Rodent models of focal cerebral ischemia: procedural pitfalls and translational problems.

ral
Experimental & Translational
ssBioMed CentStroke Medicine
Open AcceReview
Rodent models of focal cerebral ischemia: procedural pitfalls and
translational problems
Stefan Braeuninger* and Christoph Kleinschnitz
Address: Department of Neurology, Julius-Maximilians-Universitaet Wuerzburg, Josef-Schneider-Str. 11, 97080 Wuerzburg, Germany
Email: Stefan Braeuninger* - braeuninge_s@klinik.uni-wuerzburg.de; Christoph Kleinschnitz - christoph.kleinschnitz@mail.uni-wuerzburg.de
* Corresponding author
Abstract
Rodent models of focal cerebral ischemia are essential tools in experimental stroke research. They
have added tremendously to our understanding of injury mechanisms in stroke and have helped to
identify potential therapeutic targets. A plethora of substances, however, in particular an
overwhelming number of putative neuroprotective agents, have been shown to be effective in
preclinical stroke research, but have failed in clinical trials. A lot of factors may have contributed
to this failure of translation from bench to bedside. Often, deficits in the quality of experimental
stroke research seem to be involved. In this article, we review the commonest rodent models of
focal cerebral ischemia - middle cerebral artery occlusion, photothrombosis, and embolic stroke
models - with their respective advantages and problems, and we address the issue of quality in
preclinical stroke modeling as well as potential reasons for translational failure.
Introduction
Ischemic stroke is a common disease that is one of the
major causes of death and disability worldwide, as well as
being a significant economic burden. Given an aging pop-
ulation, ischemic stroke is projected to become even more
important in the future [1]. Animal stroke models have
shed light on the pathophysiology of ischemic stroke [2]
and numerous potential targets for stroke therapy have
been identified. As of 2006, a total of 7,554 results of
1,082 putative neuroprotective interventions in experi-
mental stroke have been published [3]. The translation of
these results from bench to bedside, however, has been
overall disappointing. To date, thrombolytic therapy
using recombinant tissue plasminogen activator is still the
only effective pharmacological treatment in acute
clinical and clinical study results. Inappropriate conduct
of preclinical research (especially insufficient quality) as
well as suboptimal design of clinical trials (especially
patient selection criteria and clinical evaluation) have
been identified [5]. Lessons from translational failures led
to the Stroke Therapy Academic Industry Roundtable
(STAIR) conferences of academic and industrial experts
from which recommendations and standard operating
procedures for preclinical and clinical stroke research have
been devised [6-10]. Ongoing deficits in experimental
stroke research have been intensely discussed and ana-
lyzed, and additional useful recommendations for quality
improvement have been published [11-15]. In this review
article, the commonest rodent models of focal ischemic
stroke and their respective advantages and disadvantages
Published: 25 November 2009
Experimental & Translational Stroke Medicine 2009, 1:8 doi:10.1186/2040-7378-1-8
Received: 13 August 2009
Accepted: 25 November 2009
This article is available from: http://www.etsmjournal.com/content/1/1/8
© 2009 Braeuninger and Kleinschnitz; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 11
(page number not for citation purposes)
ischemic stroke [4]. Many factors have been discussed that
may have contributed to the lack of concordance of pre-
are presented and critically discussed with a focus on qual-
ity issues and translational problems.

Experimental & Translational Stroke Medicine 2009, 1:8 http://www.etsmjournal.com/content/1/1/8
Ischemic stroke models
The classification of animal models of ischemic stroke can
be based on a variety of factors such as species, occlusion
mechanisms and stroke etiology, presence or absence of
reperfusion (temporary/transient or permanent
ischemia), the involved vascular territory and infarct dis-
tribution, or a combination of these factors.
A classification of experimental ischemic stroke according
to infarct distribution seems reasonable, and with this
approach global, focal and multifocal ischemia models
can be differentiated [16]. Global cerebral ischemia, a
model of circulatory failure, cannot be regarded as a
stroke model in the strict sense and is beyond the scope of
this review. Focal cerebral ischemia occurs when blood
flow in a specific brain region is critically limited. It is usu-
ally the result of occlusion of a major cerebral artery, such
as the middle cerebral artery. Multifocal ischemia repre-
sents a reduction of cerebral blood flow in multiple
regions and has been described as patchy reduction of cer-
ebral blood flow [16]. Multifocal cerebral ischemia can be
caused by injection of embolus material into a brain-sup-
plying artery.
The etiology of ischemia can be categorized as extravascu-
lar or intravascular [17]. Extravascular mechanisms are
vessel ligation, electrocauterization, or clipping. An inno-
vative method employing an extravascular mechanism is
local application of endothelin-1 adjacent to a brain-sup-
plying artery [18,19] or on the brain surface [20]. Intravas-
cular mechanisms include an intravascular occluding
suture in focal cerebral ischemia and injection of blood
clots and other embolus material in multifocal cerebral
ischemia models.
An overview of cerebral ischemia models is given in Table
1. For ethical and practical reasons, rats and - especially in
transgenic studies - mice are the most widely used labora-
tory animals in preclinical stroke research. In the follow-
ing, we focus on major models of focal stroke in these
rodents.
Major rodent models of focal stroke
Intraluminal thread middle cerebral artery occlusion
model
The middle cerebral artery occlusion (MCAo) model
using an intraluminal thread was developed by Koizumi
et al. in rats [21], and has subsequently been modified
[22,23] and adapted to mice [24]. Briefly, an occluding
suture or monofilament is advanced via the common
carotid artery into the internal carotid artery towards the
junction of the anterior and middle cerebral arteries. Thus
the middle cerebral artery vascular territory is subjected to
ischemia. The occluding suture may be removed after var-
ious time periods, allowing for reperfusion (transient
MCAo) or it may be left in place permanently (permanent
MCAo).
There are several advantages of this technique: it models
focal infarction in a large vascular territory and does not
require craniotomy. Thus it can be regarded as relatively
simple, though microsurgical skills are required.
Table 1: Overview of experimental stroke models
Cerebral Ischemia Etiology Reperfusion (transient ischemia) or
permanent ischemia
Examples
Global (complete or incomplete) = model
of circulatory arrest or severe
hypotension
Intravascular Transient or permanent Cardiac arrest with or without
cardiopulmonary resuscitation
Extravascular Transient or permanent Cervical compression by neck cuff;
ligation of several brain-supplying
arteries
Focal Intravascular Transient or permanent Intraluminal thread middle cerebral
artery occlusion model
Extravascular Transient or permanent Surgical middle cerebral artery
occlusion models using ligation, clipping,
electrocauterization etc.; endothelin-1-
induced middle cerebral artery
occlusion
Multifocal Intravascular Transient (spontaneous lysis or Embolization models using blood clots, Page 2 of 11
(page number not for citation purposes)
thrombolytic therapy possible in blood
clots) or permanent
microspheres or other embolus
material

Experimental & Translational Stroke Medicine 2009, 1:8 http://www.etsmjournal.com/content/1/1/8
Specific complications of the intraluminal thread MCAo
model have been reported. MCAo may be complicated by
subarachnoid hemorrhage in a substantial number of ani-
mals [25-28]. Other adverse events include ipsilateral ret-
inal injury and consequent visual dysfunction [29],
ischemia in the external carotid artery territory [30,31],
intraluminal thrombus formation [32,33], and inade-
quate MCAo and premature reperfusion [26,27,34]. If the
infarct involves the hypothalamus - as seems to be fre-
quently the case in animals subjected to permanent MCAo
or to longer times of transient MCAo - hyperthermia may
develop as a consequence of thermal dysregulation, lead-
ing to altered stroke outcome [35-37]. Technical modifi-
cations and the use of standardized monofilaments can
reduce the incidence of these complications [28,38,39]. In
line with differences in the technical approach, research-
ers may report tremendously different incidences or may
not produce certain complications in their hands at all
[27]. For example, external carotid artery territory
ischemia, such as temporal muscle necrosis resulting in
impaired outcome (decreased body weight and delayed
recovery), has been reported as frequently as in 50% of
animals [30,31], but has not been confirmed by others
and seems to be related to the use of electrocauterization
for vessel dissection [40]. In mice [25,41-43], but also in
rats [44-46], infarct volumes vary considerably between
different strains and ages, and show a steep relationship to
occlusion time [47]. For example, C57Bl/6 mice have sig-
nificantly larger infarcts than SV129 mice in the perma-
nent MCAo model [25,42], most likely due to the absence
of one or both posterior communicating arteries in many
C57Bl/6 mice [43,47,48]. These are the most widely used
parent strains for the production of transgenic mice.
Moreover, infarct volumes found in different murine
MCAo studies using the same mouse strain and same
MCAo duration can range over a fivefold difference [49].
The different absolute and relative incidences of compli-
cations, as well as minor technical differences, may con-
tribute to the considerable variability observed with the
intraluminal thread MCAo model.
Surgical middle cerebral artery occlusion models
There are also models of MCAo that do not use an intra-
luminal occluding suture, but employ an extravascular
mechanism (for example, ligation, clipping, or electrocau-
terization). These surgical MCAo models require craniot-
omy and incision of the dura. For example, the proximal
middle cerebral artery can be identified and occluded after
removal of the coronoid process of the mandible and
zygoma and opening a burr hole lateral to the foramen
ovale [50]. A more refined method that has been reported
to cause reproducible infarctions is tandem occlusion of
the distal middle cerebral and ipsilateral common carotid
of cerebral blood flow auto-regulation by damage to auto-
nomic nerves, and craniotomy. The latter has been shown
to alter brain temperature, intracranial pressure, and
blood-brain barrier permeability [52,53]. For these rea-
sons, most researchers prefer the intraluminal thread
MCAo method.
Photothrombosis model
In the photothrombosis model originally described by
Watson et al. in rats [54] and later modified for mice [55],
a cortical brain lesion is induced by systemic injection of
a photosensitive dye, such as Rose Bengal, and subsequent
focal irradiation of the skull.
Though some researchers expose the dura or brain surface
prior to illumination, the method does not require crani-
otomy if the light penetrates the skull, and it can in this
case be regarded as minimally invasive. A special advan-
tage of photothrombosis is the high reproducibility of
lesion size and location. It is even possible to determine
the region of irradiation stereotactically and thus selec-
tively induce infarcts in cortical areas representing specific
functions [56]. The technique is relatively simple and
allows high throughput.
The penumbra, however, an important target of many
putative stroke therapeutics, is lacking in photothrombo-
sis, where blood-brain barrier disruption and vasogenic
edema develop within minutes [57,58], though modifica-
tions (ring models) seem to be able to produce a penum-
bra-like lesion edge [59,60]. The mechanism by which
photochemically induced brain lesions develop is
thought to involve vascular endothelial damage, platelet
activation and subsequent thrombotic vessel occlusion
[57,58]. Recently, however, we have shown that the pho-
tothrombotic lesion develops independently of the pres-
ence of functional platelets or plasmatic coagulation [61].
Thus photothrombosis does not reflect vascular-ischemic
brain injury or stroke, but focal brain necrosis seems to be
caused directly and irrespectively of additional endothe-
lial damage leading to local thrombosis. These findings
reveal significant differences in mechanisms of tissue
injury induced by photothrombosis as compared to focal
ischemia induced by MCAo, where platelet activation and
the intrinsic coagulation pathway are instrumental
[62,63]. When evaluating antithrombotic pharmaceuti-
cals to limit tissue injury using the photothrombosis par-
adigm, one should therefore be cautious because ensuing
platelet-containing thrombosis may be present, but may
not be mandatory to induce tissue damage.
Models of embolic cerebral ischemia
The first embolic stroke model was developed in dogs [64]Page 3 of 11
(page number not for citation purposes)
arteries [51]. Disadvantages of the surgical approaches,
however, include technical intricacy, possible impairment
and has subsequently been adapted to rats [65] and mice
[66]. After extensive methodological modifications and

Experimental & Translational Stroke Medicine 2009, 1:8 http://www.etsmjournal.com/content/1/1/8
novel technical developments, various experimental mod-
els of embolic cerebral ischemia exist today. Endothelial
injury of large vessels, for example by irradiation of the
common carotid artery after administration of a photo-
sensitive dye [67], leads to local thrombus formation and
embolic stroke in the distal vascular territory. Other stroke
models use embolic materials, such as polyvinyl com-
pounds, latex microspheres, or blood thrombi that are
prepared ex vivo and injected into the common carotid
artery [68-70] or into the proximal middle cerebral artery
[66,71-73]. Thrombus formation can also be induced at
the origin of the middle cerebral artery by local thrombin
injection [74].
The main advantages of embolic stroke models are their
pathophysiological relevance - embolic vessel occlusion is
the most frequent cause of ischemic stroke in humans -
and, if blood-borne thrombi are used, the possibility to
test thrombolytic compounds. Those models not placing
emboli directly in the proximal middle cerebral artery,
however, often show high variability of infarct location,
distribution, and size (multifocal stroke model). Sponta-
neous lysis of thrombi can occur and is difficult to assess.
Moreover, placement or induction of emboli in the prox-
imal middle cerebral artery is technically demanding. A
common complication, especially in mice, is subarach-
noid hemorrhage, reported in about 40% of mice [66]. In
emboligenic endothelial lesions of the common carotid
artery, lesion type (endothelial denudation versus photo-
thrombosis) may determine the fibrin content of the
thrombi, and thus their embolic potential [75].
As most experimental stroke models address the anterior
circulation, models of vertebrobasilar occlusion are sparse
and have been developed in larger animals such as rab-
bits, cats, and dogs. Recently, however, the autologeous
thromboembolic stroke model [68] has been adapted to
the posterior (vertebrobasilar) circulation in rats [76]
allowing the study of stroke pathophysiology in the brain-
stem and cerebellum that may differ from the situation in
the anterior circulation.
Potential reasons for translational failures and
proposed actions
Species differences
There are, of course, significant physiological, neuroana-
tomical and metabolic differences between humans and
small rodents, which are the most widely used experimen-
tal animals in preclinical stroke research. For example,
small rodents usually require higher drug doses on a mg/
kg body weight basis for a similar effect than larger mam-
mals [77]. Thus, effective doses derived from preclinical
stroke studies in small rodents cannot simply be trans-
body weight. Dose-response curves in laboratory animals
and in humans can be helpful to address this problem.
Moreover, it should be kept in mind that mice and rats are
lissencephalic. Therefore, if a drug has been effective in
preclinical stroke studies in small rodents, it is recom-
mended to reproduce the result in higher species (in par-
ticular non-human primates; for review, see Fukuda and
del Zoppo [78]) prior to the initiation of clinical trials [6].
These animals are gyrencephalic and more closely resem-
ble the situation in humans.
Strain differences
Strain differences in mice [25,41-43] and rats [44-46]
must not be underestimated and imply that results in one
strain may not necessarily be reproducible in another
strain. The profound genetic differences and phenotypic
variance between mouse strains can explain that certain
isogenic strains poorly mimic human disease or even that
the effects of a targeted mutation are overshadowed,
which has raised concern about the common practice of
using a single mouse strain, mostly C57BL/6, to address
many questions [79]. In transgenic studies in particular,
not only the single gene mutation but also the genetic
background must be taken into account [79]. Thus, pre-
clinical studies of interventions in stroke should include
more than one rat or mouse strain.
Sex, age, and comorbidities
Preclinical stroke experiments are often restricted to juve-
nile male animals to avoid the variability caused by
female hormone cycling. It was assumed that pathophys-
iology and response to therapy seen in male animals
would also apply to the other sex. In various experimental
stroke models, however, young female adult rodents had
smaller infarcts than their male counterparts [80-82]. In
rats, there seems to be a correlation to the estrous cycle,
with high endogenous estradiol levels in proestrus being
correlated with smaller infarcts than low endogenous
estradiol levels in metestrus [83]. In human observational
studies, premenopausal women have a smaller incidence
of ischemic stroke than men; and stroke risk in both gen-
ders increases with age [84].
There is a complex and significant influence of age on
stroke outcome. Most studies have reported more severe
consequences of cerebral ischemia in aged as compared to
young animals. For example, older rats have been shown
to develop larger infarcts [85,86]. Others, however, have
reported contrary results with a more favorable outcome
in aged animals [45,87]. The correlation between age and
brain damage may not be linear [88]. Aging is associated
with microvascular changes in laboratory rodents [89],Page 4 of 11
(page number not for citation purposes)
ferred to the situation in humans, even if adjusted for and a small decline of cerebral blood flow has been
observed [90].

Experimental & Translational Stroke Medicine 2009, 1:8 http://www.etsmjournal.com/content/1/1/8
Rats suffering from streptozotocin-induced diabetes and
spontaneously hypertensive rat strains have been shown
to develop larger infarcts [45], illustrating the impact of
comorbidities.
Taken together, the influence of species, strain, sex and
age (and, if applicable, comorbidities) on experimental
stroke must be taken into account. Whereas preclinical
stroke research often uses healthy male juvenile rodents, a
considerable number of stroke patients in clinical studies
will be female, elderly, or suffer from comorbidities such
as diabetes, hypertension and atherosclerosis.
Stroke model
Selection of the most appropriate stroke paradigm is criti-
cal. Models requiring craniotomy, such as proximal mid-
dle cerebral artery ligation, are traumatic and do not
mimic human stroke very closely. MCAo by an intralumi-
nal thread and models using blood clot emboli are more
similar to the clinical situation. Because human stroke is
heterogeneous, however, there can be no single ideal
stroke model. The different experimental stroke para-
digms vary with respect to the underlying pathophysio-
logical mechanisms and the size or location of the lesion.
It is thus prudent to choose an approach that matches the
anticipated situation in stroke patients as closely as possi-
ble. For example, thromboembolic models using blood
clot emboli are helpful when the efficacy of thrombolytic
drugs is to be examined. Ideally, the results should be
reproducible in different models of focal ischemic stroke
and a novel substance should be tested both with and
without reperfusion [6].
Not only should the stroke model reflect the clinical situ-
ation, but the mechanism of action of a putative stroke
drug or intervention should also be relevant for human
stroke pathophysiology and, thus, has to be elucidated
prior to the initiation of clinical studies. It is, for example,
important that potential neuroprotective drugs are able to
penetrate the blood-brain barrier. Cerebral ischemia
causes disruption of the blood-brain barrier, but this
occurs only several hours after stroke onset [91].
Anesthesia
For practical and ethical reasons, experimental focal
stroke has to be induced under appropriate anesthesia
and analgesia. Any type of anesthesia, including inhala-
tion anesthesia, can, however, alter stroke outcome [92].
Barbiturates have various side effects including hypother-
mia and may mediate neuroprotection [93,94], thus inter-
fering with putatively neuroprotective interventions.
Concern has also been raised about the use of ketamine
that could overestimate the neuroprotective effect of an
ics concerning the control of the depth of anesthesia [96].
Even if inhalation anesthesia is used, however, animals
under general anesthesia that breathe spontaneously have
been shown to exhibit larger infarct volumes and a higher
variability of physiological parameters than those
endotracheally intubated and mechanically ventilated
[96].
Potential pitfalls may thus be avoided by selection of
appropriate anesthesia. Mechanical ventilation seems to
reduce possible side effects of anesthesia on stroke out-
come best, but may be too laborious to be applied rou-
tinely. Inhalation anesthesia should be preferred to
intravenous or intraperitoneal administration of anesthet-
ics, and it seems prudent to avoid barbiturates and keta-
mine.
Flaws in basic study design
Only a minority of published experimental stroke studies
have reported randomized treatment allocation and
blinding of investigators [15]. In a metaepidemiologic
approach examining 13 meta-analyses of experimental
stroke studies, those studies with unblinded induction of
ischemia reported considerably greater effects than
blinded studies demonstrating bias in the design of pre-
clinical stroke studies [97]. Thus, basic universal standards
of scientific research such as blinding of investigators
should be strictly followed in modeling cerebral ischemia.
This should also include confirmation of results by differ-
ent independent laboratories prior to the initiation of
clinical studies.
Therapeutic time window
There is a therapeutic time window in ischemic stroke
[98]. It is defined as the time interval during which an
intervention can lead to partial or total recovery by resto-
ration of perfusion (reperfusion window) or by protection
of penumbra tissue from necrosis (neuroprotective win-
dow). It has also been noted that a rigid and universal
time window for acute stroke therapy does not exist and
that the specific pathophysiological state must be taken
into account to assess the individual therapeutic potential
[99].
The therapeutic time window in ischemic stroke may dif-
fer critically between species, and, in addition, depends
on the mechanism of drug action and is influenced by fac-
tors such as temperature or collateral circulation. The time
point at which a certain drug is administered during
stroke development should thus be carefully selected and
has to balance maximum efficacy on the one hand and
clinical applicability on the other hand. Early drug admin-
istration, or even pretreatment, is usually the most prom-Page 5 of 11
(page number not for citation purposes)
intervention [95]. Inhalation anesthesia is superior to
intraperitoneal or intravenous administration of anesthet-
ising approach regarding efficacy [100], whereas for
routine clinical practice it is desirable to develop drugs