Original Paper Pediatr Neurosurg 1993; 19:202-205 A .I. Canady R .D . Fessler M .D . Klein Neurosurgery Children’s Hospital. Detroit, M ich.. U SA Key Words Ultrasound, head Extracorporeal membrane oxygenation Extra-axial brain collections Ultrasound Abnormalities in Term Infants on ECMO Abstract The charts and cranial ultrasounds of 29 infants treated with extracorporeal membrane oxygenation (ECMO) for respiratory insufficiency secondary to meconium aspiration syndrome, primary pulmonary hypertension, congenital diaphragmatic hernia and/or sepsis were examined to identify ultrasound abnormalities. Seventeen (58.6%) developed extra-axial fluid collections, only two of which were progressive. Ten (34.5%) developed evidence of intracra­ nial hemorrhage (ICH): seven caudate, one each in the thalamus, parietal and occipital lobes. Eight (27.65%) of the neonates had seizures while on ECM O . 5 of whom had concurrent ICH . Extracorporeal membrane oxygenation (ECMO) is in­ creasingly used to improve oxygenation in neonates whose response to mechanical ventilation is unsatisfacto­ ry. It is most successful in those patients who have pulmo­ nary hypertension as a major factor in their ventilatory failure. Initially, it was used in premature infants as well, but recently its use has been restricted to term or nearterm infants. This study surveys the intracranial abnor­ malities demonstrated by ultrasonography that developed in this population. infants were all full-term infants, average gestational age 39.5 weeks, with oxygenation difficulties secondary to meconium aspiration syn­ drome (10 patients), primary persistent pulmonary hypertension (7 patients), congenital diaphragmatic hernia (3 patients) and I patient with primary pulmonary hypertension and/or sepsis. The patients were on E C M O an average o f 6.5 days. Ultrasounds were performed before the start o f E C M O , daily during E C M O and as clinically indi­ cated after E C M O decannulation. Infants with evidence o f abnor­ mality on pre-ECM O ultrasounds were excluded from the study. The cranial ultrasounds were performed using an Acuson (Mountain View, Calif.) with linear high frequency transducer 5 or 7 M H z. Results Method Twenty-nine charts and ultrasounds were reviewed. Nineteen (65%) developed an intracranial abnormality documented by cranial ultrasound during EC M O thera­ py. Seventeen (58.6%) developed extra-axial fluid collec- A.I. Canady 3901 Beaubien Detroit. Ml 48201 (USA) © 1993 S. Karger AG. Basel 1016-2291/93/0194-0202 $2.75/0 Downloaded by: Univ. of California Santa Barbara 128.111.121.42 - 3/27/2018 3:16:55 PM The charts and imaging studies for 29 neonates who were placed on E C M O at the Children’s Hospital o f Michigan from 1989 to 1991 were evaluated for the presence and nature o f cranial ultrasound abnormalities. Twenty patients demonstrated abnormalities. The Discussion ECM O has been demonstrated to be a useful technique for improved oxygenation in properly selected full-term infants. Neurosurgeons are being asked to consult on the infants with abnormal ultrasounds and need to under­ stand the findings of a particular infant in the context of the ultrasound findings of children on ECM O in general. Part of this understanding must include the technical side of E C M O . Cannulation for EC M O involves the ligation of the right internal jugular vein and the right common carotid artery (fig. 1). Following cannulation, infants are heparinized and develop thrombocytopenia and coagu­ lopathy requiring infusion of platelets, fresh frozen plasma and packed red blood cells. Outflow of the blood from the patient is by gravity only and frequently these patients are elevated to as much as four or five feet above the floor to facilitate drainage. Fluid intakes still exceed outputs with elevated central venous pressures, an aver­ age of 16.7 mm Hg in our patients. The infants are all grossly edematous (fig. 2) with third spacing of fluid. In this context, the most common finding, extra-axial fluid collections (fig. 3. 4) sometimes misinterpreted as atrophy, seen in 17 patients, is not surprising. This per- 203 Downloaded by: Univ. of California Santa Barbara 128.111.121.42 - 3/27/2018 3:16:55 PM tions, two (6.9%) of which were progressive with enlarg­ ing head circumference and split sutures requiring surgi­ cal intervention. Fluid retention was a problem in ECM O patients with the intake exceeding the output in these patients by an average of 220 ml/day with the average central venous pressure being 16.7 mm Hg. Ten (34.5%) of the infants, all of whom were lull-term, developed evidence of intracranial hermorrhage (ICH): 7 caudate (24.1 % of hemorrhages) and 1 (3.4% of hemor­ rhages) each in the thalamus, parietal and occipital lobes. All patients, as expected, since they failed mechanical ventilation, had documented hypoxia prior to being placed on ECM O . No patients required cardiopulmonary resuscitation prior to EC M O cannulation. Six patients with ultrasound abnormalities developed a pneumotho­ rax in their course. 4 of whom also developed ICH on ultrasound. All studied patients had a period of hypoten­ sion in their course. Eight (27.6%) of the infants had sei­ zures while on E C M O , 5 of whom had concurrent ICH . No intracranial hemorrhage required surgical interven­ tion. ECM O was discontinued when hemorrhage was identified. Fig. 3. Ultrasound showing increased fluid in interhemispheric Fig. 5. Ultrasound o f head showing IC H in a child on E C M O . fissure. tricle. centage of 59.6% is higher than in other series. Taylor et al. [I] reported 8 cases of atrophy and 8 of a widened interhemispheric fissure in a series of 146 patients. Both of these probably represent increased extra-axial collec­ tions for a total of 11 %. An incidence of approximately 10% was also seen by Babcock et al. [2], This series may have missed transient extra-axial collections as the proto­ col called for ultrasounds before E C M O . within 12 h of E C M O and only on clinical grounds during ECM O . That explanation does not apply to the Taylor series where ultrasounds were also obtained daily during ECM O . The discrepancy cannot be explained at this time. The extraaxial collections are felt to be secondary to increased venous drainage pressures. All but two of these cases were only transient, making the ligation of the jugular vein which occurs at the time decannulation in all patients a 204 Canady/Fesslcr/KJcin Ultrasound Abnormalities in Term Infants on E C M O Downloaded by: Univ. of California Santa Barbara 128.111.121.42 - 3/27/2018 3:16:55 PM Fig. 6. IC H in posterior fossa, right temporal area, and 3rd ven­ Fig. 4. Computerized tomography o f patient who went on to require surgical intervention for extra-axial collections. References i the daily ultrasound demonstrates new hemorrhage. At the Children’s Hospital of Cincinnati, Babcock et al. [2] did not feel that small subependymal hemorrhages ex­ tended on ECM O . The 34.5% incidence of hemorrhage in this series is comparable to Taylor’s series where preterm infants were also excluded and the hemorrhagic rate by ultrasound and computerized tomography was 30% [1], Eight (37.6%) of the neonates had seizures while on E C M O , 5 of whom had concurrent ICH . So the occur­ rence of seizure should raise increased suspicion of ICH . This incidence is comparable to the series of Matamoros et al. [5] with a 25% incidence of seizures on ECM O . Conclusion Although there is a high incidence of extra-axial fluid collections in infants on ECM O , the vast preponderance is self-limited and does not require neurosurgical inter­ vention either by tapping or surgery, both of which would be contraindicated in these heparinized patients without discontinuing the ECM O therapy. In the small percentage that are progressive, intervention can be accomplished after ECM O is discontinued as determined by pulmonary' indications. In the cases of limited IC H . surgical interven­ tion was not required but ECM O therapy was discontin­ ued because of concern regarding extension of the clot. Taylor G A . Fitz C R . Glass P. Short BL: C T o f cerebrovascular injury after neonatal extracor­ poreal membrane oxygenation: Implications for neurodevclopmental outcome. Am J Radiol 1989:153:121-126. Babcock D S. Han B K . Weiss R G . Ryckman FC : Brain abnormalities in infants on extracor­ poreal membrane oxygenation: Sonographic and C T findings. Am J Radiol 1989:153:571— 576. Cillcy R E. Zwischcnbcrger JB . Andrews A F. Bowerman R A . R oloff DW . Bartlett RH : Intra­ cranial hemorrhage during extracorporeal membrane oxygenation in neonates. Pediatrics 1986:78:699-704. 4 5 Sell LL. Cullen M L . Whittlesey G C . Yedlin ST . Philippart A l. Bedard M P. Klein M D : Hemor­ rhagic complications during extracorporeal membrane oxygenation: Prevention and treat­ ment. J Pediatr Surg 1985:21:1087-1091. Matamoros A , Anderson J C . McConnell J . Bolam DL: Neurosonographic findings in infants treated by extracorporeal membrane oxygena­ tion (E C M O ). J Child Neurol 1989:4:552S61. 205 Downloaded by: Univ. of California Santa Barbara 128.111.121.42 - 3/27/2018 3:16:55 PM possible contributory but not the primary causative fac­ tor. Hydrocephalus has been seen in 2 patients, not in this series. In both those cases, the patients had superior vena cava obstruction as well. The hermorrhages that are seen now in full-term infants who have had good control of their hypertension are much more discrete (fig. 5) than those seen in the past. Early experience showed that infants less than 35 weeks gestation had rates of ICH as high as 100% [3]. A correla­ tion between hypertension and the occurrence of ICH was demonstrated at the Children’s Hospital of Michigan. With a high hypertension index (hours systolic BP greater than 90 divided by hours receiving ECM O), there was an increased risk of ICH . If ICH is present, the hypertension index was 0.37 ± 0.28 and if no IC H , the hypertensive index was 0.2 ± 0.12 [4], As these factors were recognized, the early massive ICH that came to neurosurgical attention largely disap­ peared (fig. 6). Instead, extra-axial fluid and discrete hem­ orrhages were the reason for neurosurgical consultation. The anticoagulation and thrombocytopenia undoubtedly play a role. In the current series, of the 6 patients with pneumothorax in their hospitalization. 4 had ICH . raising concern that barotrauma which predisposes premature infants to germinal matrix hemorrhages may play a role as well. Other factors postulated include rapid infusion of colloid [3], In our institution, ECM O is discontinued if