The approximate relative frequency of the main causes of ischaemic stroke and TIA

  • 50% Atherothrombosis affecting large and medium-sized arteries between the heart and the brain
  • 25% Intracranial small vessel disease (small vessel disease lipohyalinosis/microatheroma, etc.)
  • 20% Embolism from the heart
  • 5% Miscellaenous rare disorders


Atheroma mainly affects large (e.g. aortic arch) and medium-sized arteries at places of arterial branching (e.g. carotid bifurcation), tortuosity (e.g. carotid siphon), and confluence (e.g. basilar artery) (Fisher 1951, 1954; Hutchinson and Yates 1957; Schwartz and Mitchell 1961; Cornhill et al. 1980; Ross et al. 1988).


These are sites of haemodynamic sheer stress and thus endothelial trauma; boundary layer separation, blood stagnation, and the accumulation of platelets; and of turbulence, all of which are likely to promote thrombosis (Grady 1984; Reneman et al. 1985; Nicholls et al. 1989).

However, in the same individual, atheroma in one place does tend to be accompanied by atheroma in other parts of the same artery, with atheroma in other arteries to the brain, and in arteries to other organs such as the heart (Mitchell and Schwartz 1962; Miller and Cohen 1987). Presumably this reflects individual susceptibility to atheroma as a result of the presence of causal vascular risk factors (such as hypertension) and genetic predisposition which determines who will develop atheroma, while the arterial anatomy determines where the lesions occur. None the less, it is curious how severely one arterial site can be affected and yet, in the same individual, the mirror-image site on the other side of the body is still normal, perhaps because of subtle asymmetries in arterial geometry (Gnasso et al. 1997).

Atheroma starts in childhood, it is thought in response to endothelial injury (Ross 1999). Intimal fatty streaks appear first. In a gradual process stretching over many years, circulating monocyte-derived macrophages adhere to and invade the arterial wall, there is an inflammatory response with cytokine production and T-lymphocyte activation, intra- and, later, extracellular cholesterol and other lipids are deposited, particularly in macrophages which are then described as foam cells, smooth muscle cells proliferate, fibrosis occurs, and so fibrolipid plaques are formed (Fuster et al. 1992a,b; Libby 1996). Necrosis and calcification complicate advanced lesions. These atheromatous plaques invade the media, gradually spread around and along the arterial wall, and narrow the lumen, although at times the vessel dilates. The plaques are complicated by platelet adhesion, activation, and aggregation, which initiates blood coagulation and subsequent thrombosis.

Thrombus may be incorporated into the atheromatous plaques which then re-endothelialize; it may grow to obstruct the arterial lumen and then propagate proximally or distally in the stagnant column of blood as far as the next branching point or beyond; it may be lysed by natural fibrinolytic mechanisms in the vessel wall; or it may embolize in whole or in part to occlude a distal artery, usually at a branching point. Such artery-to-artery emboli vary in size and shape, and consist of some combination of cholesterol debris from the atheromatous plaque, platelet aggregates, and fibrin, which may be newly formed and relatively friable, or old and well organized. Depending on local blood flow and on the size, composition, and consistency of the impacted emboli, they may be lysed, fragment, and vanish into the microcirculation, or remain to occlude the artery and promote local thrombosis. Thrombosis is further encouraged by the release from platelets of thromboxane A2, which is also a vasoconstrictor. However, it is opposed by prostacyclin and nitric oxide, both vasodilators, released from vascular endothelium, as well as by endothelium-derived plasminogen activator (Vane et al. 1990).

The balance of pro- and antithrombotic factors determines whether a thrombus complicating an atheromatous plaque or an occlusive embolus, grows, is lysed, or is incorporated into the vessel wall.

It is likely that atheromatous plaques become ‘active’ or ‘unstable’ from time to time as a result of fissuring and cracking of thin parts of the fibrous cap which covers the rather rigid lesion; of ulceration perhaps; or sometimes of haemorrhage within the plaque, rather than the more commonly found haemorrhage entering via a crack in the endothelial surface (Fisher et al. 1987; Svindland and Torvik 1988; Richardson et al. 1989; Torvik et al. 1989; Gomez 1990; Ogata et al. 1990; Davies 1997; Lammie et al. 1999).

Any of these events exposes the highly thrombogenic necrotic core of the plaque to blood and so causes thrombus to form and then perhaps to embolize. Thus, atherothromboembolism can be regarded as an acute-on-chronic disease; at any one time a plaque may be static and quiescent with a thick fibrous cap, slowly growing but asymptomatic, or active with ongoing thrombosis and embolization, which may or may not be symptomatic depending on the depth and duration of the consequent ischaemia. This concept may explain the tendency for TIAs to cluster, for stroke to occur early after a TIA and to affect the same arterial territory, for presumed artery-to-artery embolic strokes to recur early, and for the risk of stroke to decline with time even distal to a severe symptomatic stenosis.

Complicated atheromatous lesions which eventually become fibrotic and heavily calcified make the whole artery rigid, elongated and so tortuous, and sometimes ectatic. Ectasia and aneurysmal bulging, particularly of the basilar artery, may compress adjacent structures, such as the lower cranial nerves and brainstem. Also, emboli may be released from the atheromatous walls and complicating thrombosis. However, arterial rupture is exceptional (Schwartz et al. 1993; Passero and Filosomi 1998).

Also, intraplaque haemorrhage presumably may cause quite sudden enlargement of a plaque, followed by fairly rapid shrinkage as the haematoma is absorbed. Recently it has become possible to monitor the release of emboli from carotid plaques, and elsewhere, into the cerebral circulation with transcranial Doppler of the middle cerebral artery, again an indication of plaque ‘instability’. Not surprisingly, high-intensity embolic signals are more common distal to symptomatic compared with asymptomatic stenoses, but even then the number is rather low, perhaps because of technical difficulties or insufficient recording time (Del Sette et al. 1997; Koennecke et al. 1998; Molloy et al. 1998; Wijman et al. 1998).

Symptoms of focal ischaemia occur as a consequence of reduced blood flow which, in the context of atheroma, is most commonly due to embolism from a plaque complicated by thrombosis in an extracranial artery (such as the carotid bifurcation) to occlude a smaller intracranial artery (such as the mainstem or branch of the middle cerebral artery (MCA)) (Gunning et al. 1964; Lhermitte et al. 1970; Caplan and Hennerici 1998) (Fig. 27.15). The notion that the carotid siphon might filter out emboli en route to the brain is interesting but unproven (Hugh 1987).


Occasionally emboli may reach the brain via the collateral circulation; for example from a stenosed internal carotid artery (ICA) across the circle of Willis into the MCA distal to an occluded contralateral ICA; from thrombus in the blind proximal stump of an occluded contralateral ICA, a stenosed proximal external carotid artery (ECA), or from more proximal sites of atheroma, but all via the ECA and through the ophthalmic circulation to the carotid siphon and beyond. Also, emboli may arise from the distal end of a thrombus occluding the ICA. Finally, ischaemia may occur due to haemodynamic compromise beyond an occluded ICA (Hankey and Warlow 1991a; Klijn et al. 1997). Furthermore, focal ischaemia may occur between arterial territories (i.e. in boundary zones) usually due to low flow distal to an occluded artery.

Fig. 27.15.

Various patterns of arterial occlusion causing different types of ischaemic stroke. Left-hand column, diagram of axial CT brain scan through the level of the basal ganglia; middle column, diagram of the middle cerebral artery (MCA) and anterior cerebral arteries on a coronal brain section; right-hand column, corresponding axial CT brain scan. (A) main trunk of MCA; (B) lenticulostriate perforating branches of the MCA; (C) cortical branches of the MCA; (D) cortical branches of the anterior cerebral arteries. (a) Normal arterial anatomy and CT scan; (b) occlusion (usually embolic (arrow) from heart, aorta, or internal carotid artery) of a cortical branch of the MCA and restricted cortical infarct on CT (arrows); (c) occlusion (usually embolus (arrow) as in (b) above) of MCA trunk to cause infarction of entire MCA territory (arrows); (d) occlusion of a single lenticulostriate artery to cause a lacunar infarct (arrow); note that the patient has an old lacunar infarct in the opposite hemisphere; (e) occlusion of the MCA trunk with good cortical collaterals from the anterior and posterior cerebral arteries to cause a striatocapsular infarct (arrows).

Symptomatic in situ atherothrombotic occlusion does not appear to be very common in the anterior cerebral circulation, perhaps because the most commonly affected site for atheroma is in a relatively large artery (i.e. the ICA origin) rather than in smaller arteries, such as the MCA, which are more often occluded by embolism than by in situ atherothrombosis (Lhermitte et al. 1970; Ogata et al. 1990, 1994). Another reason might be the relatively effective collateral circulation distal to any occlusion in the extracranial carotid system (Section 27.3.2). On the other hand, symptomatic atherothrombotic occlusion does appear to be more common in the posterior circulation, particularly in the basilar artery. But even here, embolism from non-occlusive thrombus into smaller arteries supplying the brainstem and elsewhere is well described (Castaigne et al. 1973; Schwarz et al. 1997; Martin et al. 1998).

In an individual patient it is relatively easy to diagnose acute focal ischaemia or infarction but, because angiography is seldom done early, or at all, it is difficult to know what the pattern of any arterial pathology is and exactly how the ischaemia has occurred. Even when angiography, or (increasingly often) ultrasound of the extra- and intracranial circulation, reveals an occluded artery, this does not necessarily mean that the occlusion was recent, or that it was embolic rather than due to in situ thrombosis, or whether any embolism had occurred via collaterals or from the distal end of the occluded thrombus, or whether the ischaemia was due to low flow distal to an old occlusion (Ringelstein et al. 1983; J. P. H. Wade et al. 1987).

To confuse matters further, it is now clear that occluded arteries can recanalize spontaneously quite quickly, particularly the mainstem of the MCA. Possibly emboli from the heart are more likely to lyse than those from atheromatous arteries, and thrombotic occlusion of the MCA or ICA is perhaps less likely to open spontaneously. Recanalization rates presumably also depend on the constituents and age of the occluding material. However, whatever the exact cause of the ischaemia, any demonstrated arterial pathology in the aorta, neck, basilar artery, or circle of Willis is usually atherothrombosis and the assumption is then reasonably made that thromboembolism has occurred at some stage and is likely to occur again. If, as is so often the case (Zhu and Norris 1990), little or no arterial pathology is demonstrated by vascular imaging, or macroscopically at post-mortem, then focal ischaemia is most likely due to embolism from the heart or to intracranial small vessel disease (Fig. 27.15).

Cholesterol embolization syndrome

This rare disorder seems to be due to the rupture of atheromatous plaques in elderly people with widespread disease, either spontaneously, but perhaps more often as a complication of instrumentation or surgery of large atheromatous arteries such as the aorta, and possibly of anticoagulation or therapeutic thrombolysis. Cholesterol debris is released and embolizes to the microcirculation of many organs throughout the body, including the brain and spinal cord. Hours or days after instrumentation or surgery there is the subacute onset of a syndrome very similar to systemic vasculitis or infective endocarditis: malaise, fever, abdominal pain, proteinuria and renal failure, stroke-like episodes, drowsiness, confusion, skin petechiae, splinter haemorrhages, livedo reticularis, cyanosis of fingers and toes, raised erythrocyte sedimentation rate, neutrophil leucocytosis, and eosinophilia. The diagnosis is made by finding cholesterol debris in the microcirculation of biopsy material, usually from the kidney but sometimes from skin or muscle (Fine et al. 1987; Cross 1991; Rhodes 1996).

Intracranial small vessel disease

The small penetrating arteries of the brain (less than about 500 µm in diameter) are not supported by a good collateral circulation; i.e. the lenticulostriate branches of the middle cerebral artery (MCA), the thalamoperforating branches of the proximal posterior cerebral artery, and the perforating arteries to the brainstem. Therefore, occlusion is rather likely to cause infarction, albeit in a small, restricted area of brain. Such ‘lacunar’ infarcts comprise about one-quarter of first ischaemic strokes and TIAs (Bamford et al. 1987; Sempere et al. 1998).

Because the case fatality is so low (about 1 per cent), there are few pathological data, but it does seem that these small arteries are much less likely to be occluded by emboli either from the heart or from extracranial sites of atherothrombosis, compared with the trunk or cortical branches of the MCA (Olsen et al. 1985; Bamford and Warlow 1988; Orgogozo and Bogousslavsky 1989; Hankey and Warlow 1991b; Tegeler et al. 1991; Boiten et al. 1996; Gan et al. 1997). Furthermore, ischaemic lacunar strokes are less often associated with MCA emboli detected with transcranial Doppler (Koennecke et al. 1998).

Although not universally accepted, it is generally thought that these small perforating arteries are occluded by thrombus complicating not atheroma but a distinct small vessel arteriopathy variously described as lipohyalinosis, complex small vessel disease, arteriolosclerosis, fibrohyalinosis or microatheroma (Millikan and Futrell 1990; Fisher 1991; Lammie et al. 1997).

The muscle and elastin in the arterial wall are replaced by collagen, there is subintimal hyalinization, the wall becomes thickened and the lumen narrowed, and the vessel becomes tortuous, possibly with the formation of microaneurysms (Charcot–Bouchard aneurysms) which may rupture. It is, therefore, conceivable that this small vessel arteriopathy can lead to small, deep haemorrhages as well as lacunar infarcts; indeed, both types of stroke have been described in the same patients (Miyashita et al. 1991; Besson et al. 1993; Samuelsson et al. 1996; Kwa et al. 1998).

Although hypertension is common in patients with lacunar infarction, it cannot explain every case and, in any event, the risk factors, including hypertension, seem to be rather similar to those in ischaemic stroke patients with presumed atherothrombotic arterial disease (Adams et al. 1989; Lodder et al. 1990; Sacco et al. 1991; Mast et al. 1995; You et al. 1995; Schmal et al. 1998).

Perhaps the same individuals are susceptible to both atherothrombosis of large and medium-sized arteries and small vessel disease, but one becomes symptomatic before the other. Or perhaps the concept of a distinct small vessel disease causing lacunar infarction is incorrect. Certainly, at least some small infarcts in the brainstem and internal capsule can be due to atheroma at the mouth of the small penetrating vessels spreading from atheroma of the larger parent artery (Fisher and Caplan 1971; Fisher 1979).