Rodent intracerebral hemorrhage models

CL MacLellan et al

522

Figure 5 (A) Representative T2-MRI scans in both ICH models from 6 h to 6 weeks after stroke. (B) Average (7s.e.m.) volume of tissue lost (mm3) determined with T2-MRI. Calculations of injury volume from MR images were inaccurate at early time points. The volume of tissue lost increased from 1 to 4 weeks in the collagenase model, but not in the blood model. The collagenase model produced significantly greater tissue loss than the in blood model. (C) Magnetic resonance imaging-determined hemispheric volume from 6 h to 4 days after ICH compared with control tissue. The ipsilateral hemisphere was significantly greater than the contralateral hemisphere due to edema and/or mass effect, especially in the collagenase model. (D) Relationship between MRI-determined injury and histopathology both at 6 weeks after stroke.

Discussion

Infusions of autologous blood or collagenase are the most widely used rodent models of ICH. Presumably, the better model is the one that more closely mimics the pathophysiology and functional consequences of ICH in humans while taking into account practical issues (e.g., surgical difficulty, assessment characteristics). In this study, we are the first to directly compare these models (matched for hematoma size) using numerous end points. Our primary findings are (1) greater primary injury occurs in the collagenase model, (2) injury also occurs distal to the hematoma in both models, but more so after collagenase infusion, (3) neurologic deficits resolve more rapidly and completely in the blood model, (4) MRI assessment fails to accurately gauge the tissue injury early after ICH, whereas it overestimates lesion size weeks later, and (5) MRI can track the evolution of the hematoma degradation, which occurs faster in the blood model. These findings underscore the necessity of using both models in testing putative therapeutics and they also illustrate the difficulties with the unimodal assessment approach common in stroke research.

It is interesting that the collagenase model caused much greater structural injury than the blood model

despite the fact that the whole-blood model had a somewhat larger hematoma. There are several possible reasons for this. First, collagenase injures many blood vessels, depending on dosage and diffusion characteristics, resulting in blood broadly dissecting throughout the parenchyma, whereas infused blood pooled within the striatum and between the striatum and corpus callosum. Accordingly, more brain cells were exposed to degenerating erythrocytes and probably inflammatory cells in the collagenase model, leading to a greater mass effect and edema, and presumably greater neurotoxicity (Atlas and Thulborn, 1998; Felberg et al, 2002). Second, BBB extravasation was greater in the collagenase model, especially within the hematoma. The temporal profile of BBB breakdown here is similar to that of other events that contribute to injury, including erythrocyte degeneration and edema formation (Xi et al, 2006). Third, the more widespread destruction of vasculature may cause some ischemic injury after collagenase infusion, at least within the diffuse hematoma, and this may be causing the greater BBB disruption. If so, this might mean that antiischemic therapies might fare better against collagenase-induced ICH than after blood infusion, as seems to be the case with the use of prolonged hypothermia in these models (MacLellan

Journal of Cerebral Blood Flow & Metabolism (2008) 28, 516–525

Rodent intracerebral hemorrhage models

CL MacLellan et al

523

Figure 6 Blood–brain barrier (BBB) disruption, measured via Gd-enhancement (see inset; mean signal intensity change7 s.e.m.), at 12 h, 2 days, and 4 days after stroke in (A) the perihematoma rim and (B) the hematoma. Collagenase caused significantly greater BBB disruption.

et al, 2004, 2006b). We did not assess cerebral blood flow in this study to determine the extent of ischemia in the hematoma or surrounding tissue and how this differs between models. Clearly, further mechanistic studies are needed to compare models regarding the contribution of ischemia among many other pathologic processes such as edema, inflammation, and oxidative stress because model differences will impact treatment efficacy.

Differences in injury might also relate to the rate of hematoma formation that is virtually instantaneous in the blood model compared with a gradual (hours) evolution in the collagenase model. However, one would predict that a slow bleed might be less devastating than a rapid one. Some ICH patients

Figure 7 Median Neurological Deficit Scale scores at baseline (day before ICH) and at 1 through 28 days after stroke in both models. Rats infused with collagenase sustained greater neurologic impairment than those infused with blood at all times after the stroke. Complete recovery was found in the blood infusion model, but not in the collagenase model where animals were still significantly impaired (versus baseline) at 28 days.

experience a single large bleed, but many undergo hematoma expansion over the first day (Fujii et al, 1994; Kazui et al, 1996). Accordingly, the collagenase model is clearly more suited to assessing treatments that affect bleeding and/or hematoma expansion such as hemostatic therapies (Kawai et al, 2006) or blood pressure manipulations. Likewise, the collagenase model might be better suited for detecting treatment side effects (e.g., elevated blood pressure caused by induced hypothermia; MacLellan et al, 2004).

The more severe and persistent functional impairment in the collagenase model undoubtedly stems from greater striatal destruction and additional injury to areas distal to the hematoma, such as the cortex and white matter tracts. Unfortunately, these other structures are not routinely assessed in experimental ICH studies (but see Felberg et al, 2002) despite calls for assessing both gray and white matter and for using models that have significant white matter injury (Report from a National Institute of Neurological Disorders and Stroke workshop, 2005). Several ICH studies report functional improvement without a reduction in striatal injury (Peeling et al, 2001), which might have resulted from a reduction in injury distal to the lesion epicenter.

Recovery from the initial functional impairment was considerably more gradual and incomplete in the collagenase model (MacLellan et al, 2006a, b, c) despite similar impairment at the outset. The rapid and apparently complete behavioral recovery in the

Journal of Cerebral Blood Flow & Metabolism (2008) 28, 516–525

Rodent intracerebral hemorrhage models

CL MacLellan et al

524

blood model is at odds with the persistent and debilitating deficits seen in ICH patients (Fieschi et al, 1988; Fogelholm et al, 1992). Furthermore, a rapid functional recovery is a significant problem in studies that seek to assess long-term outcome (e.g., floor/ceiling effects), which is of great importance in cytoprotection studies. Thus, a simple Neurological Deficit Scale should not be the only measure used in the whole-blood model; instead, or additionally, investigators should use more sensitive tests such as the forelimb use asymmetry task (Hua et al, 2002) and the single pellet reaching task (MacLellan et al, 2006c). Ideally, a sensitive functional assessment battery should be identified and used for each ICH model.

Magnetic resonance imaging is routinely used to detect hyperacute ICH, and has proven to be reliable in differentiating between ischemic and hemorrhagic insults (Fiebach et al, 2004; Schellinger et al, 1999). Furthermore, MRI is useful for identifying ICH and hematoma evolution in rodent models (Brown et al, 1995; Del Bigio et al, 1996; Hartmann et al, 2000). Long-term examinations of the temporal pattern of ICH in humans are rare, and much of the existing data are contradictory (Bradley, 1993; Fiebach et al, 2004). We cannot directly compare hematoma evolution in this study to human ICH, although it appears that both undergo the same stages of hemoglobin oxidation, which define the age of the hematoma (Bradley, 1993), and that the hematoma resolution occurs much faster in rodents. This is especially the case for the blood model as indicated by the earlier formation of a hemosiderin ring surrounding the hematoma in the chronic phase of ICH. Other pathophysiologic processes, such as edema and inflammation, also occur sooner after ICH onset in rodents compared with humans (Xi et al, 2006) and thus our findings are not surprising. Nonetheless, they raise serious concerns about predicting efficacy and therapeutic windows in patients from rodent work, thus arguing for the need to develop alternative models or to use other species.

Our findings illustrate an important technical limitation with T2-MRI in rodent ICH studies. That is, measurements of hematoma volume and/or tissue injury in the first few days after ICH were confounded by the mass effect of the hematoma and edema. Furthermore, we could not directly assess edema formation because the MR appearance of hemoglobin (hematoma) and vasogenic edema is indistinguishable soon after ICH (e.g., day 2) with T2-weighted images. Thus, alternative measures to gauge ICH size, such as hemisphere volume, may be more appropriate at such times. Finally, although MRI appears to be a valuable surrogate marker for ICH, we found that the volume of tissue lost was overestimated with MRI compared with histology. This discrepancy is probably due to the tissue thickness used here (0.8 mm for MRI versus 0.02 mm for histology), and could be partly avoided

by using a higher resolution (e.g., thinner slice thickness).

The collagenase and blood infusion models of ICH differ significantly in the extent and location of injury, and also in the recovery profile. Thus, researchers may prefer to use one model over the other in studies assessing long-term functional outcome (i.e., collagenase model) or pathophysiology of ICH (i.e., blood model). Our findings clearly show that the use of any one particular surrogate marker (e.g., histology) of benefit is fraught with limitations not the least of which is their inability to completely gauge recovery, which is the clinical end point of greatest concern. This holds true irrespective of the model and applies as well to other end points not presently used, as shown by several labs (MacLellan et al, 2006b; Wasserman and Schlichter, 2007). Advantages and disadvantages of each model, along with the fact that neither model perfectly mimics ICH in patients, strongly argue for using both models to assess putative cytoprotectants or rehabilitation therapies.

Acknowledgements

Technical assistance was kindly provided by G Chernenko, S Lee, and A Schellenberg.

References

Andaluz N, Zuccarello M, Wagner KR (2002) Experimental animal models of intracerebral hemorrhage. Neurosurg Clin N Am 13:385–93

Atlas SW, Thulborn KR (1998) MR detection of hyperacute parenchymal hemorrhage of the brain. AJNR Am J Neuroradiol 19:1471–7

Bradley Jr WG (1993) MR appearance of hemorrhage in the brain. Radiology 189:15–26

Broderick JP, Adams Jr HP, Barsan W, Feinberg W, Feldmann E, Grotta J, Kase C, Krieger D, Mayberg M, Tilley B, Zabramski JM, Zuccarello M (1999) Guidelines for the management of spontaneous intracerebral hemorrhage: a statement for healthcare professionals from a special writing group of the Stroke Council, American Heart Association. Stroke 30:905–15

Brott T, Broderick J, Kothari R, Barsan W, Tomsick T, Sauerbeck L, Spilker J, Duldner J, Khoury J (1997) Early hemorrhage growth in patients with intracerebral hemorrhage. Stroke 28:1–5

Brown MS, Kornfeld M, Mun-Bryce S, Sibbitt RR, Rosenberg GA (1995) Comparison of magnetic resonance imaging and histology in collagenase-induced hemorrhage in the rat. J Neuroimaging 5:23–33

Bullock R, Mendelow AD, Teasdale GM, Graham DI (1984) Intracranial haemorrhage induced at arterial pressure in the rat. Part 1: description of technique, ICP changes and neuropathological findings. Neurol Res 6:184–8

Choudhri TF, Hoh BL, Solomon RA, Connolly Jr ES, Pinsky DJ (1997) Use of a spectrophotometric hemoglobin assay to objectively quantify intracerebral hemorrhage in mice. Stroke 28:2296–302

Journal of Cerebral Blood Flow & Metabolism (2008) 28, 516–525

	Rodent intracerebral hemorrhage models
	CL MacLellan et al
	525
Chu K, Jeong SW, Jung KH, Han SY, Lee ST, Kim M, Roh JK	MacLellan CL, Girgis J, Colbourne F (2004) Delayed onset
(2004) Celecoxib induces functional recovery after	of prolonged hypothermia improves outcome after
intracerebral hemorrhage with reduction of brain	intracerebral hemorrhage in rats. J Cereb Blood Flow
edema and perihematomal cell death. J Cereb Blood	and Metab 24:432–40
Flow Metab 24:926–33	MacLellan CL, Gyawali S, Colbourne F (2006c) Skilled
Del Bigio MR, Yan HJ, Buist R, Peeling J (1996) Experi-	reaching impairments follow intrastriatal hemorrhagic
mental intracerebral hemorrhage in rats. Magnetic	stroke in rats. Behav Brain Res 175:82–9
resonance imaging and histopathological correlates.	Matsushita K, Meng W, Wang X, Asahi M, Asahi K,
Stroke 27:2312–9	Moskowitz MA, Lo EH (2000) Evidence for apoptosis
Felberg RA, Grotta JC, Shirzadi AL, Strong R, Narayana P,	after intercerebral hemorrhage in rat striatum. J Cereb
Hill-Felberg SJ, Aronowski J (2002) Cell death in	Blood Flow Metab 20:396–404
experimental hemorrhage: the ‘black hole’ model of	Mayer SA, Brun NC, Broderick J, Diringer MN, Davis S,
hemorrhagic damage. Ann Neurol 51:517–24	Skolnick BE, Steiner T, Investigators obotFT (2007)
Fiebach JB, Schellinger PD, Gass A, Kucinski T, Siebler M,	Randomized, placebo-controlled, double-blind phase
Villringer A, Olkers P, Hirsch JG, Heiland S, Wilde P,	III study to assess rFVIIa efficacy in acute intracerebral
Jansen O, Rother J, Hacke W, Sartor K (2004) Stroke	hemorrhage: the FAST trial. Cerebrovasc Dis 23:10
magnetic resonance imaging is accurate in hyperacute	Mayo NE, Neville D, Kirkland S, Ostbye T, Mustard CA,
intracerebral hemorrhage: a multicenter study on the	Reeder B, Joffres M, Brauer G, Levy AR (1996)
validity of stroke imaging. Stroke 35:502–6	Hospitalization and case-fatality rates for stroke in
Fieschi C, Carolei A, Fiorelli M, Argentino C, Bozzao L,	Canada from 1982 through 1991. The Canadian colla-
Fazio C, Salvetti M, Bastianello S (1988) Changing	borative study group of stroke hospitalizations. Stroke
prognosis of primary intracerebral hemorrhage: results	27:1215–20
of a clinical and computed tomographic follow-up	Newson P, Lynch-Frame A, Roach R, Bennett S, Carr V,
study of 104 patients. Stroke 19:192–5	Chahl LA (2005) Intrinsic sensory deprivation induced
Fogelholm R, Nuutila M, Vuorela AL (1992) Primary	by neonatal capsaicin treatment induces changes in rat
intracerebral haemorrhage in the Jyvaskyla region,	brain and behaviour of possible relevance to schizo-
central Finland, 1985 to 89: incidence, case fatality	phrenia. Br J Pharmacol 146:408–18
rate, and functional outcome. J Neurol Neurosurg	Peeling J, Yan HJ, Corbett D, Xue M, Del Bigio MR (2001)
Psychiatry 55:546–52	Effect of FK-506 on inflammation and behavioral
Fujii Y, Tanaka R, Takeuchi S, Koike T, Minakawa T,	outcome following intracerebral hemorrhage in rat.
Sasaki O (1994) Hematoma enlargement in sponta-	Exp Neurol 167:341–7
neous intracerebral hemorrhage. J Neurosurg 80:51–7	(2001) Recommendations for clinical trial evaluation of
Gladstone DJ, Black SE, Hakim AM (2002) Toward wisdom	acute stroke therapies. Stroke 32:1598–606
from failure: lessons from neuroprotective stroke trials	Report from a National Institute of Neurological Disorders
and new therapeutic directions. Stroke 33:2123–36	and Stroke workshop (2005) Priorities for clinical
Hartmann M, Jansen O, Deinsberger W, Vogel J, Sartor K	research in intracerebral hemorrhage. Stroke 36:e23–41
(2000) MRI of acute experimental intracerebral hema-	Rosenberg GA, Mun-Bryce S, Wesley M, Kornfeld M
toma. Neurol Res 22:512–6	(1990) Collagenase-induced intracerebral hemorrhage
Herbstein DJ, Schaumberg HH (1974) Hypertensive in-	in rats. Stroke 21:801–7
tracerebral hematoma. An investigation of the initial	Schellenberg AE, Buist R, Yong VW, Del Bigio MR, Peeling
hemorrhage and rebleeding using chromium Cr 51-	J (2007) Magnetic resonance imaging of blood–spinal
labeled erythrocytes. Arch Neurol 30:412–4	cord barrier disruption in mice with experimental
Hua Y, Schallert T, Keep RF, Wu J, Hoff JT, Xi G (2002)	autoimmune encephalomyelitis. Mag Res Med 58:
Behavioral tests after intracerebral hemorrhage in the	298–305
rat. Stroke 33:2478–84	Schellinger PD, Jansen O, Fiebach JB, Hacke W, Sartor K
Kawai N, Nakamura T, Nagao S (2006) Early hemostatic	(1999) A standardized MRI stroke protocol: comparison
therapy using recombinant factor VIIa in a collagenase-	with CT in hyperacute intracerebral hemorrhage. Stroke
induced intracerebral model in rats. Acta Neurochir	30:765–8
Suppl 96:212–7	Stroke Therapy Academic Industry Roundtable (STAIR)
Kazui S, Naritomi H, Yamamoto H, Sawada T, Yamaguchi	(1999) Recommendations for standards regarding pre-
T (1996) Enlargement of spontaneous intracerebral	clinical neuroprotective and restorative drug develop-
hemorrhage. Incidence and time course. Stroke 27:	ment. Stroke 30:2752–8
1783–7	Wahlgren NG, Ahmed N (2004) Neuroprotection in
MacLellan CL, Auriat A, McGie SC, Yan RH, Huynh HD,	cerebral ischaemia: facts and fancies—the need for
De Butte MF, Colbourne F (2006a) Gauging recovery	new approaches. Cerebrovasc Dis 17(Suppl 1):153–66
after hemorrhagic stroke in rats: implications for	Wasserman JK, Schlichter LC (2007) Neuron death and
cytoprotection studies. J Cereb Blood Flow Metab 26:	inflammation in a rat model of intracerebral hemor-
1031–42	rhage: effects of delayed minocycline treatment. Brain
MacLellan CL, Davies LM, Fingas MS, Colbourne F	Res 1136:208–18
(2006b) The influence of hypothermia on outcome after	Xi G, Keep RF, Hoff JT (2006) Mechanisms of brain injury
intracerebral hemorrhage in rats. Stroke 37:1266–70	after intracerebral haemorrhage. Lancet Neurol 5:53–63

Journal of Cerebral Blood Flow & Metabolism (2008) 28, 516–525