Lung transvascular fluid dynamics with extracorporeal membrane oxygenation in unanesthetized lambs

Lung transvascular fluid dynamics with extracorporeal membrane oxygenation in unanesthetized lambs Extracorporeal membrane oxygenation (ECMO) is used for long-term support of patients with acute respiratory failure. We investigated the effect of partial venoarterial (VA) and venovenous (VV) bypass with filler-free silicone spiral-coil membrane lungs on steady-state lung transcapillary fluid filtration in six unanesthetized lambs for periods of II to 32 hours. Using three thoracotomies we prepared animals to collect lung lymph; lymph obtained in this way is representative of lung interstitial fluid. By studying lymph flow and composition we demonstrated that the permeability of the pulmonary capillaries does not change during prolonged partial VVor VA bypass with a membrane lung. There was no accumulation of lung water during bypass. and lung protein and fluid leak neither increased nor decreased with bypass flows equivalent to those used clinically. Thus prolonged use of ECMO in unanesthetized lambs appears to be neither harmful nor beneficial to the steady-state dynamics offluid exchange in the lung. Furthermore. total pulmonary blood flow is not a determinant of net fluid filtration across the lung microci rculation.

Allen G. Delaney, M.D., Warren M. Zapol, M.D., and A. John Erdmann III, M.D.,* Boston, Mass.

Extracorporeal membrane oxygenation (ECMO) has been used for short periods in cardiac surgery and on a long-term basis to support gas exchange in patients with acute respiratory failure (ARF).1 It remains uncertain whether venovenous (VV) or venoarterial (V A) partial bypass is more beneficial to the injured lung. VV partial bypass provides improved gas transport but does not reduce pulmonary blood flow or relieve the increased pulmonary vascular pressures often seen in ARF. 2 On the other hand, VA bypass may alter lung trans vascular fluid dynamics by reducing both the intravascular pressure and the total pulmonary blood flow. Although some authors have postulated that deFrom the Departments of Surgery and Anesthesia of the Massachusetts General Hospital, Boston, Mass. Supported in part by grants from the Merrill Trust, the Magowan Foundation, an Adult Respiratory Failure Specialized Center of Research Grant No. HL 23591, and Grant No. HL 18646 from the National Heart, Lung, and Blood Institute. Received for publication May 9, 1978. Accepted for publication Oct. 3, 1978. Address for reprints: A. John Erdmann III, M.D., Department of Surgery, Massachusetts General Hospital, Boston, Mass. 02114. *Established Investigator of the American Heart Association.

252

creased pulmonary artery pressure may contribute to a reduction in lung water clinically, 1. 3 we still do not know whether or not VA bypass, with decreased pulmonary artery pressure and blood flow, in fact reduces excess lung water in ARF. Most patients treated with partial bypass for severe ARF have shown no signs of improvement in pulmonary function or gas exchange, no matter which mode of bypass has been employed. 1 Indeed, long-term bypass may itself lead to further lung injury by damaging the pulmonary capillaries, although partial VV bypass for 2 hours has not been shown to increase extravascular lung water if plasma colloid osmotic pressure is normal. 4 To explore these questions we studied the dynamics of lung fluid exchange using partial bypass in unanesthetized lambs chronically cannulated for collection of lung lymph. Lung lymph collected from lambs by the technique of Staub and colleagues" has been shown to be identical to lung interstitial fluid." The flow rate and protein composition of lung lymph are exquisitely sensitive to changes in lung capillary pressure and permeability. Previous studies have shown that the normal lymphatic system of the lung can drain up to tenfold the base-line flow rates before interstitial edema develops; further,

0022-5223179/020252+07$00.70/0 © 1979 The C. V. Mosby Co.

Volume 77 Number 2

Extracorporeal membrane oxygenation

25 3

February, 1979

any increase in pulmonary capillary permeability is invariably accompanied by increases in both lymph flow and protein content. 7-9 Fluid filtration from lung capillary to interstitium is described by Starling's fluid transport equation: Of = Kf [(Pcap - Pint) -(neap - Ilinn],

or

where is the net transvascular fluid flow, which presumably equals lymph flow (Qlyrn) in a steady state; Kf is the fluid filtration coefficient, which reflects both the lung capillary membrane area and its permeability; Pcap and Pint are the capillary and interstitial hydrostatic pressures; and Ilcap and Ilint are the capillary and interstitial colloid osmotic pressures, respectively. 10 In this study we examined the effect of VV and VA bypass on Kf for periods of II to 32 hours in six lambs. Methods In six Suffolk or Dover lambs weighing 20 to 30 kilograms we created chronic lung lymph fistulas and tested for their purity. 5 In brief, this method entails three thoracotomies. First, the caudal mediastinal lymph node (CMN) is divided medial to the right inferior pulmonary ligament to remove nonpulmonary lymphatic contributions. Second, catheters are placed in the left atrium for sampling and to allow balloon occlusion of the mitral orifice to increase left atrial pressure. A pneumatic occluding cuff of 20 mm. inner diameter is placed around the pulmonary artery to increase systemic venous pressure and to rule out a systemic contribution to the lung lymph. Third, the efferent duct of the CMN is cannulated through a right thoracotomy to collect lung lymph. The operations are spaced over a period of 3 weeks; approximately 6 days after the final procedure the animal is ready for testing the purity of the fistula and, if testing is successful, subsequent experimental study. During the latter stages of this project the preparation was shortened by omitting the left atrial balloon and pulmonary artery constrictor cuff: After division of the caudal mediastinal lymph node at thoracotomy and a I week recovery period, the CMN efferent duct was cannulated and a 19 gauge polyethylene left atrial line passed through a nonocclusive purse-string suture into a pulmonary vein transversing the mediastinal surface of the right middle lobe. Careful technique during cannulation of the CMN efferent duct eliminated lymph contamination by systemic sources. Therefore in the last two lambs we studied fistulas that had not been tested. For lymph cannulas we used silicone rubber tubing coated with filler-free silicone rubber, radiation cross-linked and bonded to glutaraldehyde cross-

linked heparin-methyl-tricapryl ammonium chloride. 11 Partial bypass was performed using a 1.5 sq. M. spiral-coil membrane lung, * carbon dioxide primed to remove gas bubbles, which may cause platelets to release vasoactive substances that provoke pulmonary hypertension. 19 New lungs were used for each experiment and were coated with filler-free silicone rubber. Prior to bypass, steel spring-reinforced, segmented polyurethane catheters were passed into the jugular vein, right atrium, and one carotid artery for perfusion by either the VV or VA route. After 6 to 24 hours to permit stabilization, bypass was begun with the animal breathing room air. All sheep were allowed free access to food and water but were restrained in their cages as previously described by Kolobow and colleagues. 12 At the time of bypass cannulation, a Swan-Ganz cathetert was passed through one jugular vein for cardiac output determinations by the thermodilution method and measurement of pulmonary artery pressure. The proximal (15 em.) injection port was positioned within the right ventricle to assess pulmonary blood flow. During VA bypass the systemic blood flow was assumed to be equal to the pulmonary blood flow plus the measured bypass flow. During VV bypass pulmonary and systemic blood flows are equal. Before placement of bypass cannulas we administered 300 U. of heparin per kilogram of body weight and then infused 50 to 100 U. per kilogram per hour to maintain the whole blood-activated coagulation timet at more than 400 seconds. One million units of penicillin G were administered intravenously every 4 hours during bypass. Four to seven days prior to bypass 500 ml. of autologous plasma were collected by plasmapheresis to prime the extracorporeal circuit; donor sheep blood was administered at that time to bring the hematocrit value to approximately 40 percent. The remaining 200 ml. of prime consisted of Ringer's lactate solution and 5,000 U. of heparin per liter. Priming was performed using carbon dioxide in both the gas and blood phases of the membrane lung. After initiation of VV bypass at 33 to 63 ml. per kilogram per minute, lymph flow was allowed to stabilize and was collected and measured; bypass was then switched to the VA route at 33 to 70 ml. per kilogram per minute. No blood transfusions were done during perfusion in order to avoid idiosyncratic pulmonary vascular reactions; bypass was therefore termi*Sci-Med Life Systems, Inc., Minneapolis, Minn. tEd wards Laboratories, Inc., Santa Ana, Calif. :j:Hemochron, International Technidyne, Inc., Edison, N. J.

The Journal of

254 Delaney, Zapol, Erdmann

Thoracic and Cardiovascular Surgery

I BASE , LYMPH COLLOID OSMOTIC PRESSURE (mmHg) CAPILLARY OSMOTIC PRESSURE (mmHg)

~~·E

10

40[ 30

20

I I I

I I I

I

I

I

VENU-ARTERIAL ECMO

VENO-VENOUS ECMO

:I LINE II I

I I

I I

I

I I

~-------f---------------i I

1

I

I

I

I I

I

I

CAPILLARY HYDROSTATIC PRESSURE (mmHg)

I

I

I

I

I

I

1

20

1 1

__~

I

: ~ WLL.-~- i1- - "' I

LUNG LYMPH FLOW (m//hr)

I I I

I I

,I

, I

i',

I

I

I

I

I I I I

I

, I

I I

I

10

I I

o 5020-76

2

4

6

8

10

12

14

16

18

20

HOURS

Fig. 1. Lung fluid dynamics in an unanesthetized sheep with a lung lymph fistula. Despite an initial brief elevation in capillary pressure with a concomitant rise in lymph flow at the onset of bypass, no significant steady-state changes from base-line are seen with either venovenous or venoarterial ECMO.

nated when the hematocrit value fell below 20 percent. Cervical cutdown-site hemorrhage developed in two lambs and a hind-limb hematoma in one. Transducers were calibrated against a mercury column, and all pressures were recorded in millimeters of mercury with zero reference at the level of the left atrium. The pulmonary capillary pressure, Pcap, was calculated using the formula: Pcap

= Pia + 0.4 (Ppa - Pia),

where Pia and Ppa are the mean left atrial and pulmonary artery pressures, respectively, and 0.4 is the fraction of pulmonary vascular resistance estimated to be downstream from the pulmonary capillary. 13 To measure the other variables in Starling's equation, lymph and blood were serially collected for analysis of total protein and albumin. Total protein was estimated using a refractometer and calibrated against standard protein solutions. Lymph and plasma albumin concentrations were measured by the BromcresolGreen Dye binding method. We used the regression equations of Landis and Pappenheirner'" to calculate

from the total protein and albumin fractions, the largevessel protein osmotic pressure, which we assumed would equal capillary protein osmotic pressure, [leap. From the lymph samples, which represent interstitial fluid in the lung," the interstitial protein osmotic pressure, flint, could also be calculated directly." In a steady state, the flow of lymph, Qlyrn, must equal the net trans vascular fluid filtration rate, or, in Starling's equation. Pint, the hydrostatic pressure in the interstitial space, which opposes fluid filtration, cannot be measured directly; however, on the average, during spontaneous breathing under base-line conditions, Pint is equal to the alveolar pressure, the mean of which is zero. That is, with no excess fluid in the interstitial space, and with the alveoli fully expanded, mean alveolar pressure and interstitial pressures are equal. We therefore assumed Pint to be zero in all our calculations." Extravascular lung water was measured at the end of the experiments. The Iamb was disconnected from bypass, anesthetized with intravenous methohexital (5 mg. per kilogram), intubated, and mechanically

Volume 77 Number 2 February, 1979

25 5

Extracorporeal membrane oxygenation

Table I. Mean steady-state hemodynamic and extracorporealflow data in six sheep Flows (L./min.) Ppa (mm. Hg ± S.E.M.)

Pia (mm. Hg ± S.E.M.)

14 ± 0.68 14.8 ± 0.75

3.3 ± 0.33

Base line (N = 6) Venovenous bypass (N = 9) Venoarterial bypass (N = 6) Legend:

11.7 ± 1.08

Qp

I

4.7 2.9

2.2 ± 0.82 0.1 ± 1.63

Qe 1.6 1.7

Ppa, Mean pulmonary artery pressure (Swan-Ganz catheter). Pia, Mean left atrial pressure (direct catheter).

thermistor).

Qe, Extracorporeal flow (direct measurement). Qs, Total systemic flow. Hct, Hematocrit.

I

Qs

Het (%)

Bypass time (hr.)

4.6 6.3 4.6

36 24 21

9.8 6.9

OIl,

Pulmonary blood flow (Swan-Ganz

Table II. Mean steady-state lung transvascular fluid exchange data in unanesthetized sheep on bypass

Base line (N = 6) Venovenous partial bypass (N Venoarterial partial bypass (N

= 9) = 6)

Peap (mm. Hg)

TIeap (mm. Hg)

TIint (mm. Hg)

7.6 7.4 4.8

30.9 22.7 23.4

18.2 12.0 12.1

Kf (ml./hr . . mm. Hg ± S.E.M.)

5.9 6.4 5.4

l.ll ± 0.26 0.85 ± 0.08 1.26 ± 0.49

Legend: Pcap, Mean pulmonary capillary hydrostatic pressure. Ilcap, Mean capillary colloid osmotic pressure of lung lymph. Oint, Mean interstitial colloid osmotic

pressure. Qlym, Mean lung lymph flow. Kf, Steady-state filtration coefficient of the lung capillary membrane.

ventilated with 100 percent oxygen. The thorax was opened rapidly and the lungs were excised at an airway pressure of 20 cm. H20. The lung wet-to-dry weight ratio and percentage of water were measured using the method of Pearce and associates.!" Correction for the intravascular blood water content was made using the hemoglobin concentration of central venous blood. Hemoglobin was measured by the cyanmethemoglobin method. All terminal lung water values reported are therefore extravascular. Animals were accepted for study only if rectal temperature was normal (39 0 C.), they were eating and drinking well, and they had normal arterial oxygen saturation and arterial blood gas values on room air. We made statistical comparisons using the t test, and values for p are presented. We accepted p < 0.05 as indicating a significant difference. Where summary data are shown, we report the mean ± I standard error of the mean. Results Hemodynamic and lymph data from a representative experiment are shown in Fig. I. A transient increase in Pcap and increase in lymph flow occurred in five of six lambs upon initiation of bypass and continued for 2 to 9 hours, although lymph flow always returned to baseline values. Increased Ppa occurred 5 to 10 seconds after the extracorporeal circuit prime reached the central circulation. The mean Ppa rose 30 mm. Hg and returned to base line in 20 to 420 minutes. In one lamb we reversed the protocol by initiating bypass in the VA mode but still found increases in Ppa and lymph flow

that again returned to normal. All data reported were obtained in a steady state after Ppa and lymph flow had returned to base-line conditions. Changes between bypass modes (i.e., VV to VA or vice versa) never caused pulmonary hypertension or an increase in lymph flow. Hemodynamic data during base-line periods and steady-state partial bypass are shown in Table I. Bypass times ranged from II to 32 hours; VV bypass periods ranged from 3 to 20.5 hours, and VA bypass periods ranged from 2.5 to 9.5 hours. The difference in mean Ppa from prebypass to VV bypass was not significant (p = 0.30). However, Ppa during VA bypass was significantly lower than either base line (p < 0.05) or VV bypass pressures (p s; 0.005). The derived data for solution of Starling's equation are seen in Table II. During VV bypass, when the lungs are potentially exposed directly to microaggregates embolized from the pump or oxygenator, we saw no increase in lung lymph flow. During VA bypass, although mean capillary pressure was 2.8 mm. Hg lower than base-line, we found no decrease in Qlyrn (Fig. 2). An important force opposing fluid filtration from the lung microcirculation is the capillary colloid osmotic pressure, IIcap. However, IIint is equally important, and it is the difference between IIcap and IIint that helps to favor fluid accumulation or resorption. No change in the difference between IIcap and IIint occurred when changing the mode of bypass. As expected, there was an acute drop in plasma colloid osmotic pressure at the initiation of bypass because the protein concentration of the ECMO prime solution was only 3.5 Gm. per 100 ml. Since the

256

...... ~

.

lOr-

<,

......

~

~ C:l

~

~ ~ ......

...... ...

• • •• • .i •• • •• • • •• •

8r-

64-

~

~......

The Journal of Thoracic and Cardiovascular Surgery

Delaney, Zapol, Erdmann

2--

10 -

• ,. ,. • .,.• ,. • ,. • ,. ,..

~ ..... 8-

••

,..

~

~ C:l

~

~ ~ ......

6e-

4r- •

~

......

• Baseline • VV Partial bypass • VA Partial bypass

,.

••

~

• Baseline ,. VV Partial bypass • VA Portial bypass

21-

o

2

.

4

6

8

10

12

14

NETFILTRATION PRESSURE (fl:ap - Tfap+ 17;'nI+ If)

o

2

4

6

8

10

12

LUNG CAPILLARY HYDROSTATIC PRESSURE (mmHg)

Fig. 2. Lung lymph flow in a steady state correlates poorly with capillary pressure alone, and neither VV bypass nor VA bypass changes lung lymph flow significantly.

•• • 20 -

o

,..

.,.,..' ••

•

• Baseline ,. VV Partial bypass • VA Partial bypass 2

.

,.

•••

4

,.

6

8

10

12

(mmHg)

14

14

LUNG CAPILLARY HYDROSTATIC PRESSURE (mmHg)

Fig. 3. Steady-state lung lymph colloid osmotic pressure falls with higher capillary hydrostatic pressures, as is seen in normal circumstances even without bypass. However, bypass itself does not favor fluid and protein shifts across the lung.

lymph-plasma protein ratio was 0.59 during base-line observations, 0.53 during VV bypass, and 0.52 during VA bypass, the total net protein flux from the lung was lower than base line during both modes of bypass. However, the protein leak rate was identical under all conditions when considered with reference to the capillary hydrostatic pressure (Fig. 3) and followed the same pattern seen in normal unanesthetized sheep."

Fig. 4. When net filtration pressure (the steady-state sum of all the factors goveming fluid exchange across the pulmonary microcirculation) is plotted against lymph flow (the net outward movement of fluid), a linear relationship is observed, with a nearly constant ratio depicting the fact that whatever the experimental condition, lung capillary permeability does not change.

To calculate the filtration coefficient, II mm. Hg was added to measured Ppa and Pia pressures in order to refer them to the ventral aspect of the lung." We found no significant difference between the mean Kf values during VV or VA bypass (p ~ 0.25). The constant relationship between net filtration pressure and lung lymph flow is graphically represented in Fig. 4, where the filtered fluid/filtration pressure ratio is the solution of Starling's equation and approaches a straight line (r = 0.80; P ::s 0.05). That is, Kf is close to unity, as has been found in all other normal sheep lungs similarly studied.I"!" 13 After bypass autopsies were done on five of the six lambs. The lungs were removed and their water content was measured gravimetrically. The extravascular lung water was 78.2 ± 2.3 percent. Extravascular lung water of seven normal lungs from lambs of comparable age averaged 78.1 ± 0.6 percent (p ~ 0.3). Bypass cannulas were removed from one lamb and the heparin infusion was stopped; this animal was a long-term SUrvIVOr.

Discussion

There has been extensive use of short-term extracorporeal circulation in cardiac surgery, and more recently attempts have been made to increase survival of adults and neonates with respiratory distress syndrome using prolonged partial bypass." 16 Membrane lung technol-

Volume

rr Extracorporeal membrane oxygenation

Number 2

25 7

February. 1979

ogy allows prolonged partial bypass because damage to blood components is minimal. However, extracorporeal bypass may affect lung fluid dynamics either by changing intravascular pressures and flows or by altering capillary permeability with pump-injured blood components. VV bypass increases the mixed venous oxygen content but does not reduce pulmonary blood flow, Qp, or the increased pulmonary artery pressure present in ARF. On the other hand VA bypass reduces Qp and pulmonary artery pressure but may be associated with maldistribution of oxygen in the systemic arterial circulation during partial bypass. In this study we measured directly the effects of partial bypass on lung capillary permeability. In lambs with normal lungs we compared fluid dynamics before and during prolonged VV and VA bypass. With a lung lymph fistula we can assay small changes in capillary permeability." This preparation has been thoroughly studied to ensure that the collected lymph reliably comes from the lung.?: 7 Furthermore there is corroborative evidence from direct microsampiing of the interstitial space in sheep lungs that lymph from our cannula is, in fact, interstitial fluid." In this preparation, changes in lymph flow secondary to increased vascular pressure can be distinguished from those due to altered microvascular permeability. 7-9 The filtration coefficient is an exquisitely sensitive index of vascular permeability; alterations in permeability and pressure can be seen from the data of Brigham and co-workers. H Under steady-state base-line conditions, Kf = 0.98 ml. per ern. H 20 per hour. When the lung capillary perfusion pressure is increased by inflating a balloon in the left atrium Kf does not change: Kf = 0.93 ml. per em. H 20 per hour. However, after intravenous administration of Pseudomonas bacteria, a capillary endothelial toxin, there is a tenfold increase in lymph flow and permeability, with Kf = 9.76 ml. per cm. H 20 per hour.f 9.16.17 In five of six lambs, we observed transient pulmonary hypertension at the initiation of bypass. In sheep systemic hypertension has been reported upon initiation of VV bypass with circuits primed with whole blood, IS and pulmonary hypertension has been shown." Recent evidence indicates that decreased platelets and white blood cells seen at the institution of bypass are related to priming techniques as well as the polymer surface. 20 A transient, reversible injury may be incurred, with a resultant increase in vascular permeability. However, since Starling's equation is valid only in a steady state, we cannot further elucidate this phenomenon. Nonetheless there was no evidence of persistent vascular injury, since Ppa and lymph flow returned to normal.

It has been hypothesized by Lemaire and others that VA bypass may decrease lung water in ARF because Ppa decreases. 1 Although we have not studied the injured lung, we nevertheless found extravascular lung water to be unchanged in the normal lung after prolonged bypass. In addition, we could find no decrease in steady-state lung lymph flow in our animals. Since VA bypass reduces the capillary hydrostatic pressure and the total lung blood flow, but not the lung lymph flow, it would appear that an increase in postcapiIIary resistance must occur simultaneously with the decrease in pulmonary artery pressure after instituting bypass; nonetheless, neither trans vascular fluid exchange nor extravascular lung water content change. The persistence of normal patterns of lung transvascular water movement, even with VA bypass, lends no promise to the hope that extracorporeal bypass will by itself decrease lung water. Furthermore, VA bypass reduces neither fluid filtration nor vascular permeability, although it may provide other benefits to the ARF patient, such as decreased pulmonary artery pressure with subsequent relief of right heart failure. With prolonged partial VV and VA bypass in unanesthetized lambs we could detect no alteration in transpulmonary fluid or protein exchange, no physiological injury to the lung microcirculation, and no accumulation of fluid in the interstitial space. In acute settings, such as might occur with short-term cardiopulmonary bypass with membrane lungs, there may be very transient alterations in lung fluid fluxes, but these appear to be immediately reversible in animals. However, definitive conclusions cannot be made concerning the non-steady-state conditions of standard VA cardiopulmonary bypass with bubble oxygenator. In addition, it remains to be seen whether or not the results of this study, in animals with normal lungs, will be relevant to the animal or patient with ARF.

2

3

4

5

REFERENCES Zapol WM, Quist J, eds.: Artificial lungs for Acute Respiratory Failure, New York, 1976, Academic Press, Inc. Hill JD, Fallat R, Cohn K, Eberhart R, Dontigny L, Bramson HL, Osborn 11, Gerbode F: Clinical cardiopulmonary dynamics during prolonged extracorporeal circulation for acute respiratory insufficiency. Trans Am Soc Artif Intern Organs 17:355, 1971 Zapol WM, Snider MT, Schneider RC: Extracorporeal t?embrane oxygenation for acute respiratory failure. Anesthesiology 46:272, 1977 Demling RH, Hicks RE, Edmunds LH Jr: Changes in extravascular lung water during venovenous perfusion. J THoRAc CARDIOVASC SURG 71:291, 1976 Staub NC, Bland RD, Brigham KL, Demling R, Erdmann

258

6

7

8

9

10 II

12

13

The Journal of

Delaney, Zapol. Erdmann

AJ III, Woolverton WC: Preparation of chronic lung lymph fistulas in sheep. J Surg Res 19:315, 1975 Vreim CE, Snashall PD, Demling RH, Staub NC: Lung lymph and free interstitial fluid protein composition in sheep with edema. Am J Physiol 230:1650, 1976 Erdmann AJ III, Vaughan TR Jr, Brigham KL, Woolverton WC, Staub NC: Effect of increased vascular pressure on lung fluid balance in unanesthetized sheep. Circ Res 37:271, 1975 Brigham KJ, Woolverton WC, Blake LH, Staub NC: Increased sheep lung vascular permeability caused by pseudomonas bacteremia. J Clin Invest 54:792, 1974 Brigham KL, Owen PJ, Bowers RE: Increased permeability of sheep lung vessels to proteins after pseudomonas bacteremia. Microvasc Res 11:415, 1976 Staub NC: "State of the art" review. Pathogenesis of pulmonary edema. Am Rev Respir Dis 109:358, 1974 Lagergren HR, Eriksson JC: Plastics with a stable surface monolayer of cross-linked heparin. Preparation and evaluation. Trans Am Soc Artif Intern Organs 17:10, 1971 Kolobow T, Spragg RG, Pierce JE, Zapol WM: Extended term (to 16 days) partial extracorporeal blood gas exchange with the spiral membrane lung in unanesthetized lambs. Trans Am Soc Artif Intern Organs 17:350, 1971 Staub NC: Steady state pulmonary trans vascular water

Thoracic and Cardiovascular Surgery

14

15 16

17 18

19

20

filtration in unanesthetized sheep. Circ Res 28:Suppl I: 1-135, 1971 Landis EM, Pappenheimer JR: Circulation. Exchange of substances through the capillary walls, Handbook of Physiology, vol 2, sec 2, WF Hamilton, P Dow, eds., Washington, D. C., 1963, American Physiological Society , Publisher, p 961 Pearce ML, Yamashita J, Beazell J: Measurement of pulmonary edema. Circ Res 16:482, 1965 Bartlett RH, Gazzaniga AB, Jefferies MR, Huxtable RF, Haiduc NJ, Fong SW: Extracorporeal membrane oxygenation (ECMO) cardiopulmonary support in infancy. Trans Am Soc Artif Intern Organs 22:80, 1976 Brigham KL, Owen PJ: Increased sheep lung vascular permeability caused by histamine. Circ Res 37:647, 1975 Spragg RG, Hill RN, Wedel MK, Masterson A, Moser KM: Platelet kinetics in venovenous membrane oxygenation. Trans Am Soc Artif Intern Organs 21:171, 1975 Birek A, Duffin J, Glynn MF, Cooper JD: The effect of sulfinpyrazone on platelet and pulmonary responses to onset of membrane oxygenator perfusion. Trans Am Soc Artif Intern Organs 22:94, 1976 Osada H, Duffin J, Nelens JM, Ward C, Cooper JD: Platelet loss with silicone rubber silica-free coated and vacuum primed membrane oxygenators. Trans Am Soc Artif Intern Organs 6:66, 1977