Classifications In Brief: The Thoracolumbar Injury Classification

History Thoracolumbar fractures present variably, ranging from simple compression fractures to complex distraction injuries with neurologic compromise. Generally these complex fracture patterns with concomitant neurologic injury are treated surgically, yet controversy remains regarding the treatment of thoracolumbar fractures, namely burst fractures, in patients who are neurologically intact. Shen et al. [16] in 2001 and Wood et al. [23] in 2005 highlighted substantial regional and national variability in the treatment of burst fractures in patients who are neurologically intact. Both groups performed small randomized trials to study the outcomes of operative versus nonoperative intervention in this setting, each finding no clinical benefit of surgery. The variety of fracture presentation and absence of clear treatment guidelines has led some authors to develop classification systems for thoracolumbar spinal trauma, to facilitate communication between providers and guide treatment [1, 3, 9, 17]. Holdsworth [3], in 1970, was the first to propose a thoracolumbar fracture classification system. This system defines injuries based on a two-column spine model: anterior and posterior columns as demarcated by the posterior longitudinal ligament. In 1983, Denis [1] subsequently introduced the concept of a middle vertebral column (posterior third of the vertebral body to the posterior longitudinal ligament), and in turn the three-column model of the spine commonly used today. This classification system describes four injury types: compression, burst, seat-belt injuries, and fracture-dislocations. Each type is based on morphologic features of the fracture and involvement of each of the three columns. In 1994, Magerl et al. [9] introduced the Arbeitsgemeinschaft für Osteosynthesefragen (AO) comprehensive classification fracture system. This system focuses on fracture morphology as the basis of its classification. The AO system classifies fractures by “Type”: Compression (Type A), distraction (Type B), and torsional injuries (Type C), with each progressive fracture type associated with increasing morbidity [9]. Each fracture type is broken into subgroups, with subgroups being further divided into subdivisions resulting in 27 unique fracture types creating a comprehensive, albeit complex, classification system. The AO classification was effective in establishing a common language for describing thoracolumbar fracture but failed to incorporate associated soft tissue and neurologic injuries, two factors crucial in guiding treatment decisions. Purpose The systems of Holdsworth [3], Denis [1], and Magerl et al. [9] provided a taxonomy for thoracolumbar fractures based predominantly on morphologic features of fractures. Vaccaro et al. [17] identified this weakness, and with a better understanding of spine biomechanics and advancements in imaging, proposed the Thoracolumbar Injury Severity Scale and Score. The Thoracolumbar Injury Severity Scale and Score was the first classification system to include neurologic status of the patient, a key determinant in any treatment algorithm. Vaccaro et al. [18] later revised the Thoracolumbar Injury Severity Scale and Score, classifying injuries with an emphasis on morphologic features of the fracture rather than mechanism, and introduced the Thoracolumbar Injury Classification and Severity Score. The Thoracolumbar Injury Classification and Severity Score, now widely used, attempts to provide a clearer framework to classify and guide the management of thoracolumbar fractures. Description The Thoracolumbar Injury Classification and Severity Score is a combined radiographic and clinical score that guides operative and nonoperative treatment of thoracolumbar fractures (Table 1). The score is calculated by investigating three principal axes: (1) morphologic features of the fracture, (2) integrity of the posterior ligamentous complex, and (3) neurologic status. Each axis is further subdivided into discreet subgroups that are scored based on increasing clinical morbidity. A final score of the Thoracolumbar Injury Classification and Severity Score classification is summated, with higher scores representing fractures with increasing fracture morbidity.Table 1.: Calculation of the Thoracolumbar Injury Classification and Severity ScoreThe first primary axis of the Thoracolumbar Injury Classification and Severity Score addresses morphologic features of the fracture as identified on radiographs, CT scans, or MR images. Like the AO classification, fractures can be classified as compression (1 point), burst (2 points), translation/rotation (3 points), or distraction (4 points). The ascending subgroups are associated with higher mechanisms of injury and increasing spinal instability. If two morphologic features are concurrently present, the most severe is scored (Fig. 1).Fig 1: A-D. The points applied in the Thoracolumbar Injury Classification and Severity scoring system are shown for (A) compression fractures, (B) burst fractures, (C) translation and rotation, and (D) distraction. (Published with permission from Arien L. Cherones, Department of Orthopaedics and Sports Medicine, University of Washington, Seattle, WA, USA.)The second axis delineates the integrity of the posterior ligamentous complex. The posterior ligamentous complex is comprised of facet joints, ligamentum flavum, and interspinous and supraspinous ligaments. The integrity is determined by radiographic findings on radiographs or CT scans, as suggested by splaying of the spinous processes or diastasis of the facet joints. Occult injuries may be identified on MRI. An intact posterior ligamentous complex receives 0 points, an indeterminate injury 2 points, and a disrupted posterior ligamentous complex receives 3 points. The final axis is the patient’s neurologic status as determined by clinical examination. Intact neurologic function receives 0 points, a single nerve root injury receives 2 points, a complete spinal cord injury 3 points, and an incomplete injury 3 points. These findings generally correlate with the American Spinal Injury Association scoring scale [11] , with “E” signifying intact neurologic function, ”A” complete injury, and “B”, “C” and “D” signifying incomplete neurologic injuries [15, 18]. Vaccaro et al. [17, 18] generally recommend patients with Thoracolumbar Injury Classification and Severity scores 3 or less can be routinely managed nonoperatively and those with scores of 5 or more should receive surgical intervention. Patients with a Thoracolumbar Injury Classification and Severity score of 4 fall into an intermediate group and the decision to treat operatively is dependent on additional patient, disease, and surgeon factors not captured by the scoring system. Validation The value of a classification system derives from ease of use, high interobserver reliability, as measured by kappa value [20] (Table 2), and prognostic value [12]. Harrop et al. [2] were the first to examine the validity of the Thoracolumbar Injury Severity Scale and Score, a precursor to Thoracolumbar Injury Classification and Severity Score, presenting 56 clinical vignettes with imaging to 30 spine surgeons. Overall, they found 73.5% agreement in operative versus nonoperative management, and showed a fair interobserver agreement in total score for the Thoracolumbar Injury Severity Scale and Score, with a kappa value (κ) of 0.24 The Thoracolumbar Injury Severity Scale and Score had very high agreement in neurologic status (κ = 0.94), but only fair agreement classifying the mechanism of injury (κ = 0.3). This finding prompted the development of Thoracolumbar Injury Classification and Severity Score, replacing fracture mechanism with morphologic features of the fracture.Table 2.: Summary of interrater reliability of the Thoracolumbar Injury Severity Scale and Score and the Thoracolumbar Injury Classification and Severity ScoreTable 2-A.: Summary of interrater reliability of the Thoracolumbar Injury Severity Scale and Score and the Thoracolumbar Injury Classification and Severity ScorePatel et al. [13] and Koh et al. [7] reported their results of studies testing the validity of the Thoracolumbar Injury Severity Scale and Score. Patel et al. [13] tested the validity of the scoring system at two times with surgeons of varying training levels and disciplines. The initial total and management scores improved from a κ of 0.19 to 0.509 and 0.46 to 0.72, respectively over 7 months. Koh et al. [7] similarly reviewed a cohort of patients and calculated the Thoracolumbar Injury Classification and Severity Score for each patient at two times. They reported a moderate overall interobserver agreement with a κ of 0.58. High agreement was seen in neurologic status (κ = 0.91) and substantial agreement regarding morphologic features of the fracture (κ = 0.61), similar to other studies [14, 17, 21]. Several groups have compared the Thoracolumbar Injury Classification and Severity Score with the Denis [1] and AO [9] classification systems. Joaquim et al. [5] compared the Thoracolumbar Injury Classification and Severity Score with the AO classification in a multicenter retrospective study of 49 patients operatively treated for thoracolumbar spine trauma. The Thoracolumbar Injury Classification and Severity Score correlated well with AO fracture types, with higher scores consistent with AO Types B and C fractures. They concluded that the Thoracolumbar Injury Classification and Severity Score better reflected patient morbidity than the AO system through evaluation of neurologic status. Lewkonia et al. [8] evaluated 54 cases and distributed them to a diverse panel of reviewers tasked to calculate the Thoracolumbar Injury Classification and Severity scores and classify the fractures using the Denis classification [1] system. They used a new statistical method called the generalizability coefficient (ep2) to assess interobserver reliability. The generalizability coefficient is similar to the kappa value but incorporates additional variables to account for sources of error. A generalizability coefficient of 0 to 0.2 is poor agreement, 0.2 to 0.4 fair agreement, 0.4 to 0.6 moderate agreement, 0.6 to 0.8 good agreement, and 0.8 to 1 very good agreement. They found good interobserver agreement among panelists, surgeons, and trainees with respect to Thoracolumbar Injury Classification and Severity Score morphologic features (ep2 -0.74), posterior ligamentous complex (ep2 -0.72), and neurologic status (ep2 – 0.84). They found a similar interobserver agreement between the Denis classification and the Thoracolumbar Injury Classification and Severity Score (ep2 – 0.82), yet highlighted the purely descriptive nature and lack of prognostic or treatment recommendation of the Denis system. Joaquim et al. [4] attempted to show the clinical utility of the Thoracolumbar Injury Classification and Severity Score through a retrospective analysis of 458 patients with thoracolumbar spinal trauma. They reviewed the cases of 458 patients between 2000 and 2010 to compare actual management with that proposed by their calculated Thoracolumbar Injury Classification and Severity scores. They identified 310 patients initially treated with conservative management. Three hundred one patients had Thoracolumbar Injury Classification and Severity scores of 1 to 3, six had a score of 4, and three had scores of 5 to 7. Overall, the Thoracolumbar Injury Classification and Severity score matched the actual management in 99% of patients treated nonoperatively and 46.3% of patients treated operatively. Conservative treatment failed in nine patients, three of whom had misdiagnosed distraction injuries. The remaining patients had Thoracolumbar Injury Classification and Severity scores of 2 to 4 and were treated for worsening kyphosis or pain. The mismatch was related to surgical treatment of 79 patients with burst fractures but who were neurologically intact, all with Thoracolumbar Injury Classification and Severity Scores of 2. The authors attribute this discrepancy to the lack of treatment guidelines in the treatment of burst fractures. Joaquim et al. [6] in 2014 were the first, to our knowledge, to prospectively investigate the safety and efficacy of treatment directed by the Thoracolumbar Injury Classification and Severity Score. They analyzed a consecutive series of 65 patients who were treated according to their Thoracolumbar Injury Classification and Severity score at presentation. Patients with scores less than 4 were treated nonoperatively while patients with scores of 4 or greater were treated operatively. Thirty-seven patients with scores of 1 to 2 points initially were treated nonoperatively. Nonoperative management failed in two patients owing to pain and progressive kyphosis. Interestingly, neither patient experienced improvements in their symptoms after surgery. No patient experienced neurologic changes owing to nonoperative intervention. Twenty-eight patients with scores of 4 to 10 were treated surgically. No patient experienced worsening neurologic function after surgery. Despite a small cohort and limited followup, that study showed the efficacy and safety of the Thoracolumbar Injury Classification and Severity Score as a tool to guide surgical management of patients with thoracolumbar spine trauma. Limitations The simplicity of the Thoracolumbar Injury Classification and Severity Score system serves as a strength and a potential weakness of its application. The system accurately directs management of thoracolumbar fractures associated with high energy, instability, or neurologic compromise, and low-energy single-column fractures. Despite this, the Thoracolumbar Injury Classification and Severity Score falls short in clearly defining management for patients with stable burst fractures who are neurologically intact, including patients with scores of 4. The patients in this intermediate group present with varied fracture patterns and neurologic status, and may be appropriately treated with operative or conservative interventions. Some authors have proposed modifications in Thoracolumbar Injury Classification and Severity Score scoring system to improve management of indeterminately scored thoracolumbar fractures. Mattei et al. [10] suggested substratification of compression injuries to define whether one or two endplates (including the posterior vertebral cortex) were affected, with two endplate injuries being more likely to fail nonoperative management. Others believe routine MRI (to better define posterior ligamentous complex injuries) should be obtained in all patients with thoracolumbar burst fractures. Winklhofer et al. [22] reviewed Thoracolumbar Injury Classification and Severity scores for 100 patients before and after CT and MRI and found that 33 of them had additional occult posterior longitudinal ligament injuries. The total Thoracolumbar Injury Classification and Severity score changed from less than 5 to greater than 5 in 24 of 33 patients. Multiple studies have shown a discrepancy between the proposed Thoracolumbar Injury Classification and Severity Score management and actual management of patients with neurologically intact burst fractures [5-7]. The largest retrospective review of patients with thoracolumbar spine trauma comparing proposed Thoracolumbar Injury Classification and Severity Score and actual management, found that 79 of 148 patients treated surgically (53%), had scores of 1 to 3, and an additional 5% had a score of 4. Joachim et al. [4-6] associated the lack of level I evidence and absence of clinical guidelines for the discordance between proposed management and delivered treatment. Conclusion The Thoracolumbar Injury Classification and Severity Score is an intuitive, clinically validated classification system designed to guide treatment of patients presenting with thoracolumbar spine trauma. Studies have shown high intra- and interobserver reliability, with a treatment algorithm for operative and nonoperative injuries validated through prospective clinical data. Despite its reliability and ease of use, the Thoracolumbar Injury Classification and Severity Score does not provide definitive insight into the treatment of neurologically intact burst fractures and indeterminate scores. Effort is continually being made to improve our understanding, description, and treatment of these injuries given the lack of universal adoption of any one thoracolumbar fracture classification system. In 2013 the AO spine group introduced a new classification scheme, the AO Spine Thoracolumbar Injury Classification System, which has been shown to have better reliability among surgeons in identifying morphologic features of fractures compared with the Thoracolumbar Injury Classification and Severity Score [19]. Further prospective multicenter trials may identify additional patient and disease factors necessary to guide treatment of indeterminate scores furthering the clinical value of the Thoracolumbar Injury Classification and Severity Score.