Ultrafiltration During Cardiopulmonary Bypass:
Laboratory Evaluation and Initial clirucal Experience
Donald J. Magilligan, Jr., M.D., and Choken Oyama, M.D.
ABSTRACT Ultrafiltration during crystalloid hemodilu-
tion cardiopulmonary bypass (CPB) was evaluated in two
groups of mongrel dogs: in one group during 2 hours of
CPB with the heart empty and beating and in the other
during 90 minutes of cold cardioplegic arrest followed by
30 minutes of recovery. In both groups, the accumulation
of extravascular lung water was less in the dogs undergo-
ing ultrafiltration than in control animals.
In 10 patients with clinical evidence of severe fluid over-
load, ultrafiltration was employed during CPB. The
amount of fluid removed ranged from 1,700 to 6,100 ml
(mean, 3,240 f 1,481 ml [standard deviation])and resulted
in an average intraoperative fluid balance of - 901 f 2,537
ml, a weight gain of 1.9 f 2.5 kg, and a decrease in ex-
travascular lung water from 1,132 f 183 ml to 919 f 267 ml
( p = 0.209). Ultrafiltration is a safe, effective means of
removing body water and of preventing further accumula-
tion of such water during hemodilution CPB.
Ultrafiltration is the removal of water and solutes from
blood by convective transport across a semipermeable
membrane. It is different from conventional dialysis,
which is a diffusion of solutes across a semipermeable
membrane into a dialysate bath. Ultrafiltration removes
water and plasma concentration of solutes from the
blood; the driving force is the pressure gradient across
the membrane. Dialysis also removes solutes from
blood, but the driving force here is the solute gradient
between the blood and dialysate bath (Fig 1). When a
dialysis patient needs fluid as well as solutes removed,
ultrafiltration is employed either simultaneously or se-
quentially with dialysis.
The use of crystalloid prime with resultant hemodilu-
tion for cardiopulmonary bypass (CPB) has the advan-
tages of decreased blood requirements, improved urine
output, and reduced risk of pulmonary and renal failure
[l]. However, dilution perfusion has been shown to re-
sult in an immediate weight gain in the patient and in an
increase of up to 33% in the measured extracellular fluid
space (21. In most patients, the excess water is well toler-
ated and rapidly excreted. However, in other patients
From the Division of Cardiac and Thoracic Surgery, Henry Ford Hos-
pital, Detroit, MI.
Presented at the Forum on Fundamental Surgical Problems, American
College of Surgeons, Chicago, IL, Oct 25, 1982.
Accepted for publication Mar 24, 1983.
Address reprint requests to Dr.Magilhgan, Division Head, Cardiac and
Thoracic Surgery, Henry Ford Hospital, 2799 W Grand Blvd, Detroit,
MI 48202.
33
with severe preoperative fluid overload, the additional
water volume, particularly in the lungs, may lead to or-
gan dysfunction. The present report describes both our
laboratory evaluation of ultrafiltration during CPB and
our initial clinical experience.
Material and Methods
Laboratory Evaluation
ANIMAL MODEL. Mongrel dogs weighing between 25 and
30 kg were anesthetized with intravenous pentobarbital
(20 mg per kilogram of body weight). Anesthesia was
maintained with halothane (1.0%). Ventilation was car-
ried out through a cuffed endotracheal tube with a tidal
volume of 20 m l k g at a rate of 8 to 12 cycles per minute
delivered by a Bird Mark 7 respirator. A polyethylene
catheter was inserted into a jugular vein and advanced
into the right atrium for infusion of fluid. A Swan-Ganz
catheter was inserted into the pulmonary artery through
a jugular vein. An American Edwards lung water cathe-
ter, a 5F catheter with a distal opening for sampling and
injection and a thermistor at its distal tip, was inserted
into the femoral artery and positioned in the abdominal
aorta. The Swan-Ganz and lung water catheters were
connected to pressure transducers, and pulmonary ar-
tery pressure, systemic arterial pressure, and electrocar-
diograms were recorded continuously. Pulmonary capil-
lary wedge pressure (PCWP) and thermodilution cardiac
output (CO) were determined before bypass and 30 min-
utes after bypass. Cardiac index (CI) was calculated us-
ing body surface area derived from animal body weight
[3]. Hematocrit, hemoglobin and electrolyte levels, and
colloid oncotic pressure (Weil Oncometer, Instrument
Laboratories, Lexington, MA) were obtained before by-
pass and 30 minutes after bypass. The hydrostatic gra-
dient between colloid oncotic pressure and PCWP,
which estimates Starling’s forces, was calculated before
and after bypass.
CARDIOPULMONARY BYPASS. A median sternotomy was
performed, and the animal was heparinized (3 mgkg).
The left femoral artery and both venae cavae were can-
nulated and connected to a Sarns modular heart-lung
machine. A Bentley Temptrol Q-200A bubble oxygen-
ator was primed with a solution of 1,500 ml of Ringer’s
lactate, 12.5 gm of mannitol, 500 mg of calcium gluco-
nate, and 1,500 units of heparin. The animal was placed
on CPB, and mechanical ventilation was turned off with
the lungs partially inflated at 20 mm Hg.The left side of
the heart was vented by a cannula inserted into the left
ventricle through the right superior pulmonary vein.
The venae cavae were snared and bypass flow was
maintained at 100 ml/min/kg; lactated Ringer‘s solution
was added to maintain this flow. In Group 1, the ani-
34 The Annals of Thoraac Surgery Vol 37 No 1 January 1984
+-----
Blood
t Filtrate
ULTRAFILTRATION
DWyaato
DIALYSIS
Fig 1 . Ultrafiltration r m o w s water and solutes (Filtrate)by conm-
tive transport across a Semipermeable membrane, with the driving
force being the transmembrane pressure gradient. Dialysis mows
solutes from blood, with the driving force being the solute gradient
between the blood and dialysate bath.
mals were kept at normothermia with hearts beating for
120 minutes and then weaned from bypass. In Group 2,
the animals were cooled to 28T, and 150 ml of cardio-
plegic solution was instilled into the aortic root im-
mediately after cross-damping and again at 20,40, 60,
and 80 minutes. The cardioplegic solution consisted of
Isolyte-S (McGaw, Irvine, CA), to which 20 mEq of
potassium chloride, 30 mEq of sodium bicarbonate, and
15 gm of dextrose were added to total 1,OOO ml. The
resultant composition included the following: K+, 25
mEq/L; Na+, 178 mEq/L; C1-, 109 mEq/L; Mg++, 2.7
mEq/L; NaHC03-, 50 mEq/L; pH, 7.9; osmolarity, 460
mOsm/L.
ULTRAFILTRATION. An Amicon Diafdter-20 (Amicon,
Lexington, MA) was interposed between the arterial
inflow and venous return of a standard CPB circuit (Fig
2). Ultrafiltration could be turned on or off without dis-
turbing the total bypass flow by clamping the filter
inflow. The rate of ultrafiltration was controlled by ad-
justing a screw clamp on the filtrate outflow of the
Diafilter. An attempt was made to remove all priming
fluid as well as all fluid added during bypass; additional
fluid was removed if bypass flows and pressures could
be maintained.
LUNG WATER. Extravascular lung water was deter-
mined from the extravascular thermal volume of the
lung as measured by the double indicator dilution tech-
nique, with indocyaninegreen dye used as the intravas-
cular indicator and the transfer of heat as the extravascu-
lar indicator. Through the right atrial catheter, 10 ml of
iced indocyanine green dye (25 mg of dye diluted to 50
ml at 5°C) was injected. Thermal curves were sensed
through the thermistor at the tip of the lung water cathe-
ter, and through this catheter, blood also was with-
drawn. Dye concentration was measured by a den-
sitometer (Waters Instrument Co., Rochester, MN), and
both curves were inscribed on a strip-chart recorder. The
thermal and dye curves were analyzed on-line by a Lung
Water Computer (American Edwards, Santa Ana, CA),
t
t
HEART LUNG
t
ULTRAr
FILTER
5r
Fig 2. 7Xe ultrafilter is inserted as a bypass circuit between the arte-
rial and uenous lines of the standard cardiopulmonary bypass a p
paratus. The amount offiltrate can be controlled by partial occlusion
of either the arterial line or the filtrate line from the ultrafilter.
which gave a digital recording of CO, dye decay time,
thermal decay time, dye mean transit time (M'TT), ther-
mal MlT, and extravascular lung water. The average of
three readings was taken, divided by body weight, and
expressed as milliliters per kilogram.
At the conclusion of seven of the experiments,
gravimetric lung water was determined by the method
of Pearce and associates [4] as modified by Holcroft and
Trunkey [5], with a correction for lung blood water.
Clinical Experience
PATIENT SE~BCI'ION. Ten patients with dinical and ra-
diographic evidence of fluid overload were selected to
have ultrafiltration during CPB. The diagnoses in these
patients included dysfunction of a prosthetic mitral
valve (3 patients); acute ischemic ventricular septal de-
fect (1); acute ischemic mitral valve incompetence (1);
mitral stenosis (1);aortic stenosis (1);mitral valve incom-
petence due to acute infective endocarditis (1); acute
myocardial infarction and pump failure (1);aortic and
mitral valve incompetence and atrial septal defect (1).
CARDIOPULMONARY BYPASS. All patients underwent
CPB with moderate systemic hypothermia (28"C), non-
blood prime, and a Sd-Med membrane oxygenator. Top-
ical cooling was employed except in reoperations when
it was technically difficult. The left side of the heart was
routinely vented. Caval tapes were not used routinely
except when a right atriotomy was performed. Crystal-
loid cardioplegia (modified St. Thomas' Hospital solu-
tion) was routinely employed. Flows were maintained at
2.5 Umin/m2, and lactated Ringer's solution was added
as needed to maintain flow. When the hematocrit
35 Magilligan and Oyama: Ultrafiltration during CPB

dropped to less than 18%, whole blood was added to the
perfusate. Intake fluid volumes were a summation of
intravenous crystalloid, blood, perfusate, and cardio-
plegic solution. Output volume was a summation of
Table 1 . Group 1: Comparison of Animals Undergoing
120 Minutes of Nonnothennic Bypass with Heart Empty
and Beating, with and without Ultrafiltrationa
~~ ~

Variable Control ultrafiltration p Value

urine, ultrafiltrate, and blood loss.
ULTRAFILTRATION. An Amicon Diafilter-20 was used in No. of animals 4 4 ...
8 patients and a Bard Hemoconcentrator in the remain- Fluid in (ml) 2,350 f 1,090 2,750 f 500 0.480
ing 2. Both filters were inserted in the CPB circuit as in Fluid out (ml) 180 z 232 2,288 f 315 0.002b
the animal experiments (see Fig 2). An attempt was A PCWP (mm Hg) t 1.0 f 3.9 t 2.5 f 6.0 0.5%
made to remove all priming fluid, all fluid added during A COP (mm Hg) 1 6.9 f 2.8 1 8.2 f 1.5 0.650
bypass, and any additional fluid as long as bypass flows A COP-PCWP N (ITI Hg) 111.2 f 4.4 110.5 f 3.2 0.860
and pressures could be maintained. A CI (Umin/m') 11.20 f 0.3 11.65 f 0.8 0.310
The Amicon device is composed of 5,000 hollow p l y - A EVLW (mvkg) t2.m 2 0.5 r 1.03 r 0.2 0.W
sulfone fibers with an internal diameter of 200 p ? lo%, ~~ ~ ~

and an active surface area of 0.25 m2. The high perme-
ability of the membrane allows flux of water and solutes
with a molecular weight of less than 50,000 daltons. It is
compact (5" long) and does not require suction to pro-
duce an effective transmembrane gradient.
The Hemoconcentrator is composed of hollow cellu-
lose acetate fibers with an internal diameter of 200 p and
has an active surface area of 1.8 m2. The membrane al-
lows a flux of water and solutes with a molecular weight
of less than 60,000 daltons and requires the addition of
suction to generate transmembrane pressures.
LUNG WATER. As in the animal experiments, extravas-
cular lung water was determined by using the Lung Wa-
ter Computer. Dye was injected into a catheter, the tip of
which was positioned in the right atium, and sampling
was done from a lung water catheter placed in the distal
aorta through the common femoral artery.
Statistical Methods
Comparisons between prebypass and postbypass values
were made using the Student t test for paired data. A
probability of less than 0.05 was considered significant
for the laboratory experiments. Values for thermal and
dye extravascular lung water and for gravimetric ex-
travascular lung water determinations were compared
using simple regression analysis.
Results
Laboratory €valuation
All animals were successfully weaned from bypass with-
out the need for catecholamine support. In Group 1(120
minutes of bypass with the heart empty and beating),
the amounts of fluid added to the control animals and to
the ultrafiltered animals were similar. The amount of
fluid removed in the control group by renal excretion
was 180 ml; in the ultrafiltered animals the amount of
filtered fluid and urine removed was 2,288 ml, which
was significantly different ( p < 0.01). Although the col-
loid oncotic pressure, PCWP, and gradient between
these variables all changed in a direction favoring the
accumulation of lung water, the differences before and
after bypass were similar in the control and ultrafiltered
animals. However, the postbypass increase in extravas-
cular lung water was significantly less in the ultrafiltered
animals (Table 1).
'Values are shown as mean f standard deviation; A represents mean
change between prebypass and postbypass values.
%tatistically signiGcant (p < 0.05).
I "= pulmonary cap- wedge pressure; COP = colloid oncotic
pressure; CI = cardiac index; EVLW = extravascular lung water.
Both control and ultrafiltered animals in Group 1
showed a marked but similar decrease in CI after by-
pass. The CO enters into the equation for the measure-
ment of lung water as follows:
EVLW = mUl-1- X CO
mqe
where EVLW = extravascular lung water.
Because the postbypass decrease in CO might have
had an effect on the accuracy of the extravascular lung
water measurement, a second group of animals was
studied using cold potassium cardioplegic arrest to
maintain the postbypass CO at nearly normal levels and
to eliminate any possible effect that a low CO might
have on measurement of extravascular lung water.
In Group 2 (90minutes of cardioplegic arrest and 30
minutes of recovery), the amount of fluid added to the
ultrafiltered animals was significantly greater than that
added to the control animals ( p < 0.01). The amount of
fluid removed in the control group was 110 ml (urine)
and in the ultrafiltered group, 2,710 ml, a significant
difference (p < 0.01). Again, the colloid oncotic pres-
sure, PCWP, and gradient between these pressures
changed in a direction favoring the accumulation of lung
water, but the prebypass and postbypass differences
were similar in control and ultrafiltered animals. In addi-
tion, the postbypass change in CI was very small and
was similar in control and ultrafiltered, animals. The
postbypass increase in extravascular lung water was
significantly less in the ultrafiltered animals (Table 2).
The concentration of solutes in plasma and ultra-
filtrate is shown in Table 3. As expected, the compo-
sition of the filtrate was similar to that of plasma and
there was no protein in the ultrafiltrate. The lack of pro-
tein in the ultrafiltrate was confirmed by the biuret test
for measuring protein.
A comparison of thermal-dye and gravimetic deter-
36 The Annals of Thoracic Surgery Vol 37 No 1 January 1984

Table 2 . Group 2: Comparison of Animals Undergoing 90

Minutes of Cold Cardioplegic Arrest Followed by 30 -
12.0
Minutes of Recovery with and without Ultrafiltration’
Variable Control Ultrafiltration p Value y 10.0:
?
d
No. of animals 2 4 f 8.0-
Fluid in (ml) 2,775 -c 25 3,425 ? % 0.001b El
Fluid out (ml) 110 ? 120 2,710 f 258 O.OOlb 6.0-
A PCWP (mm Hg) T 4.0 f 1.4 T 1.8 f 5.2 0.597
A COP (nun Hg) 110.5 f 0.4 1 7.8 f 2.4 0.205 4.0 -
A COP-PCWP (mm Hg) 114.3 t 1.5 110.5 f 4.7 0.355
I I
A CI (Umin/m*) 10.90 t 0.8 10.65 f 0.3 0.430 4.0 6.0 8.0 10.0 12.0
A EVLW (mVkg) T3.40 t 0.4 T0.95 ? 1.1 0.04Sb GRAVIMETRIC
Fig 3. There was close correlation (r = 0.96) between thermal-dye
‘Values are shown as mean ? standard deviation; A represents mean
change between prebypass and postbypass values. and gravimetric extravaxular lung water determinations over a wide
%tatisticallysignificant ( p < 0.05). range of values in 7 animals.
PCWP = pulmonary capillary wedge pressure; COP = colloid oncotic
pressure; CI = cardiac index; EVLW = extravascular lung water.
Weight on the first postoperative day ranged from 5.2 kg
more to 1.6 kg less than preoperative weight and av-
Table 3 . Comparison of Concentration of Solutes eraged +1.9 ? 2.5 kg.
in Plasma and Ultrafiltrate in Animal Experiments’ The composition of the ultrafiltrate was similar to
plasma, except for calcium and phosphorus, which for
Solute Plasma Ultrafiltrate p Value unknown reasons were higher in the filtrate than in the
plasma (Table 4). This was not true in the animal ex-
Na+ (mEqL) 142 2 0.8 140 2 0.8 0.225 periments.
K + (mEqL) 4.0 f 0.6 3.7 2 0.6 0.636 Hemodynamic data showed that before operation
CI- (mEqL) 114 2 3.3 122 f 5.4 0.051 these patients had extremely high left atrial pressures
Ca++ (mg/dl) 7.5 2 1.4 5.2 f 0.5 0.098 (25.8f 5.5 mm Hg), which decreased to 18.2 ? 6.1mm
PO4’ (mg/dl) 3.2 2 0.8 2.6 f 0.9 0.538 Hg postoperatively (p = 0.038).The mean preoperative
Urea nitrogen (mg/dl) 14.3 2 5.5 13.3 2 2.5 0.724 CI was falsely high (3.84 ? 2.1 LJmidm’) because of the
Total protein (gm/dl) 3.9 2 0.4 0 0.000b extremely high CO in the patient with sepsis due to
endocarditis (Table 5).
‘Values are shown as mean f standard deviation. Values of electrolytes and urea nitrogen did not show
bstatistically significant ( p < 0.05).
significant changes postoperatively. The change in he-
matocrit from 32 -C 5% before operation to 27 ? 5% after
operation was significant (p = 0.032). The decrease in
minations of extravascular lung water was performed in total protein from 6.3 -C 0.6 @dl to 5.3 f 0.9 g d d l
7 animals and showed close correlation ( r = 0.96)(Fig 3). postoperatively was also significant ( p = 0.019) (see
Table 5).
Clinical Experience Extravascular lung water was high in the 5 patients in
Operative procedures included mitral valve replacement whom it was measured preoperatively (mean, 15.4 -C
(MVR) (2 patients); MVR and tricuspid valve replace- 3.8 mlkg), compared with the normal 5 to 8 f i g . Since
ment (1patient); MVR and tricuspid annuloplasty (1pa- the weight of the patients was not recorded immediately
tient); MVR and coronary artery bypass grafting (CABG) after operation, the extravascular lung water was re-
(1 patient); open mitral commissurotomy and tricuspid corded as its absolute value. Although levels of this vari-
annuloplasty (1 patient); repair of ventricular septal de- able decreased from 1,132 ? 183 ml preoperatively to
fect and CABG (1patient); CABG (1patient); aortic valve 919 -C 267 ml postoperatively, the change was not
replacement (AVR) (1patient); and AVR, MVR, repair of significant (p = 0.209).
sinus of Valsalva aneurysm, and repair of atrial septal
defect (1 patient). Comment
One patient, who did not survive the first 24 hours, Ultrafiltration has previously been shown to be an ex-
was successfully weaned from bypass with a satisfactory tremely effective technique for removal of fluid in dialy-
CI so that early studies (30minutes after bypass) could sis patients [6,7.Bergstrom [8] has demonstrated that
be performed. when ultrafiltration is separated from dialysis, large
The amount of ultrafiltrate removed ranged from amounts of fluid can be removed while hemodynamic
1,700 to 6,100 ml and averaged 3,240 ? 1,481 ml (mean stability is maintained. Ultrafiltration without dialysis
f standard deviation). The fluid balance ranged from has been shown to be effective when used clinically to
+ 1,650 to - 7,850 ml and averaged - 901 -C 2,537ml. remove excess water in patients with intractable conges-
37 Magilligan and Oyama: Ultrafiltration during CPB

Table 4 . Comparison of Concentration of Solutes age rise in hematocrit from 21 to 31%. Other clinical
in Plasma and Ultrafiltrate in 5 Patients Undergoing studies likely will be forthcoming.
Ultrafiltration during Cardiopulmonary Bypass” One reason for the limited use of ultrafiltration during
bypass has been the inability to measure organ water
Solute Plasma Ultrafiltrate p Value content accurately by methods that are not destructive.
Na+ (mEqL) 132 f 4.6 131 2 3.7 0.177 Although crystalloid hemodilution has been shown to
K + (mEqL) 4.4 f 0.91 4.4 f 0.84 0.560 result in increased water content in the lungs [14], heart,
kidney, and gastrointestinal tract [ 151, the technique for
C1- (mEqL) 101 f 8.9 98 2 4.7 0.246
measuring organ water was from wet to dry weight and
Ca++ (mg/dl) 5.7 f 0.94 7.9 f 0.63 0.002b
was not applicable clinically. The thermal-dye double
PO4’ (mg/dl) 4.1 f 0.91 5.2 f 1.4 0.031b indicator dilution technique for measuring extravascular
Urea nitrogen (mg/dl) 47.6 2 24.2 40 2 30.7 0.102 lung water is simple, can be repeated, and correlates
’Values are shown as mean ? standard deviation
more closely with gravimetric lung water determinations
%tatistically significant ( p < 0.05). than the previously used isotope techniques [16]. ES-
tablishment of this technique in our laboratory, after
previous documentation of a close correlation with
Table 5 . Comparison of Preoperative and gravimetric lung water values ( r = 0.95) in 28 measure-
Postoperative Measurements in 10 Patients Undergoing ments over a wide range [17], allowed us to measure
Ultrafiltration during Cardiopulmonary Bypass” lung water repetitively. We showed an increase in ex-
travascular lung water during routine bypass in dogs
Variable Preop. Postop. p Value and demonstrated that the postbypass increase in this
variable was significantly less when ultrafiltration was
CI (L/min/m*) 3.84 2 2.1 2.87 f 0.6 0.270
employed during bypass. The correlation between ther-
LAP (mm Hg) 25.8 f 5.5 18.2 f 6.1 0.038b mal-dye and gravimetric lung water determinations was
Na+ (mEqlL) 131 f 3 132 f 5 0.643 close ( r = 0.96) in 7 of the present experiments in which
K + (mEqlL) 4.5 2 0.7 4.2 f 0.6 0.269 both were measured.
C1- (mEqlL) 97 2 4.5 97 5 6.9 0.925 We were unable to show a significant decrease in lung
Urea nitrogen (mg/dl) 38 5 33.6 37 f 30.1 0.758 water by the thermal-dye technique in the 5 ultrafiltered
Total protein (gddl) 6.3 f 0.6 5.3 f 0.9 0.019b patients in whom this variable was measured. The fail-
Hematocrit (%) 32 f 5 27 f 5 0.032b ure of the extravascular lung water level to rise after
EVLW (ml) 1,132 2 183 919 f 267 0.209 cardiac procedures, even though there is a positive fluid
balance of 4 to 5 liters, also has been noted in patients
‘Values are shown as mean 2 standard deviation. with normal ventricles undergoing simple coronary ar-
%tatistically significant ( p < 0.05).
tery bypass [18]. This underscores the fact that it is very
CI = cardiac index; LAP = left atrial pressure; EVLW = exhavascular difficult to demonstrate the beneficial effects of any
lung water.
intraoperative or postoperative intervention on pulmo-
nary function. We suspect, but cannot prove, that pa-
tients who had preoperative extravascular lung water
tive heart failure 191. It has also proven beneficial in re- values far greater than the normal 5 to 8 mYkg were
moving fluid in patients with respiratory and renal fail- better postoperatively because they did not have an ad-
ure, resulting in improved pulmonary function [lo]. ditional fluid accumulation of 4 to 5 liters during bypass.
Dilution perfusion, for all its advantages, results in The only indication that ultrafiltration was of value is
decreased colloid oncotic pressure of plasma and an ele- that 6 of the 9 extremely ill patients were extubated on
vated interstitial water content 1111. Since further in- the first postoperative day. Because it has been difficult
creases in body water might be detrimental to a patient to prove a beneficial effect of ultrafiltration on the pul-
already overloaded with fluid, ultrafiltration has a logi- monary or other organ systems, we have confined its
cal place in the CPB procedure. To date, however, there use during bypass to those patients with clinical,
has been a scarcity of clinical reports on the use of radiological, or thermal-dye evidence of excess body
ultrafiltration during bypass. Darup and colleagues [121 water.
reported using ultrafiltration during CPB in 10 patients The challenge is always raised as to whether the ther-
with either renal failure or oliguria. The average amount mal-dye measurement of extravascular thermal volume
of filtrate removed was 3.4 liters per patient during pro- of the lung truly represents extravascular lung water.
longed procedures in which the average bypass time The correlation with the “gold standard” of gravimetric
was 3 hours 56 minutes. There was minimal or no lung water was close in these experiments. The spe-
weight gain postoperatively, and ultrafiltration was ac- cific questions about the thermal-dye technique revolve
complished without sacrificing bypass flow or pressure. around the effects of thermal indicator loss and variation
Romagnoli and co-workers [13] reported using ultra- in CO on the measurement. Recently, these objections
filtration after CPB in 24 children aged 2 to 7 years. Fil- to the technique have been analyzed critically and an-
tration was accomplished safely and resulted in an aver- swered by Lewis and associates [19].
38 The Annals of Thoracic Surgery Vol 37 No 1 January 1984
It may be argued that in the animal experiments, fluid
was added to the perfusate and then filtered off. If this
was beneficial, could not the same results be accom-
plished simply by withholding the addition of fluid to
the perfusate? In order to maintain bypass flow at 100
ml/min/kg in a dog for 2 hours, substantial amounts of
fluid must be added to the perfusate, as others have
shown [ZO]. It is the dog's progressively declining vascu-
lar resistance during bypass, requiring large volume in-
fusion for maintenance of flow, that makes it an excel-
lent model for determining whether ultrafiltration can
successfully and safely remove some of this volume. The
clinical counterpart would be the patient with a normal
amount of body water who requires prolonged bypass
(more than 2 hours) for the repair of complex problems
(arch replacement, multiple valve replacement, or coro-
nary bypass procedures).
The question naturally arises as to which solutes are
filtered out with water. Solutes with a molecular weight
below 50,000 or 60,000 daltons will be filtered out. Those
with lower molecular weights, such as urea (60 daltons)
and creatinine (113 daltons), are filtered out in the same
concentrations as in plasma. Therefore, they have a siev-
ing coeficient ( S ) of 1 where:
S = solute - ultrafiltrate concentrationkolute -
blood concentration
As solutes progress in molecular weight toward 50,000
daltons, the sieving coefficient is lower. For practical
purposes, the ultrafiltrate contains plasma concentra-
tions of urea and electrolytes and no protein. Drugs that
are not bound to protein are removed as determined by
their molecular weight. Heparin has a sieving coefficient
of 1 and therefore is removed in the same concentration
as in plasma. During bypass, we monitor activated clot-
ting times to determine heparin dosage, and we have
not noted an increased heparin requirement in ultra-
filtered patients.
There is concern in using ultrafiltration during bypass,
since some dialysis membranes have been reported to
cause complement activation and a decrease in circulat-
ing polymorphonuclear leukocytes (PMNs) [21]. The de-
crease is due to PMN aggregation occurring in the pul-
monary microvasculature and may result in hypoxia
[22]. This reaction appears to occur with membranes
made of cuprophan, regenerated cellulose, and cellulose
acetate, and not with those made of polyacrylonitrile or
polysulfone [23, 24). Although the Hernoconcentrator
membrane is made of hollow cellulose acetate fibers, this
material has been used in 20 million dialysis procedures
with Cordis Dow dialyzers, and no hypersensitivity
reactions have been documented.* The Amicon device,
made of polysulfone, should be free of the problem of
hypersensitivity caused by complement activation.
'Technical bulletin entitled "Cordis Dow dialyzers carry no hypersen-
sitivity warnings." Miami, FL, Cordis Dow, 1980.
In conclusion, our laboratory evaluation of ultra-
filtration during CPB showed that large amounts of
water could be safely removed with a lowering of
extravascular lung water in the ultrafiltered animals as
compared to control animals. Ultrafiltration used clini-
cally has been safe and effective in removing up to 6
liters of plasma water during bypass. At present, its use
is reserved for patients requiring open-heart operations
who have excess body water as demonstrated by clinical
evaluation, radiographic examination, or elevated ther-
mal-dye extravascular lung water values.
Supported by a Michigan Heart Association Research Grant.
References
1. Lilleaasen P, Stokke 0: Moderate and extreme hemodilu-
tion in open-heart surgery: fluid balance and acid-base
studies. Ann Thorac Surg 25:127, 1978
2. Breckenridge IM, Digerness SB, Kirklin JW: Increased ex-
tracellular fluid after open intracardiac operation. Surg Gy-
necol Obstet 131:53, 1970
3. Davis LE: Treatment and management of burns. Vet Scope
8:7, 1963
4. Pearce ML, Yamashita J, Beazell J: Management of pulmo-
nary edema. Circ Res 16:482, 1965
5. Holcroft JW, Trunkey DD: Extravascular lung water follow-
ing hemorrhagic shock in the baboon. Ann Surg 180:480,
1974
6. Kramer P, Wigger W, Rieger J, et al: Arteriovenous
hemofiltration: a new and simple method for treatment
of over-hydrated patients resistant to diuretics. KIin
Wochenschr 55:1121, 1977
7. Paganini EP, Nakamoto S: Continuous slow ultrafiltration
in oliguric acute renal failure. Trans Am SOCArtif Intern
Organs 26:201, 1980
8. Bergstrom J: Ultrafiltration without dialysis for removal of
fluid and solutes in uremia. Clin Nephrol9156, 1978
9. Siverstein ME, Ford CA, Lysaght MJ, Henderson LW:
Treatment of severe fluid overload by ultrafiltration. N Engl
J Med 291:747, 1974
10. Lewis RM, Besso J, Henning RJ, Weil MH: Hemofiltration
for the treatment of adult respiratory distress syndrome.
Crit Care Med 8229, 1980
11. English TAH, Digerness S, Kirklin JW: Changes in colloid
osmotic pressure during and shortly after open intracardiac
operation. J Thorac Cardiovasc Surg 61:338, 1971
12. Darup J, Bleese N, Kalmar P, et al: Hemofiltration during
extracorporeal circulation. Thorac Cardiovasc Surg 27227,
1979
13. Romagnoli A, Hocker J, Keats A, Milan J: External hemo-
concentration after deliberate hemodilution. In Abstracts of
the Annual Meeting of the American Society of Anes-
thesiologists. Park Ridge, IL, American Society of Anes-
thesiologists, 1976, p 269
14. Demling RH, Hicks RE, Edmunds LH Jr: Changes in ex-
travascular lung water during venovenous perfusion. J
Thorac Cardiovasc Surg 71:291, 1976
15. Utley JR, Wachtel C, Cain RB, et al: Effects of hypothermia,
hemodilution, and pump oxygenation on organ water con-
39 Magilligan and Oyama: Ultrafiltration during CPB
tent, blood flow and oxygen delivery, and renal function.
Ann Thorac Surg 31:121, 1981
16. Holcroft JW, Hopewell P, Sheldon GF, et al: Thermodilu-
tion for measurement of lung water: comparison with
isotope and gravimetric methods. Circulation 48:Suppl 4:
447, 1973
17. Magilligan DJ Jr, Oyama C, Levin N: Interstitial and intraal-
veolar pulmonary edema: reversal by ultrafiltration. Crit
Care Med 9257, 1981
18. Sivak ED, Starr NJ, Graves JW, et al: Extravascular lung
water values in patients undergoing coronary artery bypass
surgery. Crit Care Med 10:593, 1982
19. Lewis FR, Elings VB, Hill SL, Christensen JM: The mea-
surement of extravascular lung water by thermal-green dye
indicator dilution. Ann NY Acad Sci 384:394, 1982
20. Dunn J, Kirsh MM, Harness J, et al: Hemodynamic, meta-
bolic, and hematologic effects of pulsatile cardiopulmonary
bypass. J Thorac Cardiovasc Surg 68:138, 1974
21. Craddick PR, Fehr J, Brigham KL, et al: Complement and
leukocyte-mediated pulmonary dysfunction in hemodialy-
sis. N Engl J Med 296:769, 1977
22. Jacob AI, Gavellas G, Zarco R, et al: Leukopenia, hypoxia,
and complement fixation with different hemodialysis mem-
branes. Kidney Int 18:505, 1980
23. Henderson LW, Miller ME, Hamilton RW, Norman ME:
Hemodialysis leukopenia and polymorph random mobil-
ity-a possible correlation. J Lab Clin Med 85:191, 1975
24. Bohler J, Krarner P, Gotze 0, et al: Leucocyte counts and
complement activation during pump-driven and arterio-
venous hemofiitration. In Kramer P (ed): Arterio-venous
Hernofiltration. Gtittingen, Vandenhoeck & Ruprecht,
1982, p 187