Spinal Cord Blood Supply and Its Surgical Implications

Home , Anterior spinal artery, Vertebral artery

Review Article Spinal Cord Blood Supply and Its Surgical Implications

Abstract Matthew W. Colman, MD The blood supply to the spine is based on a predictable segmental Francis J. Hornicek, MD, PhD vascular structure at each spinal level, but true radiculomedullary arteries, which feed the dominant cord supply vessel, the anterior Joseph H. Schwab, MD spinal artery, are relatively few and their locations variable. Under pathologic conditions, such as aortic stent grafting, spinal deformity surgery, or spinal tumor resection, sacrifice of a dominant radiculomedullary vessel may or may not lead to spinal cord ischemia, depending on dynamic autoregulatory or collateral mechanisms to compensate for its loss. Elucidation of the exact mechanisms for this compensation requires further study but will be aided by preoperative, intraoperative, and postoperative comparative angiography. Protocols in place at our center and others minimize the risk of spinal cord ischemia during planned radiculomedullary vessel sacrifice.

to the spinal cord. Most end within General Structure of Spinal the nerve root, dura, or pial plexus. Cord Blood Supply The radicular arteries that do contribute to the longitudinally oriented The spinal cord is nourished by single anterior spinal artery (ASA), a structure of arterial supply that fol- and therefore to the anterior two lows a generally predictable path from thirds of the spinal cord parenchyma, the great vessels to the parenchyma of the white and gray matter are named anterior radiculomedul- 1 throughout the spinal cord (Figure 1). lary vessels. Gao et al demonstrated Segmental vessels originate from the the presence of 72 of these vessels great vessels of the neck, thorax, across 20 human specimens, for an and abdomen at each segmental average of 3.6 per specimen. Con- 2 level and, with few exceptions, are versely, Martirosyan et al cited an paired bilaterally. Although the average of 10 (range, 2 to 17) segmental vessel typically divides throughout the spinal cord. Anterior into an anterior and posterior radiculomedullary arteries are usu- ramus, the posterior ramus is the ally not paired at any given level. dominant vessel and divides further Their end recipient vessel, the ASA, From the Department of Orthopaedic Surgery, Massachusetts General into a muscular branch and a spinal is mostly in continuity throughout 3,4 Hospital, Boston, MA. branch. The spinal branch becomes the entire cord and experiences both anterograde and retrograde J Am Acad Orthop Surg 2015;23: an anterior radicular artery and 581-591 a posterior radicular artery as it flow, depending on the functional demand and location of dominant http://dx.doi.org/10.5435/ traverses the neuroforamen alongside JAAOS-D-14-00219 the segmental nerve root. There are radiculomedullary feeder vessels. 31 paired radicular arteries, one for The ASA gives off many central ar- Copyright 2015 by the American Academy of Orthopaedic Surgeons. each segmental level, but relatively teries, which project and often few of them contribute meaningfully bifurcate in the sagittal plane,

October 2015, Vol 23, No 10 581

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Spinal Cord Blood Supply and Its Surgical Implications

Figure 1 ischemic functional deficits when thoughttoberedundantandwell- disrupted. As opposed to the terminal collateralized, significant variation branches of the anterior system, the exists in the number of dominant PSAs supply the cord matter in supply vessels to the ASA, and their a centripetal pattern. Most anatomic number and distribution should be studies have not identified direct well documented before considering anastomotic connections between the surgical disruption. anterior and posterior systems,2 whereas others have observed a con- nection in certain anatomic areas, Structure Unique to the such as around the conus medullaris.1 Thoracic Spine In any case, significant overlap and redundancy does occur between the In the thoracic spine, the segmental terminal supply branches of these two vessels come from the aorta or the systems within the parenchyma of the subclavian artery and continue on as spinal cord itself. intercostal arteries. The number of radiculomedullary arteries in the thoracic spine is fewer than in other Structure Unique to the areas (average, one to four), and they Illustration of typical spinal cord 3 blood supply under normal Cervical Spine are more spread out. There is poor conditions. One radiculomedullary collateral potential in this region, vessel is shown, which supplies the The ASA in the cervical spine takes its and there is virtually no direct anterior spinal artery through origin from two intervertebral ar- communication between the central a hairpin loop. teries at the very upper end of the (anterior) and peripheral (posterior) cervical spine, where it is largest, and systems.4 The ASA of the thora- heading to the center of the cord slowly tapers to a constant diameter columbar spine is fed by one or two matter. These central arteries are for the remainder of the cervical anterior radiculomedullary vessels, most dense in the lumbosacral region and into the thoracic region.7 the most dominant of which is region, followed by the cervical and In the cervical spine, segmental ves- termed the artery of Adamkiewicz thoracic regions. These terminate in sels originate from the vertebral ar- (AA).8 This feeder vessel of the ASA centrifugally oriented, inside-to- teries in the upper portion and, in the can be up to 1.3 mm in diameter and outside capillary beds, which are remainder of the cervical spine, from occurs on the left side between T9 five times as dense in gray as in white the vertebral arteries and deep cervi- and T12 in 75% to 80% of cases.9,10 matter.5 cal, costocervical, or ascending cer- The AA gives off a dominant de- The posterior one third of the cord vical branches. The segmentals then scending branch in a “hairpin” parenchyma is supplied by paired give off anterior and posterior radic- configuration and a much smaller posterior spinal arteries (PSAs), which ulomedullary arteries, the former of ascending branch as it joins the ASA. are fed by the posterior radicular ar- which supply the ASA, as in other The ASA is typically continuous teries. This arterial supply is more regions of the spine. There are typi- throughout the thoracic spine, but its similar to an arterial plexus than to the cally from one to several dominant caliber is significantly attenuated, often-envisioned two longitudinal unilateral anterior radiculomedullary especially as it approaches the paired arteries.6 Generally, these ar- vessels in the cervical spine.1,2 AA.9,11-13 These factors are thought teries are much smaller than the ASA Although the blood supply to the to be responsible for the sensitivity of and are less consequential in regard to cervical spinal cord is generally the thoracic region to ischemic insult

Dr. Hornicek or an immediate family member has received research or institutional support from Stryker; has received nonincome support (such as equipment or services), commercially derived honoraria, or other non-research–related funding (such as paid travel) from Biomet; and serves as a board member, owner, officer, or committee member of the American Association of Tissue Banks and the International Society of Limb Salvage. Dr. Schwab or an immediate family member is a member of a speakers’ bureau or has made paid presentations on behalf of Synthes and Stryker Spine; serves as a paid consultant to and serves as a board member, owner, officer, or committee member of Biom’up; and has received nonincome support (such as equipment or services), commercially derived honoraria, or other non-research– related funding (such as paid travel) from Globus Medical and Stryker. Neither Dr. Colman nor any immediate family member has received anything of value from or has stock or stock options held in a commercial company or institution related directly or indirectly to the subject of this article.

582 Journal of the American Academy of Orthopaedic Surgeons

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Matthew W. Colman, MD, et al of the anterior part of the spinal presentation of cord ischemia ranges involving open TAAA repair may not cord. from complete paraparesis to asymp- be the best model for examining the tomatic transient abnormalities anatomic causes of spinal cord ische- detectable only by neuromonitoring. mia. Endovascular techniques, such as Structure Unique to the The key clinical question driving thoracic endovascular aortic repair, Lumbar and Sacral Spine much of the anatomic research in which selectively excludes specific seg- spinal cord ischemia during surgical mental vessels by way of intra-aortic The tip of the conus medullaris may interventions is which key vessels stents, may be a better model for be nourished by the descending ter- must be preserved to maintain spinal examining the effect of segmental vessel mination of the ASA directly, but there cord perfusion pressure and therefore sacrifice on cord ischemia. In 71 pa- also exists a highly collateralized spinal cord blood flow to avoid func- tients undergoing TAAA endovascular anastomotic network at this location tional deficits. repair, Kawaharada et al20 observed termed the anastomotic loop of the less spinal cord injury in a group in conus.7 The lower lumbar and sacral whom the AA was not occluded by the radicular arteries contribute to this stent (none) compared with the group highly collateralized anastomotic Thoracoabdominal Aortic in whom the AA was occluded (10%), network rather than to the more Aneurysm Repair but the difference was not significant, cephalad ASA directly. Thus, Although thoracoabdominal aortic and the overall event rate was low although ligating a single artery or aneurysm (TAAA) repair does not (3.6%). Schurink et al21 analyzed 13 multiple sacral or lumbar radicular have direct relevance for the ortho- patients undergoing stent grafting of arteries usually does not have ische- paedic surgeon, its related literature is aneurysms at or below T8, 8 of whom mic consequences for the conus or the the most robust repository available had four or more segmental levels cephalad spinal cord, these vessels do for reports of ischemic spinal cord occluded by the stent, and 6 of whom supply end arterioles to these regions. damage resulting from vascular inter- had the segmental artery feeding the In fact, particulate matter clogging ruption. In fact, TAAA repairs have AA itself occluded. The authors these end arterioles has resulted in been described as “Russian roulette observed two cases of intraoperative lower extremity paraparesis, such as for the vascular surgeon”15 because transient motor evoked potential with inadvertent intravascular par- of a reported high incidence of spinal abnormality correctable by raising the ticulate steroid injection in the lower cord ischemia, with rates of 16% to mean arterial pressure, and no cases of lumbar or sacral spine.14 32%, depending on the location and postoperative paraplegia.21 extent of the aneurysm.16,17 Although these and other studies Spinal Cord Ischemia The mechanism of cord ischemia is indicate that the AA can be safely and During Pathologic multifactorial but involves the extent of permanently occluded at its more Conditions disruption of segmental vessels during proximal segmental supply, it is aortic mobilization and aneurysm important to remember that the aortic Spinal cord perfusion pressure is repair. Several studies have highlighted and microvascular structures of pa- simply a function of the distal aortic the importance of preserving the domi- tients with TAAA are chronically perfusion pressure (or mean arterial nant anterior radiculomedullary artery degenerative in nature and thus are not pressure) less the extrinsic cord pres- to the thoracolumbar spine or the normal. Chronic occlusion of seg- sure (or cerebrospinal fluid pressure). AA.18,19 In one study of open TAAA mental vessels may allow time for the The blood supply to the spinal cord repairs, preoperative angiographic development of collateral systems that may be interrupted iatrogenically visualization of the AA outside the supply the radiculomedullary vessels, during many different types of surgi- repair or aortic clamp area resulted in thereby making sudden occlusion less cal intervention, including aortic no spinal cord injury. In addition, when likely to cause a catastrophic ischemic aneurysm repair, tumor resection, the AA was located within the clamped event.22,23 A more rigorous model of and spinal deformity surgery. The section, microrevascularization of the critical cord blood supply may be mechanism of spinal cord injury may AA during the open procedure resulted found in other clinical scenarios. involve global hypoperfusion, as with in a 5% rate of cord ischemia com- aortic cross clamping or systemic paredwith50%whenrevasculariza- hypotension; selective ischemia from tion was unsuccessful.18 Spinal Deformity ligation of dominant segmental ves- However, because of the confound- sels; or secondary insult, such as ing effect of global hypoperfusion Study of the neurologic effects of seg- with reperfusion injury. The clinical during aortic cross clamping, studies mental vessel sacrifice demonstrates

October 2015, Vol 23, No 10 583

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Spinal Cord Blood Supply and Its Surgical Implications that segmental vessels must be ligated its rich perimedullary collateral accompanying radicular vessels can for exposure during anterior thora- supply.4 be sacrificed in dogs without cord columbar spine exposure and defor- Another study retrospectively ana- ischemia,28 Nambu et al29 and Ueda mity correction; early reports raised lyzed 1,197 patients who underwent et al30 undertook extensive experi- concerns that sacrifice of segmental anterior spinal surgery for kyphosco- mental work investigating the vessels supplying dominant radicu- liosis without using a regular tempo- importance of preserving cord- lomedullary vessels can lead to para- rary occlusion protocol before supplying vasculature during total paresis. Apel et al24 reported on three dividing segmentals. There were no en bloc spondylectomy. In several cases of congenital kyphoscoliosis in cases of postoperative paraplegia. initial reports, these authors showed which complete loss of somatosensory- However, in one patient, revision sur- only modest decreases (15% to evoked potential (SSEP) tracings was gery on the convex side of the defor- 25%) in spinal cord blood flow observed within 5 minutes after liga- mity was undertaken after prior without change in motor evoked tion of segmental vessels around the surgery on the concave side. In this potentials (MEPs) or spinal cord curve apex from T3 to T9. Each case patient, the authors did temporarily evoked potentials (SCEPs) after resulted in postoperative paraparesis. occlude the segmental vessels in the bilateral three-level segmental vessel This prompted the authors to institute surgical field, and for one vessel, there ligation at T11-T13.29,30 This area in a protocol in 44 patients whereby were immediate SSEP changes; thus, dogs corresponds to the thoracic segmental vessels being considered for the vessel was preserved.26 In addition watershed area in humans where the ligation were temporarily occluded to standard neuroprotective measures, AA is most often located. In follow- before permanent ligation. In seven such as avoidance of intraoperative up studies that sought to define the cases, SSEP amplitude changes of hypotension, these and other authors limits of segmental vessel sacrifice in .50% were observed within 5 min- advise dividing segmentals as far from this area without effect on cord utes of temporary occlusion of a seg- the neuroforamen as possible, divid- health, these authors observed blood mental vessel at the curve apex; all of ing segmentals unilaterally on the flow to the cord dropping to less thesereturnedtobaselinewithin5to convex side of the deformity only than half of normal only in dogs that 19 minutes after removal of the (whichmayplayalesserroleinvas- had bilateral five-level (44%) and occlusive clamp and preservation of cularization of the cord as a result of seven-level (25%) segmental vessel the vessel. No cases of postoperative chronic vessel tension), and consider- sacrifice; in addition, only in dogs paraparesis were observed.24 ing temporary occlusion or “soft with five or more bilateral levels A temporary occlusion protocol clamping” when a thoracic-only ligated were any MEP, SCEP, or before permanent segmental vessel deformity with intraspinal abnormal- neurologic examination abnormali- division has been advocated by other ity at the same levels is encountered. ties detected.31 In a refinement, the authors.25 However, the importance group studied segmental sacrifice of this has also been brought into over multiple levels that specifically question.4,26,27 Bassett et al4 studied Oncologic Resection of included the canine equivalent of the 15 patients undergoing anterior spi- Intraspinal Neoplasms AA based on preoperative angio- nal surgery for kyphoscoliosis. Using gram; only dogs with greater than angiography, the authors identified Resection of malignant neoplasms of three-level bilateral segmental vessel 32 total dominant radiculomedul- the spinal axis using the technique of ligations including the AA were lary vessels supplying the thora- total en bloc spondylectomy presents abnormal with regard to MEP, SCEP, columbar ASA in their population. a distinct challenge with regard to and neurologic examination.32 It is Nine of these vessels in eight patients maintenance of spinal cord blood on the basis of these studies that this were within the surgical field and supply. Typically, a more extensive, group currently advocates for safe, were temporarily occluded with no often 360° exposure is required over bilateral three-level segmental sacri- SSEP changes. However, in seven of at least two spinal motion segments fice, including the AA, in humans. these eight patients, other radicu- to isolate the resection level for uni- These authors have retrospectively lomedullary vessels outside the sur- vertebral disease. Also, the tumor reported on their experience with 180 gical field were not occluded. The extent may make it impossible to cases of total en bloc spondylectomy in eighth patient had no changes after safely preserve certain vessels, even if humans; 15 had the AA involved at the occlusion of the one dominant seg- there is preoperative concern for their resection level and underwent target mental vessel supplying the lone critical role in cord blood supply. vertebra preoperative embolization radiculomedullary vessel, but it was Based on an early report that up to and three-level segmental vessel sacri- located at L1, an area known for four unilateral nerve roots and their fice, including the AA. There were no

584 Journal of the American Academy of Orthopaedic Surgeons

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Matthew W. Colman, MD, et al

Figure 2

A through D, Illustration of disrupted spinal cord supply following ligation of a key segmental artery (shown in gray) (A) with three possible compensatory mechanisms for reconstitution of the anterior spinal artery (ASA). Without direct supply to the ASA via the typical flow from the segmental artery to the radiculomedullary artery (RMA), the ASA may be reconstituted by collaterals emanating from an adjacent segment radicular artery (B), communication between the posterior spinal arterial system and the ASA system via the pial plexus and areas of spinal cord parenchymal overlap (C), or compensatory dynamic reversal of flow in the ASA itself using supply from distant RMAs or the anastomotic loop of the conus (D). cases of neurologic deterioration or spondylectomy may increase blood for intraoperative monitoring of def- paraplegia after surgery.33 Other au- flow to the cord. Other anatomic icit include SSEPs, continuous elec- thors have also suggested that ligating explanations are that proximal liga- tromyographic monitoring (EMG), the AA may be safer than previously tion of a segmental artery or even of MEPs, and the intraoperative Stag- thought and have demonstrated post- an anterior radiculomedullary artery nara wake-up test. SSEP monitoring operative development of new collat- itself does not disrupt the afferent remains the preferred method and is erals to the ASA after AA ligation.34,35 supply to the ASA because of preex- unlikely to be affected by anesthetics, Theexplanationsgivenforthe isting, more distal contributions to the but because it predominantly mon- somewhat counterintuitive finding of critical anterior radiculomedullary itors the dorsal columns, it is at best normal spinal cord function after artery (Figure 2, B). These more distal an indirect measure of the lateral ligation of the dominant anterior ra- contributions may not be seen on corticospinal tracts and the motor diculomedullary vessel to the ASA are a preoperative angiogram because neurons, which are supplied by the theoretical. Tomita et al36 claim that they are dynamically recruited only ASA. False-negative readings can there must be anastomotic commu- after the pathologic condition is occur, with rates as high as 9%.38,39 nication between the posterior spinal initiated. Finally, it is possible that Free-running or continuous EMG artery and anterior spinal artery sys- anterograde or retrograde flow dy- monitoring, like SSEP monitoring, is tems through the intercanal, dural, or namics through the ASA itself from a good measure of individual nerve pial plexus (Figure 2, C). Some au- distant cephalad or caudad anterior root irritation during surgery and thors have corroborated this in a lim- radiculomedullary vessels compensate has the benefit of detecting sponta- ited fashion in anatomic dissections for a given anterior radiculomedullary neous nerve root activity in response by demonstrating communication at division (Figure 2, D). to stretch or compression. In con- the conus medullaris level,1 whereas trast to both, transcranial MEPs others have claimed that no functional Monitoring and Timing directly measure the anterior cord communication exists.2 Kawahara motor pathways. However, they are et al37 have demonstrated that spinal Following the sacrifice of key spinal susceptible to the anesthetic effects shortening during total en bloc cord supply vasculature, the options of halogenated compounds, nitrous

October 2015, Vol 23, No 10 585

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Spinal Cord Blood Supply and Its Surgical Implications

Figure 3 Figure 5

Spinal arterial angiography image with Midsagittal T2-weighted (A) and axial T1-weighted (B) postcontrast magnetic overlying radiograph depicting the resonance images of the spine in a 58-year-old woman with a history of small- comparatively minor contribution of cell osteosarcoma of the right hemipelvis depicting the tumor extent, with a second radiculomedullary artery pathologic collapse of the vertebra and extraosseous epidural tumor extension. locatedontheleftsideatT10inthe She presented with a solitary L1 metastasis 3 years after local and systemic same patient shown in Figures 3 and 4. treatment of her pelvic disease.

Figure 4 affect cord blood flow, but the duration is variable. MEP and SEP changes have been reported to occur within5to10minutesaftertem- porary or permanent occlusion of key segmental vessels and to return to baseline in reversible scenarios, such as “soft clamping,” within 5 to 19 minutes of clamp removal.24,43 However, although combined neurophysiologic monitoring is thoughttobeanadequateproxy for clinical motor function, the correlation is not perfect. Irrevers- ible SSEP/MEP changes may indeed A and B, Spinal arterial angiography images with overlying radiograph of the be accompanied by postoperative same patient shown in Figure 3, depicting the left-sided dominant thoracolumbar radiculomedullary artery, or artery of Adamkiewicz (black arrow). Filling of the paraplegia, but not always, and the anterior spinal artery is also seen (white arrow). reverse is true, as well, whereby false-negative readings occur.39 The report by Svensson et al16 of 1,509 oxide, or neuromuscular blockade ischemia of 100% and 91%, respec- TAAA procedures had a 16% clinical and are associated with complica- tively.41 The Stagnara wake-up test is postoperative neurologic deficit rate, tions from muscle contraction, effective but used less commonly but only 21% were observed in the intraoperative movement, and aber- today because a relatively long lag immediate postoperative period; rant electrical signals affecting the time is required to reverse anesthesia 32% were not observed for several brain or cardiac systems.40 Most and make an accurate determination days after surgery, and the range of authors feel that a combined of cord function clinically.42 onset was 1 to 21 days.16 Neurologic approach using SSEP and MEP is the Many authors have commented on damage from altered blood supply is safest configuration, with a sensitivity the timing of neurophysiologic mon- complex and may be related to local and specificity for detecting cord itoring changes after maneuvers that ischemia, global perfusion dynamics,

586 Journal of the American Academy of Orthopaedic Surgeons

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Matthew W. Colman, MD, et al reperfusion injury, or a combination Figure 6 of the three.

Protective Maneuvers

Standard neuroprotective maneuvers for any surgical procedure involving the spinal cord include maintenance of hemostasis, maintenance of mean arterial pressure, maintenance of end-tissue oxygenation, and avoidance of mechanical stressors such as distraction or rapid deformation. However, because spinal cord perfusion pressure is simply a function of mean arterial pressure minus the cerebrospinal fluid pressure, other methods have been developed to balance this equation in a favorable way. One example is controlled cerebrospinal fluid drainage via lumbar cannulation of the intrathecal space. A recent Cochrane review of three large trials investigating cerebrospinal fluid drainage in the setting Postoperative PA (A) and lateral (B) radiographs depicting the spinal reconstruction in the same patient shown in Figures 3 through 5, following of TAAA repairs found evidence, total spondylectomy of L1, with partial spondylectomies of T12 and L2, including albeit limited, that cerebrospinal fluid posterior segmental fixation, femoral allograft interbody reconstruction, Kaneda- drainage improves neurologic out- type rod fixation of the anterior vertebrae, and linkage of the posterior and come in open TAAA.44 Although less anterior systems via a crosslink device. germane to tumor resection or deformity surgery, another neuro- Figure 7 protective maneuver used during TAAA repair involves distal aortic perfusion during cross-clamping of the aorta to maintain distal aortic perfusion pressure. A Cochrane review found no high-quality randomized studies to support this technique but did highlight several observational studies that suggest a neuroprotective effect with improved outcomes.45 In addition, other authors have reported low rates of neurologic complications during TAAA using a protocol that involves reimplantation of segmental vessels previously divided for expo- 46-49 A, Axial short tau inversion recovery (STIR) magnetic resonance image of the sure; however, this protocol has cervical spine in a 30-year-old woman who presented with a 2-year history of left been called into question by others,19 periscapular shoulder pain demonstrating tumor encasement of the left vertebral who have reported low rates of artery and extensive left neuroforaminal involvement. B, Left-sagittal STIR neurologic compromise without re- magnetic resonance image demonstrating the tumor extent across multiple segments in the midcervical spine. implantation, and it is not commonly

October 2015, Vol 23, No 10 587

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Spinal Cord Blood Supply and Its Surgical Implications

Figure 8 limited support available for the use of cerebrospinal fluid drainage, evidence for most modalities is lacking.

Common Protocols at the Authors’ Center

Currently, to minimize the risk of acute or postoperative ischemic injury to the spinal cord during tumor resection, our group uses several methods in the preoperative, intraoperative, and postoperative stages. Preoperatively, we use angiography to identify critical anterior radiculomedullary vessels, and we favor a multidisciplinary approach to the preoperative planning involving thoracic surgery, vascular surgery, and Standard (A) and reversed (B) spinal angiogram images in the same patient neurointerventional radiology. We shown in Figure 7, demonstrating the lone dominant C5 radiculomedullary artery frequently use preoperative emboli- and its contribution to the anterior spinal artery. zation of the tumor levels, depending on the presence of visualized collat- Figure 9 eral supply to the cord. Intraoperatively and postoperatively, we maintain the mean arterial pressure at .90 mm Hg; however, relatively higher pressures must be balanced with intraoperative bleeding risk. We use combined SSEP/MEP monitoring for every case. In high-risk patients, such as those with comorbid risk factors for ischemia (eg, old age, peripheral vascular disease, kidney disease) or with disease-related risk factors (eg, planned sacrifice of major radiculomedullary vessel or vertebral artery), we occasionally insert prophylactic cerebrospinal fluid drains, which are initially clamped and used simply to monitor cerebrospinal fluid pressure. Depending on the patient risk profile and any neurologic changes that occur intraoperatively or postoperatively, we Postoperative AP (A) and lateral (B) radiographs of the cervical spine initiate strict maintenance of cere- reconstruction, in the same patient shown in Figures 7 and 8. brospinal fluid pressures ,10 mm Hg52,53 with continued maintenance done at our center. Other potentially mals has demonstrated a neuro- of the mean arterial pressure at .90 clinically useful maneuvers include protective effect of intravascular mm Hg. In the case of a postoperative epidural or systemic cooling;50 also, injection of a direct free-radical neurologic change, we also advise the recent experimental evidence in ani- scavenger agent.51 Aside from the use of axial imaging to rule out other

588 Journal of the American Academy of Orthopaedic Surgeons

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Matthew W. Colman, MD, et al causes of neurologic change, such as spinal disease, there was no evidence Case 2 postoperative hematoma. of other metastasis or local recurrence A neurologically normal 30-year-old Intraoperatively, when dividing in the pelvis. She underwent pre- woman presented with a 2-year his- segmental vessels or nerve roots operative radiotherapy with 19.8 Gy tory of left periscapular shoulder pain. containing radicular vessels, we to the L1 tumor site. She underwent She had imaging and biopsy studies attempt to divide only the vessels that standard preoperative staging and consistent with a conventional-type are necessary for exposure and safe imaging procedures, including pre- chordoma involving the C4 and C5 resection of the tumor, and we divide operative thoracolumbar angiogra- cervical segments, as well as the left them as far laterally as possible. We phy. Angiography demonstrated C4-5 neuroforamen (Figure 7). The regularly use a “soft clamping” pro- a dominant anterior radiculomedul- tumor extent was predominantly left- tocol, in which key segmental or lary vessel originating from the left L1 sided with encasement of the left radicular vessels to be divided are segmental vessel and supplying the vertebral artery and was not metas- temporarily clamped and the neu- ASA (Figure 4). There was a minor tatic. She underwent preoperative rophysiologic monitoring observed contribution to the ASA via a second cerebral angiography, which demon- for 10 minutes before permanent radiculomedullary artery on the left strated a co-dominant vertebral division. If neuromonitoring changes at T10 (Figure 5). artery system with adequate contra- are observed during this “soft With a thorough understanding of lateral reflux through the circle of clamping,” the clamp is removed the neurologic risks, the patient elec- Willis and the presence of a single and attempts to spare the segmental ted to undergo en bloc resection of dominant radiculomedullary vessel vessel in question are undertaken. the isolated metastasis at L1. This Often, the vessel in question cannot was performed as a two-stage pro- emanating from the left vertebral be safely spared, given the tumor cedure, with the first stage involving artery at C5 (Figure 8). There were no extent, and although attempts at posterior decompression and partial identifiable tumor supply vessels for revascularization can be considered, transpedicular osteotomies at T12 embolization. She underwent pre- a difficult decision must at times and L2 and posterior segmental operative combinatorial proton- and be made regarding paraplegia risk instrumentation and fusion with iliac photon-based radiation to 50.4 Gy. versus oncologic resection margin. crest autograft from T10 to L4. T12 With a thorough understanding of This is discussed preoperatively with and L1 nerve roots along with their the neurologic risks, the patient elec- every patient. Advances in the techni- radicular vessels were ligated bilat- ted to undergo en bloc resection of ques of radiotherapy delivery both in erally, given the extent of the tumor. her tumor. This was performed as and out of the operating room have The second stage was performed 2 a two-stage procedure. The first stage helped address close or intentionally days later and consisted of left-sided involved posterior decompression, positive margins, but the clinical sce- thoracotomy and en bloc excision of partial posterior osteotomies at C3 nario remains challenging. L1, along with partial segments of and C6, posterior instrumentation T12 and L2, with femoral strut from C2 to C7, and ligation of the left allograft reconstruction and Kaneda- vertebral artery at C3 and C6. In Case Examples type anterior instrumentation (Figure addition, the C4 and C5 nerve roots 6). The patient did receive 10 Gy of were ligated, along with the domi- The following case examples illustrate intraoperative P32 dural plaque nant radicular vessel. Prior to liga- sacrifice of dominant radiculomedul- radiation, as well as a postoperative tion, however, temporary vascular lary vessels without postoperative boost with protons, for a total of clips were used for 10 minutes to test paraplegia. 68.4 Gy. Intraoperative margins for neurologic potential changes, were microscopically positive at the with no events observed. Throughout Case 1 dural margin, as planned. Given the the remainder of the first-stage oper- A neurologically normal 58-year-old L1 nerve root sacrifice, the patient ation, there were no significant woman with a history of small-cell had postoperative hip flexor weak- changes from baseline in the SSEPs or osteosarcoma of the right hemipelvis ness, but there was no evidence of MEPs. A wake-up test was not presented with a solitary L1 metas- global paraparesis, as might be pre- performed. tasis 3 years after local and systemic dicted following sacrifice of the AA The second stage of the resection treatment of her pelvic disease (Figure at L1. One explanation for this is involved completion osteotomies and 3). Based on a necrosis rate of 50%, that the minor T10 radiculomedul- a hemivertebrectomy of the most her tumor responded poorly to sys- lary artery compensated for the caudad portion of C3, the entirety of temic chemotherapy. Aside from the sacrificed dominant AA. C4 and C5, and the most cephalad

October 2015, Vol 23, No 10 589

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Spinal Cord Blood Supply and Its Surgical Implications portion of C6. The right vertebral spinal surgery. J Bone Joint Surg Br 1974; 17. Crawford ES, Crawford JL, Safi HJ, et al: 56(2):225-235. Thoracoabdominal aortic aneurysms: artery, including segmental vascula- Preoperative and intraoperative factors ture, was not dissected free because it 4. Bassett G, Johnson C, Stanley P: determining immediate and long-term Comparison of preoperative selective spinal was not involved by tumor. The results of operations in 605 patients. J Vasc angiography and somatosensory-evoked Surg 1986;3(3):389-404. tumor specimen was removed en potential monitoring with temporary bloc, and the defect was re- occlusion of segmental vessels during 18. Kieffer E, Richard T, Chiras J, Godet G, anterior spinal surgery. Spine (Phila Pa Cormier E: Preoperative spinal cord constructed with free autogenous 1976) 1996;21(17):1996-1999. arteriography in aneurysmal disease of the fibula graft and an anterior plate descending thoracic and thoracoabdominal 5. Turnbull IM: Microvasculature of the aorta: Preliminary results in 45 patients. (Figure 9). The postoperative course human spinal cord. J Neurosurg 1971;35 Ann Vasc Surg 1989;3(1):34-46. was not significant for any sensori- (2):141-147. 19. Etz CD, Halstead JC, Spielvogel D, et al: motor neurologic deficit except for 6. Gillilan LA: The arterial blood supply of the Thoracic and thoracoabdominal aneurysm weakness (3/5) in the left deltoid and human spinal cord. J Comp Neurol 1958; repair: Is reimplantation of spinal cord 110(1):75-103. subtle weakness (4/5) in the left arteries a waste of time? Ann Thorac Surg 2006;82(5):1670-1677. biceps, given the oncologically 7. Tveten L: Spinal cord vascularity: III. The spinal cord arteries in man. Acta Radiol 20. Kawaharada N, Morishita K, Kurimoto Y, scheduled root sacrifice at C5. By 8 Diagn (Stockh) 1976;17(3):257-273. et al: Spinal cord ischemia after elective months postoperatively, the motor endovascular stent-graft repair of the 8. Adamkiewicz AA: “Die Blutgefässe des thoracic aorta. Eur J Cardiothorac Surg function in the left biceps and deltoid menschlichenRückenmarkes: II. Die 2007;31(6):998-1003. muscle groups had normalized. Gefässe der Rückenmarksoberfläche”. Sitzungsberichte der Kaiserlichen Akademie 21. Schurink GW, Nijenhuis RJ, Backes WH, der Wissenschaften. Mathematisch- et al: Assessment of spinal cord circulation Naturwissenschaftliche Classe 1882;85: and function in endovascular treatment of Summary 101-130. thoracic aortic aneurysms. Ann Thorac Surg 2007;83(2):S877-S881. 9. Koshino T, Murakami G, Morishita K, The blood supply to the spine is com- Mawatari T, Abe T: Does the Adamkiewicz 22. Akasaka J, Sakurai M, Takase K, artery originate from the larger segmental plex and may have dynamic autor- Kumagai K, Tabayashi K: Another blood arteries? J Thorac Cardiovasc Surg 1999; supply to Adamkiewicz’s artery. Jpn J egulatory or anastamotic mechanisms 117(5):898-905. Thorac Cardiovasc Surg 2004;52(9): 432-434. to compensate for loss of critical vessels 10. Lazorthes G, Gouaze A, Zadeh JO, that feed the ASA during surgical pro- Santini JJ, Lazorthes Y, Burdin P: Arterial 23. Jacobs MJ, de Mol BA, Elenbaas T, et al: cedures. Elucidation of the exact vascularization of the spinal cord: Recent Spinal cord blood supply in patients with studies of the anastomotic substitution thoracoabdominal aortic aneurysms. J Vasc mechanisms for this compensation re- pathways. J Neurosurg 1971;35(3): Surg 2002;35(1):30-37. quires further study but will be aided by 253-262. 24. Apel DM, Marrero G, King J, Tolo VT, preoperative, intraoperative, and post- 11. Svensson LG, Klepp P, Hinder RA: Spinal Bassett GS: Avoiding paraplegia during operative comparative angiography. cord anatomy of the baboon: Comparison anterior spinal surgery: The role of with man and implications for spinal cord somatosensory evoked potential Relatively little study has been dedi- blood flow during thoracic aortic cross- monitoring with temporary occlusion of catedtothecervicalspine,andthis clamping. S Afr J Surg 1986;24(1):32-34. segmental spinal arteries. Spine (Phila Pa 1976) 1991;16(suppl 8):S365-S370. region in particular could be an area of 12. Biglioli P, Spirito R, Roberto M, et al: The future investigation. anterior spinal artery: The main arterial 25. Leung YL, Grevitt M, Henderson L, supply of the human spinal cord. A Smith J: Cord monitoring changes and preliminary anatomic study. J Thorac segmental vessel ligation in the “at risk” Cardiovasc Surg 2000;119(2):376-379. cord during anterior spinal deformity References surgery. Spine (Phila Pa 1976) 2005;30 13. Morishita K, Murakami G, Fujisawa Y, (16):1870-1874. et al: Anatomical study of blood supply to References printed in bold type are the spinal cord. Ann Thorac Surg 2003;76 26. Winter RB, Lonstein JE, Denis F, those published within the past 5 (6):1967-1971. Leonard AS, Garamella JJ: Paraplegia resulting from vessel ligation. Spine (Phila years. 14. Kennedy DJ, Dreyfuss P, Aprill CN, Pa 1976) 1996;21(10):1232-1233. Bogduk N: Paraplegia following image- 1. GaoL,WangL,SuB,WangP,YeJ,ShenH: guided transforaminal lumbar spine 27. Tsirikos AI, Howitt SP, McMaster MJ: The vascular supply to the spinal cord and its epidural steroid injection: Two case reports. Segmental vessel ligation in patients relationship to anterior spine surgical Pain Med 2009;10(8):1389-1394. undergoing surgery for anterior spinal approaches. Spine J 2013;13(8):966-973. deformity. J Bone Joint Surg Br 2008;90(4): 15. Doppman JL: Paraplegia after surgery for 474-479. 2. Martirosyan NL, Feuerstein JS, thoracoabdominal aneurysms: Russian Theodore N, Cavalcanti DD, Spetzler RF, roulette for the vascular surgeon. 28. Woodard JS, Freeman LW: Ischemia of the Preul MC: Blood supply and vascular Radiology 1993;189(1):27-28. spinal cord: An experimental study. reactivity of the spinal cord under normal J Neurosurg 1956;13(1):63-72. and pathological conditions. J Neurosurg 16. Svensson LG, Crawford ES, Hess KR, Spine 2011;15(3):238-251. Coselli JS, Safi HJ: Experience with 1509 29. Nambu K, Kawahara N, Kobayashi T, patients undergoing thoracoabdominal Murakami H, Ueda Y, Tomita K: 3. Dommisse GF: The blood supply of the aortic operations. J Vasc Surg 1993;17(2): Interruption of the bilateral segmental spinal cord: A critical vascular zone in 357-368. arteries at several levels: Influence on

590 Journal of the American Academy of Orthopaedic Surgeons

Copyright ª the American Academy of Orthopaedic Surgeons. Unauthorized reproduction of this article is prohibited. Matthew W. Colman, MD, et al

vertebral blood flow. Spine (Phila Pa 1976) 37. Kawahara N, Tomita K, Kobayashi T, 46. Jacobs MJ, Mess W, Mochtar B, 2004;29(14):1530-1534. Abdel-Wanis ME, Murakami H, Nijenhuis RJ, Statius van Eps RG, Akamaru T: Influence of acute shortening Schurink GW: The value of motor evoked 30. Ueda Y, Kawahara N, Tomita K, on the spinal cord: An experimental study. potentials in reducing paraplegia during Kobayashi T, Murakami H, Nambu K: Spine (Phila Pa 1976) 2005;30(6):613-620. thoracoabdominal aneurysm repair. J Vasc Influence on spinal cord blood flow and Surg 2006;43(2):239-246. function by interruption of bilateral 38. Dawson EG, Sherman JE, Kanim LE, segmental arteries at up to three levels: Nuwer MR: Spinal cord monitoring: 47. LeMaire SA, Miller CC III, Conklin LD, Experimental study in dogs. Spine (Phila Pa Results of the Scoliosis Research Society Schmittling ZC, Coselli JS: Estimating 1976) 2005;30(20):2239-2243. and the European Spinal Deformity Society group mortality and paraplegia rates after survey. Spine (Phila Pa 1976) 1991;16 thoracoabdominal aortic aneurysm 31. Fujimaki Y, Kawahara N, Tomita K, (suppl 8):S361-S364. repair. Ann Thorac Surg 2003;75(2): Murakami H, Ueda Y: How many ligations 508-513. of bilateral segmental arteries cause 39. Deutsch H, Arginteanu M, Manhart K, et al: ischemic spinal cord dysfunction? An Somatosensory evoked potential monitoring in 48. Acher CW, Wynn MM, Hoch JR, Popic P, experimental study using a dog model. anterior thoracic vertebrectomy. J Neurosurg Archibald J, Turnipseed WD: Combined Spine (Phila Pa 1976) 2006;31(21): 2000;92(suppl 2):155-161. use of cerebral spinal fluid drainage and E781-E789. naloxone reduces the risk of paraplegia in 40. Stecker MM: A review of intraoperative thoracoabdominal aneurysm repair. J Vasc 32. Kato S, Kawahara N, Tomita K, monitoring for spinal surgery. Surg Neurol Surg 1994;19(2):236-246. Murakami H, Demura S, Fujimaki Y: Int 2012;3(suppl 3):S174-S187. Effects on spinal cord blood flow and 49. Griepp RB, Ergin MA, Galla JD, et al: neurologic function secondary to 41. Pelosi L, Lamb J, Grevitt M, Mehdian SM, Looking for the artery of Adamkiewicz: A interruption of bilateral segmental arteries Webb JK, Blumhardt LD: Combined quest to minimize paraplegia after which supply the artery of Adamkiewicz: monitoring of motor and somatosensory operations for aneurysms of the descending An experimental study using a dog model. evoked potentials in orthopaedic spinal thoracic and thoracoabdominal aorta. Spine (Phila Pa 1976) 2008;33(14): surgery. Clin Neurophysiol 2002;113(7): J Thorac Cardiovasc Surg 1996;112(5): 1533-1541. 1082-1091. 1202-1213.

33. Murakami H, Kawahara N, Tomita K, 42. Owen JH: The application of intraoperative 50. Cambria RP, Davison JK: Regional Demura S, Kato S, Yoshioka K: Does monitoring during surgery for spinal hypothermia with epidural cooling for interruption of the artery of Adamkiewicz deformity. Spine (Phila Pa 1976) 1999;24 spinal cord protection during during total en bloc spondylectomy affect (24):2649-2662. thoracoabdominal aneurysm repair. Semin neurologic function? Spine (Phila Pa 1976) Vasc Surg 2000;13(4):315-324. 43. Mizrahi EM, Crawford ES: 2010;35(22):E1187-E1192. Somatosensory evoked potentials during 51. Herlambang B, Orihashi K, Mizukami T, 34. Boriani S, Bandiera S, Gasbarrini A: The reversible spinal cord ischemia in man. et al: New method for absolute spinal cord role of cord vascularization in planning Electroencephalogr Clin Neurophysiol ischemia protection in rabbits. J Vasc Surg spine oncologic surgery. ArgoSpine 2008; 1984;58(2):120-126. 2011;54(4):1109-1116. 18(1):40. 44. Khan SN, Stansby G: Cerebrospinal fluid 52. Svensson LG, Hess KR, D’Agostino RS, 35. Toribatake Y: The effect of total en bloc drainage for thoracic and et al: Reduction of neurologic injury after spondylectomy on spinal cord circulation. thoracoabdominal aortic aneurysm high-risk thoracoabdominal aortic Nihon Seikeigeka Gakkai Zasshi 1993;67 surgery. Cochrane Database Syst Rev operation. Ann Thorac Surg 1998;66(1): (11):1070-1080. 2012;10:CD003635. 132-138.

36. Tomita K, Kawahara N, Murakami H, 45. Hsu CC, Kwan GN, van Driel ML, 53. Coselli JS, LeMaire SA, Köksoy C, Demura S: Total en bloc spondylectomy Rophael JA: Distal aortic perfusion Schmittling ZC, Curling PE: Cerebrospinal for spinal tumors: Improvement of the during thoracoabdominal aneurysm fluid drainage reduces paraplegia after technique and its associated basic repair for prevention of paraplegia. thoracoabdominal aortic aneurysm repair: background. J Orthop Sci 2006;11(1): Cochrane Database Syst Rev 2012;3: Results of a randomized clinical trial. J Vasc 3-12. CD008197. Surg 2002;35(4):631-639.

October 2015, Vol 23, No 10 591