Extracorporeal membrane oxygenation causes significant changes in intracranial pressure and carotid artery blood flow in newborn lambs

Extracorporeal Membrane Oxygenation Causes Significant Changes in Intracranial Pressure and Carotid Artery Blood Flow in Newborn Lambs By Charles J.H. Stolar and Cynthia Reyes N e w York, N e w York 9 The effects of ECMO on cerebral dynamics, particu- 9 larly in the face of asphyxia, are largely unknown. We " inquired as t o w h e t h e r carotid a r t e r y blood f l o w (CABF) and intracranial pressure (ICP) w e r e \ a f f e c t e d by carotid a r t e r y / j u g u l a r vein ligation, asphyxia, ECMO, and ECMO w i t h asphyxia. Lightly sedated n e w b o r n lambs (two to four days old, 3 t o 4 kg) in four groups w e r e monitored for mean ICP by an epidural sensor, mean CABF by a f l o w probe, and J mean arterial pressure. Mean values w e r e determined for t h e duration of each step of the experiment. ECMO was venoarterial at 100 t o 120 m L / k g / m i n . CABF and ICP w e r e measured in group 1 before and after C A / J V ~ ligation; in group 2 during n o r m o x i a / n o r m o c a p n i a followed by hypoxia (30 to 40 t o r r ) / h y p e r c a p n i a (70 t o 90 torr); in group 3 before, during, after ECMO w h i l e n o r m o x i c / n o r m o c a p n i c throughout; and in group 4 as ECMO was begun w h i l e hypoxic/hypercapnic. Vessel ligation alone caused no significant CABF/ICP changes. Asphyxia caused physiologic increases in CABF IP < .03) and ICP (P < .01). ECMO alone caused a significant decrease in ICP (P < ,(303). ECMO w i t h asphyxia caused an even m o r e severe decrease in ICP (P < .001) combined with augmented CABF (P < .03). The ICP decrease was limited t o the duration of ECMO. Possible explanations include loss of cerebral a u t o r e g u l a t i o n induced by hypoxia/hyperd~bia and alterations in cerebral venous drainage necessitated by this method of cardiopulm o n a r y bypass. 9 1988 by Grune & Stratton, Inc. INDEX WORDS: Extracorporeal membrane oxygenation.

T R A C O R P O R E A L membrane oxygenation E X(ECMO) is an established treatment modality

for newborns suffering from intractable pulmonary hypertension and lung injury. Despite its salutary effects on newborn humans, little is known about the effects of ECMO on normal or pathophysiologic states in newborn animal models. ECMO causes increased pulmonary and systemic arterial pressure? but the brain may be the organ most vulnerable to gross fluid dynamic changes, particularly in the presence of asphyxia, which in turn may be further compounded by alterations in arterial blood supply and venous drainage required for ECMO cannulation. Some investigators report that the obligatory major vessel ligation associated with ECMO may exacerbate an evolving cerebral injury. 2'3 To examine the effects of ECMO on cerebral dynamics in an animal model, we simultaneously and continuously measured carotid artery blood flow (CABF) and intracranial pressure (ICP), and asked

four questions. First, do carotid artery and jugular vein ligation alter contralateral CABF and ICP if the animal remains normoxic and normocapnic? Second, in the newborn lamb model, are the responses to hypoxia and hypercarbia physiologic in the absence of ECMO? Within certain limits, CABF should increase with parallel changes in ICP in response to asphyxia. Others 4'5 have shown a predicted autoregulatory response of cerebral blood flow (a consequence of CABF) to a range of arterial PCO2 and PO2 as well as BP. Tweed et al 6 demonstrated in lambs that relatively brief periods of hypoxia disturbed autoregulation of cerebral blood flow, but that it was reversible. We combined hypoxia and hypercapnia to mimic the frequent clinical situation. Third, what are the effects of ECMO alone on CABF and ICP in the face of normoxia and normocapnia? This could address the effects of ECMO on cerebral dynamics independent of pathophysiologic variables. Fourth, in the asphyxiated animal, how does major vessel ligation combined with ECMO in the asphyxiated animal affect CABF and ICP differently from the preceding situations? Newborn lambs were studied because the cerebral circulation has unique features. It is characterized by posterior arterial supply that accounts for only a trivial amount of cerebral perfusion. The anterior circulation arises from a common carotid trunk at the aortic arch. This bifurcates, and in turn, gives several branches to the head and neck before entering the brain as a common carotid artery. This common carotid artery communicates with the circle of Willis by way of a rete system and consequently supplies the majority of arterial blood to the brain. If one carotid artery is ligated, most blood arrives in the brain via the contralateral artery.

From the College of Physicians and Surgeons, Columbia University, and the Division of Pediatric Surgery, Babies Hospital, Columbia-Presbyterian Medical Center, New York City. Supported by The Charles Edison Fund and The Anya Fund. Presented at the 19th Annual Meeting of the American Pediatric Surgical Association, Tucson, Arizona, May 11-14, 1988. Address reprint requests to C. Stolar, MD, Babies Hospital, 203 N, 3959 Broadway, New York, N Y 10032. 9 1988 by Grune & Stratton, Inc. 0022-3468/88/2312-0014503.00/0

Journal of PediatricSurgery. Vol 23, No 12 (December), 1988: pp 1163-1168

1163

1164

STOLAR AND REYES

Table 1. Results for Group 1

MATERIALS AND METHODS

Newborn mixed breed lambs, two to four days old and 3 to 5 kg in weight, were divided into groups of six each to address the aforementioned questions. Group 1 was normoxic and normocapnic so the effects of vessel ligation alone on CABF and ICP could be measured. With group 2, we examined the effects of asphyxia without ECMO in animals that were normoxic and normocapnic initially, followed by sustained hypoxia (30 to 40 torr) with hypercapnia (70 to 90 torr). Group 3 animals had CABF/1CP measured while normoxic and normocapnic throughout initiation and continuation of ECMO. Group 4 animals were made asphyxic (PO2 30 to 40 torr), PCO~ (70 to 90 torr) before vessel ligation and initiation of ECMO. They were subsequently resuscitated with ECMO with CABF and ICP were monitored throughout. ECMO cannulation was for venoarterial bypass through the right neck. The techniques, as well as ECMO circuit configuration, were described by Bartlett et al. ~ A 0.8-m 2 silicone membrane oxygenator (Sci-Med Life Systems, Minneapolis) was combined with a polyvinyl chloride tubing circuit. Standard crystalloid/albumin/whole sheep blood prime was used. ECMO was initiated at 15 mL/kg/min and increased by that increment every ten minutes until 120 mL/kg/min flow was achieved. The oxygenator was ventilated with mixtures of 02, CO2, and compressed air fractions sufficient to generate postoxygenator PO2 150 tort, PCO2 35 torr, pH 7.35. CABF was measured in the left carotid by an appropriately sized electromagnetic flow probe (Biotronix sine-wave flowmeter #$46474; Boston, MA). ICP was measured by a fiberoptic epidural transducer (Ladd Co). This was placed in the epidural space through a frontal burr hole and sealed in position with bone wax. Both devices were calibrated by standard techniques. Arterial blood pressure was monitored through polyvinyl chloride catheters in the femeral artery by hydraulic transducers (Trantee/Bentley, model 800; Minneapolis). Venous access was by femoral vein. All analog physiologic signals were recorded by multichannel physiograph (Sensormedics, Anaheim, CA)~ Animals were prepared by pentobarbitol (1 to 2 mg/kg) induction and maintained lightly sedated (0.5 mg/kg every two hours as needed). Local anesthesia for catheter placement was maintained by

MAP (mmHg) CABF (cc/min) ICP (cmH20)

500,

20 ~ 1 0 0 1

, L

G R O U P I:

0

V E S S E L LIGATION O N L Y NO ASPHYXIA

NO ECMO Fig 1. Results for group 1, effect of vessel ligation only. ICP, intracranial pressure; MAP, mean arterial pressure; CABF, mean carotid artery blood flow; B, baseline; L, ligation.

72 • 16 295 • 142 15 • 2

NS <.06 (n = 6, t test)

In this experiment, each variable was measured serially multiple times, yielding a repeated measures design. In groups 1 and 2, a paired t test was used to compare prequestion and postquestion mean values (_+SEM) for each variable. For groups 3 and 4, individual t

20"

B

79 -+ 17 263 +- 141 15 -+ 1

Statistical Methods

9 ICP

0 ....

P

0.5% lidocaine infiltration. Intravascular volume and hematocrit were maintained by whole sheep blood and crystalloid as needed. All animals were ventilated by endotracheal intubation on a pressurecycled ventilator (Healthdyyne, Rahway, N J) and maintained in an upright prone position once all catheters were secured. In groups 1 and 3, the animal was ventilated with compressed air and 02 mixtures of FiO~ ffi 0.35 at intermittent mandatory ventilation (IMV) 30 per minute, positive end-expiratory pressure (PEEP) 5 cmH20 and peak inspiratory pressure (PIP) 10 cmH20. In groups 2 and 4, asphyxia was generated in the animals with hypoxic gas mixture.-FiO2 .10, FiCO~ .20, and FiN2 .70. A minimum tenminute stable baseline was obtained in all groups before and at each step of the experiment before CABF/ICP recordings were initiated. In groups 3 and 4, ECMO duration was two hours. Animals were euthanized by pentobarhitol and potassium chloride overdose. Autopsy confirmed arterial cannula tips positioned at the carotid trunk/aortic junction, not occluding the main carotid trunk. All protocols had prior review and approval of the Institutional Animal Care and Use Committee at Columbia-Presbyterian Medical Center.

9 CABF

0

Ligation

NOTE. Ligation of the vessels causes no significant changes in ICP and MAP; CABF augmentation approaches significance (P < .06, t test). Abbreviations: NS, not significant; MAP, mean arterial pressure; CABF, carotid artery blood flow; ICP, intracranial pressure.

9 O ICP MAP 100"1

Baseline

0

O MAP

0

9 CABF 500-

"

0"

B

A

G R O U P Ih ASPHYXIA O N L Y NO LIGATION NO ECMO Fig 2. Results for group 2, effect of vessel ligation only. ICP, intraoranial pressure; M A P , mean arterial pressure; CABF, mean

carotid artery blood flow; B, baseline; A. asphyxia.

EFFECTS OF ECMO

1165

Table 2. Results for Group 2

MAP (mmHg) CABF (cc/min) ICP (cmHaO)

9 O ICP MAP

Baseline

Asphyxia

P

66 _+ 8 153 _+ 104 11 +_ 2

72 + 11 191 _+ 60 16 • 5

NS <.02 <.01 (n = 6, t test)

CABF

20"1001

t

NOTE. CABF (P < .02, t test) and ICP (P < .01, t test) have physiologic responses to asphyxia. Abbreviations: NS, not significant; MAP, mean arterial pressure; CABF, carotid artery blood flow; ICP, intracranial preSSure,

tests could not compare the three to five means for the multistaged experiments, and consequently it was not possible to judge significance of the overall experiment. Instead, we used a multivariable analysis of overall variance (SPSS-X-MANOVA) to test the significance of our repeated measures design. If significance was determined for a multistep experiment, a paired t test was applied to the mean value during each step, for each variable, to investigate the source of the significant variance.

RESULTS

0"

500

t

I. i

B

E

P

GROUPIIh ECMO ONLY NO ASPHYXIA Fig 3. Results for group 3. effect of ECMO only. ICP, intracranial pressure; MAP, mean arterial pressure; CABF, mean carotid a r t e r y blood flow; B, baseline; E, ECMO; P, post-ECMO.

Group 1 Although ligation of the carotid artery and jugular vein caused transient changes in mean arterial pressure (MAP), CABF and ICP during normoxia and normocapnia, these changes were not significant. CABF increased by 12%, but only approached significance ( P < .06, t test) (Fig 1, Table 1).

Group 2 Although asphyxia alone did not significantly alter MAP, CABF and ICP had predicted physiologic responses. CABF increased by 25% (P < .02, t test) and ICP increased by 45% (P < .01, t test) (Fig 2, Table 2).

Group 3 ECMO alone caused no significant sustained changes of MAP or CABF when a healthy animal was perfused for two hours. However, the acute changes that accompanied initiation of ECMO were characterized by a fall in ICP of 29% (P < .003, t test). This fall in ICP was significant only while the animal was on ECMO. Once off ECMO, ICP gradually returned toward the pre-ECMO baseline and lost its significance (P = NS, MANOVA) (Fig 3, Table 3).

Group 4 The initial asphyxia in this group caused mean CABF to increase by 51% compared with the preasphyxic baseline (P < .03, t test). In the asphyxiated animal, ligation of the carotid artery and jugular vein again did not significantly alter MAP. Significant overall changes in CABF for this multistep experiment were observed (P < .03, MANOVA). Once ECMO was initiated, CABF returned toward a level not significantly different from the baseline. Mean ICP was increased 60% compared with the preasphyxic baseline (P < .03, t test) in the asphyxic phase of this experiment. Subsequent vessel ligation caused no further significant changes compared with asphyxic levels, but were more significant when compared with baseline levels (P < .008, t test). More importantly, mean ICP during perfusion on ECMO was associated with a dramatic and sustained decrease to 75% below the asphyxic level (P < .001, t test) and 60% below the normoxic baseline (P < .001, t test). Termination of ECMO resulted in a gradual return toward the baseline. Although MAP had no significant changes for

Table 3. Results for Group 3

MAP (mmHg) CABF (cc/min)

ICP(cmH20)

Baseline

On ECMO

Off ECMO

P

81 _+ 9 292-+ 105 14 • 2

79 _+ 23 264 + 147 10_+ 3

66 + 14 251 -+ 165 11 -+ 3

NS NS NS (n = 6, MANOVA)

NOTE. ECMO causes ICP to fall (P < .003, t test), but ICP returns to baseline after ECMO. Abbreviations: NS, not significant; MAP, mean arterial pressure; CABF, carotid artery blood flow; ICP, intracranial pressure.

1166

STOLAR AND REYES a

9 0 ICP MAP

CABF

20" 1003

500

+ O-

0

B

NL

E

P

A

GROUPIV: ECMO

ASPHYXIA Fig 4. Results for group 4, effect of ECMO and asphyxia. ICP, intracranial pressure; MAP, mean arterial pressure; CABF, mean carotid artery blood flow; B, baseline; A, asphyxia; L, ligation; E, ECMO; P, post-ECMO.

this entire experiment, CABF (P < .02, MANOVA) and ICP (P < .001, MANOVA) clearly change. Physiologic integrity of the animal was demonstrated by intact responsiveness of CABF and ICP to a second asphyxic insult (Fig 4, Table 4). DISCUSSION

Little is known about the effects of cardiopulmonary bypass on cerebral dynamics, particularly in the face of asphyxia. The literature is confusing because of inconsistent animal models and study methods. What is reported is usually in the context of nonpulsatile flow during total cardiopulmonary bypass. ECMO is only partial cardiopulmonary bypass, and although the pulse pressure is narrowed, pulsatile perfusion is maintained. Nevertheless, neurologic disturbances that do occur can be thought of, in general, as failures of maintenance of cerebral blood flow during cardiopulmonary bypass. In cats, Santillan et al, s used a radiolabeled microsphere technique to assess global cerebral blood flow; they showed that cardiopulmonary bypass was associated with increased cerebral blood flow, decreased resistance, and alterations in MAP. BP and blood flow were interdependent and showed direct proportionality. Govier et al, 9 studied cerebral blood flow in human subjects using xenon 133 clearance techniques as related to MAP and PCO2. They

observed an expected interdependence of physiologic variables. Specifically, cerebral blood flow responded to PCO2 variations as expected, despite augmentations of flow on bypass. The work of Walker et al/~ is relevant. In studying lambs on ECMO, they reported that autoregulatory responsiveness of cerebral blood flow to PCO2 as measured by microsphere technique is intact over a range of bypass flows. We observed increased CABF in groups 2 and 4. Extrapolation to the human setting is suggested by Lundar et al, 1~ who observed in humans that flow velocity in the middle cerebral artery was increased by cardiopulmonary bypass, as measured by noninvasive Doppler techniques. The Doppler-recorded changes in the middle cerebral artery were corroborated by simultaneous electromagnetic flow probe measurements of the ipsilateral internal carotid artery. They also demonstrated that an autoregulatory control of flow velocity was intact despite variable nonpulsatile perfusion. This is the same method we used in this report to imply cerebral blood flow in the newborn lamb model. Reports from our laboratory 1 suggest that initiation of ECMO may precipitate a hyperdynamic state characterized by systemic as well as pulmonary hypertension, which is at least temporally related to activation of a variety of vasoactive mediators. In this study, the hyperdynamic state is reflected in CABF changes. The self-limited nature of the effect is reflected in the overall statistical nonsignificance of the CABF changes throughout the entire experiment. How the hyperdynamic state and vasoactive activation induced by ECMO interface with the changes in cerebral dynamics is speculative. Not only is autoregulation of cerebral blood flow preserved during cardiopulmonary bypass, but in our experiments, it is also preserved, as measured by CABF during asphyxia. Although not directly measured, this observation is implied in group 4 by the normal physiologic response of ICP/CABF to a second asphyxic insult. The usual response of cerebral blood flow to asphyxia has been characterized in lambs by Rosenberg/2 who describes the early postasphyxic period by increased cerebral blood flow and increased oxygen consumption. Subsequently, there is a depression of cerebral blood flow and oxygen consumption, which may contribute to the development of an

Table 4. Results for Group 4 Baseline MAP (mmHg) CABF (cc/min) ICP(cmH20)

73 + 11 110 +_ 37 10_+ 3

Asphyxia/Ligation 77 +_ 16/ 84 + 19 166 _+ 61/287 _+ 149 16-+ 2 / 17 -+ 6

ECMO 76 +_ 14 165 +_ 84 4_+ 1

P 69 +- 15 152 +- 75 7 _+ 4

A 90 +- 25 205 +- 74 14-+ 5

P <.08 <.02 <.001 (n = 8, MANOVA)

NOTE. Ligation causes increased CABF (P < .03); ECMO causes significant intracranial hypotension (P < .001) limited to ECMO. Abbreviations: MAP, mean arterial pressure; CABF, carotid artery blood flow; ICP, intracranial pressure.

EFFECTS OF ECMO

1167

asphyxic CNS lesion. Another possibility is impaired vasodilitation in the postasphyxic state with or without hypercapnia, which prevents augmentation of cerebral blood flow. It is not clear how ECMO modifies the postasphyxic state. Short et al ~3 studied nonasphyxiated lambs and reported that cerebral blood flow and oxygen consumption were significantly decreased during the early hours of ECMO in a lamb model. Our results are different, as they show increased CABF. These differences may be explained by differences in methodology as well as the use of an asphyxiated model. Diversion of blood from the carotid to the head/neck region before it enters the brain was not quantitated in our model. Consequently, CABF measurements are not absolute measures of cerebral blood flow but proportional ones. The continuous dynamic changes associated with initiation of ECMO preclude more discrete but static determinations of cerebral blood flow using radio-labeled microspheres. This mandated continuous analog measurements with electromagnetic flow probes. However, if the CABF changes are proportional to cerebral blood flow changes, then it can be suggested that ECMO may

minimize postasphyxic changes by generating adequate cerebral blood flow and oxygen delivery to the brain during a very critical time. Although the responses of ICP/CABF to asphyxia were as predicted, the responses of ICP to ECMO were surprising. ICP, in general, should parallel changes in cerebral blood flow. Although CABF was at least transiently increased with the initiation of ECMO, ICP was acutely and dramatically depressed for as long as the animal was on bypass. The reasons for this are not dear. In fact, ligation of the jugular vein might be expected to actually increase ICP at least transiently. This did happen, but not to a significant degree in both the normoxic and asphyxiated groups. ICP in the asphyxiated group was already significantly elevated prior to ECMO initiation. We speculate that intracranial hypotension may be due to preferential drainage and decompression of the right atrium, superior vena cava, and its tributaries from the brain to a greater degree than the inferior vena cava. The implications of intracranial hypotension for the early postasphyxiated brain when combined with ECMO are speculative.

REFERENCES 1. Stolar CJH, Dillon PW: Do venovenous and venoarterial ECMO cause different modifications of the vassoactive profile and cardiopulmonary hemodynamics in newborn lambs? Pediatr Res 16:1306, 1985 (abstr) 2. Brann AW, Myers RE: Central nervous system findings in the newborn monkey following severe in utero partial asphyxia. Neurology 25:327-338, 1975 3. Rice JE, Vannucci RL, Brierly JB: The influence of immaturity on hypoxic-ischemic brain damage in the rat. Ann Neurol 18:131-141, 1981 4. Rosenberg AA, Jones D, Traystman R J, et al: Response of cerebral blood flow to changes PCO2 in fetal newborn, and adult sheep. Am J Physiol 242:H862-H866, 1982 5. Papile L, Rudolph A, Heyman MA: Autoregulation of cerebral blood flow in the pre-term fetal lamb. Pediatr Res 19:159-161, 1985 6. Tweed A, Cote J, Lou H, et al: Impairment of cerebral blood flow autoregulation in the newborn lamb by hypoxia. Pediatr Res 20:516-519, 1986

7. Bartlett RH, Gazzaniga AB, Tommasian J, et al: Extracorporeal membrane oxygenation in neonatal respiratory failure--100 cases. Ann Surg 204:236-242, 1986 8. Santillan GG, Chemnitius JM, Birg R J: The effect of cardiopulmonary bypass on cerebral blood flow. Brain Res 345:1-9, 1985 9. Govier AV, Reves JG, McKay, RD: Factors and their influence on regional cerebral blood flow during nonpulsatile cardiopulmonary bypass. Ann Tborac Surg 38:592-600, 1984 10. Walker LK, Short BL, Gleason CA, et al: Cerebrovascular response to CO2 during ECMO. Pediatr Res 23:429A, 1988 (abstr) 11. Lundar B, Lindegaand KF, Froysak T, et al: Cerebral perfusion during cardiopuimonary bypass. Ann Thorae Surg 40:144150, 1985 12. Rosenberg AA: Regulation of cerebral blood flow after asphyxia in neonatal lambs. Stroke 19:239-24,1, 1988 13. Short BL, Walker LK, Gleasen CA, et al: Effects of extracorporeai membrane oxygenation on cerebral blood flow and oxygen consumption in the newborn. Pediatr Res 23:425A, 1988 (abstr)

Discussion A. deLorimier (San Francisco): Could you tell us what produces the diminished intracranial pressure? Is it possible that there is arteriovenous shunting, a phenomenon that is not normally associated with intracranial blood flow mechanics, or is it perhaps a loss of water volume in the brain tissue itself? T. Weber (St Louis): We see a number of candidates for ECMO that are hypoxic, but are hypocarbic due to vigorous ventilation. Can you tell us whether it is the

hypoxia or the hypercarbia that causes the cerebral blood flow alterations in your study? T. Moore (Los Angeles): I would like to inquire about the effect of the blood in terms of other responses, particularly an immunologic response. As you know, there has been substantial interest in recent years in the involvement of whole blood as an immunosuppressant beneficial in transplantation and adverse in cancer patients. We have been looking at the role of

1168

whole blood transfusion in sheep; the adults seem to be sure, but we find there is a rather marked reduction and it looks like traffic through primary peripheral nodes in adult sheep when they are given central venous whole blood transfusions. This is one possible aspect of using ECMO not only in the sheep model but in the adults. Do you have any observations or comments regarding this matter? D. Tapper (Seattle): Would it be appropriate to study this same model using venoveno ECMO? I am concerned if people read titles, particularly neonatologists who are looking for alternate approaches, that this may be interpreted along with some other data as a major reason to suspend the use of ECMO until controlled studies indicate a clear cut advantage. Also, further evaluation of CNS effects may limit its usefulness. C.J.H. Stolar (closing): All those factors that Dr DeLorimier mentioned may be pertinent. With discrete microspheric studies, you could look at regional blood flow changes. We speculate that the etiology of

STOLAR AND REYES

the intracranial hypotension is related to the selective decompression of the superior vena cava with the ECMO cannulation and drainage from the head. The reason you see hypocarbic babies may be because you ventilate newborns differently from the way we do. The way to address it would be to create hypoxic hypocarbia and hypercapnic hypoxia and study those two issues separately, which we have not done in this experiment. A large body of data suggests that the autoregulatory response of the brain may be more sensitive to PCO2 rather than oxygen content, and responds to overall oxygen delivery by change in carotid artery flow. In this setting, the brain is more sensitive to the oxygen delivery than the CO2 content. CABF changes with venoveno bypass would be the same since it is a hypoxic response. The political comments at the SPR meeting--some work about ECMO altering 02 consumption and cerebral flow by Billy Short et al in Washington (Society of Pediatric Research, Annual Meeting, April 1988)--were listed under the ECMO complication section.