This article is mentioned in:

1. Introduction

In neurosurgical setting, intracranial hematomas (ICH) are a frequent condition with acute intracranial expansive effect (1), that lead to secondary injuries and complications, such as intracranial hypertension (IHT) (2,3), which may become the cause of death for these patients (4). In the last decades, spontaneous ICH prevalence has remained stable, with an average of 24.6 cases per 100,000 inhabitants per year in developed countries (5). Systemic arterial hypertension is the most important risk factor for non-traumatic bleeding, especially among subjects no adherent to antihypertensive treatment (4). Other causes for intracranial hemorrhages are aneurism rupture, arteriovenous malformation, vasculitis, coagulopathies, venous thrombosis, cocaine use, amyloid angiopathy and hemorrhagic complications after ischemic stroke thrombolysis (4).

ICH may present with a wide variety of signs and symptoms depending on the location and severity of the bleeding. Patients can be asymptomatic or have mild local deficits, whereas others present with IHT syndromes and complete loss of consciousness (4). Spontaneous ICH has an overall mortality rate of 50% after 30 days (6,7), and approximately half of deaths occur within the first 24 h of the initial bleeding (8). The functional prognosis for survivors is poor, and only 20% of patients are expected to be functionally independent at 6 months (9).

ICH may promote cerebral hemodynamic impairment with a shift from aerobic to anaerobic metabolism that leads to: i) Lactate and free radicals accumulation (10); ii) activation of inflammatory cascades (interleukin 1β and tumor necrosis factor) promoted by the complement, microglia, macrophages and neutrophils (11); iii) immune responses (and systemic immunosuppression) that contribute to blood-brain barrier impairment; and iv) additional swelling of the brain tissue, leading to IHT, which adds even more damage to the brain tissue (11,12).

When therapeutic strategies such as hypertonic saline or mannitol infusions, hyperventilation and mild hypothermia fail to control IHT, the spreading ischemia may contribute to brain death unless an emergency neurosurgical procedure as decompressive craniectomy is performed (13). Therefore, animal experimentation and testing remains an important tool to improve understanding of the different pathophysiological mechanisms of injury in order to investigate the techniques of intervention and neuroprotection improvement. Currently, to the best of our knowledge, the available literature has few models of experimental ICH in animals, which are characterized by heterogeneity and varied methodology. The aim of the present review was to discuss the main animal models for ICH investigation, emphasizing the advantages and disadvantages of each method.

2. Methods

The present comprehensive review included experimental original studies published in English. Reviews were assessed to build the body of evidence. The present study's aim was to identify relevant studies on animal models of ICH. The search was effectuated in October 2021 and updated in May 2022 through the electronic databases PubMed/MEDLINE (https://pubmed.ncbi.nlm.nih.gov/), Google Scholar (https://scholar.google.com.br/), LILACS (https://lilacs.bvsalud.org/en/) and EMBASE (https://www.embase.com/landing?status=grey), by two investigators (WSP and SB). The following terms were applied to identify potential eligible articles: ‘Animal model’ OR ‘experimental model’ AND ‘intracranial hematoma’ OR ‘cerebral hematoma’ AND ‘intracranial hypertension’, and were then selected by title and abstract. In addition, a manual search was done using the reference lists of included studies to identify others relevant papers, as the ‘Related Articles’ tool for the selection of additional relevant articles. Inclusion criteria included experimental studies of any animal species with the purpose of assessing the effects of hematomas and/or intracranial hypertension over the brain. Exclusion criteria comprised of experimental interventional studies for the assessment of systemic effects of ICH and studies not published in English. The funneling process for selection of studies is presented in Fig. 1.

Figure 1 Manuscript search and selection funneling.

3. Models

Regarding the animals used for modeling ICH, the following different species were used: Mice or rats (14-33), pigs (34-40), dogs (41-43), monkeys (44), sheep (45), cats (46) and rabbits (47,48). The majority of studies (63%) utilized rats, whereas the model most used is autologous blood intracranial injection (51%). The assessment of neuroinflammation cascades was the outcome studied in 26% of the studies. Fig. 2 summarizes the representation of each animal type, model and outcomes assessed in reproducing intracranial hematomas.

Figure 2 Models, animals and outcomes assessed after an intracranial hematoma simulation.

Although each model recreates the fundamentals of human hematoma with good precision, they differ in ways that influence outcome. The present study described the technique details of ICH induction and compared the technical and pathological advantages and disadvantages of the existing models (Table I). Each model is presented in further detail bellow.

Table I Comparison of intracranial hematomas modeling techniques.

Table I

Comparison of intracranial hematomas modeling techniques.

Model	Advantages	Disadvantages
Autologous blood injection	Reproducible	Does not mimic rupture of blood vessels
	Homogeneous volumes	Does not allow to evaluate rebleeding
	Clinically relevant comparison
Bacterial collagenase	Represent hematoma expansion	Overestimate inflammatory reaction
	Produces vasogenic edema	Neurotoxic effects
Cerebral blood vessels damage	Reproduces blood toxicity	Promotes ischemia
Intracranial balloon inflation	Reliable mass effect	Does not reproduce blood toxicity

Intracerebral injection of autologous blood

Blood injection models allow the observation of hemolysis-induced toxicity assessment, as the following immune-inflammatory responses (49). This model also allows the assessment of interventions to reduce swelling, hematoma expansion and the IHT effects on brain hemodynamics (49). Experimental models of brain hematomas have been described since the 1960's, and typically involve the intracerebral injection of autologous blood, which is a simple and effective technique for the production of brain parenchymal hematoma (50). This type of model has been developed for larger animals (cats, dogs, pigs, sheep and monkeys) (36,37,41-43,45,49,51,52) through the injection of blood in the frontal lobe or basal ganglia. There are several variations within studies, such as blood source, the amount of blood injected and depth of injection. Whisnant et al (43) induced an intracerebral collection in dogs by infusing venous blood in the deep white matter. Sussman et al (41), by contrast, injected arterial blood superficially into the right frontal lobe (0.5 cm beneath the cortex) in dogs. A similar technique was presented by Wagner et al in 1996, 1998 and 1999 (49,50), who produced lobar hematomas in pigs by infusions of arterial blood into right frontal white matter.

For smaller animals (rats and mice) blood is injected into the caudate nucleus in the majority of studies (53-60). Yang et al (53) injected autologous blood into the caudate nuclei of rats in order to study the formation and resolution of brain edema. As an alternative for basal ganglia, Bullock et al (61) injected autologous blood into the lateral ventricles to compare the consequences of contained and uncontained hemorrhage. They also evaluated the impact of the intracerebral collection on the IHT.

The autologous injection of blood method has a major advantage, which is allowing the production of hematomas with homogeneous volumes. It also mimics the rapid accumulation of blood noted to occur in the clinical setting (62). However, it does not reproduce the rupture of blood vessel present in human brain hematomas and can lead to intraventricular hemorrhage and/or subarachnoid hemorrhage by ruptures in the ventricular and subdural spaces (50). Moreover, there is a risk for the infused blood to back flow along the needle track (62). Nevertheless, it can be used for the study of biochemical and pathophysiological effects in patients with acute ICH and IHT as it permits the infused blood volume to be controlled, enabling the generation of hematomas sizes and mass effects (45).

Autologous blood injection allows the observation of hematoma capsule formation at the boundary tissue around the blood clot (53). It is composed by a necrotic layer of brain tissue, fibrin deposits, secondary capillary hemorrhages and white matter vacuolation in the first week (42), up to 1 cm from the ICH site. Impregnation of metalloproteinases and oxidative stress induced by ischemia contribute to these hemorrhagic phenomena and disruption of the blood-brain barrier (BBB) (63). Microglia and astrocytes seem to be more resistant to ICH effects, otherwise, neurons and axons are more sensitive to hypoxic-ischemic changes (45). At 3 days after ICH, red blood cell (RBC) morphology is significantly changed to spheric, with complement system activation responsible for RBC lysis following phagocytosis by macrophages and microglia (37).

The blood injection model was developed as a single intracerebral injection (64), but often produces inconsistent results due to the backflow of blood through the needle in rats (53). To minimize this complication, a dual-injection method has been developed, in which a small amount of blood is injected immediately above the target area of the brain, followed by a second injection of blood into the basal ganglia (64). The clot of the first injection prevents backflow along the needle (64). This model was first applied by Deinsberger in 1996(65), and since then has been used in numerous studies with small animals (66-69). The double injection model mimics a hematoma developing rapidly but does not induce real rupture of cerebral blood vessels (62). A major advantage of this method is that it helps to prevent the backflow of blood into the subarachnoid space (62). However, it is not a spontaneous hemorrhage and does not mimic vessel rupture and, therefore, it does not allow researchers to evaluate rebleeding.

Bacterial collagenase model

Bacterial collagenase is a protease that damages the extracellular matrix around the brain capillaries, weakening them and causing rupture of the vessel and consequent extravasation of blood (64). This model produces spontaneous intracerebral bleeding, which develops over several hours, as shown in ~30% of patients with ICH (65). It is considered appropriate for studying hematoma expansion and vasogenic edema, anticoagulation, axonal degeneration, iron-induced apoptosis, endothelial disturbances and BBB impairment, in addition to systemic complications following brain aggression (62,64,70).

The injection of bacterial collagenase in the basal ganglia, leading to the breakdown of the basement membrane of blood vessels was first introduced in the early 1990s using mice (71,72) and has been widely used ever since (21,73-77). Collagenases spread and penetrate into the brain parenchyma instead of remaining on the site of infusion; therefore, although the pathophysiologic mechanisms of injury are the same of those disclosed in the section of autologous blood infusion, they seem to be enhanced in this model (76).

Adjustments to procedural parameters have been seen between multiple studies. The majority of authors inject bacterial collagenase type IV (78), but some have adopted bacterial type VI (71), VII S (17,79-81) or XI (38) collagenase. The infusion period also varies, ranging from 2 to 16 min (71,78,80). This technique is preferably used in small animals (rats and mice). Mun-Bryce et al (38) was one of the few to try this model in larger animals by injecting collagenase into the primary somatosensory cortex of swine. It is considered a simple and reproducible method since it evokes a dose-dependent hemorrhage size and can be used in multiple species. As it can mimic hematoma expansion and vasogenic edema (51,70,82), it is commonly used to investigate the mechanisms of increased bruising and for developing treatments that affect cerebral homeostasis (23,83,84).

Using this model, it is possible to imitate vessel rupture and represent hematoma expansion. It allows to assess long term outcomes. Even though it evokes an inflammatory response earlier and for a prolonged period (when compared with the blood injection model), it seems that the inflammatory reaction is too severe and is different from the observed in the human brain (85,86). Furthermore, extensive bleeding resulting from the intracerebral injection of collagenase may produce an unplanned ischemic brain injury (87). As expected from the above, this technique leads to significantly increased severe neurological deficits with poorer recovery compared with blood infusion (88).

Cerebral blood vessels damage model

In this model, rats have their cortical veins exposed via a craniotomy and damaged using a curved needle, resulting in cortical hemorrhages (89,90). This model has been very little used in the recent research as it has a large variability of brain injury created due to ischemic infarction, which limits the reliability of the experimental results (85). Xue and Del Bigio (90) compared the three models and observed that the relative magnitude of the inflammatory phenomena, molecular and cellular changes may differ between the models, although within similar patterns. These authors described damaged DNA neurons and CD8α immunoreactive lymphocytes to be maximal at 3 days after injury, but because the microglial/macrophage reaction peaked early between 3 to 7 days and persisted for weeks, dying neurons were seen in small quantities after 21 to 28 days. Neutrophils were substantially lower in this cortical vessel avulsion model compared with the blood and collagenases infusion models (90)

Alternative to this procedure is rupturing the vessels using a laser in order to produce microbleeds and assess coagulation outcomes (91). Zhou et al (86) induced an intracerebral hematoma by puncturing the middle cerebral artery. This procedure has been performed in 12 dogs under the ultrasound guidance with a high success rate. The main limitations of this model are that it can only be performed using an open bone window, which can underestimate the effects of intracranial hypertension and, as discussed above, it seems to produce a less severe histological damage.

Intracranial balloon model

Sinar et al (16) developed a model with micro balloon insertion in rats as a way to study the effects of a mass leading to IHT. A micro balloon 25 Fr mounted on a needle was inserted into the right caudate nucleus through a burr hole in the skull of the animal. The micro balloon was inflated to 0.05 ml over a period of 20 sec and maintained for 10 min after intervention was inflated and deflated. At the end of the study, the authors examined the histology of the brain, intracranial pressure (ICP) and cerebral blood flow. They found the micro balloon model to be successful in producing an effective brain injury, causing a reduction in cerebral blood flow and an ICP increase at the site of injury. The main advantage of the balloon model according to Alharbi et al (92) is that it imitates the mass effect of an acute cerebral hematoma in humans.

Andrade et al (93) also evaluated the variation on intracranial pressure and showed that this method reliably leads to IHT. The present study describes below the hands-on experience of the authors of this review, with previously unpublished information and images (Fig. 3, Fig. 4 and Fig. 5). It was submitted to and approved by the Institutional Animal Care and Use Committee of the School of Medicine at the University of São Paulo (USP) (approval no. 019/14; being developed according to the recommendations of the National Council for the Control of Animal Experimentation and the Ethics Committee on Animal Use.

Figure 3 Piglet surgery model with instrumentation implanted in the skull. Black star, ICP monitor with external ventricular drain; blue star, intracranial balloon; red star, transcranial Doppler probe; green star, intraparenchymal probe for ICP, brain oximetry and temperature monitoring. ICP, intracranial pressure.

Figure 4 Piglet brain after experiment with cortical injury produced by the intracranial balloon.

Figure 5 Piglet brain sagittal section after surgery. The balloon inflation produces injury by intracranial hypertension with local expansion, but not much adjacent injury. The arrow shows brain injury caused by the balloon inflation. The white stars mark the lateral ventricles of the animal.

Piglets were separated into 3 groups: First group with mild intracranial hypertension, a second group with severe intracranial hypertension and for the third group a cerebral rebleeding model. Prior to surgery, the animals fasted for 12 h but had free access to water. Intramuscular ketamine was co-administered at a dose of 15 mg/kg and xylazine at a dose of 2 mg/kg as a preanesthetic. Once intravenous (IV) access was obtained, anesthesia was induced with propofol at a dose of 5 mg/kg. The animals also received an initial IV volume of 20 ml/kg physiological saline (NaCl 0.9%) to compensate for volume loss due to fasting, and fluid support was continued throughout at a rate of 5 ml/kg/h. Anesthesia was maintained with IV propofol. In this protocol, ICP, direct brain oxygen pressure and results of transcranial Doppler exams were performed (Fig. 3). To induce IHT, a balloon 8Fr was used (Fig. 4) in piglets (average weight, 20 kg). The injury caused by balloon produces IHT with local expansion (Fig. 5).

Studies for ultrasound assessment of the optic nerve sheath diameter (34,94) and transcranial Doppler pulsatility index (35) have taken advantage of this method because of sudden IHT development. Thus, the balloon model appears to be sufficient for generating oscillations in intracranial pressure measurements as well as for causing an impact on cerebral hemodynamic conditions and valuable for studying the effects of mechanical damage to the brain tissue (43). Although it's an easily reproducible method, it has numerous limitations as follows: i) Does not reproduce blood toxicity in the brain parenchyma, this could be the reason for a lesser degree of ischemia in this model compared with the injection of an equivalent volume of blood; ii) does not promote blood brain barrier disruption: and iii) is not able to induce edema formation (16).

4. Experimental studies in small animals

Rodents are the pillar of modeling in ICH (16,72). They were first used as modeling for brain hematomas by Bullock et al in 1984(61), where ICH was induced by femoral artery cannulation, which provides blood at arterial pressures directly to the brain. Ever since, these animals have been used for different experimental models of brain hematoma, such as autologous blood injection, bacterial collagenase and cerebral balloons (70,93-96).

There are numerous advantages of using mice and rats as models for ICH, such as the excellent cost effectiveness (70), accurate paradigms for testing and outcomes (95) and an extensive sample of reagents for immunohistochemistry and molecular biology (96). Moreover, the availability of transgenic systems in the mouse allows for genetic studies on ICH and its mechanisms of lesion (96). The main disadvantage of rodents is the small size of the brain, which limits the clot volume that can be created and complicate its use in surgical studies (70,96). Furthermore, the lack of brain gyrus and the small amount of white matter limits the correlation with the human brain.

5. Experimental studies in developed brain gyri animals

Piglets have well developed white matter, a relatively low cost and no major difficulties with protective animal societies, being an excellent species for ICH studies (93). The possibility of using hematoma volumes 20 to 30 times higher in this model compared with rodent species also allows the test of hematoma removal (surgery simulation) in addition to providing the rebleeding simulation (93). Shi et al (97) described the balloon technique in the subcortical white matter, instead of blood injected into the gray matter or the basal ganglia. The use of this method turns obtaining a more uniform and reproducible hematoma volume possible, facilitating the extrapolation of information to humans, especially because of the greater volume of white matter found in larger animals and the developing edema adjacent to the hematoma.. This model is clinically relevant since the ICP information can be extrapolated from models to clinical treatment.

Wagner et al (49) developed a lobar hemorrhage model in pigs, where 1.7 ml of autologous arterial blood is slowly injected in the frontal white matter using an infusion pump. The slow injection reduces the likelihood of ventricular rupture or leakage of blood along the needle path. Compared with rapid infusions at high pressures, this method seems to reproduce more reliably deep hematomas in humans, which usually arise from bleeding of small parenchymal arteries (49). Küker et al (98) described a model injecting 0.5 to 2.0 ml of blood with a venous blood reservoir into the previous frontal lobe bruising to study the characteristics using magnetic resonance imaging. In another study (39), the authors modified this model into a double injection procedure (with a main injection of 2 to 3 ml of autologous venous blood reservoir) to better prevent post-injection reflux. The use of pigs is still a useful model of intracranial hypertension due to the presence of cortical gyri, which makes the model more similar to the human brain compared with murine models (99). Infusion of autologous blood in the piglet model has been used to investigate the ICP, cerebral blood flow, development of edema, brain metabolism, transcription factor and gene expression for inflammatory activation (96,99,100). It has also been used to study the possibility of surgical hematomas (101) and clot lysis induced by tissue plasminogen activator (102). It has also been used to test the effects of brain injury induced inhibition of heme oxygenase (14) and iron chelators deferoxamine (103). While the use of autologous blood allows the evaluation of the biochemical effects of blood on neurons, these models do not evaluate in a simple and direct way the effects of surgery for the removal of hematomas.

Piglets are also used in models of collagenase injection. Mun-Bryce et al (38) described a 10 ml collagenase pump infusion for 20 to 30 min in the right somatosensory cortex. This study examined the excitability of the tissue around the hematoma and evaluated the results of brain injury using magnetic resonance imaging and magnetoencephalography. The use of collagenase does not address the clinically relevant phenomenon of vasogenic edema associated with cerebral hematoma and so does not offer the ideal conditions for studies on intracranial pressure (102). However, collagenase levels in the pig model are much higher compared with those found in human brain injury, making it difficult to derive any correlation to clinical information (103).

Based on classical studies with a balloon, Shi et al (97) published an experimental model using a pediatric urinary catheter inflated with saline to achieve anatomical space and then infusing autologous blood. However, the analyses of cerebral hemodynamics were not carried out and its variables only evaluated the clinical condition of the animals days after the surgical procedure that simulated a deep intracerebral hematoma. Another study in pigs (104) was intended to simulate a diffuse intracerebral injury with the method of acceleration and rapid deceleration compared with a control group. In this study ICP values were analyzed; however, it is a model that cannot reverse brain damage by surgical procedure and also does not apply to intracerebral hematoma models.

6. Controlling and sham

A few studies compared the index population with appropriate controls (56,57,94,104). Some studies report the control group to be composed by intact animals, the comparison of different infusions as autologous blood or collagenase, and even the reproduction of ICH with the exclusive insertion of an intracranial canula without any intervention (77,88,93,103). The present review observed that using a sham group is not a common practice in this regard but used more often for the assessment of immune-inflammatory factors and cerebral blood flow, in the autologous blood infusion and balloon inflation models (16). Hua et al (56), for example, aimed to assess edema reduction and the inhibition of complement factors induced by autologous blood injection in rats by comparing groups with and without the addition of N-acetyl heparin after ICH. Xi et al (57) infused the same amount of saline in contrast to different preparations of red blood cells in order to assess brain swelling and the brain blood barrier permeability, also in rats. With distinction from the models described, Azevedo et al (94) assessed ultrasound of the optic nerve sheath (ONSD) in piglets to observe these animals ONSD standard values and its variation according to different anesthetics infusion. These values also served as references to non-invasive screening for ICH simulation; whereas Friess et al (104) evaluated neurological multimodal monitoring (ICP, oximetry and microdialysis) in randomized pigs with no-impact rotational traumatic brain injury and no injured instrumented (the same techniques) sham.

7. Limitations

The difficulties in transferring information from experimental models to clinical settings derive from specific aspects of each species, as well as limitations of the models and methods. For the most part, experiments use younger animals with greater functional reserve and hemodynamics. Moreover, rodents disclose an outstanding capacity of regeneration and rehabilitation that is not comparable with humans (11). The complex pathophysiology of brain hematomas involving vascular injury as well as apoptosis and molecular aspects involved hinder the translation of information. The majority of studies are conducted in murine models and there are important differences in cerebral hemodynamic changes on the acute phase of intracranial hypertension for these animals compared to humans. Regarding the studies, a wide methodologic heterogeneity for injury inducing was found, likewise, appropriate controlled study designs are lacking since a few studies used suitable controls to evaluate their endpoints.

8. Conclusions

Animal models have the potential to enhance our understanding of the pathophysiology and treatment of intracranial hypertension and brain hematomas, being essential to develop and evaluate new therapeutic strategies in preclinical settings. To decide which is the best model for each research, some points must be considered: The purpose of each model, the primary outcome of the study, considering whether hematoma is in the acute or chronic phase and molecular vs. hemodynamic analysis.

Acknowledgements

Not applicable.

Funding

Funding: No funding was received.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors' contributions

WSP, SB and IN conceptualized and planned the execution of the study, collected references and prepared the manuscript. GCP, EZ, DAG, AFDA and MJT collected and compiled data and prepared the manuscript. SB, CM and RD performed manuscript review and online search. WSP, AFDA and SB confirm the authenticity of all the raw data. All authors have read and approved the final manuscript.

Ethics approval and consent to participate

This study was approved by the Institutional Animal Care and Use Committee of the School of Medicine at the University of São Paulo (USP) (approval no. 019/14; being developed according to the recommendations of the National Council for the Control of Animal Experimentation and the Ethics Committee on Animal Use.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Authors' information

Dr Sérgio Brasil, ORCID 0000-0003-2397-9947.

References