Although cerebral haemorrhage represents only 10%-15% of all strokes, it has been estimated to cause more than 50% of the overall stroke mortality (Bamford et al. 1990). The 30-day mortality rate at around 45% is similar for intracranial and subarachnoid haemorrhage (SAH) (Broderick et al. 1993). SAH represents one quarter to one third of all cerebral haemorrhages (Nilsson et al. 2000). The more recent studies have estimated an annual rate of SAH between 6 in North America and 10 in Europe per 100,000 population (Broderick et al. 1993; Nilsson et al. 2000). Women are affected about twice as often compared with men (Broderick et al. 1993; Nilsson et al. 2000). Approximately 75% of spontaneous SAHs are caused by intracranial aneurysms (Nilsson et al. 2000).
In a recent population-based European study, 16% of the patients with SAH died before reaching the hospital. Another 5% deceased at the day of haemorrhage after admission (Pobereskin 2001). Among patients surviving the initial haemorrhage treated without surgery, rebleeding is the major cause of morbidity and mortality. The risk is between 15% and 20% within the first two weeks. The goal of early surgery is to reduce this risk. Of those reaching neurosurgical care, secondary deterioration caused by delayed ischaemic neurological deficits (DINDs) is assumed to be the predominant complication after SAH. In the international cooperative study on the timing of aneurysm surgery, DINDs were responsible for disability in 6.3% and mortality in 7.2% among patients with SAH (Kassell et al. 1990a). The recent randomised, double-blind, vehicle-controlled trials of tirilazad mesylate found a rate of 33% to 38% for DINDs in all patients of the vehicle group. Cerebral infarctions occurred in 10% to 13% (Haley et al. 1997; Lanzino and Kassell 1999; Lanzino et al. 1999).
DINDs are clinically characterised by confusion or decreased level of consciousness with a focal neurological deficit. The diagnosis is one of exclusion. In contrast to the usual clinical presentation of stroke, symptoms typically develop gradually, waxing and waning over several hours. Increased headache and meningismus, low-grade fever, seizures, and a disturbance of consciousness are usually followed by the appearance of focal neurological symptoms that typically reflect ischaemia in the anterior circulation. Characteristic signs include mental status changes, abulia, akinesia, anosagnosia, Witzelsucht, aphasia, hemiparesis, paraparesis or urinary incontinence (Adams and Love 1992).
It has been hypothesised that intracellular products of red blood cells induce DINDs (MacDonald and Weir 1991). This hypothesis has been based on: (a) the temporal correlation between the occurrence of DINDs (day 4 - 12 after the haemorrhage, maximum at day 7) and haemolysis in the subarachnoid space (Pluta et al. 1998), and (b) the correlation between the amount of subarachnoid blood in the initial computed tomogram (CT) and the risk to develop a DIND (Fisher et al. 1980; Kistler et al. 1983; Brouwers et al. 1993). Up to 70% of patients develop angiographically visible delayed arterial spasm after SAH (Brouwers et al. 1993). While the absence of angiographic spasm has a high negative predictive value, the positive predictive value of moderate/severe angiographic spasm for the development of a DIND is only between 33% and 50% (Vora et al. 1999; Unterberg et al. 2001). This positive predictive value is less than that of a coin toss suggesting additional pathogenetic factors.
Most “classical” vascular risk factors seem to increase the likelihood for a DIND. [Seite 3↓] Thus, in cigarette smokers, a slightly higher risk for the development of DINDs has been observed. Weir et al. (1998) analysed 3500 patients from five multi-centre trials and found an odds ratio of 1.2 (95% confidence interval 1.1-1.4). A higher prevalence of hypertension in patients developing a DIND was suggested in case-control studies (Brandt et al. 1991; Öhman et al. 1991). Even though age was negatively correlated with angiographic spasm, the incidence of DINDs was either similar to that observed in younger patients (Fortuny et al. 1980) or increased with advancing age (Lanzino et al. 1996). This is possibly related to the finding that the subarachnoid clot is thicker in elderly patients due to parenchymal atrophy allowing for a larger quantity of blood to collect (Inagawa et al. 1988; Sakaki et al. 1989). In addition, the prevalence of hypertension is higher in the elderly. As another risk factor for the evolution of a DIND, cocaine use was identified (Conway and Tamargo et al. 2001). To my knowledge, a particular study related to diabetes and DIND has not been performed, yet. A causal link between hyperglycaemia and outcome after DIND was not established in the Cooperative Aneurysm Study (Lanzino et al. 1993).
Only the gender relation for DINDs is clearly different from that of most other variants of ischaemic stroke, which show a significant male preponderance. In contrast, the incidence of DINDs was either similar in men and women with SAH (North American Study) (Kongable et al. 1996) or significantly higher in women (European Studies) (George et al. 1989; Rosenorn et al. 1993). DINDs share this epidemiological feature with migraine-related stroke (Tzourio et al. 2000). However, although migraine, particularly migraine with aura, has been established as a risk factor for ischaemic stroke in all recent population- and hospital-based studies (reviewed by Tzourio et al. 2000), it has not been investigated so far whether the coexistence of migraine is linked to an increased risk for DINDs. To my knowledge, there are also no epidemiological studies regarding female hormones and incidence of DINDs.
Maximal frequency of re-bleeding after aneurysmal SAH is in the first day (4% on the first day, then 1.5% daily for 13 days). Fifteen to 20% re-bleed within the first two weeks (Winn et al. 1977). The risk was neither altered by blood pressure on admission nor site of the aneurysm (Inagawa et al. 1987). Early surgery prevents re-bleeding.
Acute hydrocephalus was observed on admission CT in 15% of SAH patients with 40% of these being symptomatic (Graff-Radford et al. 1989). Factors assumed to contribute to acute hydrocephalus include: blood interfering with cerebrospinal fluid circulation through the Sylvian aqueduct, fourth ventricle outlet, subarachnoid space, and/or with re-absorption at the arachnoid granulations. Chronic hydrocephalus is due to pia-arachnoid adhesions or permanent impairment of the arachnoid granulations. Acute hydrocephalus does not inevitably lead to chronic hydrocephalus. Cerebral lesions typically associated with hydrocephalus are located at the ventricle poles.
Hyponatraemia commonly occurs following SAH. Although it had been attributed to a rise in antidiuretic hormone (ADH) producing hypervolaemic hyponatraemia, the ADH increment is usually transient, lasting for only about 4 days. Another theory was based on the fact that there is often a delayed peak in atrial natriuretic factor (ANF), a 28-amino acid [Seite 5↓] polypeptide, after an initial smaller rise (Wijdicks et al. 1991). Increased ANF secretion leads to a urinary loss of sodium (cerebral salt wasting syndrome) resulting in hypovolaemic hyponatraemia. Although cerebral salt wasting syndrome has clearly been shown to be the cause of hyponatraemia in the majority of these patients (Harrigan 1996), it has remained unclear whether ANF is the operative natriuretic factor in SAH (Kröll et al. 1992). The neurological effects of hyponatraemia can mimic DINDs.
A particular problem in studying DINDs is the differentiation from peri- or intra-operative complications associated with aneurysm clipping. Direct or indirect surgical complications have a different pathogenesis from DINDs and, hence, are likely to be different in pathoanatomy of the cerebral lesions, age and gender distribution, risk factor profile and response to medical treatment. In other words, surgically related ischaemic infarcts represent an important confounding factor in studies on DINDs. Unfortunately, in many studies on DINDs, surgical complications were not adequately controlled. E.g., Millikan, in 1975, expatiated upon his experience from a series of 198 patients that “any diffuse or focal brain damage, which occurs after the first few hours following the bleed or particularly following a neurosurgical procedure, is automatically explained as being due to vasospasm”. This clinical problem is further complicated by the interaction between delayed vasospasm with parenchyma partially damaged during the surgical procedure since partial damage may result in increased vulnerability. That interaction might also result in delayed neurological deficits although the pathogenesis would be different from that of DINDs in non-operated patients. Serial magnetic resonance imaging (MRI) could help to differentiate between surgically related and unrelated lesions when images taken at the second post-operative day are [Seite 6↓] compared with those at, e.g., day 12 after the haemorrhage. Unfortunately, the serial MRI study of Shimoda et al. (2001) is inconclusive in this respect. Concerning aneurysm closure and its ischaemic consequences in MR images, the recent, comparative study of aneurysm coiling or clipping by Hohlrieder et al. (2002) is interesting. Of the 112 patients treated after aneurysm rupture, 57% suffered from ischaemic infarcts. However, also the 32 patients in whom the aneurysm was electively coiled or clipped showed secondary cerebral infarcts in 41%.
Direct intra-operative complications resulting in ischaemia are, e.g., spatula pressure, stenosis of the aneurysm-bearing vessel by the surgical clip around the aneurysm neck, generalised brain swelling or aneurysm rupture. The most important surgical complication is probably thrombosis occurring post-operatively and requiring post-operative control angiography to be distinguished from a DIND. E.g., Proust et al. (1995) investigated 230 consecutive patients with ruptured aneurysm. Whereas post-operative thrombosis occurred in 25 cases (11%), DINDs caused only 15 secondary complications in this study (7%). Similar percentages for post-operative thrombosis were reported in other studies with systematic control angiography (11% in a series of 100 subjects [Creissard et al. 1990], 12% in a series of 66 subjects [MacDonald et al. 1993] and 11% of 63 patients who died after aneurysm surgery [Karhunen et al. 1991]), whereas the incidence was considerably lower in studies without systematic control angiography (3% in the International Cooperative Study on Aneurysm Surgery [3521 patients, Kassell et al. 1990b] and 4% in a series of 150 patients [Gilsbach et al. 1988]). That comparison suggests that post-operative thrombosis is the most important confounding factor for DINDs in studies without systematic control angiography.
In summary, the differentiation between DINDs and surgically related ischaemic infarcts is difficult and controversial. Only for ischaemic lesions in the cortex associated with [Seite 7↓] DINDs, it was shown that surgery was not a confounding factor (Neil-Dwyer et al. 1994). Most surgically related complications including post-operative thrombosis will produce territorial infarcts.
The incidence of different variants of ischaemic stroke is significantly increased for any kind of surgery during the post-operative period from the hospitalisation until 30 days after surgery. This may be related to changes in coagulation, bed rest, medication, diagnostic tests etc. (Wong et al. 2000). In addition, it is likely that the rate of stroke caused by cardiac embolism is particularly increased after SAH since SAH is well known to produce cardiac arrhythmias (reviewed by Sakr et al. 2002). However, the role of cardiac embolism, atherosclerosis or other frequent aetiologies of ischaemia for ischaemic deficits after SAH has never been adequately studied to my knowledge. These aetiologies may particularly contribute to ischaemic complications after SAH in the elderly. The majority of these aetiologies, in particular cardiac embolism, will produce territorial infarcts.
Robertson performed the first pathoanatomical study on DINDs in 1949. A microscopically detected infarct of motor cortex in case number 8 of the original publication laid the foundation of the hypothesis that ischaemic changes can occur remote from the territory of the artery bearing the aneurysm. From a large autopsy series, Falconer reported similar findings in 1954. As a typical example, he presented a case of a non-operated patient with a left anterior cerebral artery aneurysm in whom a cortical necrosis was detected in the territory of the left middle cerebral artery. Birse and Tom, in 1960, presented a detailed [Seite 8↓] histological analysis of these widespread ischaemic lesions after SAH in 8 selected autopsy cases of non-operated patients. They provided clear diagrams emphasizing the characteristic focal cortical distribution of the infarcts, which were often arranged around a fissure or a sulcus (Fig. 1A and B ). Deeper structures of the brain showed small ischaemic patches. Stoltenburg-Didinger and Schwarz published a large autopsy study of operated and non-operated patients in 1987. They analysed 207 cases after aneurysmal SAH collected from 1969 to 1985. In 106 of 139 patients properly examined microscopically, either bell-shaped or laminar infarcts in the cortex were a constant finding. The large majority of these cortical infarcts occurred in areas covered with a subarachnoid clot. Of the 156 patients not undergoing surgery, only 9 demonstrated a larger infarct in the territory of one of the three major cerebral arteries (6%). In contrast, 13 of the 51 patients in whom a clipping of the ruptured aneurysm was performed, showed a territorial infarct (26%) (Chi-Square-Independence-test, p < 0.0001). This indicated some pathogenetic role of the surgical intervention for a considerable number of territorial infarcts. Neil-Dwyer and colleagues, in 1994, performed a prospective autopsy study including 53 cases. This study aimed to differentiate between the lesions induced by the initial haemorrhage and that by delayed ischaemia. A significant correlation was confirmed between delayed ischaemia and cortical ischaemic lesions. Interestingly, moderate to severe cortical ischaemic lesions were also detected in 56% of patients without evidence of vasospasm in the angiogram.
A and B
The upper two pictures are diagrammatic representations of macroscopic and microscopic cerebral infarction in a 37-year old man with DIND at day 8 after subarachnoid haemorrhage. The DIND was characterised by a rapid deterioration with development of paraplegia. The patient died 4 days later. The pictures are taken from a paper of 8 cases selected from a larger autosy series by Sheila Birse and Mary Tom published in Neurology in 1960.
The lower two pictures are turbo inversion recovery magnitude magnetic resonance images from a 33-year old man who developed a DIND with mental status changes and right-sided hemiparesis about 6 days after subarachnoid haemorrhage (Dreier et al. 2002b). The patient was treated in the intensive care unit with a good recovery. In both patients, the aneurysm was un-operated at the time point when the pictures were made so that the lesions represent the typical pattern occurring in the natural course of the disease. The most characteristic feature of the lesions is their distribution in the cortex band around a sulcus (arrows in A) or a fissure (arrows in B). In both patients, the aneurysm was located at the anterior communicating artery in which case the blood is typically pushed upwards along the interhemispheric fissure.
C A similar focal laminar cortical necrosis is observed after cortical spreading ischaemia in response to the red blood cell products haemoglobin and increased subarachnoid K+ in the rat (*). The bar illustrates size and position of the cranial window. It is unclear why the lesion can extend beyond the borders of the window. Likely, haemoglobin and K+ have diffused to areas neighbouring the window (Dreier et al. 2000).
A recent systematic review of all randomised controlled clinical trials on Ca2+ -antagonists in patients with SAH (2756 patients in total) confirmed a significant reduction in frequency of poor outcome by nimodipine. This resulted from a reduction in the occurrence of DINDs (Feigin et al. 1998). Nimodipine prevents 1 of 3 bad outcomes due to a DIND. One of 7 patients develops a bad outcome due to a DIND after SAH (Kassell et al. 1990a). Hence, about 21 patients have to be treated with nimodipine to prevent one bad outcome after SAH. This beneficial effect was shown for oral nimodipine. It is unclear why the more expensive intravenous nimodipine is routinely used in many centers.
At the cellular level, nimodipine directly inhibits slowly inactivating voltage-sensitive Ca2+ channels (L-type). Blockade of Ca2+ entry from the extracellular space into vascular smooth muscle leads to cerebral vasodilatation and increases cerebral perfusion (Kazda and Towart 1982; Andersson et al. 1983; Haws et al. 1983). It was proposed that Ca2+ antagonists may reverse chronic vasospasm of the major cerebral arteries (Allen and Bahr 1979). However, no significant effect on cerebral vasospasm was angiographically detected following nimodipine treatment in several randomised clinical trials (Feigin et al. 1998). In addition, chronic vasospasm induced by autologous blood in the subarachnoid space did not resolve in response to high doses of nimodipine in primates (Espinosa et al. 1984; Nosko et al. 1985; Lewis et al. 1988). Haemoglobin is assumed to play a major role in the pathogenesis of the large artery spasm (MacDonald and Weir 1991; Pluta et al. 1998). However, nimodipine poorly antagonised haemoglobin-induced arterial constriction in different species such as monkey, dog and man (Nosko et al. 1986).
These clinical and experimental findings suggested that the anti-ischaemic effect of [Seite 11↓] nimodipine may be related to a direct protection of neurons through blockade of excitotoxic Ca2+ entry (Brandt et al. 1988; Pisani et al. 1998). However, if a cytoprotective effect on neurons was the key function of nimodipine, the drug would also act on other types of ischaemia than DINDs. In several animal studies of focal cerebral ischaemia, nimodipine was not consistently effective (Langley and Sorkin 1989; Scriabine et al. 1989). In a recent meta-analysis, no evidence was available to justify the use of nimodipine in patients with ischaemic stroke unrelated to SAH (Horn and Limburg 2001).
The risk reduction over the last 20 years to develop delayed ischaemia has not only been related to the prophylactic use of nimodipine but also to an improved fluid management. In approximately one-third of the patients after SAH, excessive natriuresis and intravascular volume contraction occurs (van Gijn and Rinkel 2001). In the past, hyponatraemia was erroneously attributed to water retention. Therefore, fluid restriction was applied which was later found to increase the risk for DIND. Two non-randomised studies with historical controls suggested that a daily fluid intake > 3 l of saline (against 1.5 – 2 l in the past) was associated with a lower rate of DINDs (Hasan et al. 1989; Vermeij et al. 1998). In addition, a randomised study was performed comparing controls and patients treated with volume expansion using albumin and crystalloid to obtain a haematocrit of 45% (Rosenwasser et al. 1983). In this small trial, moderate hypervolaemic haemodilution resulted in a significant reduction of DINDs. A more intense prophylactic volume expansion with elevated cardiac filling pressures was not superior to a moderate volume expansion (Lennihan et al. 2000; Egge et al. 2001). Hyponatraemia per se does not increase the risk of a DIND (Qureshi et al. 2002).
In patients with ischaemic stroke unrelated to SAH, haematocrit levels were significantly higher than in matched controls (Toghi et al. 1978; Harrison et al. 1981). However, in contrast to the benefit of patients with DIND, a meta-analysis did not find any evidence that moderate volume expansion improved outcome (Asplund et al. 2001).
Megyesi et al. (2000) counted 57 different SAH models in animals to study the arterial spasm. All known models share the feature that DINDs or ischaemic lesions are essentially never observed despite the presence of significant arterial spasm (Megyesi et al. 2000). This, once again, suggests that the arterial spasm alone is probably not sufficient to induce DINDs. That DINDs do not occur in animal SAH models in contrast to SAH in man may be due to the better collateralisation in smaller mammals. However, the main reason is probably related to the clinical observation that the amount of blood in fissures and cisterns in the initial CT after SAH is well correlated with the incidence of DINDs (Fisher et al. 1980; Kistler et al. 1983; Brouwers et al. 1993). This suggests that a considerable quantity of blood has to collect to induce a DIND. It is obvious that the subarachnoid space of man can accommodate a significantly larger volume of blood than that of the species used for SAH models (rat, rabbit, cat, pig, dog, smaller primates). The observation by Stoltenburg-Didinger and Schwarz (1987) that the lesions typically occurred in areas covered with subarachnoid clot indicated a direct effect of the red blood cell products on the underlying cortex in addition to inducing arterial spasm at the base of the brain. A possibility to circumvent the problem of the limited subarachnoid space in the rat and to model the accumulation of red blood cell products at the cortical surface is to directly superfuse these products. With aid of this approach, we have demonstrated that red blood cell products are indeed able to produce ischaemia and ischaemic [Seite 13↓] cortical lesions (Fig. 1C ) (Dreier et al. 1998; 2000). Relevance of this model for the clinical condition was supported by the therapeutic effect of nimodipine and increase of the systemic volume, respectively (Dreier et al. 2002a).
|© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.|
|DiML DTD Version 3.0||Zertifizierter Dokumentenserver|
der Humboldt-Universität zu Berlin
|HTML-Version erstellt am: