**4. Discussion**

The mouse hind limb ischemia model is the most used model for basic research and for pre-clinical studies investigating therapeutic neovascularization. It is considered a proper model, as it is reproducible, and strongly mimics the specific features of human peripheral arterial disease.

To perform this model, occlusions at di fferent anatomical levels of the iliac and femoral artery are used, as well as di fferent technical approaches. Based on the knowledge that arteriogenesis can only occur after increasing shear stress in the pre-existing arterioles proximal to the occlusion, the correct level of ligation of the femoral artery is distal to the origin of the collateral branches. However, in methods in which the femoral artery is ligated proximal to the origin of the collateral branches, arteriogenesis is also observed. Kochi et al. demonstrated that the number of collaterals is higher than previously thought. This may explain the unexpected arteriogenesis after ligation of the femoral artery at the proximal end. Kochi et al. also demonstrated the presence of collateral arteries in other muscle groups in the upper thigh besides the adductor muscle group. Furthermore, recent MRI research demonstrated that the collateral arteries in the quadriceps muscles are better developed than those in the adductor muscle group. This suggests that the analysis of the other muscle groups may yield valuable information on arteriogenesis, and should be included in experimental studies.

The hind limb ischemia mouse model has limitations. One of the major limitations is the acute nature of the ischemia induced in this model, while the PAD in patients is a chronic process, and critical limb ischemia arises slowly as a result of a gradual build-up of atherosclerosis. Yang et al. [28] and Padgett et al. [6] used ameroid constrictors to progressively induce hind limb ischemia and mimic the gradual occlusion that occurs in patients. However, this method is not ye<sup>t</sup> representative of the cellular response in the thigh after occlusion in PAD patients; therefore, further optimization of the method is needed.

Another limitation related to the hind limb ischemia model is that the procedure is predominately performed in young and healthy mice, which do not reflect patients with PAD. PAD patients are old and have co-morbidities like diabetes mellitus, hypercholesterolaemia, and hypertension. Westvik et al. showed that old mice show a slower perfusion recovery after inducing hind limb ischemia compared to young mice [26]. Young wild-type mice do not show the comorbidities of PAD patients. It is well known that these comorbidities accelerate the development of atherosclerosis and a ffect vascular remodeling [12]. The extent to which these comorbidities a ffect vascular remodeling varies. Van Weel et al. investigated arteriogenesis using a hind limb ischemia model in di fferent mouse types and showed that hypercholesterolaemia is more than hyperglycemia or hyperinsulinemia associated with impaired arteriogenesis [12]. Hypercholesterolaemia leads to impaired arteriogenesis due elevated blood cholesterol levels which a ffect monocyte chemotaxis, whereby monocyte influx is reduced [42]. These data sugges<sup>t</sup> that the changed lipid metabolism in diabetes patients may a ffect the arteriogenesis more than a disturbed glucose metabolism. Moreover, diabetes mellitus causes endothelial dysfunction which is multifactorial and results in reduced angiogenesis [43].

Hypertension activates angiotensin, which is associated with endothelial dysfunction due increased oxidative stress. However, Angiotensin activation can initiate and stimulate arteriogenesis due to inflammation regulation. Hypertension can also stimulate arteriogenesis through the increase of shear stress and activation of the renin-angiotensin system [42].

The mouse hind limb ischemia model and PAD patients frequently show similar neovascularization patterns with arteriogenic responses that are close to the occlusions to form collaterals arteries and an angiogenic response in the distal ischemic tissue, as was illustrated by comparing the angiogenic response and VEGF expression in the muscle biopsies of CLI patients and mice after induction of HLI [31,44,45]. However, there may also be di fferences in neovascularization patterns observed. In the

mouse HLI model, neovascularization occurs in a standard, homogenous fashion, due to the fairly standard way of blocking the blood flow, i.e., by occluding the major arteries in the upper limb in an acute way. This may be a crucial di fference with the situation in PAD patients, where occlusion usually occurs gradually, but most importantly, it may occur at various locations in the vascular tree. It is obvious that an occlusion of one of the major arteries above the knee may have di fferent consequences on endogenous collateral formation and neovascularization than an occlusion below the knee. Therefore, the neovascularization required to restore the blood flow in PAD patients with critical limb ischemia will be quite heterogeneous in nature, whereas the mouse models usually focus on a more standardized neovascularization induction. Unfortunately, mimicking the human situation is still complex, and really predictive (larger) animal models are not available yet, making the mouse model the most used model at present.

Furthermore, to choose an appropriate model, it is essential to consider which mouse strain to use, because the genetic background influences the outcomes. Several studies have shown high variability between mouse strains in restoring the limb perfusion after the induction of ischemia, and variability in ability of neovascularization [13,14]. Di fferent strains show di fferent blood recovery patterns after surgically-induced hind limb ischemia. Many studies have exhibited fast recovery of limb perfusion in C57BL/6 after inducing hind limb ischemia in comparison to the slow recovery of Balb/C mice. It has been suggested that the di fference in vascular remodeling is due a specific gene locus in chromosome 7 of the mouse [46,47]. Also, a wide variation in the extent of the native pre-existing collaterals is observed in di fferent mouse strains, whereby C57BL/6 and BALB/c demonstrate the largest di fference [40]. Due to their slow recovery, Balb/c mice are often used for hind limb ischemia studies, based on the assumption that this slow response better mimics the situation in patients with PAD. Recently Nossent et al. showed that especially in Balb/c mice, a stronger upregulation of pro-angiogenic and pro-arteriogenic genes is observed when compared to C57Bl6 mice, despite the poorer blood perfusion recovery in Balb/c [7]. This suggests Balb/c mice lack a thus far unknown factor that is crucial for vascular remodeling, rather than that this model better mimics the situation in patients with peripheral arterial disease.

Schmidt et al. recently demonstrated more muscle injury within 6 h after inducing ischemia in Balb/c mice compared to C57BL/6 mice [48]. The muscle injury may contribute to the ability of the vascular bed to recover after ischemia, and to regenerate. This suggests that understanding of etiology of the ischemic muscle and its contribution to the restoration of blood flow is needed, and that more research on this topic is required.

The techniques for the assessment of the results of the hind limb ischemia model are diverse. However, LDPI remains the essential readout method because of feasibility and reproducibility. The other techniques can serve as additional techniques since they mostly lack the capability to perform robust and fast analysis of the blood flow in larger series of mice.
