OUP user menu

Experimental models for the investigation of brain ischemia

Konstantin-Alexander Hossmann
DOI: http://dx.doi.org/10.1016/S0008-6363(98)00075-3 106-120 First published online: 1 July 1998
Keywords
  • Brain ischemia
  • Animal models
  • Stroke
  • Circulatory arrest
  • Resuscitation
  • Treatment

Time for primary review 28 days.

1 Introduction

More than any other organ of the body, brain integrity depends on the continuous blood supply of oxygen and glucose for covering the energy demands of the tissue. Cessation or severe reduction of blood flow results in almost instantaneous biochemical and functional deficits which become rapidly irreversible unless blood flow is promptly restored. The ischemia time that is tolerated by the brain depends, among others, on the density of ischemia, the tissue concentration of primary and secondary energy stores and the rate of energy consumption which, in turn, depends on the temperature, the degree of functional activity and the absence or presence of anesthetics and other drugs. There is also a major difference between the ischemic vulnerability of different regions of the brain. As an example, the CA1 sector of hippocampus may suffer irreversible injury after ischemia of as short as 5 min [94]whereas other nerve cell populations survive normothermic cerebrocirculatory arrest of as long as 1 h [77].

The pathophysiology of cerebral ischemia is further complicated by the fact that the severity of cell injury is modulated by numerous indirect or secondary consequences of the primary ischemic impact. Recirculation disturbances, stress responses, peroxidative changes, or the activation of genomic responses are only few examples of the large number of hemodynamic and molecular responses which determine the final outcome [153, 97]. The interpretation of experimental data of cerebral ischemia, therefore, requires intimate knowledge of the models and conditions under which the data have been collected. The number and diversity of such models, has become very large over the years because researchers have sought in vain for the ideal approach to study ischemic injury and its therapeutic reversal. Such an ideal model should be relevant to the clinical situation, it should be highly reproducible, it should avoid complicating side effects, and it should be easy to perform. Obviously, these requirements are not easily met, which explains the continuous search for new experimental procedures. Interestingly, similar solutions have been proposed repeatedly, apparently without knowledge of the previous observations. The present review, therefore, also includes earlier reports to draw attention to this frequently forgotten knowledge.

As regards the pathophysiology of cerebral ischemia, three major categories of flow reduction have to be distinguished: transient global ischemia, permanent or transient focal ischemia, and microembolism. The most important clinical cause of global ischemia is cardiac arrest which produces complete cessation of cerebral blood flow, followed by more or less efficient recirculation, depending on the success of cardiac resuscitation. Global ischemia also results from strangulation, severe shock or intracranial hypertension, but under these conditions flow decline is incomplete and heterogeneous. The clinical prototype of focal brain ischemia is stroke which is most frequently caused by thrombotic occlusion of the middle cerebral artery. Transient forms of focal ischemia are produced by vascular clipping in the course of neurosurgical interventions or by severe vasospasms. Microembolisms, leading to multiple ischemic microfoci are caused by fat microemboli after bone fractures, by release of platelet aggregates and thrombotic materials from ulcerating atherosclerotic plaques, or by air bubbles during cardiac surgery.

All these ischemic conditions exhibit different pathophysiologies and require different therapeutic approaches. This has not always been appreciated in the past, and attempts have been made to search for universal mediators of ischemic injury which are equally important for the manifestation of delayed neuronal death in selectively vulnerable brain regions after transient global ischemia as for the evolution of brain infarcts during permanent focal ischemia. Examples of such mediators are lactic acid, calcium, glutamate, nitric oxide, arachidonic acid or oxygen free radicals, all of which have become targets of therapeutic interventions. In fact, potent drugs have been developed that provide powerful protection in certain experimental models but failed to improve ischemic injury under clinical conditions [53, 35]. This is not surprising because the experimental models used in ischemia research must be carefully selected in accordance with the clinical problem under investigation, and the therapeutic recommendations derived from such studies should be restricted to this particular pathophysiological situations.

In the following, an overview of the most important models of brain ischemia will be presented, and the relevance of these models for cerebrovascular disease will be discussed. Some of these models have been gradually improved over the years. In such instances Tables 1 and 2 will provide multiple citations which refer to the first and the most widely used modification of this model.

View this table:
Table 2

Global brain ischemia and in vitro ischemia1

Means of induction of ischemiaSpeciesReference
Cardiac arrest
 with KClmonkeyMyers and Yamaguchi, 1977 [127]
ratBlomqvist and Wieloch, 1985 [12]
 by ventricular fibrillationcatHossmann and Hossmann, 1973 [81]
dogLin and Kormano, 1977 [108]
dogGurvitch et al., 1972 [56]
monkeyBacalzo and Wolfson, 1969 [8]
 by intrathoracal hookratKorpatchev et al., 1982 [100]
 by exsanguinationdogHeymans and Bouckaert, 1935 [69]
dogKirimli et al., 1968 [93]
 by asphyxiaratKatz et al., 1995 [90]
 by drowningdogMakarenko, 1972 [114]
 by arresting extracorporeal
 circulationcatIijima et al., 1993 [84]
Induced hypotension
catGygax et al., 1978 [57]
catPasztor et al., 1972 [139]
monkeyBrierley et al., 1969 [17]
Occlusion of aorta
ratWade et al., 1975 [172]
catTenCate and Horsten, 1952 [161]
 with vent into blood reservoirrabbitCantu et al., 1969 [22]
 with aorto-atrial bypassdogArai et al., 1986 [6]
 distally of left subclaviandogPontius et al., 1954 [142]
 artery
 and vena cavadogMarshall et al., 1956 [117]
dogBrockman and Jude, 1960 [19]
dogJackson et al., 1981 [85]
monkeyMiller and Myers, 1970 [122]
  and vena azygosdogZimmermann and Spencer, 1958 [190]
dogGoldstein et al., 1966 [52]
 and arteria pulmonaliscatGänshirt and Zylka, 1952 [48]
rabbitLäwen and Sievers, 1910 [105]
Occlusion of arteria pulmonalis
catWeinberger et al., 1940 [175]
Occlusion of vena cava
dogCohen et al., 1952 [29]
dogRead et al., 1956 [144]
 and vena azygosgroundhogBigelow and McBirnie, 1953 [11]
Pneumatic cuff around neck
catHart et al., 1978 [63]
rabbitHirsch, 1958 [73]
dogKabat and Dennis, 1938 [87]
 and hypotensionratNemoto and Frinak, 1981 [132]
ratSiemkowicz and Hansen, 1978 [152]
catNemoto et al., 1981 [133]
monkeyNemoto et al., 1977 [131]
Occlusion of carotid arteries
mouseBarone et al., 1993 [10]
hypertensive ratFujishima et al., 1981 [45]
neonatal
ratsMitsufuji et al., 1996 [123]
gerbilKobayashi et al., 1977 [95]
awake gerbilChandler et al., 1985 [26]
gerbilTomida et al., 1987 [164]
sheepTerlecki, 1967 [163]
monkeySengupta et al., 1973 [149]
 and hypotensionratSmith et al., 1984 [156]
catWelsh et al., 1977 [176]
 and hypoxianeonatal ratSchwartz et al., 1992 [148]
 and subclavian arteriesratde la Torre, 1991 [34]
 and basilary arterymutant mousePanahian et al., 1996 [138]
ratKameyama et al., 1985 [88]
catGinsberg et al., 1978 [50]
 and pterygopalatine arteryratGumerlock et al., 1989 [54]
 and jugular veinsratAragno and Doni, 1976 [5]
Occlusion of carotid and
vertebral arteries
ratPulsinelli and Brierley, 1979 [143]
catSugar and Gerard, 1938 [157]
rabbitYang et al., 1991 [183]
monkeyTeraura et al., 1972 [162]
 and retrograde drainagedogFujishima, 1971 [44]
monkeyWolin et al., 1972 [181]
 and hypotensionrabbitKolata, 1979 [99]
Occlusion of innominate and
subclavian arteries
catGildea and Cobb, 1930 [49]
dogPike et al., 1908 [141]
 and hypotensioncatHossmann and Sato, 1970 [79]
monkeyHossmann and Zimmermann, 1974 [80]
Intracranial hypertension
 by fluid infusionratSiesjö and Zwetnow, 1970 [154]
catvan Harreveld, 1947 [168]
rabbitMarshall et al., 1975 [116]
dogNeely and Youmans, 1963 [130]
monkeyHeyreh and Edwards, 1971 [70]
  and hypotensionratLjunggren et al., 1974 [111]
dogKramer and Tuynman, 1967 [101]
dogHallenbeck and Bradley, 1977 [61]
  and peritoneal dialysisratKawakami and Hossmann, 1977 [92]
 by ballooncatHekmatpanah, 1970 [68]
monkeyLangfitt et al., 1965 [104]
Isolated head
ratKrieglstein et al., 1987 [102]
catHirsch et al., 1955 [73]
dogHinzen et al., 1972 [72]
monkeyWhite et al., 1967 [179]
Decapitation
 in situ revivalguinea pigOkada, 1974 [135]
In vitro ischemia
 neuronal culturesratGoldberg and Choi, 1993 [51]
 organotypic culturesratVornow et al., 1994 [171]
 brain slicesratDong et al., 1988 [37]
guinea pigWhittingham et al., 1984 [180]
  • 1 The presently most widely used models have been highlighted.

View this table:
Table 1

Focal brain ischemia and microembolism1

Means of inducing ischemiaSpeciesReference
Occlusion of common carotid
arterygerbilLevine and Payan, 1966 [107]
hypertensive
ratChoki et al., 1977 [28]
sheepTerlecki, 1967 [163]
 and contralateral external
 carotid arterygerbilBosma et al., 1981 [14]
 after previous ligation of
 contralateral carotid and both
 vertebral arteriesratDougherty et al., 1982 [38]
 and contralateral anastomoses
 between carotid artery and
 jugular veinratBannister and Chapman, 1984 [9]
 and intracranial hypertensionratBusto and Ginsberg, 1985 [21]
 and anoxiaratLevine, 1960 [106]
 and hypoxianeonatal ratAndiné et al., 1990 [3]
 and CO-poisoningratMacMillan, 1978 [112]
Occlusion of anterior cerebral
artery
baboonLiu et al., 1992 [110]
Occlusion of middle cerebral
artery
 transorbital ligationcatO'Brien and Waltz, 1973 [134]
dogFisk et al., 1969 [41]
monkeyHudgins and Garcia, 1970 [83]
  and internal carotid arteryratBrint et al., 1988 [18]
  occlusioncatBose et al., 1984 [13]
dogDiaz et al., 1979 [36]
 postorbital approachcatvan der Sprenkel, 1988 [167]
 transcranial approachratRobinson, 1981 [146]
ratTamura et al., 1981 [160]
hypertensive
ratCoyle et al., 1984 [31]
catSundt and Waltz, 1966 [158]
dogAnthony et al., 1963 [4]
monkeyMeyer and Denny-Brown, 1957 [120]
  and both carotid arteriesratChen et al., 1986 [27]
  and olfactory branchratShiino et al., 1989 [151]
  and posterior communicating
  arterygerbilYoshimine and Yanagihara, 1983 [185]
 intraluminal occlusion
  with silicon rubberratTurner, 1975 [165]
rabbitZhang et al., 1995 [188]
dogMolinari, 1970 [124]
monkeyMolinari et al., 1974 [125]
  with cyanoacrylate-monomersmonkeyBrassel et al., 1989 [15]
  with silver/gold ballsrabbitHegedüs et al., 1985 [67]
  with retractable threadratKoizumi et al., 1986 [98]
ratZarow et al., 1997 [186]
mouseConnolly et al., 1996 [30]
mouse mutantHara et al., 1997 [62]
ratCsiba et al., 1992 [33]
rabbitMolnar et al., 1988 [126]
monkeyBremer et al., 1975 [16]
 clot embolism
  with autologous blood clotratKudo et al., 1982 [103]
dogHill et al., 1955 [71]
  by photothrombosisratMatsuno et al., 1993 [118]
hypertensive
ratYao et al., 1996 [184]
  by thrombin infusionratZhang et al., 1997 [187]
 vasospastic occlusion
  with endothelin-1ratSharkey et al., 1993 [150]
catMacrae, 1992 [113]
Occlusion of vertebral arteries
gerbilHata et al., 1994 [64]
 and spinal arterydogGuo et al., 1995 [55]
Occlusion of basilary artery
 with clipgerbilYamada et al., 1984 [182]
 with silicon rubbercatNakahara et al., 1991 [128]
 by magnetic localization of
 iron filingsdogFujishima et al., 1970 [46]
Occlusion of the Circle
of Willis
dogSuzuki et al., 1980 [159]
monkeyWest and Matsen, 1972 [177]
Occlusion of posterior
choroidal arterymonkeyVajda et al., 1985 [166]
Occlusion of internal maxillary
arterygoatMiletich et al., 1975 [121]
Occlusion of pial arteries
 by embolization with steeldogPenry and Netsky, 1960 [140]
 balls
Occlusion of pial veins
 by photothrombosisratNakase et al., 1996 [129]
Microcirculatory occlusion
 by local endothelin-1ratAgnati et al., 1991 [1]
 by focal photothrombosisratWatson et al., 1985 [174]
 by ring photothrombosisratWester et al., 1995 [178]
Microembolism
 with microspheresratKogure et al., 1974 [96]
catVise et al., 1977 [170]
 with arachidonate-induced
 platelet aggregatesratFurlow and Bass, 1976 [47]
 with ADP-induced platelet
 aggregatesrabbitFurlow and Bass, 1976 [47]
 with air bubblesratJohansson, 1980 [86]
catFritz and Hossmann, 1979 [42]
dogSimms et al., 1970 [155]
monkeyMeldrum et al., 1971 [119]
Spontaneous infarcts
stroke-prone hypertensive ratOkamoto et al., 1974 [136]
  • 1 The presently most widely used models have been highlighted.

2 Focal cerebral ischemia

2.1 Intracranial vascular occlusions

One of the clinically most relevant models of animal experimental stroke is the occlusion of the middle cerebral artery. This vessel can be exposed by transcranial, postorbital or transorbital approaches, and can be clipped or ligated under the operating microscope. The transorbital approach requires removal of the eyeball but is atraumatic to the brain because it is not necessary to retract the tissue for exposure of the vessel. Surgical procedures for this approach have been described for cat, monkey, dog and rat but the present use is mainly restricted to cats and monkeys. Following implantation of an occluding device, vascular occlusion can even be performed in unanesthetized, unrestrained animals [66, 109]. Occlusion is also possible by local application of the potent vasoconstrictive agent endothelin-1 which produces severe vasospasm for a half maximal restoration time of 45 to 60 min [113].

Over the past years, the surgical exposure of the middle cerebral artery for occlusion has been widely replaced by intraluminal placement of flow-obstructing devices. The first attempts were made with permanent or retractable macroemboli consisting of silicon rubber or metal balls. Later, fine nylon threads with silicon-thickened tips have been introduced, particularly in rats and mice. These threads have several advantages: They avoid the need of craniotomy because they can be inserted from the external carotid artery, and they are easily withdrawn to produce transient focal ischemia. Placement of threads can also be carried out under remote control which enables continuous recordings of electrophysiological or even nuclear magnetic resonance tomographic and spectroscopic data [58]. The disadvantages of the thread technique are the difficulty of precise placement, the risk of vascular puncture and the large size of the infarct because the thread occludes the origin of both, the middle and the anterior cerebral artery. Several authors, therefore, insert the thread for a limited duration in order to grade the size of infarcts [20, 89]but this reversible type of focal ischemia produces a pathophysiology which is different from that of clinical stroke (see below).

The thread insertion approach has initially been described for use in rats but can also be applied to mice. This has opened the way to produce infarcts in transgenic animals which is one of the most powerful approaches to dissect the complex molecular interactions contributing to infarct evolution. The literature in this field is exploding (see [25, 32, 62, 173, 187, 189], to list only a few examples), but it should be remembered that the use of mutant animals also poses series conceptual and methodological problems. Some of these will be discussed in more detail in the chapter on therapeutical studies.

In focal ischemia, the reduction of blood flow is most severe in the center of the territory of the occluded artery [7]. This is basically different from incomplete global ischemia in which ischemia is most pronounced in the peripheral borderzone [17]. The size of the ischemic region depends on the efficacy of collateral blood supply. The territories supplied by the three major brain vessels – the anterior, middle and posterior cerebral arteries – are interconnected by the pial network of Heubner's anastomoses. Blood supply by this system depends on the anatomical configuration of the network, the vascular tone, blood viscosity and the blood pressure [75]. Under unfavourable conditions, ischemia may develop in the total distribution of the occluded vessel (maximal infarct), but the ischemic region may also be very small when collateral blood supply is optimal (minimal infarct) [192].

For this reason, numerous modifications of the middle cerebral artery occlusion model have been described for improving reproducibility of the lesion. As a general rule, infarcts become the more uniform the lower the animal species and the more proximal to the internal carotid artery the MCA is occluded. This is the reason for the frequent use of proximal MCA occlusion in rats which results in large, uniform infarcts [160].

A focal ischemia model of major clinical interest is clot embolism, not only because of the high incidence of thrombotic occlusion in clinical stroke but also because of the increasing use of thrombolytic therapy [115, 59]. However, production of clots and selection of the animal species has to be carefully done to account for differences in clot composition and interspecies differences of the coagulation system. Studies of this kind should, therefore, be carried out in close collaboration with experienced hematostasiologists.

2.2 Microcirculatory occlusion

In an attempt to produce ischemic lesions of constant size, Watson and co-workers [174]introduced the model of focal cortical photothrombosis. In this model the dye rose bengal is infused intravenously, followed by focal illumination of the cerebral cortex with laser light. Since rose bengal is a potent photosynthesizer, laser illumination results in massive microvascular coagulation of the exposed tissue. A modification of this technique produces a ‘ring’ lesion in which the center exhibits certain characteristics of the so-called ischemic penumbra [178].

2.3 Extracranial vascular occlusion

The technical difficulties encountered in all intracranial vascular occlusion models have prompted numerous attempts to produce focal brain ischemia by extracranial occlusion of the common carotid artery. However, in most species such an occlusion does not produce cerebral ischemia without further surgical interventions because the circle of Willis provides sufficient collateral blood supply from the other non-occluded neck vessels. The only exceptions are the Mongolian gerbil in which the circle of Willis is incomplete [107], the sheep in which the vertebral arteries do not unite to form a basilary artery [163], and the stroke-prone spontaneously hypertensive rat in which the collateral system is compromised [28]. But even in these animals, infarcts do not consistently develop. Extracranial ligation of a carotid artery in the gerbil produces infarcts in only about 30% of the animals. This incidence can be raised to 70% by additional coagulation of the contralateral external carotid artery in order to reduce collateral blood supply from the opposite side [14]. In other species, more complicated surgical interventions have to be carried out in order to produce focal brain ischemia by manipulation of extracranial arteries. In rats, carotid artery occlusion has been combined with intracranial hypertension [21]or by anastomosing the contralateral carotid artery with the jugular vein in order to reduce collateral blood perfusion pressure [9]. In monkeys and dogs, interruption of the circle of Willis has been carried out by clipping the anterior and posterior communicating arteries [177]. Subsequent extracranial ligation of the carotid artery produces a regional reduction of blood flow in the ipsilateral hemisphere. In the goat blood flow to the brain is via a rete mirabile which is supplied with blood from the carotid and the internal maxillary artery. After ligation of the internal maxillary artery and thrombosis of the extracerebral branches, blood supply to the brain is confined to the carotid artery, and subsequent ligation of this vessel causes ischemia of the ipsilateral side of the brain [121].

With the increasing use of the intraluminal middle cerebral artery thread occlusion technique, these extracranial occlusion methods have lost much of their appeal. However, the large size of thread-induced infarcts and the restriction of this model to small rodents are limitations that are not easily overcome. Experimental researchers should, therefore, be aware of these alternative approaches which have proved in the past to produce reliable and reproducible focal ischemic lesions.

A basically different way to enhance the ischemic impact following extracranial vascular occlusion is the combination with respiratory hypoxia or carbon monoxide poisoning. The classical model of anoxic-ischemic encephalophathy introduced by Levine [106]combines permanent unilateral carotid artery occlusion with repeated exposures to 100% nitrogen. This model has lost much of its interest because of poor reproducibility and the complexity of the pathophysiology which includes major cardiocirculatory abnormalities. An exception is the application to neonatal rats where the technical simplicity and the opportunity of longterm evaluation are major advantages [60]. The Levine model is, therefore, still the most often used model of hypoxia-ischemia in immature animals [169].

3 Global cerebral ischemia

3.1 Cardiac arrest

Global brain ischemia is a reduction or cessation of total cerebral blood flow. The clinically most relevant model is cardiac arrest which in animal experiments is readily produced by ventricular fibrillation or intracardiac injection of cardioplegic agents. A technically easy although rather traumatic approach is the intrathoracic compression method of Korbatchev [91, 100]where a hook is introduced into the chest to squeeze the large mediastinal vessels against the inner wall of the thorax. Cardiac arrest has also been produced by asphyxia [90], exsanguination [69]or drowning [114]. Finally, body circulation has been interrupted in animals connected to heart-lung-machines by temporarily switching the pumps off [84].

Cardio-circulatory failure affects not only the brain but all organs of the body which may interfere with the post-ischemic recovery of the brain in various ways. For this reason a number of models have been developed which reduce or interrupt cerebral blood flow without or with little interference of heart or other peripheral organs.

3.2 Complete brain ischemia

Examples of complete global ischemia are compression of the blood vessels in the neck by strangulation or inflation of a pneumatic cuff [87], the intra-thoracic occlusion of the innominate and left subclavian arteries which results in the interruption of blood supply to both carotid and vertebral arteries [49], or the increase of intracranial pressure above blood pressure by infusing fluids under high pressure into the cisterna magna [154]. In all models of selective cerebrocirculatory arrest care has to be taken to avoid collateral supply of blood to the ischemic brain [74]. This has been done by occluding, in addition to the main arterial supply, the internal mammary arteries [80], the pterygopalatine arteries [54], or by retrograde drainage of the occluded vessels [44, 181]. Collateral flow can also be reduced by lowering the blood pressure to hypotensive levels, either by bleeding or application of ganglion-blocking agents [80, 131].

For certain purposes, isolated head or brain preparations are used for the production of complete ischemia. In these preparations blood flow can be varied over a wide range simply by adjusting the pump speed of the extracorporeal circulation system [72].

Although the common pathogenic factor in all these ischemia models is global interruption of cerebral blood flow, considerable pathophysiological differences exist. When arterial blood supply is interrupted without simultaneous blockade of venous outflow, most of the blood escapes from the brain during ischemia, leading to anemic ischemia. The additional interruption of venous outflow, e.g. during strangulation ischemia, causes hyperemic ischemia with massive congestion of the cerebral vasculature. This difference is of importance for the recirculation after ischemia because the increased viscosity of stagnant blood requires a higher reperfusion pressure than following anemic ischemia [40].

Another difference is the extracellular fluid reservoir during ischemia. In anemic ischemia this reservoir is smaller than in hyperemic ischemia or compression ischemia with cisternal infusion of artificial CSF. Ischemic brain swelling, in consequence, is less pronounced during anemic ischemia and, therefore, interferes less with post-ischemic recirculation.

3.3 Incomplete brain ischemia

Incomplete global ischemia (oligemia) is produced by extracranial ligation of the carotid and vertebral arteries, by lowering arterial blood pressure in combination with bilateral carotid artery occlusion or by increasing intracranial pressure slightly below blood pressure. In gerbils severe oligemia can be produced by ligation of both common carotid arteries without additional occlusion of the vertebral arteries because in this species the circle of Willis is incomplete, and there is no connection between the basilary and the internal carotid artery [107]. By implantation of carotid artery occluders, this intervention can also be carried out in awake gerbils [26]. This is of particular interest for the investigation of repetitive ischemia because animals do not have to be re-anesthetized for each ischemic episode [164].

The main pathophysiological peculiarity of incomplete global ischemia is the fact that the critical perfusion pressure necessary for the maintenance of blood flow in the brain is compromised at first in the peripheral regions of the supplying territories of the cerebral arteries. Since these regions are located in the borderzones between these territories, the resulting decrease in blood flow is referred to as ‘border line’ or ‘border zone' ischemia [17]. Under clinical conditions, the temporo-parietal region is mostly endangered because it is the common borderzone between the territories of all three cerebral arteries [191].

Oligemia does not result in even reduction of flow at the microcirculatory level. Quite contrary, the local blood flow may vary greatly with low flow-rates in some and relatively high flow-rates in other parts of the brain. Without appropriate imaging techniques ischemic injury in a particular region is difficult to assess [176]. Another complicating factor of incomplete ischemia is the fact that the vasculature is slowly perfused with blood during the ischemic period. The resulting supply of fluid causes ischemic brain swelling and a compression of the microcirculation which may compromise the recirculation of the brain following ischemia. This factor is presumably one of the reasons why severe incomplete ischemia causes more pronounced damage than complete ischemia [80, 145].

3.4 In vitro ischemia

Recently, in vitro models have found increasing application for ischemia research. In these models primary neuron cultures [51], organotypic cultures [171]or brain tissue slices [180]are incubated in deoxygenated, glucose-free medium in order to mimic the interruption of the supply of oxygen and nutrients to brain parenchyma. Brain slices have also been prepared ex vivo from decapitated animals to study the post-ischemic recovery process in vitro [37]. All these models have several major disadvantages. First, the preparation of the slice is associated not only with severe tissue trauma but also with a period of ischemia before the slice is brought into the incubation medium. The control recordings of such preparations represent a post-traumatic, post-ischemic recovery state which may be basically different from the normal situation. This is particularly disturbing in studies of the hippocampus where brain trauma or a few minutes of ischemia are known to cause irreversible injury. Furthermore, in vitro preparations require incubation media which differ substantially from the normal extracellular environment and which, in contrast to the in vivo situation, provide an unlimited supply of extracellular solutes, notably sodium and calcium. Results obtained in such models have, therefore, little in common with the in vivo situation of brain ischemia and should be interpreted with caution.

4 Microembolism

Although not the dominating pathology under clinical conditions, several experimental models have been developed to study this particular pathophysiology. The most common method is the intracarotid injection of calibrated microspheres which have the advantage that the number and diameter can be standardized [170]. Other models of microembolism are the intracarotid injection of air bubbles [42]and the infusion of adenosine diphosphate [39], arachidonic acid [47]or phorbol-myristate-acetate (PMA, [2]) in order to induce platelet aggregation. An important pathophysiological difference between microembolism and other forms of ischemia is the fact that following microembolism the blood-brain barrier instantaneously breaks down [170]. Barrier damage presumably is the consequence of an ‘irritation’ of the vascular wall and not an ischemic event because, under pure ischemic conditions, the barrier breaks down only after several hours [137].

Another particularity of microembolism is the development of multifocal reactive hyperemia in the surrounding of occluded microvessels [42]. This leads to the sudden swelling of the brain, and an extremely inhomogeneous distribution of microcirculation with very low-flow rates in embolized and high-flow rates in non-embolized vessels. The net blood flow of the total brain, therefore, may remain in the normal range although a substantial number of brain vessels has been occluded.

5 Experimental models for therapeutic studies

The ultimate reason of brain ischemia research is the development of new therapeutic strategies for the treatment of cerebrovascular disease. Progress in the understanding of the pathophysiology of brain ischemia and the discovery of a great number of drugs that are able to interfere with the hemodynamic and molecular consequences of ischemia have aroused considerable expectations for the successful treatment of stroke or other clinically relevant manifestations of cerebrovascular disease. However, the treatment of brain ischemia in the clinic has been disappointingly unsuccessful [53, 35], indicating that the experimental conditions for the evaluation of drugs are difficult to match to the clinical situation.

Obviously, the failure to replicate the experimental data must be due to incompatibilities of the experimental models. Therefore, the most important consideration for the selection of the appropriate model is the identification of the correct pathophysiology. Other selection criteria are reproducibility, sensitivity, invasiveness and technical simplicity but none of these should be traded for the more important requirement that the pathophysiology of the model is relevant to the clinical situation.

In view of the large number of models that have been proposed for ischemia research, any recommendation must be highly subjective. The following proposals for therapeutic studies are mainly based on own experience and, therefore, are not necessarily congruent with other opinions.

5.1 Models for the treatment of stroke

The dominating hemodynamic pathophysiology of acute ischemic stroke is a gradient of declining blood flow from the periphery to the center of the supplying territory of the occluded artery [7]. According to the threshold concept of cerebral ischemia, blood flow correlates in a reproducible way with the sequel of biochemical disturbances, that progress from disturbances of protein synthesis at relatively mild degrees of ischemia to impairment of oxidative phosphorylation and energy failure at flow rates below 10–15% of control [78]. During the initial few hours of stroke, blood flow changes little; therefore, the appropriate animal experimental model for the evaluation of early treatment is a permanent type of vascular occlusion. The size of ischemic infarcts in such models is a function of the efficacy of collateral flow and may greatly vary in individual animals. As described above, attempts have been made to improve the reproducibility of the infarcts by reducing collateral blood supply but this maximizes the size of the lesion and reduces the potentials of therapeutic interventions. Such ‘maximal’ infarct models should only be used when treatments are tested which are targeted to improve flow-independent secondary effects in the surrounding of the ischemic infarct.

Recently, temporary occlusion of the middle cerebral artery for 1 to 3 hours has found increasing use because the size of the ischemic infarct can be varied as a function of the duration of vascular occlusion [89]. However, such models are of disputable relevance for treatment studies because the flow reduction in clinical stroke rarely resolves during this interval, and if so, the clinical symptoms are likely to disappear spontaneously without any interventions. Recirculation due to the inherent thrombolytic activity of the blood or neo-angiogenesis occurs after days rather than hours and is not as prompt as after experimental vascular occlusion. This difference is of considerable importance because the sudden recirculation after short periods of focal ischemia leads to the generation of oxygen radicals which have been implicated in the pathophysiology of reoxygenation injury [24]. Such injury, however, is of much lesser importance during permanent ischemia which explains that spin-traps or other drugs which ameliorate free radical induced injury reduce lesion size after transient but not after permanent middle cerebral artery occlusion [23].

A possibly more relevant indication for the use of transient vascular occlusion models is the development of adjuvant therapies in connection with thrombolysis. The high incidence of thrombotic vascular occlusions in clinical stroke has prompted research into thrombolytic therapy but previous attempts were of little benefit because improvement of flow was offset by an increased rate of hemorrhagic transformation of the ischemic lesion. Only recently, evidence has been provided that a carefully selected subgroup of patients responds favourably to recombinant tissue plasminogen activator, provided therapy is initiated within three, or better within 1.5 hours after the onset of symptoms [59, 115]. Obviously, the potentials of thrombolysis would greatly improve if secondary complications could be eliminated. It is this consideration which has led to the increasing use of transient focal ischemia models for pharmacological studies. However, the pathophysiology of reversible vascular occlusion differs also markedly from thrombolysis. In fact, the release of surgically-induced vascular occlusion results in the sudden restoration of the full arterial reperfusion pressure, whereas thrombolysis produces a gradual reopening of the vessel which is complicated by the formation of microemboli that are generated during lysis of the clot material. The appropriate experimental model of this type of reperfusion is, therefore, thrombolytic treatment of clot embolism rather than reversible thread insertion or other types of transient surgical occlusion.

5.2 Models for the treatment of global ischemia

As discussed above, the clinically most important and certainly the most dangerous form of global cerebral ischemia is that associated with cardiac arrest. Cardiac arrest differs from the more widely used experimental models of isolated cerebro-circulatory arrest by at least three factors: Flow arrest is complete, the recirculation after cardiac arrest is initiated at reduced blood pressure (particularly during the time of heart resuscitation), and the composition of the blood is more severely disturbed because whole body flow arrest produces much more pronounced alterations than selective interruption of brain circulation. The use of isolated brain ischemia models may, therefore, lead to serious underestimation of the importance of general pathogenic factors that determine the severity of brain injury under clinical conditions.

On the other hand, the various global brain ischemia models are well suited to detect brain-specific alterations including their responsiveness to therapeutic interventions. The success of such interventions should, however, be verified in a cardiac arrest model to confirm clinical relevance.

Among the brain-specific responses to circulatory arrest, two different categories of lesions have to be differentiated: 1. delayed neuronal death in the selectively vulnerable regions of the brain, notably the hilus and CA1 sector of hippocampus, the dorsal striatum and the reticular nucleus of thalamus, and 2. the early necrotic lesions in other parts of the brain. The first category of lesions is consistently produced by brief episodes of flow arrest in all models of global ischemia, whereas the manifestation of necrotic lesions outside the vulnerable areas depends not only on the duration, completeness and temperature of ischemia but also on the state of post-ischemic reperfusion [76]. Therapeutic intervention may, therefore, lead to greatly divergent results in different models of global ischemia, and have to be replicated in relevant disease models to predict their usefulness for the treatment under clinical conditions.

5.3 In vitro models for treatment studies

In following the above argumentation, screening of new drugs in in vitro models of ischemia is of little relevance because the basic pathophysiological differences between ischemia produced in vivo and in vitro may lead to false positive or false negative results. However, such models may be useful for exploring the mechanisms of drugs that have been previously shown to improve ischemic brain damage under clinically more relevant experimental conditions. Similarly, refinements of drug actions can be more easily achieved in vitro, provided that results are verified in vivo. Using this strategy, combination of in vivo and in vitro approaches has proved very successful in the past and should be used more frequently in the future.

5.4 Transgenic animals for treatment studies

The increasing availability of genetically modified animals has stimulated intensive research into the mechanisms of ischemic injury. The most widely used species for gene targeting is the mouse, and clinically relevant experimental models of both focal and global ischemia are available for this species [30, 10].

Similar to most in vitro approaches, transgenic animals are of greater interest for evaluating the molecular concepts of therapeutic strategies than for screening new drugs. An example of this is the use of neuronal and endothelial nitrous oxide synthase (NOS) knockouts for dissecting the complex therapeutic effects of NOS inhibitors [82]. Similarly, the observation that infarct size is smaller in Cu–Zn superoxide dismutase overexpressing mice led to the use of free radical scavengers for the treatment of transient ischemia [24].

A major methodological problem of transgenic animals is the fact that the parent strains used for gene targeting may exhibit different susceptibilities to ischemia. As an example, middle cerebral artery occlusion produces larger infarcts in C57Black/6 than SV-129 mice [30], an effect which can be related to differences in the vascular anatomy of the two strains (Maeda et al., in preparation). The ischemic susceptibility of gene-manipulated and wild-type animals derived from these strains may, therefore, differ, and the resulting difference in the severity of the ischemic injury may interfere with the gene-dependent molecular influence of infarct evolution.

Another problem is the body weight, which may markedly differ between mutant and wild-type littermates. This is particularly important for the widely-used thread occlusion of middle cerebral artery because the size of the artery changes with body weight. In order to assure similar reduction of blood flow in mutant and wild-type animals, the size of the thread has to be matched individually to the body weight [65].

Finally, the possibility of hemodynamic or molecular susceptibility differences exist, as recently demonstrated in C57Black/6 and SV-129 strains for global ischemia [43]and for kainate-induced excitotoxicity [147]. The use of transgenic animals for testing the efficacy of anti-ischemic drugs requires therefore careful validation experiments to exclude non-specific side effects.

6 Conclusions

The large number of animal experimental models that have been developed for the study of the pathophysiology and therapy of brain ischemia provides the opportunity to select an appropriate model for each task. However, this richness also bears the risk that without proper knowledge of the individual pathophysiology, mis-selections may lead to erroneous conclusions. Examples of these have been given. An important consideration is also the strict control of physiological variables because even minor changes in body temperature, blood pressure, blood gases or glucose levels are sources of variability which may confound the validity of the model. In order to avoid misconceptions and to ensure a maximum of medical relevance, animal models should be selected according to the following criteria: 1. The medical problem under investigation (e.g. ischemic or hemorrhagic stroke, brain resuscitation after cardiac arrest, etc.) should be clearly defined and the characteristic pathophysiological criteria of this condition should be established. 2. The experimental model should be chosen in accordance with these criteria, to match the kind and pattern of hemodynamic and/or metabolic disturbances with that of the clinical condition. 3. The leading mechanism of the pathological process should be identified and clearly differentiated from epiphenomena of the lesion which do not require specific treatment. 4. Treatment should interfere with the suspected cause, and the intended effect of the intervention should be documented (e.g., hemodynamic interventions should result in an increase of blood flow, application of free radicals scavengers should result in improvement of free radical mediated biochemical alterations, etc). 5. Clinical recommendations should be given only when endpoint evaluations of the therapeutic interventions were made in the appropriate clinically relevant model.

Regrettably, not all of the treatment studies comply to these criteria, if rigorously tested. This explains the disappointing results of clinical trials and stresses the need to clarify the relevance of the experimental model with regard to cerebrovascular disease. It would be highly desirable if a consensus could be reached to use standardized experimental models in different laboratories in order to facilitate communication and to avoid that data obtained in inadequate models are misused to promote the application of inefficient drugs for the treatment of cerebral ischemia under clinical conditions.

References