Applied Radiology

The introduction of transcranial Doppler ultrasound (TCD) by Aeslid in 1982provided a new window for the evaluation of cerebral blood flow velocities inthe brain.1 Alterations in blood flow following acute hydrocephalus, asphyxia,trauma, and subarachnoid hemorrhage can be followed with TCD. The potentialclinical utility of this technique in evaluating vascular injury and cerebralautoregulation in the neonatal and pediatric population is addressed below.

Technique

The anterior fontanelle usually remains open through the first year of aninfant's life and provides a window for imaging the neonatal brain. A 3 to 5MHz sector transducer can be used via the anterior fontanelle to insonateintracranial vessels utilizing duplex ultrasound with color imaging. In thesagittal midline plane, the anterior cerebral arteries (ACA) can be visualizedcoursing over the corpus callosum (figure 1). The vessels of each side are soclosely adjacent that the signals from them cannot be differentiated. In thecoronal plane, both internal cerebral arteries can be insonated just above thecranial base (figure 2). The middle cerebral arteries (MCA) are insonated asfar laterally as possible (figure 3). Higher frequency transducers can be usedto evaluate superficial vessels, such as the sagittal sinus.

After fontanelle closure, blood velocity and resistive indices of theintracranial arteries can be obtained via the transtemporal window. This isachieved with a nonimaging pulsed Doppler technique or duplex ultrasound withcolor imaging using a 2 MHz transducer. The area in which the temporal bone isthinnest is demarcated by an imaginary line drawn 1 cm above the zygomaticarch, anterior to the external auditory meatus. The ultrasound beam is thendirected horizontally. The circle of Willis lies anterior to the cerebralpeduncles in the suprasellar cistern (figure 4). Color Doppler and spectralanalysis will show MCA flow towards the transducer (figure 5).

In infants, the most distal portion of the M1 segment is generallyidentified at a depth of 30 mm. In older children, this progressively deepens,reaching an average adult depth of 50 mm by the age of 10 years. At deeperdepths, the bifurcation of the ACA and MCA will show both antegrade andretrograde flow. Tilting the probe anteriorly and deeper allows fordemonstration of ACA flow away from the transducer. The posterior cerebralartery (PCA) can be visualized caudally, with flow toward and then away fromthe transducer as it circles around the cerebral peduncles. The peak systolicand diastolic velocities and mean peak velocities of these vessels can then bemeasured.

The resistive index (RI) (systolic velocity-diastolic velocity/systolicvelocity) or pulsatility index (PI) (systolic velocity-diastolic velocity/meanvelocity) minimizes the effect of angulation of the probe. Age dependentreference values are available for velocities and RIs of the variousintracranial vessels. In term infants, mean RI of the intracranial vesselsmeasures 0.7 ± 7%. This decreases to 0.5 ± 15% by the age of two.2 Anincrease in diastolic flow will result in a decrease in the RI, while adecrease in diastolic flow will result in an increase in the RI. As ICPincreases above mean arterial pressure, diastolic flow may become reversed,demonstrating an RI of greater than 1.0.

Hydrocephalus

The ability to differentiate between patients with ventriculomegaly andthose with hydrocephalus (increasing ventricular size and increasedintracranial pressure) can be difficult. Radiographic findings and clinicalparameters, such as increasing head circumference, bulging fontanelle,lethargy, and bradycardia, may not be adequate in evaluating the need forshunting or shunt revision.

When hydrocephalus develops, intracranial pressure increases, resulting in adecrease in end-diastolic flow. Stable ventriculomegaly is associated withnormal pulsatility. Thus, an elevation in RI may imply the need for aventriculoperitoneal shunt. In a series by Chadduck and Seibert, childrenrequiring shunts had a mean RI of 84 ± 13% prior to shunting. The mean RIfell to 72 ± 11% after shunting (p <0.001).2 Patients who neverrequired shunts had RIs in the normal range. In infants with ventriculomegalysecondary to intraventricular hemorrhages, RI may be used in planning thefrequency of tapping and assessing effectiveness of the taps prior to theconsideration of permanent shunting (figure 6).3

Goh recommended using an RI of greater than 0.8 as a sign of increased ICPin the neonate and an RI of greater than 0.65 in children.4 Unfortunately,there is a wide variation of normal values, with an overlap between abnormaland normal. Thus, baseline values are most useful in following a patient'scourse. Limitations in RI reliability include multiple intra- and extracranialfactors that can alter the RI. One common reason for lack of RI correlationwith the hydrocephalus can be due to the presence of a meningomyelocele. Insuch patients, low RIs are noted despite the presence of hydrocephalus due toleaking spinal lesions. A falsely high RI may be the result of a patent ductusarteriosus, pneumothorax, or indomethacin. Thus, the risk of false positive orfalse negative values requires close correlation with clinical findings.5

TCD also has been shown to be useful for predicting shunt malfunction.6,7Postoperative baseline studies can be used to assess adequacy of the shuntingprocedures. With the expectation that a patient's RI should decrease with age,any increase in RI could be considered significant in terms of shuntmalfunction. Knowing the baseline of each individual patient may allow for moresuccessful monitoring than is normally possible due to the wide range of normalRI values. Excessive thickness of the calvarium in some of these patients,however, may prevent their RIs from being obtained successfully. False normalvalues may be the result of CSF fluid tracking along the shunt.

Vascular malformations

In the infant with congestive heart failure of unknown etiology, thedifferential diagnosis includes intracranial vascular malformations. TCD is asimple method of identifying these arteriovenous malformations portably via theanterior fontanelle. If flow is increased because of shunting from a deep AVMor by direct arterial communication, the vein of Galen may become massivelydilated. On grey scale sonography these so-called "vein of Galenaneurysms" usually appear as cystic masses. However, duplex color andpower Doppler imaging can confirm that these cystic masses are indeedvascular.8-10 Spectral waveforms will demonstrate high velocity and lowpulsatility with elevated diastolic flow. Peak and mean systolic velocitieswill be elevated, and RIs will be low. While MRI is more sensitive in screeningfor vascular malformations and angiography is still required to evaluate theanatomy of the feeding vessels, Doppler is useful in following the effects ofembolization or surgical ligation (figure 7). The decreasing size of themalformation can be followed, as can the decrease in systolic velocity andincrease in RI of the feeding vessels.8,9

Asphyxia

TCD can be useful in the evaluation of hypoxic-ischemic brain injury.Occurrence of asphyxia may result in impairment of cerebral autoregulation,producing an increase in diastolic blood flow and a decrease in cerebrovascularresistance (figure 8). In a series of term infants studied by Archer et al, alow RI within the first 48 hours of asphyxia correlated with a poor neurologicoutcome.11 Stark and Seibert demonstrated similar findings in a series of 16term infants with low RIs, two of which died and 11 of which experienced severeneurodevelopmental delay.12 Interestingly, only 50% of these cases had abnormalultrasound findings. Bode, in a study investigating an asphyxiated neonatalpopulation, also noted low RIs and elevated cerebral blood flow velocities.Hyperventilation failed to alter the waveform pattern, indicating vasomotorparalysis.13 This loss of autoregulation also has been described in the olderchild after head injury or cardiac arrest.14 This increase in diastolic flowfollowing asphyxia in children is also likely due to impairment in cerebralautoregulation, and may be useful in predicting cerebral injury prior to CTfindings.

Vasospasm

In adults, TCD has been used in the evaluation of vasospasm followingsubarachnoid hemorrhage. Vasospasm typically develops in the first two daysafter the hemorrhage, peaks two weeks later, and then gradually declines duringthe subsequent three weeks. As the cross sectional area of the affected vesselsdecreases, blood flow velocity will increase. TCD has been found to be highlyspecific in the diagnosis of vasospasm in patients with increased flowvelocities preceding clinical manifestations of cerebral ischemia. Thus, TCDcan be used to guide optimal timing of surgery and early institution oftherapy. Serial TCD studies showing reduction in velocities indicate anappropriate time to withdraw therapy, minimizing complications and shorteningthe patient's stay in the ICU.15

TCD is most accurate in the evaluation of the proximal MCA. Vasospasm isconsidered severe if velocities are greater than 200 cm/sec. If a rapidincrease in velocity (>50 cm/sec per day) is noted in the early daysfollowing a subarachnoid bleed, the prognosis is guarded.16 Errors in diagnosisof vasospasm can be the result of increased intracranial pressure, low volumeflow, or peripheral vasospasm. Thus, TCD results should always be combined withclinical and laboratory data.17

Sickle cell patients

Children with sickle cell disease are at risk of cerebral infarctionsecondary to occlusive vasculopathy.18 Stenosis of the ICA, MCA, and ACA mayprogress for years before symptoms develop. Stroke prevention may be possibleby hypertransfusion therapy in these patients who are at risk.

Adams and coworkers demonstrated that non-duplex Doppler can be effective inscreening for cerebrovascular disease. In 190 asymptomatic sickle cell patientswho were followed, a mean peak flow velocity in the MCA of greater than 170 wasan indicator of a patient at risk for stroke.18 The researchers then comparedTCD to cerebral angiography in 33 neurologically symptomatic patients anddetermined criteria for cerebrovascular disease that included a mean peakvelocity of 190 cm/sec, a low MCA velocity of less than 70 cm/sec, and anACA/MCA ratio of greater than 1.2.19

In their study, Seibert and coworkers described several indicators ofcerebrovascular disease using duplex Doppler imaging. These included mean peakvelocity of the MCA of greater than 170 cm/sec, ophthalmic artery (OA) velocityof greater than 35 cm/sec, RI in the OA of less than 0.5, and OA, PCA,vertebral, or basilar artery velocities that were greater than the MCAvelocity.20,21

In the STOP protocol by Adams et al, TCD studies were considered abnormal ifthe time average mean velocity of the MCA, bifurcation, or distal ICA wasgreater than 200 cm/sec.22 A total of 3929 TCD studies were performed on 1934children. In this series, 9.4% of patients had abnormal studies. A total of 130children were randomized to transfusion or standard care. There were 10cerebral infarctions and one hemorrhage in the standard care group, and oneinfarction in the treated group. Current recommendations suggest that twoabnormal studies, separated by at least a week, should be recorded beforehypertransfusion is advised.

Traumatic brain injury

Head trauma initiates several pathologic processes, often resulting insignificant changes in cerebral hemo- dynamics. Timely and accurate diagnosisof these abnormalities is crucial in the management of head injury. Evenpatients with minor head injuries have been found to develop impaired cerebralautoregulation, and may be at increased risk for secondary ischemic neuronaldamage.23 Blood flow velocity may then increase due to vasospasm fromsubarachnoid hemorrhage or posttraumatic hypervolemia. If cerebral edemadevelops, diastolic flow will decrease, resulting in an elevation of the RI. AsICP increases above mean arterial pressure, reversal of blood flow in diastolemay occur (figure 9).

Serial TCD readings can be helpful in evaluating the presence of cerebraledema and in following the treatment course.24 One method of treating cerebraledema is hyperventilation. Due to the vasodilatory effect of CO2, the higherthe pCO2, the greater the diastolic flow and the lower the RI. With lowerlevels of CO2, vasoconstriction occurs, with a decrease in diastolic flow andan increase in RI. Cerebral blood flow should increase as CO2 rises and can beused to assess CO2 reactivity. If the RI does not change as the patient ishyperventilated, severe brain injury may be present.25 However, cerebral edemaalso may result in increasing RI and therefore, correlation with clinical andlaboratory findings is important.

Brain death

Establishing brain death in a timely manner can be problematic. Besides theneurologic examination, EEG, brain stem evoked potential, and nuclear bloodflow studies, the use of TCD adds another noninvasive method of determiningbrain death. Advantages of the TCD study include noninvasiveness, the abilityto repeat the study as often as required, portability, low cost, and relativeease of performance. For patients in phenobarbital coma, in which an EEG is notdiagnostic, TCD is particularly helpful in demonstrating the degree ofcerebrovascular compromise.26,27

Following a severe asphyxiating event, there may be an initial decrease inRI due to loss of autoregulation and resultant vasodilatation. As cerebraledema develops, the mean intracranial arterial pressure drops and vesselsdecrease in size, resulting in an increase in RI. Arrest of cerebral blood flowmay first occur at the microcirculation level. The larger vessels will distendor constrict and eventually thrombose or collapse. As ICP increases above meanarterial pressure, arrest of cerebral circulation will result in a reversal ofdiastolic blood flow and a decrease in antegrade systolic velocity. Small earlysystolic spikes and complete arrest of antegrade flow may then develop (figure10).26

Reversal of diastolic flow can be characteristic of essentially absenteffective cerebral circulation in the adult and in older children. However, ina few pediatric cases with mild diastolic reversed flow, recovery of forwarddiastolic flow and brain stem function has been described. Kirkham suggestedusing a direction of flow index (DFI=1 - maximum diastolic velocityarea/maximum systolic velocity area).28 In his series, all children withsubstantial diastolic reverse flow (DFI < 0.8) and a time averaged velocityof less than 10 cm/sec over a 30 minute period died without recovering brainstem function. Other patterns that have been described in brain death includesmall early systolic spikes and absent flow in the MCA with reversal ofdiastolic flow in the extracranial internal cerebral artery.

There has been some concern as to the reliability of TCD in the assessmentof infant brain death. In neonates, low RIs have been described in clinicallydead patients, while infants with an extremely high RI have survived.13,28-30Thus, careful clinical correlation is crucial in determining brain death, asarrest of supratentorial flow is not synonymous with brain death.

The TCD examination should never be used in isolation to supplant clinicalneurologic findings in children or neonates. Rather, it should be used toprovide data indicating the severity of cerebrovascular arrest. It may beprudent to repeat the study to confirm that cerebral blood flow arrest has beenpresent for a sufficient amount of time to cause irreversible damage to thesupratentorial structures.

Conclusion

There are many reasons why trans-cranial Doppler ultrasound is an effectivemodality for monitoring children. It is portable, noninvasive, and can berepeated as often as necessary in a relatively short amount of time, providinga useful adjunct in the clinical assessment of children with neurologicabnormalities.

References