of Stroke

Pathogenesis of Stroke

Arterial Ischemia

Pathophysiology

Classification of Arterial Stroke

Microembolic Signals

Spontaneous Microemboli

Detection of Microemboli in Patent Foramen Ovale

Venous Ischemia

The pathophysiologic correlate of cerebral ischemia is the inadequate delivery of glucose and oxygen to the brain. This is caused by a critical reduction of cerebral blood flow (CBF), mostly due to occlusion of a brain-supplying vessel. On the basis of early animal studies (Astrup et al 1981, Heiss 1983) and positron emission tomography (PET) analyses in acute stroke patients (Baron 1999) a “three-compartment” model of stroke comprising different degrees of CBF reduction has been developed. The three compartments of the model are the ischemic core, the penumbra, and a surrounding region of oligemia (Fig. A4.1). Normal CBF is ~50–60 mL/100 g per minute (Kety 1950). CBF within the ischemic core is <20% of normal values (<10 mL/100 g per minute) which leads to irreversible tissue damage. The core is surrounded by the penumbra—an inhomogeneous zone with a critical reduction down to 20–40% of normal CBF values (10–20 mL/100 g per minute), which is below the functional threshold but above the threshold for morphologic integrity. Depending on the pace and magnitude of reperfusion and the functionality of collaterals, the flow in the penumbra may either completely normalize without induction of structural damage and with improvement of clinical symptoms, or it may further decrease and lead to an enlargement of the ischemic core. The penumbra itself is surrounded by a region of oligemia with only mildly reduced CBF values (20–50 mL/100 g per minute) which is equally influenced by the above factors. In addition, the time factor is as important as the magnitude of hypoperfusion. Within 3 hours of stroke onset a penumbra can be found in the majority of patients which may persist for more than 16 hours (Baron 1999), but may also last for more than 24 hours as we know from clinical experience.

Fig. A4.1 Schematic of the three compartments of cerebral ischemia: 1 = ischemic core (CBF <10 mL/100 g per minute); 2 = penumbra (CBF 10–20 mL/100 g per minute); 3 = oligemia (CBF 20–50 mL/100 g per minute).

For analysis of the ischemic penumbra in a clinical setting, magnetic resonance imaging (MRI) using diffusion-weighted imaging (DWI) and perfusion-weighted imaging (PWI) has almost completely replaced the PET technique. The MR-defined penumbra is determined by the mismatch between the area of impaired diffusion (i.e., ischemic core) and the area of impaired perfusion (DWI–PWI mismatch). However, there are some methodological peculiarities of MRI that need to be considered. Not all areas with impaired diffusion will result in infarction. In fact, there is some regression of the MRI-defined infarct core in up to 20% of cases within a 6-hour time window (Fiehler et al 2004). Furthermore, MRI and PET-defined penumbra is not congruent although an MRI-determined time-to-peak (TTP) delay between 4 and 6 seconds seems to correspond well with a PET-derived CBF <20 mL/100 g per minute (Heiss et al 2004).

Despite these shortcomings, the MRI mismatch concept allows us to sufficiently identify the brain tissue with critically low perfusion and therefore enables selection of patients with regard to intravenous thrombolysis within the 4.5–6-hour time window.

In contrast with myocardial ischemia, in which local atherosclerotic vessel wall disease is practically the only underlying pathomechanism, in ischemic stroke a variety of etiologies have to be considered. Cerebral ischemia can be classified according to several different criteria, e.g., by the temporal pattern, by the infarct pattern, by the affected vascular territory, by its etiology, and finally by its pathogenesis. Usually all these criteria will be incorporated into the final diagnosis, although an exact classification is not always possible. Also, stroke mimics such as migraine with aura, Todd paresis following focal seizures, peripheral vestibular syndromes, neuropathies, acute hypoglycemia, and cerebral venous thrombosis may be challenging, especially in the acute stage of the disease. The following section presents the different approaches to stroke classification.

Classification of Arterial Stroke

Temporal Pattern

The temporal pattern is an important aspect from both the clinician’s and the patient’s perspective. If clinical symptoms completely cease within 24 hours of stroke onset, the episode is defined as a transient ischemic attack (TIA), whereas persisting symptoms are defined as a completed stroke. The concept of a reversible ischemic neurologic deficit (RIND)—symptoms that do not last longer than 7 days—has been abandoned as it falsely suggests transient ischemia without a morphologic correlate. The definition of a TIA, which was developed at a time when the current imaging methods were not available, is also under critical review now, as it suggests that no structural damage has occurred. The term “acute cerebrovascular syndrome,” analogous to the “acute coronary syndrome,” was recently proposed by Japanese groups instead of the term TIA (Okada 2014). MRI sequences, including DWI, show that small structural lesions can be found in up to 60% of cases after a TIA (Brazzelli et al 2014). TIA-related DWI abnormalities are associated with prolonged duration of TIA (Inatomi et al 2004). Although they may regress completely in a short time (Carpentier et al 2012), their presence indicates a higher risk of subsequent stroke (Redgrave et al 2007). In addition, the arbitrary 24-hour cut-off seems problematic. About half of all TIAs are limited to 30 minutes duration. If the symptoms last longer than an hour, the probability of a clinical deficit that will persist beyond the 24-hour cut-off reaches 86% (Levy 1988). Furthermore, the start of symptoms may represent not the onset of vessel occlusion but the onset of collateral failure, which also indicates the need to redefine our concepts of cerebral ischemia by shifting from a clinical time-based to a morphology-based view of stroke.

Nevertheless, the term TIA is of great practical importance as it points out the risk of developing a subsequent completed stroke and therefore requires urgent etiological clarification. In a meta-analysis of patients with completed stroke, 23% reported a prior TIA. Of these, 43% had their last TIA within the week before and 17% on the day of the completed stroke. The TIAs and the subsequent strokes were mostly located within the same vascular territory (Rothwell and Warlow 2005). The risk of having a completed stroke after TIA is especially high in patients with an intracranial vessel occlusion or with a lesion on DWI MRI. Coutts et al (2005) found that the 90-day stroke risk in patients without intracranial vessel occlusion and without a DWI lesion was 4.3%, increasing to 10.8% in those with a positive DWI finding alone and to 32.6% in those with an additional intracranial vessel occlusion. The overall 90-day stroke risk was 11.7%. In patients with a symptomatic ICA stenosis the 90-day stroke risk was 20.1% after a hemispheric TIA (Eliasziw et al 2004). Based on clinical data alone and considering age (A), blood pressure (B), clinical signs (C), duration of symptoms (D), and diabetes (D) an ABCD2 score was developed (Johnston et al 2007). The stroke risk after a TIA within 2 days was highest with 8.1% probability in patients aged 60 years or over with vascular risk factors, motor symptoms, and duration longer than 1 hour. Considering the heterogeneous data it becomes clear why the current concept of TIAs is under debate. A useful proposition might be to limit the diagnosis of TIA to neurologic deficits persisting less than 1 hour and in those in whom DWI MRI does not depict structural lesions (Albers et al 2002).

Infarct Pattern and Vascular Territory

Stroke requires cerebral imaging. In most places the first imaging modality is cranial CT (CCT) which is often followed by cerebral MRI. CCT is a well-established method of excluding intracranial bleeding (e.g., intracerebral hematoma, subarachnoid hemorrhage, subdural and epidural hematoma) which can be found in up to 15% of stroke patients. More recently, MRI with its blood-sensitive susceptibility weighted and T2* weighted, as well as fluid attenuated inversion recovery (FLAIR) sequences, have been shown to be able to detect intracranial hemorrhage with a sensitivity equal to that of CCT or even better. CCT, and better MRI, allow classification of ischemic infarcts according to their pattern and corresponding vascular territory. We consider territorial infarctions, lacunar infarctions, and border zone infarctions as independent entities; the latter may develop within one vascular territory or between several vascular territories.

Territorial Infarction

The brain comprises circumscribed regions that are supplied with blood via one main artery and its tributaries (Fig. A4.2). The classic concept that blocking a certain artery leads to infarction of the complete territory of supply has been refuted. Due to the extensive pattern of extra- and intracranial collateral flow a territorial infarct may therefore either incompletely (partial territorial infarction) or completely (total territorial infarction) involve the area of a brain-supplying artery. The underlying pathogenesis in these cases is an intracranial arterial occlusion of the dedicated artery. In elderly patients these occlusions are mostly of embolic nature, for example deriving from a cardiac source or from upstream macroangiopathic vessel wall alterations. However, an in-situ atherothrombosis on the basis of preexisting macroangiopathy may also be present (Lhermitte et al 1970). In younger patients rare conditions such as vasculitis, dissection, and vasoconstriction or nonatherothrombotic in-situ thrombosis with underlying genetic predisposition have to be considered.

Fig. A4.2 Schematic of the arterial territories of the brain. Left: Axial plane. Right: Coronal plane. (Adapted from Duus et al 2005.).

The infarct size depends on several factors. In cases with an embolic event the location of the occlusion (proximal or distal), the duration of the occlusion, and the quality of the leptomeningeal collaterals (LMC) determine the dimension of the induced lesion; for further details about leptomeningeal collaterals, see Chapter 5, “Secondary Collaterals (Ophthalmic Artery and Leptomeningeal Collaterals)” under “Intracranial Collateral Pathways”/“Intracranial Collateral Pathways in ICA Occlusive Processes.” In the following, examples are given of the most commonly affected intracranial artery, the middle cerebral artery (MCA). In case of a main-stem M1-MCA occlusion, which includes the lenticulostriate arteries (LSAs), total MCA territorial infarction will occur if timely recanalization does not occur and if the LMC are insufficient (Fig. A4.3). If good LMC are present, the infarct size may—despite an identical location of the occlusion—be restricted to the striatocapsular area. In M1-MCA occlusion, this region is almost always affected as the LSAs supply blood to end zone territories that are not reached by other vessels (Fig. A4.4). Occasionally, LSAs may in part or completely arise from a proximal M2-MCA branch, especially if the M1 segment is short, in which case an M1-MCA occlusion might not affect the basal ganglia. In between these two extremes different patterns of partial territorial infarctions with variable sizes, with or without subcortical involvement, can be observed in patients with a proximal MCA occlusion. Preserved “cortical islands” that derive their blood supply from LMC are a frequent finding (Fig. A4.5). In case of an M1-MCA occlusion distal of the LSA origin, the basal ganglia will be preserved. In a “best case scenario” a circumscribed distal M1-MCA occlusion could be endured without infarction— but only in the case of excellent leptomeningeal blood supply and if embolus fragments have not migrated into the periphery. Most commonly, however, an inhomogeneous partial territorial infarct is seen, involving a variable amount of the corresponding cortex (Fig. A4.6). Apart from large and almost total striatocapsular infarction other subcortical infarct patterns may also occur in the centrum semiovale which may be difficult to distinguish from lacunar stroke (Wessels et al 2005). If an MCA branch occlusion is present, the size (large or small) and extent (complete or incomplete) of infarction depends on the site of occlusion and the time-course of recanalization (Fig. A4.7). Because of their mostly clear localization, supratentorial infarctions are usually named according to the affected vascular territory.

Fig. A4.3 Variants of MCA territorial infarction in proximal M1-MCA occlusion. Cranial CT, axial plane. (A) Hyperdense media sign (arrow). (B) Complete MCA infarction.

Fig. A4.4 Variants of MCA territorial infarction in proximal M1-MCA occlusion. Cranial CT, axial plane. (A) Hyperdense media sign (arrow). (B) Partial MCA infarction leading only to a large striatocapsular infarct.

Fig. A4.5 Variants of MCA territorial infarction in proximal M1-MCA occlusion. (A) Cerebral DW MRI, axial plane. Large MCA infarct with a vital parenchymal “island.” (B) Cerebral MRI, FLAIR image, axial plane. Large MCA infarct predominantly within the frontal and parietal opercula sparing most of the temporal parenchyma.

Fig. A4.6 Variants of MCA territorial infarction in distal M1-MCA occlusion. Cerebral DW MRI, axial plane. (A) Partial, predominantly subcortical MCA infarct with spots of cortical involvement. (B) Partial MCA infarct affecting mainly the insula and sparing the basal ganglia.

The distribution pattern of strokes in the anterior and posterior circulation roughly resembles the relation of both territories with regard to the total CBF. In a clinically orientated, community-based study of 675 patients, 68% of territorial infarctions were assumed to affect the anterior and 32% the posterior circulation (Bamford et al 1991). An identical proportion (68%) of strokes in the anterior circulation was found in the hospital-based Lausanne stroke registry, which included 1,000 consecutive stroke patients who underwent CCT diagnostics. Distribution of infarcts within the anterior circulation was as follows: 96% within the MCA, 3% within the ACA, and 1% combined within the MCA and ACA territories. Embolic ACA infarctions are rare because of the ACA’s unfavorable angulation at the distributary from the terminal ICA. Emboli therefore tend to follow the flow into the MCA. Large ACA vessels, e.g., also feeding the contralateral ACA or having a diameter similar to the ipsilateral MCA or larger, favor emboli into the ACA territory (Shoamanesh et al 2014). Vertebrobasilar and PCA territory ischemia was seen in 26% of patients. Hemodynamic infarction and mixed patterns were seen in 3% of cases. Lacunar infarcts in this study were attributed to the corresponding vascular territory (Bogousslavsky et al 1988). Clinically, supratentorial territorial and lacunar or infratentorial ischemia can be differentiated as the former demonstrates cortical signs such as aphasia, apraxia, hemineglect, visuospatial impairment, and hemianopsia in combination with motor or sensorimotor deficits. However, large subcortical and thalamic infarcts can present similar signs but have a somewhat better prognosis.

Fig. A4.7 Variants of territorial MCA infarction in MCA branch occlusion. Cerebral DW MRI, axial plane. (A) Large left anterior partial MCA infarction of an M2/3 segment leading to Broca’s aphasia. (B) Small left cortical MCA infarction within the left precentral gyrus (Rolandic artery, M4) leading to right brachiofacial hemiparesis.

Fig. A4.8 Microcirculation of the brain. (A) Schematic. (Adapted from Spatz 1939, Fig. 1, with kind permission of Springer Science and Business Media.) (B) Postmortem angiogram, axial plane: Note the distinct ramification of small perforating arteries in the basal ganglia. (Adapted from http://www.radnet.ucla.edu/sections/DINR/Part%2018/Part18B11.htm by courtesy of Professor G. Salamon, Radiology, UCLA, Los Angeles, USA).

Lacunar Infarction

Lacunar infarctions result from occlusion of small arterial branches or a single perforating artery itself. They have an average diameter between 100 μm and 400 μm, and arise directly from much larger arterial vessels in a perpendicular direction (Fig. A4.8). Causes of lacunar stroke may vary and are subject of ongoing debate (Del Bene et al 2013, Wardlaw 2005). Histopathologic analysis of smaller lacunar infarcts often reveals nonatherosclerotic subintimal vessel wall hyalinosis, lipohyalinosis, and fibrinoid necrosis (Lammie et al 1997). Hypertension and diabetes mellitus are predisposing factors for this type of vessel affection but may clinically be overlooked. Larger perforating arteries may show a proximal intraluminal atheroma, also called arteriolosclerosis, or an atherosclerotic plaque located at the origin of the vessel. Also a small embolus may enter and occlude an LSA as has been demonstrated in a monkey model (Macdonald et al 1995). However, a single small subcortical infarction is most unlikely to be caused by a proximal embolic source, as the embolus would need to pass the unfavorable branching-off of the perforating artery from the parent artery. Even in the presence of a suggestive cardioembolic cause, such as atrial fibrillation other etiologies should be taken into account whenever cortical areas are spared by the ischemia. The most frequently affected perforating arteries are the LSA arising from the MCA stem, the thalamoperforating and thalamogeniculate arteries arising from the proximal PCA and posterior communicating artery, and the paramedian branches of the BA. Corresponding infarcts are usually found within the basal ganglia, the internal and external capsule, the centrum semiovale, the thalamus, and the brainstem, mainly pontine (Fig. A4.9 and Fig. A4.10). Lacunar infarcts are by definition small, not exceeding 15 mm, most of them being smaller than 10 mm. The true dimension of a subcortical infarct can only be determined if at least two MR planes are used. Infarctions with a small round or oval shape in all imaging planes are true microinfarctions due to nonatherosclerotic vessel diseases (Fig. A4.10 and Fig. A4.11). In contrast, an infarct appearing small and oval-shaped in one plane but tubular, club-, or fan-like-shaped in at least one of the other two planes and exceeding the above-defined 15 mm can be considered as in-situ (local) thrombotic or atherothrombotic (in atherosclerosis), i.e., affecting the complete vascular area of a single perforating artery (see “Infarctions of Uncertain Classification” below).

Fig. A4.9 Morphologic variants of supratentorial microangiopathy. (A) Cerebral DW MRI, axial plane. Right-sided lacunar thalamic infarct assumed to correspond to small-vessel disease. (B) Cerebral MR T2-weighted image, axial plane. Pronounced periventricular confluent white matter changes (leukoaraiosis), predominantly in the parietooccipital area, and small lacunar lesions of the basal ganglia.

Most lacunar strokes present as characteristic syndromes as they affect the above-mentioned circumscribed brain regions. These are generally pure motor strokes, found in up to 50% of cases; pure sensory strokes; sensorimotor strokes; dysarthria–clumsy hand syndrome; and ataxic hemiparesis. However, these clinical syndromes are not pathognomonic of lacunar stroke. A study in 73 patients with clinically lacunar syndromes revealed a different pathomechanism in 23% of cases, half of them attributable to a cardiac embolic source, and half of them due to a relevant proximal arterial stenosis (Wessels et al 2005). It may be helpful to note that sub-cortical strokes usually lead to complete motor, sensory, or sensorimotor signs affecting limbs and face, whereas cortical infarctions usually spare one limb or the face. In pure motor stroke a proportional hemiparesis equally affecting arm and leg indicates a subcortical lesion while cortical MCA or ACA infarctions usually present a more arm- or leg-dominating hemiparesis.

Fig. A4.10 Variants of lacunar paramedian pontine stroke assumed to correspond to small-vessel disease. Top: Cerebral DW MRI, axial plane. Bottom: T2-weighted images, sagittal plane. (A) Medium-sized left-sided paramedian pontine infarct. (B) Small left-sided pontine infarct (arrow).

Fig. A4.11 Cerebral DW MRI, axial plane. Right-sided lacunar thalamic infarct assumed to correspond to small-vessel disease. (B,C) Cerebral MR T2-weighted image, coronal plane (B) and sagittal plane (C), both showing the infarct small and oval-shaped, suggestive of a microangiopathic lesion (arrow).

Fig. A4.12 Variants of anterior external BZI. (A) Cranial CT, axial plane. Hypodense BZI located between the left MCA and ACA territory. (B) Cerebral DW MRI, axial plane. Left anterior BZI, larger and located more laterally.

Border Zone Infarction

Border zone infarction (BZI) is considered to be caused by a low-flow state in large brain-supplying arteries due to high-grade stenosis or occlusion of an upstream artery or profound hypotension. An incomplete cerebral arterial circle (circle of Willis) is probably another important risk factor for BZI (for further details, see Case 30). Brain lesions may be located in the anterior and in the posterior circulation at the boundary of the territorial blood supply from the major intracranial vessels. They are best discussed in the anterior circulation, but even here there is considerable concern about the nature and significance of these lesions (Caplan and Hennerici 1998, Momjian-Mayor and Baron 2005). Within the supratentorial parenchyma, two categories can be distinguished: external and internal BZIs, the first also referred to as cortical BZIs. External BZIs are located between two or all three cortical territories of the MCA, ACA, and PCA. The lesions affect mainly the cortical area in a wedge-shaped manner, but they may extend into the subcortical areas and may vary considerably in size. An anterior external BZI between the MCA and ACA is mainly observed in ICA pathology (Fig. A4.12). A posterior external BZI, located between the MCA and PCA territory, may be present in ICA pathology in combination with fetal-type PCA or in additional steno-occlusive disease of the vertebrobasilar circulation (Fig. A4.13). However, there is no extensive data correlating the distribution of cortical BZI with the vascular status. The internal border zone involves a subcortical area within the corona radiata between the superficial and deep perforators of the MCA or between the superficial perforators of the MCA and ACA (Fig. A4.14). An internal BZI may have a distinct rosary-like pattern of small inline white matter lesions or a more prominent cigar-shaped confluent pattern (Fig. A4.15). Both patterns may occur separately or in combination (Fig. A4.16). The reported proportion of hemodynamic strokes varies. From clinical and autopsy studies it is assumed that ~10% of all brain infarctions are of hemodynamic origin (Bladin and Chambers 1994, Jörgensen and Torvik 1969). In the Lausanne stroke registry only 3% of patients were considered to suffer from BZI (Bogousslavsky et al 1988). In symptomatic high-grade ICA stenoses or ICA occlusions of atherosclerotic origin ipsilateral hemodynamic lesions have been observed in ~50% of cases (Szabo et al 2001).