Risk Factors, Incidence and Mortality of Vertebral Artery Injury in Patients Undergoing Anterior Cervical Corpectomy: A Retrospective Large National Data Base Study

Abstract

Study Design

Retrospective Cohort Study

Objective

To determine the incidence of vertebral artery injury (VAI), risk factors, intervention, associated complications, and mortality in patients undergoing anterior cervical corpectomy.

Methods

We performed a retrospective review of the incidence of VAI during anterior cervical corpectomy using the PearlDiver database from 2010-2017. The CPT code 63 081 to identify corpectomy patients. Patient data extracted included, incidence of VAI, demographic factors, intervention, and future complications of death and stroke. The risk were calculated compared with those patients who did not have VAI.

Results

26 126 patients were identified to have undergone cervical corpectomy. Multivariate analysis of risk factors showed that younger age and male sex were associate with higher rate of injury (t = -11.5; P < .0001 and t = 3.8; P = .0001, respectively). Vertebral artery injuries occurred in 78 patients at an incidence of .3%. 11 (14%) VAI patients had a cerebral infarction compared with 1705 (7%) for non-VAI patients (OR = 2.13; 95% CI = [1.18 – 3.85; P = .0179]) during the follow up period. 1-year mortality rates were higher in patients who suffered a VAI (14%) compared to those who did not suffer a VAI (4%; OR = 3.85; CI = [2.04 – 7.14]; P < .0001).

Conclusion

Consequence of VAI may not be known for months following the injury. Although the same admission mortality is rare with this injury, there is a significant increase in post-discharge complications. This study suggests that further investigations into long term health risk of VAI is needed.

Introduction

Anterior cervical decompression by removal of vertebral body, also known as a corpectomy, is a commonly used surgical technique to treat cervical spondylotic myelopathy or radiculopathy. Vertebral artery injury (VAI) during anterior corpectomy is a well-recognized complication of surgery with a reported incidence ranging from .07% to 2.2%.^1,2 Given that the aberrancy in vertebral artery course occurs at the mid-vertebral body level, concerns for VAI are more significant with full or partial corpectomies as opposed to more routine discectomies.

Vertebral artery injury can be of significant consequence as it can result in stroke, comatose states, or death if blood loss or hemodynamic instability is severe.^3-5 Although there are reports of these complications associated with VAI, much of the literature consists primarily of small case series.^1-6 Because of the rarity of the complication the true incidence of VAI remains unclear, and the demographic of patients and long-term mortality and morbidity is not well described in the literature. This present study aims to quantify the incidence of VAI and identify risk factors and related complications using a large commercial database.

Methods

This study was a retrospective cross-sectional review of anterior cervical corpectomy procedures during the years 2010 to 2017 and their corresponding rates of VAI using the PearlDiver Inc database. ⁷ Institutional IRB approval (study 00023136) was obtained as a low risk non-human study and was deemed exempt from informed consent requirements.

The PearlDiver database includes anonymized and longitudinal HIPAA-compliant medical record claims data from all payers except Kaiser and Tricare throughout the United States. The current database includes medical records during the years 2010 and 2020 from 91 million randomly selected individual patients ranging from ages 2 to 85.

Inclusion criteria consisted of any patient who underwent a corpectomy procedure, identified using the CPT code, 63 081, while absent of primary and secondary neoplasm of the bone. VAI were identified using their corresponding ICD-10 and ICD-9 diagnosis codes (Table 1). Patient data, including age, gender, ICD-10 diagnosis codes, CPT codes, obesity status, Elixhauser comorbidity Index (ECI) scores, surgical levels, previous spine surgery, and mortality rates were collected. Obesity was identified using its respective ICD-10 and ICD-9 diagnosis codes (Table 1).

Table 1.

Diagnostic and procedural codes used to identify variables necessary for analysis.

Code(s)	Description
CPT 63081	Vertebral corpectomy (vertebral body resection), partial or complete, anterior approach with decompression of spinal cord and/or nerve root(s)
ICD-9-D-44324, ICD-10-D-S15101A, ICD-10-D-S15102A, ICD-10-D-S15109A, ICD-10-D-S15111A, ICD-10-D-S15112A, ICD-10-D-S15119A, ICD-10-D-S15121A, ICD-10-D-S15122A, ICD-10-D-S15129A, ICD-10-D-S15191A, ICD-10-D-S15192A, ICD-10-D-S15199A, ICD-10-D-I7774	Vertebral Artery Injury (includes left, right, and unspecified laterality of vertebral artery injuries)
ICD-10-D-I639, ICD-10-D-I6300, ICD-10-D-I63011, ICD-10-D-I63012, ICD-10-D-I638, ICD-10-D-I6389, ICD-10-D-I63013, ICD-10-D-I63019, ICD-10-D-I63111, ICD-10-D-I63112, ICD-10-D-I63113, ICD-10-D-I63119, ICD-10-D-I63211, ICD-10-D-I63212, ICD-10-D-I63213, ICD-10-D-I63219, ICD-9-D-43321, ICD-9-D-43491	Cerebral infarction
ICD-10-D-I6050, ICD-10-D-I6051, ICD-10-D-I6052, ICD-9-D-430	Subarachnoid Hemorrhage
ICD-9-D-7980, ICD-9-D-7981, ICD-9-D-7982, ICD-9-D-7989, ICD-10-D-R99	In-Hospital Death
ICD-9-CM 278.0, ICD-10-CM E66	Obesity

Descriptive statistics were obtained comparing the group with VAI to the group with no VAI. Chi-squared analysis and odds ratios were used to compare univariate categorical data between the two aforementioned groups. Multivariate logistic regression models looking at factors impacting the occurrence of VAI included the diagnosis of VAI as the dependent variable, with predictor variables of age and ECI score as continuous variables and male gender, obesity, previous spine surgery, multiple level surgery, and anatomy altering diagnoses as categorical variables. Anatomy altering diagnoses were defined as having the diagnosis of pyogenic vertebral osteomyelitis, spinal epidural abscess, or discitis at the time of surgery. Quantitative data, including age and ECI, between the two groups were analyzed using independent t-tests. Additionally, associations with one, two, and three-year mortality, cerebral infarction rate, and subarachnoid hemorrhage rate between the two groups of VAI vs no VAI were analyzed using univariate chi-squared analysis and odds ratios. Multivariate logistic regression models looking at factors impacting the three mortality rates included mortality as the dependent variable, with predictor variables of age and ECI as continuous variables and male gender, obesity, multiple surgical levels, previous spine surgery and occurrence of VAI as categorical predictor variables. After propensity matching the two groups by gender, age, ECI, and obesity, rates of mortality, cerebral infarction, and subarachnoid hemorrhage were analyzed again using univariate chi-squared analysis and odds ratios. Analysis was conducted in PearlDiver, which uses R coding (2021, version 4.1.0).

Results

There were 26 126 patients who underwent a cervical corpectomy identified. The mean age was 56.8 (±11.8) and 13 569 (52%) were female.

Vertebral artery injury occurred in 78 patients at an incidence of .3%. In most cases, the laterality was not identified (63/78, 80%). Univariate analysis (Table 2) showed increased risk of VAI occurrence with younger age (41.1 ± 17.4 years in patients with VAI compared to 56.8 ± 11.8 years in patients without VAI, P < .0001) and male sex (OR = 2.17; CI = [1.35 – 3.45]; P = .0015). Obesity showed no association with the incidence of VAI. Patients with lower ECI showed higher rates of VAI (2.8 ± 2.7 in patients with VAI compared to 3.5 ± 3.1 in patients without VAI, respectively; P = .0172). This difference was most likely due to the younger aged patients who sustained VAI. Multivariate logistic regression analysis that included predictor variables of age, ECI, gender, obesity, previous spine surgery, multiple levels, and anatomy altering diagnoses, showed that age (P < .0001), male gender (P = .0003), and multiple level surgery (P < .0001) were significantly associated with VAI occurrence.

Table 2.

Univariate analysis of factors associated with rates of vertebral artery injury.

Variable	WITH VAI	NO VAI	P-VALUE	ODDS RATIO	95% CI LOWER LIMIT	95% CI UPPER LIMIT
	n = 78	n = 26 048
Age (±SD)	41.1 (17.4)	56.8 (11.8)	< .0001	—	—	—
ECI	2.8 (2.7)	3.5 (3.1)	.0172	—	—	—
Male	52 (67%)	12 505 (48%)
Female	26 (33%)	13 543 (52%)	.0015	2.17	1.35	3.45
Obese	25 (32%)	8882 (34%)
Not obese	53 (68%)	17 166 (66%)	.7939	1.1	.68	1.77
Previous surgery	5 (6%)	1744 (7%)
No previous surgery	73 (94%)	24 304 (93%)	1	.95	.39	2.37
Multiple level surgery	16 (21%)	14 641 (56%)
Single level surgery	62 (79%)	11 407 (44%)	< .0001	4.97	2.87	8.62
Cerebral infarction	11 (14%)	1705 (7%)
No cerebral infarction	67 (86%)	24 343 (93%)	.0138	2.13	1.18	3.85
Subarachnoid hemorrhage	2 (3%)	63 (0%)
No subarachnoid hemorrhage	76 (97%)	25 985 100%)	—	—	—	—

Of the patients in this series who suffered a VAI, 11/78 (14%) had a cerebral infarction and 2 (3%) had a subarachnoid hemorrhage. Patients who suffered a VAI were more likely to have a cerebral infarction (OR = 2.13; 95% CI = [1.18 – 3.85; P = .0138) after the corpectomy procedure compared with patients who did not suffer a VAI.

One-year mortality rates were higher in patients who suffered a VAI (14%) compared with those who did not suffer a VAI (4%; OR = 3.85; CI = [2.04 – 7.14]; P < .0001). This difference remained statistically significant when the groups were compared after propensity matching for gender, age, ECI, and obesity (OR = 6.24; CI = [1.33 – 29.16]; P = .02) (Table 3) as well as after multivariate logistic regression analysis (P < .0001).

Table 3.

Non-propensity matched and propensity matched vertebral artery injury mortality rates.

All corpectomies	With VAI n = 78	No VAI n = 26 048	P-value	OR	Lower limit	Upper limit
Deceased (1-year)	11	1056
Not deceased (1-year)	67	24 992	> .0001	3.85	2.04	7.14
Deceased (2-year)	13	1834
Not deceased (2-year)	65	24 214	.002	2.64	1.45	4.80
Deceased (3-year)	19	3031
Not deceased (3-year)	59	23 017	.0009	2.45	1.46	4.11
Propensity Matched	With VAI n = 78	No VAI n = 78	P-value	OR	Lower limit	Upper limit
Deceased (1-year)	11	2
Not deceased (1-year)	67	76	.02	6.24	1.33	29.16
Deceased (2-year)	11	2
Not deceased (2-year)	67	76	.02	6.24	1.33	29.16
Deceased (3-year)	13	4
Not deceased (3-year)	65	74	.04	3.7	1.15	11.91

Two-year mortality rates were higher in patients who suffered a VAI (17%) compared with those who did not suffer a VAI (7%; OR = 2.64; CI = [1.45 – 4.80]; P = .002). This difference remained statistically significant when the groups were compared after propensity matching for gender, age, ECI, and obesity (OR = 6.24; CI = [1.33 – 29.16]; P = .02) (Table 3) as well as after multivariate logistic regression analysis (P < .0001).

Three-year mortality rates were higher in patients who suffered a VAI (24%) compared with those who did not suffer a VAI (12%; OR = 2.45; CI = [1.46– 4.11]; P = .0009). This difference remained statistically significant when the groups were compared after propensity matching for gender, age, ECI, and obesity (OR = 3.7; CI = [1.15 – 11.91]; P = .04) (Table 3) as well as after multivariate regression logistic analysis (P < .0001).

Discussion

Damage to the vertebral artery during anterior cervical corpectomy is rare but can lead to grave complications including death and stroke. When an aberrant vertebral artery exists, it is primarily situated at the mid-vertebral body level, thus VAI can be associated with a corpectomy procedure. ⁸ This study reviews the incidence of VAI, risk factors, and related complications including mortality and stroke.

Our results demonstrate that VAI occurred at a rate of .3% during anterior cervical corpectomy which is within the reported rate of other previous studies.^1,2 Indications for anterior cervical corpectomy, for example, pyogenic vertebral osteomyelitis, spinal epidural abscess, and discitis were included in the multivariant model and determined to not influence mortality or VAI occurrence. A previous surgical procedure was determined to not influence the occurrence of adverse events or alter the mortality associated with VAI.

From prior studies, it is expected that older patients with more degeneration would lead to higher rates of VAI, but our data determined there was not an increased incidence of VAI in older patients. ⁹ Spinal degeneration may influence a VAI, but potential hypotheses for the opposite finding include the degree of musculoskeletal tissue strength during dissection, neck circumference, and retraction-based difficulties to afford adequate exposure. Imaging studies for patients are not available through the database, and therefore, it was impossible to determine the extent of spinal degeneration. Male sex and younger individuals are associated with a greater muscle mass and stronger musculoskeletal system.^10-12 As a result, increased force may be necessary during the surgery, which may increase the possibility of eccentric removal of a vertebral body in a younger population. ¹³ In older individuals’ atrophy of muscle mass allows a surgeon to navigate the anatomy with increased visualization. ¹¹ Additionally, corpectomies may be required due to trauma, where normal anatomy may be disrupted. Younger people and men tend to have higher rates of cervical trauma, providing another possible explanation. ¹⁴ In addition, variation in the vertebral artery may contribute to increased injury. A study analyzing vertebral artery origin and entrance reported males to have three times more variability in C5 entrance level than females. ¹⁵

Men are also known to have higher rates of cervical osteophytes compared to females. ¹⁶ Osteophytes change the anatomy of the spine, which causes the corpectomy procedure to be more difficult with their presence. ⁹ Therefore, males with osteophytes are at an increased risk of damage to additional anatomical structures. Alternatively, women have less muscle mass and smaller body dimensions, which allows the surgeon to navigate their anatomy without excess force.^10,11 Patients with a lower ECI show lower rates of VAI, potentially attributed to a younger population included in the study. Furthermore, spinal instability from age-related degeneration, spondylolisthesis, spinal deformity, congenital defects, and tumors can alter the anatomy and lead to VA variation and shifting, thereby influencing VAI. ¹⁷ Finally, the exact vertebral levels operated on were not available to be determined as the database did not indicate specific vertebral bodies. In our analysis, single vs multiple level corpectomies were assessed, and significantly increase the occurrence of VAI (Table 2).

Post-operatively, patients with VAI are at increased risk for cerebral vascular impairment and death.^1,3 In a comparison of patients with and without VAI 1 year post-operatively, our findings were notable for an increased risk of cerebral infarction and subarachnoid hemorrhage in the VAI group. Vertebral artery injury may include pseudoaneurysms and arterial dissections which predispose a patient to embolic events. Large occlusive thrombus formation can be particularly devastating in the dominant vertebral artery or the contralateral vertebral artery if atherosclerotic or congenitally diminutive. Furthermore, there was a statistically significant increase in one-year, two-year, and three-year mortality in patients with VAI compared to those without. Three years following VAI is the extent allowed by the database. The higher incidence of cerebral infarction following VAI increases their risk for mortality and warrants a longer follow-up of VAI patients. Long-term monitoring for patients with VAI potentially allows earlier detection of fatal sequela.

The consequence of VAI may not be known for months following the injury. Although most of these patients do not die immediately from the injury, this study suggests that further investigations into the long-term health risk of VAI are needed. Repeated Magnetic Resonance angiography (MRA) or Computed Tomography Angiography (CTA) of patients with identified VAI to assess vertebral artery integrity may be beneficial to understanding appropriate treatment regimens including antiplatelet or anticoagulation interventions. Intervention including prescribing aspirin to patients with VAI may decrease their risk for thrombosis but increase their risk for bleeding. ¹⁸

The retrospective nature of the study adds limitations to the data and potentially leads to selection bias and a lack of control of additional variables. Inconsistency of surgical technique and procedure results in a lack of standardization. Training level and the surgeons’ level of experience were not noted. The indication for surgery was not monitored or noted and a population of patients may consist of those with concomitant non-iatrogenic VAI and cervical corpectomies. For instance, the trauma population may be subjected to an unstable cervical injury requiring a corpectomy as well as a blunt traumatic vertebral artery injury. We did not collect information about patients’ race or socioeconomic status, which could influence how they were treated by their healthcare team. Because this is a claims database, it relies on the accuracy of reporting from the clinical team. Because CPT codes are used in professional billing, we believe it would accurately capture the corpectomy patient using the CPT code. Vertebral artery injury is a significant complication and it is unlikely to be missed in coding however VAI injuries may occur that are relatively quiescent until the patient has a delayed ischemic deficit secondary to an embolic event. Thus, this may result in an underdiagnosis of VAI injuries.

In conclusion, the national incidence of VAI occurs once every 335 cervical corpectomies, with younger age and male sex significantly increasing the risk. A significant increase in stroke and mortality associated with VAI requires close monitoring of the patients following identification of this complication.

ORCID iD

Spencer Smith https://orcid.org/0000-0003-4703-557X