CURRENT PERSPECTIVES

Endovascular Ischemic Stroke Models in Nonhuman Primates

Di Wu1,2,3& Ankush Chandra4,5 & Jian Chen1,6

& Yuchuan Ding1,4 & Xunming Ji1,2,3

Published online: 8 November 2017# The American Society for Experimental NeuroTherapeutics, Inc. 2017

Abstract To bridge the gap between rodent and human stud-ies, the Stroke Therapy Academic Industry Roundtable com-mittee suggests that nonhuman primates (NHPs) be used forpreclinical, translational stroke studies. Owing to the fact thatvast majority of ischemic strokes are caused by transient orpermanent occlusion of a cerebral blood vessel eventuallyleading to brain infarction, ischemia induced by endovascularmethods closely mimics thromboembolic or thrombotic cere-brovascular occlusion in patients. This review will make athorough summary of transient or permanent occlusions of acerebral blood vessel in NHPs using endovascular methods.Then, advantages and disadvantages, and potential applica-tions will be analyzed for each kind of models. Additionally,we also make a further analysis based on different kinds ofemboli, various occlusion sites, infract size, abnormal hemo-dynamics, and potential dysfunctions. Experimental modelsof ischemic stroke in NHPs are valuable tools to analyze spe-cific facets of stroke in patients, especially those induced byendovascular methods.

Key Words Ischemic stroke . animal model . thrombolysis .

prognosis . nonhuman primate

Introduction

Cerebral stroke is one of the leading causes of death and dis-ability, accounting for almost 9.6% of all deaths worldwide[1]. In the USA, someone has a stroke every 40 s and some-body dies from this disease every 4 min, demonstrating theomnipresence, frequency, and lethality of this disease [2].Stroke can broadly be divided in to ischemic and hemorrhagic,with diverse clinical and radiographic presentations. Ischemicstroke is a highly complex and heterogeneous disorder, whichcauses neurological damage and dysfunction due to lack ofoxygen delivery to the brain. The final infarct size and theneurological outcome depend on a multitude of factors, in-cluding duration and severity of ischemia; the existence ofcollateral vascular systems; and localization of the infarct,such as the proximal end of intracranial artery (ICA), the maintrunk of middle cerebral artery (MCA), and the branch ofMCA (M2) [3]. In spite of the pervasiveness of this disease,only tissue plasminogen activator (t-PA) has been approvedby the Food and Drug Administration for the acute treatmentof ischemic stroke [4]. Therefore, translation from bench re-search to bedside treatment of patients with ischemic stroke isa necessity for newer and more effective therapies [5, 6].Proposed recommendations to improve the translation ofacute stroke therapies have included evaluation of agents inmultiple animal models of stroke, especially in nonhumanprimates (NHPs) [7].

NHPs are unique owing to their phylogenetic proximity tohumans, and similar physiological functions and life course(just on a shorter time scale) to humans, which provide anadvantage in translational and preclinical biological studies

* Xunming Jijixm@ccmu.edu.cn

1 China–America Institute of Neuroscience, Xuanwu Hospital, CapitalMedical University, Beijing, China

2 Beijing Key Laboratory of Hypoxia Conditioning TranslationalMedicine, Beijing, China

3 Center of Stroke, Beijing Institute for Brain Disorders, Beijing, China4 Department of Neurosurgery, Wayne State University School of

Medicine, Detroit, MI, USA5 Department of Neurological Surgery, University of California San

Francisco, San Francisco, CA, USA6 Department of Neurosurgery, XuanWu Hospital, Capital Medical

University, Beijing, China

Neurotherapeutics (2018) 15:146–155https://doi.org/10.1007/s13311-017-0586-z

[8]. Similarities in life longevity, age-related disorders, andcognitive abilities between NHPs and humans are inherentadvantages for these animal models in translational medicalstudies [9, 10]. Moreover, similar brain size, anatomy, andcompositions have made the use of radiology a strong toolin research with NHPs [9, 10]. In fact, owing to their similarityto humans, many human diseases, including coronary sclero-sis, emphysema, degenerative joint diseases, cancer, and dia-betes, have been found and reported in the middle-aged andold NHPs [11, 12]. For these reasons, NHPs have been de-scribed to be an ideal model for translational research studiesfor various human diseases. Our review will specifically focuson endovascular ischemic stroke models in NHPs and discussthe use of NHPs in stroke research.

NHP stroke models are mainly produced by 2 methods:endovascular approaches and craniotomy. The open surgicalmodels have advantages such as to lower variability and mor-tality because of precise blockage of the target vessel [13].However, the traumatic aspect of invasive surgery might beviewed as a confounding factor because ischemic stroke inhumans is not commonly associated with head trauma [14].Compared with ischemic stroke models achieved by craniot-omy in NHPs, ischemia induced by endovascular methodsclosely mimics thromboembolic or thrombotic cerebrovascu-lar occlusion in patients. However, previous studies adoptedvarious methods to induce focal ischemia in NHPs with var-iable outcomes. Owing to the shared features of pathophysio-logical mechanisms of ischemic stroke in NHPs and humanpatients, we will summarize NHPs models of focal cerebralischemia developed by endovascular methods, and exploretheir respective strengths and weaknesses so as to describethe optimum model for translational studies.

Ischemia Models in NHPs

Based on previous studies and reports, an ideal ischemicstroke model in NHPs needs to meet 3 basic criteria: 1) for-mation of a relatively large infarct size; 2) notable neurologicimpairments; and 3) a long-term survival period (at least 7days) of the animal [13, 15, 16]. Since these models simulatethe pathophysiological process of focal ischemia as seen inpatients, they provide a unique opportunity for developingpotential treatments.

Brains of gyrencephalic primates resemble human brainsmore closely than lissencephalic species. Thus, for the sake ofsimplicity, our review will primarily summarize studies usinggyrencephalic NHPs, including commonly used Macacafascicularis (cynomolgus monkey), Macaca mulatta (rhesusmonkey) and Papio anubis (baboon) [5].

Several features of NHPs account for their great benefits inpreclinical research. Genetically, Macaca and baboons share92% of their genome sequence with that of humans, whereas

mice and rats share only 66% and 64% of their genome withthat of humans, respectively [10]. Anatomically, macaquemonkeys and baboons have a gyrencephalic brain with similarcortical and subcortical structure to human brains. Moreover,NHPs also have a much higher percentage of white matterthan mice, rat, and rabbits, making it similar to the whitematter content in human brains [3, 17]. Further, NHPs havea complete cerebral arterial ring and distributions of the inter-nal carotid and vertebral arteries as those in humans [10, 18].

Studies have shown that leukocyte composition, whole-genome mRNA, and microRNA expression in leukocytes,as well as inflammatory molecules expressed during stoke,differ in rodents comparedwith humans [19]. However, unlikerodents, NHPs have a similar immune composition, whichresponds to ischemic stroke as in humans [6]. Componentsof vascular hemostasis also share great similarities betweenNHPs and humans, including platelets, coagulation proteins,and fibrinolytic proteins [15]. Additionally, NHPs also havethe ability to perform complex motor planning tasks just as inhumans. Lastly, consequences of infarction of cerebral vesselsare very similar to those in humans and are predictable, de-pending on the site of occlusion. For instance, infarction at themain trunk (M1) of the MCA in NHPs causes ischemia pri-marily in the basal ganglia and white matter, similar to patientswith MCA stroke [15]. Moreover, ischemic damage of thewhite matter is a prognostic factor for stroke outcomes.However, only NHP stroke models are able to mimic whitematter pathology in stroke patients most closely to pathologyseen in humans and is recognized as an important limitation torodent ischemic stroke models [20–24]. Thus, NHP strokemodels have numerous intrinsic advantages over rodents, es-pecially owing to their similarity in different fronts of thehuman anatomy and physiology, making them an irreplace-able and indispensable animal model for studying cerebralischemia [25].

Ethical Considerations

While NHPs serve as the closest animal models to humans,the ethical challenges inherent in primate research must becarefully considered when designing a primate stroke model.To begin with, endovascular methods to induce ischemicstroke in NHPs, as a minimally invasive surgery, can eliminateor reduce traumatic pain when invasive surgery or craniotomyare adopted. However, it does not reduce pain, distress, oremotional anxiety due to neurological deficits following theonset of ischemia. Analgesia, environmental enrichment,treatment for antiedema and antibacterial infection, parenteralnutrition, and intensive veterinary care are basic requirementsfor NHP models [16, 26]. Another critical issue is to calculatethe minimal number of animals needed to detect any signifi-cant effect of a neuroprotective treatment in NHPmodels [27].

Endovascular Ischemic Stroke Models in Nonhuman Primates 147

This can be resolved by the research group by consulting abiostatistician, as well as the Institutional Animal Care andUse Committee. In addition, the Institutional Animal Careand Use Committee should also abide by the recommenda-tions of the Stroke Therapy Academic Industry Roundtable,which states that stroke recovery studies inNHPsmay be donewith a gyrencephalic species, similar to humans (e.g., ma-caque monkeys) once sufficient evidence of efficacy has beenobtained for a given therapy in rodent models [5, 28, 29].

Permanent MCA Occlusion

Permanent models of stroke often involve blocking ≥ 1 of themajor cerebral arteries or simply leaving an intraluminal em-bolus in the artery (surgical suture or thrombus) [30]. InNHPs, permanent MCA occlusion is generally developedwith emboli that could not be resolved with t-PA treatment,such as cyanoacrylate adhesive, nylon thread, silk sutures, andpolystyrene spheres. The experimental paradigms using dif-ferent materials for permanent occlusion models are summa-rized in Table 1 [26, 31–35].

Early studies used cyanoacrylate adhesive as emboli, whichwas injected to the distal branch of the MCA. These studiesreported a large variation in infarct size (3.38–20.5 ml), sug-gesting a possibility of migration of the final occlusion site fromthe original site of injection [26]. Thus, in order to restrict theocclusion site, later studies obstructed the distal M2 section ofMCA using silk sutures as the emboli. This approach for per-manent MCA occlusion successfully induced a reproducibleand appropriate cortical infarction in adult rhesus monkeys[31]. In another study, multiple segments of silk sutures werefirst injected within the desired M3 branch, followed by a finalsuture injected into an M1 segment to control for collateralarteries. Such an approach, called endovascular trapping, suc-cessfully occluded collateral vessels to the desired vessel ofocclusion (M3), leading to production of reproducible ischemia

inMCA territory with clinically quantifiable neurologic deficits[32]. Such studies demonstrated the feasibility of producing aprecise permanent occlusive model in NHPs with advance-ments in techniques and materials. Thus, radiographic datafrom the aforementioned study and those from stroke patientsstrongly suggest that precise control of occlusion site is animportant prerequisite for a successful permanent MCAocclusion model in NHPs [16, 36].

Common in practice, most researchers chose to occlude thedistal M1 orM2 for permanent models in order to preserve theperforating branches of lenticulostriate arteries [26, 31].Moreover, studies have found that when the M1 trunk isblocked with glue or silk sutures, these models mimic large-vessel ischemic stroke with an unfavorable prognosis.However, when a distal M1 or M2 segment is blocked, thosemodels generally have a modest infarct size and do not cause afatal ischemic stroke in the animal [26, 31]. These discussedpermanent models satisfy the 3 criteria for an ideal ischemicstroke model and have been successfully used to generatepreclinical and basic science data for investigating promisingneuroprotective therapies, novel restorative strategies, andbrain plasticity in poststroke recovery [37].

An alternative approach for developing permanent occlu-sive models has been the injection of embolic spheres. In astudy by Cook et al. [34], the authors injected polystyrenespheres into the ICA of cynomolgus monkeys to induce focalischemia. They found that emboli with larger diameter spheres(400 μm) produced severe neurological impairments due toischemia in the basal ganglia and internal capsule, whereasspheres with a modest diameter (200 μm) only induced a mildand lasting neurologic deficit; spheres with a smaller diameter(100 μm) did not induce any noticeable neurologic deficit,even when used in large quantities [34]. In another study, alacunar-type stroke model was developed in macaque mon-keys by injecting agarose spheres (50 μm in diameter) into theICA of these monkeys [35]. The discussed sphere-inducedmodels of occlusion did not exhibit obvious neurological

Table 1 Summary of permanent middle cerebral artery occlusion (MCAo) in nonhuman primates

Source Species Location MCAo duration Emboli Size (ml) Outcome

D'Arceuil et al. [26] Cynomolgus M1 Permanent Glue 3.38–20.5 Death (n = 2)Survival (n = 2)

Zhang et al. [31] Rhesus Distal M2 Permanent Silk suture 1.08–7.8 Death (n = 2)Sacrifice (n = 3)

Tong et al. [32] Rhesus M3 and M1 Permanent Silk suture 3.0–12.9 Death (n = 2)Sacrifice (n = 6)

Sato et al. [35] Cynomolgus ICA Permanent(lacunar stroke)

Sephacryl beads NA NA

Cook et al. [34] Cynomolgus ICA Permanent Polystyrene spheres 10–40 Death (n = 1) Sacrifice (n = 18)

Rodriguez-Mercado et al. [33] Rhesus M1 Permanent Silk sutures 0.2–15.0 Sacrifice (n = 3)

ICA = internal carotid artery; NA = not applicable

148 D. Wu et al.

impairments. Although the number and volume of infarct sitecould be determined based on magnetic resonance images, itis not possible to control for the occlusion site and retrievespheres in these models. However, these models can be usedto mimic small embolic strokes that occur in multiple clinicalscenarios, such as cardiopulmonary bypass, carotid endarter-ectomy or stenting, and other endovascular procedures [34].

While permanent MCA occlusion models in NHPs canmimic pathological processes and outcomes in ischemicstroke, they precluded the possibility of t-PA thrombolysisand could not simulate the treatment process in ischemic pa-tients. Thus, to understand the outcomes of treatments in is-chemic stroke, transient MCA occlusion models are needed.

Transient MCA Occlusion

TransientMCA occlusion in NHPs is generally induced with aretrievable tool such as a microcatheter, an inflatable balloon,and a microcoil. Ischemia occurs as a result of consequentblockade of blood flow, whereas reperfusion is achieved whenthe tool is withdrawn. Transient MCA occlusion in NHPs byendovascular methods from various studies are summarized inTable 2 [14, 16, 26, 37–45].

It is relatively easy and cost friendly to create focal ische-mia in NHPs with a microcatheter. A previous study reporteda small infarct size in cynomolgus monkeys using a Prowler-10 microcatheter (0.55 mm outer diameter) to block the distalbranch ofM1 [37]. The samemethod was also widely adoptedto develop a reversible MCA occlusion model in NHPs [46].However, the infarct size in this model was relatively small

(0.4–3.2 ml, Diffusion Weighted Imaging (DWI) at 3 h) andthere was no information about neurologic deficit after the onsetof ischemia. To improvise upon this model, another study useda larger microcatheter (Rebar 18 microcatheter) to induce focalischemia in NHPs. However, infarct sizes were significantlylarger than the expected vascular territory and correlated witha significant neurologic deficit when the microcatheter wasnavigated into the M1 segment of the MCA in rhesus monkeys[16]. Such findings can be explained by the fact that a largemicrocatheter serves as both an embolus and irritant for theMCA. While it is difficult to induce a homogenous infarct sizejust by inserting a microcatheter, it is possible to control ische-mia and reperfusion as needed, making this method desirable.Thus, in order to resolve the issue of heterogeneous infarct sizewith this model, a combination of a microcatheter and addition-al emboli can help control the occlusion site and duration.

Using a microcatheter for transient occlusion confers theadvantage of delivering agents to the occlusion site. Thus, thisapproach makes it easy to evaluate the efficacy and feasibilityof local or regional infusion of fibrinolytics, neuroprotectivedrugs, or ice-cold saline before or after the reperfusion [47]. Infact, studies have suggested that microcatheter delivery mayprovide more benefits for the analysis of penumbral freezing,prevention of reperfusion injury, and enhanced recanalizationefficacy in NHPs models [47].

Ischemia and reperfusion in NHPs can also be induced byinflation and deflation of a catheter-delivered balloon into ablood vessel. In 2006, Gao et al. [42] developed a novel bal-loon catheter (0.61 mm outer diameter for microcatheter,1.1 mm outer diameter for a deflated balloon) to temporarilyocclude and reperfuse the MCA by inflation and deflation of

Table 2 Summary of transient middle cerebral artery occlusion (MCAo) in nonhuman primates

Source Species Location MCAoduration

Approach Size Outcome

Del Zoppo et al. [44] Baboons M1 3 h Balloon 3.2 ± 1.5 mL Survival (n = 5)

Hamberg et al. [43] Baboons M1 3 h Balloon (n = 4)Microcoil (n = 1)

475–880 mm2 Sacrifice (n = 5)

Guo et al. [45] Rhesus M1 2 h Microcoil NA Survival(n = 6)

de Crespigny et al. [37] Cynomolgus M2 3 h Microcatheter 0.3–1.9 ml Sacrifice (n = 4)

Gao et al. [42] Rhesus M1 2 h Balloon NA NA

Zhao et al. [14] Rhesus M1 2 h Microcoil 9.8–24.5 ml NA

M2 2 h Microcoil 1.6–3.1 ml NA

Wu et al. [16] Rhesus M1 3 h Microcatheter 9.7–20.1 ml Death (n = 5)Survival (n = 1)

Zhang et al. [38] rhesus M1 2 h Microcoil 2.95 + 0.48% Survival (n = 17)

Yi et al. [39] Rhesus M2 NA Microcatheter NA Survival (n = 7)

Jungreis et al. [40] Nemestrina M1 NA Balloon NA NA

Schwartz et al. [41] Baboons ACA 3 h Microcoil 1.23, 1.65 ml Survival (n = 2)

NA = not applicable; ACA = anterior cerebral artery

Endovascular Ischemic Stroke Models in Nonhuman Primates 149

the balloon, respectively, in cynomolgus monkeys. The ani-mals in this study exhibited hemiplegia and had a relativelylarge infarct size (mainly at basal ganglia and cortex) [42].Along similar lines, another study developed an ischemia–perfusion model to induce a 3-h ischemic stroke using thecatheter-based balloon technology discussed above. Thisgroup also reported relatively large stroke lesion sizes com-pared with other ischemic–reperfusion models in NHPs, dem-onstrating the robustness and reliability of this partial occlu-sion model [43]. Most importantly, transient ischemic strokemodels form large lesion, causing neurological impairmentalong with a long survival period of the animal, satisfyingthe 3 key elements required to be an ideal stroke model, asdiscussed above. Although this model allows us to control theduration of occlusion, one of the main limitations of this mod-el is that it only allows occlusion of the proximalMCA. This isdue to the large diameter of the balloon catheter comparedwith that of the distal segments of MCA.

Another microcatheter-based MCA occlusion can be donevia the introduction of a microcoil. A microcoil can be pre-cisely positioned by advancing the microcatheter, and it can berepositioned until appropriate complete MCA occlusion isachieved. By introducing the microcoil into the M1, re-searchers can block blood flow into the blood vessel and canremove the coil to simulate reperfusion of M1. Studies havefound that microcoil-induced M1 occlusion caused a largeinfarct in deep brain structures such as the basal ganglia andinternal capsule, as well as the white matter and cortex [38,45]. In addition, these models caused severe neurological im-pairments as seen as in stroke patients. In contrast to the neu-rological and anatomical outcomes with microcoil-inducedM1 occlusion, M2 occlusion using the same technique actu-ally caused a smaller infarct size, primarily in the cortex andthe Sylvian fissure, and much milder neurological impairment[14]. The microcatheter-derived microcoil deployment meth-od simulated the ischemia–reperfusion process in patients, inwhich the occlusion site and duration could be precisely con-trolled. In addition, this model also exhibited increased stabil-ity, safety, and reproducibility across various studies.Moreover, this transient ischemic stroke model qualifies asan ideal ischemic stroke model, as it satisfies the 3 requiredcriteria as discussed above.

In all the transient ischemic stroke models discussedabove, complete reperfusion is achieved when the retriev-able tool is withdrawn in NHPs. However, in an intriguingstudy, when a balloon was used to induce focal ischemia inbaboons, in situ thrombosis in perforating branches wasobserved [44]. Based on previous reports and our ownexperience, thrombosis and vasospasm could significantlyaffect ischemia size and clinical outcomes and thus it isimportant to keep these factors in mind when developingsuch models in NHPs [14]. All in all, transient MCA oc-clusion in NHPs is ideal for studying the effect of

promising therapies that require proper reperfusion of theoccluded cerebral vessel.

Thrombus MCA Occlusion in NHPs

Thrombi and emboli are the main causes for ischemicstroke in humans, with about 50% due to large-vessel ath-erosclerosis and 20% caused by cardioembolism [3]. Clotmodels in NHPs closely mimic the pathogenesis of strokepatients and hold potential for thrombolysis treatment.Thrombus models by endovascular methods are summa-rized in Table 3 [16, 40–58]. In order to achieve an idealischemic stroke model in NHPs using thrombi, 2 key fac-tors are needed: 1) accurate control of thrombus; and 2)timely recanalization of the vessel.

As reported in rodent models of thromboembolic stroke,the major concern of these models in NHPs is control over theintroduced embolus, which may be determined by the site ofocclusion and nature of the clot [55]. Such concerns arosewhen 2 studies reported an experimental thromboembolicmodel in NHPs in which the clots introduced were not at theinjection site but migrated down to a nearby site [48, 49]. Inorder to selectively block the MCA, 1 study combined neuro-surgical and endovascular methods in rhesus monkeys [50].However, when the clot was delivered under pressure througha microcatheter, the clot was prone to break down owing to itsfragile quality, compromising the model [50, 52]. In a recentstudy, a single fibrin-rich clot was used to induce ischemia inrhesusmonkeys, a technique derived from rodent models [16].Using a single fibrin-rich clot significantly increased thetoughness and flexibility of the clot as red blood cells andplatelets were washed out during clot preparation [56]. As aresult, clots did not break down when they were deliveredfrom the microcatheter to the site of interest. Moreover, therewas much better control over these clots with the aid of amicrocatheter and were mainly located at M1 or M2 segment[16]. While it is challenging to control exactly the occlusionsite using only the endovascular method of delivery, improvedthrombus composition, a definite load (clot size), delicate op-erative skills, and a microcatheter-aided delivery increase thecontrol of clot embolus significantly.

In spite of the omnipresence of ischemic stroke, throm-bolysis with t-PA is the only approved treatment for pa-tients with ischemic stroke, with a narrow therapeuticwindow of 3 to 4.5 h following onset of symptoms [47].Successful recanalization with t-PA is imperative in bothpatients and NHPs when the MCA is permanently occlud-ed by a clot [16, 57]. However, only 1 paper reported thatintra-arterial reteplase appeared to be effective in achiev-ing recanalization in NHP models of intracranial throm-bosis [54]. Unfortunately, small sample size and signifi-cant variability within each treatment group hindered

150 D. Wu et al.

statistical analysis in this study. In another study, uroki-nase was reported to improve neurologic deficits and re-duce cerebral infarction in the thromboembolic strokemodel of cynomolgus monkey [53]. However, a high doseled to cerebral edema and hemorrhage despite a betterneuroprotective effect. Moreover, another paper showedthat urokinase reduced infarct volume owing to its throm-bolytic effect on in situ thrombosis of perforatingbranches formed in a reversible eccentric balloon com-pression (3 h) baboon model [44]. While there are somepromising studies favoring the use thrombolytics inNHPs, additional studies are needed to establish an algo-rithm, similar to the thrombolysis in myocardial infarctionalgorithm, to evaluate response to the thrombolysis treat-ment in NHPs and humans. Moreover, future studies arealso needed that investigate safe administration of recom-binant t-PA and therapeutic effect of recombinant t-PAfollowing acute ischemia in NHPs.

In the past decade, 5 randomized trials showed efficacy ofendovascular thrombectomy in patients with acute ischemicstroke caused by occlusion of arteries of the proximal anteriorcirculation [58]. Thrombus occlusion models in the proximalof MCA or ICA have been established in NHPs but haveproven to be fatal without successful recanalization of theblood vessel [16, 52]. Moreover, to date, there are noestablished NHPs models that simulate endovascularthrombectomy in humans. Thus, research to develop thesemodels is imperative.

The greatest advantage of thromboembolic models is 2-fold: 1) simulation of the pathogenesis of cerebrovascular oc-clusion and focal ischemia; and 2) the opportunity for recan-alization treatment of occluded vessels. On the flip side, thesemodels may not replicate the characteristics of spontaneousthrombus or thrombus superimposed on ruptured atheroscle-rotic plaque, conferring the data from such models limited.

Infarct Size, Symptoms, and Prognosis

Endovascular techniques appear to have increased variabilityin stroke size with a higher mortality rate than open vascularocclusions in focal ischemia models of NHPs [5]. Thus, manyearly studies only reported the infarct size and neurologicalexamination scores in the acute state [5]. However, recentreports have found an association between infarct size andoutcomes in NHP stroke models developed usingendovascular methods [16, 26, 30, 31]. Studies have previous-ly reported the correlation between functional outcome andinfarct size in both rodents and NHPs [13, 59, 60]. As a result,neurological impairments and prognosis in NHPs models arefirst characterized based on the infarct size, and then furtheranalyzed for the application of behavioral tests in this paper.Studies have shown that when infarct size is < 3.0 ml, NHPsexhibit a mild, short-term functional deficit or abnormal be-havior following recovery from anesthesia, according toSpetzler Scores, in transient and permanent MCA occlusionmodels [26, 34, 37]. These animals mimic patients with minorstrokes showing abnormal radiographs but with or withoutfunctional deficit. On the contrary, when infarct size is > 15ml, NHPs exhibit severe neurological impairments in tran-sient, permanent, and thrombus MCA occlusion models, in-cluding symptoms of hemiplegia of contralateral limbs, comaor delirium, decreased muscle tone, and salivation [14, 16,26]. These NHPs simulate clinical features of patients withsevere stroke with a poor prognosis. Moreover, when infarctsize is between 3.0 and 15.0 ml, the most common symptomobserved is contralateral motor weakness in all transient, per-manent, and thrombus MCA occlusion models [14, 32, 37,44]. These animals mimic a majority of stroke survivors hav-ing a permanent functional impairment.

Stroke patients generally exhibit sensorimotor impair-ments and paralysis. In clinical studies of acute ischemic

Table 3 Summary of thrombus middle cerebral artery occlusion (MCAo) in nonhuman primates

Source Species Location Thrombolysis Emboli Size Outcome

Kito et al. [52] Cynomolgus M1 No Clot 28.3 + 12.4% Sacrifice (n = 10)

Qureshi et al. [54] Rhesus M1 Yes, 2h (t-PA) Thrombin injection NA Sacrifice (n = 16)

Susumu et al. [53] Cynomolgus M1 No Clot 24.7 + 3.5% Sacrifice (n = 10)

M1 Yes, 2 h (UA) Clot 11.9 + 3.9% Sacrifice (n = 6)

M1 Yes, 6 h (UA) Clot 7.6 ± 2.5% Death (n = 2)Sacrifice (n = 3)

Wu et al. [16] Rhesus M1 No Clot 7.7–13.5 ml Death (n = 2)Survival (n = 2)

M2 No Clot 3.2–7.1 ml Survival (n = 6)

Kuge et al. [51] Cynomolgus ICA No Clot NA NA

Xie et al. [49] NA ICA No Clot NA NA

t-PA = tissue plasminogen activator; NA = not applicable; UA = urokinase; ICA = internal carotid artery

Endovascular Ischemic Stroke Models in Nonhuman Primates 151

stroke, the modified Rankin Scale is used to assess func-tional status at 90 days following the stroke event or theindex event [59]. The ability to perform motor planningtasks is unique to NHPs and humans, making NHP modelsextremely important to understand the functional implica-tions of stroke. Behavioral tests for postischemic stroke-induced functional impairments have been reported inNHPs models undergoing craniotomy. These include a sen-sorimotor battery of tasks and object retrieval detour task[13, 61]. Similar function tests could also be performed inNHP stroke models using endovascular methods whenthey meet all 3 criteria for an ideal NHP model. Thus, toreiterate, NHPs offer unique opportunities to model motorimpairments in stroke patients and to understand corticalcontrol of motor tasks in such patients.

Abnormal Hemodynamics, Occlusion Siteand Clinical Outcomes

Platelets play an important role in the pathogenesis of ce-rebral ischemia. This begins with aggregation of plateletsfollowing their activation. Moreover, platelet secretoryproducts have been implicated in the evolution of stroke[62]. As shown in Fig. 1, when the M2 segment of theMCA is permanently blocked with silk sutures (Fig. 1A)

or a clot (Fig. 1B), there is normal blood flow from the ICAto M1, and subsequent blood flow to the second branch ofM2. Thus, a local perfusion defect will lead to focal ische-mia and mild neurological deficit. However, when M1 seg-ment is blocked with a clot (Fig. 1C) or a retrievable tool(Fig. 1D), there is normal blood flow only from the ICA tothe anterior cerebral artery and not to MCA branches, asexpected. When reperfusion is achieved in these modelsusing thrombolysis treatment or by the withdrawal of thetool, infarcts are primarily found at the putamen and inter-nal capsule, supplied by the horizontal segment of MCA[48]. Fortunately, these 4 models discussed above meet all3 criteria for an ideal NHP model, validating their use inresearch. However, researchers should keep in mind thatunexpected abnormal hemodynamics may occur in rarecases. Figure 2 shows abnormal hemodynamics in anNHP model, caused by injecting a clot. Immediately afterinjection of a clot into the MCA, a perfusion deficit wasinitially observed in the proximal segment of the artery.However, 3 h later this clot extended into the ICA,blocking the perfusion of regions beyond the occludedICA. As a result of the abnormal blockade of ICA, theseNHPs will exhibit severe neurological impairments.

Pathologically, these NHPs models exhibited cerebraledema, which was significant within 7 days following theonset of ischemia both in stroke patients and NHP models

Fig. 1 The evolution of abnormal hemodynamics in different inducingmethods. (A) When silk sutures (in blue) are injected to distal branch ofM1 (M2), platelets (in pink) will aggregate to the proximal of M2. (B)When a clot (in red) is injected to M2, platelets (in pink) will aggregate tothe proximal of M2. Normal blood flow is preserved. (C) When a clot (inred) is injected to main trunk of middle cerebral artery (M1), platelets (in

pink) will aggregate to the proximal of M1. Blood flow from MCAwillonly go to the anterior cerebral artery (ACA). (D) when a retrievable tool(e.g., a microcatheter in blue) is inserted at M1 segment of the middlecerebral artery (MCA), blood flow from the MCA will only go to theACA. LSA = lenticulostriate arteries

152 D. Wu et al.

[26, 63]. Clinically, this time period correlated with ahigher mortality rate in both humans and NHPs, givingus a deeper understanding of the implications of poststrokeedema [26, 57].

Lastly, it is important to note that although there aremany reports on NHP models of stroke focusing in anteriorcerebral circulation, those models almost exclusively focuson MCA occlusion. Only few studies reported vertebralartery or basilar artery occlusion models in NHPs withendovascular techniques. Based on reports from caninemodels, studies have found that vertebral or basilar arteryocclusion models exhibited a severe and fatal neurologicaldeficit, making it difficult to study the clinical and patho-physiological effects of such strokes [64]. Thus, there is aneed to develop additional NHP models that can help sim-ulate clinical consequences of vertebral and basilar arterystrokes in human. International cooperation in preclinicalstroke research, especially in NHPs models, is needed [65].

Conclusions

Ischemic stroke is a complex and heterogeneous disorder thataffects a large population in the world. Ischemic stroke is alsoa long-term pathologic process, and includes 3 phases: 1)acute, 2) subacute; and 3) chronic. Experimental models ofischemic stroke are valuable tools with which to analyze spe-cific clinical, pathological, and morphological facets of strokeparalleling those in humans. Over the years, several NHPsstroke models have been developed using various differenttechniques. Each NHPs model has its own pros and cons,and thus we should select the appropriate NHPs model basedon the goals of our study and outcomes we would like tomeasure and analyze in a preclinical context. Among otheranimal models, NHPs models provide the most valuable pre-dictions for human trials due to the anatomical, immunologi-cal, and pathophysiological homology between NHPs andhumans. As the mainstay of NHP scientific research and

Fig. 2 The thrombus extended tointernal carotid artery (ICA) fol-lowing the original injection of aclot at main trunk of middle cere-bral artery (M1). (A) Anterior–posterior and (B) lateral views ofnormal perfusion in adult rhesusmonkey when the microcatheterwas placed at the main trunk (M1)of middle cerebral artery (MCA).(C) Anterior–posterior and (D)lateral views showing perfusiondeficit in the proximal end ofMCA just following the injectionof clot. (E) Anterior–posteriorand (F) lateral views exhibitingno perfusion in the ICA segmentat 3 h following the injection ofclot

Endovascular Ischemic Stroke Models in Nonhuman Primates 153

potential preclinical studies, M. fascicularis (cynomolgusmonkey) andM. mulatta (rhesus monkey), and the clot meth-od are recommended as preferred species and the appropriatemethod for testing human neuroprotectants in translationalstroke studies, respectively [50, 66].

Acknowledgments This work was supported by National NaturalScience Foundation of China for Outstanding Youth (81325007);National Natural Science Foundation of China (81500997, 81771260);Chang Jiang Scholars Program (#T2014251) from the Chinese Ministryof Education; National Natural Science Foundation of China(81620108011) ; Na t i ona l Key R&D Prog ram of Ch ina(2017YFC1308401); and the Bmission^ talent project of BeijingMunicipal Administration of Hospitals (SML20150802); BeijingMunicipal Administration of Hospitals’ Youth Programme(QML20170802). Ankush Chandra is a Howard Hughes MedicalInstitute Medical Research Fellow.

References

1. Truelsen T, Begg S, Mathers C. The global burden of cerebrovas-cular disease. Geneva: World Health Organization, 2000.

2. Mozaffarian D, Benjamin EJ, Go AS, et al. Executive summary:heart disease and stroke statistics—2016 update: a report from theAmerican Heart Association. Circulation 2016;133:447-454.

3. Sommer CJ. Ischemic stroke: experimental models and reality. ActaNeuropathol 2017;133:245-261.

4. O’Collins VE,MacleodMR, Donnan GA, Horky LL, van derWorpBH, Howells DW. 1,026 experimental treatments in acute stroke.Ann Neurol 2006;59:467-477.

5. Cook DJ, Tymianski M. Nonhuman primate models of stroke fortranslational neuroprotection research. Neurotherapeutics 2012;9:371-379.

6. Herson PS, Traystman RJ. Animal models of stroke: translationalpotential at present and in 2050 Future Neurol 2014;9:541-551.

7. Jickling GC, Sharp FR. Improving the translation of animal ische-mic stroke studies to humans Metab Brain Dis 2015;30:461-467.

8. Wu D, Yue F, Zou C, Chan P, Zhang YA. Analysis of glucosemetabol i sm in cynomolgus monkeys dur ing ag ing .Biogerontology 2012;13:147-155.

9. Kemnitz JW. Calorie restriction and aging in nonhuman primates.ILAR J 2011;52:66–77.

10. Grow DA, McCarrey JR, Navara CS. Advantages of nonhumanprimates as preclinical models for evaluating stem cell-based ther-apies for Parkinson’s disease. Stem Cell Res 2016;17:352-366.

11. Bruns CM, Baum ST, Colman RJ, Kemnitz JW, Weindruch R,Abbott DH. Insulin resistance and impaired insulin secretion inprenatally androgenized male rhesus monkeys. J Clin EndocrinolMetab 2004;89:6218-6223.

12. Colman RJ, Anderson RM, Johnson SC, et al. Caloric restrictiondelays disease onset and mortality in rhesus monkeys. Science2009; 325:201-204.

13. Cook DJ, Teves L, Tymianski M. Treatment of stroke with a PSD-95 inhibitor in the gyrencephalic primate brain Nature 2012;483:213-217.

14. Zhao B, Zhao B, Shang G, et al. A more consistent intraluminalrhesus monkey model of ischemic stroke. Neural Regen Res2014;9:2087-2094.

15. Fukuda S, del Zoppo GJ. Models of focal cerebral ischemia in thenonhuman primate. ILAR J 2003;44:96-104.

16. Wu D, Chen J,Wang B, et al. Endovascular ischemic stroke modelsof adult rhesus monkeys: a comparison of two endovascularmethods. Sci Rep 2016;6:31608.

17. Borlongan CV, Jolkkonen J, Detante O. The future of stem celltherapy for stroke rehabilitation. Future Neurol 2015;10: 313-319.

18. Chandra A, Li WA, Stone CR, Geng X, Ding Y. The cerebralcirculation and cerebrovascular disease I: Anatomy. Brain Circ2017;3:45-56.

19. Sharp FR, Jickling GC. Modeling immunity and inflammation instroke: differences between rodents and humans? Stroke 2014;45:e179-e180.

20. Bihel E, Roussel S, Toutain J, Bernaudin M, Touzani O. Diffusiontensor MRI reveals chronic alterations in white matter despite theabsence of a visible ischemic lesion on conventional MRI: a non-human primate study Stroke 2011;42:1412-1419.

21. Sozmen EG, Hinman JD, Carmichael ST. Models that matter:White matter stroke models. Neurotherapeutics 2012;9: 349-358.

22. Hara K, Yasuhara T, Matsukawa N, et al. Hippocampal CA1 cellloss in a non-human primate model of transient global ischemia: apilot study. Brain Res Bull 2007;74:164-171.

23. Hara K, Yasuhara T, Maki M, et al. Anomaly in aortic arch alterspathological outcome of transient global ischemia in Rhesus ma-caques. Brain Res 2009;1286:185-191.

24. Shindo A, Liang AC, Maki T, et al. Subcortical ischemic vasculardisease: Roles of oligodendrocyte function in experimental modelsof subcortical white-matter injury. J Cereb Blood FlowMetab 2016;36: 187-198.

25. Fan J, Li Y, Fu X, Li L, Hao X, Li S. Nonhuman primate models offocal cerebral ischemia. Neural Regen Res 2017;12:321-328.

26. D'Arceuil HE, Duggan M, He J, Pryor J, de Crespigny A. Middlecerebral artery occlusion in Macaca fascicularis: acute and chronicstroke evolution. J Med Primatol 2006;35:78-86.

27. Muller PY, Brennan FR. Safety assessment and dose selection forfirst-in-human clinical trials with immunomodulatory monoclonalantibodies. Clin Pharmacol Ther 2009;85:247-258.

28. Stroke Therapy Academic Industry Roundtable (STAIR).Recommendations for standards regarding preclinical neuroprotec-tive and restorative drug development. Stroke 1999;30:2752-2758.

29. Fisher M, Feuerstein G, Howells DW, et al. Update of the stroketherapy academic industry roundtable preclinical recommenda-tions. Stroke 2009;40:2244-2250.

30. Krieger DW, Yenari MA. Therapeutic hypothermia for acute ische-mic stroke: what do laboratory studies teach us? Stroke 2004;35:1482-1489.

31. Zhang X, Tong F, Li CX, et al. Temporal evolution of ischemiclesions in nonhuman primates: a diffusion and perfusionMRI studyPLOS ONE 2015;10:e0117290.

32. Tong FC, Zhang X, Kempf DJ, et al. An enhanced model of middlecerebral artery occlusion in nonhuman primates using anendovascular trapping. AJNR Am J Neuroradiol 2015;36:2354-2359.

33. Rodriguez-Mercado R, Ford GD, Xu Z, et al. Acute neuronal injuryand blood genomic profiles in a nonhuman primate model for is-chemic stroke Comp Med 2012;62:427-438.

34. Cook DJ, Teves L, Tymianski M. A translational paradigm for thepreclinical evaluation of the stroke neuroprotectant Tat-NR2B9c ingyrencephalic nonhuman primates. Sci Transl Med 2012;4:154ra133.

35. Sato Y, Chin Y, Kato T, et al. White matter activated glial cellsproduce BDNF in a stroke model of monkeys Neurosci Res 2009;65:71-78.

36. Wein TH, Bornstein NM. Stroke prevention: cardiac and carotid-related stroke. Neurol Clin 2000;18:321-341.

37. de CrespignyAJ, D'Arceuil HE,Maynard KI, et al. Acute studies ofa new primate model of reversible middle cerebral artery occlusionJ Stroke Cerebrovasc Dis 2005;14:80-87.

154 D. Wu et al.

38. Zhang Y, Fan F, Zeng G, et al. Temporal analysis of blood-brainbarrier disruption and cerebrospinal fluid matrix metalloproteinasesin rhesus monkeys subjected to transient ischemic stroke. J CerebBlood Flow Metab 2017;37: 2963-2974.

39. Yi KS, Choi CH, Lee SR, et al. Sustained diffusion reversal with in-bore reperfusion in monkey stroke models: Confirmed by prospec-tive magnetic resonance imaging J Cereb Blood Flow Metab2017;37:2002-2012.

40. Jungreis CA, Nemoto E, Boada F, Horowitz MB. Model of revers-ible cerebral ischemia in a monkey model. AJNRAm J Neuroradiol2003;24:1834-1836.

41. Schwartz AE, Pile-Spellman J. New model of reperfused stroke byocclusion of the anterior cerebral artery in baboons Acta Neurochir(Wien) 2011;153:327-331.

42. Gao H, Liu Y, Lu S, Xiang B,Wang C. A reversible middle cerebralartery occlusion model using intraluminal balloon technique inmonkeys. J Stroke Cerebrovasc Dis 2006;15:202-208.

43. Hamberg LM, Hunter GJ, Maynard KI, et al. Functional CT perfu-sion imaging in predicting the extent of cerebral infarction from a 3-hour middle cerebral arterial occlusion in a primate stroke model.AJNR Am J Neuroradiol 2002;23:1013-1021.

44. Del Zoppo GJ, Copeland BR, Harker LA, et al. Experimental acutethrombotic stroke in baboons. Stroke 1986;17:1254-1265.

45. Guo J, Zheng HB, Duan JC, et al. Diffusion tensor MRI for theassessment of cerebral ischemia/reperfusion injury in the penumbraof non-human primate stroke model. Neurol Res 2011;33:108-112.

46. LaVerde GC, Jungreis CA, Nemoto E, Boada FE. Sodium timecourse using 23Na MRI in reversible focal brain ischemia in themonkey J Magn Reson Imaging 2009;30:219-223.

47. Fisher M, Saver JL. Future directions of acute ischaemic stroketherapy Lancet Neurol 2015;14:758-767.

48. Molinari GF, Moseley JI, Laurent JP. Segmental middle cerebralartery occlusion in primates: an experimental method requiringminimal surgery and anesthesia. Stroke 1974;5:334-339.

49. Xie Y, Munekata K, Seo K, Hossmann KA. Effect of autologousclot embolism on regional protein biosynthesis of monkey brainStroke 1988;19:750-757.

50. Gauberti M, Obiang P, Guedin P, et al. Thrombotic stroke in theanesthetized monkey (Macaca mulatta): characterization byMRI-apilot study. Cerebrovasc Dis 2012;33:329-339.

51. Kuge Y, Yokota C, Tagaya M, et al. Serial changes in cerebral bloodflow and flow-metabolism uncoupling in primates with acute throm-boembolic stroke J Cereb Blood Flow Metab 2001;21:202-210.

52. Kito G, Nishimura A, Susumu T, et al. Experimental thromboem-bolic stroke in cynomolgus monkeys. J Neurosci Methods2001;105:45-53.

53. Susumu T, Yoshikawa T, Akiyoshi Y, Nagata R, Fujiwara M, KitoG. Effects of intra-arterial urokinase on a non-human primatethromboembolic stroke model. J Pharmacol Sci 2006;100:278-284.

54. Qureshi AI, Suri MF, Ali Z, et al. Intraarterial reteplase and intra-venous abciximab for treatment of acute ischemic stroke. A prelim-inary feasibility and safety study in a nonhuman primate model.Neuroradiology 2005;47:845-854.

55. Orset C, Macrez R, Young AR, et al. Mouse model of in situ throm-boembolic stroke and reperfusion Stroke 2007;38:2771-2778.

56. Zhang L, Zhang RL, Jiang Q, Ding G, Chopp M, Zhang ZG. Focalembolic cerebral ischemia in the rat Nat Protoc 2015;10:539-547.

57. Hacke W, Schwab S, Horn M, Spranger M, De Georgia M, vonKummer R. ‘Malignant’ middle cerebral artery territory infarction:clinical course and prognostic signs. Arch Neurol 1996;53:309-315.

58. Goyal M, Menon BK, van Zwam WH, et al. Endovascularthrombectomy after large-vessel ischaemic stroke: a meta-analysisof individual patient data from five randomised trials. Lancet2016;387:1723-1731.

59. Banks JL, Marotta CA. Outcomes validity and reliability of themodified Rankin scale: implications for stroke clinical trials: a lit-erature review and synthesis. Stroke 2007;38:1091-1096.

60. Rogers DC, Campbell CA, Stretton JL, Mackay KB. Correlationbetween motor impairment and infarct volume after permanent andtransient middle cerebral artery occlusion in the rat. Stroke 1997;28:2060-2066.

61. McEntire CR, Choudhury GR, Torres A, Steinberg GK, RedmondDE Jr, Daadi MM. Impaired arm function and finger dexterity in anonhuman primate model of stroke: motor and cognitive assess-ments Stroke 2016;47:1109-1116.

62. Joseph R, Welch KM, Oster SB, Grunfeld S, D’Andrea G A plas-matic factor may cause platelet activation in acute ischemic stroke.Circ Res 1989;65:1679-1687.

63. Neugebauer H, Witsch J, Zweckberger K, Jüttler E. Space-occupying cerebellar infarction: complications, treatment, and out-come Neurosurg Focus 2013;34:E8.

64. Atchaneeyasakul K, Guada L, Ramdas K, et al. Large animal canineendovascular ischemic stroke models: a review. Brain Res Bull2016;127:134-140.

65. Dirnagl U, Hakim A, Macleod M, et al. A concerted appeal forinternational cooperation in preclinical stroke research. Stroke2013;44:1754-1760.

66. Cyranoski D. Monkey kingdom. Nature 2016;532:300-302.

Required Author Forms

Disclosure forms provided by theauthors are available with theonline version of this article.

Endovascular Ischemic Stroke Models in Nonhuman Primates 155