Transcranial Doppler Sonography and Magnetic Resonance Angiography in the Assessment of Collateral Hemispheric Flow in Patients With Carotid Artery Disease
Background and Purpose The aim of this investigation was to compare the respective efficacy of transcranial Doppler sonography (TCD) and magnetic resonance angiography (MRA) for the assessment of intracranial hemodynamics in patients with extracranial occlusion or severe stenosis of the internal carotid artery (ICA).
Methods Twenty-five patients with unilateral ICA occlusion (n=20) or tight stenosis (n=5) demonstrated by duplex scanning or angiography were studied with both TCD and MRA. Three-dimensional time-of-flight MRA was used for the evaluation of extracranial-intracranial ICAs. Collateralization through the circle of Willis was investigated by means of selective two-dimensional MRA with presaturation of the carotid siphon, ophthalmic artery, or basilar artery. TCD was performed according to published standards: Anterior, middle, and posterior cerebral arteries were insonated through the temporal window, and carotid siphon and ophthalmic artery were assessed through a transorbital approach. Collateralization through the anterior circle of Willis was assumed if anterior cerebral artery flow was reversed, through the external carotid artery if ophthalmic artery flow was reversed, and through the basilar artery if the ratio of ipsilateral to contralateral posterior cerebral artery velocity was greater than 50%. TCD and MRA were performed by different investigators unaware of the results obtained with the other technique.
Results In every case time-of-flight MRA demonstrated the ICA occlusion or stenosis. There was an excellent correlation (κ=0.64) between TCD and MRA in assessing the hemodynamic contribution of the anterior part of the circle of Willis, whereas MRA was unable to detect the anastomotic pathway of the ophthalmic artery (κ=0.32). The contribution of the posterior communicating artery was difficult to assess with both techniques, but in three cases only MRA showed unequivocal evidence of collateralization. In three cases of middle cerebral artery stenosis TCD was superior to MRA in demonstrating the patency of the vessel.
Conclusions TCD and MRA should be considered complementary techniques. Combining the findings of both examinations may help to better understand the changes in intracranial hemodynamics produced by extracranial carotid occlusion. The contribution of the ophthalmic pathway, although important for the intraorbital structures, is probably of limited functional significance to the hemispheric blood supply.
In recent years, as a result of technological advancement, magnetic resonance angiography (MRA) and transcranial Doppler sonography (TCD) have been developed and have proved to be the two most useful techniques for the noninvasive assessment of intracranial hemodynamics. MRA is by far more expensive, is time-consuming, and requires at least some cooperation by the patient, but when technically satisfactory it can provide an anatomic display of cerebral vessels that is fairly accurate although still not comparable to that obtained with conventional angiography. In comparison, TCD is cheaper, relatively easy to perform, and suitable for real-time monitoring of blood flow velocities of cerebral vessels. However, its reliability is heavily dependent on the operator’s skill, it may be physically impossible to perform because of an impenetrable bony window, and so far the morphological imaging of the brain vessels, despite very recent advances in the software, has yet to become routinely available.1
Given these reciprocal limitations, we have tried to compare the respective efficacy of TCD and MRA in assessing the hemodynamics of intracranial vessels in patients with occlusion of the extracranial internal carotid artery (ICA).
Subjects and Methods
The characteristics of the study sample are shown in Table 1⇓. We studied 25 patients consecutively admitted to the Clinica Neurologica of the University of Brescia for either a transient ischemic attack or a stroke in the carotid territory and with occlusion of the symptomatic ICA at the neck demonstrated on duplex scanning or angiography. In 21 patients the symptomatic ICA was occluded, and in the remaining 4 it showed a greater than 90% stenosis. Regarding the asymptomatic ICA, total occlusion was detected in 2 patients and a stenosis greater than 90% in 2. These were all patients with occlusion of the symptomatic carotid artery. In the remaining patients the asymptomatic ICA was either normal or had less than 60% stenosis.
In 9 cases intra-arterial digital subtraction angiography with selective injection of the common carotid and vertebral arteries (IADSA) was available for comparison. In 11 additional cases IADSA was incomplete in the intracranial study. There was complete agreement in the 20 patients in whom angiography could be compared with duplex scanning of the neck vessels.
All MRA studies were performed on a 1.5-T system (Magnetom SP 63, Siemens Medical Systems). Time-of-flight (TOF) MRA was performed to obtain a combined anatomic-functional study of cerebral circulation. A circularly polarized head coil was used for both extracranial and intracranial circulation. Three-dimensional (3D) Fourier transform FISP gradient-echo sequences were used for the evaluation of the ICA from the bifurcation through the siphon with the following scan parameters: repetition time, 60 milliseconds; echo time, 7 milliseconds; flip angle, 20°; field of view, 250 mm; 64-mm coronal slab with 64 partitions, including vertebral and basilar arteries. For the baseline study of the anatomy of intracerebral vessels, two-dimensional (2D) TOF MRA (fast low-angle shot sequences; repetition time, 30 milliseconds; echo time, 10 milliseconds; flip angle, 30°; sequential acquisition of 3-mm-thick slices with 40% overlap) was preferred to 3D TOF MRA, at the expense of spatial resolution, for two reasons: (1) it is less sensitive to saturation of slow flow, which can represent a potential source of error in the identification of major vessel occlusion, and (2) it permits a considerable reduction of the examination time.
The evaluation of flow dynamics in the circle of Willis was obtained with 2D selective MRA, which consists of a combination of 2D techniques with radio frequency presaturation pulses.2 Ten-millimeter-thick presaturation bands are oriented along the course of the desired vessel to evaluate the dependent vascular territory. Presaturation causes a loss of flow signal intensity in the presaturated vessel and in its branches, including collateral flow, without affecting the vessels not coursing through the band.3 4 The middle cerebral artery (MCA) flow was visually assessed in the axial plane before and after presaturation of (1) the contralateral carotid siphon, for collateral flow through the anterior communicating artery (ACoA); (2) the ophthalmic arteries (OPHT), for collateral flow through the ipsilateral external carotid artery; and (3) the vertebral arteries, for collateral flow through the posterior communicating artery (PCoA). It was assumed that the presaturated vessel was carrying a significant portion of collateral flow if the signal of the MCA either disappeared or was significantly dimmed on visual inspection. Nineteen patients underwent a complete study, and in 16 of them it was possible to obtain a simultaneous presaturation of two or even all three putative collateral vessels. In the remaining 6 patients the study was incomplete and included the following: PCoA plus ACoA in 1 patient, OPHT plus ACoA in 2 patients, and ACoA alone in 3 patients. Therefore, the ACoA was assessed in all 25 patients, the OPHT in 21, and the PCoA in 20.
The imaging volumes were processed by a maximum intensity projection algorithm and reviewed together with the single slices to avoid loss of information due to artifacts in electronic reconstruction.
The Figure⇓ is an example of the usefulness of MRA for the noninvasive assessment of the anatomic variants of the circle of Willis.
As for TCD, the anterior cerebral artery (ACA), MCA, and posterior cerebral artery (PCA) were insonated through the temporal window, carotid siphon, and OPHT through the transorbital approach, according to published standards.5 6 Collateralization through the anterior part of the circle of Willis was assumed if ACA flow ipsilateral to the carotid occlusion was reversed (this usually occurred in conjunction with the acceleration of the contralateral ACA), through the external carotid artery if ophthalmic flow was reversed, and through the basilar artery if the ratio of ipsilateral to contralateral velocity in the PCA exceeded 50%.7 8 The criterion set for PCA was somewhat more conservative than others adopted in the literature and was based on our data for the normal population, which yield a PCA mean flow velocity of 42±11 cm/s and an average side-to-side asymmetry of 16±16% (G.P.A. et al, unpublished data, 1991). In 24 patients all three possible collateral sources were assessed; in 1 patient the ophthalmic flow could not be studied.
Carotid compression was performed on the asymptomatic side when deemed appropriate and safe. It was done in 3 patients in whom the asymptomatic ICA was undamaged and the collateral supply to the symptomatic hemisphere was carried by both the ACoA and the ipsilateral PCoA.
MRA and TCD were always done within 24 hours of each other and were performed an average of 18±7 days after the onset of symptoms. The results of one exam were unknown to the operator of the other exam.
We preliminarily assessed the concordance of the two techniques in the evaluation of the patency of the symptomatic MCA. Both techniques assessed the same number of normally patent MCAs (n=47; Table 2⇓). In contrast, in three cases of abnormality MRA tended to overestimate the degree of stenosis and incorrectly classified as occluded two MCAs that were actually stenosed (Table 2⇓), as subsequently confirmed by contrast angiography.
Overall the results of selective MRA hemodynamic evaluation were technically satisfactory for comparison with TCD in all 25 patients for the ACoA, in 21 for the OPHT, and in 20 for the PCoA (see “Subjects and Methods”). For the ACoA the two methods were concordant in 84% of cases (Table 3⇓). In 3 cases the reversal of flow in the ACA, as demonstrated by TCD, was apparently without functional significance, as presaturation of the contralateral siphon on MRA did not affect the signal in the MCA. In 1 patient TCD was unable to detect the reversal of flow in the ACA. For the PCoA the agreement was somewhat less satisfactory (80%) but still significant. Three of 4 discordant cases were due to the relative insensitivity of TCD in demonstrating the contribution of the PCoA. In contrast, for the OPHT the agreement in 62% of cases was counterbalanced by the disagreement in 38% (Table 3⇓). In all cases of mismatch, MRA failed to confirm the activation of the ophthalmic collateral pathway indicated by TCD.
Among the 9 patients in whom IADSA was also performed, TCD and MRA were concordant in 5 cases: Arteriographic findings confirmed TCD and MRA in 4 of these cases, but IADSA failed to demonstrate the contribution of the contralateral internal carotid system through the ACoA in 1 patient with a 95% stenosis of the right ICA. In 4 patients TCD and MRA were discordant: In 2 TCD was unable to show the PCoA pathway, in 1 it failed to appreciate the inversion of the flow in the ACA, and in 1 MRA missed the ophthalmic contribution. The IADSA findings were in agreement with MRA in the first 3 cases, but in the last they confirmed the relative insensitivity of MRA in imaging the ophthalmic collateral pathway. Furthermore, in 2 patients IADSA showed additional collateral pathways through pial anastomoses.
Finally, Table 4⇓ shows the distribution of the different patterns of collateralization as shown by TCD and MRA. In approximately 50% of cases two collateral pathways were simultaneously at work, as shown by both TCD and MRA. TCD indicated as equally frequent the activation of one or three collateral pathways (25% each), whereas according to MRA the single collateral pathway accounted for 31% of all patients compared with 11% for the three concomitant pathways. No collateralization was found in 1 patient by TCD and in 2 patients by MRA. The single most frequent collateral pathway was the ACoA, demonstrated in 71% of patients by TCD and in 73% by MRA.
The present study was aimed at comparing the type of information that can be collected by the two most useful noninvasive examinations available for the study of intracranial circulation. It was not a validation study in that the reference gold standard, ie, angiography, was available in less than half the cases, but it was meant to assess whether the type of information afforded by the two methods was comparable or rather gave two different views of brain circulation. In a sense, the answer to this question is neither positive nor negative. On the one hand, with respect to the patency of MCA, the two techniques are quite interchangeable in their apparent accuracy in assessing the normality of the vessel. On the other hand, when the MCA lumen is reduced as a result of either local thrombosis or recanalizing embolus, TCD seems more sensitive in detecting the residual patency of the vessel. However, it must be remembered that for TCD the diagnosis of MCA occlusion relies on the lack of MCA signal together with the reliable appreciation of ACA and PCA signals, according to the accepted parameters of normality for the latter vessels.8 It is clear that in conditions of altered intracranial hemodynamics the ultrasound identification of the single vessel may be complicated by alterations in direction and velocity of flow. Particularly in those cases in which the TCD diagnosis of MCA occlusion is uncertain, MRA can become essential for diagnosis because it provides a direct visualization of the cerebral vessels.
TCD and MRA showed excellent correlation in assessing the contribution of the anterior circle of Willis and the high prevalence of the ACoA (>70%) as a collateral pathway (Tables 3⇑ and 4⇑). The three cases in which MRA failed to appreciate the contribution of ACoA were those in which the ICAs were bilaterally occluded (two cases of bilateral thrombosis, one of unilateral thrombosis with contralateral >90% stenosis) and the blood supply to the brain was actually carried by the basilar or the external carotid artery. This explains the apparent discrepancy of the findings: The blood flow in the A1 tract of the ACA was actually reversed on one side, but this was without hemodynamic consequences because the main blood supply came from alternative sources.
The correlation between TCD and MRA in the appreciation of the contribution of posterior-to-anterior anastomoses was slightly less promising than for the anterior circle of Willis (84% versus 80%, respectively). This we could attribute to the relatively arbitrary criterion that we had adopted to assume the posterior contribution (>50% increase in ipsilateral PCA velocity compared with contralateral). Given the variability of diameter of the posterior vessels, this criterion is likely to have reduced the sensitivity of TCD with respect to MRA, as TCD failed to recognize the contribution of PCoA in three cases, which was confirmed by angiography in two patients. In summary, TCD actually underestimated the anastomoses between carotid and basilar arteries. The adoption of a more refined criterion, such as the addition of the systematic compression of vertebral arteries, might improve the accuracy of TCD in the future.
Finally, it is apparent from Table 3⇑ that the assessment of collateral flow through the OPHT represents the major mismatch between the two techniques. In all discordant cases TCD showed flow reversal in the OPHT, suggesting the activation of the physiological external-internal carotid artery bypass, whereas MRA presaturation of the OPHT failed to affect the signal in the MCA. Therefore, it is clear that either TCD overestimates or MRA underestimates the contribution of the ipsilateral compensatory pathway. One intrinsic limitation of TCD is that the reversal of flow in the OPHT actually represents the reversal of the pressure gradient between external and internal carotid artery territories, but is no proof by itself that the external carotid artery provides an effective collateralization to the ipsilateral hemisphere. It is thus likely that at least in some patients, ophthalmic reversal, although necessary for the blood supply to the intraorbital structures, especially the eye, may have little functional significance for the hemispheric blood flow. It is perhaps worth noting that in all but one of the eight cases in which TCD indicated the activation of external-internal carotid artery bypass and MRA was negative, at least one other collateral pathway was at work (ACoA four times and PCoA twice, both in one case). Moreover, as shown in Table 4⇑, the MCA was recanalized by the OPHT alone in only 2 of 18 cases of ophthalmic reversal. This agrees with those studies that have reported on the activation of ophthalmic collaterals in patients with reduced cerebral perfusion pressure but not in those with normal cerebrovascular reserve.9 10 On the other hand, the disappearance of the MRA signal in the MCA after presaturation of the putative collateral vessel does not completely rule out the possibility that other sources carry a small fraction of flow, as MRA technology is still insufficient in imaging very slow flow. If, for instance, both the ACoA and OPHT contribute to the blood supply of one MCA, but the former is quantitatively preponderant, MRA may fail to appreciate the ipsilateral contribution simply because it is unable to image the residual slow flow that remains after presaturation of the contralateral siphon. A very recent study on intracranial collateral blood flow in patients with ICA occlusion, the design of which is similar to the present report, came essentially to the same conclusion as to the inadequacy of MRA in imaging of ophthalmic collaterals.11
In conclusion, our findings suggest that TCD and MRA are complementary techniques. The former is probably more sensitive in depicting all the potential sources of collateralization, and the latter is more specific in delineating which source, among multiple possible pathways, has the greatest functional importance.
- Received July 7, 1994.
- Revision received November 7, 1994.
- Accepted November 8, 1994.
- Copyright © 1995 by American Heart Association
Bogdahn U, Becker G, Winkler J, Greiner K, Perez J, Meurers B. Transcranial color-coded real-time sonography in adults. Stroke. 1990;21:1680-1688.
Edelman RR, Mattle HP, O’Reilly GV, Wentz KU, Cheng Liu, Bin Zhao. Magnetic resonance imaging of flow dynamics in the circle of Willis. Stroke. 1990;21:56-65.
Spencer MP, Whisler D. Transorbital Doppler diagnosis of intracranial arterial stenosis. Stroke. 1986;17:916-921.
Schneider PA, Rossman ME, Bernstein EF, Ringelstein EB, Otis SM. Noninvasive assessment of cerebral collateral blood supply through the ophthalmic artery. Stroke. 1991;22:31-36.
Schneider PA, Rossman ME, Bernstein EF, Torem S, Ringelstein EB, Otis SM. Effect of internal carotid artery occlusion on intracranial hemodynamics: transcranial Doppler evaluation and clinical correlation. Stroke. 1988;19:589-593.
Powers WJ, Press GA, Grubb RL, Gado M, Raichle ME. The effect of hemodynamically significant carotid artery disease on the hemodynamic status of the cerebral circulation. Ann Intern Med. 1987; 106:27-35.
Norrving B, Nilsson B, Risberg J. CBF in patients with carotid occlusion: resting and hypercapnic flow related to collateral pattern. Stroke. 1982;13:155-159.