Comparison of bubble and membrane oxygenators in short and long perfusions

J THORAC CARDIOVASC SURG 78:655-666, 1979

Original Communications

Comparison of bubble and membrane oxygenators in short and long perfusions Eighty patients had cardiopulmonary bypass (CPB), half having short (109 :':: 11 minutes) perfusions and half having long (188 :':: 14 min) perfusions . Twenty patients in each group were perjused with bubble oxygenators (Bentley, Harvey, or Galen) and 20 with membrane oxygenators (Modulung or Tejfo). Hemodilution to a hematocrit value of 22 .5% ± 1.4% and hypothermia to 28° ± Z" C were used in all patients. Complete hemograms, sequential multiple analyzer 18 tests, coagulation profiles, blood gases and pH, three immunoglobulins, and two complement fraction proteins were sampled as follows: three times before perfusion, one to ten times during perfusion, 1 hour immediately after perfusion, and 4, 24, and 48 hours postoperatively. Data in concentration terms were compared statistically and reported as mean and standard error for each subset. Additionally, rates of gain or loss were calculated in terms of quantity per liter of blood pumped per minute. During perfusion for both duration sets, use of a membrane oxygenator resulted in greater pump flows (4.55 ± 0.15 Llmin versus 3.75 ± O.ll Llmin], lower total peripheral resistances (1,125 ± 63 dynes . sec . cm:> versus 1,652 ± 115 dynes' see' cm i''}, and greater urinary outputs (9.4 ± 1.1 mllmin versus 2.2 ± 0.6 mllmin) than in the bubble oxygenator subsets. Comparisons of measured and calculated data in the immediate postperjusion interval showed 110 differences between bubble and membrane oxygenator subsets for short perfusions. In long perfusions, the membrane subset had lower plasma hemoglobin and white cell concentrations and generation rates, smaller (3 to 8V:z times) lossses of IgG. IgM. C3 and shed blood necessitating less transfusion, and greater C4 losses. The membrane oxygenator systems used were more complex and costly and offered no advantages for short perfusions in adults. In anticipated long perjusions or where bleeding may be a problem, a membrane oxygenator appears more efficacious than bubble systems. For perjusions of less than 2 hours, membrane oxygenators had no biochemical or hematologic advantage over the bubble devices used in this study.

Richard E. Clark, M.D., Richard A. Beauchamp, B.S. (by invitation), Robert A. Magrath, B.S. (by invitation), John D. Brooks, B.A. (by invitation), Thomas B. Ferguson, M.D., and Clarence S. Weldon, M.D., St. Louis, Mo.

BlOOd-gaS interface oxygenators have been the principal type of artificial lung in clinical open-heart operations since the first operation in 1953. 1 Gibbon's screen oxygenator," the disc oxygenator designed by Anderson with modifications by Bjork'': 4 and Kaye and From the Division ofCardiothoracic Surgery, Washington University School of Medicine. St. Louis. Mo. 63110. Read at the Fifty-ninth Annual Meeting of The American Association for Thoracic Surgery. Boston. Mass., April 30 to May 2. 1979. Address for reprints: Richard E. Clark, M.D., Suite 3108 Queeny Tower. 4989 Barnes Hospital Plaza. St. Louis. Mo. 63110.

Cross," and the bubble oxygenators of DeWall" and Rygg and Kyvsgaard? were later supplanted by commercial hard-shell devices, which presently are the most widely used throughout the world. The initial physiological alterations observed with blood-gas interface devices, inlcuding peripheral oligemia and central hypertension, bleeding diatheses, and the complications of cerebral, renal, and cardiac failure, were attributed to protein denaturation, hemolysis, platelet destruction, and liberation of various unidentified polypeptides.

0022-5223/79/110655+12$01.20/0 © 1979 The C. V. Mosby Co.

655

The Journal of

656

Clark et al.

These clinical complications from total cardiopulmonary bypass (CPB) with the blood-gas interface devices prompted investigation of oxygenators which separated blood from oxygen. The observation by Kolff and associates" in 1956 of gas transfer in renal dialysis devices prompted development of the membrane oxygenator during the infancy of clinical CPB procedures. Clowes and colleagues" first reported use of a membrane oxygenator in patients in 1956. The early work of Lee's group'? spurred further development of membrane oxygenators, the practical advent of which held great hope of eliminating clinical complications of perfusion. The efforts of Kolobow, 11 Peirce," Lande;" and their associates have culminated in the manufacture of commercial membrane oxygenators. After almost 20 years of development, only two or three membrane oxygenators are in wide use in this country at the present time. During this decade the hard-shell bubble-film devices have improved markedly, with reduction of solid particulates present after manufacture and generation of gaseous microemboli. A better understanding of the need to avoid excessively high partial pressures of oxygen and to maintain partial pressures of carbon dioxide near physiological levels has resulted in improved bubble devices which use lower gas-blood flow ratios, a parameter first used in 1970 to characterize performance." These newer bubble-film oxygenators had integrated heat exchangers, filters in the cardiotomy suction circuits, and internal reservoirs which improve safety and ease of use and were of reasonable cost. The literature contrasting the membrane oxygenators and the bubble oxygenators is contradictory. Claims for the superiority of membrane oxygenators have not been substantiated by carefully controlled studies of the newer bubble devices and similar perfusion techniques. The purpose of this randomized, prospective clinical study was to determine the hematologic and biochemical alterations resulting from the use of bubble and membrane oxygenators for matched patient populations and perfusion times during and after operation. Emphasis was placed on study of immunologic and coagulation protein abnormalities engendered by the devices.

Patients Since 1972, membrane oxygenators have been used in 113 patients and bubble oxygenators in 2,126 patients undergoing cardiac operations at the Bames Hospital, St. Louis, Missouri. Two groups of adults requiring cardiac operation were selected prospectively-one group expected to have "short" perfusion

Thoracic and Cardiovascular Surgery

times and a second group expected to have "long" perfusion times. The patients were matched for age, sex, New York Heart Association Clinical Classification, and type of operation. Patients anticipated to have short perfusions, i.e., less than 2 hours, were those requiring atrial septal defect repair, single valve replacement, and single or double coronary artery bypass. Patients with multivalvular disease, valvular and ischemic heart disease, ischemic heart disease necessitating three or more coronary artery bypass grafts, and all requiring reoperation were placed in the long perfusion group. Prospective randomization of the patients to the bubble and membrane subsets in the anticipated short and long perfusion groups was made by random table. The patients were equally divided between long and short perfusion times and bubble and membrane groups.

Methods CPR apparatus and methodology. The CPB technique was similar in all patients. Two caval cannulas (No. 34 Fr.) were joined to a Tygon tube (1/2 inch inner diameter) connected directly to the venous port of the bubble oxygenator* or passed through a roller pump in the case of the membrane oxygenator. t In some of the latter cases, volume displacement transducers were used to control independently the venous and arterial roller pumps. IS All cardiotomy suction return and vent lines passed through roller pumps and were connected to a reservoir. The outlet tube of the reservoir passed through a roller pump and was connected to the venous side of the oxygenator with a cardiotomy Pioneer CA-100 filter interposed between the pump and the oxygenator. A No. 24 Fr. cannula was inserted into the ascending aorta through an aortotomy circled by two concentric purse-string sutures. All oxygenators were filled with crystalloid solution containing dextrose. The priming volumes of the bubble and membrane oxygenators were 2,105 ± 154 ml and 2,734 ± 262 ml, respectively. The priming volume was circulated slowly (~500 ml/min) prior to bypass and cooled to 15° ± 2° C in all cases. Perfusion was begun slowly and increased to maximal flow within 3 minutes. The tourniquets about the caval cannulas were tightened. *Bubble oxygenators: 14 Bentley Q-IOO and Q-200 units, Bentley Laboratories, Inc., Irvine, Calif.; 14 Harvey H-lOOO units William Harvey Research Corp., Santa Ana, Calif.; 12 GalenCobe Optifto 200 units, Cobe Laboratories, Inc., Lakewood, Colo. tMembrane oxygenators: 32 Travenol TMO units and eight Travenol Modulung units, Travenol Laboratories, Inc., Deerfield, III.

Volume 78

Comparison of bubble and membrane oxygenators

Number 5 November, 1979

657

Table I. Characteristics of patient groups Short perfusions Bubble

Membrane

Bubble

(N = 20)

(N = 20)

106±11* 16 47.3 ± 2.3 1.85 ± 0.15

112 ± 10 15 50.4 ± 2.9 1.88 ± 0.11

179 ± 12 12 52.6 ± 5.4 1.84 ± 0.14

197 ± 16 11 50.1 ± 4.7 1.81 ± 0.12

16 4

15 5

11 9

13 7

10

9

2

3

8

10

8 2

6

(N = 20)

Perfusion time (min) Males Age (yr) Body surface area (rn") N.Y.H.A. Class III

IV

Long perfusions

I

I

Membrane (N = 20)

Operations Single valve replacement, including reoperations Double valve replacement, including reoperations Repair of congenital defect Coronary artery bypass Coronary artery bypass plus valve replacement, including reoperations

2 8

I

10

1

*Mean ± SEM.

The esophageal temperature was lowered to 28° ± 2° C and the heart was bathed in 4° C Ringer's lactate solution. Pericardial fluid was pumped continuously by a roller pump and collected in a large waste receptacle. Continuous and intermittent ischemic arrest was used. Rewarming was begun nearly synchronously with removal of the aortic cross-clamp. The remaining contents of the oxygenator were slowly infused during the first 60 minutes after termination of CPB. Partial pressures of oxygen and carbon dioxide and pH in arterial blood were maintained in a range of 95 to 185 mm Hg (mean 155),31 to 54 mm Hg (mean 43), and 7.19 to 7.46 (mean7.31), respectively. Variables measured. Variables were measured I to 3 days preoperatively, immediately after the induction of general endotracheal anesthesia, immediately prior to initiation of CPB, at 30 minute intervals during CPB, and I, 4, 24, and 48 hours after CPB. The following values were measured; hematocrit, total hemoglobin, plasma hemoglobin, red blood cell count, white blood cell count, platelet count, and concentrations of immunoglobulins IgA, IgG, and IgM, complement fractions C3 and C4 total protein, fibrin split products, fibrinogen, sodium, potassium, chloride, carbon dioxide combining power, blood urea nitrogen, creatinine, alkaline phosphatase, serum glutamic oxaloacetic transaminase, lactic dehydrogenase, creatine phosphokinase, and glucose. The partial pressures of oxygen and carbon dioxide and pH for both arterial and venous blood were measured. Serial protamine sulfate precipi-

tation tests were performed, and partial thromboplastin time and prothrombin time were measured. Red cell indices were calculated. Records were maintained of total crystalloid fluid and colloid volume administered during operation, perfusion, and the postoperative interval; urinary output, blood loss, and blood replacement were recorded during the operative and postoperative intervals. Hemodynamic variables recorded included pump flow rate, systolic, diastolic, and mean arterial pressures, central venous pressure, pulmonary wedge pressure or left atrial pressure, heart rate, esophageal, blood, and water temperatures during perfusion, and rectal temperature in the postoperative course. Data treatment. Nearly all of the hematologic and biochemical raw data were expressed in terms of concentration. Concentration terms are greatly influenced by hemodilution during perfusion, administration of various fluids, urinary output, and blood loss during and after operation. One of the major purposes of this study was to ascertain gains or losses of specific variables as a function of time and type of oxygenator. Absolute quantities of various variables required the calculation of preoperative erthrocyte and plasma volumes based on surface area, sex, and age.!" Corrections were made for additions and losses as a function of time during perfusion. The gain or loss of a substance by destruction and liberation or polymeric surface absorption, aggregation, and trapping necessitated a scheme to normalize the data for the duration of the perfusion

The Journal of

658

Clark et al.

Thoracic and Cardiovascular Surgery

Table II. Pertinent immediate postperfusion data: Measured values and expected values by dilution alone (mean ± SEM) Short perfusion timest Bubble (N Variable

Hematocrit (%) Plasma hemoglobin (mg/lOO m1) White blood cells (per mrn" x 103 ) Platelets (per rnm" x 103) IgA (mg/lOO ml) IgG (mg/lOO ml) IgM (mg/100 ml) C3 (mgll 00 ml) C4 (mg/lOO ml)

Range of normal values

37-52 ~IO

4.8-10.8 200-400 90-450 800-1800 60-250 101-189 20-40

Preperfusion values* (N = 80)

36.7 7.7 6.8 208 206 1,236 155 110 41

± ± ± ± ± ± ± ± ±

2.1 2.7 0.6 31 27 144 29 10 6

Measured

22.6 50 10 107 117 751 91 62 27

± ± ± ± ± ± ± ± ±

2.1 6 3 16 17 119 8 6 5

I

= 20)

Membrane (N

Expected

4.8 4.2 128 127 763 96 68 25

± ± ± ± ± ± ± ±

1.6 0.4 19 17 89 18 6 4

Measured

21.3 43 9.1 109 115 703 86 69 22

± ± ± ± ± ± ± ± ±

1.9 5 1 13 16 134 14 10 4

I

= 20) Expected

4.4 3.9 121 120 718 90 64 24

± ± ± ± ± ± ± ±

1.5 0.3 18 16 84 17 6 4

·Plus or minus (±) 2 standard deviations. tp = NS for all measured variables (bubble versus membrane oxygenators). *p < 0.001 for measured variables (bubble versus membrane oxygenators). §p < 0.02 for measured variables (bubble versus membrane oxygenators). lip < 0.05 for measured variables (bubble versus membrane oxygenators). For all other measured values, p = NS.

and the volume of blood pumped. The amount of blood pumped was calculated by integrating the flow ratetime data from the perfusion record. Thus changes in variables of the small patient with a small blood volume who required small pump flows and a short period of perfusion could be compared to the changes in a patient who had a large pump flow, blood volume, and long duration of perfusion. Linear and nonlinear changes were examined by comparisons of the changes found in short perfusions to those found in long perfusions. Data analysis. All data were analyzed in terms of (1) raw data, (2) expected values of all changes caused by hemodilution, and (3) normalization for blood volumes, duration of perfusion, and volume of blood pumped. Data were analyzed for each subset for each time interval. The mean value, the standard deviation, and the standard error of the mean were determined for each variable in each time period for each patient group. A paired t test was used to test for significance when values were compared for the same patient groups. The unpaired Student's t test was used for determining statistical significance between subsets of patients for the same time intervals. Results Preoperative assessment. Preoperative values for patients in the two major sets (short versus long perfusions) and the two major subsets in each set (bubble versus membrane) were statistically tested for significance and no differences were found. Additionally,

the immediate preperfusion values were compared to those obtained prior to the day of operation and no statistically significant differences were found. Thus the subsequent data for all groups were comparable with respect to alterations resulting from perfusion and operation, and the influences of oxygenator type and duration of perfusion could be determined. The characteristics of the patient groups are given in Table I. These data show that a greater number of women, Class IV patients, and patients having double valve replacement were present in the long perfusion group, but they were similarly distributed between bubble and membrane subsets. Skewness would be anticipated on the basis of patient selection by procedure, wherein the more difficult or complex operations would require longer perfusion times and patients with significantly decreased preoperative ventricular function would require longer periods of extracorporeal support during weaning from the bypass system. Intraperfusion interval. There were no statistically significant differences in the raw data between the sets for a variable measured in concentration terms during CPB. These data reflect similar decreases in all groups as a result of hemodilution to a mean hematocrit value of 22.5% ± 1.4% (SEM). Data for hemoglobin and total protein concentrations were nearly identical in all groups (±5%) in reflecting the degree of hemodilution. Of all the variables analyzed, only three showed statistically significant differences between the bubble and membrane oxygenator groups during both short and long perfusions; pump flow rate, total peripheral

Volume 78

Comparison of bubble and membrane oxygenators

Number 5 November, 1979

Long perfusion times Bubble (N = 20) Measured

24.4 ± 119 ± 16 ± 69 ± 115 ± 534 ± 52 ± 64 ± 25 ±

2.3 8 2 17 18 92 19 12 3

Membrane (N = 20)

Expected

Measured

± ± ± ± ± ± ± ±

21.9 ± 2.0 60 ± 7 8 ± I 96±1I 102 ± 14 686 ± 119 77 ± 13 63 ± 9 13 ± I

5.1 4.5 139 137 824 103 73 27

1.8 0.4 21 18 96 19 7 4

Expected

4.6 4.0 124 123 736 92 65 24

± ± ± ± ± ± ± ±

1.6:1: 0.4§ 18 16 86 17 6 411

resistance, and urinary output. The mean pump flow rates were 3.75 ± 0.11 Llmin and 4.55 ± 0.15 Llmin (p < 0.05) for the bubble and membrane groups, respectively. The bubble oxygenator (short and long) group had a total peripheral resistance of 1,652 ± 115 dynes' sec' cm" and the membrane (short and long) group had a total peripheral resistance of 1,125 ± 63 dynes . sec . cm" (p < 0.05). Urinary outputs were 2.2 ± 0.6 ml/min and 9.4 ± 1.1 ml/min (p < 0.01) for the bubble and membrane groups, respectively. Perfusion times were similar for bubble and membrane subsets in short and long perfusions: 109 ± 11 minutes and 188 ± 14 minutes for short and long sets, respectively. The longest perfusion time was 316 minutes. Immediate postperfusion interval. The measured variables and those expected by dilution alone are shown in Table II and contrasted to the preperfusion values. The following values were measured in the immediate postperfusion period: hematocrit, total hemoglobin, total protein, fibrin split products, partial thromboplastin time, prothrombin time, fibrinogen, sodium, potassium, chloride, carbon dioxide combining power, blood urea nitrogen, creatinine, alkaline phosphatase, serum glutamic oxaloacetic transaminase, lactic dehydrogenase, creatine phosphokinase, red blood cell indices, and the arterial blood gases and pH. These valves were not statistically different between bubble and membrane groups in either short or long perfusions . Comparisons of the measured data for the bubble and membrane subsets showed no significant differences in any measured variable for short perfusions. However, in long perfusions, lower concentrations of plasma hemoglobin, white cell count, and C4 were found for the membrane oxygenator subset. The only significant difference found between long and short perfusions for

659

both oxygenators was a greater plasma hemoglobin concentration in long perfusions. Next, the measured values and the expected values were multiplied by the total of the sum of the calculated erythrocyte and plasma volumes of each patient corrected for additions and losses integrated with respect to time. The differences in total quantities were divided by the total volume of blood pumped and the perfusion time to yield a change in a value per liter of blood pumped per minute of perfusion. This analysis demonstrated significant differences in some variables not evident by statistical comparisons of raw data. These differences for long perfusions are shown in Table III and illustrated in percent change from the preperfusion values for short perfusions in Fig. 1 and for long perfusions in Fig. 2. No statistically significant differences were found between oxygenator subsets for the short perfusion interval by this analysis. New variables of statistical significance appeared in the long perfusion set. Differences of threefold to fourfold were found between bubble and membrane groups for rates of increase in quantities of plasma hemoglobin and white cells and rates of decrease of platelets and C4. Five to 8.5 fold differences between oxygenator groups were found for IgG, IgM, and C3. These data show that when perfusion time is extended to a mean duration of 3 hours, destruction, aggregation, and absorption of red cells and platelets and absorption and/or denaturation of two immunoglobulin and one complement fraction proteins occurred to a significantly greater degree with bubble than with membrane devices. There was one exception. C4 was decreased in concentration terms, increased in "loss per liter per minute" terms, and greatly decreased as a percentage of the preperfusion value for the membrane group in long perfusi ons . The percentage of gain or loss from expected values showed no real differences between oxygenator subsets for perfusions of 2 hours or less. These findings are in keeping with the results of other workers. Measurements of coagulation profiles showed a trend toward lower concentrations of fibrin split products in the membrane subset of the long perfusion group, although these differences were not statistically significant. Calculated losses and gains (per liter per minute) of variables were plotted versus duration of perfusion to separate linear versus nonlinear behavior. These plots demonstrated a nearly linear function for hemoglobin generation, white cell increase, and platelet decrease and exponential functions for losses of IgG, IgM, C3, and C4.

660

The Journal ot Thoracic and Cardiovascular Surgery

Clark et al.

Plasma HgB

WBC ~.....,=~"",

m!i!i!i!i!i!i!imim!!i!i!i!i!imi!imi!i!i!i!i![ei!imi!!!!!i!i!i!i!i!i!i!!!!!i!i!m: _ 5 8 3 M

~532

Hi!!!!!iHHmi!immHm!!mHemmim!!m!!!mH!m!m!!!i!!! M

Platelets

% Gain

% Loss IgM

I

15

I

I

20

10

I

I

I

I

40

60

80

100

;L.---

Fig. 1. Gains and losses during short perfusions as percent of preperfusion values. B, Bubble. M, Membrane. HgB, Hemoglobin. WBC, White blood cells.

Table III. Alterations of variables for bubble and membrane oxygenator groups for long perfusions: Data normalized for volume of blood pumped and duration of perfusion Gain ( +) or loss ( -) of variable Variable Plasma hemoglobin (mg/Llmin) WBC (X 1000/LImin) Platelets (X 10"ILl min) IgG (mg/Llmin) IgM (mg/Llmin) C3 (mg/Llmin) C4 (mg/Llmin)

Bubble

+4.18 +4.0 -25.6 -10.6 -18.7 -3.3 0.7

I

Membrane

10- 2

+ 1.35

10- 2 x 10- 4 x 10- 4 x 10- 4

-7.0 -1.2 -3.7 -0.5 -2.7

X

x 10- 4 x 10- 4 X

Postoperative interval. The concentration values at the 24 and 48 hour intervals were in or near the normal range, although some variables did not reach the lower limits. The variables out of normal range or not similar to preperfusion values 48 hours after operation were the white cell and platelet counts for all patients and the concentrations of C4 IgG, and IgM in the long perfusion group. Two important variables were different for the short perfusion versus the long perfusion group: volume of blood products transfused and the amount of mediastinal and pleural cavity drainage. Blood replacement was 2.17 ± 0.10 Land 1.63 ± 0.09 L (p < 0.05) per patient in the bubble and membrane groups, respectively. The quantities of mediastinal and

X

10- 2

+ 1.1 x 10- 4

x 10- 4 X 10- 2 x 10- 4 x 10- 4 x 10- 4

Factor

p Value

3.1 3.7 3.7 8.5 5.0 6.6 3.7

<0.01 <0.01 <0.01 <0.001 <0.001 <0.001 <0.01

pleural cavity drainage were 0.96 ± 0.09 Land 0.51 ± 0.06 L (p < 0.05) for bubble and membrane subsets, respectively, in the long perfusion set. Discussion The central hypothesis tested by this study was that a membrane oxygenator was less traumatic to blood than a bubble oxygenator. It was expected that differences would be found between bubble and membrane oxygenators with respect to platelet and protein concentrations and losses, coagulation profiles, and the quantitites of shed and transfused blood in both short and long clinical perfusions. Differences in total peripheral resistance, urinary output, and pump flows during perfusion

Volume 78

Comparison of bubble and membrane oxygenators

Number 5 November, 1979

WiWHiWHHiiiWiiiWiiiWiiiiiWiWiHiiWiiltWHWWHHiWiWm!

Plasma Hg a

%

M

66 1

:ii WWii Hi WWWH: 1480

~

713

WH 162 wac iiWiWiWiiWWWimWiWiiWimWiimimaWWWWiiimHiiiiiWii M

Loss

I

liiiiWWiiiiWWWWmWiimimWWWmiia.HiimWiiiimimWiiiimimmiiiiiiimiiWii

I

I

Platelets

M

liHWmWm:eWmiiwm IgA

I

M

liiWiHmiiiiiWiiii!HiHiiiHBHWiiiiiHWHHiHHWmHH IgG

~

% Gain

EimHiiiiiiiHHiiiHiHiiimliiiHWiilHHiiisHiiiiimiiiiiiii!iiHHiHimiiiWiiiiimm Ig M

I

M

IHWmWe:mwm c'3

I i

35

I

30

~

M 25

I

I

20

15

I

10

I

5

lM

C'4 I

o

I

I

I

I

I

20

40

60

80

100

,

Fig. 2. Gains and lossesduring long perfusions as percent of preperfusion values. For abbreviations see Fig. 1.

were unexpected. Equally unexpected were the few differences found in both raw and computed data between oxygenator subsets and especially the lack of significant differences for perfusions lasting 1 to 2 hours. The difference in mean pump flow between two oxygenator subsets was not a result of differences in heights between the right atrium and the oxygenator portal for venous blood. The height of the operating table was similar in the two oxygenator subsets. The bubble oxygenator venous portal was approximately 9 inches from the floor, and the membrane oxygenator system had a flexible reservoir bag in the venous line proximal to the venous pump which was 12 to 16 inches from the floor. Hence the membrane cases had less gravitational height. No restrictions of venous flow were made and maximal flow rates were sought in all cases. Central venous pressure was similar in both groups during perfusion (0 to 2 em H 2 0 ). The combination of statistically significantly greater pump flow and slightly but not statistically significant lower mean blood pressures in the membrane group resulted in a 47% difference in total peripheral resistance between the membrane and bubble oxygenator groups. The urinary output was increased significantly in the membrane group, this increase initially being thought to be a result of larger priming volumes. However, the rate of output of urine was sustained and each oxygenator subset received similar quantities of crystalloid and colloid solutions in each perfusion time set.

A large amount of data was collected and analyzed for four subsets of 20 patients each for three preperfusion intervals and four postperfusion intervals. One to ten blood samplings were made during perfusion. From these analyses only a few changes not evident from the raw data could be found by calculation. Platelets and three proteins decreased less and one protein decreased more when normalized for time and the volume of blood pumped with membrane oxygenators than with bubble oxygenators for long perfusions. The three immunoglobulins and two complement fractions were chosen for study because of a wide molecular weight range (150,000 to 900,000) and previous reports of significant depression lasting 5 to 7 days after CPB Y It was thought that denaturation and absorption of these proteins during CPB would provide a more sensitive index of trauma than study of the cellular components alone. Parker and associates 18 measured complement and immunoglobulin levels in serum before and after CPB. They showed changes in total complement at 4 hours, complement and IgG at 20 hours, and IgM at 44 hours after perfusion. IgA was not altered. The marked loss of C4 in the long perfusion membrane group was not in keeping with the bulk of the data. This loss may be a result of a binding specificity for this protein for both silicone rubber and expanded polytetrafluoroethylene used in the membrane oxygenators employed in this study.

The Journal of

662

Clark et al,

Most clinical perfusion studies report observations of variables expressed in concentration terms. With variability in dilution, concentration terms do not necessarily reflect a gain or loss. Gain or loss estimations require the estimation of total quantities before and after an intervention. Protein absorption data are not necessarily linear and are a function of time of exposure, temperature, concentration, surface, shape and size, surface-to-volume ratio, and flow conditions. Data were normalized for comparison of subsets to account for volume of blood pumped and duration. The estimations of gain or loss were obtained by comparing the measured concentration of a variable with that expected by dilution of the preperfusion value using the mean dilution factor based on hematocrit, hemoglobin, and total protein. Use of apostperfusion value corrected by the dilution factor for comparison to the preperfusion value incorporates a gain or loss and hence is inaccurate. Further, this maneuver overestimates a loss or underestimates a gain. The significant morbidity accompanying CPB at the outset of cardiac operations was clearly associated with the oxygenator component of the circuitry 19. 20 and generated many studies concerning protein denaturation in blood-gas interface devices. With the advent of disposable, low-prime, high-flow bubble oxygenators, the morbidity from CPB decreased and hemodilution became common practice. Dilution alone may not have been the key. It is probable that use of the new lowvolume bubble devices resulted in cessation of use of large volumes of homologous unfiltered blood and that the substitution of crystalloid for homologous blood was responsible for the decreased morbidity. Membrane oxygenators have found clinical use in organ preservation systems and in the long-term gas exchange in pediatric and adult patients with respiratory distress syndrome. These data demonstrate the efficacy of membrane devices when long perfusion times are required. 21-28 The operative experience with membrane oxygenators is growing. Although initiated in 1956 by Clowes and colleagues, 9 a decade passed before more than singular experiences were reported. Since 1970, more than 15 clinical reports have described use of membrane oxygenators for infant, pediatric, and adult patients with cardiac disease. Extended claims have been made for various membrane devices. For example, McKenzie and co-workers'" compared a group of patients perfused with a disc oxygenator which required a large volume of homologous blood to a group of patients perfused with a membrane oxygenator primed with crystalloid solution. Most of the variables ana-

Thoracic and Cardiovascular Surgery

Iyzed were significantly improved in the latter group of patients. More discrete studies by Williams and associates'" found no discernible differences with respect to shed or transfused blood, plasma hemoglobin, platelet, fibrinogen, blood urea nitrogen, or glucose concentrations in 80 patients perfused with the Travenol Teflo and Harvey No. 200 oxygenators for periods of 30 to 150 minutes (mean 96 ± 5). The results of our study are very similar. Byrick and Noble'" compared Travenol's bubble (VF 1) and membrane (Teflo) devices with respect to lung water and pulmonary dynamics in 30 patients perfused for 2 hours. The bubble group had increased pulmonary vascular resistance and lung water and the membrane group had normal values. These results were not found using the General Electric membrane oxygenator in puppies. 32 Liddicoat and colleagues" reported a study of 91 patients perfused with the VF1 and Teflo oxygenators and showed increased flow rates and mean perfusion pressures and less blood loss for those patients who had perfusion with a membrane oxygenator. Beall and associates;" reporting on 285 patients perfused with the Teflo membrane oxygenator for a mean of 71 minutes, provided no objective data to support use of the device. The Teflo membrane lung has been the subject of three reports by Karlson,3s-37 who reported use in 258 patients, one of whom was perfused for 563 minutes. For the group, platelet counts and fibrinogen concentrations at the start and end of perfusion were similar. The earlier version of the Travenol membrane lung (Modulung) has been used in 21 infants and children, II days to 6 years of age and 1.7 to 21.6 kg in weight, with satisfactory results." This artificial lung has also been used without the presence of heparin in the perfusate by Fletcher and associates'": 40 in dogs and baboons. Other reports of laboratory and clinical experience have appeared concerning the capillary, spiral coil, and small sheet membrane oxygenators. 41-49 There are distinct disadvantages to the two membrane systems used in this study. First, the priming volumes were increased because of the eight Modulung cases which used eight 0.75 m2 units in parallel. The Teflo system used two reservoir bags with a maximal volume of approximately 500 ml each. As experience was gained, the priming volume decreased to approximate that of bubble oxygenators. The membrane systems used two pumps which required a recirculation line or volume displacement transducers to maintain a constant volume between roller pumps. Closed, highflow, low-volume perfusion systems have a small margin of safety without some type of reservoir, trap,

Volume 78 Number 5 November, 1979

or arterial line filter should air enter the venous line. The membrane system used was more costly and complex and offered no advantage in the routine perfusion lasting 2 hours or less. The membrane oxygenator has distinct advantages in long perfusions. It is less damaging to the cellular elements and results in less leukocyte response. There is less removal of the two immunoglobulins (lgG and IgM) that comprise 80% of the total of these important proteins, and there is less loss of C3. The hemodynamic characteristics of the perfusion were indicative of improved tissue perfusion. Whether or not this is the result of less gaseous microemboli generation by these membrane systems remains to be proved. The decrease in shed and transfused blood products was small but can be significant over a year in terms of volume of blood products saved. Since compilation of these data, a single pump and spiral coil system have been used for reasons of simplicity. Our experience now includes small children and adults. This membrane system is currently used for difficult cases of congenital disease and those adult patients in whom long perfusion times or problems with bleeding are anticipated. We acknowledge the statistical computations of Lisa B. Wright, B.S., and Steven A. Saltz, M.D., and the secretarial assistance of Frances L. Grubbs.

2

3 4 5

6

7

8

9

REFERENCES Gibbon JH Jr: Application of a mechanical heart and lung apparatus to cardiac surgery. Minn Med 37:171, 1954 Kirklin JW, Dushane JW, Patrick RT, et al: Intracardiac surgery with the aid of a mechanical pump-oxygenator system (Gibbon type). Report of eight cases. Mayo Clin Proc 30:20 I, 1955 Bjork YO: An artificial heart or cardiopulmonary machine. Lancet 2:491, 1948 Bjork YO: Brain perfusions in dogs with artifically oxygenated blood. Acta Chir Scand 11:27, 1959 Kay EB, Cross FS: Direct vision repair of intracardiac defects utilizing a rotating disc reservoir oxygenator. Surg Gynecol Obstet 104:701, 1957 DeWall RA, Warden HE, Gott VL, Read RC, Varco RL, Lillehei CW: Total body perfusion for open cardiotomy utilizing the bubble oxygenator. J THORAC SURG 32:591, 1956 Rygg IH, Kyvsgaard E: A disposable polyethylene oxygenator system applied in a heart-lung machine. Acta Chir Scand 112:433, 1956 Kolff WJ, Effler DB, Groves LK, et al: Disposable membrane oxygenator (heart-lung machine) and its use in experimental surgery. Cleve Clin Q 23:69, 1956 Clowes GHA Jr, Hopkins AL, Neville WE: An artificial lung dependent upon diffusion of oxygen and carbon

Comparison of bubble and membrane oxygenators

663

dioxide through plastic membranes. J THORAC SURG 32:630, 1956 10 Lee WH Jr, Krurnhaar D, Fonkalsrud EW, Schjeide OA, Maloney JV Jr: Denaturation of plasma proteins as a cause of morbidity and death after intracardiac operations. Surgery 50:29, 1961 II Kolobow T, Zapol WM, Sigman RL, Pierce J: Partial cardiopulmonary bypass lasting up to seven days in alert lambs with membrane lung blood oxygenation. J THORAC CARDIOVASC SURG 60:781, 1970 12 Peirce EC II: A comparison of the Lande-Edwards, the Peirce, and the General Electric-Peirce membrane lungs. Trans Am Soc Artif Intern Organs 16:358, 1970 13 Lande AJ, Carlson RG, Patterson RH Jr, et al: Cardiac surgery with disposable membrane lungs. Trans Am Soc Artif Intern Organs 18:532, 1972 14 Clark RE, Mills M: The infant Temptrol oxygenator. J THORAC CARDIOVASC SURG 60:54, 1970 15 Clark RE, Ferguson TB, Hagen RW, et al: Experimental and clinical use of an automated perfusion system and a membrane oxygenator. Circulation 49, 50:Suppl 2:213, 1974 16 Altman PL, Dittmer DS: Blood and Other Body Fluids, Bethesda, Md., 1961, Federation of American Societies for Experimental Biology, p 493 17 Hairston P, Manes JP, Graber CD, Lee WH: Depression of immunologic surveillance by pump oxygenator perfusion. J Surg Res 9:587, 1969 18 Parker OJ, Cantrell JW, Karp RB, et aI. Changes in serum complement and immunoglobulins following cardiopulmonary bypass. Surgery 71:824, 1972 19 Anderson MN, Hambraeus G: Physiologic and biochemical responses to extracorporeal circulation. Experimental studies during four-hour perfusions. Ann Surg 153:592, 1961 20 Ashbaugh DG, Paton BC, Swan H: Prolonged partial perfusion in the dog. JAMA 179:128, 1962 21 Santiago-Delpin EA, Moberg AW, Mozes MF, Campos RA, Mason RV, Najarian JS: Comparative analysis of perfusion and nonperfusion methods for renal preservation. Surgery 72:793, 1972 22 Birnbaum D, Soyer T, Eiseman B: Studies in prolonged liver preservation. Surgery 72:772, 1972 23 Hill JD, O'Brien TG, Murray 11, et aI. Prolonged extracorporeal oxygenation for acute post-traumatic respiratory failure (shock-lung syndrome). Use of the Bramson membrane lung. New Engl J Med 286:629, 1972 24 Pois AJ, Katzman L, Weston CB, et al: Long term support with a membrane lung. Chest 62:690, 1972 25 Fallat RJ, Hill JD, Eberhart EC, et al: Treatment of acute respiratory failure with prolonged extracorporeal oxygenation. Chest 66:Suppl 38S, 1974 26 Pyle RB, Helton WC, Johnson FW, et al: Clinical use of the membrane oxygenator. Arch Surg 110:966, 1975 27 Geelhoed GW, Adkins PC, Corso PJ, Joseph WL: Clinical effects of membrane lung support for acute respiratory failure. Ann Thorac Surg 20: 177, 1975

The Journal of

664

Clark et al.

Thoracic and Cardiovascular Surgery

28 Bartlett RH, Gazzaniga AB, Fong SW, Jefferies MR, Roohk HV, Haiduc N: Extracorporeal membrane oxygenator support for cardiopulmonary failure. J THORAC CARDIOVASC SURG 73:375, 1977 29 McKenzie N, Wall W, Robert A, et al. Blood-free open heart surgery. The blood sparing effect of an atraumatic circuit and a membrane oxygenator (abstr). Circulation 51, 52:Suppl 2:73, 1975 30 Williams DR, Tyers GFO, Williams EH, et al: Similarity of clinical and laboratory results obtained with microporous Teflon membrane oxygenator and bubble-film hybrid oxygenator. Ann Thorac Surg 25:30, 1978 31 Byrick RJ, Noble WH: Postperfusion lung syndrome. Comparison of Travenol bubble and membrane oxygenators. J THORAC CARDIOVASC SURG 76:685, 1978 32 Rhodes EL, Kirsh MM, Howatt W, O'Rourke PT, Straker J, Sloan H: A comparison of pulmonary function in puppies undergoing total cardiopulmonary bypass with bubble or membrane oxygenators. J THORAC CARDIOVASC SURG 68:658, 1974 33 Liddicoat JE, Bekassy SM, Beall AC Jr, et al: Membrane VS. bubble oxygenator. Clinical comparison. Ann Surg 181:747, 1975 34 Beall AC Jr, Solis RT, Kakvan M, et al: Clinical experience with the Teflo disposable membrane oxygenator. Ann Thorac Surg 21:144, 1976 35 Karlson KE, Murphy WR, Kakvan M, Anthony P, Cooper GN Jr, Richardson PO, Galletti PM: Total cardiopulmonary bypass with a new microporous Teflon membrane oxygenator. Surgery 76:935, 1974 36 Karlson KE, Massimino RJ, Cooper GN, et al. Respiratory characteristics of a microporous membrane oxygenator. Ann Surg 185:397, 1977 37 Karlson KE, Vargas LL, Cooper GN Jr et al: Clinical evaluation of a microporous membrane oxygenator. J Cardiovasc Surg 18:71, 1977 38 Sugg WL, Fennig, JS, Platt MR: Heart surgery with a membrane oxygenator in infants. J THORAC CARDIOVASC SURG 67:593, 1974 39 Fletcher JR, McKee AE, Mills M, Snyder KC, Herman CM: Twenty-four hour membrane oxygenation in dogs without anticoagulation. Surgery 80:214, 1976 40 Fletcher JR, McKee AE, Herman CM: Membrane oxygenation in baboons without anticoagulants. J Surg Res 22:273, 1977 41 Kaye MP, Pace JB, Blatt SJ, Ferguson RJ: Use of a capillary membrane oxygenator for total cardiopulmonary bypass in calves. J Surg Res 14:58, 1973 42 Saxena NC, Hillyer P, Edmunds LH: Use of the spiral coil membrane oxygenator during open heart surgery in infants and children. J Cardiovasc Surg 18: I, 1977 43 Helton WC, Johnson FW, Howe JB, et al: Development of a practical membrane lung system. Ann Thorac Surg 26:55, 1978 44 Rawitscher RE, Dutton RC, Edmunds LH: Evaluation of hollow fiber and spiral coil membrane oxygenators de-

45

46

47

48

49

signed for cardiopulmonary bypass in infants. Circulation 47, 48:Suppl 3:105, 1973 Wright JS, Fisk GC, McCulloch CH, Torda TA, Stacey R, Hicks RG: Pediatric application of Laude-Edwards membrane oxygenator. J THORAC CARDIOVASC SURG 65:503, 1973 Siderys H, Herod GT, Halbrook H, Pittman IN, Rubush JL, Kasebaker V, Berry GR Jr: A comparison of membrane and bubble oxygenation as used in cardiopulmonary bypass in patients. J THORAC CARDIOVASC SURG 69:708, 1975 Holdefer WF, Tracy WG: The use of rated blood flow to describe the oxygenating capability of membrane lungs. Ann Thorac Surg 15:156, 1973 Murphy DA, Gillis DA, Lau H: Use of a membrane oxygenator for open heart surgery in infants. Can J Surg 19:103, 1976 Ochsner JL, Lawson NW, Mills NL, et al: Technic of deep hypothermia and circulatory arrest in the neonate and infant. South Med J 69:607, 1976

Discussion DR. GRANT V. S. PARR Hershey. Pa.

Our experience with relatively short perfusion times echoes that of Dr. Clark. In a prospective matched series of 80 patients, 40 perfusions were done with a bubble oxygenator (the Harvey 1000) and 40 were done with a membrane oxygenator with expanded tetrafluoroethylene. Bypass time, weight, and body surface area were quite similar. Patient age and operative diagnosis were also similar. We found no difference in preoperative or postoperative platelet count with either oxygenator, no difference in fibrinogen levels before or after bypass, and no difference in prothrombin time before or after bypass in the two groups. Furthermore, plasma hemoglobin levels determined 60 minutes after bypass were 5 to 10 mg/ I00 ml /hr of bypass regardless of the oxygenator employed. No patient in either group had clinical hemoglobinuria. Chest tube drainage, blood usage and predischarge hematocrit valves were nearly identical in the two groups. Clinical results were also quite similar. There were no deaths in either group. The number of days on a ventilator and the number of days in an intensive care unit requiring any form of electrocardiographic monitoring were quite similar. Thus, like Dr. Clark, we found no advantage in relatively short perfusions. Subsequent to this experience, but before the availability of a clinical left heart assist device, we attempted long-term patient support employing, in part, a membrane oxygenator. In all three such patients an increase in oxygen gradient across the membrane began after about 4 hours. In these three patients we noted a foamlike substance being extruded from the oxygenator at about 6 hours. We believe this was albumin which initially coated the membrane but was finally extruded

Volume 78 Number 5

Comparison of bubble and membrane oxygenators

665

November, 1979

but perhaps it was in fact C4 which coated the oxygenator. Dr. Clark, have you had a similar experience with a membrane oxygenator? If so, then your data and ours would indicate that if there is an advantage to a membrane oxygenator, it is to be limited to perfusion of greater than 3 hours but less than 5 to 6 hours. Thus your observation as to cost and complexity seems all the more valid. DR. L. HENRY EDMUNDS, JR. Phi/adelphia. Pa.

The data that I am going to mention have been compiled by Dr. Paul Addonizio in collaboration with Dr. Robert Colman and myself. We studied fresh heparinized human blood in an in vitro circuit with either a membrane or a bubble oxygenator. I would like to make two points regarding platelets in extracorporeal perfusion systems. My first point is that the mechanism of platelet injury in membrane and bubble oxygenators differs. On this slide [slide] is plotted on the left thromboxane B2 , which is a metabolite of thromboxane A 2 , a very potent vasoconstrictor released from platelets. On the right is plasma low-affinity platelet factor 4 (LA-PF4), which is an antiheparin factor released exclusively by platelets. In a bubble oxygenator, the release of thromboxane B 2 is progressive during the hours of perfusion. Also, release of platelet factor 4 is not affected by aspirin. Aspirin inhibits cyclooxygenase and prostaglandin synthesis and would be expected to inhibit the release of platelet factor 4 if in fact it occurred by this mechanism. In the membrane oxygenator there are two differences. first, there is a delay in the release of thromboxane B2 of at least 2 minutes, which indicates a different mechanism of release. Second, aspirin does make a significant difference in that is released. Thus the release of this the amount of LA-PF~ protein occurs by at least two mechanisms in membrane oxygenator systems. My second point is that, after brief contact with blood, extracorporeal perfusion systems become passivated, so that normally functioning platelets will not react with the extracorporeal surfaces. [Slide] This slide shows the release of LA-PF4 in the in vitro perfusion system. The dotted line shows the amount of LA-PF4 released in control membrane oxygenator systems. If prostacycline, or PGI 2 , is added as a single bolus before recirculation, the release of LA-PF4 is very small and no different from that which is released by blood incubated in a test tube at 37° C for 2 hours. Prostacycline is fully metabolized in plasma within about 8 minutes, and we have shown that the recirculating platelets function normally and are normal morphologically at 1 hour. These platelets could react with the surface if in fact the surface was the same as it was at the beginning of the perfusion, but they do not. I enjoyed Dr. Clark's paper and I wish to concur with his conclusion that the potential and theoretical advantages of membrane oxygenators have not yet been realized during short perfusions.

DR. JOE D. MORRIS Ann Arbor. Mich.

I thank the authors for their timely study and wish to contribute to the oxygenator comparison from the cost perspective. At Ann Arbor's St. Joseph Mercy Hospital, our consensus was that membrane oxygenator patients had smoother postoperative courses and required less care than bubble oxygenator patients. Clinical review, however, failed to establish any difference between the systems. For cost comparison, TMO membrane oxygenators were employed in 100 consecutive operations between November, 1978, and January, 1979. A cost analysis compared this group with 100 consecutive operations performed with the bubble system between August and November, 1978. Patient charges were obtained from the business office. The one death in the study occurred in the bubble series and was unrelated to perfusion. Perfusion times were 2 hours, 9 minutes in the bubble series and 2 hours, 11 minutes in the membrane series. For all services performed, average patient costs were lower in the membrane series than the bubble series. Reductions in costs for the membrane group were as follows: intensive care unit nursing, 7%; respiratory therapy, 16%; laboratory, 10%; pharmacy, 10%; central supply, 13%; and monitoring, 9%. The average total hospital charge for each patient in the bubble oxygenation series was $12,043 compared to $ I 1,503 per patient in the membrane series. We believe the average saving of $450 with the membrane system justifies its routine use. President Kirk/in: May I ask if your method of myocardial preservation changed during this period? Dr. Morris: No. All patients in the study were managed uniformly. DR. DELOS M. COSGROVE Cleveland, Ohio

Until recently there has been a paucity of information comparing membrane and bubble oxygenators in cardiac surgery. Like Dr. Clark, we at the Cleveland Clinic were interested in the merits of the two systems. To evaluate the clinical effectiveness of bubble and membrane oxygenator systems we undertook a randomized prospective study in 80 patients having elective revascularization. We used the Travenol Variflo bubble oxygenator in 40 and the Travenol TMO membrane oxygenator in the other 40. No statistically significant differences could be demonstrated in hematologic, renal, hepatic, cardiac, respiratory, or neurologic parameters. Similarly, the amount of chest tube drainage, the blood and colloid requirements, and the incidence of re-exploration were not statistically significantly different. Considering that a short perfusion might not stress the systems adequately to demonstrate potential differences we used the same criteria to compare a subset of 20 patients, 10

The Journal of

666

Clark et al.

Thoracic and Cardiovascular Surgery

in each group, having perfusions greater than 2 hours. A favorable trend in all parameters was noted for membrane oxygenators, but this achieved statistical significance only for blood replacement. In this subgroup the hematocrit values preoperatively, immediately postoperatively, and on the first and seventh postoperative days were similar. The blood required to maintain these hematocrit values differed markedly. A statistically greater amount of packed cells, whole blood, and total blood was transfused in the bubble oxygenator group. We believe the superiority of the membrane oxygenator can be demonstrated only in perfusions greater than 2 hours. DR. C LA RK (Closing) I thank the discussers. In answer to Dr. Parr, our longest perfusion was 316 minutes, with survival, on a membrane oxygenator. We have not seen albumin exiting from the oxy-

genator. The ECMO program used membrane oxygenators for many days without this type of complication occurring, at least to my knowledge. In response to Dr. Edmunds' concern with thromboxane and the antiheparin agent released from platelets, LA-PF4, we measured antiheparin levels in some patients. The differences between bubble and membrane groups did not achieve statistical significance but tended to be higher in the bubble group. The studies that Dr. Edmunds and his associates are doing on platelets are important and will lead to further improvement of extracorporeal perfusion devices. Dr. Morris, I did not analyze hospital costs for our patients. I know that the membrane systems used have been more expensive than bubble systems. I would point out that there was a 3-day difference in hospital stay between his two groups of patients. A 33% increase in hospitalization cannot be accounted for by use of a bubble oxygenator.

Copyright information The appearance of a code at the bottom of the first page of an original article in this journal indicates the copyright owner's consent that copies of the article may be made for personal or internal use, or for the personal or internal use of specific clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc., P.O. Box 765, Schenectady, N. Y. 12301,518-3744430, for copying beyond that permitted by Sections 107 or 108 of the U. S. Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale.