Histological Findings After Aortic Cross-Clamping in Preclinical Animal Models

Abstract

Spinal cord ischemic injury and paralysis are devastating complications after open surgical repair of thoracoabdominal aortic aneurysms. Preclinical models have been developed to simulate the clinical paradigm to better understand the neuropathophysiology and develop therapeutic treatment. Neuropathological findings in the preclinical models have not been comprehensively examined before. This systematic review studies the past 40 years of the histological findings after open surgical repair in preclinical models. Our main finding is that damage is predominantly in the grey matter of the spinal cord, although white matter damage in the spinal cord is also reported. Future research needs to examine the neuropathological findings in preclinical models after endovascular repair, a newer type of surgical repair used to treat aortic aneurysms.

INTRODUCTION

Spinal cord ischemic injury (SCII) is a serious complication of open repair (OR) surgery of thoracoabdominal aortic aneurysms (TAAA) and may result in motor and sensory deficits, paraparesis, and paraplegia (1). Numerous animal models of SCII have been used to develop potential therapeutic agents and surgical techniques to attenuate neurological injury. Unfortunately, most of these treatments have not been translated to clinical practice, and the rates of SCII following OR of TAAA are still considerable. There is a need for clinically feasible pharmacological treatments for SCII (2–7).

Consolidating and reviewing animal experimentation of SCII is useful to guide future studies. To our knowledge, a systematic review on histological outcomes of preclinical animal models of SCII has not been published. Our objective is to provide a useful resource that details histological outcomes of SCII models that simulate aortic cross-clamping during OR of TAAAs. Better experimental design and therapeutic development will emerge from an understanding of the neuropathophysiology of SCII.

The inability to translate effective therapies from animal models of spinal cord ischemia to treatments for patients undergoing TAAA repair is multifactorial. First, animal models may inadequately represent spinal cord ischemia in humans; this inadequacy is due to biological differences between humans and animal models and differences in spinal cord (SC) blood supply. Pharmaceuticals may demonstrate efficacy in animal models but may not be neuroprotective in the context of spinal cord ischemia in humans. Zones of ischemic damage in the SC due to arterial territories and collateral blood supply may also impede translation of biochemical mechanisms and pharmaceutical therapies from animal models to humans. Moreover, consistent histological methodology must be employed to gain a cohesive understanding of SCII. Employing perfusion fixation and appropriate histological staining techniques (elaborated on further in the Discussion) are essential for generating accurate histological outcomes. Accordingly, we believe that a summary of all animal models may better support translation of molecular mechanisms and pharmaceutical therapies from animal models to patients undergoing OR of TAAAs; noting patterns that are conserved across species may suggest mechanisms that contribute to SCII in humans.

MATERIALS AND METHODS

This systematic review conforms to guidelines established by the Systematic Review Center for Laboratory Animal Experimentation (SYRCLE) and the Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies (CAMARADES). Our protocol was registered on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PROSPERO).

Search Strategy

To develop a search strategy to obtain potentially relevant articles, we followed the PICO (Population, Intervention, Comparator, Outcome) model: our population of interest is all preclinical controlled animal studies; the intervention is cross-clamping across the descending aorta; comparison groups include naïve- and sham-operated animals; and our sole outcome of interest is a histological examination and findings in the spinal cord. From this model, we developed a query string to return all potentially relevant articles comprised of the following search components (SCs): (1) SC1: A string that returns all animal studies; (2) SC2: A string that references the aorta and cross-clamping (with synonyms); and (3) SC3: A string that references spinal cord ischemic injury (with synonyms). More detail is provided in Supplementary Data.

Screening Process

To conduct the screening process, we imported articles identified by our query string to covidence after which a title/abstract screen and a full-text screen was conducted. Title/abstract and full-text screening were accomplished using the inclusion/exclusion criteria provided in Table 1.

TABLE 1.

Complete Inclusion and Exclusion Criteria for Title/Abstract and Full-Text Screening Employed to Identify Articles Included in this Review

	Category	Inclusion Criteria	Exclusion Criteria	Rationale
Title/abstract screen	Population	Preclinical animal studies	Clinical studies, studies conducted in humans, ex vivo studies	This is our target population in which to study SCII
	Study design	Primary studies	Letters to the editor, presented abstracts, commentaries, reviews, chapters	Our goal is to look at primary literature
	Intervention	Obstruction of the aorta to simulate open surgical repair of aortic aneurysms by placing at least two clamps on the aorta	Applying only one clamp on the aorta, utilizing a balloon to occlude the aorta, utilizing ligatures to occlude the aorta, non-cross-clamping models, non-aortic aneurysm models	We believe that application of multiple cross-clamps best resembles OR of TAAA (hemodynamically and pathologically) while ensuring that our data are manageable yet not so narrow that important findings are missed
Full-text screen	Population	Preclinical animal studies	Clinical studies, studies conducted in humans, ex vivo studies	This is our target population in which to study SCII
	Study design	Primary studies	Letters to the editor, presented abstracts, commentaries, reviews, chapters	Our goal is to look at primary literature
	Co-intervention	Co-intervention study includes at least one experimental group in which co-intervention was not conducted	All experimental groups are subject to a potentially confounding co-intervention	Ensures that the study provides histological outcomes of animals undergoing SCII without confounding co-intervention
	Intervention	Obstruction of the aorta to simulate open surgical repair of aortic aneurysms by placing at least two clamps on the aorta	Applying only one clamp on the aorta, utilizing a balloon to occlude the aorta, utilizing ligatures to occlude the aorta, non-cross-clamping models, non-aortic aneurysm models	We believe that application of multiple cross-clamps best resembles OR of TAAA (hemodynamically and pathologically) while ensuring that our data is manageable yet not so narrow that important findings are missed
	Experimental control group	A naïve or sham-operated control or some comparator group	No comparator group	Controls are essential in ensuring that pathological findings are concomitants of SCII and not due to other factors
	Spinal cord ischemia	Demonstrable indication that the spinal cord was subjected to ischemic insult (histology, paraplegia, etc.)	No indication that the spinal cord was subject to ischemic insult	Other evidence of SCII suggests that pathological findings in the spinal cord are congruous with ischemia-induced injury and not due to other factors
	Outcomes	Histological investigation is conducted	Histological investigation is not conducted	Allows analysis of target data
	Other	All languages provided an English translation is available	No English version of the article is available

In order for an article to pass screening, it must be consistent with all of the inclusion/exclusion criteria detailed in Table 1. During the title/abstract and full-text screening phases, two independent reviewers assessed each article. Conflicts were first attempted to be resolved through discussion and consensus; if a consensus was not reached, a third reviewer cast the decisive vote. One important exclusion criterion was the use of non-cross-clamp methods (intra-aortic balloon, ligature, snares, chemicals, and single clamp) of inducing SCII because the cross-clamping best mimics the procedure performed in humans during OR. Whether other methods accurately mimic the exact pathophysiology of SCII after OR are unclear. To prevent bias, only articles that used aortic cross-clamping were included in this study.

All data extracted for the included articles are available in Supplementary Data Appendix SA. In total, 122 articles met inclusion/exclusion criteria and the screening process is shown in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram (Fig. 1). Information regarding data extraction, data analysis, and risk of bias is provided in Supplementary Data Table 1.

FIGURE 1.

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram depicting the number of articles obtained from searches, the number of articles excluded through screening, and the number of included articles.

Risk of Bias Assessment

Finally, each article was subject to a risk of bias (ROB) assessment using the SYRCLE ROB tool which is predicated on the Cochrane ROB tool. This ROB tool is a questionnaire consisting of 10 items designed to gauge selection bias, attrition bias, performance bias, detecting bias, reporting bias, and other biases and is provided by Hooijmans et al (8) (Supplementary Data Appendix B). One reviewer assessed the ROB utilizing the SYRCLE ROB tool for each article and a second reviewer checked over the ROB assessment to ensure accuracy. Conflicts between reviewers were first attempted to be resolved by a consensus; if a consensus was not achieved, a third reviewer made the final decision.

RESULTS

In the 122 reviewed articles, there were models in mice, rats, rabbits, dogs, pigs, baboons, and sheep. The majority of studies were performed in rabbit models followed by rat and mouse, whereas sheep and baboon studies were rare (Fig. 2 [9–129]).

FIGURE 2.

A graphical depiction of the number of included articles that utilize each animal model (9–129).

Experimentation in rabbits was most frequently reported (57 articles), while the sheep (1 article) and baboon (1 article) models were most infrequently reported. These data suggest a paucity of experimentation in large animal models and an abundance of experimentation in small animal models. We analyzed the histological observations using cross-clamp time and reperfusion time as parameters. Cross-clamp location and histological stain parameters were also analyzed but were not correlated to histological observations. Information regarding each parameter is provided in Supplementary Data Tables 2–9.

In general, the risk of bias (Supplementary Data Appendix B) across all articles was quite low. The most common source of bias was the lack of sham animals in 9% of the articles (40, 42, 45–48, 51, 64, 66–68). This was followed by detection bias caused by the lack of histopathologist blinding in 2% of articles (66, 96) and reporting bias caused by not outlining histology in the methods in 2% of the articles (73, 81). Analysis of selection bias (determining if allocation was concealed to the investigator), performance bias (random housing of animals and determining if caregivers/researchers were unaware of cross-clamped/sham), and detection bias (determining if animals were selected at random for histological outcome) revealed that almost no papers discussed these topics.

Sixty-eight percent of the articles verified a reduction of blood flow during and/or a resumption of blood flow after clamping the aorta (Supplementary Data Appendix C). Some papers used Doppler to measure at least a 90% decrease in blood flow in the femoral artery, some articles measured a distal blood pressure of <10 mmHg, while most did not provide a numerical value of decrease in pressure or flow. Finally, some groups verified aortic occlusion visually or by palpation. Eighteen percent of the articles measured only the changes in the systemic blood pressure (Supplementary Data Appendix C), and the remaining 14% of articles did not report on blood pressure or blood flow through the clamped aorta (Supplementary Data Appendix C). However, the efficacy of aortic occlusion in these articles was otherwise verified through neurological deficits or biochemical alterations in the SC.

Histological Methods

Sixty-nine percent of the articles employed only immersion fixation. Of the articles that used immersion fixation, 8% perfused with a non-fixative agent, such as saline or phosphate buffered saline before immersion. Seventeen percent of the total articles reported perfusion fixation. The remaining 14% articles did not provide the method of fixation (Supplementary Data Appendix C).

The various histological methods utilized in the articles are summarized in Table 2. Note, the categorization of other staining methods in Table 1 refers to Nissl stain, toluidine blue, Luxol blue, methylene blue, Klüver-Barrera, Bielschowsky method, glial fibrillary acidic protein (GFAP) stain, lead citrate uranyl acetate, osmium fixed Spurr’s resin, Cresyl violet, and Fluoro-Jade B (FJB) stain.

TABLE 2.

Number of Articles Based on Histological Stain

Histological Stain Based on Species
Staining Method	Mouse	Rat	Rabbit	Dog	Pig	Sheep	Baboon	Overall
H&E	15	19	53	5	9	1	1	103
IHC	8	13	12	0	1	0	0	34
TUNEL	3	4	7	0	0	0	0	14
Other	4	5	8	5	4	0	0	26

H&E, hematoxylin and eosin; IHC, immunohistochemistry; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling.

Hematoxylin and eosin (H&E) was the most commonly used histological stain in all animal models. Across all included articles, immunohistochemistry (IHC) was the second most commonly utilized histological stain; IHC was conducted for Caspase-3, Caspase-9, GFAP, CD31, Iba-1, PAS, TIMP-2, Collagen-I, Collagen-IV, MFSD2a, NeuN, CD11b, HuC/HuD, IL-6, myeloperoxidase, poly-(adenosine diphosphate ribose) polymerase (PARP), hypoxia-inducible factor-1a (HIF-1a), NFKB, HSP70, Bax, B-cell lymphoma-2 (Bcl-2), microtubule-associated protein-2 (MAP2), ubiquitin, GRP78, neuronal nitric oxide synthase (nNOS), endothelial nitric oxide synthase (eNOS), and inducible nitric oxide synthase (iNOS). Of these IHC stains, Caspase-3 (8 articles) IHC staining was most commonly conducted. In general, a diversity of staining techniques is conducted in small animal models, while larger animal models employ a smaller variety of histological techniques.

Histological Results

We categorized histological observations of SCII made by articles as grey matter (GM), white matter (WM) damage and changes, neuronal degeneration, necrosis, apoptosis, eosinophilia, pyknosis, loss of Nissl bodies, vacuolization, edema, changes to cytoplasm and nuclei, shrunken neurons, vascular changes, and inflammation and gliosis. These histological observations are summarized in Table 3. Of note, the article in the baboon model observed no histological abnormality postreperfusion (80).

TABLE 3.

Different Categories of Histological Observations

Most Common Histological Observations		122 Total Articles
Observation	Number of Articles	Percent of Total
Eosinophilia	23	19%
Pyknosis	18	15%
Loss of Nissl bodies	24	20%
Neuronal degeneration	83	68%
Inflammation and gliosis	48	39%
Vacuolization	40	33%
White matter damage and changes	33	27%
Edema	15	12%
Changes to cytoplasm/nuclei	15	12%
Shrunken neurons	17	14%
Necrosis	33	27%
Apoptosis	22	18%
Vascular changes	22	18%
Grey matter damage	118	97%

Grey Matter Pathology

Histological observations corresponding to the GM damage included references to eosinophilia in the GM, neuronal pyknosis, loss of Nissl bodies, neuronal degeneration, inflammation and gliosis in the GM, vacuolization in the GM, edema, shrunken neurons, and neuronal necrosis and apoptosis. A depiction of the number of articles that observed damage to the GM for each postreperfusion time and cross-clamp time is provided in Supplementary Data Tables 10–20.

Across all animal models (except the baboon), GM damage is much more frequently reported than WM damage. This contrast between the reporting of GM versus WM damage is most apparent in mouse and rabbit models, while the article that utilized a sheep model did not report any damage to the WM (103).

GM damage was reported in 100% of mouse, rabbit, and sheep articles. Ninety-five percent of rat articles reported GM damage; 92% of pig articles reported GM damage; and only 90% of dog articles reported GM damage. In comparison, the highest percentage of WM damage observations was 36% in the rat articles.

The earliest signs of GM damage in the mouse models were seen at 24 hours postsurgery, after cross-clamp times of 5, 7.5, and 11 minutes. These observations include areas of necrosis (16, 17), infiltration of necrotic areas by microglia (16, 17, 72), neuronal degeneration (64), and proliferation of astrocytes in necrotic areas (16, 17).

In articles involving rat models, the earliest observation of damage was at 8 hours postsurgery in a rat model that underwent cross-clamping for 45 minutes (120). That article reported TUNEL positivity in the lumbar SC, which is an indicator of neuronal apoptosis.

In articles involving rabbit models, the earliest signs of GM damage in the rabbit models were seen at 4 hours postsurgery, after cross-clamp times of 15 and 40 minutes. These observations include inflammation of the GM (55) and reduced number of motor neurons (49, 55).

In articles involving dog models, the earliest signs of GM damage were seen at 2 hours postsurgery, after cross-clamping for 40 minutes. Heavy argyrophilic interneurons were observed in Laminae VII (67). Pyknotic neurons were observed at 3 hours postsurgery and 60 minutes of cross-clamping (127).

In articles involving pig models, the earliest signs of GM damage were seen at 1 hour postsurgery after 60 minutes of cross-clamping. These observations include ischemia/reperfusion injury, edema, and vacuolization of neurons (113).

In the article involving the sheep model, infarction of GM was observed 168 hours postsurgery, after 50 minutes of cross-clamping (103). In contrast, no GM damage occurred in the baboon model after undergoing cross-clamping for 60 minutes. SC was recovered 48 hours postsurgery (80).

White Matter Pathology

Histological observations corresponding to WM pathology and changes included references to degenerative changes to the white matter, axonal degeneration and swelling, damage to the funiculi, loss of myelin, edema in the WM, infiltration by macrophages, and damage to the white tracts and spinal columns. The number of articles that reported damage to the WM for each postreperfusion time and cross-clamp time is provided in Supplementary Data Table 21.

In articles involving mouse models, WM damage was reported as early as 48 hours postsurgery in mouse models that underwent cross-clamping of 11 minutes. In mouse models, only 32% of the articles report WM changes or damage.

In articles involving rat models, WM damage and changes were observed in the form of interruptions in the myelin sheath as early as 24 hours postsurgery in a rat model that underwent cross-clamping for 30 minutes (37). Changes or damage in WM occurred in only 36% of articles.

In articles involving rabbit models, WM changes in the form of axonal swelling were reported as early as 24 hours postsurgery in rabbit models that underwent cross-clamping for 40 minutes (15). Only 28% of rabbit articles report WM changes or damage.

In articles involving dog models, WM damage and changes were reported in the form of axonal degeneration as early as 72 hours postsurgery in a dog model that underwent cross-clamping for 30 minutes (105). In dogs, only 10% of articles found WM changes or damage.

In articles involving pig models, WM damage and changes was reported in the form of degeneration of myelin as early as 1 hour postsurgery in a pig model that underwent cross-clamping for 60 minutes (113). In pig models, 17% of articles had WM changes or damage.

In articles involving sheep and baboon models, WM damage was not observed. The summary of the prevalence of GM and WM damage in every animal species is listed in Table 4.

TABLE 4.

Distribution of GM and WM Among the Animal Species

GM Damage			WM Damage
Animal Species	Number of Articles	Percent	Animal Species	Number of Articles	Percent
Mouse	19	100%	Mouse	6	32%
Rat	21	95%	Rat	8	36%
Rabbit	57	100%	Rabbit	16	28%
Dog	9	90%	Dog	1	10%
Pig	11	92%	Pig	2	17%
Sheep	1	100%	Sheep	0	0%

Axonal Degeneration

Histological observations corresponding to axonal degeneration included references to myelin damage, anterograde, and retrograde degeneration. Axonal degeneration was observed in mouse, rat, rabbit, dog, and pig models. In the mouse, anterograde and retrograde axonal degeneration and axonal spherocytosis were noted and are indicative of axonal degeneration (16, 61, 128). Lang-Lazdunski et al (61) cross-clamped the aortic arch, left subclavian artery, and the internal mammary artery for 11 minutes and found little axonal degeneration (using H&E), localized around the GM at 48 hours postreperfusion.

In the rat, the severity of WM damage is variable: cross-clamping the descending thoracic aorta, left subclavian artery, and the right subclavian artery for approximately 11 minutes resulted in rare axonal degeneration, which involved swollen axons (22); cross-clamping the aorta distal to the renal arteries and the aorta proximal to the aortic bifurcation for 30 minutes contributed to interruption in the myelin sheaths (37); and cross-clamping the aorta distal to the renal arteries and the aorta proximal to the aortic bifurcation for 45 minutes contributed to destruction of the entire white matter (110).

All articles that utilized rabbit models and reported axonal degeneration cross-clamped the aorta distal to the renal arteries and proximal to the aortic bifurcation, suggesting that location of aortic clamping plays a role in the SC damage. Observations utilized H&E and Cresyl-violet stains, and noted swollen axons, swollen axonal terminals, and vacuolar degeneration; these observations were made between 24 and 72 hours postreperfusion (15, 26, 28, 35, 44, 52, 53, 71, 90, 119). On the other hand, Herold et al (46) noted sparing of the myelin sheaths after cross-clamping the aorta distal to the renal arteries and the aorta proximal to the aortic bifurcation after 96 hours postreperfusion.

In the dog, observations indicative of axonal injury included swollen axons, myelin fragmentation, and spheroids (105). In the pig, Hellberg et al (1 hour postreperfusion, 60 minutes cross-clamp time, aorta distal to the subclavian artery and the aorta proximal to the aortic bifurcation, electron microscopy) noted separation of myelin lamellae and myelin degeneration (113), whereas Christiansson et al (1.5 hours postreperfusion, 30 minutes cross-clamp time, aorta distal to the subclavian artery and the aorta proximal to the aortic bifurcation, electron microscopy) noted that axon terminals were normal in appearance (104). These data once again suggest that observations of axonal sparring after OR are inconsistent among animal species.

Localization of the Damage Within the GM

Histological observations corresponding to damage to the anterior GM included references to damage to the anterior or ventral horn, to Rexed laminae VIII and IX, and to motor neurons. On the other hand, histological observations corresponding to damage to the intermediate and dorsal GM included references to damage to the lateral and dorsal GM, the central GM, and abnormalities in ependymal cells surrounding the central canal of the spinal cord. Provided in Figure 3 is a depiction of the number of articles of every indicated species that reported damage to the anterior GM and the dorsal and intermediate GM for species.

FIGURE 3.

Number of articles that observed anterior and intermediate/dorsal GM damage in included articles using mouse (A), rat (B), rabbit (C), dog (D), and pig models (E).

Among the articles that used mouse models, 16% did not specify the localization of GM damage. In the other animal models, the percentage of articles that did not specify the localization of GM is as follows: 27% in rat, 26% in rabbit, 50% in dog, and 42% in pig.

In the mouse, 14 articles observed damage to the anterior GM, whereas 2 articles noted damage to the intermediate/dorsal GM; in the rat, 10 articles observed damage to the anterior GM while 6 articles noted damage to the intermediate/dorsal GM; in the rabbit, 35 articles observed damage to the anterior GM while 7 articles noted damage to the intermediate/dorsal GM; in the dog, 4 articles observed damage to the anterior GM while 1 article noted damage to the intermediate/dorsal GM; in the pig, 4 articles observed damage to the anterior GM while 3 articles noted damage to the intermediate/dorsal GM. Across all species (except the baboon), articles more frequently noticed damage to the anterior GM compared with damage to the intermediate and dorsal GM.

In the mouse, damage to the anterior horn and dorsal horn was observed as early as 24 hours postsurgery in mice that underwent cross-clamping for 7.5 minutes. These observations included necrosis, infiltration, and morphological changes of microglia and astrocytes in the anterior horn (16). Similar observations were made in the dorsal horn (17).

In the rat, damage to the anterior horn was observed as early as 24 hours postsurgery in rats that underwent cross-clamping for 14, 20, and 30 minutes (31, 37, 57, 58). These observations include increased TUNEL positivity, increased Bax, HSP70, and NFKB immunoreactivity, neuronal degeneration, grey mater necrosis, and reduced number of motor neurons. Damage to the dorsal horn in the rat model was also observed as early as 24 hours postsurgery in rats that underwent cross-clamping for 14 minutes. These observations include increased TUNEL positivity and increased NFKB immunoreactivity (57, 58, 60).

In the rabbit, damage to the anterior horn was observed as early as 4 hours postsurgery in rabbits that underwent cross-clamping for 15 minutes. A reduction in the number of motor neurons was observed (49). Increased GRP78 and caspase-12 immunoreactivity was also observed 8 hours postsurgery in rabbits that underwent cross-clamping for 15 minutes (122). Damage to the dorsal horn in the rabbit was observed as early as 24 hours postsurgery in rats that underwent cross-clamping for 30 minutes. Vacuolar degeneration of the dorsal horn was observed (101).

In the dog, damage to the anterior horn was observed as early as 24 hours postsurgery in dogs that underwent cross-clamping for 30, 45, and 60 minutes. These observations included ischemic changes, cell degeneration, and reduced number of normal motor neurons in the anterior horn (40, 42, 129). Damage to the dorsal horn was observed as early as 2 hours postsurgery in dogs that underwent cross-clamping for 40 minutes. Neurons with heavy argyrophilic somata were observed (67). In articles involving dog models, anterior GM damage was not observed through TUNEL staining or IHC staining.

In the pig, damage to the anterior and dorsal horn was observed as early as 1 hour postsurgery in pigs that underwent cross-clamping for 30 and 60 minutes. These observations include shrinkage of motor neurons, ischemia/reperfusion injury, edema, and vacuolization (81, 113).

Anterior GM damage was not specifically noted in the article involving the sheep model.

Damage to the Thoracic and Lumbosacral Segments of the SC

Histological observations corresponding to damage to the thoracic and lumbosacral segments of the SC were identified in mouse, rat, rabbit, dog, pig, and sheep models. Reporting of the SC level where the neuropathology was observed was infrequent for all animal models, with the exception of articles dealing with dog models. Fifty percent of articles dealing with dog models reported the SC level where grey mater damage was observed, while only 21% of mouse model, 14% of rat model, 5% of rabbit model, and 25% of pig model articles reported the location. The sheep model article also reported the SC level where damage was observed. The number of articles that observed damage to the thoracic and lumbosacral regions is provided in Table 5.

TABLE 5.

Observations of Damage to the Thoracic and Lumbar Regions

Damage to Thoracic and Lumbar SC Regions
	Mouse	Rat	Rabbit	Dog	Pig	Sheep
Thoracic	2	0	2	1	1	1
Lumbar	2	3	1	4	2	1

In the mouse, studies noted that the ischemic lesion extended to the thoracic and lumbar spinal cord (17) and degenerative changes to neurons and axons were present in the thoracolumbar cord (128). Casey et al (128) describe sparing of the dorsal nuclei and the posterior column in the thoracolumbar segments. The earliest damage to the thoracic and lumbar spinal cord was observed as early as 12 hours postsurgery in mice that underwent cross-clamping for 7.5 minutes. Spinal cord damage was not observed above T6 level (17).

In the rat, neurodegenerative changes were noted in the lumbar spinal cord (110, 120) along with GRP78, caspase-3, caspase-9, Bax, and bcl-2 immunopositivity (120, 125). The earliest damage to the lumbar spinal cord was observed as early as 8 hours postsurgery in rats that underwent cross-clamping for 45 minutes. These observations include GRP78 immunopositivity and TUNEL positivity (120). Interestingly, GRP78 immunopositivity was decreased in spinal cords obtained 24 hours postsurgery.

Neurodegenerative changes, ischemic infarct, inflammatory infiltrates, vacuolization, edema, and vascular alterations were all noted in the rabbit thoracic and lumbosacral spinal cord (15, 24, 83, 84, 90, 92, 101, 108, 109, 116, 121, 122). Damage to the thoracic and lumbar spinal cord was observed as early as 24 hours postsurgery in rabbits that underwent cross-clamping for 30 and 40 minutes. These observations include destruction of motor neurons in the anterior horn of the thoracic spinal cord (15), vacuolar degeneration of the anterior and posterior horn in the lumber segments, neuronal degeneration, perineural edema, swollen ependymal cells with euchromatic nuclei, proliferation of glial cells (101), necrosis, marked proteolysis, and over 50 percent motor neuron death (121). Surprisingly, some saw increased Nissl bodies in the damaged lumbar segment (101).

In the dog, neurodegenerative changes, inflammatory infiltrates, vascular hyperplasia, hemorrhage, and astrocytosis were all observed in the thoracic and lumbosacral spinal cord (94, 106, 117, 129). Damage to the thoracic spinal cord was observed as early as 72 hours postsurgery in dogs that underwent cross-clamping for 45 minutes. Severe necrosis was reported (106). In the lumbar spinal cord, damage was reported as early as 24 hours postsurgery in dogs that underwent cross-clamping for 45 minutes. These observations include ischemic changes, shrunken neurons, pyknotic neurons, neurons with vacuolated and eosinophilic cytoplasm, astrocytosis, and mild spongiosis (129). Interestingly, some neurons were reported to contain hyperbasophilic cytoplasm.

In the pig, ischemic infarction, axonal degeneration, edema, vascular changes, and iNOS immunoreactivity were observed in the thoracic and lumbosacral cord (107, 113, 115, 126). Damage to the thoracic spinal cord was observed as early as 1 hour postsurgery in pigs that underwent cross-clamping for 60 minutes. Ischemia/reperfusion injury was observed (113). In the lumbar spinal cord, damage was observed as early as 48 hours postsurgery in pigs that underwent cross-clamping for 45 minutes. These observations include severe necrosis, edema, vacuolization, gliosis, and iNOS positivity in the lumbar spinal cord (115). Swollen neurons without iNOS positivity were also reported. In the sheep, ischemic changes were most common in the high lumbar and low thoracic segments (103).

Endothelial Integrity Disruption and Angiogenesis

Histological observations corresponding to changes in endothelial integrity and vascularization included observations of hemorrhaging and capillary proliferation in the spinal cord tissue. Supplementary Data Table 22 provides information regarding the number of articles that observed vascular changes for each postreperfusion time and cross-clamp time.

Vascular changes in the rat include hemorrhaging, vascular congestion, and PARP immunoreactivity in the vascular endothelium (9, 10, 29, 37, 50, 56); in the rabbit, vascular and capillary proliferation are noted (11, 38, 44, 45, 84, 101); in the dog, multifocal hemorrhage and vascular hyperplasia are observed (105, 117, 129); and in the pig, bleeding, vascularization, petechial hemorrhaging, and perivascular inflammatory changes are observed (78, 79, 115, 123).

Categories of Grey Matter Damage

Histological observations corresponding to neurodegeneration, or GM damage, included references to neuronal vacuolization, pyknotic nuclei of neurons, cytoplasmic eosinophilia, necrotic and apoptotic changes, abnormal neuronal morphology, and chromatin condensation; certain staining methods have been found to offer insight into neurodegenerative changes, such as IHC (caspases, Bax, bcl-2, hsp70, MAP2, PARP), TUNEL, Nissl, and FJB staining. The number of articles that observed various forms of neurodegeneration and GM damage for each postreperfusion time and cross-clamp time is provided in Supplementary Data Tables 10–20.

Neuronal Death

Observations reporting neuronal death are ubiquitous across all animal models except the baboon. We defined neuronal death as any observation that mentions neuronal degeneration, neuronal death, ischemic changes to neurons, or reduced number of neurons. Earliest observations in the mouse and rat were made 24 hours postsurgery. The respective cross-clamp times were 10 minutes in the mouse and 14, 20, and 30 minutes in the rat (31, 37, 60, 64). In the rabbit, dog, and pig, neuronal death was reported at earlier times: 4 hours postsurgery in the rabbit, 2 hours postsurgery in the dog, and 1 hour postsurgery in the pig. The respective cross-clamp times were 15 and 40 minutes in the rabbit, 40 minutes in the dog, and 60 minutes in the pig (49, 55, 67, 113). In the sheep study, neuronal death was observed 168 hours postsurgery after the sheep were cross-clamped for 50 minutes (103).

Necrosis

Widespread necrosis is a common neurodegenerative change reported in mouse models (16, 17) and in the grey matter of rat models (9, 51, 57, 60). In the rabbit, necrosis of the global GM, central GM, the anterior horn, the lumbar cord, and motor neurons is observed (24, 45–47, 84, 88, 111, 121). The earliest postoperative reporting of GM necrosis in the mouse, rat, and rabbit occurred 24 hours postsurgery. The respective cross-clamp times were 7.5 minutes in the mouse, 20 minutes in the rat, and 40 minutes in the rabbit (15–17, 31, 121). In a dog model, Bonillo et al noted that the severity of necrosis varied between animals: 2 dogs displayed severe necrosis of the GM while 3 animals displayed mild to moderate necrosis (106). Observation of necrosis in the dog occurred 72 hours postsurgery after cross-clamping for 45 minutes. In the pig, necrosis of individual motor neurons was reported (100, 115), and Van Voorst et al (126) report that central cord necrosis and intermediate necrosis are present in paralyzed animals. The earliest reporting of necrosis in pigs occurred 48 hours postsurgery in pigs that underwent cross-clamping for 40 minutes (100, 115). Necrotic changes were not directly reported in the sheep, although infarction was noted (103).

Apoptosis

Apoptosis following SCII is reported in mouse, rat, and rabbit models and may contribute to neurological deficits. Ni et al observed apoptosis of motor neurons in the mouse (73) and neuronal apoptosis is also reported in the rat (19, 110). Iwamoto et al reported anterior horn neuronal apoptosis through the TUNEL assay and early indications of apoptosis were observed by Tetik et al through the citrate-uranyl acetate method (116, 124); on the other hand, Takahasi et al reported no evidence of apoptosis through electron microscopy and Hamaisi et al utilized the TUNEL assay to note that anterior horn motor neurons undergo necrosis but not apoptosis (88, 111). It is interesting to note that all 4 rabbit models that comment on apoptosis utilized similar cross-clamping methods; however, articles that noted apoptosis occlude the blood supply for 30 minutes whereas studies that claim a paucity of evidence suggesting apoptosis apply the cross-clamp for 15–20 minutes. Apoptosis was observed as early as 48 hours in the mouse, 8 hours in the rat, and 24 hours in the rabbit. The corresponding cross-clamp times were 4 and 5 minutes for the mouse, 45 minutes for the rat, and 15 and 21 minutes for the rabbit (12, 69, 72, 95, 97, 120).

Shrunken Neurons

Several abnormal morphological characteristics of neurons postischemic insult are indicative of neurodegenerative processes. Shrunken neurons or shrunken neuronal organelles were observed in mouse, rat, rabbit, dog, and pig models (10, 11, 15, 20, 22, 24, 31, 40, 41, 47, 48, 57, 100, 105, 114, 117, 122, 129). Shrunken neurons were observed as early as 24 hours postsurgery in the rat, rabbit, and dog. The respective cross-clamp times were 20 minutes for the rat, and 30 and 40 minutes for the rabbit, and 30 and 45 minutes for the dog (15, 31, 40, 48, 129). In the mouse, shrunken neurons were observed at a later time of 60 hours postsurgery in mice that underwent cross-clamping for 4 minutes (20). In the pig, shrunken neurons were observed as early as 1 hour postsurgery in pigs that underwent cross-clamping for 30 minutes (81).

Cytoplasmic Eosinophilia

Cytoplasmic eosinophilia is a characteristic of cells undergoing necrosis. Cytoplasmic eosinophilia was observed in mouse, rat, rabbit, and dog models (24, 27, 32, 36, 37, 39, 40, 43–45, 48, 49, 51, 60, 61, 70, 71, 74–76, 89, 102, 109, 111, 114, 129). Cytoplasmic eosinophilia was observed as early as 24 hours postsurgery in the rat, rabbit, dog, and pig, whereas it was reported as early as 48 hours postsurgery in the mouse. The corresponding cross-clamp times were 11 minutes in the mouse, and 30 minutes in the rat, rabbit, dog, and pig (37, 40, 48, 61, 66).

Nuclear Pyknosis

Nuclear pyknosis signifies degradation of DNA due to either programmed or unprogrammed cell death. Nuclear pyknosis was observed in mouse, rat, rabbit, and dog models (14, 19, 25, 33, 37, 40, 45, 49, 65, 85, 86, 89, 90, 110, 129). In the rat and rabbit, pyknosis was observed as early as 24 hours postsurgery, while in the mouse pyknosis was reported only as early as 60 hours postsurgery. The corresponding cross-clamp times were 4 and 5 minutes for the mouse, 30 and 45 minutes for the rat, and 20 minutes in the rabbit (19, 30, 33, 37, 86). In the dog and pig, pyknosis was reported earlier: 3 hours postsurgery in the dog and 1 hour postsurgery in the pig. The corresponding cross-clamp times were 60 minutes in the dog and 30 minutes in the pig (81, 127).

Changes to the Cytoplasm and Nucleus

Evidence of chromatin condensation was identified in the rat and mouse through H&E and Cresyl-violet staining (57, 77, 86, 110). While neurodegenerative processes were reported in nearly every article, survival of small motor neurons and intermediate neurons was observed by Guler et al (110); moreover, Lafci et al (56) observe that cross-clamped animals are similar to sham-operated animals in terms of neurodegeneration. Both studies utilized rat models, cross-clamped below the left renal artery and above the aortic bifurcation for 45 minutes and made the aforementioned observations 48 hours postreperfusion through H&E. Changes to the cytoplasm and nucleus were also reported in the rat, rabbit, and dog models. Loss of cytoplasmic features in neurons was reported in rats 24 postsurgery after undergoing cross-clamping for 30 minutes (37). Enlargement of the cytoplasm of neurons was reported in rabbits 24 hours postsurgery after undergoing cross-clamping for 30 minutes (48). Triangular nuclei were reported in neurons of dogs 24 hours postsurgery after undergoing cross-clamping for 30 minutes (40).

Edema

Edematous changes include parenchymal edema in the mouse, which was reported as early as 48 hours postsurgery after undergoing cross-clamping for 11 minutes (61). Perineural edema was noted in the rat as early as 24 hours postsurgery after undergoing cross-clamping for 30 minutes (37). In the rabbit, interstitial edema, edema around the perikaryon, perineural edema, and edema in the anterior horn and white matter were observed (68, 90, 92, 101). The earliest reporting of edema in the rabbit occurred 24 hours postsurgery after undergoing cross-clamping for 30 minutes (101). Lafci et al (56) observe that cross-clamped animals are similar to sham operated animals in terms of edema. In the dog and pig, edema was noted as early as 3 hours and 1 hour postsurgery, respectively (113, 127). Both animals underwent cross-clamping for 30 minutes.

Loss of Nissl Bodies

The loss or disappearance of Nissl bodies was not reported in the mouse. In the rat, the disappearance of Nissl bodies was observed as early as 48 hours postsurgery in rats that underwent 45 minutes of cross-clamping (10, 114). In the rabbit, the disappearance of the Nissl bodies was observed as early as 24 hours postsurgery in rabbits that underwent 30 minutes of cross-clamping (101). In the dog, the disappearance of the Nissl bodies was observed as early as 24 hours postsurgery in dogs that underwent 30 minutes of cross-clamping (40). In the pig, the disappearance of the Nissl bodies was observed as early as 24 hours postsurgery in pigs that underwent 30 minutes of cross-clamping (66).

Vacuolization

Vacuolization was observed, in the mouse, as early as 46 hours postsurgery in mice that underwent cross-clamping for 7.5 minutes (130). In the rat, vacuolization was observed as early as 24 hours in rats that underwent cross-clamping for 30 and 45 minutes (30, 37). In rabbits, vacuolization was observed as early as 24 hours postsurgery in rats that underwent cross-clamping for 21 and 30 minutes (12, 48). In the dog, vacuolization was observed as early as 24 hours postsurgery in dogs that underwent cross-clamping for 45 minutes (129). In the pig, vacuolization was observed as early as 1 hour after surgery in pigs that underwent cross-clamping for 60 minutes.

Inflammation and Gliosis

Inflammation and gliosis manifested histologically as the presence of inflammatory infiltrates; IHC for nitric oxide synthase, IL-6, NF-KB, HIF-1a, PAS, Iba-1, myeloperoxidase, CD11b, CD31, GFAP, and hsp70 are also markers of inflammatory responses. Inflammation and gliosis are observed in the mouse, rat, rabbit, pig, and dog. Reactive microglia, detected by IHC CD11b, CD31, and Iba-1, are reported in necrotic lesions in the GM and in WM tracts in mouse models (16, 17, 64, 72, 86); CD11b immunostaining also revealed reactive microglia in the intact GM (17). Through IHC CD31 and IHC Iba-1, Lobenwein et al report that microglial response decreases as a function of time up to 168 hours postreperfusion (64). Reactive astrocytes, detected through IHC GFAP are reported in necrotic lesions in the GM, WM tracts, and in the intact GM as early as 24 hours postsurgery in mice that underwent cross-clamping for 7.5 minutes (16, 17). Reactive microglia were detected using Iba-1 immunostaining as early as 24 hours postsurgery in mice that underwent cross-clamping for 5 minutes (72). Interestingly, no increase of reactive microglia was observed 48 hours postsurgery. Lang-Lazdunski et al (61) observe preservation astrocytes and oligodendrocyte 24 hours postreperfusion and 48 hours postreperfusion.

In the rat, macrophages, detected by Cresyl-violet, PAS and IHC Iba-1, infiltrated the lateral and ventral columns and the GM (22, 25, 57). Lang-Lazdunski et al (57) further detected the infiltration of polymorphonuclear monocytes and neutrophils. Dong et al (25) detected increased IHC GFAP immunoreactivity 48 hours postreperfusion. Inflammatory pathology was further corroborated by increased postinjury immunoreactivity of myeloperoxidase, IL-6, HIF-1a, caspase-3, NF-kB, and hsp70 (9, 30, 31, 56, 57). Inflammation and glial cell infiltration were observed as early as 24 hours postsurgery in rats that underwent cross-clamping for 45 and 60 minutes (29, 54, 56). Reactive microglia were also observed using Iba-1 immunostaining after 24 hours postsurgery in rats that underwent 60 minutes of cross-clamping (25).

In the rabbit, astrocytes, microglia, and neutrophils were implicated in inflammation following SCII (26–28, 32, 36, 41, 43, 46, 76, 119). The presence of infiltrates was detected in the GM, the ventral horn, WM, central GM, the anterior, dorsal, and later columns, and the parenchyma (12, 18, 41, 45, 46, 76, 111, 119). IHC caspase-3, NF-kB, nNOS, and eNOS further suggest and inflammatory pathology (11, 18, 28, 52, 53, 71). The earliest reporting of inflammation was 4 hours postsurgery in rabbits that underwent cross-clamping for 40 minutes (55).

Infiltration by macrophages and astrocytes were observed in the GM and WM in articles using dog models (67, 105, 106, 117, 129). In the dog, gliosis in Rexed laminae 4-7 was reported as early as 48 hours postsurgery in dogs that underwent cross-clamping for 40 minutes (67).

In the pig, inflammatory changes were reported as early as 2 hours postsurgery in pigs that underwent cross-clamping for 60 minutes (78, 79).

A summary of the earliest histological observations in every animal species is provided in Table 6. The number of articles that reported certain common histological observations is provided in Supplementary Data Table 23.

TABLE 6.

Earliest Histological Finding in the Spinal Cord of Different Species After Aortic Cross Clamping

	Earliest Reporting Post-Surgery (hours)
Observation	Mouse	Rat	Rabbit	Dog	Pig	Sheep
Eosinophilia	48	24	24	24	24	NA
Pyknosis	60	24	24	3	1	NA
Loss of Nissl bodies	NA	48	24	24	24	NA
Neuronal degeneration	24	24	4	2	1	168
Inflammation and gliosis	24	48	4	48	2	NA
Vacuolization	46	24	24	24	1	NA
White matter damage and changes	48	24	24	72	1	NA
Edema	48	24	24	3	1	NA
Changes to cytoplasm/nuclei	NA	24	24	24	NA	NA
Shrunken neurons	60	24	24	24	1	NA
Necrosis	24	24	24	72	48	168
Apoptosis	48	8	24	NA	NA	NA
Vascular changes	NA	24	24	3	0	NA
Grey matter damage	24	8	4	2	1	168
Damage to thoracic segment	24	NA	24	72	1	168
Damage to lumbar segment	24	8	24	24	48	168
Damage to anterior horn	24	24	4	24	1	NA
Damage to dorsal horn	24	24	24	2	1	NA

GM and WM Sparing

Selective sparing of the GM and WM was observed in articles that utilized mouse, rat, rabbit, and dog models. In the mouse and rat, both GM and WM sparing was noted, while in the rabbit and the dog, only WM sparing and only GM sparing was observed, respectively. A depiction of the number of articles that observed sparing of the GM and the WM for each postreperfusion time and cross-clamp time is provided in Supplementary Data Table 24.

DISCUSSION

Several preclinical animal models have been established to study the pathophysiological mechanisms of SCII following OR surgery. Each of these models offers unique advantages and disadvantages: ease of use, expense, biological similarity to humans, anatomical features, practical applicability, and pathophysiology/outcomes vary across different models. Previous work has been done to look into the technical considerations, anatomical features, functional and behavioral outcomes. Awad et al (131) comprehensively described the technical considerations as well as the clinical/behavioral outcomes across numerous animal models for SCII; Mazensky et al (132) offered insight into the anatomical features of preclinical SCII models by providing a detailed review on the arterial supply of the SC for various animal models; and Fouad et al and Zhang et al (133, 134) reviewed the functional and behavioral outcomes for different SCII models.

To our knowledge, no work has been done that performed a comprehensive analysis of the histological findings after aortic cross-clamping. Histological analysis of SCII models is particularly useful because the study of SC morphology and its cytoarchitecture offers evidence on the presence/absence of particular disease mechanisms, as well as helps guide future therapeutics.

We found that animal experimentation of SCII is largely conducted in small animal models, while large animal models are utilized less frequently. The paucity of studies that utilize dog, pig, sheep, baboon, and other large animal models may be attributed to higher cost of care and more complex experimental protocols (135). However, because large animal models typically better replicate human physiology, large animal experimentation is imperative in translatability of potential treatments from preclinical models to patients (135). Moreover, compared to small animal models, histological analysis of the dog and pig SCs are typically conducted at earlier postreperfusion times. While this approach may reduce the costs and difficulties of post-SCII care in large animal models, this impedes our understanding of the temporal progression of damage. The temporal progression is relevant in understanding how human pathology evolves after aortic cross-clamping surgery.

Mouse models of OR surgery for TAAA are useful in the study of biochemical mechanisms of SCII due to the ability to generate genetic knock-out strains (135). However, the pathophysiological response to traumatic SC injury in the rat may better approximate human SC injury, perhaps explaining the popularity of SCII models in the rat (136). It remains unclear how well rat SCII models compare to larger animal models.

In our set of included articles, the rabbit model of SCII is most common. The unique arterial supply to the rabbit SC is conducive to a simple and straightforward cross-clamping procedure that produces highly reproducible SC ischemia and concomitant neurological deficits, which explains the popularity of the rabbit model (137). However, the longer and more invasive surgical procedure in the rabbit model may result in confounding complications (137).

Collateral blood supply has a great effect on the histological outcome after aortic cross-clamping. Therefore, considering both cross-clamping method and collateral blood supply offers insight into the differential results following spinal ischemia in different animal species. The supply to the cervical SC is accomplished by vertebral arteries while the thoracolumbar SC is supplied by dorsal intercostal arteries and lumbar arteries (132). The ventral spinal artery supplies the central intrinsic system, which perfuses the ventral GM, the dorsal columns, part of the dorsal GM, and part of the ventral and lateral columns (132). The dorsal spinal arteries supply the peripheral intrinsic system, which perfuses the posterior dorsal GM and outer ventral and later columns (132). However, there is a difference between the collateral blood supply of widely used ischemic SCII animal models and the human blood supply.

The pig has larger internal thoracic arteries and subscapular arteries than the human, which provide extensive collateral blood flow to the lower body and offer blood supply to the SC through the collaterals (138). Moreover, the median sacral artery in the pig offers blood supply to the SC. On the other hand, the rabbit has no collateral branches to the SC. The main source of blood in the caudal portions is segmental supply from the abdominal aorta, which results in reproducible paralysis after infra-renal aortic occlusion (139).

In the rat, the ventral and the dorsal spinal arteries represent 2 independent systems of SC arterial supply, being more delineated than those in humans (140).

The blood supply to the thoracolumbar SC of the mouse is highly varied (141). Flesarova et al found a high variability in the occurrence of ventral and dorsal branches of the rami spinales, as well as irregular occurrence of segmental arteries in the lumbar SC of the mouse.

In dogs, Pais et al (142) found the artery of Adamkiewicz was present in only 50% of specimens. In contrast, a meta-analysis by Taterra et al (143) found that the artery of Adamkiewicz was present in almost 85% of human subjects. Moreover, Mezensky et al (132) found that the level of origin of the artery of Adamkiewicz varies among the animal species, although it usually originates from a lumbar artery.

Overall, there is no one best model for SCII due to the unique advantages and disadvantages that each poses. Large animal models have a vasculature that better resembles that of humans compared to small animal models. However, large animal models (dog and pig) have increased collateral blood flow to the spinal cord compared to humans. Results from small animal models are more reproducible (rabbit) and allow for genetic knock-outs to be made (mouse, rat). These characteristics of small animal models allow for easier elucidation of the mechanisms of SCII.

Histological analysis indicates GM damage in 97% of all reviewed articles and WM damage and changes in 27% of review articles, with GM damage being reflected in a variety of histological observations. Results regarding damage to the GM and WM are controversial; both GM and WM damage were noted across all animal models, yet selective sparing of the grey and white matter are reported in the mouse, rat, rabbit, and dog. Although it is tempting to conclude that GM pathology is the primary mechanism of neurological injury, the preponderance of GM injury compared to WM injury may be due to the focus on GM pathology following SCII. Moreover, studies in ischemic stroke in the central nervous system offer evidence for both grey and white matter damage (144). However, a difference in blood supply to the GM and WM may also explain predominance of GM pathology (145).

Several factors may contribute to differential results regarding sparing and damage of the GM and WM. Failure to adequately decrease perfusion in the ventral spinal artery, which supplies a majority of the GM, may produce GM sparing. An adequate decrease in aortic blood flow was verified by some way in 68% of the articles. The remaining did not report an adequate assessment of a reduction in aortic blood flow during clamping. However, every article demonstrated clamp efficacy by subsequent neurological and behavioral deficits. Additionally, sufficient perfusion in peripheral intrinsic system and the central intrinsic system may result in sparing of the dorsal, ventral, and lateral columns of the WM. Finally, improper handling of SC specimens may produce the dark neuron artifact, which may be mischaracterized as ischemia-induced injury (146). This risk is best reduced by perfusion fixation which was performed in 17% of the articles; however, the majority employed immersion fixation, which creates an increased risk for the dark neuron artifact.

Another factor contributing to the heterogeneity of the data is the histological techniques used for SC cytoarchitectural analysis. While a variety of histological methods are employed, H&E staining is the most popular. H&E staining is a versatile histological method and can be used to identify degenerative changes in neurons, the presence of astrocytes, the presence of oligodendrocytes, and the presence of normal and reactive microglia and other monocytes (147). However, caution should be exercised when drawing definitive conclusions regarding pathophysiological processes through H&E (147). While cytoplasmic eosinophilia, pyknosis, and loss of Nissl substance are indicative neuronal damage, these features may not be apparent in smaller neurons such as interneurons (147). Additionally, reporting a specific neuronal degenerative pathway (such as apoptosis) is difficult, as different mechanisms of neuronal death and injury may appear morphologically similar in H&E staining (147). These notions are supported by Jordan et al (146), who recommend the use of supplementary special stains to identify cellular injury, which may not manifest through H&E staining. Degeneration of these smaller neurons is more reliably noted through FJB and silver staining techniques (147). Furthermore, Garman recommends the use of GFAP immunostaining to corroborate an observation of gliosis predicated on H&E staining (147). Non-reactive microglia are also difficult to detect in H&E-stained sections and Iba-1 is a better marker of microglia (147). The heavy reliance on H&E staining for histopathological analysis and the lack of diverse staining methods, particularly in large animal models, may contribute to the conflicting results regarding SCII pathophysiology and treatments.

Differential histological outcomes are possibly due to the use of different histological staining techniques. H&E staining was used to report apoptosis in the mouse (73); however, as noted previously, additional staining techniques should be utilized to corroborate a diagnosis of apoptosis (147). Guler et al (110) reported the survival of small motor neurons and intermediate neurons through H&E staining in the rat, while Marsala et al (67) noted damage to interneurons in the dog; this discrepancy may be attributed to the use of H&E staining by Guler et al (147) or due to differences in the animal models and experimental protocols. While astrogliosis is not ubiquitously reported, Garman recommends GFAP staining to confirm the presence and proliferation of reactive astrocytes; it is possible that, in most models, astrogliosis did occur but was not reported because a diagnosis of astrogliosis through H&E staining may be inappropriate. Reactive microglia and inflammatory infiltrates were identified through immunostaining for CD11b, CD31, Iba-1, and H&E staining. While immunostaining is a more reliable marker for the presence of reactive microglia (146, 147), Garman explains that activated but not resting microglia are reliably identifiable through H&E staining (147). Axonal injury is identifiable through a variety of histological methods particularly through FJB (146, 147), and is reported in the mouse, rat, rabbit, and pig. By recognizing that neurodegeneration, inflammation, and white matter damage contribute to the pathophysiology of SCII, a failure to report these processes due to a particular staining method may hinder the elucidation of the disease mechanism.

The localization of GM injury is also controversial: while observations regarding anterior GM damage are most common, articles also note damage to the lateral, central, and dorsal GM. Moreover, selective sparing in these regions are also observed. Inadequate occlusion of the ventral spinal artery may result in sparing of the anterior, intermediate, and dorsal GM while inadequate occlusion of the dorsal spinal arteries may preserve the posterior aspect of the dorsal GM (132). Again, appropriate and diverse staining techniques must be employed to completely comment on the presence or absence of pathology in different regions of the GM. However, Krogh explained that increased damage to the anterior horn may be due to its less favorable blood supply (148).

Ischemic injury to the thoracic and lumbosacral regions of the SC is most likely a function cross-clamp location. According to Mazensky et al (132), occlusion of the thoracic aorta in the mouse, rat, rabbit, dog, and pig should result in hypoperfusion of the thoracic SC. Occlusion of the abdominal aorta in the mouse, rabbit, dog, and pig should result in hypoperfusion of the lumbar SC (132). Detailed information on the spinal vascular anatomy could not be found for sheep. However, histological observations from included articles did not fully align with clamping technique and damage to the lumbosacral cord was more frequently reported. Preferential damage to the lumbosacral cord may be due to the narrowing of ventral spinal artery in the lower thoracic and lumbar regions (132) and the higher metabolic demands of motor neurons in the lumbosacral cord (149). It is also possible that damage to the thoracic SC may have occurred but not reported or analyzed.

What does the data tell us?

Although the heterogeneity of the data renders identifying definitive correlations difficult, we would like to discuss some of the observed trends and patterns that may provide a global overview of the mechanisms of injury in SCII.

A common pattern of tissue injury was gray GM, seen in almost all cases and across species (except baboon). In over two-thirds of these cases, neuronal degeneration was observed. The time postclamping to observe these injuries varied, but generally occurred within 24 hours of the ischemic event. Damage tended to be worse in the anterior GM, although other portions of the GM were also affected. The damage to both anterior and dorsal GM segments was observed to occur similarly, within 24 hours.

WM damage was less common, although axonal damage was shared in common across all animal types (except sheep). Injury was seen within 1-3 days, but generally took longer to manifest than gray matter changes. The changes seen included axonal degeneration/swelling, and/or myelin degeneration.

Limitations of the Study

One limitation of this review is our decision to analyze animal models that used aortic cross-clamps and vascular clips to model SCII. Other models designed to simulate OR surgery of TAAAs include the use of an intra-aortic balloon, ligatures, snares, and chemicals to induce SCII. Second, other outcome measures apart from histology may be more effective in determining the exact pathophysiological mechanisms of neurological injury following SCII. Histological findings may not provide a holistic portrayal of spinal injury due to the inherent limitations of histological stains; moreover, histological findings may be incorrectly characterized due to the morphological similarities of different pathophysiological processes and due to incorrect handling SC specimen. Moreover, because of the qualitative nature of histological outcomes we could only comment on what observations were made through histology as opposed to what observations were not made: absence of evidence is not evidence of absence. Accordingly, histological outcomes are best understood concurrently with other outcomes such as qPCR, western blot, RNA seq, metabolomics, and flow cytometry. We also did not include a histological analysis of various drug treatments for SCII; such analysis would aid in the understanding of pathophysiological processes and the efficacies of various therapies. These limitations imply that more research and experimentation is required to elucidate the pathophysiological processes of SCII and develop therapies.

With regard to the risk of bias, it does not consider the bias that may result from the interpretation of various methods of staining. The risk of bias only allows us to consider the biases listed in items #1-10 in Supplementary Data Appendix B. This prevents us from completely identifying all of the risk factors for bias in each article.

Concluding Remarks and Future Directions

Despite decades of research, an effective therapy aimed at attenuating neurological injury following OR surgery of TAAAs has not been identified. In this systematic review, in an effort to elucidate the mechanisms of SCII and provide guidance for the development of a treatment for SCII, we summarize the histological outcomes of articles that conducted cross-clamping of the aorta to simulate OR surgery of TAAAs. Our results offer insight into the pathophysiological processes of SCII and are useful for future experimentation aimed at attenuating the biochemical mechanisms that contribute to neurological injury. Recently, we published a study comparing the neuropathological outcomes after OR and endovascular repair in a dog model (150). We found that there is a neuropathological difference after each approach. The past 40 years of research on OR have focused on therapeutic development without consideration for the molecular mechanism of the pathophysiology. In order to properly develop therapeutics and interventions, a precise understanding of the molecular mechanisms is required beforehand. By targeting specific cellular pathways and mechanisms, the translatability and efficacy of therapeutics will be improved. We hope that future studies will take this into account and elaborate more on these neuropathological differences.

 

This study was supported by a National Institute of Health (NIH) National Institute of Neurological Disorders and Stroke (NINDS) R03NS102861 Grant to ET and HA, a National Institute of Health (NIH) National Institute of Neurological Disorders and Stroke (NINDS) R21NS113097 Grant to ET, HA, and MDB, a Development fund to HA (315279), and by the College of Medicine at the Ohio State University.

The authors have no duality or conflicts of interest to declare.

Supplementary Data can be found at academic.oup.com/jnen.