Basilar Vasospasm Diagnosis
Investigation of a Modified “Lindegaard Index” Based on Imaging Studies and Blood Velocity Measurements of the Basilar Artery
Background and Purpose— Numerous studies have shown that cerebral vasospasm is one of the leading causes of death and neurological disability after subarachnoid hemorrhage. Most of these studies, however, have focused on anterior circulation vessels. Since the introduction of the transcranial Doppler (TCD), increasing attention has been given to basilar artery (BA) vasospasm, especially in traumatic subarachnoid hemorrhage. As shown for the anterior circulation, however, the significance of elevated flow velocities (FVs) in the posterior vessels may be ambiguous, so vasospasm may not be reliably differentiated from hyperemia. The purpose of the present study was to evaluate the potential additional value of an intracranial/extracranial FV ratio in the posterior circulation to cope with this shortcoming of the TCD in the diagnosis of BA vasospasm.
Methods— FV in the extracranial vertebral artery (VA) was measured in 20 healthy volunteers. Normative values of an intracranial/extracranial VA FV ratio (IVA/EVA) and a BA/extracranial VA FV ratio (BA/EVA) were calculated. Thirty-four patients with subarachnoid hemorrhage were then evaluated with TCD and CT angiography (CTA). The value of the IVA/EVA and BA/EVA ratios in the diagnosis and assessment of vertebrobasilar vasospasm was investigated.
Results— The extracranial VA could be insonated in all subjects at depths ranging from 45 to 55 mm. The average FV for the extracranial VA was 26 cm/s. The ratios between intracranial and extracranial VA FVs were 1.6 on both sides, whereas the ratio between the BA FVs and the mean extracranial VA FVs was slightly higher at 1.7. Fourteen patients (41.2%) had CTA evidence of BA vasospasm. Vasospasm was severe in 7 patients, moderate in 1, and mild in the remaining. An FV threshold of 80 cm/s was indicative of BA vasospasm in 92.8% with 3 false-positive results that could be related to vertebrobasilar hyperemia. Comparative analysis between CTA and TCD findings showed that BA/EVA was >2 in all patients with BA vasospasm (100% sensitivity) and <2 in all but 1 patient without BA vasospasm (95% specificity). Furthermore, the BA/EVA ratio showed a close correlation with BA diameter (r=−0.8139, P<0.0001) and was >3 in all patients with severe vasospasm.
Conclusions— The results of the present study showed that the BA/EVA ratio may contribute to an improved discrimination between BA vasospasm and vertebrobasilar hyperemia and enhance the accuracy and reliability of TCD in the diagnosis of BA vasospasm. Our data further suggest that the BA/EVA ratio may provide an approximation of vasospasm severity and help in identifying patients who are likely to suffer from hemodynamically significant vasospasm.
- basilar artery
- cerebral angiography
- subarachnoid hemorrhage
- tomography, x-ray computed
- ultrasonography, Doppler, transcranial
Since its introduction by Aaslid, the transcranial Doppler (TCD) has emerged as an accurate, noninvasive, and effective tool for the diagnosis of cerebral vasospasm.1 Since the pioneer report on the use of TCD for the evaluation of patients suffering from subarachnoid hemorrhage (SAH) by Aaslid et al,2 numerous studies have provided substantial evidence and conclusive clinical data relating the elevation of
See Editorial Comment, page 78
blood flow velocities (FVs) in the main basal arteries of the brain and cerebral vasospasm.3,4 Increased blood FVs, however, may not necessarily imply arterial narrowing. Indeed, both increasing flow and reduced vessel diameter may lead to high flow velocities. Consequently, cerebral vasospasm may not be safely and definitely differentiated from cerebral hyperemia by the mere assessment of FV in the basal arteries.3–5 To cope with this diagnostic shortcoming of TCD, Lindegaard et al3 suggested the use of the ratio between flow velocities in the middle (MCA) or anterior (ACA) cerebral arteries and the internal carotid artery (ICA). In a convincing report comparing the results of angiographic and TCD studies in patients with SAH, Lindegaard et al3 were able to improve significantly the diagnostic accuracy of TCD in cerebral vasospasm whenever an MCA/ICA threshold of 3 was used as diagnostic criterion. Moreover, in the same report, these authors showed that the likelihood and severity of cerebral vasospasm further increased as the MCA/ICA ratio increased. The MCA/ICA ratio has since been used in most clinical studies on the implementation of TCD for the diagnosis of cerebral vasospasm. These studies, however, as well as that of Lindegaard et al,3 focused on the anterior circulation vessels and did not address the vertebrobasilar system. In a comprehensive angiographic and ultrasonographic study, Sloan et al6 first attempted to define TCD criteria for the diagnosis of vertebrobasilar vasospasm. Using a criterion of 60 cm/s, they reached a 76.9% sensitivity level with 21.7% false-positive results, whereas vertebrobasilar vasospasm could be found in all patients with FV >95 cm/s. In their report, however, the authors did not address either the diameter or the FVs in the extracranial portion of the vertebral arteries (VAs).
In this study, we present a comparative study between FVs recorded along the vertebrobasilar axis and the findings of vascular neuroimaging studies in an attempt to refine the TCD criteria for the diagnosis of vertebrobasilar vasospasm after SAH.
Subjects and Methods
Twenty healthy volunteers had a thorough clinical examination, excluding history or evidence of neurological or vascular disease. There were 11 men and 9 women whose ages ranged from 19 to 46 years (mean, 33.5±7.8 years). All had TCD examination according to the technique described below.
Thirty-four consecutive patients suffering from SAH were included in this study. There were 20 men and 14 women, ranging from 18 to 74 years of age (mean, 44.7±16.1 years). SAH was spontaneous (sSAH) in 25 patients, posttraumatic (tSAH) in 8 patients, and postoperative in 1 patient. Clinical data are summarized in Table 1.
In all patients, fluid regimen was aimed at the maintenance of mean arterial pressure at 95 to 100 mm Hg and hematocrit at 30% to 35%. Nimodipine (2 mg/h) was administered routinely only in the sSAH group. In the tSAH group, only patients with TCD evidence of vasospasm for ≥48 hours and jugular bulb oxymetry <70% were treated with nimodipine. Mechanical ventilation, sedation, and intracranial pressure monitoring were performed as indicated. Increased intracranial pressure was treated by hyperventilation guided by jugular bulb oxymetry and intravenous infusion of mannitol 20% and propofol. Propofol and atracurium were selected for ventilation and intracranial pressure control because of their rapid onset and offset, allowing neurological evaluation as needed.
TCD evaluations were performed with an Intraview system (Rimed Inc) with a 2-MHz pulse-waved range-gated transducer according the technique described by Aaslid et al.1 Insonation of the VA was performed from above the posterior cervical triangle. Assessment of the extracranial VA at the first cervical vertebra level was performed at a depth ranging from 45 to 55 mm, whereas the intracranial portion of the vessel was evaluated at ≥70-mm depth. Basilar artery (BA) location was defined at an insonation depth >80 mm according the technique described by Fujioka and Douville.7 FV in the BA was recorded at ≥3 different locations, the highest of which was located >90 mm. Two distinct intracranial/extracranial flow velocity ratios were then calculated: 1=highest recorded BA FV over the averaged extracranial VA FV (BA/EVA ratio), and 2=highest recorded intracranial VA FV over the ipsilateral extracranial VA FV (IVA/EVA ratio). TCD criteria for vasospasm were defined by mean FVs >120 cm/s and ≥3-fold that of the internal carotid artery for anterior vessels and >80 cm/s for the vertebrobasilar system.
In the patient group, TCD recordings were performed within the first 48 hours after admission and thereafter every day until the patient’s discharge or TCD stabilization at normal FV values.
Neurological status was evaluated on admission by means of Hunt and Hess score8 after resuscitation measures and offset of sedative drugs. During the hospitalization course, the Hunt and Hess score was recorded at least every 8 hours.
SAH severity was assessed on admission in each patient by use of the classification described by Fisher et al9: grade 1, normal CT scan; grade 2, diffuse SAH or thin vertical layer <1 mm thick; grade 3, localized clots or vertical layer of subarachnoid blood >1 mm thick; and grade 4, intraventricular or intracerebral hemorrhage.
All patients had high-resolution CT angiography (CTA) with maximum intensity projection reconstruction of the vertebrobasilar system. In sSAH patients, CTA was performed within 48 hours of admission regardless of TCD findings as part of the routine evaluation. In all patients, however, CTA was also obtained to achieve a definite diagnosis when TCD showed sustained elevated FVs for ≥48 hours in the basal arteries, suggesting vasospasm. Forty-one CTA studies were performed in 34 patients. In 5 instances, a second study was indicated by clinical and TCD findings suggestive of symptomatic vasospasm; in 2 patients, however, a second CTA was performed as a follow-up evaluation. In all 7 patients, the CTA selected for analysis was the one demonstrating the smallest BA diameter.
CTA studies were performed with an MX8000 scanner (Marconi) with 1-mm-thick sections and a 0.6-mm section gap. A 250-mm field of view was used with a matrix of 512×512. All CTAs were independently reviewed by a neuroradiologist blinded to the ultrasonographic data. Measurements included the extracranial portion of the VA at the C1 level, the intracranial portion of the VA, and the BA. For each location, the value selected for analysis was the average of 3 different measurements performed at the narrowest diameter to obtain the best correlation with the highest FV. BA vasospasm was considered mild when the BA diameter was <3 mm, moderate when it was <2.5 mm, and severe when it was <2 mm.
Both VAs and BAs could be successfully insonated in all subjects. The posterior cervical triangle provided practical access to the extracranial portion of the VA at an insonation depth ranging from 45 to 55 mm. The average FV for the extracranial VA was 26 cm/s (Table 2). The ratios between intracranial and extracranial VA FVs were 1.6 on both sides; the ratio between the BA FVs and the mean extracranial VA FVs was slightly higher at 1.7.
Of 34 patients, 14 (41.2%) had CTA evidence of basilar vasospasm. Basilar vasospasm was severe in 7patients, moderate in 1 patient, and mild in 6 patients. Associated VA vasospasm was found in 13 (38.2%) of these patients. Vertebral vasospasm was bilateral in 11 patients and unilateral in the remaining 2 patients. VA vasospasm was mild in 3 patients, moderate in 4, and severe in 6. Consequently, the average diameter of the BA and VA in its intracranial portion was significantly smaller in patients with elevated FVs in the vertebrobasilar system compared with the remaining patients (Table 2). On the contrary, the average diameter of the extracranial portion of the VA was very similar in both patient subgroups (Table 2). As detailed earlier, 7 of 14 patients had 2 CTA studies. In all 7 patients, the BA diameter was significantly larger in the CTA made remote from the vasospasm peak: BA diameter was in the normal range in 4 patients, mildly reduced in 2 patients, and moderately decreased in 1 patient.
FVs in the extracranial VA were similar in patients with and without CTA evidence of vasospasm, although they were moderately elevated in patients as a group compared with healthy subjects (Table 3). On the contrary, both BA FV (100.3±26.6 cm/s) and intracranial VA FVs (85.5±33.3 cm/s and 89.5±34.5 cm/s) were markedly increased in patients with BA diameters <3 mm compared with FVs in the control group and in SAH patients without BA vasospasm on CTA (Table 3). In 13 of 14 patients with BA vasospasm, BA FV was >80 cm/s (sensitivity, 92.8%). In contrast, 3 patients with BA FV >80 cm/s had no CTA evidence of BA vasospasm (specificity, 85%). In 2 patients, CTA disclosed an arteriovenous malformation supplied by the posterior cerebral artery. The third patient was admitted because of severe head injury and had suggestive evidence of cerebral hyperemia, ie, increased intracranial pressure, reduced arteriovenous difference on jugular bulb oxymetry, and elevated FVs in both internal carotid arteries. As a result, positive and negative predictive values of elevated FVs in the vertebrobasilar system were 81.2% and 94.4%, respectively.
Within the patient group, BA FVs showed a satisfactory correlation with the BA diameter measured on CTA (r=−0.74, P<0.0001; Figure 1). Intracranial VA FVs demonstrated a similar correlation with the diameter of the intracranial portion of the VA on CTA (r=−0.73, P<0.0001; Figure 1).
Intracranial/Extracranial FV Ratios and Vasospasm
Despite a slight increase in FVs in the vertebrobasilar vessels, the BA/ECVA and ICVA/ECVA FV ratios remained practically unchanged in SAH patients without CTA evidence of vasospasm compared with healthy subjects. On the contrary, the BA/ECVA FV ratio showed a 58.8% increase in patients with BA diameter <3 mm (2.8±0.7, P<0.0001; Table 3). Moreover, this increase in the BA/ECVA FV ratio showed an almost linear correlation with vasospasm severity (P<0.0001; Figure 2). Similarly, the ICVA/ECVA FV ratio was 56.3% (left) and 43.8% (right) higher in patients with vertebrobasilar vasospasm than in those without vasospasm and in the control group.
Comparative analysis between CTA and TCD results showed that the BA/ECVA FV ratio was >2 in all patients with BA vasospasm (sensitivity, 100%), whereas all but 1 patient without a significant reduction of the BA diameter had a BA/ECVA FV ratio below the same threshold (specificity, 95%). Positive and negative predictive values of elevated FVs in the posterior vessels were therefore improved to 93.3% and 100%, respectively. Similarly, the BA/ECVA FV ratio resulted in a substantial enhancement of the linear correlation between CTA and TCD findings (r=−0.8139, P<0.0001; Figure 3). This correlation could be further improved by relating the BA/EVA FV ratio to the BA diameter expressed as a percentage of the averaged extracranial VA diameter instead of the BA diameter. Such a calculation is likely to minimize the influence of anatomic variations in the size of the VAs. These variations may affect the intrinsic value of the BA diameter and raise the need for a correction of threshold values for BA vasospasm. Indeed, the BA/EVA FV ratio showed a very tight correlation with the percentage of diameter reduction from the extracranial VA to the BA (r=0.8715, P<0.0001; Figure 3).
Despite the early study of Echlin10 based on an animal model of traumatic SAH, BA vasospasm has been widely neglected by most authors in favor of anterior circulation vessels, and very few clinical observations have been reported. Until the introduction of the TCD, cerebral vasospasm was evidenced only by cerebral angiography, so it was related to aneurysm rupture in most instances and consequently involved mainly anterior vessels. Only sporadic studies have addressed the issue of BA vasospasm with special reference to tSAH.11 Since these studies, TCD has allowed the diagnosis of vasospasm at bedside and extended the field of investigation of SAH to head injury in which angiographic studies are seldom performed.12–20 These studies showed that posttraumatic vasospasm was more common than previously thought, involving prominently the vertebrobasilar system.12,13,18 After the comprehensive report of Sloan et al6 defining TCD criteria for the diagnosis of BA vasospasm, few studies have provided circumstantial evidence linking elevated FVs in the vertebrobasilar system with poorer neurological outcome in patients suffering from SAH, especially secondary to severe head injury.12,13,18 Combining blood flow studies and TCD, Lee et al13 showed that posttraumatic BA vasospasm correlated significantly with poorer neurological outcome. More recently, Soustiel et al18 reported on TCD findings in a large cohort of SAH patients and further supported the hypothesis that BA vasospasm may be responsible for secondary brain damage, especially after tSAH. In both series, however, elevation of BA FVs alone could not reliably predict a poor neurological outcome on an individual basis, so the clinical significance of elevated blood FVs in the VA and the BA remains to be clarified. In this regard, several studies have emphasized the difficulty in relying only on TCD to distinguish between cerebral hyperemia and vasospasm, especially in head injury. In particular, Lee et al13 showed that increased FVs in the BA were associated with reduced blood flow in only 71%, whereas in the remaining patients (29%), there was evidence for cerebral hyperemia.
In the present study, BA vasospasm was present in 41.2% of the investigated patients. This incidence is in close accordance with the findings of our previous study but is significantly higher than that reported by Sloan et al.6 A plausible explanation for this discrepancy is that the angiography performed by Sloan et al was at a time not related to TCD findings. In contrast, elevated BA FVs sustained for ≥48 hours was considered in the present study a prerequisite for an initial CTA to be performed in head-injured patients or a follow-up examination in sSAH patients. This may, in turn, account for the higher sensitivity of TCD in the diagnosis of BA vasospasm in the present series compared with that of Sloan et al,6 as also assumed by these authors in the discussion of their own results. Conversely, 3 patients with elevated BA FVs had no CTA evidence of vasospasm. An arteriovenous malformation supplied by the posterior cerebral artery could be demonstrated in 2 patients, whereas the third patient had clinical and laboratory findings suggestive of posttraumatic cerebral hyperemia. As a result, the specificity of elevated FVs as a diagnostic criteria of BA vasospasm was lower than that reported by Sloan et al6 (85%), enhancing the challenge of a clear distinction between cerebral hyperemia and vasospasm relying only on BA FVs. Interestingly, there was no arteriovenous malformation or tSAH in the patients investigated by these authors, so the 100% specificity that they found may be at least in part the consequence of a selected series. Similarly, TCA was performed in tSAH patients only whenever indicated.
As demonstrated by Lindegaard et al3 for the anterior circulation, the use of an intracranial/extracranial FV ratio significantly improved the accuracy of TCD in the diagnosis of BA vasospasm and reduced the number of false-positive results. The use of a threshold value of 2 for the BA/ECVA FV ratio yielded 100% sensitivity and 95% specificity with only 1 false-positive result in a patient with ruptured arteriovenous malformation. Particularly interesting is the case of the head-injured patient suffering from cerebral hyperemia. This patient had very high BA FVs (93 cm/s) for several days after admission because of diffuse brain injury. Such a high BA FV may be considered diagnostic according to the criteria of Sloan et al6 and therefore may justify special therapeutic measures such as triple-H therapy in the absence of CTA that eventually ruled out BA vasospasm. FVs in both extracranial VAs, however, were found to be significantly elevated (54 and 44 cm/s) at the same time with a BA/ECVA ratio of 1.9 precluding the diagnosis of vasospasm.
Another important factor that may challenge the clinical implications of TCD findings is the absence of a definite correlation between FVs and the diameter of the vasospastic artery as demonstrated by imaging studies. In the series of Sloan et al,6 patients with moderate BA vasospasm had higher BA FVs as a group than patients with severe BA vasospasm, whereas 2 of 5 patients in the latter subgroup had BA FVs <80 cm/s.6 In the present series as well, despite the improvement achieved in the correlation between BA FVs and BA diameter by a TCD-oriented timing for CTA evaluation, BA FVs alone would not prevent substantial individual errors in the estimation of the vasospasm severity. This shortcoming of TCD may be even more distressing when considering the absence of a linear correlation between arterial narrowing and cerebral blood flow in the affected vessel. In an early report on carotid stenosis, Moore and Mallone21 found that poststenotic flow was compromised only in the presence of severe diameter reduction. More recently, experimental studies based on phantom models of progressive arterial narrowing showed that flow distal to stenosis was almost unaffected by a diameter reduction <50% to 60%.21,22 In this regard, the BA/ECVA FV ratio proved to enhance significantly the correlation between TCD and anatomical findings, therefore allowing firmer conclusions about vasospasm severity. But because no blood flow measurements were included in this study, statements on brainstem ischemia relying on TCD estimation of BA diameter will remain speculative. This may be even more relevant when taking into consideration the imaging modality selected for this study. Indeed, even with high-resolution CT, precise measurement of vessel diameter remains challenging at <0.2 mm, so the categorization of some patients might have been erroneous. Nevertheless, as suggested in Figure 3, a BA/ECVA threshold value of 3 would accurately delineate patients suffering from high-grade BA vasospasm (≥50% diameter reduction). As such, the BA/ECVA ratio may therefore contribute to the differential diagnosis between hemodynamically significant BA vasospasm and nonsymptomatic BA narrowing. This assumption, also advocated by Lindegaard et al,3 has since been supported by measurements of cerebral blood flow made in tSAH patients with elevated MCA FVs and high MCA/ICA ratios.5,23
In conclusion, the BA/ECVA ratio proved to enhance the clinical value of TCD in BA vasospasm. According to the results of the present study, the BA/ECVA ratio may improve the accuracy of TCD in the diagnosis of vertebrobasilar hyperemia, especially in SAH caused by head injury and ruptured arteriovenous malformation, and provide reliable data on vasospasm severity.
We thank Hava Wolf, Shmuel Weizman, and Pesah Ladoviz, technologists in the CT unit, for their outstanding technical support and cooperation.
- Received April 20, 2001.
- Revision received September 4, 2001.
- Accepted September 18, 2001.
Sloan MA, Burch CM, Wozniak MA, Rothman MI, Rigamonti D, Permutt T, Numaguchi Y. Transcranial Doppler detection of vertebrobasilar vasospasm following subarachnoid hemorrhage. Stroke. 1994; 25: 2187–2197.
Fujioka KA, Douville CM. Anatomy and freehand examination.In: Newell DW, Aaslid R, eds. Transcranial Doppler. New York, NY: Raven Press; 1992: 9–32.
Soustiel JF, Levy E, Bibi R, Lukaschuk S, Manor D. Hemodynamic consequences of cerebral vasospasm on perforating arteries: a phantom model study. Stroke. 2001; 32: 629–635.
The Need for a Quantifiable Normalized Transcranial Doppler Ratio for the Diagnosis of Posterior Circulation Vasospasm
In clinical practice, one often applies certain knowledge without going back to the articles where such “knowledge” originated. This is undoubtedly the case for many who regularly use the “Lindegaard index,” without ever having seen the original publication by Lindegaard and coworkers.1 For the writing of this editorial comment, I looked up the article, and frankly, I do not recall having seen it before! Such may also be the fate of the present article: in my opinion, the authors have provided us with a “tool” that will be used by many clinicians who have learned to take the transcranial Doppler (TCD) measurement results into account in their practice. In fact, the need for a normalized value, rather than the absolute flow velocities, is much greater in the posterior than the anterior circulation for a few key reasons.
First, the clinical signs of vasospasm in the anterior circulation are, in most cases, fairly easy to recognize. When the clinical impression is supported by increasing blood velocities or simply by a single measurement showing high velocity and a Lindegaard index above 3, the clinician can (start to) treat the patient for vasospasm. In the posterior circulation, however, the clinical signs are much subtler and nonspecific, and thus we have had to rely more on nonclinical parameters, such as TCD, which until now were rather unreliable.
Second, there is a patient population much larger than that with aneurysmal subarachnoid hemorrhage (SAH), ie, those with severe head injuries. For these patients, basilar artery vasospasm is of paramount prognostic (and probably clinical) significance: Approximately 50% of patients with severe head injuries have SAH, and in half again of those, basilar artery vasospasm develops. In patients with basilar vasospasm, unfavorable outcome is twice as frequent as in those without spasm.2 It is important to realize that these data in head injury were generated in the UCLA Brain Injury Research Center, an extremely specialized, well-funded unit, with scores of personnel, Doppler units, 133Xenon CBF machines, positron emission tomography scanning, etc, and there is no way this can be done in regular, nonresearch clinical practice.
Third, in head-injured patients vasospasm often occurs on posttrauma day 2 or 3, at which time most patients are chemically paralyzed and ventilated, making clinical recognition near impossible. With the method proposed here, it should be relatively simple for anyone with some experience with TCD to make the diagnosis of (isolated) basilar artery vasospasm, and maybe even treat it!
I have no idea where the term “Lindegaard index” originated. The present authors, modestly, do not propose to name this new index the “Soustiel index.” In their original publication, Lindegaard coined the term “VMCA/VICA ratio.” Similarly, one could use the term “VBA/VECVA ratio”; “posterior circulation Lindegaard index” or “posterior Lindegaard index” would be acceptable as well. After acceptance of this commentary by the editor of Stroke, I gave the manuscript to my TCD operators and asked them to try it (find the extracranial VA) on some patients and volunteers. They succeeded in only half of the cases. More experience, I assume, is needed to approach the 100% success rate cited by the authors. The TCD operators, without any suggestion from my side, termed it “VBA/VECVA ratio.” Only time will tell which term prevails, but that the authors have written a “classic” seems almost certain to me.
Lindegaard KF, Nornes H, Bakke SJ, Sorteberg W, Nakstad P. Cerebral vasospasm after subarachnoid hemorrhage investigated by means of transcranial Doppler ultrasound. Acta Neurochir. 1998; 42 (suppl): 81–82.
Lee JH, Martin NA, Alsina G, McArthur DL, Zaucha K, Hovda DA, Becker DP. Hemodynamically significant cerebral vasospasm and outcome after head injury: a prospective study. J Neurosurg. 1997; 87: 221–233.