blood transfusions with aprotinin treatment is not much
greater that that seen with tranexamic acid or «-aminoca-
proic acid.
With such a minimal benefit and a large price differ-
ential, many centers are using the less expensive drugs
for patients at risk. The cost of aprotinin is so large ('
CAN$1000, compared to tranexamic acid ' CAN$100 to
CAN$275, and «-aminocaproic acid ' CAN$50), that
direct recovery of its cost by reducing blood transfusions
is difficult [12]. One of the problems is actually the
unnecessary blood transfusions that 27% of patients
receive [13]. A recent publication has revealed that the
cost per unit of allogeneic blood for inpatient red blood
cell transfusion is CAN$210 [14]. This includes the cost of
collection, production, distribution, and delivery. How-
ever, one must remember that the cost of bleeding
includes not only the cost of drug/transfusion therapy,
but also the materials and manpower costs of reopera-
tions, prolonged intensive care, and the treatment of
complications of large volume allogeneic blood product
transfusions. Other problems will always exist in con-
junction with blood product transfusions: a limited sup-
ply; need for cross-matching; ABO blood group compat-
ibility error; septic unit transfusion; alloimmunization;
possible immunosuppression; dilutional coagulopathy;
and a short shelf life. These costs are enormous (reported
as high as CAN$60,000) in comparison to the cost of the
drug therapy [13]. Fortunately, these complications are
infrequent, so the added costs are spread out among a
large number of patients. Thus by reducing blood trans-
fusions, one can easily realize an economical benefit,
better patient care, and reduced stress on a valuable and
chronically short-supplied blood banking system.
One cannot discuss the benefits of these hemostatic
drugs without discussing their potential risks. It would
seem logical that along with increased efficacy, there may
be increased thrombotic risk. A large multicenter, ran-
domized, placebo controlled study was commissioned by
Bayer to answer the question of early graft occlusion with
aprotinin therapy. Alderman and colleagues [15] re-
814 WONG ET AL Ann Thorac Surg
APROTININ AND TRANEXAMIC ACID 2000;69:808 –16

Table 6. Comparison Trials

Total dose Proportion
Authors Procedure Drug N (70 kg/4 h) Blood Loss Blood Transfusion Transfused

Bennett-Guerrero et al Repeat ACB/ Aprotinin 99 6 3 106 KIU AP , EACAa AP , EACAb

[18] and/or valve EACA 105 19 g
Blauhut et al [19] Primary ACB Aprotinin 14 5 3 106 KIU AP , Plaa AP , Plaa AP , Plaa
Tranexamic 15 1 g TA , Plab TA , Plab TA , Plaa
Placebo 14 AP , TAb AP , TAb AP , TAa
Boughenou et al [20] Primary valve Aprotinin 17 6 3 106 KIU TA , APb TA , APb
Tranexamic 18 2 g
Corbeau et al [21] Primary ACB Aprotinin 43 6 3 106 KIU AP , Plaa AP , Plab
or AVR Tranexamic 41 2.1 g TA , Plaa TA , Plab
Placebo 20 AP , TAa AP , TAb
Cousin et al [22] Primary or Aprotinin 20 6 3 106 KIU AP H , Plaa Pla , AP Ha
repeat ACB Aprotinin L 30 1.75 3 106 KIU AP L , Plaa Pla , AP Lb
and/or valve Tranexamic 20 2.1 g TA , Plab Pla , TAb
Placebo 20 AP , TAb TA , APb
Eberle et al [23] Primary ACB Aprotinin 20 6 3 106 KIU AP , Plaa AP . Plab AP . Plab
(nonrandomized EACA 20 40 g EACA , Plaa EACA . Plab EACA . Plab
control) Placebo 10 AP , EACAb AP , EACAb AP , EACAb
Jamieson et al [24] Repeat valve Aprotinin 24 6 3 106 KIU AP , Plaa AP , Plaa
(two trials) Placebo-AP 36
Tranexamic 22 10 mg TA , Plaa TA , Plaa
Placebo-TA 19 AP 5 TA AP , TAb
Landymore et al [25] Primary ACB Aprotinin L 48 1 3 106 KIU AP , Plaa AP , Plaa
Tranexamic 56 1 g TA , Plaa TA , Plaa
EACA 44 9 g EACA , Plaa EACA , Plaa
Placebo 50 AP , EACA , TAa AP 5 EACA , TAa
Menichetti et al [26] Primary ACB Aprotinin 24 6 3 106 KIU AP , Plaa AP , Plaa
Tranexamic 24 1.5 g TA , Plaa TA , Plaa
Placebo 24 AP , TAa AP , TAa
Mongan et al [27] Primary ACB Aprotinin 75 7 3 106 KIU AP , Plaa AP , Plaa AP , Plaa
(nonconcurrent Tranexamic 75 2 g TA , Plaa TA , Plaa TA , Plaa
control) Placebo 30 AP , TAb AP 5 TA AP 5 TA
Penta de Peppo et al Primary ACB Aprotinin 15 6 3 106 KIU AP , Plaa AP , Plab AP , Plab
[28] and/or valve Tranexamic 15 1.4 g TA , Plab TA , Plab TA , Plab
Placebo 15 AP , TAa AP , TAb AP , TAb
Pugh et al [29] Primary ACB Aprotinin 21 2 3 106 KIU AP , Plaa AP , Plaa
(single-blinded) Tranexamic 22 5 g TA , Plaa TA , Plaa
Placebo 23 AP , TAb AP , TAb
Speckenbring et al Primary ACB Aprotinin 15 6 3 106 KIU AP , Plaa AP , Plab
[30] Tranexamic 15 1 g TA , Plaa TA , Plab
Placebo 15 AP , TAb AP , TAb
Trinh-Duc et al [31] ACB, valve Aprotinin 29 6 3 106 KIU AP , EACAb AP , EACAb AP , EACAb
Aorta, ASD EACA 27 20 g
a b
p # 0.05. Not significant.
ACB 5 aortocoronary bypass surgery; AVR 5 aortic valve replacement; Aorta 5 aortic dissection repair; AP 5 aprotinin group; Aprotinin
H 5 high dose aprotinin; Aprotinin L 5 low dose aprotinin; ASD 5 atrial septal defect repair; EACA 5 e-aminocaproic acid; KIU 5
kallikrein inhibitory units; Pla 5 placebo; TA 5 tranexamic acid group; Valve 5 valve replacement surgery.

ported the results from the International Multicenter
Aprotinin Graft Patency Experience (IMAGE) trial which
involved 13 international sites and enrolled 870 patients
in 1994 to 1995. Among 703 patients with assessable
saphenous vein grafts, occlusions occurred in 15.4% of
aprotinin-treated patients and 10.9% of the placebo
group (p 5 0.03). The occlusion rate of distal saphenous
vein grafts was 7.5% (897 saphenous vein graft insertions
studied) in the aprotinin group versus 4.8% (837 studied)
in the placebo group (p 5 0.02). The conclusion by the
authors was that early vein graft occlusion was increased
by aprotinin, but this outcome was promoted by multiple
risk factors for graft occlusion (mainly small or poor
distal vessel quality).
The synthetic antifibrinolytics have not been studied as
intensely as aprotinin, but the question of graft throm-
bosis is just as pertinent. Recently, Karski and associates
[16] presented an abstract on their experience with tran-

Downloaded from ats.ctsnetjournals.org by on June 8, 2013

Ann Thorac Surg
2000;69:808 –16
examic acid. They assessed early saphenous vein graft
patency (5 to 30 days) with cine magnetic resonance
imaging techniques in 146 patients randomized to re-
ceive tranexamic acid (n 5 76) versus placebo (n 5 70).
The saphenous vein graft patency was 85.1% in the
tranexamic acid group versus 86.3% in the placebo group
(clearly no difference but a relatively high graft occlusion
rate). In terms of blood sparing effect, 14.3% of the
tranexamic acid group required red blood cell transfu-
sions versus 24.0% of the placebo group (p , 0.05).
Unfortunately, this study did not use angiography, which
is considered the gold standard for patency trials. How-
ever, at least there is no early evidence that high dose
tranexamic acid is causing harm when given prophylac-
tically (100 mg/kg). Earlier aprotinin trials with ultra-fast
computed tomographic or magnetic resonance imaging
studies also did not show any difference with graft
patency (small groups also). It took a large, multicenter,
angiography trial (IMAGE) to show that there can be an
increase in graft occlusions when patients receive
aprotinin.
As Westaby and Katsuma [17] pointed out in an edi-
torial, “occluded [coronary artery bypass grafts] are a
high price to pay for an average blood saving of 250 mL.”
In the preoperative and postoperative period, we are
obsessed with the use of platelet inhibitors and antico-
agulants to treat patients with acute coronary ischemia,
prosthetic valve implantation, and after ACB operation to
promote graft patency. Yet in the operating room, with
the onslaught of surgically induced thrombin formation,
we are using agents which may combat physiologic
fibrinolysis, a vital process to maintain vessel patency. In
addition, we are giving agents that promote platelet
adhesion and aggregation. A balance must be reached
with these two opposing goals. Routine use of these
hemostatic agents may lead to an increase in adverse
events. A more logical approach may be to reserve these
pharmacologic therapies to patients who are at high risk
for transfusions, and thus may receive the most benefit at
the lowest risk. The recent public scrutiny of Canada’s
blood system has certainly increased interest in alterna-
tives to allogeneic blood product transfusions to decrease
the risk of transfusion-born infectious diseases. How-
ever, like any other therapy, there may be significant
risks.
Our report utilized a high-risk group of patients to
compare the efficacy of two blood-sparing drugs (aproti-
nin and tranexamic acid). Patients who are at risk of
greater blood loss after cardiac operation have a greater
potential to benefit from prophylactic pharmacologic
treatment. Aprotinin and tranexamic acid had similar
efficacies in reducing blood loss and blood transfusions
for our group of high transfusion risk cardiac procedures.
The authors thank C. David Naylor, DPhil, and Kathy Sykora,
MSc for their expert advice with the statistical analyses and
critical review of the manuscript. Also, this study would not be
possible without the assistance of Dr Tom Paton, Department of
Pharmacy, Mr Ahmed Coovadia, Blood Bank, SD Laboratory
Services, Dr Peter Pinkerton, Department of Hematology, and
Downloaded from ats.ctsnetjournals.org by on June 8, 2013
WONG ET AL 815
APROTININ AND TRANEXAMIC ACID
the staff of the Cardiovascular Operating Rooms and Intensive
Care Unit of Sunnybrook & Women’s College HSC. This study
was supported by a grant from the J.P. Bickell Foundation,
National Trust Company.
References
1. Holloway DS, Summaria L, Sandesara J, Vagher JP, Alex-
ander JC, Caprini JA. Decreased platelet number and func-
tion and increased fibrinolysis contribute to postoperative
bleeding in cardiopulmonary bypass patients. Thromb Hae-
most 1988;59:62–7.
2. Despotis GJ, Filos KS, Zoys TN, Hogue CW, Spitznagel E,
Lappas DG. Factors associated with excessive postoperative
blood loss and hemostatic transfusion requirements: a mul-
tivariate analysis in cardiac surgical patients. Anesth Analg
1996;82:13–21.
3. Hardy J-F, Desroches J. Natural and synthetic antifibrinolyt-
ics in cardiac surgery. Can J Anaesth 1992;39:353– 65.
4. Fremes SE, Wong BI, Lee E, et al. Metaanalysis of prophy-
lactic drug treatment in the prevention of postoperative
bleeding. Ann Thorac Surg 1994;58:1580– 8.
5. Laupacis A, Fergusson D. Drugs to minimize perioperative
blood loss in cardiac surgery: meta-analyses using periop-
erative blood transfusion as the outcome. Anesth Analg 1997;
85:1258– 67.
6. Bidstrup BP, Harrison J, Royston D, Taylor KM, Treasure T.
Aprotinin therapy in cardiac operations: a report on use in 41
cardiac centers in the United Kingdom. Ann Thorac Surg
1993;55:971– 6.
7. Bidstrup BP, Royston D, Sapsford RN, Taylor KM, Cosgrove
DM. Reduction in blood loss and blood use after cardiopul-
monary bypass with high dose aprotinin. J Thorac Cardio-
vasc Surg 1989;97:364–72.
8. Karski JM, Teasdale SJ, Norman P, et al. Prevention of
bleeding after cardiopulmonary bypass with high-dose tran-
examic acid. Double-blind, randomized clinical trial. J Tho-
rac Cardiovasc Surg 1995;110:835– 42.
9. Feindt P, Seyfert UT, Volkmer I, Straub U, Gams E. Celite
and kaolin produce differing activated clotting times during
cardiopulmonary bypass under aprotinin therapy. Thorac
Cardiovasc Surg 1994;42:218–21.
10. Brown RS, Thwaites BK, Mongan PD. Tranexamic acid is
effective in decreasing postoperative bleeding and transfu-
sions in primary coronary artery bypass operations: a dou-
ble-blind, randomized, placebo-controlled trial. Anesth
Analg 1997;85:963–70.
11. Westaby S. Aprotinin in perspective. Ann Thorac Surg 1993;
55:1033– 41.
12. Bennett-Guerrero E, Sorohan JG, Gurevich ML, et al. Cost-
benefit and efficacy of aprotinin compared with «-aminoca-
proic acid in patients having repeated cardiac operations.
Anesthesiol 1997;87:1373– 80.
13. Harmon DE. Cost/benefit analysis of pharmacologic hemo-
stasis. Ann Thorac Surg 1996;61:S21–5.
14. Tretiak R, Laupacis A, Rivière M, McKerracher K, Souêtre E,
Canadian Cost of Transfusion Study Group. Cost of alloge-
neic and autologous blood transfusion in Canada. Can Med
Assoc J 1996;154:1501– 8.
15. Alderman EL, Levy JH, Rich JB, et al. Analyses of coronary
graft patency after aprotinin use: results from the Interna-
tional Multicenter Aprotinin Graft Patency Experience
(IMAGE) trial. J Thorac Cardiovasc Surg 1998;116:716–30.
16. Karski J, Iwanochko M, Kucharczyk W, et al. Tranexamic
acid and early graft patency in elective aortocoronary bypass
(ACB) surgery. Can J Anesth 1999;5:A29.
17. Westaby S, Katsumata T. Editorial: aprotinin and vein graft
occlusion—the controversy continues. J Thorac Cardiovasc
Surg 1998;116:731–3.
18. Bennett-Guerrero E, Sorohan JG, Gurevich ML, et al. Cost-
benefit and efficacy of aprotinin compared with «-aminoca-
816 WONG ET AL
APROTININ AND TRANEXAMIC ACID
proic acid in patients having repeated cardiac operations.
Anesthesiol 1997;87:1373– 80.
19. Blauhut B, Harringer W, Bettelheim P, Doran JE, Späth P,
Lundsgaard-Hansen P. Comparison of the effects of aproti-
nin and tranexamic acid on blood loss and related variables
after cardiopulmonary bypass. J Thorac Cardiovasc Surg
1994;108:1083–91.
20. Boughenou F, Madi-Jebara S, Massonnet-Castel S, Benmos-
bah L, Carpentier A, Cousin MT. Fibrinolytic inhibitors and
prevention of bleeding in cardiac valve surgery. Comparison
of tranexamic acid and high dose aprotinin. Arch Mal Coeur
1995;88:363–70.
21. Corbeau JJ, Monrigal JP, Jacob JP, et al. Comparison of
effects of aprotinin and tranexamic acid on blood loss in
heart surgery. Ann Fr Anesth Réanim 1995;14:154– 61.
22. Cousin MT, Boughenou F, Madi-Jebara S, Massonnet-Vastel
S, Benmosbah L. The use of antifibrinolytics in heart sur-
gery. 3 prospective studies. Cahiers d’Anesthésiologie 1993:
41;473– 84.
23. Eberle B, Mayer E, Hafner G, et al. High-dose «-aminocap-
roic acid versus aprotinin: antifibrinolytic efficacy in first-
time coronary operations. Ann Thorac Surg 1998;65:667–73.
24. Jamieson WRE, Dryden PJ, O’Connor JP, Sadeghi H, Ansley
DM, Merrick PM. Beneficial effect of both tranexamic acid
and aprotinin on blood loss reduction in reoperative valve
replacement surgery. Circulation 1997;96(Suppl II):96 –101.
25. Landymore RW, Murphy T, Lummis H, Carter C. The use of
low-dose aprotinin, «-aminocaproic acid or tranexamic acid
Downloaded from ats.ctsnetjournals.org by on June 8, 2013
Ann Thorac Surg
2000;69:808 –16
for prevention of mediastinal bleeding in patients receiving
aspirin before coronary artery bypass operations. Eur J Car-
diothorac Surg 1997;11:798 – 800.
26. Menichetti A, Tritapepe L, Ruvolo G, et al. Changes in
coagulation patterns, blood loss and blood use after cardio-
pulmonary bypass: aprotinin vs tranexamic acid vs epsilon
aminocaproic acid. J Cardiovasc Surg 1996;37:401–7.
27. Mongan PD, Brown RS, Thwaites BK. Tranexamic acid and
aprotinin reduce postoperative bleeding and transfusions
during primary coronary revascularization. Anesth Analg
1998;87:258– 65.
28. Penta de Peppo A, Pierri MD, Scafuri A, et al. Intraoperative
antifibrinolysis and blood-saving techniques in cardiac sur-
gery. Prospective trial of 3 antifibrinolytic drugs. Tex Heart
Inst J 1995;22:231– 6.
29. Pugh SC, Wielogorski AK. A comparison of the effects of
tranexamic acid and low-dose aprotinin on blood loss and
homologous blood usage in patients undergoing cardiac
surgery. J Cardiothorac Vasc Anesth 1995;9:240– 4.
30. Speekenbrink RGH, Vonk ABA, Wildevuur CRH, Eijsman L.
Hemostatic efficacy of dipyridamole, tranexamic acid, and
aprotinin in coronary bypass grafting. Ann Thorac Surg
1995;59:438– 42.
31. Trinh-duc P, Wintrebert P, Boulfroy D, Albat B, Thevenet A,
Roquefeuil B. Efficacy of «-amino-caproic acid versus high-
dose aprotinin on intra- and postoperative blood loss in
cardiac surgery. Ann Chir Thorac Cardiovasc 1992;46:
677– 83.
 Aprotinin and tranexamic acid for high transfusion risk cardiac surgery
Bill I. Wong, Richard F. McLean, Stephen E. Fremes, Katherine A. Deemar, Ellen M.
Harrington, George T. Christakis and Bernard S. Goldman
Ann Thorac Surg 2000;69:808-816

Updated Information including high-resolution figures, can be found at:

& Services http://ats.ctsnetjournals.org/cgi/content/full/69/3/808
References This article cites 27 articles, 16 of which you can access for free at:
http://ats.ctsnetjournals.org/cgi/content/full/69/3/808#BIBL
Citations This article has been cited by 16 HighWire-hosted articles:
http://ats.ctsnetjournals.org/cgi/content/full/69/3/808#otherarticles
Permissions & Licensing Requests about reproducing this article in parts (figures, tables) or
in its entirety should be submitted to:
http://www.us.elsevierhealth.com/Licensing/permissions.jsp or
email: healthpermissions@elsevier.com.
Reprints For information about ordering reprints, please email:
reprints@elsevier.com

Downloaded from ats.ctsnetjournals.org by on June 8, 2013