Differentiating Lumbar Fractures from Thoracolumbar Fractures

Jonathan Belding, Darrel S. Brodke, and Brandon D. Lawrence

Introduction

Nearly 90% of spinal fractures due to trauma occur in the thoracolumbar region. These fractures present a substantial burden on the patient from the standpoint of pain and disability, as well as a substantial economic burden on society. Given these burdens, a large body of literature exists as to the optimal treatment of these fractures, specifically addressing the controversies that occur with fractures in the thoracolumbar (T11–L2) region, as these are the most common given the transition between the rigid thoracic spine and the flexible lumbar spine. Although this has led to an increased understanding of the optimal treatment paradigm for this region, these principles cannot be simply applied to the lower lumbar region due to the differing anatomy and bio-mechanics of the lower lumbar spine. These anatomic and biomechanical considerations substantially distinguish the area below L2 from that above it, which creates challenges for the spine surgeon. This chapter discusses the differences between the thoracolumbar and lower lumbar regions and how they affect treatment options and subsequent patient outcomes. This chapter synthesizes the relatively small amount of literature on this subject and extrapolates from the literature on thoracolumbar injuries to provide a reproducible algorithm for dealing with traumatic injuries to the lower lumbar spine.

Anatomic Considerations

The adult spine shows anatomic variation as it moves from the thoracic to the lumbar region. The thoracic spine is more rigid due to the stability of the rib cage and its articulations, as well as its coronal facet orientation. Moving in a caudal direction, the facets become more sagittal, enabling flexion and extension in the lumbar spine. The vertebral bodies become larger in the lower lumbar region and are further stabilized by the iliolumbar ligaments and the pelvic brim. More importantly, the apex of lordosis exists at L3, which places the center of gravity more anterior to the vertebral bodies in the thoracic and thoracolumbar spine. This creates compression of the anterior column through L2 and a tensile force on the important posterior ligamentous complex (PLC). In the lower lumbar region, compressive axial forces are distributed throughout the body because the center of gravity is more posterior and therefore leads to less kyphotic deformity with less emphasis on the PLC in overall stability.1 The lower lumbar spine also has a higher degree of mobility, 20 degrees at L5/S1 compared with the thoracolumbar junction of 12 degrees. Finally, the spinal canal widens moving from the narrow thoracic cord level region to the capacious lumbar cauda equine region, providing more room for the neural elements during injury. All of these factors contribute to a relatively high area of stress concentrated at the thoracolumbar junction. This theoretical weak point has been borne out in multiple epidemiological studies. Most recently, Reinhold et al 2 looked at 733 patients with thoracolumbar fractures of all types and found a distribution of 19.8% thoracic injuries, 13.2% lumbar injuries, and 67% thoracolumbar junction injuries in operatively treated patients.

In addition to the relative mobility of the lower lumbar spine, it is important to remember that this region provides the majority of the total lumbar lordosis and therefore has a large effect on overall global sagittal alignment. Fusion studies and research on flat-back syndrome have shown the importance of maintaining lordosis, and it remains to be seen whether the lower lumbar region is able to tolerate significant focal kyphosis from lumbar fractures.

Evaluation

As with any traumatic spine injury, the initial assessment should follow the trauma protocols outlined in Advanced Trauma Life Support, starting with airway, breathing, and circulation (the ABC’s), followed by a thorough trauma workup while maintaining spine precautions. The incidence of associated injuries is quite high, around 20%, and includes blunt trauma to the chest causing pneumo- or hemothorax and hollow and solid visceral injuries with lower lumbar fractures.³ It is especially important to obtain and document a thorough neurologic examination, including a rectal examination, because a variety of neurologic patterns can present in thoracolumbar and low lumbar fractures.

Classification

Thoracolumbar spine fracture classification dates back to 1968, with Holdsworth⁴ defining a two-column model based on radiographs and biomechanics. This was refined by Denis⁵ into a three-column model with the introduction of computed tomography (CT) scans that is still often used for its descriptive nature of compression and burst fractures, flexion-distraction, and fracture dislocations. Although descriptive, it engendered controversy in terms of the stability of certain injuries, especially burst-type fractures. A more recent, comprehensive pattern has been established by the AO group, which categorizes these fractures as A (compression), B (distraction), or C (rotation), with further subdivisions. Although very complete from a descriptive and mechanistic standpoint, this classification is more useful in the research setting than in clinical practice. The Spine Trauma Study Group proposed the Thoracolumbar Injury Classification and Severity Score (TLICS) to provide a more useful classification in terms of outcomes and treatment plans. It assigns points based on the morphology of the fracture, the neurologic status, and the integrity of the PLC. A score of 5 or more indicates the need for surgery, whereas a score of 3 or below indicates a stable injury that is suitable for nonoperative treatment. A score of 4 leaves treatment to the surgeon’s discretion. This classification has been very helpful and has been shown to be reliable and reproducible for most thoracolumbar injuries, with the exception of multilevel contiguous injuries and extension injuries in stiff spines such as ankylosing spondylitis (AS) or diffuse idiopathic skeletal hyperostosis (DISH).⁶ (For more information on classification, see Chapter 1.)

Recently, however, authors have also called into question its reliability in the low lumbar spine. Lehman et al⁷ proposed a modification to the TLICS system, which they called the Lumbosacral Injury Classification System (LSICS). They revised the injury morphology and neurologic injury descriptions to make it more appropriate for lumbosacral injury patterns and found it to have good interobserver reliability. Although it is useful, this classification focuses more on the lumbosacral region and may not as adequately predict treatment for the low lumbar region without an associated sacral injury. Moore et al⁸ performed an interesting study to evaluate the reliability of the TLICS system in 20 fractures at L3–L5. They found only 28% agreement in the final TLICS score among 15 trained spine surgeons, although there was still 80% agreement in terms of overall decision on whether to operate or not. Interestingly, when looked at separately, the L3 fractures showed more consistency in the TLICS assessment, and this may prove useful when assessing them clinically.

Given the anatomic considerations previously described and the possible classification flaws, low lumbar fractures may be better described either by a combination of the above classifications or by the load-sharing classification system.9 This system assigns points based on increasing anterior column comminution, kyphosis correction (anterior gapping), and apposition of fracture fragments. Patients who scored greater than 7 had increased hardware failure if only posterior instrumentation was applied, and they may benefit from anterior column support. Because the load is more evenly distributed across the lumbar bodies in the low lumbar spine than in the thoracolumbar area, the importance of the PLC as described by the TLICS, may not as adequately determine the need for surgical stabilization. The most recent development, the AOSpine classification (see Chapter 1), tries to overcome the shortcomings of the previous mentioned classification systems by combining their strengths. Whether or not this system will be valuable for the lower lumbar spine still needs to be evaluated.

We will discuss later how these principles may also influence the surgeon’s decision on the approach to use for these low lumbar fractures in comparison to thoracolumbar fractures.

Neurologic Injury Patterns

The changing anatomy of the spinal canal also contributes to the difference in management between thoracolumbar and lumbar fractures. The spinal cord ends between L1 and L2 and terminates in the conus medullaris, with the branches of the cauda equina descending through the rest of the lumbar spinal canal. Injury patterns in the thoracolumbar region can therefore be on a wide spectrum from complete spinal cord injury to isolated nerve root injury. The most basic classification of these injuries is whether or not the injury is complete or incomplete, with the majority of practitioner’s utilizing the American Spinal Injury Association (ASIA) or Frankel classification to determine the grade of the severity of the injury. Injuries at the thoracolumbar junction can present with a typical upper motor neuron spinal cord injury pattern. As the injury moves lower into the upper and lower lumbar regions, the cauda equina can be affected. This presents with a lower motor neuron flaccid paralysis and possible neurogenic bladder secondary to the injury of the sacral upper motor nerve cells. Damage at the level of the conus medullaris itself can result in a conus medullaris syndrome with loss of bowel and bladder control, but sparing of the low motor neurons and near-normal muscle control of the legs.¹ It has been shown that spinal root injuries have a better prognosis for recovery than spinal cord injuries, because of their mobility and other biochemical properties. However, it is unclear how this finding should be considered in the discussion of treatment options in the neurologically injured patient.¹⁰

It has been shown that timely neurologic decompression and stabilization is beneficial in thoracolumbar injuries, but given the relatively good prognosis of nerve root injury, some authors have debated if this can be applied to low lumbar fractures. Several studies solely on low lumbar burst fractures show a spontaneous recovery similar to peripheral nerve injuries with nonoperative care.¹¹ They also showed no difference in the incidence of neurologic deterioration between operative and nonoperative groups, although neurologic deficit was used as the indication for surgery based on the surgeons bias. The one caveat to treating low lumbar fractures nonoperatively with neurologic injury is stability of the fracture. Nonoperative care is not recommended in unstable neurologically compromised patients. Additionally, Finn and Stauffer 12 found no development of late-onset stenosis from nonoperative treatment of low lumbar fractures due to the ability of the canal to remodel even in the setting of significant canal encroachment.

A recent evidenced-based review chapter attempted to delineate if operative decompression influenced the outcome of conus medullaris (CM) and cauda equina (CE) level lesions.¹³ A thorough review of the literature was performed, and the authors found only low-quality studies that identified neurologic injury with subclassification into spinal cord injury (SCI), CM, and CE injuries, and there was no statistical difference in outcome between patients treated with surgery and those treated without surgery. However, the authors did find evidence in several studies of improved recovery with anterior decompression compared with posterior decompression, and this was especially true for patients with CM lesions and bowel and bladder control symptoms. But this came at the cost of increased complications. This finding is particularly intriguing in that laminar fracture and nerve root entrapment is often cited as the cause of neural injury in these patients. These studies highlight one of the differences between thoracolumbar and lumbar injuries, in that neurologic injury may not be as important of a determinant in the decision to operate in the low lumbar spine. Therefore, one must carefully evaluate fracture stability of the low lumbar spine rather than neurologic injury.

Nonoperative Management

The majority of thoracolumbar trauma and lower lumbar fractures can be treated conservatively with very good outcomes. In the upper lumbar and thoracolumbar junction, the brace can be a custom-molded total contact cast, Jewett extension type brace, or a standard thoracolumbosacral orthosis (TLSO). As the fracture level descends past L3, the brace needs to be a TLSO with possible thigh extensions, as the lumbosacral joint must be included for stability. Bracing was historically initiated after a period of bed rest, but is now usually started immediately. Upright films are obtained to ensure no occult instability, and the brace is worn for a period of 8 to 12 weeks.¹⁴

A substantial amount of literature has been devoted to the indications for nonoperative care in the thoracolumbar spine, with good outcomes reported in patients with simple compression fractures, stable burst fractures, bony Chance fractures, or flexion-distraction–type injuries. Controversy surrounds the question of stability of the burst fracture, but most authors agree that greater than 25 to 30 degrees of local kyphosis, greater than 50% bone loss, or greater than 50% canal compromise suggest the need for surgery due to the instability of the PLC. Wood et al¹⁵ performed a prospective study on 53 neurologically intact patients with thoracolumbar burst fractures randomized to brace versus posterior stabilization. They found no difference in radiographic or pain outcomes, but there were fewer patient-reported complications in the nonoperative group. The most recent literature review supports this decision as well, especially in neurologically intact patients, as only one high-quality study showed improved outcomes in surgically treated patients.¹⁶

These indications have been extrapolated to the treatment of the lumbar spine with fairly good results. Several retrospective studies have shown good outcomes with nonoperative treatment of the low lumbar spine burst fractures. Knight et al¹⁷ reviewed patients with low lumbar fractures treated nonoperatively and those treated operatively and found equivalent outcomes, with a decreased time to return to work for the nonoperatively treated patients. The two groups showed significant differences, however, in preoperative injury severity, with those undergoing surgery having greater injury severity scores. Seybold et al¹¹ and Andreychik et al¹⁸ also showed equivalent outcomes in patients with low lumbar fractures treated both ways. These outcomes occurred despite an increase in kyphosis and settling among the non-operative group, which was most common at L3, and an improved kyphosis correction in the surgical group. This can be partially explained by the final follow-up alignment being relatively equal in the two groups in the study by Andreychik et al. Finally, a study concerning the fifth lumbar vertebra found improved radiographic alignment as well as return to work and pain status in the conservatively treated group.19

One caveat is the relative instability of significant coronal split compression fractures in the low lumbar spine. These fractures tend to require surgery in contrast to the majority of compression fractures.

Given the success of nonoperative care and relative stability, some authors question the need for bracing at all. Bailey et al²⁰ undertook a prospective randomized study to look at this question, and compared patients with neurologically intact burst fractures, who underwent nonoperative management either with or without a brace. They found no difference at 2-year follow-up in patient outcomes, pain scores, or average local kyphosis. Although this study did not delineate between thoracolumbar and low lumbar fractures, we can assume that the increased stability of the abdominal musculature and the pelvic girdle would make lumbar fractures even more amenable to not using an orthosis. It is also important to remember that brace treatment does carry certain risks associated with soft tissue injury and pressure ulcers. Patients and caregivers need to be instructed on the appropriate way to wear the brace, and if bracing is used, it should be removed as soon as the fracture is deemed appropriately healed, generally in 8 to 12 weeks.

Operative Management

Although many injuries can be treated conservatively, it is understood that injuries that are either mechanically unstable or neurologically unstable with deficit will benefit from operative fixation. Fracture dislocations, ligamentous flexion-distraction injuries, and unstable burst fractures of the thoracolumbar and low lumbar spine all show improved outcomes with surgical stabilization and decompression of the neural elements when required. There are, however, controversies concerning the approach, the timing, and the method of fixation. This section discusses the differences as they relate to both the thoracolumbar and lumbar injuries. Most of this controversy centers around the burst fracture, as the definition of its stability is much debated.

Multiple approaches have been utilized for both decompression and stabilization of the thoracolumbar spine. An anterior approach offers direct visualization of the fracture and the ability to directly decompress the neural elements. Anterior vertebral body reconstruction can be supported with posterior instrumentation for a 360-degree fusion or can be a stand-alone construct. Alternatively, a posterior approach can be utilized for reconstruction and stabilization of the posterior tension band combined with a laminectomy for dorsal decompression or the transpedicular approach for ventral 360-degree decompression (Fig. 9.1). More recently, costotransversectomies (see Chapter 6) have been utilized to perform a 360-degree decompression and fusion with the placement of an anterior cage placed from the back. Several studies have been performed to examine the relative benefits, and have found both anterior and posterior approaches to be equivalent in terms of neurologic recovery, outcomes, return to work, and deformity correction in the thoracolumbar spine.¹⁶ A meta-analysis found longer operative times, greater blood loss, and more complications with the anterior approach, and these findings seem to also hold true for the lower lumbar spine, except that there are studies that suggest better surgical correction with a combined approach. Korovessis et al²¹ examined burst fractures from L2 to L4 treated with either short segment posterior fixation alone (Fig. 9.2) or with an anterior cage (Fig. 9.3). They found that although the posterior-alone approach was associated with slightly better outcomes and fewer complications, a significant loss of correction of 5 degrees occurred. Given the importance of maintaining sagittal balance, they concluded that this loss was unacceptable and recommended anterior column support in fractures with segmental angulation greater than 12 degrees. Three-column support also enables a short segment of fixation, which is beneficial for preserving motion segments, an important consideration in the mobile lumbar spine. If fractures occur in the low lumbar spine, we recommend short segment fixation, one level above and one level below the fracture, with vertebral body reconstruction if required (Fig. 9.3).

Concerns regarding the effects of fusion on the thoracolumbar spine have led authors to examine nonfusion methods of fixation. Kim et al²² retrospectively examined 23 patients with thoracic and lumbar fractures who were stabilized with pedicle screws that were later removed after fracture healing. They found that the initial sagittal angle changed from 17.2 degrees of kyphosis to 2.8 degrees of lordosis after fixation of fractures. This angle was 1.7 degrees of kyphosis just before implant removal, 2.4 degrees of kyphosis just after implant removal, and 5.9 degrees of kyphosis at the final follow-up. These findings were associated with a very good maintenance of segmental motion, with better results in the lumbar compared with the thoracolumbar spine. Wang et al²³ performed a prospective study showing no significant difference in maintenance of sagittal alignment between a standard posterolateral fusion with pedicle screws and a nonfusion group. In the nonfusion group there was less loss of vertebral body height and less blood loss, and the low back outcome scores were equivalent. Further long-term studies are needed to determine if this loss of kyphosis will be more significant in the low lumbar spine given its importance in sagittal balance, or whether these equivalent outcomes will continue to stand up over long-term follow-up.