Nonhuman Primate Models of Ischemic Stroke and Neurological Evaluation After Stroke

Nonhuman primates are closer to human beings than rodents in genetics, neuroanatomy, physiology and immunology. Nonhuman primates are therefore considered an ideal preclinical model to replicate various aspects of human stroke. Ischemia stroke models in nonhuman primates can better fit the physiological symptoms and changes in humans after cerebral ischemia. Currently, various construction methods and neurological evaluation methods have been developed and applied to stroke models of nonhuman primates, including craniectomy models, endovascular stroke models, autologous thrombus models and intraluminal filament models. Meanwhile, new innovative methods have emerged, such as the endothelin-1 model and photothrombosis model. In the past thirty years, these model studies have explored various mechanisms that are initiated in the first minutes, hours, and days after a stroke. Permanent and temporary middle cerebral artery occlusion models have been trying to simulate the complex situation of human stroke. However, a comprehensive comparison of the above methods, including their advantages and disadvantages, difficulty and application fields, is limited. Here, we introduce various modeling methods that are currently available for nonhuman primate stroke models, compare the differences between these different preparation methods, and analyze the advantages and disadvantages of the various methods and the fields of application. The imaging detection methods of nonhuman primates after cerebral ischemia and the neurological evaluation methods after stroke are also discussed briefly. Methods are sorted and compared so that scholars can choose appropriate modeling methods and evaluation methods to establish nonhuman primate stroke models.


Introduction
Every year worldwide, 15 million people suffer from stroke, resulting in 5 million deaths, and another 5 million people are permanently disabled. Stroke is the main cause of long-term disability in the United States, and more than half of stroke survivors over 65 years old have limited mobility 1 . Ischemic stroke accounts for more than half of the current stroke cases and has become the main cause of death and disability in patients without effective treatment 2 . Stroke is caused by thromboembolic occlusion of the cerebral aorta or its branches. Vascular occlusion leads to hypoxia and loss of energy, followed by the formation of reactive oxygen species, the release of glutamate, the accumulation of intracellular calcium, and the induction of inflammatory processes 3 . Eventually, irreversible damage to the brain tissue occurs in this area. The main treatment method is the use of neuroprotective drugs and thrombolysis drugs. However, currently approved thrombolysis drugs that can be used for acute ischemic stroke are very limited. The only one permitted by the Food and Drug Administration (FDA) and used in stroke is recombinant tissue plasminogen activator (rt-PA). If the application time exceeds 4.5 hours, it is basically invalid and will increase the risk of hemorrhage 4 . This has resulted in 90% of patients being unable to be treated with thrombolytic drugs in time 5 . Therefore, there is an urgent need to develop new thrombolytic drugs or extend the effective time of existing drugs.
Developing new thrombolytic drugs requires the use of appropriate animal thrombosis models. Rodents (rats, mice) are widely used in medical research for studying cerebral ischemia, by the beneficial of its low cost, abundant sources, and simplicity of operation 6,7 . Many treatment strategies have been proved working after stroke in rodents, but have not been successful in clinical translation. The main reason for the failure is large differences between the brains of rodents and humans, causing rodents to respond differently to the same ischemic damage. Large animals (such as pigs, canines, nonhuman primates) are similar to humans in genetic background, behavioral characteristics, risk factors for cerebrovascular disease, anatomy of brain tissue and blood vessels 8 . Therefore, large animals more closely resemble the pathophysiology and clinical manifestations of stroke in humans. Generally, different volume of white matter and gray matter triggers different sensitivities to cerebral ischemia, and larger brain tissue is suitable for complex surgical operations and facilitates postoperative imaging evaluation. At the same time, large animals, especially nonhuman primates (NHPs), have corresponding neurological dysfunction after stroke, which is very suitable for the observation of stroke treatment and rehabilitation process. However, the research of large animals also has certain limitations, such as high cost and certain ethical issues, and requires relatively sophisticated testing equipment. Therefore, it is very difficult to select the most suitable animal to study stroke among many experimental animals. Nonhuman primates (macaque, baboons, etc.) are closer to humans than rodents in physiological structure and blood vessel supply of the brain. At present, due to the limits of economic and ethical factors, the main application areas of nonhuman primate models are limited. However, researchers using NHP models have got the important achievements of the pathogenesis and the recovery of the treatment, especially about major diseases that currently affect human health, such as human viral infections, respiratory disease, diabetes, stroke and neuroprotection, etc [9][10][11][12] . Cerebral ischemia stroke models in NHPs can better fit the physiological symptoms and changes in humans after cerebral ischemia, meaning this option has become the best choice for studying human stroke disease. The distribution area of the middle cerebral artery (MCA) is also similar to that of humans, and the size of the brain is quite suitable for magnetic resonance imaging (MRI) research 13 ; therefore, numerous scientists have used nonhuman primates to simulate human brain disease 14,15 .
In neuroscience, NHPs are often used to prepare ischemic stroke models (mostly the macaque and baboon) because they have an intelligence quotient. NHPs can use their arms to handle various tools to complete motor tasks. Because of their degree of compliance and irritability after training, the evaluation of neurological function in an ischemic stroke model can be closer to that in real clinical patients 16 . Therefore, more NHPs are being used to study various poststroke treatment methods, such as acute and chronic drug treatments and stem cell treatments after cerebral ischemia 17,18 . After establishing the model, researchers then use neurological behavioral tests and imaging tests to evaluate the treatment effects 19 . The middle cerebral artery is a frequent site of cerebral ischemia in humans, and many cerebral ischemia models have been developed for MCA. At present, the method of establishing NHP models of local cerebral ischemia can be subdivided into the transorbital approach method, transcranial approach method, balloon compression model, snare ligature model, etc. Although different methods can lead to stroke, the applicable research field is different.
This article briefly introduces the modeling methods of several NHP models of focal cerebral ischemia and briefly describes their advantages and disadvantages. Simultaneously, based on the model preparation results, this article sorted the evaluation method of imaging and neurological function in NHPs after cerebral ischemia.

Transorbital approach method (Clamp model)
The earliest established NHP stroke model was the operation through the orbital approach. In 1970, Hudgins first established this model in squirrel monkeys, which removed the eyeball, periorbital tissue, and optic nerve during the operation 20 . After, the skull on the posterior orbital wall was removed to expose the anterior part of the circle of Wills, the squirrel monkey's middle cerebral artery was identified, and an arterial clip was used to block the proximal M1 segment of the MCA. Then, in 1975, Symon used the orbital approach method in baboon but blocked the proximal M1 and distal M1 of the MCA to compare the difference between these two segments 21 . To date, this method has been applied to large NHPs through the orbital approach. In 1992, Liu and Symon again used this method to block the bilateral A1 in baboon, which aimed to test the blood of the territory of the anterior cerebral artery 22 . Before 1997, researchers only used a transorbital approach to permanently clamp the cerebral artery. Research on ischemia-reperfusion in the brains of NHPs is still lacking. Young anesthetized baboons were subjected to 6 hours of either reversible or permanent middle cerebral artery occlusion. The results confirmed that early reestablishment of cerebral blood flow after focal ischemic insult is not detrimental but indeed is beneficial regarding the final infarct volume 23 . Then, in 2008, Murphy chose to clamp the M1 segment of the MCA, and after 90 minutes, the methods were not changed, thereby forming an ischemia-reperfusion model in male and female rhesus macaques 24 . Magnetic resonance angiography (MRA) technology was then used to identify cerebral infarction differences between sexes. At present, the transorbital approach method is used both to study permanent occlusion of blood flow and ischemia-reperfusion in the NHP brain. The ischemia time can be adjusted according to the needs of the research and remains a popular method of modeling a stroke in NHPs. The transorbital approach is better than other techniques requiring a craniotomy for exposure of the middle cerebral artery, which leaves a significant residual dural defect. The transorbital approach requires less operation time with decreased attendant morbidity and mortality. In addition, vascular structures are not disrupted by this method, in contrast to the other craniotomy approach 25 .

Transcranial approach method (electrocoagulation model)
This method emerged because the blood vessels in the deeper parts of the brain cannot be reached through the transorbital approach; for example, in the M2 segment of the middle cerebral artery and the transorbital approach, the available operation space is too small, it easily damages the surrounding tissue, and the failure rate is high. This method uses a lateral transcranial approach to open the skull and incise into the corresponding area of tissue to expose the artery and operate on it. After entry from the pterion (left side or right side), the subcutaneous tissue is cut, and the flap is turned to the sphenoid pterion. After cutting the meninges, the brain is gently retracted to expose the lateral fissure. After carefully dissecting the subarachnoid space and trabecula of the related brain cistern under an operating microscope, the target artery is gradually exposed without affecting the brain tissue. Then, operations such as electrocoagulation of the artery (clamp can also be used to block the artery so that ischemia-reperfusion can be performed). In 1996, Marshall first used electrocoagulation through the lateral transcranial approach to permanently occlude the M2 segment of the middle cerebral artery in marmosets 26 , which showed that all subjects showed unilateral dyskinesia of the contralateral forelimb and neglected contralateral tactile stimuli. Then, in 2003, Jonathan improved their original method by using a monkey stereotactic frame to fix the marmoset brain. The position of electrocoagulation is adjusted from the M2 segment to the M1 segment 27 . Because the occlusion is located in front of the lenticular artery, it blocks blood flow to the entire MCA area, including the subcortical structure, caudate nucleus, and putamen. Subsequently, scholars used this method as the basis for their research, using electrocoagulation or clamp blockade of blood flow to establish vascular occlusion through a transcranial approach [28][29][30] . In the same postoperative cynomolgus monkeys, a distal occlusion of the M1 branch of the middle cerebral artery occurred, and an infarction of the same size was observed, mainly involving the cerebral cortex and subcortical white matter distributed along the middle cerebral artery 31 . Monkeys with partial middle cerebral artery occlusion and stable infarct areas have been observed mainly in the basal ganglia and subcortical areas. A transcranial approach to occlude the artery greatly broadens the operating range and enables the establishment of various ischemia models. This model is suitable for studying secondary damage in remote regions or drugs that have neuroprotective effects after stroke. However, the mechanism of blocking blood vessels by electrocoagulation or clamp is significantly different from the pathogenesis of human cerebrovascular embolism, and blocked blood vessels cannot be used for drug thrombolysis or mechanical recanalization experiments after vascular embolization. Additionally, this method is demanding and can easily result in bleeding or other injuries during the operation.

Balloon compression model of MCAO
The balloon compression model, as the name suggests, uses a special airbag to wrap the separated middle cerebral artery. After exposure of the targeted cerebral artery, the airbag is inflated to pressurize the blood vessel and physically and mechanically block blood flow. In 1980, Spetzler first used this method in a baboon; the middle cerebral artery was first separated by the transorbital approach, and an inflatable balloon around the middle cerebral artery was then wrapped 32 . The inflated balloon causes vascular occlusion by mechanically compressing the blood vessel. After 1-30 minutes (different times according to the ischemia reperfusion requirement) of temporary occlusion, the balloon was removed for subsequent testing. The balloon blocking method depends solely on external physical and mechanical forces, which differ from middle cerebral artery occlusion (MCAO) in humans, and its physiological and pathological mechanisms are also inconsistent. Meanwhile, the airbag has a serious impact on the surrounding tissues and may cause mechanical damage to blood J o u r n a l P r e -p r o o f vessels. All of these disadvantages limit its use. Moreover, the effect of the balloon can be achieved using a micro arterial clip, so the method has gradually been replaced.

Snare ligature model of MCAO
In 1981, Crowell invented this method to temporarily block the middle cerebral artery in macaques using a specific snare ligature 33 . The method is operated at the base of the brain after the transorbital approach, requires separation of the middle cerebral artery, connecting a silk thread and corresponding catheter, and then exposing the silk to the wound and suturing the skin. The eyeball also needs to be removed. This model has several advantages. Most importantly, it avoids the huge trauma caused by transcranial approach craniotomy, minimized the trauma by using a special ligature implanted in the orbit. There is less damage than mechanical or balloon compression equipment. Another advantage is that it permits reperfusion in awake monkeys, which releases occlusion by refining the snare ligature design. However, the disadvantages are also obvious, as it is expensive and requires extremely high surgical operation skill. These shortcomings limit its application in NHP cerebral ischemia models.

Endovascular stroke model of MCAO
Endovascular stroke models usually use interventional methods, and various emboli are applied to induce cerebral artery embolism models, such as using coils or wires to block blood vessels or utilizing an inflating balloon to block blood vessels endogenously. In 1974, Molinari first used the endovascular stroke model in macaque mulatto, in which the embolus is not a coil or wire but a small amount of silicone rubber compound and the size was fixed (1.6 mm) 34 . However, in that era, digital subtraction angiography (DSA) was developed slowly, and DSA was not used to confirm the location of the embolus, but a palpable pressure drops signaled clearance of the embolus from the cannula into the internal carotid artery. Finally, a monkey stroke model was successfully formed but where the embolus entered could not be accurately determined, and there may be a risk of blood vessel rupture caused by over injection of the embolus. Subsequently, the embolus was gradually replaced with a detachable coil and a retractable airbag. Subsequent development of DSA helped determine the exact location of the embolism in this intervention.
At present, the most classical and interventional method is to insert a catheter percutaneously through the femoral artery under angiography or enter the internal carotid artery. After reaching the internal carotid artery, the guidewire in the catheter is withdrawn. The micro guidewire is used to introduce the tip of the catheter into the M1 segment of the middle cerebral artery, and a contrast agent is used to confirm the distance of the catheter tip to the MCA. A detachable coil is inserted for temporary occlusion. After occlusion of the MCA with imaging methods, the embolus is then removed or stays in the blood vessels permanently according to the closure time required by the surgeon. DSA is employed to confirm the recanalization of the blood vessel 35 . Another embolus is an endogenous balloon, which utilizes a special airbag after entering the intracranial artery of the macaque through the catheter that can temporarily inflate to block the blood vessel in the artery. Then, the airbag is deflated and removed when the ischemic time is sufficient (it differs from balloon wrapping the artery; this method, the airbag is located in the artery) 36 . Similarly, some researchers separate the internal carotid artery and enter the MCA through a catheter directly 37 . It is worth mentioned that Tong using the trapping technique to deliver the 4-0 silk suture instead the coils to occlusion the MCA-M3 and MCA-M1 segments of rhesus monkeys, which effectively eliminates retrograde collateral flow in the MCA branch and produced consistent infarct results 38 .
In brief, the most common embolus to induce an MCAO model in NHPs with the use of balloons, guidewires, coils and silk to block the MCA has advantages and disadvantages. The balloon is easy to control and completely blocks the lumen of the blood vessel but is difficult to manipulate and is not suitable for bending in the cerebrovascular system, making it difficult to reach blood vessels with a smaller diameter, and it is not suitable for the study of hypothermic treatment after ischemiareperfusion (the balloon is blind and cannot be filled with refrigerant). There is a risk of rupture of blood vessels. The wire has a constant diameter, making it difficult to fit various vessel diameters; therefore, the same position of the guidewire may lead to different embolization locations in blood vessels. The coil is relatively soft, easy to operate, and can be accurately positioned. Apart from the cost factor, the wire has certain advantages compared to other occlusion methods. Silk is easy to obtain and costs less than coils. Because of its smaller diameter, it allows to accessed to finer sections of the middle cerebral artery, such as the M2 and M3 sections. Of course, because it is too soft, it needs more research to confirm whether occlusion is stable if only to use silk occludes the initial segment (M1) of middle cerebral artery. Although the endovascular stroke model of MCAO has certain shortcomings, compared with the orbital approach and craniotomy, this method does not require craniotomy and can be operated on by nonsurgical personnel, so it has been widely used. Justine used coil on cynomolgus macaques to simulate ischemia-reperfusion injury in the middle cerebral artery and developed the nonhuman primates neurological scale:(NHPNS) for evaluating monkeys after stroke. He evaluated the macaque's stroke per-occlusion and post-recanalization impairment on PET-MRI, in addition to acute and chronic neuro-functional assessment 19 . Kyung used the endovascular stroke model to demonstrate that ischemic brain tissue could be salvaged after recanalization of the occluded artery (within 3h), which with diffusion-perfusion matching 39 . Zhang used silk to occlude rhesus monkeys MCA-M2 and examined diffusion weighted imaging (DWI) and T2-MRI at 1-6 hours and 48 and 96 hours after post-occlusion, confirming that the diffusion-perfusion mismatch was visible generally at 6 hours but nearly diminished at 48 hours post occlusion 40 . In short, endovascular stroke model as a widely used method is suitable for research ischemia-reperfusion after stroke.

Intraluminal filament model of MCAO
This method was originally only used in rat MCAO models. In 1989, Enrique established this method in the rat middle cerebral artery 41 . The most important procedure is introducing a 4-0 nylon intraluminal suture into the internal carotid artery (ICA) and advancing it to the MCA to block blood flow. Now, scholars apply this method to monkeys and imitate the same steps. In 2008, Thomas first used an intraluminal filament model in marmosets, and the common carotid artery of monkeys was separated through a median neck incision 42 . After blocking the blood flow by ligating the distal end of the external carotid artery, the special nylon embolus was slowly inserted into the blood vessel from the proximal end of the external carotid artery. The blocking site in the middle cerebral artery was confirmed using a doppler blood flow instrument. After blockage with a specific embolization time, the nylon embolus is then withdrawn. Ebeline compared the difference between permanent and transient chronic ischemic stroke in marmosets based on this method 43 . The results showed that neurogenesis was increased in both permanent and transient animals. After a marmoset stroke, permanent MCAO causes more severe sensorimotor impairment than temporary MCAO in the long-term. The advantage of the intraluminal filament model of MCAO is that it can strictly control ischemic time, and the embolus can be easily removed to obtain reperfusion with no transcranial operation. This method has lower requirements for operators and is easy to learn. By adopting this method, the animals have less traumatic injury and no obvious adverse reactions. The surgical procedure was achieved directly without postoperative mortality or cerebral hemorrhage. This method can ideally simulate various physiological reactions in cerebral ischemia patients after J o u r n a l P r e -p r o o f embolization and thrombolysis and has been widely used in stroke research on NHPs. The only disadvantage is that it damages the external carotid artery and affects the blood supply parts of the external carotid artery, such as the face, which may cause facial muscle paralysis after surgery.

Autologous thrombus model of MCAO
The autologous thrombus method requires a certain catheter for delivery. Venous blood from the animal must be collected before the operation and prepared in vitro into white thrombi of the required length using thrombin. In 2001, Kito first used autologous thrombus to model a cynomolgus monkey stroke 15 . Thrombus occurs through the internal carotid artery injected into a specific location using a catheter and saline, leading to the thromboembolism model. Notably, autologous thrombi can be injected into the brain via the femoral vein catheter or directly through the internal carotid artery to localize to the MCA in rhesus monkeys and cynomolgus monkeys 36,44 . The difference between these two approaches is that the former must be navigated by DSA; because the arteries that pass from the femoral artery to the internal carotid artery have a complicated shape, it would be impossible to reach without DSA. The thrombus is delivered after guiding the catheter to the aim artery. The latter does not require the help of DSA and only needs to enter a fixed length to reach the middle cerebral artery. In 2006, Kito used this autologous thrombus model of MCAO cynomolgus monkeys again to test whether urokinase dissolves blood clots in stroke and improves neurological function after stroke 45 . There are numerous and meaningful studies using this method. Wu used autologous thrombus to establish a rhesus monkey stroke model, and compared the differences between the stroke models established by the endovascular stroke model (balloon method) and autologous thrombus model 46 . It confirms the advantages of the thrombus model include clinically translatable pathogenesis, relatively small infarct size, better survival and the possibility of thrombolytic therapy. The disadvantage of thrombus injection method is technical complexity and poor control over the exact location of occlusion. Based on autologous thrombus model, Song found that remote ischemic conditioning results in significant modulations of the plasma proteome in rhesus monkeys 47 . Infarct formation with autologous thrombus model is the closest to clinical cases. Therefore, it has practical value in preclinical applications for the research, diagnosis, and treatment of the occurrence and development mechanism of stroke, especially thrombolytic therapy. In addition, the autologous thrombus model can be used in ischemia-reperfusion(I/R) studies, and the clots can also be indwelled to form a permanent vascular occlusion model if the blood clot ends up in the blood vessel without thrombolysis or withdraw it.

Photothrombosis model of MCAO
At present, in addition to conventional physical methods to induce the middle cerebral artery, there are also methods utilizing a photothrombosis model of MCAO in NHPs. Photothrombosis is a newly discovered method in the last twenty years that can use a specific light source to illuminate the dye to form a local thrombus. Rose bengal, a photosensitive dye, was infused intravenously via the femoral vein and then circulated throughout the body. When irradiated by a cold light source, the dye is activated and causes endothelial damage, platelet activation and thrombosis, leading to local blood flow interruption. A photothrombosis model was first used to establish stroke models in rats because the method is simple and does not require complicated operations such as embolization or clipping. In 2005, Masashi first used this method in cynomolgus monkeys 48 . The photo illuminate dye on the surface of the middle cerebral artery for 20 minutes resulted in cyclical flow reduction (CFR), which produced a significantly larger infarction and more severe neurological deficits compared with clamp J o u r n a l P r e -p r o o f model, even though the decline in blood flow was smaller. Later, Tomizawa confirmed that the thrombus formed by a photothrombosis model of MCAO can be dissolved by drugs and improved thrombus formation and hemodynamic changes 49 . These applications showed that the focal cerebral ischemia model made by rose-bengal dye irradiated with cold light produced changes in nerve function, such as consciousness disorder and limb movement disorder. In addition, cerebral cortical edema, telangiectasia, nerve cell degeneration and necrosis appeared in the irradiated area, which were similar to clinical changes. The primate model prepared by this photothrombosis method is simple and fast. The method provides important tools for neuroprotective research and exhibits important clinical significance. Yang established the photothrombosis model in rhesus monkeys and treated with CircSCMH1-extracellular vesicles which found that it improved neurological deficits after stroke 50 . However, this method is not conducive to the establishment of collateral circulation and ischemia-reperfusion injury, nor is it convenient for observing changes in the ischemic penumbra or the prevention and treatment effects of related drugs. Its ischemia is not permanent but intermittent and is accompanied by a certain degree of reperfusion. The consistency of the thrombus was poor because of the flow of dye.

Endothelin-1 injection model of MCAO
Endothelin-1 (ET-1) is a 21-amino acid peptide with potent and long-acting vasoconstriction properties. It was first identified as an endothelial contractile factor by Yanagisawa in 1988 51 . Researchers have found that ET-1 overrides cerebral autoregulatory mechanisms to constrict a number of cerebral vessels in vivo and reduce cerebral blood flow (CBF) below the ischemic threshold to induce infarction. David first used ET-1 injected in the M2 segment of the MCA in marmosets, which produces a robust M2 territory infarction that is correlated with contralateral motor and sensory/neglect impairments 52 . Other studies have also confirmed that endothelin-1 can cause local occlusion of the cerebral arteries in marmoset 53 . These studies showed the same result: local injection of endothelin-1 can form a strong contraction of this segment of blood vessels, thereby forming an ischemic state within a certain period of time. When the effect of endothelin fades, blood flow is restored, resulting in ischemia-reperfusion injury. Intracerebral injection of ET-1 into the M1 segment of the middle cerebral artery successfully induced motor cortex infarction in rhesus monkeys 54 . Through MRI, they precisely observed that infarct size reached its peak at 7 days post-operation and then gradually decreased. The endothelin-1-induced cerebral infarction model has good stability and a consistent infarct focus and thus has good application prospects. However, relatively high surgical requirements and precise blood vessel positioning are accompanied by risks of blood vessel rupture and bleeding during injection positioning. Meanwhile, the mechanism by which ET-1 causes cerebral infarction is continuous vasoconstriction and reduced blood flow, but thrombosis is uncertain and not suitable for thrombolytic research. Another problem is that it is difficult to control the consistency of vasoconstriction time. These shortcomings limited its further application.

Ferric chloride-induced model of MCAO
Increasing knowledge of the mechanism of thrombosis in mice has led to the wide use of ferric chloride to induce blood vessel formation thromboembolism. Researchers have found that 10% ferric chloride can induce thrombosis in the mouse middle cerebral artery and that it can be dissolved by tissue-type plasminogen activator (t-PA) 55 . Inducing thrombi in the mouse carotid artery with ferric chloride solution is also feasible 56 . Our recent study used ferric chloride solution to induce stable thrombus formation in rat carotid arteries and demonstrated that the thrombus could be dissolved by t-PA 57 . Some scholars have tried to use ferric chloride to build a larger artery thrombosis model, such J o u r n a l P r e -p r o o f as the canine carotid artery 58 . Currently, the main mechanism of ferric chloride-induced thrombosis is to cause local vascular endothelial cells to detach and expose the basement membrane; simultaneously, a large amount of iron ions accumulates on endothelial cells to promote the adhesion of platelets and tissue factors to form aggregates. This triggers the thrombin reaction and leads to thrombus formation 59 . Therefore, the thrombus induced by ferric chloride can be dissolved by thrombolytic drugs, which has good application value in thrombolytic drug detection. The size of the thrombus can be adjusted by controlling the filter paper soaked in ferric chloride. The local thrombus formed by ferric chloride is dense and uniform, similar to the thrombus that can form in human blood vessels, with no obvious side effects. Therefore, this method has good application prospects. This method is easy to operate and does not easily cause additional damage; thus, it may appear in the construction of NHP artery thrombosis models in the future.

Comparison of the above stroke models of nonhuman primates
The multiple different methods mentioned above can produce NHP middle cerebral artery cerebral ischemia models. We have drawn a brief schematic diagram of the main methods ( Figure 1). Each of them has certain advantages and disadvantages. According to the different application directions and operation and implementation capabilities of scholars, all plans have certain application value. We compared the fields in which various monkey brain ischemia model induction methods have been applied in their respective studies in recent years (Table 1). For most research methods, ischemia reperfusion, destruction of brain tissue, imaging changes and neurological deficits are involved, excluding recanalization of blood vessels and thrombi. From the current application results of multiple models, the autologous thrombus injection model and the photothrombosis model could be used in the study of thrombosis. New methods, such as ferric chloride-induced thrombus, are worth considering and have good application prospects.

Permanent MCAO or Transient MCAO
Some of the methods introduced above can be used to build both permanent MCAO models and transient MCAO models. Permanent stroke models usually require physical methods to permanently block major arteries in the brain or leave the emboli that cannot be dissolved by the animal themselves 60 . For example, in the clamp model and electrocoagulation model, when clip is permanently placed to block the blood vessel or the blood vessel is blocked by electrocoagulation, the experimental subjects cannot restore the blood flow through their own strength. If the remaining emboli and clot are stable enough, permanent infarction can also be caused without withdrawing the emboli. Although the mechanisms of vascular occlusion caused by these methods are different, the final outcome will lead to permanent stroke in NHPs. It is suitable for simulating exceed the thrombolysis time window (most patients) or patients with failed thrombectomy, for studying the J o u r n a l P r e -p r o o f treatment and rehabilitation process after permanent stroke. As the transient MCAO models, the most popular used methods are endovascular stroke model (balloon), intraluminal filament model and clamp model 61 . Because these three methods can achieve accurate ischemia time and reperfusion time, so as to simulate the ischemia-reperfusion situation of patients after complete thrombolysis at any point time. When the occlusion time reaches the requirement of ischemia, the balloon, suture and clamp are removed to restore blood flow and simulate the effect of timely thrombolysis after stroke. Subsequent studies related to ischemia-reperfusion, such as reducing ischemia-reperfusion injury, and rescue brain tissue which in the ischemic penumbra (IP) area.

Imaging evaluation of postoperative blood flow and infarct focus
Imaging evaluation is also important for assessing NHP models. Imaging evaluation methods, including DSA, MRI, computed tomography (CT), positron emission tomography (PET) and transcranial doppler blood flow (TCD) detection, are required to confirm the success of constructed cerebral ischemia models. These methods show changes in vascular occlusion and cerebral ischemia. MRI is the most common imaging method for evaluating the success of the NHP ischemia model. This method can visually show the size of ischemic infarction and the volume involved. Liu used MRI to detect transient and permanent cerebral infarctions after middle artery occlusion in rhesus monkeys 62 . Similarly, diffusion tensor imaging(DTI) is used to visualize the conductive fibers in the neurofunctional region, and the intracranial blood vessels can be reviewed by MRA, which has many advantages over other examination methods 63 . Meanwhile, MRI can test more comprehensive brain function by performing other sequences. Alexander West established the rhesus monkey cerebral ischemia model through the orbital approach and successfully detected cerebral blood flow (CBF), apparent diffusion coefficient (ADC), and DWI items by MRI 64 . In the acute ischemic stroke (AIS), the infarcts shown by MRI does not represent the final infarct volume. And sometimes T1-MRI, T2-MRI and FLAIR-MRI are fails to show the infarct areas (within 4.5 hours) 65 . DWI and perfusion imaging (PWI) can reflect ischemic changes in brain tissue in the early hours after stroke. Therefore, after establishing a non-human primate stroke model, it is more appropriate to use DWI and PWI to detect infarct lesions in the acute stage, especially within a few hours after a stroke. Fang detected a clear infarct area by DWI at 3 hours after stroke in rhesus monkeys, and found that administration of Tat-CIRP could reduce the brain infarct volume and preserve neurological function 66 . Wu and his colleagues demonstrated that when an autologous blood clot was used to block the middle cerebral artery in rhesus monkeys, administration of t-PA at 2.5 hours after stroke presented different thrombolytic outcomes according to the result of DSA; complete thrombolysis, partial thrombolysis, unable to dissolve thrombolysis 67 . Meanwhile, the results demonstrate that t-PA can be effectively combined with selective intra-arterial cooling (SI-AC) to improve both structural and functional recovery after NHPs stroke 68 . This result is clearly presented by DWI detection in the acute stage at 4 hours after stroke. However, sometimes MRI results may exist DWI/PWI mismatch 69 . Current studies believe that there is an obvious ischemic penumbra area after ischemic stroke. Although these areas are affected by decreased blood flow, the portion of the ischemic territory that is still potentially salvageable, if an appropriate treatment is given 70 . Some researchers think mismatch between the larger hypoperfused area(PWI areas) and the smaller ischemic lesion(DWI areas) on MRI shows the ischemic penumbra 71 , timely post-stroke treatment can reduce the damage in this area which is necessary. Cook used the clamp model to establish a macaque stroke model and confirmed the mismatch between the DWI and PWI after surgery 3 hours. While using Tat-NR2B9c (a neuroprotective agent) to treat macaque after stroke, the confirmed ischemic penumbra which is J o u r n a l P r e -p r o o f DWI/PWI mismatched can be rescued 30 . In the chronic stage after stroke, both FALIR-MRI and T2-MRI can show the corresponding infarct area 68,72 . Some researchers have found that functional magnetic resonance imaging (fMRI) signal is importance for prognosis in the subacute disease period of stroke patients, which also could be used for assessment of rehabilitation 73,74 . Although the application of fMRI to non-human primates after stroke has not yet been found. However, some studies have begun to focus on fMRI signal changes in the awake, behaving NHPs, especially when video and image stimuli are given, it can be combined with the detection of fMRI to assess the changes in neural activity 75,76 . We believe that in the future, there will be new studies using fMRI to detect the changes of brain state in non-human primate before and after stroke. And using fMRI to detect neural repair in damaged areas during use of effective rehabilitation therapy after stroke is possible. Therefore, after a NHP stroke model is established, it is necessary to select an appropriate sequence when applying MRI to detect infarcts for acute ischemic stroke and chronic ischemic stroke. The shortcomings of MRI are also obvious: it is expensive, and the animal must be anesthetized throughout the process, anesthesia accident may occur due to prolonged examination. DSA can clearly show the interruption in blood flow after cerebrovascular thrombosis or vascular occlusion but not the overall effect of postoperative cerebral ischemic infarction 77 . The method is suitable for judging whether the blood vessel is occluded during modeling. The disadvantage is that a professional technician is required to perform the operation, and there may be a risk of vascular rupture. PET can quantify regional cerebral blood flow, regional fractional oxygen extraction, and regional oxygen metabolic rate. With these variables, irreversible tissue damage and perfusion after stroke can be clearly defined, which show the potentially salvageable areas 78 . However, the complex logistics limit it applications, and at the same time, it is too expensive as an experimental detection tool, few studies currently use PET detection alone for study of non-human primates after stroke. TCD can visually display the changes in blood flow during the operation to determine whether the embolization is complete. It is the most intuitive method of detection during surgery 79 . However, the equipment is expensive and relatively complicated to operate. For the sake of a smaller brain size in monkeys, CT scans might not be clear enough; moreover, MRI can recover the function of CT imaging to a certain extent, so there are few studies using CT alone to evaluate cerebral infarction after brain surgery in NHPs. Collectively, there are relatively few imaging methods for evaluating vascular occlusion and cerebral infarction in NHPs. Combined application of MRI, DSA and TCD imaging, as well as DWI, DTI, and MRA, are increasingly used to show changes in cerebral ischemic infarction and vascular obstruction 80 .

Neurological Evaluation after Stroke of nonhuman primates
NHPs stroke, except for imaging examination to determine the formation of infarcts and vascular embolism, also needs behavioral evaluations to prove the corresponding neurological damage. However, the current neurobehavioral testing items for NHPs are not clearly unified, and the testing standards are also quite different. Here, we present several evaluation methods that are currently used in many applications.

Nonhuman primate stroke scale
The NHP stroke scale (NHPSS) score (Table 2) is a comprehensive assessment method with 11 items of NHPs. The assessment includes consciousness, defense response, grip reflex, limb movement, gait, rotation, slow movement, balance, neglect, vision cutting/hemianopia, and facial weakness. This test is included in the stroke scoring system of the National Institutes of Health. Out of a total of 41 points, 0 points corresponded to normal behavior, while 41 points corresponded to severe neurological impairment 30 .
NHPSS declares that NHPs can clearly reflect the contralateral side, limbs, vision, and touch caused by the corresponding middle cerebral artery ischemic infarction after induced ischemia. The impact of the changes in perception is more comprehensive for postoperative neurological function evaluation. However, there are still some shortcomings, such as differences in the severity of each score not being particularly obvious, some items may have inaccurate scores, and the test is not applicable if animals had their eyeballs removed.

The neurological deficit score in monkeys after stroke (neurological scale for middle cerebral artery infarction)
The neurological deficit score can be applied to many types of NHPs, such as macaques, cynomolgus monkeys, and rhesus monkeys. This test addresses consciousness, sensory system, motor system, and body skeletal muscle coordination to evaluate neurological impairment in monkeys after cerebral ischemia (Table 3) 81 .

J o u r n a l P r e -p r o o f
The evaluation content of the neurological deficit score sheet of stroke monkeys mainly focuses on the consciousness state of neurological function, physical activity, and coordination. Visual field evaluation and spatial orientation are not involved. Detailed scoring criteria and clearly defined scoring items help to evaluate neurological damage accurately in monkeys after cerebral ischemia. However, a large evaluation content requires professional behavior evaluation personnel and more careful records.  Ataxia, but able to climb the wire net 6 Stands spontaneously, falls with a few steps 10 Sits, just able to circle 12 Posed lateral or dorsal recumbency 16 No movement 18 Total 100

Standard neurological scale
Spetzler first introduced this neurological test to study NHPs in 1980. This method mainly examines motor function, along with behavioral changes and eye and cranial nerve damage 32,82 . The innovative test content is listed in Table 4.

Comparison of the above three neurological evaluation methods and prospect
Through comparison and integration of the above three neurological evaluation methods (Tables  2-4), we presented a comprehensive evaluation (Table 5), which can more intuitively reflect the advantages and disadvantages of each scoring table. This evaluation can help scholars choose the appropriate behavioral evaluation method according to their own experimental programs and surgical methods. In short, NHPSS is suitable for the overall neurological assessment after stroke. It has a good evaluation of the NHP overall motor balance and the ability to respond to stimuli, but it has defects in the detailed evaluation of the fine movements of the limbs, such as the movements of the fingers. The neurological scale for middle cerebral artery infarction provides a more detailed evaluation of the movement of the finger joints, but it lacks visual evaluation, and the content is complex. The standard neurological scale can simply reflect movement and balance after a stroke, but it is not comprehensive and sometimes requires other evaluation methods as a supplement. To the abovementioned three commonly used NHP poststroke behavioral overall scores, there are also some separate evaluation items that test the limb mobility and space exploration ability of NHP animals, such as the hill and valley task, two-tube task, and six-well task 27,83 . However, it requires some special cages and props that limit its application so that it is sometimes used to supplement the above three evaluations. In addition, a newly scoring method called modified primate Rankin Scale (pRS) was introduced for NHPs (Table 6), which was quite different from NHPSS or neurological scale in evaluation of middle cerebral artery infarction and standard neurological scale, and focus on neurological impairment about their daily activities in chronic stage of NHPs after stroke 84 . 84 . As a simple and reliable functional test, this method is similar to modified Rankin Scale (mRS) for human, and can be used for evaluating long-term neurological defect in NHPs.

Ethical Considerations and the rule of 3R
Although NHP is the closest animal model to humans, the ethical challenges of this type of experimental animal and the rule of 3R (Reduction, Refinement, Replacement) must be carefully considered when establish a NHP stroke model 85 . No matter what method is used to establish a stroke model in NHPs, invasive surgery or craniotomy should be involved, but the traumatic pain caused by operation must be eliminated or minimized. And in the postoperative observation period, pain, distress or emotional anxiety due to neurological deficits after ischemic stroke should be minimized as much as possible. Intensive veterinary care must be provided for post-operative NHPs, along with appropriate pain relief, environmental improvement, anti-edema, and antibacterial treatment 61 . If post-stroke animals are unable to feed themselves in the short term, parenteral nutrition should be considered. Another key question is how to control the number of non-human primates used for NHP models to the minimum required for the experiment, thereby reducing abuse of non-human primates.
Here, we recommend that researchers can address this issue by consulting with a biostatistician and institutional animal care and use committees before conducting experiments. At the same time, prepare elaborately before the model is established to minimize the operative mortality. It is better to Follow the guidelines for planning animal research and testing during the period of project 86 . In addition, researchers should follow the recommendations of the Stroke Therapy Academic Industry Roundtable (STAIR) which instructs that stroke recovery in NHPs should only be performed when enough evidence of efficacy has been obtained from the research of rodent models 87 .

Perspectives and other large animal models
As highlighted throughout this review, NHP animal models provide significant benefits in improving clinical translation including the utilization of imaging technologies (such as MRI, DSA), neurological dysfunction after stroke, study of thrombolysis and ischemic penumbra. The advantages and disadvantages are summarized in Table 7. When consider to select the NHPs to establish stroke models, researchers must assessment whether they can meet the high requirements of surgical skills, professional veterinary, behavioral testers and imaging evaluation equipment. Meanwhile, other large animals have also been used for stroke models, such as canines and pigs. Canines have a proportion of gray/white matter so it is more closely to human's brain than small animals. Compared with NHP models, canine model cost less maintenance fees and raises less ethical concerns than that of NHPs 88 . The method for establishment of canine stroke model is similar to the method for NHP stroke model, which also covers endovascular stroke model and autologous thrombus model [89][90][91] . Researchers have developed a canine stroke score(CSS) which is similar to the National Institute of Health stroke scale (NIHSS) for evaluate the canine neurological function after stroke 92 . Pig brains are larger than canine brains while presence of obviously cerebral sulci and gyri, and the large intracranial vessel diameter is easy for surgery. The methods for establishment of the pig stroke model include clamp model, electrocoagulation model and autologous thrombus model [93][94][95][96] . Because of the large diameter of cerebral vessels in pigs, embolization with emboli (such as coils, guidewires) is not suitable. Stroke models in pigs have fewer fee and ethical concerns, but can also be used to examine post-stroke imaging changes by MRI and DSA 96,97 . Neurological changes after stroke as assessed by cognitive, memory function and behavioral test in pigs 98 . However, because pigs exhibit high resistance to t-PA, the pig stroke model is contraindicated to study the effect of t-PA combination with other drugs 99 . In short, non-human primates are still the best choice of experimental animal stroke models, it is superior to other large animals while meeting the conditions of high fee, precise equipment and professional operators. Corresponding neurological training can be carried out before operation for the evaluation of neurological dysfunction after stroke.
High costs for purchase and feeding the animals.
Post-stroke evaluation can be performed by using relevant clinical imaging evaluation methods (such as MRI、DSA).
Requirement for relevant test equipment with high standard and the testing fee is expensive.
Ability to undertake acute/chronic stroke, ischemic penumbra changed after ischemia-reperfusion.
Ethical considerations higher.
Increased clinical translation. Required professional veterinarian for animal anesthesia and post-operative care of animals.

Conclusion
At present, the application of NHPs in medical research is limited because of economic and ethical factors. Therefore, when using NHPs to study stroke models, it is important to choose the appropriate model methods for study. This article mainly introduces the various methods currently used for constructing NHPs cerebral ischemia models and briefly evaluates and compares their advantages and disadvantages. The craniotomy clamp method and the interventional embolization method that can control the time of blood flow blockade are suitable for the study of ischemia-reperfusion injury, as well as the study of secondary edema and ischemic penumbra. The use of electrocoagulation and other methods to permanently block blood flow is more suitable for the study of neurological deficits caused by long-term cerebral infarction. The photothrombosis model can simulate the effect of intermittent blood flow blockage and is suitable for simulating the symptoms of intermittent blood flow reduction in human vascular stenosis. The autologous thrombus model can be used to study thrombolytic drugs. In addition, evaluation methods of imaging evaluation and neurological deficits methods of NHPs after stroke are also important. Researchers must choose the most suitable method according to the purpose of the experiment. Although there are many kinds of other animal models used for stroke study, the majority of clinical translations is failed. Translation from bench research to bedside treatment of patients with ischemic stroke is a necessity for newer and more effective therapies. The advantages of non-human primates are in longevity, age-related diseases and human-like cognitive abilities, which make them more reliable for clinical translation. In addition, the similar brain structure and suitable for evaluation with imaging tools make it easier to study for human diseases. Therefore, NHPs are important to improve the translation of stroke therapies, the NHP experiment is necessary despite many difficulties.