What is the best experimental model for developing novel therapeutics in peripheral artery disease?

Abstract

Clinical problem:

More than 200 million people worldwide have peripheral artery disease (PAD). PAD affects quality of life and is associated with significant morbidity and mortality. Standard treatment for severe cases of PAD is surgical or endovascular revascularization. However, up to 30% of patients are not candidates for open or endovascular procedures, due to high operative risk or unfavorable vascular involvement. Furthermore, revascularization procedures may be insufficient to adequately improve microvascular tissue perfusion, wound healing, or limb salvage. Accordingly, regardless of advances in treatment modalities, outcomes of patients with PAD have remained unfavorable. Therefore, new medical therapeutic approaches are much needed. Small animal models are indispensable tools for the understanding of PAD physiopathology and the development of novel medical therapies.

Recommendations for increasing translation from animal models:

Development of animal models that more closely mimic the pathophysiology (with occlusive atherothrombosis and chronic development of limb ischemia), can incorporate the cardiovascular risk factors associated with this disease state, and focus on more clinically-relevant outcomes are critical. In practice, this means using both animals that develop atherosclerosis and methods for the application of gradual arterial occlusion to induce hind limb ischemia. Doing so will likely help identify novel targets for intervention and overcome some principal challenges confronted by previous clinical trials. Whilst various rodent models are discussed, the optimal animal model is yet to be defined.

Introduction

More than 200 million people worldwide have peripheral artery disease (PAD). PAD dramatically affects quality of life and is associated with significant morbidity and mortality with an age-adjusted death rate of 17.0 per 100,000.^1,2 Standard treatment for severe cases of PAD is surgical or endovascular revascularization. Although open surgery has been the standard of care, endovascular surgery is an increasingly accepted means of treatment and recent advances in technology and technique have allowed endovascular surgery to be applied to an ever-increasing number of patients.³ Preclinical studies in large animal models and clinical studies have served well in the evaluation of proof of concept studies, confirmation of safety and characterization of performance of these endovascular devices.

Unfortunately, revascularization is often of limited benefit, as bypass grafts can have high failure rate that are worse in some situations (smokers, prosthetic grafts), while endovascular approaches are compromised by high restenosis rates. Moreover, up to 30% of patients are not candidates for either open or endovascular procedures, due to high operative risk or unfavorable vascular involvement. Over the years, there have been numerous clinical investigations that have sought to stimulate angiogenesis and improve outcomes in PAD patients.⁴ These trials were initially driven by promising preclinical studies conducted in small animal models, that were primarily focused on the impact of intramuscular or intraarterial administration of pro-angiogenic growth factors on regulating perfusion recovery and blood vessel “growth” after hindlimb ischemia.^5–10 Unfortunately, these clinical investigations were largely met with translational failure – as individuals with PAD did not experience the therapeutic angiogenesis or improved outcomes that were expected of such treatments. Cell therapies additionally emerged as theoretically superior alternatives to growth factor therapies given their capacity to produce broader arrays of cytokines and exert paracrine angiogenic effects, but these too proved to be ineffective.^11,12 These deficiencies may be attributed to the lack of concordance in pathology between PAD patients and the animals in whom the preclinical studies were conducted.^13,14 In fact, most trials were conducted in young and healthy animals without either arteriosclerosis or notable cardiovascular risk factors. Conversely, PAD patients are often older individuals with numerous comorbidities; these characteristics can independently exert negative effects on vascular function and tissue reparative mechanisms. New therapeutic approaches still need to be developed, but such new approaches require development of animal models that closely mimic the population of interest and focus on more clinically-relevant outcomes. Here, the focus is on rodent models, since small animal models are suitable for mimicking prevention of progression of PAD to critical limb threatening ischemia (CLTI), or medical treatment of CLTI, or both. Large animal models used for regulatory development of interventional strategies such as endovascular devices need to be discussed elsewhere.

Available methods for the induction of hindlimb skeletal muscle ischemia

Most limb ischemia animal models ignored the arterial pathology and are based on the induction of hindlimb ischemia by surgical methods in normal arteries. However, there are huge variations in the methods employed to induce hindlimb ischemia as well as the choice of animal species. Mice, rats, rabbits, pigs, and sheep have been used, but the most commonly used experimental animals for limb ischemia research are rodents, because mice and rats are easily available and modulation of genetic and relevant risk factors is relatively easy to implement.^15–17

The approach used to induce ischemia determines how closely the hindlimb ischemia model represents the condition it is simulating. Arterial ligation is one method: an incision is made at the origin of the inguinal crease and the femoral artery and vein are separated by blunt dissection. The femoral artery is then ligated with a single ligature just distal to the origin of the profunda femoris. However, this method leaves most of the collateral circulation to the lower limb intact and consequently blood flow to the limb is fully restored within 7 days.¹⁸ Accordingly, this model only produces a mild ischemia that is akin to intermittent claudication. Alternatively, a second ligature can be placed below the first and the intervening segment of femoral artery can be excised to produce more severe ischemia. Double ligation and excision of a segment of femoral artery in this manner removes the collateral bed, which means that angiogenesis and arteriogenesis are required to restore blood flow. Under these conditions only one-third of the original blood flow is restored after 7 days. This model has been used primarily to study the angiogenic and arteriogenic responses in peripheral limb muscle tissue and the effects of various therapies such as drugs, gene delivery, or stem cells on these responses.¹⁹ Concurrent femoral artery excision and ligation of the profunda femoris and circumflex branches increases the severity of ischemia even further resulting in rapid necrosis and auto-amputation of the limb, and is therefore too severe to use as a representative model for the study of CLTI.¹⁸ Although many variations of these models have been described (single, double ligation, plus or minus collaterals ligation, plus or minus collaterals excision), acquiring an optimal model that accurately ressembles human pathology remains a significant challenge, since all of these models are based on a 1-stage procedure.

Another important weakness of these models has been that they are acute limb ischemia models, whereas PAD in humans generally occurs as a result of a chronic process associated with the building of atherosclerotic plaques over many years leading to arterial stenosis. In an attempt to circumvent this weakness, models of gradual arterial occlusion were developed. Sequential arterial ligation consists of femoral artery and collateral bed ligation, followed by iliac artery ligation four days later.²⁰ In an analogous approach, ameroid constrictors have been used. These devices have an outer metal sleeve encasing an inner layer of hygroscopic material, usually casein, which, when they are placed around an artery, induce gradual vessel occlusion as they absorb moisture from the surrounding tissues²¹. These models, inducing a more gradual arterial occlusion with nadir of blood flow typically occurring over 3–5 days and sustained over weeks, might provide a model more relevant to human PAD (Table 1).^21,22 A 2-stage procedure of initial gradual femoral artery occlusion by ameoid constrictors for 14 days and subsequent excision of the femoral artery has been developed.²¹ Hind limb ischemia was more severe in mice receiving the 2-stage compared to the 1-stage ischemia induction procedure. This 2-stage model also exhibited a marked decreased in collaterals and in the expression of angiogenic mediators (e.g. VEGF, VEGF-R1, VEGF-R2) in the ischemic skeletal muscles. Accordingly, a 2-stage model seems more relevant to PAD in humans, but it requires further validation, with replication by others. A 2-stage model may increase animal attrition rates (which are seldom reported). Further research is needed before they become the main preclinical model employed in PAD research.

Table 1.

Methods TO induce hind limb ischemia in rodents

Characteristic	Femoral artery ligation18	Arterial excision and removal of collateral bed19	Sequential arterial ligation20	Ameroid constriction22	Ameroid constriction and arterial excision21
Mimics human pathology	−	−	+	++	++
Mimics human disease progression	−	−	++	++	+++
Mimics human disease chronicity	−	+	+	++	+++
Responsive to age	++	+	+	NR	NR
Responsive to hypercholesterolemia	++	+	+	+	+
Responsive to diabetes	++	+	NR	+	NR
Responsive to hypertension	++	+	NR	NR	NR
Responsive to obesity	+	+	NR	NR	NR
Responsive to nicotine exposure	+	+	NR	NR	NR
Repeatability & consistency	+	−	+	+	+
Dependence on sex	+	−	NR	NR	NR
Dependence on genetic background	+++	+	NR	+	NR
Attrition of animals	−	−	+	++	++
Learning curve	−	−	++	++	++
Suitable for local & systemic therapeutics	++	+	++	++	++
Optimal focus / application	Angiogenesis Cell and molecular therapy	Angiogenesis Cell and molecular therapy	Functional parameters PAD associated myopathy	Functional parameters Cell and molecular therapy	Functional parameters Exercise therapy

⁻no,

⁺yes,

⁺⁺moderately representative of this;

⁺⁺⁺strongly representative of this, NR not reported.

Endpoints and measurements

To simulate the wide variation in clinical manifestation of PAD, a combination of endpoints should be sought using animal models including assessments of limb perfusion, limb function, tissue injury, and detailed histological analyses of the arteries as well as the ischemic muscle and tissue niche.

Blood perfusion of the hindlimb is commonly assessed using Laser Doppler Perfusion Imaging, a reasonable technique for quantitative blood flow assessment since it enables real-time assessment of the microcirculation throughout the time frame imaged, as compared to the un-operated contralateral hindlimb, considered as control. It is non-invasive and measurements are reproducible.¹⁵ Laser light, directed by the Laser Doppler Perfusion Imaging device at the tissue under study is scattered by moving blood within the tissue causing a change in the wavelength and frequency of the light that is reflected back, which is then used to calculate the distribution of velocity of blood cells in the tissue. Laser Doppler Perfusion Imaging is, however, a measure of blood flux rather than a measure of perfusion; thus, valid comparisons of measurements taken at different time points relies on ensuring that ambient conditions such as temperature and activity of the animal immediately prior to taking the measurement have been standardized, as Laser Doppler Perfusion Imaging is affected by vasomotor tone. Other imaging modalities such as intra-arterial injections of radio- or fluorescently-labelled microspheres, transcutaneous oxygen pressure, and laser speckle flowmetry have been trialed to study muscle oxygenation; however, these techniques can be challenging.²³ Dynamic contrast-enhanced magnetic resonance imaging, computer tomography (CT), or micro-CT have also been proposed for the 3D anatomic study of post-ischemia neo-vascularization, although a dedicated room is required for this equipment and in general is more expensive.²⁴

In vivo assessment of limb function has been evaluated using semiquantitative assessment of impaired use of the ischemic limb (3=dragging of foot, 2=no dragging but no plantar flexion, 1=plantar flexion and 0=flexing the toes to resist gentle traction on the tail). Ambulatory activity can also be assessed using an open field test, as well as treadmill tests.²¹ However, in small animal models, function still remains more difficult to assess than in humans since quadrupeds can manage ably on 3 functional limbs in many cases.

Semi-quantitative measurement of the ischemic damage was also proposed (0=no difference from the contralateral hindlimb, 1=mild discoloration, 2=moderate discoloration, 3= severe discoloration or subcutaneous tissue loss or necrosis, and 4=any amputation).^21,25 Histological muscular assessment to measure angiogenesis/arteriogenesis is the most common technique used to understand pathological changes within ischemic hindlimbs. Immunohistochemical methods are used to quantify capillary density as compared to contralateral limb, and angiogenesis is often assessed as the capillary to fiber ratio within the muscle.²⁶ Histological correlates from human studies should be explored for operative mechanisms in animal models. For example, emerging human histological studies of patients with chronic limb-threatening ischemia revealed paradoxically higher numbers of abnormally enlarged capillaries; in addition, these vessels were detached from skeletal myofibers with decreased capillary transit time and expressed markers of an endothelial-mesenchymal transition (Endo-MT) process, indicating impaired microvascular perfusion and function.²⁷ Identification of animal models recapitulating these findings would be invaluable for future investigations to uncover novel therapeutic interventions.

Accordingly, endpoint assessment of preclinical PAD models should involve a combination of blood perfusion assessment using Laser Doppler Perfusion Imaging, limb function and tissue injury scores, and histological muscular examination. The most clinically-relevant human PAD endpoints are improvements in symptoms (such as walking distance and pain) and quality of life. In models of hindlimb ischemia, experiments would likely focus only on functional improvement. While quality of life cannot be adequately quantified in animals, outcomes to evaluate limb blood flow, hindlimb ischemia and limb function should be used, which are both informative and more translatable to real patient outcomes.²⁸

Incorporation of modifiable and non-modifiable risk factors into animal models of peripheral artery disease

PAD patients generally are not young and healthy individuals, but rather people with multiple cardiovascular risk factors including age, hypercholesterolemia, diabetes, hypertension, obesity, and smoking. Moreover, stressing the arterial pathology underlying PAD is an important point to consider in animal models. In fact, in patients, the pathology stems from chronic progression of occlusive atherothrombosis, often with medial calcification, which restricts the supply of oxygen and other blood nutrients reaching the skeletal muscle. Mimicking this pathology in animals is a major challenge and demands getting the right animal background to study limb ischemia caused by diseased atherosclerotic arteries. The pathophysiologic changes associated with these variables exert independent effects on human endothelial health and vascular function – often leading to impaired angiogenesis secondary to increased endothelial cell dysfunction, as well as reduced capacity for revascularization and wound healing both at baseline and in response to intervention.^26,29 In this context, different animal models have aimed to integrate these disease states (e.g. dyslipidemia, diabetes, hypertension, smoking) to better characterize the abnormalities associated with some presentations and to discover novel targets for therapy.

Since increased circulating total cholesterol levels or increased cholesterol incorporated in low-density lipoproteins (LDL) and very low-density lipoproteins (VLDL) are major risk factors for atherosclerosis, most of the animal models that exhibit increased total cholesterol, LDL or VLDL cholesterol levels are considered as models of atherosclerosis (Table 2). Different mice models were therefore developed: Apo E deficient mice, LDL receptor deficient mice, apo E3-Leiden transgenic mice, mice lacking hepatic lipase, and human apo B100 transgenic mice.^30–36 Furthermore, breeding experiments have been carried out to generate additional mouse models of human dyslipidemia: cross breeding of human apo B₁₀₀ transgenic mice with LDL receptor deficient mice produced a highly susceptible strain with hypercholesterolemia and atherosclerosis.³⁰ These models have been demonstrated to induce a number of vascular abnormalities including impaired vasodilation, defective wound healing, and decreased angiogenesis.³⁵ Within preclinical in vivo investigations of PAD, these models have collectively shown that in response to hindlimb ischemia animals with hypercholesterolemia experience significant attenuation in blood flow recovery, capillary density, microvascular relaxation, and collateral vessel development when compared with normal controls.^36–38 Such defects have been attributed to several mechanisms including increased oxidative stress in endothelial cells due to reduced nitric oxide (NO) bioavailability, reduced pro-angiogenic growth factor responsiveness, and increased pro-inflammatory macrophage signaling.³⁹ However, a major limitation of atherosclerotic mice models is that the plaques usually manifest in the aorta but not in the femoro-popliteal segment, limiting the translational value of this models.

Table 2.

Examples of current atherosclerosis mouse models

Background	Features
Apo E deficient mice29	Spontaneous development of atherosclerosis Elevated levels of circulating VLDL particles
LDL receptor deficient mice30	Dietary cholesterol is required to develop hypercholesterolemia and atherosclerosis Elevated levels of LDL and VLDL particles
Apo E3-Leiden transgenic mice31	Dietary cholesterol is required to develop hypercholesterolemia and atherosclerosis Elevated levels of circulating cholesterol
Hepatic lipase deficient mice32	Elevated levels of plasma cholesterol, phospholipids and HDL cholesterol
Human apo B100 transgenic mice33	Elevated levels of LDL particles
Cross breeding of human apo B100 transgenic mice with LDL receptor deficient mice29	Spontaneous development of atherosclerosis Severe hypercholesterolemia

Similarly, diabetes has been known to disrupt many cellular processes required for effective revascularization – with decreased angiogenesis, collateral formation, and endothelial cell vasodilation identified as common culprits.^40,41 Within in vivo preclinical models, diabetic (db/db) mice and streptozotocin- or alloxan-induced hyperglycemic animals largely demonstrate poor blood flow and functional recovery, increased limb necrosis, and diminished vascular density following hindlimb ischemia when compared with non-diabetic counterparts.^35,42 One significant mechanism governing this pathogenesis does not necessarily appear to be deficiency of pro-angiogenic growth factors itself, but rather maladaptive changes in vascular endothelial growth factor (VEGF) and angiopoietin growth factor ligand/receptor expression and subsequent downstream signaling.^43,44 Additionally, endothelial oxidative stress emerges as a notable contributor to the impaired angiogenic response in pre-clinical diabetic models, as some studies reveal that introduction of potent Akt/eNOS activators such as liraglutide and apelin can significantly improve revascularization in these animals.^45,46

Both genetic and non-genetic mouse models of hypertension exist as well. Non-genetic models of hypertension include surgical induction of hypertension via renal artery constriction, high-salt diet induced hypertension, or drug-induced hypertension via angiotensin II administration. Alternatively, genetic models include cross-bred selected spontaneously hypertensive mice (BPH/2 mouse), transgenic mice, and mice “knocked out” for genes such as nitric oxide synthase, endothelin, natriuretic peptides, and components of the renin-angiotensin-aldosterone and sympathetic nervous system pathways. Hypertension-induced endothelial dysfunction results in microvascular rarefaction through capillary and arteriolar regression, which exacerbates hypertension through subsequent increase in total peripheral resistance.⁴⁷

Nicotine – the addictive component of tobacco and e-cigarette products – is known to impair angiogenesis under chronic exposure and has been studied in animal models by adding nicotine to drinking water or via subcutaneously-implanted pumps in rodents, before inducing hindlimb ischemia.⁴⁸ These experiments have demonstrated that chronic exposure to nicotine impaired cholinergic angiogenesis through downregulation of the vascular nicotinic acetylcholine receptor and attenuation of nicotine-induced VEGF release. Chronic nicotine exposure can also impair the ability of cell therapies to exert therapeutic benefits.⁴⁹

Future Directions - Filling the Experimental Model Gap for PAD

The current discordance between positive preclinical results and failure of subsequent human trials has been ascribed to several shortcomings including limited exploration of drug dosing and delivery strategies, and complexities pertaining to patient and endpoint selection. In particular, the key limitation relevant to our discussion is the lack of pathophysiological concordance between human patients with PAD and the animals in whom these preclinical models were conducted.^13,14 The initial growth factor and cell-based experiments which propelled human therapeutic angiogenesis trials were mostly conducted in young, healthy animals with healthy arteries and no notable cardiovascular risk factors. Conversely, PAD patients are often older individuals with widespread atherosclerosis and numerous comorbidities; these characteristics can independently exert negative effects on vascular function and health, as can be seen in recent animal models of PAD described above. Thus, it is ultimately unsurprising that patients in past trials showed markedly reduced responses to pro-angiogenic therapies when compared to what was anticipated. Accordingly, having an established model of hindlimb ischemia that alters function and puts the limb at risk before a proposed therapy is assessed in a preclinical randomized trial, as would be needed in patients, is mandatory. Endpoints measurements should then include limb perfusion, function, and healing.

Pre-clinical models that recapitulate the anatomic and pathological considerations observed in human subjects along the continuum of PAD will inform a stage-specific approach for potential therapeutic intervention. For example, modulating severity and duration of ischemia can markedly alter limb necrosis and collateralization. Future studies should consider both the animal susceptibility to atherosclerosis and complementary models that rely less on acute limb ischemia that leave collateral development intact, and more on alternative sub-acute ischemia models that phenocopy the more gradual occlusion of PAD over time with less associated limb skeletal muscle necrosis. Thus, while no perfect model mimicking PAD exists at this time, methods using either a two-stage sequential approach or application of ameroid constrictors are likely more suitable for creating the gradual arterial occlusion typically seen in human chronic limb and critical limb-threatening ischemia. These experiments should be carried out for a minimum of 14 days and optimally up to four weeks to adequately observe recovery of blood flow along with changes in angiogenesis, arteriogenesis, limb necrosis, and limb function. In addition, preclinical models that are able to incorporate multiple common risk factors observed in patients with PAD such as diabetes, hypercholesterolemia, hypertension, smoking, and older age will likely provide more stringent conditions for translational success. Lastly, future therapeutics and cell-based therapies, including bioengineering strategies to sustain therapeutic efficacy, should test such efforts in stringent in vivo conditions for survival and long-term function with broad readouts encompassing limb revascularization, skeletal muscle and tissue necrosis, inflammatory and metabolic responses, and both local and systemic organ impact.

Recommendations for preclinical PAD models

• A suitable PAD animal model should be as anatomically and physiologically similar to humans

• Animal models susceptible to atherosclerosis are important, since PAD occurs in atherosclerotic arteries

• The animal should be capable of modulation by specific human cardiovascular risk factors

• A two-stage sequential approach or methods using ameroid constrictors are more suitable for creating a gradual arterial occlusion

• The observation period following the onset of ischemia should have a duration of at least 14 days

• Endpoints measurements should include perfusion assessment, functional and tissue injury scores, and histological analysis.

Conclusions

Patients with PAD reflect a heterogenous population of individuals who are simultaneously burdened by multiple cardiovascular comorbidities that significantly impact their vascular microenvironments and interactions with pro- and anti-angiogenic mediators and other drugs. This resulting complexity and the failure of initial pre-clinical models and human trials to grasp these intricacies substantively contributed to the ineffectiveness of therapeutic angiogenesis in clinical studies thus far. To overcome such challenges, scientists must build upon their cellular and animal disease models of PAD which better mimic the pathological and functional characteristics of common human presentations of intermittent claudication and critical limb ischemia. When possible both sexes should be studied, and reported separately, to ensure thorough preclinical assessment. Only through such efforts can we begin to close the gap in “bench-to-bedside” treatments for this pathology.