Venovenous compares favorably with venoarterial access for extracorporeal membrane oxygenation in neonatal respiratory failure

Venovenous compares favorably with venoarterial access for extracorporeal membrane oxygenation in neonatal respiratory failure Traditional extracorporeal membrane oxygenation via the venoarterial route requires cannulation and ligation of the internal jugular vein and common carotid artery. Concerns about ligation of the common carotid artery prompted development of a 14F double-lumen internal jugular vein cannula for venovenous oxygenation for neonates with respiratory failure. We retrospectively compared 22 patients supported by venovenous bypass and 20 patients supported with traditional venoarterial bypass. The two groups of patients were selected to be comparable in terms of diagnosis and severity of respiratory insufficiency. The diagnoses in both groups were limited to meconium aspiration syndrome or persistent pulmonary hypertension of the newborn. The average oxygenation indexes in the two groups were similar (46.6 venovenous, 47.2 venoarterial, p = not significant). Venovenous access allowed flow rates of more than 100 m1jkg per minute, which were adequate for gas exchange support. One patient required conversion from venovenous to venoarterial bypass because of hemodynamic instability. The average time of bypass support was 115 hours (range 24 to 338 hours) for venovenous bypass and 134 hours (range 47 to 361 hours) for venoarterial bypass (p < 0.05). The time to extubation after decannulation from extracorporeal membrane oxygenation was 133 hours (range 38 to 720 hours) for venovenous support and 100 hours (range 27 to 192 hours) for venoarterial support (p = not significant). One patient supported with venoarterial bypass had an intracranial hemorrhage. There were no documented neurologic injuries in the patients managed with venovenous bypass. There were no deaths in either group. Venovenous extracorporeal membrane oxygenation through a double-lumen cannula can adequately provide respiratory support for neonates with pulmonary failure and effectively avoids ligation of the common carotid artery. (J THoRAc CARDIOVASC SURe 1993;106:329-38)

Ralph Delius, MDa (by invitation), Harry Anderson III, MDa (by invitation), Robert Schumacher, MDb (by invitation), Michael Shapiro, MDa (by invitation), Tetsuro Otsu, MDa (by invitation), Kenneth Toft, BSa (by invitation), Jennifer Hirsch, Bsa (by invitation), and Robert Bartlett, MD,a Ann Arbor, Mich.

Extracorporeal membrane oxygenation (ECMO) has become an increasingly accepted technique for the supFrom the Departments of Surgery" and Pediatrics," University of Michigan Medical Center, Ann Arbor, Mich. Supported in part by grants from the National Institutes of Health. The University of Michigan Extracorporeal Circulation Laboratory receives a consulting fee from Kendall Health Care Products, Mansfield, Mass., the manufacturer of the double-lumen cannula. Read at the Seventy-second Annual Meeting of The American Association for Thoracic Surgery, Los Angeles, Calif., April 26-29, 1993. Addressfor reprints: Robert H. Bartlett, MD, University of Michigan Medical Center, 2920 Taubman Center, Box 0331, 1500 East Medical Center Drive, Ann Arbor, MI 48109-0331. Copyright © 1993 by Mosby-Year Book, Inc. 0022-5223/93 $1.00

+ .10

12/6/46213

port of neonates with life-threatening respiratory failure. I Conventional ECMO entails drainage of venous blood from the right atrium through a cannula placed in the right internal jugular vein and return of oxygenated blood via a cannula placed in the right common carotid artery. Concerns have been raised, however, about the potential complications of the technique, especially the need to ligate the right common carotid artery? The prevalence of neurologic complications in patients supported with ECMO appear to be similar to that in comparably ill neonates receiving standard medical treatment, although when unilateral hemispheric cerebral injury is seen, it is predominantly localized to the right hemisphere.l-" Although the ligation of this artery seems to be tolerated well by most patients," the long-term effects of common

329

330

The Journal of Thoracic and Cardiovascular Surgery August 1993

Delius et al.

nates supported with conventional venoarterial (VA)

ECMO.

Methods Patient population. Twenty-two neonates supported with VV ECMO were retrospectively compared with 20 patients treated with VA ECMO. Data for this report were gathered from hospital and ECMO records. Patients were selected for bypass support when they failed to respond to conventional mechanical ventilation and pharmacologic therapy. The oxygenation index was used to more clearly define the neonates at highest risk for death. The oxygenation index (01) was computed by the following forrnula'': . Fi0 2 X 100 01 = Mean airway pressure X --,:p:-----a02

Fig. 1. Diagram of double-lumen cannula. Note the eccentrically placed septum, allowing for a large drainage lumen and a smaller return lumen.

carotid ligation are unknown. Venovenous (VV) bypass, in which oxygenated blood is reinfused through a vein, can provide respiratory support and eliminate the need for ligation of the common carotid artery.' VV bypass has been described in infants in whom the oxygenated blood was reinfused through the femoral vein." Although this method provided adequate oxygenation for most patients, significant morbidity from the groin dissection and femoral vein ligation, as well as the extra time involved in the dissection, discouraged widespread acceptance of this technique. A double-lumen cannula that is inserted into the internal jugular vein has recently been developed. 7 This cannula has a larger lumen for venous drainage and a smaller lumen for reinfusion of oxygenated blood. A preliminary report describing the use of this cannula in a small group of neonatal patients was encouraging," Use of this cannula avoids the problems of femoral vein cannulation and common carotid artery ligation. The purpose of this report was to compare the physiology and short-term outcome of a group of neonates with simple respiratory failure supported by VV ECMO with the double-lumen cannula with a comparable group of neo-

where Fi0 2 is inspired oxygen concentration and Pa02 is arterial oxygen tension. In our neonatal intensive care unit an oxygenation index greater than 25 defines a 50% risk of death and an oxygenation index consistently over 40 represents an 80% risk of death.!? All patients included in this report had an oxygenation index greater than 25 before initiation of bypass support; three VV-supported patients and two patients receiving VA bypass had an oxygenation index less than 40. The diagnoses were limited to meconium aspiration or idiopathic persistent pulmonary hypertension of the newborn, because other conditions commonly requiring ECMO, such as congenital diaphragmatic hernia or neonatal sepsis, have factors other than respiratory failure (i.e., pulmonary hypoplasia, renal failure) that affect the physiology and clinical outcome. The patients in both study groups were carefully matched in regard to hemodynamic status, severityof respiratory failure, and lack of other complicating features such as prematurity or congenital abnormalities. In most of the VA-supported patients ECMO was begun during the 6 months immediately before the introduction of the double-lumen cannula; in a few patients VA bypass was begun after the introduction of the double-lumen cannula because of the unfamiliarity of some of the attending staff with the use of this cannula shortly after it was introduced. All patients included in this report would currently be considered for VV bypass. Description of thedouble-lumen cannula. A diagram ofthe cannula is shown in Fig. I. The cannula (marketed by Kendall Health Care Products, Mansfield, Mass.) was constructed of thin-walled polyurethane. The septum separating the drainage and infusion lines is eccentrically placed, so that it creates a larger lumen for drainage and a smaller lumen for reinfusionof oxygenated blood. The holes for the reinfusion lumen are located on the side of the cannula; when the cannula is properly placed in the right atrium, a significant portion of the perfusate is directed through the tricuspid valve. This helps minimize recirculation. The drainage lumen is able to maintain flowrates of 400 to 500 nil/min." 8 Preoxygenator pressures are approximately 60 mm Hgj I00 ml bypass flow when used in a clinical setting," The bypass circuit used for VV bypass is identical to that used for VA bypass with the exception that a slightly larger membrane oxygenator (0.8 m 2 versus 0.4 or 0.6 m'; SciMed Life Systems, Inc., Minneapolis, Minn.) is used to allow greater oxygen and carbon dioxide transfer. Cannulation technique. The cannulation technique for VA ECMO has been described previously and will not be discussed

The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2

Delius et al.

331

80

70

60

MEAN ARTERIAL PRESSURE (mmHgl

50

~

K

~

~

P

>--

--

~~

.....,

--

Io&.a.

--

-" <, ~

--..,,.._

vv VA

40

30

20

o

24

48

72

96

120

144

TIME ON BYPASS (Hour.)

Fig. 2. Mean arterial pressure during ECMO course.

here.? The cannulation technique for VV ECMO is similar in many respects to VA cannulation. The neck of the neonate is extended by a roll placed beneath the shoulders. Local anesthesia is used and the initial dissection is identical to that performed for VA bypass. Even though the common carotid artery is not usedfor VV bypass, the artery is dissected out and encircled with a vesselloop before heparinization in case VA bypass is required because of cardiac arrest during cannulation or inadequate support by VV bypass. Heparin (100 U /kg) and a paralyzing agent (succinylcholine or pancuronium) are given after the dissection is completed but before cannulation. A 2-0 silk marking ligature is placed around the cannula 5.5 to 6 ern from the tip. Special care has to be taken when securing this or any subsequent ligature around the cannula, because the very thin wallsare easily compressed by an excessivelytightened ligature. The connecting tubing is placed on the cannula at this point. The tubing for the reinfusion lumen must have a Luer-Lok device with an attached stopcock. Any inotropic agents should be infused through intravenous tubing into the lumen of the connecting tubing via the stopcock before the start of bypass. Early clinical use of the double-lumen cannula was sometimes complicated by a brief period (3 to 4 minutes) of hypotension, which occurred immediately after VV bypass was begun. Hypotension was occasionally severe enough to require chest compressions because of the pressor agents (usually infused through an umbilicalvenous catheter) being drawn into the drainage lumen of the double-lumen cannula, leaving the patient without inotropic support during the time blood from the patient circulated through the bypass circuit. Infusing inotropic agents directly into the reinfusion cannula avoided this brief period of hypotension altogether. To minimize recirculation, it is imperative to ensure that the reinfusion port is placed anteriorly. Failure to do so results in nearly complete recirculation. Once the cannula is in place, it is secured and the skin closed. Cannula placement is confirmed by chest roentgenogram. Treatment of patients. Treatment of patients in the VA and VV groups was similar after the initial stabilization period.

Bypass flowswere initiated at 100 ml/kg per minute for both the VA and VV groups. The bypass flows were titrated to keep arterial saturations over 90%. Activated clotting times were checked every hour and maintained between 180 and 200 seconds. The platelet count was kept over 100,000/mm 3 . The ventilator was weaned to rest settings (usually inspired oxygen fraction 30%, airway pressures 20/4 mm Hg, rate 10 breaths/ min), with the goal of maintaining an arterial oxygen saturation greater than 90%. Fluid status was carefully monitored, because bypass flow rates are dependent on the amount of blood drained from the right atrium, which in turn is determined by the volume status. Colloid solutions were given as necessary to maintain the bypass flow rates needed for adequate oxygenation. Dopamine or dobutamine (or both) was used as needed to maintain a mean arterial pressure greater than 40 mm Hg. Patients underwent cranial ultrasonography before bypass support was begun. A follow-up ultrasonogram was obtained during bypass, usually on the third or fourth bypass day, and also before discharge from the hospital. The procedure for weaning from bypass differed between the VA- and VV-supported patients. Once bypass flow rates were reduced to approximately 50 ml/rnin, the VA-supported patients underwent a trial off bypass by unclamping of the bridge and clamping of the venous and arterial lines above the bridge. This allowed blood to flow through the bridge during the trial off VA bypass. For trials off VV bypass the oxygen flow to the membrane oxygenator was shut off; the bridge remained clamped and blood flow was continued through the circuit and cannula. Cannulas for the VA and VV groups were removed when arterial saturations were maintained above 92% while off bypass with the following ventilator settings: inspired oxygen fraction 30% to 40%, rate less than 40 breaths/min, and peak inspiratory pressure less than 30 Cm H 20 . Recirculation measurements. Recirculation was determined in nine patients supported by VV bypass. The technique for estimating recirculation has been described earlier'' but will be reviewed here because understanding the concept of recircu-

The Journal of Thoracic arid Cardiovascular Surgery August 1993

3 3 2 Delius et al.

Fig. 3. Inotropic requirements during ECMO course. A, Dobutamine. B, Dopamine. (*p

lation is imperative for understanding the physiology of VV bypass. Recirculation measurements were performed immediately before decannulation from VV bypass. Recirculation was estimated by using a fiberoptic catheter (Abbott Critical Care Systems, Abbott Laboratories, North Chicago, III.) to monitor the oxygen saturation of the blood drained from the right atrium into the bypass circuit. This was performed at several flow rates when no oxygen was flowing through the membrane lung. Since no gas exchange across the membrane lung occurs under these circumstances, the bypass circuit serves as a VV shunt, allowing continuous measurement of the oxygen saturation of the blood being drained from the right atrium. Next, the flow

< 0.05.)

of 100%oxygen was restored to the membrane oxygenator, and the increase in oxygen saturation of blood draining from the right atrium was measured at several bypass flow rates. Recirculation was calculated at each bypass flow rate by the following equation: Recirculation (%)

=

SQoz - Svo, I S X 100 -

vOz

where SQoz is venous saturation with membrane lung oxygen on and Svo , is venous saturation at the same bypass flow with the oxygen off.

The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2

Delius et al.

333

50

40

30 PULSE PRESSURE

., ..-

(mmHg)

20

"

10

-- -- ;....- .....-

.,"

~V

V

., .,.,

"

/1 ----f

1

V

vv VA

O+-----,..--.,...-..-----'T-.........-r---.--,..-----,..----r-.,....----,

o

24

48

72

96

120

144

TIME ON BYPASS (Hours)

Fig. 4. Differences in pulsepressureduring ECMO course for VA and VV bypass. Note that differences are significant onlyduringfirst36 hoursof bypass support,althoughVVbypass appearsto havea consistently greater pulse pressure during the entire period of bypass support. (*p < 0.05.) Statistics. Alldata are expressed as mean ± standard deviationunless otherwise stated.Comparisons between VVand VA groups at each time pointduring bypass supportweredone by the paired t test. Differences in outcomeand patient characteristics werecompared by Student's t test. Results Patient population. There were 22 patients in the VV group and 20 in the VA group. Meconium aspiration syndrome was the diagnosis in 16 patients in the VA group and 21 in the VV group; the rest of the patients, 4 in the VA group and I in the VV group, had persistent pulmonary hypertension of the newborn. There were 9 male and II female infants in the VA group and II male and II female infants in the VV group. The average weights were 3.7 ± 0.6 kg and 3.9 ± 0.5 kg for patients in the VA and VV groups, respectively. All patients were term infants, and bypass support was begun within the first 72 hours of life. The average time on ECMO for VA-supported patients was 134 hours (range 47 to 361 hours) and liS hours (range 24 to 338 hours) for VV-supported patients (p = NS*). The time to extubation after decannulation from ECMO was 133 hours (range 38 to 720 hours) for VV bypass and 100 hours (range 27 to 192 hours) for VA support (p = NS). The average oxygenation index was 47 ± 12 for the VV group and 47 ± 13 for the VA group (p = NS). All patients *NS = Not significant.

received an echocardiogram to rule out congenital heart disease. Signs of pulmonary hypertension were seen on the echocardiogram in all patients, but no structural defects except for patent ductus arteriosus or patent foramen ovale (or both) were noted. All patients were receiving dopamine (5 to 10 fJ.g/kg per minute) or dobutamine (5 to 15 fJ.g/kg per minute) (or both) before bypass. One patient in this study group required conversion from VV to VA bypass within the first 2 hours of bypass support because of persistent hypotension (mean arterial pressure less than 35 mm Hg) that did not improve with dopamine, dobutamine, and fluid administration. This patient was ultimately weaned from bypass and discharged home. Hemodynamics. The mean arterial pressure was comparable in the VV and VA groups during the entire time bypass support was used (Fig. 2). Early in the bypass course patients in the VV group required greater inotropic support to maintain a mean arterial blood pressure greater than 40 mm Hg, but by 48 hours the pressor requirements were similar (Fig. 3, A and B). As expected, the pulse pressure in the VV group was significantly greater than in the VA group during the first 36 hours, and it remained greater than the pulse pressure in the VA group during the entire course, although the differences were not statistically significant after 36 hours (Fig. 4). Gas exchange. The arterial oxygen tension was comparable in the VA and VV groups except immediately

The Journal of Thoracic and Cardiovascular Surgery August 1993

3 3 4 Delius et al.

400

350

300

250

ARTERIAL p02 (mmHg)

200

\ r-,

150

100

"

"

---9--

_____

r-, --

~

-,

....--' V "

-- --

'-

VV VA

~0

50

o

o

24

48

96

72

144

120

TIME ON BYPASS (Hours)

Fig. 5. No significant differences in oxygen tension (POl) were noted between VV and VA bypass.

100 90 80 70 60 RECIRCULATION (%)

50 40 30 20

10

O+----.-..,...-....--r-.......--,.-........--.-.......-,--.........----, o

100

200

BYPASS

300

PUMP

FLOW

400

500

600

(ml/mlnute)

Fig. 6. Recirculation at various ECMO bypassflow rates. Recirculation increases as bypass flow rate increases, leading to less efficiency at high bypassflow rates. after initiation of bypass (Fig. 5), during which the oxygen tension in the V A group was significantly greater than that in the VV group (202 ± 160 mm Hg versus 105 ± 90 mm Hg, p < 0.05). The oxygen tension was maintained at 80 to 100 mm Hg in most patients. The carbon dioxide tension was comparable during the entire ECMO course for the VV and VA groups. The inspired

oxygen fraction was also virtually identical in the two groups during the time of ECMO support. Oxygen delivery. Recirculation describes the amount of blood that has already been oxygenated by the ECMO circuit and is reaspirated into the drainage lumen of the double-lumen cannula. Fig. 6 shows that recirculation increases as the bypass flow rate increases. Recirculation

The Journal of Thoracic and Cardiovascular Surgery Volume 106. Number 2

Delius et al.

335

100

90

..............

i\.

80

f '\

MIXED VENOUS OXYGEN SATURATION (%)

V

--__

vv VA

70

......

60 +--~

o

-~---,;--~-...,....-~~-r-~---';--~--'

24

A

48

72

96

120

144

TIME ON BYPASS (Hours)

130 120

110

BYPASS PUMP FLOW (ml/kg/mln)

---e-_

VV VA

40 30 +-~~-r-~---,;--~-...,....-~~......-~---,;--"""'--,

o

B

24

48

72

96

120

144

TIME ON BYPASS (Hours)

Fig. 7. A, Mixedvenous oxygensaturation during bypasssupport. VVbypassresultsin significantly greater mixed venous oxygen saturation levels during the first 60 hours of support because of recirculation. B, During the first 36 hours of bypass the flow rates required for VV bypass are significantly higher than that required for VA bypass. (*p < 0.05.)

has two potential effects: (I) an increase in oxygen saturation of blood drained from the right atrium because of aspiration of oxygenated blood and (2) higher bypass flow rates needed to maintain oxygenation. The effects on venous oxygen saturation can be seen in Fig. 7, A. The right atrial oxygen saturation is significantly higher in VV-supported patients during the initial 60 hours of

bypass. Bypass flow rates were also significantly higher in VV-supported patients during the first 48 hours of bypass support (Fig. 7, B). As native lung function improved, bypass flow rates were decreased. Recirculation diminished as bypass flow rates were decreased, leading to right atrial oxygen saturations comparable with those seen with VA bypass. The amount of oxygen transported per min-

The Journal of Thoracic and Cardiovascular Surgery August 1993

3 3 6 Delius et ai.

8

6

5

OXYGEN TRANSPORT (ml/kg/mln)

4

-- ......-

VV

-

VA

3

2

O+--~--""T""--.,.......-~---r---.,.......----,

o

2

3

4

5

TIME ON BYPASS (Days)

Fig, 8. The amountofoxygen transported across the membrane lungwasnearlyidentical for VV and VA ECMO.

ute across the membrane lung represents the volume of oxygen supplied to the patient per minute. Total oxygen transport across the membrane lung was identical regardless of whether VA or VV bypass was used (Fig. 8). Renal function. Serum blood urea nitrogen and creatinine concentrations were not significantly different in the VA and VV groups. The weights of VV- and VA-supported patients peaked during the third day of bypass support, followed by a gradual decrease during the ensuing 3 to 4 days. There were no significant differences in weights between the two groups, nor was there any difference in the amount of fluids administered during bypass support. Anticoagulation. The desired goals for anticoagulation and platelet counts were identical for both the VV and VA groups. There was no difference between the two groups in the amount of platelets transfused or the amount of heparin required. Mechanical factors. Preoxygenator line pressures were significantly greater in the VV group than in the VA group (235 ± 61 mm Hg versus 189 ± 39 mm Hg, p < 0.05). This discrepancy was anticipated, because the outflow lumen of the double-lumen cannula is smaller than the 10F arterial cannula, which was used for all patients supported by VA bypass. These higher line pressures did not result in any circuit complications, nor were serum hemoglobin levels elevated when compared with those in the VA group (26 ± 12 mgjdl for the VV group versus 30 ± 26 mgjdl for the VA group, p = NS).

Complications. There was one case of grade I intracranial hemorrhage in the VA group; no cases of intracranial hemorrhage were documented in patients supported with VV bypass. Mechanical complications in the VA group included two heat exchanger malfunctions. Two patients in the VA group and three in the VV group required exploration of the cannulation site for bleeding. There were no long-term sequelae from the wound reexplorations. Early in the series three patients supported with VV bypass required repositioning of the cannula because of excessive recirculation. This problem was eliminated once we realized that correct positioning (i.e., perfusion holes directed toward the tricuspid valve) is necessary to avoid excessive recirculation. One patient initially placed on VV bypass was converted to VA bypass 12 hours after bypass was initiated because of persistent hypotension that did not respond to pressor support. This patient was weaned from bypass support and did not appear to have any sequelae resulting from delayed initiation of VA ECMO. There were no instances of superior vena cava thrombosis or obstruction resulting from the use of the double-lumen cannula. Discussion Although VA ECMO has been successful in supporting neonates with respiratory failure, concerns have been raised regarding the safety of the procedure, in particular, the effects ofligating the common carotid artery. The long-term consequences of common carotid ligation are

The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2

uncertain at this time; some reports reveal no significant adverse effects from carotid ligation, whereas other studies have discovered subtle neurologic findings that seem to correlate with common carotid artery ligation.': 4 VV ECMO avoids this problem, because arterial access is unnecessary. VV ECMO, as described in neonates.o 6 involvesa circuit in which oxygenated blood was reinfused through a cannula placed in the femoral vein. Although most patients were satisfactorily supported with femoral VV ECMO, the procedure was more complex and time-consuming than VA ECMO because of the extra dissection involved, More important was the significant morbidity from dissection of the femoral vein. It was thought that the disadvantages of femoral VV ECMO outweighed the advantages." Prior animal experiments had demonstrated that VV ECMO was feasible and was capable of supporting respiratory function. I I One approach to VV ECMO without femoral vein dissection has been the single-lumen tidal-flow system reported by Kolobow, I 2 Durandy, I 3 and their associates. With this system, blood is alternatively drained and reinfused into the right atrium, controlled by a pair of valves, Tidal-flow VV ECMO has been used effectively in Hospital Trousseau in Paris. An alternative technique for VV ECMO is the use of a double-lumen cannula that can be placed easily in the jugular vein." The feasibility of clinical use of this cannula has been reported in an earlier report. 8 The aim of this retrospective comparison was to compare the ECMO course of patients supported with traditional VA bypass with that of patients supported with VV ECMO using the double-lumen cannula. Because VV ECMO does not provide circulatory support, hemodynamic stability must be assured before bypass. Patients who are in hemodynamically unstable condition at the time of cannulation should be placed on VA bypass. However, moderate pressor requirements are not a contraindication to VV bypass. Inotropic agents can usually be withdrawn as pulmonary hypertension resolves and ventilation settings are reduced. The increased pulse pressure seen in patients in the VV group was anticipated, because systemic blood flow is entirely dependent on native heart function. In VA bypass the oxygenated blood is returned to the arterial circulation by a nonpulsatile pump. As greater amounts of blood are diverted through the bypass circuit, arterial flow is increasingly due to flow through the ECMO circuit, resulting in dampening of the pulse pressure. As VA bypass is weaned, the circulation becomes more dependent on native heart function and pulse pressure increases. No grossly detectable physiologic effects on renal

Delius et al.

337

function resulted from the dampened pulse pressure of VA bypass. Oxygenation was easily maintained in these patients with VV bypass. No patients included in this study required conversion to VA bypass because of hypoxemia, The ventilator settings used during bypass support were virtually identical in the two groups of patients. Higher bypass flows were required with VV bypass because of recirculation, but adequate oxygenation was still feasible with a 14F double-lumen cannula for all patients in this report. At the time of this writing the cannula is available only in the l4F size; 10 and 12F sizes have been tested and are awaiting Food and Drug Administration approv-

al. The elevation in right atrial oxygen saturation in the VV group can be attributed to recirculation. Although the absolute values of venous oxygen saturation are "spuriously" elevated as a result of recirculation, it is still useful to follow the trend in right atrial oxygen saturation, provided that other factors (i.e, bypass flow, hematocrit value, ventilator settings) are not altered. The increased bypass flow rates and higher circuit pressures seen in the VV group did not appear to have any adverse effects. The higher line pressures were due to the smaller return lumen of the double-lumen cannula and could potentially cause increased hemolysis. However, hemolysis was not increased in patients supported with VV bypass, There were no statistically significant differences in neurologic complications between the VA and VV bypass groups, although the only intracerebral hemorrhage occurred in a patient supported by VA bypass. Internal jugular vein ligation could potentially contribute to neurologic injury by increasing cerebral venous pressure, although this remains unproved. Larger studies and long-term follow-up of patients supported with VV bypass will be needed to establish if VV ECMO offers any neurologic benefits over VA support. VV ECMO with a double-lumen cannula can effectively support hemodynamically stable neonates with respiratory failure. Bypass flows must be higher than those needed for VA bypass, but no ill effects are seen from the increased flow rate. The l4F double-lumen cannula can provide 450 to 500 ml/rnin flow rates and can be used to support infants as large as 4 kg. In addition to increased flow rates, differences in right atrial oxygen saturation and pulse pressure are also seen in VV bypass. In most other respects the clinical physiology of VV bypass in hemodynamically stable patients is similar to that seen in patients receiving VA support. Because the long-term effects of common carotid artery ligation are

338

The Journal of Thoracic and Cardiovascular Surgery August 1993

Delius et al.

unknown, it may be preferable to use VV bypass with a double-lumen cannula whenever feasible. Thanks to Sandy Snedecor, MS, for help with statistical analysis. REFERENCES 1. Stolar CJ, Snedecor SM, Bartlett RH. Extracorporeal membrane oxygenation and neonatal respiratory failure: experience from the Extracorporeal Life Support Organization. J Pediatr Surg 1991;26:563-71. 2. Elliott SJ. Neonatal extracorporeal membrane oxygenation: how not to assess novel technologies. Lancet 1991; 337:476-8. 3. Schumacher RE, Barks JD, Johnston MV, et al. Rightsided brain lesions in infants following extracorporeal membrane oxygenation. Pediatrics 1988;82:155-61. 4. Glass P, Miller M, Short B. Morbidity for survivors of extracorporeal membrane oxygenation: neurodevelopmental outcome at I year of age. Pediatrics 1989;83:72-8. 5. Andrews AF, Klein MD, Toomasian JM, Roloff OW, Bartlett RH. Venovenous extracorporeal membrane oxygenation in neonates with respiratory failure. J Pediatr Surg 1983;18:339-46. 6. Klein MD, Andrew AF, Wesley JR, et al. Venovenous perfusion in ECMO for newborn respiratory insufficiency. Ann Surg 1985;210:520-6. 7. Otsu T, Merz SI, Hultquist KA, et al. Laboratoryevaluation of a double lumen catheter for venovenous neonatal ECMO. ASAIO Trans 1989;35:647-50. 8. Anderson HL, Otsu T, Chapman RA, Bartlett RH. Venovenous extracorporeal life support in neonates using a double lumen catheter. ASAIO Trans 1989;35:650-3. 9. Hirschi RB, Bartlett RH. Extracorporeal membrane oxygenation (ECMO) support in cardiorespiratory failure. In: Tompkins R, et ai, eds. Advances in surgery. Chicago: Year Book,1987:189-211. 10. Bartlett RH. Extracorporeallife support for cardiopulmonary failure. Curr Probl Surg 1990:27:627-705. II. Zwischenberger JB, Toomasian JM, Drake K, Andrews AF, Kolobow T, Bartlett RH. Total respiratory support with single cannula venovenous ECMO: double lumen continuous flow vs single lumen tidal flow. ASAIO Trans 1985;31:610-5. 12. Kolobow T, Borell M, Spatola R, et al. Single catheter venovenous membrane lung bypass in the treatment of experimental ARDS. ASAIO Trans 1988;34:35-8.

13. Durandy Y, Chevalier JY, LeCompte Y. Venovenous extracorporeallung support: initial experience in pediatric patients. In: Gille JP, ed. Neonatal and adult respiratory failure: mechanisms and treatment. Paris: Elsevier, 1989: 159-72.

Discussion Dr. L. Henry Edmunds (Philadelphia, Pa.). The authors deserve our thanks for incrementally increasing both the simplicity and safety of ECMO for the therapy of respiratory distress in infants. Dr. Bartlett, as you know, is both the parent and grandparent of ECMO for neonates and has taken a group of diseases with nearly 100% mortality and reduced that mortality to about 6%. That is quite an achievement. The VV technique is interesting and the double-lumen catheter is very innovative. Most interesting is the fact that the technique provides the same oxygen delivery as the VA technique despite the recirculation of what I calculated to be about 60% to 70% of the flow at the flow rates that are required. Arterial carbon dioxide tension remains stable but is not usually a problem in respira tory insufficiency.The successof the VV technique with a high percentage of recirculation is due to the fact that oxygen is added to the steep part of the oxygen satura tion curve and that mixed arterial and venous blood still has a great deal of capacity to add oxygen. I think the technique has wider application, and perhaps you will comment. It should be applicable to adults and should compete with the IVOX system of adding oxygen directly to the inferior vena cava (W.T. Farley, Inc., Simi Valley, Calif.). The single neck cannula is a big advantage, because an adult can walk around and not be bedridden. Do you envision the eventual development of a long-term oxygenating device based on your new VV technology to be used with a compact portable membrane lung and perhaps a chemical source of oxygen? Dr. Richard E. Clark (Pittsburgh. Pa.). How does your double-lumen cannula differ from that which Dr. Kolobow has been using for 3 or 4 years and which he reported at the ASAIO meeting? It looks very similar to me, and he gets the same kind of results in juvenile sheep. Could you comment, please? Dr. Delius. Thank you, Dr. Edmunds and Dr. Clark, for your comments. There are plans for adult-sized cannulas but they have not come to fruition yet. Your ideas are very interesting, but I do not know if there are any plans at this point to make the system portable. It certainly is an interesting concept. Dr. Kolobow's work entails using a single-lumen cannula with tidal flow,which involvestaking a certain amount of blood, let's say 200 ml, running it through the circuit, and reinfusing it into the right atrium. That system also works well, but it is much more complex than our system. In our system, the double-lumen cannula readily attaches to our standard VA bypass circuit.