Matsushita and Smith developed a rat model to study the effect of AXC on hind limb spasticity (19). In 1987, Le May et al. developed a rat model of ISCI (20) to replace the rabbit model. At that time, this rabbit model was widely used as a nonspecific representation of neuronal ischemic injury (29, 40–42). Soon after, Le May et al. made additional modifications to the rat model (43). They used the same incision to explore the thoracic aorta, but instead occluded the right and left subclavian arteries and the thoracic aorta for 10 min. Neurological deficits were recorded at different time intervals.

Coston et al. developed another rat model of ISCI using an endovascular approach (35). Specifically, a 2F Fogarty catheter was inserted through the femoral artery into the aorta and was inflated. Fifteen minutes of aortic occlusion caused permanent hind limb motor deficits in 75% of the operated rats. Marsala and Yaksh modified the Coston et al. model by inserting the 2F Fogarty catheter through the tail artery into the thoracic aorta until the catheter tip reached the level of the left subclavian artery (35).

Kanellopoulos et al. evaluated three different approaches to develop a new rat model, claiming that the previous rat models were not reproducible (44). They concluded that a reproducible model of ISCI was possible in rats by combining aortic arch occlusion with induced hypovolemia (44).

Models in which cross clamping of the abdominal aorta has been used to cause spinal cord ischemia have produced variable results. Usul et al. and Guler et al. reported significant neurological deficits in rats after 45 min of aortic occlusion (28, 31). In contrast, Carrillo et al. reported that sub-diaphragmatic AXC is not suitable for causing spinal cord ischemia in rats because this method does not increase blood and tissue concentrations of pyruvate, glucose, or lactate relative to a non-ischemic control population (45). Nouri et al. also concluded that abdominal AXC is not a reliable method to induce spinal cord ischemia in rats (46) because the clamp is placed distal to the origin of the artery of Adamkiewicz (ARM). In rats, the origin of the artery of Adamkiewicz lies slightly more rostral to segments of the thoracic spinal cord (relative to human). Consequently, blood supply to the lumbosacral spinal cord is largely unaffected (46).

Zivin and DeGirolami developed the first reproducible rabbit model of ISCI in 1980 (29). Their intention was to develop a central nervous system ischemia model to avoid variability and clinical complications associated with modeling infarction in the cerebral arteries. They attributed the reproducibility of their model to the uniformity of the spinal cord vasculature in rabbit models, i.e., the main source of blood in the caudal spinal cord is segmental from the abdominal aorta and can be interrupted by a single AXC (29). They later modified their technique of AXC by placing a snare ligature around the abdominal aorta, with the free end of the snare accessible outside the rabbit (40). In related studies, a subcostal incision was performed to expose the infra-renal aorta through a retroperitoneal approach, instead of a midline laparotomy. The use of an externalized snare ligature enabled investigators to cause repeated bouts of ischemia in the conscious animal (47–51).

Cheng et al. used a less invasive approach (52). A balloon catheter was introduced through the femoral artery and advanced into the abdominal aorta until it was just distal to the left renal artery. Occlusion of the aortic lumen was accomplished by inflation of the balloon. This approach has been applied in diverse rabbit models of SCI because it is simple, fast, and less invasive than ligating the aorta through a laparotomy (53–59).

Maeda et al. developed a new rabbit model of SCI (32). Instead of the classical technique of infra-renal AXC, they selectively occluded the five lumbar arteries (L3–L7), utilizing a flank incision. Their goal was to avoid the hemodynamic changes associated with AXC, which include proximal hypertension with subsequent increase in CSF pressure and distal hypotension with subsequent ischemia of the bowel and lower extremities. By selective occlusion of the lumbar arteries, blood flow in the aorta was maintained, but the blood supply to the lumbar spinal cord was still interrupted (Table 3).

The rabbit model with infra-renal AXC is widely used as a model of ISCI. Although the rabbit model is reproducible, owing this quality in part to the simplicity of the surgical approach and the low incidence of adverse reactions associated with the procedure, it does not mimic the consequence of AXC in humans. Indeed, infra-renal AXC in humans undergoing abdominal aortic aneurysm repair rarely causes paraplegia (60, 61).

Nylander et al. developed a canine model in which permanent occlusion of the infra-renal aorta was used to cause ISCI (30). The location of aortic ligation was chosen to avoid bowel infarction and kidney damage, i.e., complications that occur with occlusion of the aorta above the celiac artery or the superior mesenteric artery (64, 65).

Coles et al. described a temporary thoracic aorta occlusion model of dog ISCI (25). By clamping the proximal and distal thoracic aorta, as well as the origin of the left subclavian artery (LSA), it was discovered that collateral blood flow to the spinal cord also was interrupted, thereby increasing the reproducibility of ISCI and paraplegia (63). Coles used this model to evaluate the neuroprotective effects of regional spinal cord cooling and systemic injection of a putative free radical scavenger (63, 66). Subsequent experimental studies by other groups have shown that double aortic cross clamping is also a suitable approach (with minor variations) for causing ISCI in dogs (67–75). Other researchers used single AXC (via a small thoracotomy) just distal to the origin of the LSA, to induce spinal cord ischemia in dogs (21, 76–84). Similarly, Grossi et al. (85), Sugiyama et al. (86) and Shokoku et al. (87) determined the effects of ligating the LSA, right subclavian, and/or double AXC in their models. Kumagai et al. described a different model of canine ISCI (88). This research group was able to simulate TAAA surgery through extensive clamping of the aorta and every other visceral branch. Sugawara et al. (22) added proximal hypovolemia to their model. Specifically, a significant reduction in proximal aortic blood pressure (reduced to ∼80 mmHg) was created by withdrawing blood into a reservoir while simultaneously cross clamping the descending aorta. This technique caused substantial ischemia and loss of motor function.

Because of the extensive collateralization of blood supply to the canine aorta, it is difficult to cause prolonged and consistent ischemia. To overcome this innate property of the canine blood supply, a number of researchers evaluated the effects of ligating lumbar and intercostal segmental arteries. By interrupting the segmental blood supply to 3–5 consecutive segments of the spinal cord (without AXC), Kato et al. (9), Fujimaki et al. (10), and Zhang et al. (89) created reproducible ISCI models. Awad et al. used AXC in addition to permanent interruption of the segmental blood supply to produce irreversible paralysis (90).

Endovascular occlusion techniques have been used to induce ISCI in dogs. Both Kaplan et al. and Ishizaki et al. used an endo-aortic occlusion approach to produce spinal ischemia in dogs (38, 91). In both cases, a balloon catheter was introduced through the left common carotid artery into the thoracic aorta, then advanced caudally to the lower abdominal aorta. The celiac and superior mesenteric arteries were ligated to eliminate collateral blood flow to the spinal cord. Graded ISCI was caused by stepwise alternation of the level of aortic occlusion via drawing the inflated balloon in the cephalic direction (91).

Overall, because dogs have extensive collateral blood supply to the spinal cord, their spinal cord neurons may escape ischemic injury after single level AXC. Additional ligation of segmental arteries (90), with or without proximal hypovolemia (22), may be required to consistently develop robust spinal cord ischemia in dogs.

Wadouh et al. used both endovascular and open approaches to induce spinal cord ischemia in the pig. From their research, the anatomical location of the ARM and the efficacy of the collateral circulation were defined as critical determinants of ischemic spinal cord injury during aortic occlusion (94, 97, 98). They also studied hemodynamic changes in the spinal and aortic collateral circulation after single and double AXC (95, 99, 100).

Like smaller species, the relative simplicity of the single AXC approach has been used by many laboratories to induce spinal cord ischemia in pigs (101–107). The proximal thoracic aorta is usually exposed through a small thoracotomy that is cross clamped distal to the LSA. Other researchers used the double cross clamping approach as a more reliable method to induce ischemia in pig models, and to mimic the surgical technique used during aortic aneurysm surgery (108–110). For double AXC, the aorta is exposed through two thoracotomies in the fourth and seventh intercostal spaces, then is cross clamped just distal to the LSA above the diaphragm (109, 110).