Fate of Patients with Spinal Cord Ischemia Complicating Thoracic Endovascular Aortic Repair

Abstract

Objective

Spinal cord ischemia(SCI) is a potentially devastating complication of thoracic endovascular aortic repair(TEVAR) that can result in varying degrees of short-term and permanent disability. This study was undertaken to describe the clinical outcomes, long-term functional impact, and influence on survival of SCI after TEVAR.

Methods

A retrospective review of all TEVAR patients at the University of Florida from 2000–2011 was performed to identify individuals experiencing SCI as defined by any new lower extremity neurologic deficit not attributable to another cause. SCI was dichotomized into immediate or delayed onset, with immediate onset defined as SCI noted upon awakening from anesthesia, and delayed characterized as a period of normal function followed by development of neurologic injury. Ambulatory status was determined using database query, chart review and phone interviews with patients and/or family. Mortality was estimated using life-tables.

Results

607 TEVARs were performed for various indications, with 57 patients(9.4%) noted to have postoperative SCI(4.3% permanent). SCI patients were more likely to be older (63.9±15.6 vs. 70.5±11.2;p=.002) and have a number of comorbidities including: COPD, hypertension, dyslipidemia and cerebrovascular disease(P<.0001). Fifty-four patients(95%) had a CSF drain placed at some point in their care, with the majority placed postoperatively(54%). In-hospital mortality was 8.8% for the entire cohort(SCI vs. No SCI;P=.45). Twelve patients developed immediate SCI, 40 had delayed onset, and 5 were indeterminate due to indiscriminate timing from postoperative sedation. Three(25%) immediate SCI patients had measurable functional improvement (FI), while 28(70%) of the delayed-onset patients experienced some degree of neurologic recovery(P=.04).

Of the 34 patients with complete data available, 26(76%) reported quantifiable FI, while only 13(38%) experienced return to preoperative baseline. Estimated mean survival(±standard error) for patients with and without SCI was 37.2±4.5 and 71.6±3.9 months(P<.0006), respectively. Patients with FI had a mean survival of 53.9±5.9 months compared to 9.6±3.6 months for those without improvement(P<.0001). Survival and return of neurologic function were not significantly different when comparing patients with pre- and postoperative CSF drains.

Conclusions

The minority of patients experience complete return to baseline function after suffering SCI with TEVAR, and outcomes in patients without early functional recovery are particularly dismal. Patients experiencing delayed SCI are more likely to have FI and may anticipate similar life-expectancy with neurologic recovery compared to patients without SCI. Timing of drain placement does not appear to have an impact on post-discharge FI or long-term mortality.

Introduction

Thoracic endovascular aortic repair (TEVAR) has become a mainstay of therapy for diseases of the thoracic aorta over the last decade. Despite being less invasive than open aortic repair, TEVAR still results in spinal cord ischemia (SCI) in 2–15% of patients^1–4. There are a variety of reported patient and procedure-related risk factors for SCI after TEVAR, including aortic treatment length^{3, 5}, left subclavian artery coverage^{1, 5}, obesity⁶, blood loss⁶, procedural urgency⁶, adjunct procedures⁶ (e.g. conduit, embolization), renal insufficiency⁷, hypotension⁵ and indication⁸. Further, a number of adjunctive therapies for the prevention and treatment of SCI after TEVAR have been reported and include: cerebrospinal fluid (CSF) drainage, left subclavian and/or hypogastric artery revascularization, augmentation of oxygen delivery and pharmacologically induced hypertension⁹. Despite increased awareness of this problem and judicious application of these interventions, some patients continue to suffer this devastating complication.

SCI leads to varying degrees of short and long-term disability, ranging from mild transient paraparesis to permanent flaccid paralysis, and the occurrence of this complication has a known negative impact on long-term survival^{1, 10}. Additionally, previous reports have suggested that neurologic recovery with delayed paraplegia compared to immediate paraplegia has a more favorable outcome^{7, 9, 11–13} but small sample sizes make it difficult to draw definitive conclusions about the natural history of SCI after TEVAR. Moreover, few reports have focused on the long-term functional outcome of patients with SCI after TEVAR, and the prognostic implications of the degree and speed of functional recovery after SCI.

The purpose of this study is to define the outcomes of patients experiencing SCI after TEVAR and determine differences in the evolution of long-term functional recovery, as well as the impact on survival.

Methods

Subjects and database

All TEVAR patients at the University of Florida between September 2000 and November 2011 were identified from a prospectively maintained endovascular aortic database. Patients diagnosed with postoperative SCI were further analyzed and compared to patients without SCI. Patients with neurologic deficits confirmed to be secondary to stroke or peripheral neuropathy were excluded from the analysis. Demographics, comorbidities, history of previous aortic surgery, and preoperative ambulatory status were determined by review of the electronic medical record. Procedure-related data pertaining to indication, aortic coverage zone(s), device type, timing of spinal drain use, anesthetic classification, as well as adjunct utilization were obtained from the database. Comorbidities and complications were defined and retrospectively recorded using the SVS reporting guidelines¹⁴.

Thoracic endovascular aortic repair technique and procedural adjuncts

The endovascular technique, sequence of graft implantation, need for adjuncts and spinal drain use was left to the judgment of the operating surgeon. Typically, patients with ≥ 150mm of aortic coverage³, a prior history of open/endovascular aortic surgery¹⁵, or an unrepaired infrarenal aneurysm¹⁵ were considered high-risk for paraplegia, and preoperative CSF drainage was employed. Peri-operative management of the spinal drain was based on a previously published standardized protocol^{3, 16}, and management of symptomatic patients is briefly outlined below. Patients undergoing elective repair were systemically heparinized (100U/kg) to achieve an activated clotting time of ≥300 seconds, although heparin was used selectively in urgent/emergent cases (e.g. aortic transection or aneurysm rupture). Protamine (1mg/100U heparin) was generally used at case completion to achieve normalization of clotting parameters.

Spinal cord ischemia diagnosis and management

SCI after TEVAR was defined as any new lower extremity motor and/or sensory deficit not attributable to other causes (e.g. epidural hematoma, intracranial pathology, peripheral neuropathy, or neuropraxia). Patients underwent a gross neurologic exam in the operating room whenever possible. Patients who had a documented change from their preoperative neurologic exam noted at the time of the first postoperative exam were considered to have immediate SCI, whereas those who experienced an interval of normal postoperative function followed by injury recognition were considered to have a delayed presentation. Consultation with Neurology and/or confirmatory imaging with spinal MRI were obtained in equivocal cases.

Patients were admitted postoperatively to a dedicated cardiothoracic ICU for hourly neurologic assessment, as well as continuous hemodynamic monitoring. If the patient developed SCI and did not have a CSF drain, a drain was placed immediately (usually within one hour of consultation) by the regional anesthesia service, and managed based on our institutional protocol^{3, 16}. CSF is drained to keep the pressure ≤ 10mmHg (14cm H₂O), with a serial decrease in pressure titrated to neurologic recovery.

To further optimize spinal cord perfusion, other maneuvers employed include volume resuscitation and vasopressor support (goal mean arterial pressure ≥ 90mmHg) with or without a drop in spinal drain height, as well as augmentation of oxygen delivery with maintenance of a cardiac index greater than 2.0 L/minute using vasoactive medications (if patients have invasive hemodynamic monitoring), pulse oximetry ≥ 96%, and maintenance of hemoglobin above 10 mg/dL.

Although drains are typically removed 36–48 hours after placement in asymptomatic patients, in those with documented SCI, the drain remains open for at least 72 hours after the onset of symptoms (irrespective of return of function), or up to 5 days, depending on whether neurologic recovery was observed.

After recovery in the ICU, patients were transferred to a dedicated cardiovascular nursing ward and received intensive inpatient physical and occupational therapy, with disposition to home or to a rehabilitation unit determined by the degree of neurologic recovery and functional assessment at time of discharge.

Study endpoints and definition of functional outcome

The primary end points included perioperative mortality, long-term survival, and overall functional outcome measured by ambulatory status. Perioperative mortality was defined as any death ≤30 days of the procedure or any death occurring during the initial hospitalization. Functional status and survival was determined by a review of the electronic medical record, as well as discussion with the patient or a close family member. When patients could not be contacted directly or through family, current survival status was verified by query of the Social Security Death Master file. Phone interviews were completed with a standardized questionnaire focusing on functional outcomes defined by ambulation status and the subjective assessment of functional improvement (see appendix). Patients were asked describe if they had return to preoperative global functional status, as well as best ambulation status. Ambulation status was divided into four categories: (1) ambulating independently, (2) ambulating with assistance (cane, walker, etc.), (3) non-ambulatory, but mobile (stand/pivot and transfer, wheelchair use), and (4) bedridden.

Statistical analysis

The SAS (V.9; Cary, NC) statistical software package was used to calculate means, standard deviations and frequencies. χ² or Fisher’s exact test were used to compare 2 groups (patients with or without SCI) on categorical variables, and Student t- or Mann-Whitney tests were used to compare them on continuous or integer variables, when indicated. The SPSS (V.20; Chicago, IL) statistical software package was used to estimate long-term survival using Kaplan-Meier curves. If the subject died, the time (in months) between TEVAR and death was analyzed as the survival time. If the subject did not die, time between TEVAR and December 1, 2011(the date of SSDI review) was the survival time. Because the groups were significantly imbalanced with respect to age, gender and comorbidities, a Cox logistic regression analysis was performed to account for potential confounding of these 2 variables. A P-value < .05 was considered significant. This study was approved by the Institutional Review Board at the University of Florida College of Medicine.

Results

Study cohort characteristics

Between September 2000 and November 2011, 607 TEVARs were performed and SCI was noted in 57 patients (9.4%). A permanent deficit from preoperative baseline was documented in 26 patients (4.3%). Analysis of the demographic and clinical variables of patients with and without SCI is demonstrated in Table I. Notably, development of SCI was more frequent in older patients (70.5±11.2 vs. 63.9±15.6 years, P=0.002) and there were a variety of comorbidities that were more prevalent in the SCI group (mean total number of comorbidities SCI=2.9 ±1.5 vs. No SCI=0.9±1.6, P< .0001).

Table I.

Patient characteristics and co-morbidities

Feature, No. (%)a	No SCI	SCI	OR (95% CI)	P-value
Patients	550 (90.6)	57 (9.4)
Age, mean ± SD, years	63.9±15.6	70.5±11.2		.002
Male, No. (%)	377 (69.2)	36 (63.2)		.34
Comorbidities, No. (%)b
Hypertension	149 (27.1)	50 (87.7)	18.7 (8.2–42.7)	<.0001
Smoking	78 (14.2)	21 (36.8)	3.4 (1.9–6.2)	<.0001
Dyslipidemia	66 (12.0)	25 (43.9)	5.1 (2.8–9.3)	<.0001
Renal Insufficiency	37 (6.7)	18 (31.6)	6 (3.1–11.7)	<.0001
COPD	34 (6.2)	16 (28.1)	5 (2.5–9.9)	<.0001
CVOD	14 (2.6)	11 (19.3)	8 (3.4–18.9)	<.0001
PVOD	18 (3.3)	7 (12.3)	3.3 (1.3–8.4)	.01
CAD	47 (8.6)	10 (17.5)	n/a	.10
CHF	15 (2.7)	2 (3.5)	n/a	.89
Diabetes	26 (4.7)	3 (5.3)	n/a	.90
Composite Total, mean ± SD	0.9±1.6	2.9±1.5	n/a	<.0001

^aChi-square or t-test when appropriate; OR, odds ratio; CI, confidence interval

^bMultivariable regression analysis to control for age and gender.

CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; CVOD, cerebrovascular occlusive disease; PVOD, peripheral vascular occlusive disease.

Table II further details differences in SCI vs. no SCI patients, including previous history of open or endovascular aortic surgery, procedural indications, and operative details. Notably, previous aortic surgery history was not more frequently associated with development of SCI (P=.23), and acute dissection was the only indication more likely associated with SCI (OR=2.6, 95%CI 1.3–5.2; P=.007). Overall, 261 (43.6%) of patients received a preoperative spinal drain and no difference in the rate of preoperative spinal drainage was noted between patients experiencing SCI (21/57;36.8%) vs. No SCI (240/550;43.6%)(p = .52). No differences were detected when comparing patients with and without SCI regarding procedural urgency, proximal aortic coverage zones, open or endovascular conduit use, prophylactic spinal drain utilization or frequency of subclavian revascularization. However, there was a significant difference in the mean number of aortic stent grafts used in the repair of patients with SCI (No SCI, 2.1±1.1 vs. SCI, 2.4±0.93; P=.0007).

Table II.

Previous aortic surgery history, distribution of anatomic indications for TEVAR, and procedure characteristics^ä

Feature, No. (%)	No SCI (n=550)	SCI (n=57)	p-value
Previous AAA repair	99 (18)	14 (24.6)	.23
Indication, No. (%)
Thoracic aortic aneurysm	248 (45.4)	25 (43.9)	.16
cTBAD	74 (13.6)	9 (15.8)	.16
Acute-dissection	67 (12.3)	13 (22.8)	.007
Penetrating aortic ulcer	62 (11.4)	8 (14)	.93
Traumatic aortic transection	40 (7.3)	0	.97
Post-surgical	24 (4.4)	1 (1.8)	.39
Other	24 (4.4)	0	.96
TAAA	7 (1.3)	1 (1.8)	.79
Procedural Variables, No (%)
Urgency
Elective	362 (65.9)	32 (56.1)
Urgent (symptomatic)	91 (16.6)	14 (24.6)
Emergent (rupture)	96 (17.5)	11 (19.3)	.25
Proximal coverage zone
0	42 (7.8)	4 (7)
1	15 (2.8)	0
2	217 (40.5)	30 (52.6)
3	134 (25)	12 (21.1)
4	128 (23.9)	11 (19.3)	.37
Number stents, mean±SD	2.1±1.1	2.4±0.93	.0007
Open conduit	128 (23.9)	17 (29.8)	.28
Carotid-subclavian bypass	74 (13.5)	10 (17.5)	.39
Preoperative spinal drain	240 (43.6)	21 (36.8)	.324

^äChi-square or Fischer’s exact test when appropriate; TEVAR, thoracic endovascular aortic repair; cTBAD, chronic type B aortic dissection; TAAA, thoracoabdominal aortic aneurysm

Procedural outcome data and length of stay (LOS) are highlighted in Table III. As expected, SCI patients had significantly greater LOS (15.5±12.3 vs. 8.4±11.5 days; P<.0001). Additionally, SCI patients more commonly had pulmonary complications (OR=3.2, 95%CI 1.5–6.7; P=.002) and experienced a greater average number of total complications (exclusive of SCI: 0.7±1 vs. 0.4±.8; P=.04), even when controlling for age, gender and procedural urgency. Despite differences in complication rates, in-hospital mortality was similar between the 2 groups (P=.49).

Table III.

Categorization of complications and outcomes after TEVAR^ä

Outcomes	No SCI (n=550)	SCI (n=57)	p-value
Length of stay (days±SD)	8.4±11.5	15.5±12.3	<.0001
Death (in-hospital)	35 (6.4)	5 (8.8)	.49
Category, No. (%)
Other*	60 (10.9)	5 (8.8)	.62
Pulmonary	36 (6.6)	11 (19.3)	.002
Renal	27 (4.9)	6 (10.5)	.11
Ischemic	24 (4.4)	2 (3.5)	1
Bleeding	22 (4.0)	4 (7.0)	.29
Cardiac	19 (3.5)	3 (5.3)	.45
Gastrointestinal	14 (2.6)	2 (3.5)	.66
Wound	7 (1.3)	2 (3.5)	.20
Total number of complications (mean±SD) [95% CI]	.4±.81 [.3–.5]	.65±1 [.4–.9]	.04

^äChi-square or Fischer’s exact test when indicated; TEVAR, thoracic endovascular aortic repair;

^*Other, includes mycotic, endoleak and device failure complications of the endograft

Functional outcomes

Table IV demonstrates outcomes of the SCI patient cohort (N=57) and depicts condition on discharge, disposition, and functional improvement (FI) during follow-up. One-third were discharged home while the remaining patients (57.9%; N=33) were transferred to an inpatient facility. Delayed SCI was documented in 40 patients, with 62.5% (N=25) of these patients reporting some degree of neurologic recovery and either independently ambulating or ambulating with minimal assistance (e.g. cane) upon discharge from the hospital. Three additional patients within the delayed SCI subgroup improved their functional status after discharge (N=28/40;70% with FI, in total). This is in sharp contrast to subjects with immediate onset of SCI (N=12), where only 25% had documented neurologic recovery at time of hospital discharge (P=0.04), and only one additional patient having any FI after discharge.

Table IV.

Disposition and functional outcomes for SCI patients

Outcome, No. (%)	n = 57*
Disposition
Home	19 (33.3)
Inpatient facility	33 (57.9)
In-hospital death	5 (8.8)
Condition on discharge
Ambulating independently	7 (12.3)
Ambulating w/assistance**	20 (35.1)
Non-ambulatory	9 (15.8)
Bedridden	10 (17.5)
Unknown	6 (10.5)
Best post-operative ambulation status
Ambulating independently	15 (26.3)
Ambulating w/assistance	16 (28.1)
Non-ambulatory	9 (15.8)
Bedridden	10 (17.5)
Unknown	2 (3.5)
Return to preoperative baseline functional status
Yes	13 (22.8)
No	21 (36.8)
Unknown	18 (31.6)
Any improvement in-hospital
Yes	23 (40.3)
No	26 (45.6)
Unknown	8 (14)
Any additional improvement after discharge
Yes	26 (45.6)
No	8 (14)
Unknown	18 (31.6)
Assistance with ADLs
Yes	13 (22.8)
No	17 (29.8)
Unknown	22 (38.6)

^*51 patients had complete hospital records documenting ambulatory status; 40 delayed and 12 immediate SCI patients

^**Cane or walker; ADLs, activities of daily living

Due to late deaths and inability to complete follow-up in some patients (e.g. moved/no contact information), comprehensive post-discharge data (including patient questionnaires) on FI were available for only 34(60%) patients. In total, 28 (82%) of these patients reported some degree of additional FI after hospital discharge, but only 13(38%) experienced complete return to preoperative baseline functional status. In total, 15 of 34 patients (44%) achieved independent ambulation at last follow-up. The remaining 6 patients reported no improvement in their lower extremity deficit after discharge.

An analysis was performed to determine the relationship of the patient’s functional status at time of hospital discharge to neurologic recovery as an outpatient. Not surprisingly, the patient’s ambulatory ability at discharge strongly correlated with their ability to gain measurable improvement in ambulation status during the follow-up interval (Spearman r = 0.89, P<.0001)(Figure 1). Specifically, if patients had a need for only a cane or walker at discharge, a higher proportion reported subsequent post-hospitalization improvement in ambulatory status compared to patients bedridden at discharge (P=0.07). Of note, no patient who was bedridden and unable to ambulate at discharge from the hospital achieved any neurologic recovery after leaving the hospital.

Figure 1.

Demonstrates ambulation status improvement after hospital discharge based on the ambulation status determined at discharge. There were no significant differences between the groups, but notable trends toward less likelihood for improvement with various degrees of reduced functional status at discharge. No patients reported a decline in function after discharge, and no patients with complete paralysis reported any functional improvement after discharge.

CSF Drainage

Among the patients who developed SCI, 54 (94.7%) had placement of a spinal drain either pre- or postoperatively. Three patients did not receive a drain due to coagulopathy and/or hemodynamic instability. Preoperative spinal drains were placed in 40.4% (N = 23) of cases. Fifteen of these 23 patients had data regarding long-term functional outcome after discharge, of whom 12(80%) experienced some element of neurologic recovery. Within the postoperative drain placement cohort [N = 31(54.4%); 19 with outpatient follow-up data], 73.7% experienced some degree of FI, which was not different than the preoperative drain group (P=1). Further, timing of spinal drain placement was not differentially associated with survival, ambulation status on discharge, best postoperative ambulation status, return to preoperative baseline function, or subjective functional improvement on long-term follow-up.

Survival

Long-term survival was estimated using Kaplan-Meier life table methods and is depicted in Figure 2. Estimated mean survival time (±standard error; SE) for all SCI patients was 37.2±4.5 months, which was significantly less than the 71.6±3.9 months observed in non-SCI patients(P<.0006). Notably, the significant survival difference remained when controlling for the increased age of the SCI cohort (HR = 1.7, 95%CI 1.2–2.6; P=.007). Overall survival of patients with any SCI was 64% at 12 months, compared to 82% in the non-SCI group (log-rank P<.001).

Figure 2.

Demonstrates a significant difference in survival between patients with SCI vs. those without (P<0.001).

Survival times were compared between patients who reported some degree of FI and those who had no recovery, with survival being dramatically better for those with FI. The mean survival was 53.9±5.9 months for those with FI and 9.6±3.6 months for those without (HR = 7.6, 95%CI 2.2–25.8; P=.001) (Figure 3). The long-term survival of individuals experiencing any measurable improvement was significantly better than those who did not, even when controlling for differences in comorbidities and demographics between the 2 subgroups. Strikingly, survival of SCI patients without FI was only 25% at 12 months, compared to 92% in those that reported some degree of neurologic recovery (log-rank P<.0001). Interestingly, there was a stepwise increase in survival as the degree of neurologic recovery increased.

Figure 3.

A significant difference in survival (P<0.0001) is present depending on whether the patients demonstrated any functional improvement after suffering SCI. The one-year survival of those patients with functional improvement was 92%, and 25% for those without any neurologic recovery.

Finally, to determine the impact of timing of neurologic injury on long-term survival, a comparison between immediate and delayed onset SCI is depicted in Figure 4a. A significant difference in all-cause mortality (log-rank P<.0001) is noted between patients experiencing immediate onset of SCI compared to patients who develop delayed SCI or no evidence of neurologic injury.

Figure 4.

Figure 4a. Differential survival is noted between those patients with return to preoperative neurologic baseline status versus those without return to baseline after suffering SCI (P=0.03) with TEVAR.

Figure 4b. This figure demonstrates differential survival between those patients without SCI vs. those with immediate or delayed SCI. Delayed SCI patients can anticipate similar long-term life expectancy compared to patients without SCI.

Despite the detrimental impact of SCI on survival Figure 4b, shows that patients who returned to baseline function had a further trend toward improved survival over those who did not return to baseline and their survival approximates that of patients without SCI.

Discussion

This study analyzed a large cohort of TEVAR patients with and without SCI and further compared patients based on the timing of SCI onset, as well as the degree of functional improvement attained after suffering this complication. Consistent with previous reports, SCI patients were noted to have poorer long-term survival than those without this complication. Interestingly, the timing of SCI and trajectory of functional improvement were important indicators of the overall prognosis.

SCI is a well-known complication of TEVAR ^{1, 17, 18}, and despite advancements in risk stratification and management, the incidence of this complication still ranges between 2–15%^{1, 3, 4, 19–21}. Indeed, in our own practice, despite a heightened awareness, liberal CSF drainage, judicious use of adjuncts such as subclavian revascularization and intensive monitoring, the rate of SCI has been consistent over time at 9%, with a permanent deficit rate of 4.3%. Although lower than the usual reported rate of this complication in open aortic repair, this is certainly not insignificant given the devastating impact of SCI.

A variety of risk factors have been associated with SCI after TEVAR and, although this not the focus of this analysis, many of those associations have been corroborated in this analysis. Many previously identified risk factors have been reported including: advanced age, male gender, a history of renal insufficiency, presence (or previous repair) of an abdominal aneurysm, acute dissection, lumbar/hypogastric artery patency, urgency of TEVAR, aortic coverage length, and left subclavian artery coverage^{1, 3, 10, 15, 18, 21, 22}. Notably, a prior history of open or endovascular aortic repair was not more frequently associated with the SCI group (P=.23), although an association was found in a previous analysis from our group of repaired or unrepaired AAA with smaller patient numbers¹⁵.

Once SCI was recognized, multiple adjuncts were utilized in our management algorithm, including spinal drainage, vasopressor induced hypertension and volume resuscitation. Unfortunately, despite aggressive use of spinal drainage, and the positive impact on short-term neurologic recovery in some patients, the timing of drain placement (preoperative vs. postoperative) did not appear to effect survival or functional recovery. While we strongly support the use of spinal drains for the management of SCI due to the clear evidence that this can improve neurological outcomes, it is not clear what benefit a preoperative drain has over a postoperatively placed drain. Additionally, using this analysis, we are not able to determine whether any of the available preventative maneuvers affect the long-term outcomes of SCI.

Perhaps the most interesting findings from this study include the prognostic importance of timing and rate of recovery of SCI after TEVAR. The majority of patients who suffer SCI after TEVAR have a delayed presentation (e.g. interval of normal neurologic function followed by development of neurologic deficit)^{21, 23}, and this was indeed the case in our series where 70.2%(N=40) of SCI patients presented in this manner. Of the delayed SCI patients, 62.5% experienced at least some neurologic recovery and were either independently ambulating or ambulating with minimal assistance (e.g. cane) at the time of discharge. This is in sharp contrast to patients with immediate onset of SCI (N=12), where only 25% (P=0.04) had any documented improvement upon discharge. Further, although occurrence of any SCI is a marker of poor long-term survival, this analysis suggests that patients experiencing rapid return of function can anticipate similar life-expectancy compared to patients without SCI (53.9±5.9 vs. 71.6±3.7 months;P=.41), and those with full return to their baseline function, not surprisingly, fare the best. Notably, whether their SCI was delayed or immediate in onset, only 25% of patients without functional improvement were alive at 1-year after TEVAR, compared to 92% of those with functional improvement.

Interestingly, despite the higher overall rate of complications in patients with SCI, the in-hospital mortality rate was not different between patients with and without SCI. Therefore, the majority of deaths occurred after hospital discharge. Unfortunately, we do not know the cause of death for most of our patients, and can only speculate about the explanation for poorer prognosis in SCI patients. As demonstrated in this analysis, SCI rarely occurs in isolation, and it is difficult to account for the confounding effect of other complications on mortality. Additionally, there are many known risks to the immobility caused by spinal cord injury that may impact long-term outcome, including venous thromboembolism/pulmonary embolus, decubitus ulceration, pneumonia, urinary tract infections, and chronic institutionalization²⁴. The pulmonary morbidity associated with spinal cord injury is also well known²⁵ and indeed our analysis demonstrated a higher rate of pulmonary complications in our SCI patients (P=.002).

There are several important limitations to this analysis including a heterogeneous patient population who were clinically diagnosed with SCI, many times without supportive imaging to confirm SCI. The methodology with which we identify SCI lends itself to high sensitivity and potentially low specificity and perhaps accounts for the elevated documented rate of SCI in the series. However, all indications and all pathologies were included in this analysis which further confounds the observed rate of SCI, as well as the potential trajectory of recovery. A dditionally, missing data elements in the clinical follow-up for the SCI cohort may introduce the possibility of bias that we are unable to account for in long-term follow-up. Unfortunately, there is no prevailing comprehensive definition of SCI after TEVAR, and we feel that our method is more likely to overestimate rather than underestimate the occurrence of this complication. Due to inconsistent documentation in medical records, no standardized metric for stratifying the severity of SCI was possible in this analysis. Further, the functional outcome assessment tool was based on a non-validated patient questionnaire that was administered retrospectively. Prospective analysis of a larger cohort of TEVAR patients at high-risk of SCI with standardized quality of life and physical therapy regimens would likely yield more definitive assessment about the impact of timing and severity on the potential rate of recovery after hospital discharge.

Conclusions

Spinal cord ischemia continues to be a challenging complication of TEVAR. Although patients suffering this complication have significantly reduced long-term survival, the subset of patients with delayed onset SCI with in-hospital recovery can anticipate similar life expectancy compared to TEVAR patients without SCI. Conversely, those patients with immediate onset SCI and no improvement in functional status before discharge have a dismal prognosis. The findings of this study are important clinical factors that can be used in planning of postoperative rehabilitation and patient/family discussions, as well as being taken into account in preoperative decision-making when considering patients with high-risk for development of SCI.