Article Text
PDF
Abstract
Introduction Endovascular treatments such as transluminal balloon angioplasty and intra-arterial nimodipine represent rescue therapy for cerebral vasospasm (CVS) after aneurysmal subarachnoid haemorrhage (SAH). Both indication and data regarding its efficacy in the prevention of cerebral infarct are, however, inconsistent. Therefore, an MR based perfusion weighted imaging/diffusion weighted imaging (PWI/DWI) mismatch was used to indicate this treatment and to characterise its effectiveness.
Methods MRI was performed for suspicion of CVS. For quantitative evaluation, the brain was partitioned into 19 arbitrary segments of comparable volume. Segments with PWI/DWI mismatch were defined as ‘segment at risk (SR)’. In these cases, MRI was followed by angiography (digital subtraction angiography (DSA)) including endovascular treatment. 48±12 h after treatment, a second MRI was performed and the treatment was repeated if new or remaining SR were observed. Efficacy was classified as the percentage of reduced diameter of the proximal cerebral arteries on DSA following the treatment: mild (≥33%), moderate (34–66%) or severe (≥67%).
Results 48 treatment cycles, each consisting of MRI, DSA and a second MRI, were performed in 25 patients. During these cycles, 95 SR were identified. The infarct rate was significantly higher in SR (37%) compared with segments without risk (4%). The infarct rate in SR was significantly reduced if mild proximal CVS could be achieved. In the case of persistent severe CVS, infarcts occurred in all SR.
Conclusion The present series suggests that PWI/DWI mismatch is predictive of the development of infarct in the case of CVS. The infarct rate could, however, be improved if proximal CVS was sufficiently reduced.
Statistics from Altmetric.com
Introduction
Delayed cerebral vasospasm (CVS) after aneurismal subarachnoid haemorrhage (SAH) was classically defined as the narrowing of the large capacitance arteries of the brain which results in reduced cerebral blood flow (CBF), ischaemic neurological deficits and, finally, cerebral infarction.1 2 Typical onset of CVS occurs 3–5 days after SAH, reaching a maximum after 5–14 days and it is resolved after 2 –4 weeks.3 4 Current investigations suggest, however, more complex mechanisms involving a disturbed microcirculation or electrocortical phenomena in addition to the spasm of the capacitance arteries.5 6 The common denominator of the definition of CVS is, therefore, currently the manifestation of delayed ischaemic neurological deficits after SAH which ends in cerebral infarction in 15–20% of patients after SAH, even in contemporary series.7–9
The pathophysiological background leading to CVS is still poorly understood. Several promising approaches to specific prophylaxis of CVS, such as the application of clazosentan,10 were, however, reported recently. In spite of these options, delayed ischaemic lesions due to CVS were still observed in 5–13%10 of patients receiving such prophylactic treatment.
In the case of manifest CVS, the only evidence based conventional treatment is the so called haemodynamic therapy which includes the manipulation of blood pressure, volume and viscosity in order to improve CBF.3 Complication rates between 20% and 30% occur, however, during this treatment.4 In the remaining patients with severe CVS resistant to conventional treatment, endovascular therapy such as transluminal balloon angioplasty (TBA) or selective intra-arterial application of vasodilators, seem to be the last rescue option to avoid cerebral infarction.3 Accordingly, the more or less successful treatment of severe CVS by TBA,11 application of vasodilators12–14 or a combination of both15–17 was reported in several case series. Mainly after the use of vasodilators, the effect was temporary and multiple interventions were necessary in some cases. The indications and efficacy control of this endovascular therapy, however, remain inconsistent in the literature.11–17 The aim of the present investigation was, therefore, to evaluate the efficacy and feasibility of an endovascular treatment with a prospectively defined protocol including indication, efficacy control and, if necessary, repetition of the interventional procedure. The indication was based on the tissue at risk concept18 which is defined as perfusion weighted imaging/diffusion weighted imaging (PWI/DWI) mismatch in MRI and is widely used for the definition of the therapeutic target in ischaemic stroke.19–21
Patients and methods
Patient screening and standard treatment
Between January 2006 and November 2007, 187 patients with aneurismal SAH, clinical grade 1–4 according to Hunt and Hess, were admitted to our department.
After admission, digital subtraction angiography (DSA) was performed, and the aneurysm responsible for the bleeding was occluded by operative or endovascular treatment. Twenty-four to 48 h after this treatment, periprocedural lesions were detected or excluded by CT scan. Subsequently, all patients were treated using a standardised protocol in an intensive care unit. The protocol included prophylactic oral (6×60 mg/day) or intravenous (2 mg/h) nimodipine until day 21, maintenance of normovolaemia (central venous pressure 5–10 cm H2O) and normotonia (mean arterial blood pressure (MAP) 70–80 mm Hg). Transcranial ultrasound Doppler sonography (TCD) was performed daily. In patients under anaesthesia, intracranial pressure was monitored, and cerebral perfusion pressure (CPP) was maintained between 60 and 70 mm Hg.
Suspicion of CVS was defined as any form of neurological deterioration, a TCD mean velocity ≥150 cm/s or an increase in velocity ≥50 cm/s in 24 h. If neurological deterioration was observed, detailed clinical examinations, blood tests and, if necessary, an EEG were performed to rule out other causes of the deterioration. Within 2 h after suspicion of CVS, patients were investigated by MRI, including DWI and PWI, to exclude further causes of clinical deterioration or to confirm impaired CBF. Patients without suspicion of CVS who were not completely alert received MRI routinely on day 7±2 after the ictus. A DWI/PWI mismatch with a PWI delay ≥2 s compared with the contralateral side was defined as manifest impaired CBF which is adopted from the initial tissue at risk definition after SAH.18 In patients with suspicion of CVS and/or manifest impairment of CBF, haemodynamic therapy was initiated aiming at a MAP of 100–120 mm Hg or, in patients with ICP monitoring, CPP of 90–110 mm Hg.11 22 Subsequently, they were included in the present investigation.
Endovascular treatment and DSA protocol
The protocol of the present study is summarised in figure 1 and was approved by the local ethics committee. Immediately after impaired CBF was proven by MRI, DSA was performed. Angiographic CVS was evaluated by diameter changes of the proximal cerebral artery segments and the cerebral circulation time (CCT). CCT was determined as the time taken for a contrast medium bolus between the arterial intradural inflow at the level of the carotid siphon and filling of the bridging veins. Proximal angiographic CVS was classified by the diameter reduction of one or more proximal artery segments compared with the baseline DSA: mild (≤33%), moderate (34–66%) or severe (≥67%).23 The following segments were classified: the proximal part of the anterior, middle and posterior cerebral artery, the intradural internal carotid artery and the basilar artery.
Flowchart of the diagnosis and treatment protocol of cerebral vasospasm (CVS) in the present case series. In any suspicion of CVS or routinely on day 7±2, an MRI, including perfusion weighted imaging (PWI) and diffusion weighted imaging (DWI), was performed. In the event of PWI/DWI mismatch (‘tissue at risk’) the MRI was directly followed by digital subtraction angiography (DSA) and an endovascular treatment of any severe CVS by transluminal balloon angioplasty (TBA), nimodipine or a combination of both. The effect of this treatment was controlled 48±12 h later by MRI. If tissue at risk was detected again, the treatment cycle consisting of MRI, DSA (including endovascular treatment) and a second MRI (48±12 h) was repeated until further tissue at risk could be excluded. CCT, cerebral circulation time; DIND, delayed ischemic neurological deficit; H&H, Hunt and Hess; SAH, subarachnoid haemorrhage; TCD, transcranial Doppler sonography.
Invasive endovascular treatment was performed only in the case of proven impaired CBF on MRI and either severe proximal angiographic CVS or a CCT ≥5.5 s.23 24 Whenever obtainable, proximal CVS was treated by TBA, as described previously,11 24 and distal or diffuse CVS by intra-arterial application of nimodipine. Nimodipine was selectively applied to the affected vascular territory. An initial bolus of 0.8 mg was followed by an intra-arterial infusion of 4 mg for 1 h. At the end of the procedure, DSA was repeated to control the efficacy of the endovascular treatment. The response of CVS to the intervention was determined by classification of proximal CVS before and after treatment, and by CCT, as described above. DSAs were read out for classification by the first and senior author. A follow-up MRI was taken 48±12 h after endovascular therapy. If the tissue at risk was proven again, a second DSA, including endovascular therapy, was performed. This treatment cycle was repeated until remaining tissue at risk could be excluded by MRI. Thereafter, haemodynamic therapy was tapered.
Efficacy assessment of the endovascular treatment and MRI protocol
A standardised MRI protocol on a 1.5 T (Magnetom Vision, Siemens, Germany) was performed, including axial T2 weighted images (WI), T2* weighted images and T2 fluid attenuated inversion recovery sequences. In addition, all patients received PWI and biplane DWI. For DWI, a single shot echoplanar imaging spin echo sequence was used. Acquisition of bolus tracking PWI was performed with a gradient echo echoplanar imaging sequence, as described previously.23
For the assessment of the DWI and PWI features, the territories of the cerebral arteries were divided into segments, as demonstrated in figure 2. The middle cerebral artery was divided into four segments—the anterior cerebral artery and the posterior cerebral artery in two segments, respectively. The infratentorial vertebrobasilar territory was divided into a brainstem and two cerebellar segments. Accordingly, the cerebrum was partitioned into 19 segments. This segmentation was used to define regions of interest in DWI and PWI. In any of the defined segments, the time to peak (TTP) value in PWI was determined. A 50% DWI lesion ≥1 segment was defined as major infarct (MI); <50% was defined as minor ischaemic lesion. A direct correlation of a TTP delay in the MRI before and a DWI lesion in the MRI after endovascular treatment could thus be achieved for each segment.
For the placement of regions of interest (ROI) for determination of the perfusion weighted imaging (PWI) and to provide a regional correlation of an initial PWI delay and the diffusion weighted imaging (DWI) lesion, the territories of the major cerebral arteries were divided into segments: the middle cerebral artery territory into four segments (MA–D), the anterior cerebral artery (AA–B) and the posterior cerebral artery territory (PA–B) into two segments, respectively. The vertebrobasilar (VB) territory was divided into brainstem (VBA) and cerebellar (VBB) parts.
The indication for endovascular treatment was detection of at least one ‘segment at risk (SR)’. This was defined as a TTP delay ≥2 s in one segment compared with the contralateral side, the exclusion of MI and relevant CVS on DSA in correlation with the supplied territory.18 In the posterior fossa, the regions of interest for comparison of the PWI delay were placed outside the vertebrobasilar territory in segments of the anterior cerebral artery and middle cerebral artery. The effectiveness of endovascular treatment was assessed by the SR developing MI or minor ischaemic lesions compared with the segments without infarcts which were defined as ‘recovered tissue’.
A clinical follow-up examination, including grading of patients using the Karnofsky Performance Score and the modified Rankin scale were done 6±2 month after SAH.
Statistical analysis
Statistical significance of different infarct rates in SR versus ‘segments not at risk (SnoR)’, mild, moderate or severe angiographic CVS, and CCT ≤5.5 s after interventional treatment were tested by Fisher's exact test. A p value <0.05 was considered significant.
Results
A series of 25 patients, aged 50±10 years (mean±SD; range 27–66), were included in the present investigation. Of these, 16 were women and nine were men. Hunt and Hess grade was 3±0.86 (median±mean deviation from the median); 23 of the patients were Fisher grade 3 in the initial CTS and the remaining two patients were Fisher grade 4. The aneurysm responsible for the SAH was treated by clipping in 10 and by coiling in 14 patients. In one case, no initial aneurysm treatment could be achieved. Clinical baseline data are given in table 1.
Clinical baseline data and outcome in 25 patients with cerebral vasospasm
Forty-eight treatment cycles, each consisting of initial MRI, DSA including interventional treatment and follow-up MRI, were performed. Details of the treatment modalities, classification of angiographic CVS, effect of endovascular treatment and comparison of the infarct rate in SR to SnoR are given for each cycle in table 2. Twelve patients received one cycle, seven patients two cycles and six patients between three and five cycles. Accordingly, in 13 of 25 patients (52%), the effect of the endovascular intervention on impaired CBF was only transient. Seventeen patients were permanently under anaesthesia during the investigation. The remaining eight patients developed delayed ischaemic neurological deficits. In six, the symptoms were progressive confusion or reduced consciousness, and anaesthesia was necessary for their further treatment. A reasonable and reproducible neurological assessment was possible in only two patients. Accordingly, the neurological state for the monitoring of CVS could only be used in three of the 48 treatment cycles.
Modalities and effect of endovascular treatment on angiographic cerebral vasospasm, tissue at risk and infarcts
During the treatment cycles, 95 SR were identified and treated subsequently. Treatment modalities were: TBA only, one intervention; selective intra-arterial nimodipine only, 34 interventions; combination of both, 13 interventions. A sufficient MAP or CPP (as defined above) was maintained in all patients. In three cases, however (patient Nos 5, 13 and 24) complications occurred leading to one minor and three major segment lesions which were compromised in the infarct rate in the SR. In patient No 5, two MI and in patient No 13 one minor ischaemic lesion was caused by embolic complications. In patient No 24, a reperfusion bleeding resulted in one major segment lesion.
Overall, 32 (3.5%) segments with minor ischaemic lesions and 36 (4%) with MI were observed. Comparison of the infarct rate in SR to SnoR is shown in figure 3 and details are given in table 3. From 912 scanned segments, 95 (10%) were identified as SR and 817 (90%) as SnoR. The proportion of SR was independent of the number of treatment cycles. The rate for both minor ischaemic lesion and MI was, however, significantly higher in SR (p<0.001) in spite of the interventional therapy (table 3, figure 3). Overall, in 33 of 817 (4%) SnoR and in 35 of 95 (37%) SR, infarcts were detected. Accordingly, any DWI lesion could be prevented in 60 of 95 (63%) SR and MI in 77 of 95 (81%) SR.
Comparison of the infarct rate in segments at risk versus not at risk. The infarct rate in segments at risk was significantly higher in major and minor infarcts compared with segments not at risk. *p <0.0.5 versus the infarct rate in segments not at risk.
Comparison of the number and percentage of infarcts in segments with and without perfusion weighted imaging/diffusion weighted imaging mismatch (segments at risk/ not at risk)
Before intervention, proximal angiographic CVS was classified as mild in four, moderate in 15 and severe in 29 cases. Angiographic improvement was achieved in 39 cycles. Accordingly, angiographic CVS was classified as mild post-interventionally in 35, moderate in 12 cases and severe in only one case. The CCT was improved to <5.5 s in 39 cases but remained extended after nine endovascular treatments. The infarct rate depending on the effect of the endovascular treatment on the proximal CVS and CCT in DSA is given in table 4 and figure 4. In all groups except those with moderate proximal CVS and normalised CCT after intervention, the infarct rate was significantly higher in SR compared with SnoR (table 4). Only in the SR in this group, however, was the infarct rate significantly lower in the case of a CCT ≤5.5 s compared with >5.5 s. A remaining severe proximal CVS after intervention was only observed in one cycle in which the CCT was also not improved (table 4, figure 4). Compared with this group (severe proximal CVS, CCT >5.5 s), the overall and major infarct rate in SR was significantly reduced in mild proximal CVS (with and without CCT improvement) and in moderate CVS, but only with normalised CCT (table 4, figure 4). In SnoR, the overall and major infarct rate of all groups was also significantly lower compared with that in severe proximal CVS.
Comparison of the infarct rate depending on the response to endovascular treatment
Comparison of the infarct rate depending on the effect of endovascular treatment. Proximal cerebral vasospasm (CVS) after treatment was classified in mild, moderate or severe. Each of the proximal CVS grades was subdivided into cerebral circulation time (CCT) <5.5 s (yes) or not (no). +p<0.05 versus the severe CVS group; #p<0.05 versus CCT ≤5.5 s in the group of the same proximal CVS grading.
In five of the included patients, any delayed DWI lesion could be prevented, in 11 patients only minor lesions occurred and nine patients suffered from MI caused by CVS. Clinical outcome together with baseline data are given in table 1. A favourable outcome (modified Rankin scale ≤2 and Karnofsky Performance Score ≥70) was achieved in 15 (60%) patients. Of patients receiving one cycle, nine of 12 (75%) had a favourable outcome; of those receiving two cycles, three of seven (43%) had a favourable outcome; and of those receiving multiple cycles, three of six (50%) had a favourable outcome.
Discussion
Several case series reported previously on the treatment of severe CVS by TBA,11 vasodilators12–14 or a combination of both.15–17 The efficacy of the interventional treatment was, however, mainly evaluated by the effect on angiographic vessel diameter and to a lesser extent on the prevention of cerebral infarction or improved clinical outcome. Furthermore, indications for endovascular treatment of CVS are inconsistent in the literature and sometimes not even comprehensible.11–17 The present investigation is, therefore, the first to employ the MRI based ‘tissue at risk’ concept18 for the indication, re-evaluation and efficacy control of this invasive endovascular therapy.
The tissue at risk concept using PWI/DWI mismatch as an estimation for the penumbra in ischaemic stroke has been proposed since the late 1990s.19 According to the findings that early DWI lesions may be partly reversible in ischaemic stroke25 and SAH,26 and that PWI may overestimate the penumbra under some circumstances, this mismatch hypothesis remains unproven19 and a consistent definition27 is lacking. It is, however, a widely accepted practical tool for the indication of recanalisation after ischaemic stroke,19–21 which was used in an increasing number of clinical stroke trials recently.28 29 The transfer of the tissue at risk concept for the indication of endovascular treatment of CVS seems, therefore, to be obvious.18 30
Diagnosis of impaired CBF in good time to prevent cerebral infarction remains challenging in patients after SAH. Patients with higher neurological grades have an increased risk of developing CVS.31 In these patients in particular, CBF cannot be sufficiently monitored by neurological assessment, and CVS is usually screened by TCD. Increasing flow velocities are used as surrogate for reduced CBF in these cases.3 The positive predictive value of this method for the development of delayed ischaemic neurological deficits is, however, limited to less than 50%.3 32 TCD may, therefore, not be a convenient single tool for the indication of an invasive procedure such as DSA and endovascular treatment. Accordingly, it was used in the present investigation for the screening of CVS which had to be confirmed by PWI before the indication for an invasive treatment was assumed.
For evaluation, TTP values were used because correlating maps can be rapidly assessed after PWI, and a good correlation of TTP values with cerebral perfusion measurements by positron emission tomography was proven.33 Several previous investigations using the TTP measurement suggested a smaller threshold for a perfusion delay in the case of CVS compared with thrombotic or arteriosclerotic ischaemia.11 23 34 35 Accordingly, an infarct was observed in any territory with a TTP delay of more than 3.3 s during CVS,11 whereas in arteriosclerotic or thrombotic strokes, a lesion enlargement occurred mainly at a delay of more than 4–6 s.34 35 In applying these observations, including an additional safety zone, a TTP delay of ≥2 was used for the definition of tissue at risk in the present investigation. The high infarct rate in the SR compared with the relatively low one in the SnoR confirms the validity of the defined threshold in the present data.
A possible background for the lower threshold compared with arteriosclerotic ischaemia may be the complex pathophysiological mechanisms during CVS,3 involving a diameter reduction of the proximal arteries and additional disturbance of the microcirculation.36 37 These mechanisms may lead to bilateral disturbance of CBF during CVS, which is accentuated in one or more vascular territories, resulting in the development of ischaemic stroke.36 37 The definition of tissue at risk based on the contralateral side may, therefore, even underestimate the maximum extension of the reduced CBF. The overall infarct rate in SnoR was, however, only 4% in the present series. Accordingly, a predictive value of 95% could be calculated statistically for missing of any infarct in previously detected SnoR. This confirms the diagnostic validity of the tissue at risk concept in spite of the semiquantitative nature of PWI in MRI.
TBA represents an established therapy in the case of CVS resistant to conventional treatment. Accordingly, a permanent effect on proximal angiographic CVS15 17 24 and the CBF of the supplied brain tissue has been proven.11 15 17 TBA is, however, limited to proximal vessels and is an invasive method with a complication rate that is not negligible.16 38 Furthermore, new infarcts were observed in the dependent vascular territory in spite of a successful TBA in between 7%16 and 32%.11 Ischaemic lesions independent of the vascular territory were even observed in 40%24 to 75%.39 We, therefore, performed a combination of TBA and intra-arterial application of nimodipine in the present investigation. Nimodipine is a widely used vasodilator for the intra-arterial treatment of CVS.12 13 40 A significant improvement of angiographic CVS was observed in 43–90% of patients.12 13 Duration and efficacy of this effect was, however, variable.12 13 Accordingly, the data regarding the efficacy of endovascular treatment of severe CVS are inconsistent. Furthermore, the retrospective matched pair analysis of the tirilazad study yielded no significant effect of interventional treatment in the prevention of infarcts caused by CVS or an improved clinical outcome.39
In the present series, severe proximal CVS resistant to interventional treatment resulted in prolonged CCT in all cases, infarcts in all SR and infarcts in nearly half of the SnoR. The infarct rate was significantly reduced if proximal CVS was converted to mild CVS, independent of the CCT. In the case of moderate proximal CVS, the infarct rate was, however, significantly lower if the CCT was ≤5.5 s compared with a prolonged CCT and severe proximal CVS. Therefore, the data from the present series suggest that the development of ischaemic lesions can be significantly reduced by interventional treatment if severe proximal CVS or moderate proximal CVS with an enhanced CCT can be eliminated. The present data thus support the pathophysiological concept that infarction due to CVS is chiefly a haemodynamic problem.3 Impaired CBF may, however, be caused by a combination of vasospasm of the proximal cerebral arteries and microcirculatory problems. Ischaemic lesions may only occur if the compensatory possibilities of both mechanisms are exhausted.
The restriction of a case series like the present one is obviously—lack of a control group—and therefore there is some difficulty in the quantitative estimation of the efficacy of this therapeutic approach. The present data, however, support the diagnostic validity of the tissue at risk concept18 and the therapeutic effectiveness of the repetitive use of a combined endovascular treatment of severe CVS based on this concept.
Conclusion
The first new aspect of the present series is the demonstration of the predictive value of a defined DWI/PWI mismatch for the development of delayed infarcts in the case of CVS after SAH. Furthermore, it provides evidence for the effectiveness of an invasive interventional treatment of CVS, including repetitive MRI and DSA. Without doubt, such complex, time and resources consuming treatment options have to be confirmed in prospective randomised studies. Accordingly, the data from the present series could support the design of such an investigation to establish this type of rescue therapy in the clinical routine.
Acknowledgments
The authors are grateful to Marina Eberhardt for excellent technical support and to Anne Sicking and Johannes Otto for grammatical help.
References
Footnotes
Competing interests None.
Ethics approval This study was conducted with the approval of the ethics committee of Goethe-University, Frankfurt, Germany.
Provenance and peer review Not commissioned; externally peer reviewed.