Diagnosis of carotid arterial injury in major trauma using a modification of Memphis criteria

The incidence of Blunt Cerebrovascular Injuries (BCVI) varies from 0.5% to 1% of all admissions for blunt trauma, but this relatively small percentage of patients is affected by a stroke rate ranging from 25% to 58% and a mortality rate ranging from 31% to 59% [1‐5].

Although BCVIs are related to severe complications and high mortality rate, controversy exists in literature when defining the patient at risk for these injuries. Four-vessel angiography has been considered the gold standard diagnostic test for the presence of BCVI for a long time. With the increasing availability and accuracy of computed tomography (CT), computed tomography angiography (CTA) has largely supplanted angiography as the primary means of diagnosing BCVI in many institutions, and it has recently been described as a reliable method of screening for the presence of BCVI [3, 6‐11].

A protocol used for identifying the highest risk patients for BCVI is the one proposed by Miller and co-workers ("Memphis approach") [2]. The aim of the present study was to evaluate if a modification of one of the original Memphis criteria can increase the sensitivity of this screening protocol in BCVI diagnosis,.

This study was conducted at the Intensive Care Unit (ICU) of a referral trauma center (Careggi Teaching Hospital, Florence, Italy), and includes trauma patients admitted from January 2008 to October 2009. The ICU database (FileMaker Pro 5.5v2, FileMaker, Inc, USA) was used for data registration and collection. To identify patients at highest risk for BCVI, the Memphis approach [2] was adopted and modified: all petrous bone fractures instead of lesions limited to the involvement of the carotid canal were included (Table 1), as previously suggested as risk factor for carotid arterial injury by Biffl and co-workers [12]. The study was conducted in accordance with the principles of the Declaration of Helsinki and Internal Review Board approved the study.

For each patient, demographic and clinical data were collected: age, Glasgow Coma Scale (GCS), Injury Severity Score (ISS), Revised Trauma Score (RTS), Trauma and Injury Severity Score (TRISS), length of stay (LOS) in ICU and in hospital, intra-ICU and hospital mortality, and neurological deficit at 6-months follow-up. Reports of all the relevant radiographic studies were also examined to determine diagnostic and confirmatory studies, grade of injury, evidence of cerebral or cerebellar infarction, and progression of the injury on follow-up imaging.

At admission, patients were examined in the 8-slice CT of Emergency Department as provided by the institutional protocol for major trauma admission. After radiological identification of the presence of modified Memphis criteria, a 64-slice CTA (64-slice multi-detector, General Electric, Fairfield, Conn.) was performed.

Protocol consists of a timed contrast injection, with images obtained from the aortic arch to the clinoids. Imaging parameters were the following: axial slices were collimated at 2.5 mm with a pitch of 1, but additional data acquisition from multiple slices per rotation allowed reconstruction in the sagittal plane to a resolution of 0.63 mm. Contrast was injected at a rate of 3 mL per second after a 25 second delay for 25 seconds. Physicians involved in the analysis of the CT scans were the radiologist, surgeon, intensivist. Grading of carotid arterial injury was assessed using the scale proposed by Biffl and colleagues (Table 2) [13].

Despite car safety systems, prevention programs, and pre-hospital/in-hospital strategy improvement of patient care, trauma caused by road accidents is still the major cause of death in the under-40s population [14, 15]. The most frequent fatal lesions are brain injury (45.8%), followed by thoracic, abdominal and pelvic trauma (41.6%) [3, 4]. The recognition and treatment of BCVI has dramatically evolved over the past two decades. Cerebrovascular injuries have been sporadically described since 1967 through a number of cases [16].

The main finding of our experience consists in the increased sensitivity, even though with he limitation of the sample size, of the Memphis approach by including all patients with petrous bone fracture, independently from the involvement of carotid canal. The percentage of patients screened in our population (19.9%) among those admitted for trauma was higher than previously reported (3.5%-10%) [2, 17]. As a difference from original Miller work [2], decision to perform CT angiography was based on CT scan and not on clinical findings. Besides, a single modification of original criteria permitted to extend the screening of the population and the diagnosis and treatment of BCVI in one patient who would not have been considered at risk if the original Memphis criteria were used. Using a modification of the original Memphis criteria, a number of patients underwent to additional radiological examination without any direct benefit: in our opinion, the risk/benefit ratio of X-ray exposure in case of severe trauma can be justified by the possibility to make a potentially lifesaving diagnosis.

Notably, we did not find any vertebral artery injury in our population. This maybe be attributed to the sample size and, consequently, to the limited number of cervical vertebra injuries observed (Table 4). More recently respect to this study period, the importance of petrous bone fracture has been identified in a multivariate logistic regression analysis of a large cohort [18] and underlined in the guideline of the Eastern Association for the Surgery of Trauma [19].

Three basic means of BCVI have been encountered: 1) extreme hyper-extension and rotation; 2) a direct blow to the vessel; 3) vessel laceration by adjacent bone fractures [20]. The most common causes of blunt carotid injury are: 1) hyper-extension of the carotid vessels over the lateral articulation of C1-C3 at the base of the skull; 2) a direct blow to the artery; 3) basilar skull fractures involving the petrous bone or sphenoid portions of the carotid canal. The accepted mechanism for vertebral artery injury results in secondary damage, due to fractures of the transverse foramen through which the vessel courses (C2-C6). Based on the nature of the injury, the traumatic event may cause intimal disruption, pseudo-aneurysm, dissection, and/or thrombosis. Moreover, the lesion can evolve despite a small intimal injury [1].

The overall incidence of BCVI observed in our population (2.2%) is higher if compared to the most recent study on a large trauma population, reporting a BCVI incidence of up to 1% [2, 15]. This observation may be related to the use of extended screening together with an improvement of sensibility in diagnosing vessel injury. Nevertheless, the real incidence rate of BCVI could be higher than previously reported. A superior BCVI incidence rate can be found in another recent article by Stein and co-workers [21]. The Authors, analyzing a large population of 12,667 patient in a 30 month period, reported an incidence of BCVI of 2.4% [21]. However, it must be noted that the study of Stein and colleagues was carried out without a defined screening protocol for BCVI, instead being based on physicians' judgment during CT-scan execution.

Literature shows that the most common associated injuries include closed head injuries (50-65%), facial fractures (60%), and thoracic injuries (40-51%). Nearly half of all patients had cervical spine fractures at the time of diagnosis [3, 4]. While these data appear different from our sample (Table 3), data shown confirmed that the carotid lesions are often associated to thoracic lesions, and in our case series, half of the patients with BCVI had thoracic region injuries.

The mortality rate of BCVI patients in our experience was 0 compared to the 13% reported in the largest study available [21]. This data is perhaps influenced by the median age which is noticeably lower (31 years old). Another explanation consists in the low-medium grade of lesions according to the grading system described by Biffl and co-workers [13] (Table 2). A timely anti-coagulant treatment was reported to reduce mortality due to the lesions, and prevent blood clots in asymptomatic patients [2]. In our experience, treatment with LMWH was effective in preventing stroke evolution, and none of the patients had bleeding events.

The early identification of BCVI remains a challenge in trauma patients. However, due to its potentially dramatic consequences, a standardized protocol to guarantee a prompt diagnosis is needed. Here we have shown our experience in which the inclusion of petrous bone fracture might have improved the sensibility of screening criteria. In consideration of the small sample and the single center setting, largest studies are needed to identify a common and shareable screening program based on well defined risk factors.