From Cothren CC, Moore EE, Biffl WL, et al: Cervical spine fracture patterns predictive of blunt vertebral artery injury. J Trauma 55:811–813, 2003. Reproduced with permission.

Diagnosis

Numerous groups have developed screening protocols to identify patients who are at risk of developing either VAI or carotid artery injury.¹⁴^–¹⁶ In general, the presence of any of the following features were used to select patients for most of the screening protocols: a cervical spine fracture or subluxation, Horner syndrome, skull base fracture, Lefort 2 or 3 facial fractures, extensive neck soft tissue injury, or the presence of neurologic deficits that are unexplained by intracranial imaging findings.

The gold standard for the diagnosis of VAI is angiography. However, angiography is invasive and carries a small but significant risk of complications. Since the late 1990s, CTA ¹⁷ and magnetic resonance angiography (MRA)¹⁸ were enthusiastically promoted as possible replacements for conventional angiography. Beginning in the early part of this decade, however, reports comparing the sensitivity and specificity of CTA and MRA with those of traditional angiography concluded that these newer tests were not appropriate alternatives because they missed a substantial proportion of VAI in addition to having a high false positive rate.¹⁹^–²⁰ Two important papers are worth mentioning here. Over a five-year period, Biffl et al. evaluated 46 patients with carotid and/or VAI and found that the sensitivity of CTA was only 68% and the specificity was 67%.¹⁹ With regard to MRA, the same group reported a sensitivity of 75% and a specificity of 67% when compared with conventional angiography. Similarly, Miller et al. described 43 VAIs in a group of 216 patients meeting screening protocol requirements.²⁰ All 43 patients underwent four-vessel cerebral angiography, and 30 of these 43 underwent CTA as well.²⁰ In these 30 patients, CTA revealed only 16 VAIs, leading to a sensitivity of only 53%.²⁰

In the same study, Miller et al. also compared MRA to four-vessel cerebral angiography and found a 47% sensitivity rate for the former study.²⁰ However, these early studies were performed with four-channel detector technology, with limited three-dimensional reconstruction capability.

Initially, many reports compared both MRA and CTA to conventional angiography. Magnetic resonance imaging has the added advantage of imaging the brain to identify acute stroke resulting from VAI. However, the difficulties in performing MRA on acutely ill trauma patients and the possible presence of orthopedic implants in this population largely detract from its otherwise favorable attributes.

Newer Technology Computed Tomographic Angiography

With newer technology that incorporated 8- and 16-slice multidetector systems, the ability to accurately detect arterial injury was vastly improved. Schneidereit et al. and Berne et al. in 2006 published the first reports on the use of 8-slice and 16-slice CTA, respectively, as the initial imaging modality to screen for blunt cerebrovascular injury.^21,²² The reports concluded that newer CT technology was effective in diagnosing more instances of both carotid artery injury and VAI; in addition, these authors stated that no patient with a negative CTA developed ischemic symptoms that prompted study with catheter angiography.^21,²² Langner et al. in 2008 performed CTA as part of their whole-body imaging protocol in 368 patients and discovered three VAIs; they concluded that since no patient with a negative CTA developed clinical symptomatology indicative of VAI, CTA was a sufficient screening study.²³ Similarly, Biffl et al. detected nine VAIs in a group of 331 patients via CTA that were then confirmed by catheter angiography; however, patients with normal-appearing CTAs did not undergo confirmatory catheter angiography.²⁴ Instead, the group concluded that none of the patients who had normal CTA went on to develop clinical symptomatology related to cerebrovascular injury.²⁴ Utter et al. studied a group of 82 patients who underwent 16-slice CTA and had normal findings, which was confirmed in 75 of these patients using catheter angiography, giving CTA a negative predictive value of 92%.²⁵ In the four aforementioned studies, CTA and four-vessel cerebral angiography were not compared head to head; as a result, it was difficult to draw conclusions related to the ability of CTA to replace four-vessel cerebral angiography as the screening modality of choice. In addition, these groups all relied on the presentation of clinical symptomatology to define the false negativity of CTA.

However, Eastman et al. in 2006 laid many of these concerns to rest when they prospectively evaluated 162 patients with 16-slice CTA, of whom 146 subsequently underwent follow-up catheter angiography using their screening protocol for trauma victims.⁶ CTA successfully diagnosed 25 of the 26 VAIs subsequently seen on catheter angiography for a sensitivity of 96%.⁶ Malhotra et al. studied 7000 trauma patients over a 40-month period and used a more conservative screening protocol than other groups, resulting in 119 patients selected for screening with CTA, of whom 92 later underwent catheter angiography.²⁶ Thirteen patients were diagnosed with VAI, and the authors reported 75% sensitivity and 92% specificity for CTA.²⁶ However, CTA sensitivity increased over the latter half of the study period, leading this group to advise caution in the use of CTA as a sole means of diagnosing VAI, because there may be a learning curve associated with CTA evaluation, especially with regard to reconstructed images.²⁶

Natural History

Until recently, the failure to diagnose a significant number of VAIs limited the ability to predict, with any degree of certainty, the natural history of VAI, especially in asymptomatic cases. Louw et al. studied a group of 12 patients in South Africa who suffered from cervical spine dislocation and found that 9 of these patients (75%) suffered from VAI.²⁷ Only 2 of 9 patients suffered from neurologic deficits, but these resolved spontaneously during the follow-up period.²⁷ In the early 1990s, Willis et al. reported on 26 patients who had suffered from either cervical spine fracture or subluxation over a 21-month period and who were enrolled in a prospective study to evaluate the incidence of VAI using catheter angiography.⁸ Twelve of 26 patients (46%) were found to have VAI; in 9, it consisted of occlusion, and the remaining three suffered from intimal flap, arterial dissection, and pseudoaneurysm, respectively.⁸ None of the 9 patients with occluded VAI were treated, and none suffered from a delayed neurologic deficit attributable to the VAI during the course of their hospitalization. However, the study did not evaluate patients for the development of neurologic symptoms attributable to the VAI following discharge.⁸ Vaccaro et al.²⁸ and Parbhoo et al.²⁹ have reported on patients suffering from VAI secondary to cervical spine trauma; on the basis of follow-up MRA, they have concluded that the vast majority of injured vessels do not recanalize.

In a prescreened group of trauma patients who were at high risk of developing VAI, Miller et al. in 2001 reported a 14% rate of stroke attributable to VAI in a group of 50 patients.¹⁵ Similarly, Biffl et al. in 2000 reported a 24% rate of stroke among a group of 38 patients,¹⁴ and then in 2002, the same group reported a 20% rate of stroke when they evaluated patients over an 11-year period.¹¹ However, these rates are illustrative only of patients who develop neurologic deficits around the time of the traumatic injury, and it is still uncertain what the rate of stroke may be in asymptomatic patients who suffer from VAI and who are followed for long periods of time subsequent to their injury. It is logical to theorize that patients with lower-grade injuries may be at higher risk of subsequent stroke, but numerous reports have shown that there is no correlation between the grade of VAI and stroke.^11,^14,¹⁵

A recent report by Stein et al, may cast more light on the long-term outcome of VAI. This group evaluated 68 VAIs in 61 patients out of more than 12,000 admissions to a major urban trauma center.¹² Six of these 68 injuries (8.8%) resulted in stroke referable to either one or both injured vertebral arteries around the time of the initial trauma. At first follow-up (<1 month) to evaluate the VAI over the long term in 20 injuries, no injury worsened radiographically in either the medically treated group or the untreated group; all injuries in both groups either improved or were found to be stable radiographically. These results were echoed at later follow-up (>1 month).

Most important, however, at a mean clinical follow-up of 6.9 months, only one of 103 patients who suffered from either a carotid artery injury or VAI associated with trauma developed delayed stroke. The authors concluded therefore that (1) most strokes associated with VAI occur prior to any intervention and (2) treatment may decrease the rate of stroke, but up to one third of trauma patients may have contraindications to instituting antithrombotic therapy.

Management

The most controversial issue related to VAI is management. At this time, there is no class I evidence to support or refute any form of therapy. Some researchers ^11,^14,^15,²⁰ have advocated varying degrees of anticoagulation, antiplatelet agents, and/or endovascular intervention for most VAI lesions unless contraindications exist, while others^9,^12,^27,²⁹ have advocated observation in asymptomatic cases. There are two major centers advocating medical and/or endovascular treatment for VAIs, and these recommendations are borne out of the following studies from those groups.

In one of the first studies comparing medical treatment with heparin or antiplatelet agents and observation, Biffl et al. in 2000 ¹⁴ evaluated a group of 38 trauma patients with 47 total VAIs in Denver, Colorado. Of the 21 patients who were treated with heparin, 3 (14%) suffered stroke attributable to the VAI; and in the group of 17 patients with VAI who did not receive medical treatment, 6 patients suffered stroke (35%). This difference did not reach statistical significance (P = .13). The authors concluded that heparin and, if heparin was contraindicated, antiplatelet agents prevent stroke in patients who suffer from VAI and who may be asymptomatic. Heparin was given as a bolus dose of 15 U/kg/hour and then was administered to maintain a partial thromboplastin time of between 40 and 50 seconds. The study had many limitations. First, the study did not adequately define what was meant by “mild” or “severe” deficit in patients with neurologic signs and symptoms related to VAI. Second, the number of patients studied was too small to draw conclusions. Furthermore, radiographic improvement of VA characteristics, as a result of what the authors believed was the result of heparin treatment, had no bearing on clinical improvement.

The same group evaluated patients over an 11-year period ¹¹ to report on the effect of treatment on follow-up arteriography. In this report, there was no difference in the rate of healing of vessels in the group of patients given heparin (or antiplatelet drugs) or those who were given no treatment.¹¹ The authors continued to call for follow-up arteriography in this report, since they noted that 8% of grade I and 43% of grade II VAI progressed to pseudoaneurysm formation leading to endovascular treatment of these lesions; in addition, they noted that grade III and IV injuries scarcely changed over time. It is unclear, however, what the nature of changes were with regard to the anatomy of pseudoaneurysm formation in grade I and II injuries. We are also not aware as to the decision-making process of balancing the benefit of preventing further neurologic events and the risks inherent in stenting or angioplasty, which may cause strokes, worsen the vessel injury, or condemn patients to be maintained on antiplatelet therapy for years. There was no statistically significant difference (P = .15) in neurologic outcome between the three treatment groups: heparin, antiplatelet drugs, or no treatment.¹¹ In addition, there was a 1% rate of stroke as a result of follow-up catheter angiography in addition to a 22% incidence of bleeding complications—intracranial, gastrointestinal, and retroperitoneal—in the group of patients receiving heparin, leading the authors to omit the starting bolus dose.

A center in Memphis, Tennessee, has reported on the management of VAI.^15,²⁰ The group evaluated 64 VAIs in 50 patients over a 4-year period and noted that 7 of these patients (14%) developed stroke.¹⁵ Eighty-eight percent of the patients received some form of antithrombotic treatment; 62% received heparin, and 26% received antiplatelet drugs.¹⁵ Only four patients out of 50 received no treatment, one as a result of VAI resulting in vessel occlusion, one as a result of spinal cord contusion, and the other two as a result of poor prognosis.¹⁵ There was an 8% rate of complications associated with heparin use. Of the seven patients who developed stroke, two improved to resume normal life, and two died. The authors go on to conclude that since there were no strokes in the group treated with heparin, one stroke in the group treated with aspirin, and six strokes in patients who were not treated, patients with VAIs should be treated with heparin, and if that is contraindicated, some form of antiplatelet regimen should be instituted. However, the conclusions drawn by this study are weakened as a result of several study limitations. First, a comparison between asymptomatic VAIs that were then assigned to either treatment or no treatment groups was not performed. Second, the strokes included in the no treatment group included those in patients who received no treatment in addition to those in patients who did not receive treatment before the onset of ischemia. In addition, it is not clear whether the patients in the no treatment group were more critically ill or had severe brain injury that ultimately resulted in worse neurologic outcomes than those who were treated, further confounding the results. Obviously, it is very difficult to analyze, much less conclude, which mode of management is best if patients presenting with neurologic deficits are included in the analysis, thereby limiting the scope of treatment in these patients as compared with others.

The same group later used their previous group of patients ¹⁵ as a historical control and compared them with 43 patients who suffered a total of 49 VAIs,²⁰ the majority (74%) receiving treatment with antiplatelet drugs, 19% receiving systemic anticoagulation, and 7% receiving no treatment. The authors noted a 0% rate of stroke in these 43 patients and compared it to their historical control,¹⁵ in which they noted a 14% rate of stroke; with this observation, they concluded that treatment with antiplatelet medication reduces the rate of stroke in VAI, even though there was no difference in the stroke rate (0%) in any of the three groups. In addition, comparison with the historical control is flawed as a result of the inherent shortcomings in the first study, as mentioned previously.

A move toward standardization of treatment paradigms in the setting of VAI, therefore, is severely limited by the lack of definitive data supporting one management strategy over another. Studies reporting on VAI usually describe small sample sizes, are performed at individual centers, and succumb to the inherent intricacies of personalizing care in complex trauma patients in a nonrandomized fashion. Often, these studies include patients who present with neurologic injury secondary to the VAI, prior to the consideration of instituting anticoagulation or antiplatelet medication, in addition to having one or more contraindications to the institution of medical therapy. This group of patients is then compared with patients who are diagnosed with VAI who may be clinically asymptomatic but who nonetheless are then begun on anticoagulation or antiplatelet medication. Conclusions are then made with regard to the “lower stroke rate” of patients with VAI who are treated, compared to those who are not, all the while disregarding the important idea that patients who are asymptomatic from these injuries likely will remain asymptomatic regardless of therapy. In addition, the radiographic course of these lesions is one of gradual healing or stabilization, with few instances of recanalization; furthermore, radiographic appearance has not been shown to be significantly related to clinical course.

With regard to the design of trials evaluating the merits of various management options, several important factors, including the time course, the scope of antecedent neurologic injury in patients newly diagnosed with VAI, and the clinical relevance of injuries, have not been taken into consideration. In contrast to the majority of studies described earlier, cerebrovascular neurosurgeons and stroke neurologists must be involved in the design of future trials intending to answer the question of VAI management.