Key points

Introduction

An ideal spine injury classification permits clear, easy, accurate, and reproducible communication between surgeons, residents, fellows, researchers, and other health care professionals. Many historical classifications were either mechanistic or based solely on the radiographic morphology of the injury. In an effort to more accurately guide treatment and possibly predict long-term outcomes, multiple recent classifications have been developed that consider the patient’s entire clinical picture. Although many thoracolumbar injury classifications have been proposed, currently there is no single, globally accepted classification. In North America, many surgeons use the Thoracolumbar Injury Classification System (TLICS) ; in contrast, many European surgeons commonly use the Magerl system. Most recently the AOSpine Thoracolumbar Injury Classification system was published and validated; however, it remains unclear if this classification will be able to achieve global acceptance. The failure of surgeons to agree on a unified classification system for these injuries may initially seem unimportant, but the lack of a universal classification system has resulted in dramatically different treatment algorithms for similar fractures throughout the world; furthermore, the regional treatment algorithms tend to be similar in areas that use the same classification system.

Introduction

An ideal spine injury classification permits clear, easy, accurate, and reproducible communication between surgeons, residents, fellows, researchers, and other health care professionals. Many historical classifications were either mechanistic or based solely on the radiographic morphology of the injury. In an effort to more accurately guide treatment and possibly predict long-term outcomes, multiple recent classifications have been developed that consider the patient’s entire clinical picture. Although many thoracolumbar injury classifications have been proposed, currently there is no single, globally accepted classification. In North America, many surgeons use the Thoracolumbar Injury Classification System (TLICS) ; in contrast, many European surgeons commonly use the Magerl system. Most recently the AOSpine Thoracolumbar Injury Classification system was published and validated; however, it remains unclear if this classification will be able to achieve global acceptance. The failure of surgeons to agree on a unified classification system for these injuries may initially seem unimportant, but the lack of a universal classification system has resulted in dramatically different treatment algorithms for similar fractures throughout the world; furthermore, the regional treatment algorithms tend to be similar in areas that use the same classification system.

Historical classifications

The first published thoracolumbar injury classification in the English literature was by Watson-Jones in 1938. He identified 3 distinct fracture types—the simple wedge fracture, the comminuted fracture, and the fracture dislocation—and he recommended different treatments for these fracture types. Additionally in the middle of the 20th century, many unique thoracolumbar fractures were identified, and a different treatment algorithm was proposed for the individual fractures. One such injury is the Chance fracture, which was originally described in 1948 by G.Q. Chance as a flexion injury resulting in a wedge deformity of the vertebral body that may result in the disruption of the posterior elements. Another classic thoracolumbar fracture is the burst fracture. The term burst fracture was coined by Holdsworth in 1970 in a series of more than 1000 spinal injuries, and he defined a burst fracture as any vertebral body compression fracture that disrupted the posterior vertebral wall. In the same publication, Holdsworth proposed the first mechanistic classification. He divided fractures into 6 basic types (simple wedge, dislocation, rotational fracture–dislocation, extension, burst, and shear injuries). Perhaps the most important and controversial finding in this publication was that Holdsworth reported that all fractures with an intact posterior ligamentous complex (PLC) were stable. Although this classification offered basic treatment guidelines, the classification has never been validated independently, and so although the term burst fracture has persisted, the remainder of the classification is no longer used.

Two- and three-column classifications

Kelly and Whiteside proposed the next major classification in 1968 when they divided the spine into 2 columns. The anterior column, which they considered the entire vertebral body and intervertebral disc, and the posterior column, which comprised the neural arch and ligamentous complex. They postulated that any injury that involved only one of the columns was stable, but any injury resulting in disruption of both columns was unstable. This classification was never validated independently, and it was challenged in 1983 when Denis published a comprehensive classification for thoracolumbar fractures based off of 412 patients with a thoracolumbar injury. The Denis classification is commonly thought of as dividing each spinal segment into 3 columns: the anterior column—from the anterior longitudinal ligament through the anterior two-thirds of the vertebral body; the middle column—from the posterior third of the vertebral body/intervertebral disc to the posterior longitudinal ligament; and the posterior column—everything posterior to the posterior longitudinal ligament. However, the classification actually divides fractures into 4 major types (compression fractures, burst fractures, seatbelt-type injuries, and fracture–dislocations), and then subdivides each fracture into 1 of 16 total subtypes ( Table 1 ). The 3-column theory was described in the same publication as an alteration to the 2-column theory of stability proposed by Kelly and Whiteside ; Denis reported that the individual fracture pattern should not dictate treatment, but rather the treatment was determined by the integrity of the middle column. Denis proposed that isolated anterior or poster column injuries were stable, but if the injury resulted in concomitant disruption of the middle column, the fracture was unstable. This 3-column concept of stability achieved widespread acceptance, and to this day is responsible for many surgeons recommending operative treatment of thoracolumbar burst fractures in a neurologically intact patient ; despite its widespread use and moderate to substantial interobserver reliability of identifying the 4 main types of fractures, when attempting to classify the fractures into these subtypes, the reliability is poor. Furthermore, the importance of the middle column as the critical piece in spinal stability continues to be debated in the scientific community today.

Building on the work of Denis, McAfee and colleagues published a classification that was both a morphologic and mechanistic classification. Fractures were initially classified into 1 of 6 fracture types: wedge compression, stable burst, unstable burst, flexion distraction, Chance, and translation injuries. Similar to the Denis classification, the spine was divided into 3 columns, and any fractures that involved the middle column were considered unstable; however, McAfee and colleagues proposed that the mechanism of failure of the middle column was critical to determining the appropriate treatment. Injuries of the middle column that occurred from either a compressive (burst fractures) or a distractive (Chance and flexion distraction injuries) could be treated posteriorly with distraction or compression instrumentation. Conversely, translational injuries were a contraindication to posterior distraction and required segmental instrumentation of each level.

Following the Denis classification, the Ferguson and Allen classification was published in 1984; importantly, this is a mechanistic classification that did not challenge Denis’ 3-column biomechanical concept of stability. Because of this, although the classification contributed significantly to the nomenclature of thoracolumbar injuries, it failed to alter the treatment paradigm for any fractures. This classification identified 7 injury types with 5 additional injury subtypes ( Box 1 ), and presented treatment recommendations based on the type of injury. Likely because this classification did not alter the treatment paradigm and is derived from an inferred injury mechanism rather than verifiable imaging characteristics, it has never been validated independently.

The next substantial contribution to thoracolumbar fracture classification came in 1994 when Magerl and colleagues published a hierarchical classification. The Magerl system divides fractures into 3 types based on mechanism (compression/A-type fractures, distraction/B-type fractures and rotational/C-type fractures) and then it further incrementally subdivides the fractures into a total of 53 subtypes based on specific morphologic characteristics of the fracture. Although this classification was designed such that there was an incremental increase in severity of the injury when increasing the overall fracture type (A → B → C) and fracture subtype, the classification is still designed around the 3-column concept of spinal stability. Because of this, it had little impact on the treatment algorithm of controversial fractures. Furthermore, owing to the complexity of the classification, multiple authors have demonstrated either poor or fair interobserver reliability, so a clinically validated treatment algorithm based off of this classification has not been possible. Additionally, although the Magerl system is criticized for being overly complex with its 53 unique injury types, it still fails to formally consider to critical factors in determining the treatment algorithm, the integrity of the PLC, and the neurologic status of the patient.