From Radiographic Evaluation to Treatment Decisions in Neurologically Intact Patients With Thoraco-lumbar Burst Fractures

Abstract

We propose that the key to improving care for these patients is to truly understand the processes that take place from the interpretation of radiographic findings, through the assessment of the severity of various injuries, to inclusion within a classification category and finally to selecting a specific treatment.

A result of the complex anatomy of the spinal column is that fractures of the thoracolumbar spine, by their very nature, present with an almost infinite heterogeneity of morphologic characteristics affecting one or several vertebral segments, facets and posterior musculoligamentous structures as well as the anatomically unique intervertebral disc with its supporting longitudinal ligaments. When faced with this seemingly endless array of injury complexity, surgeons have attempted to identify patterns based on either implied injury mechanisms ¹ or radiographic morphology^2,3 that they use to classify patient injuries into categories within classification systems and then use these to guide therapeutic recommendations. Despite multiple iterations of classification systems, algorithms and scoring systems, there remains wide variability in the treatment of injuries, even those that may appear to be similar morphologically.^4–7

The AO Spine Trauma Knowledge Forum (AOSKFT) is an international group of spine surgeon who all focus a substantial part of their clinical and academic activity on the management of spine trauma. The AOSKFT has devoted significant time and attention to the processes of classification leading to treatment recommendation and utilizing various scoring systems to guide recommendations of specific treatments for specific categories of injury.^6–18 Often these recommendations are based upon the best available evidence, but when the evidence is lacking or contradictory, clinicians rely on their personal experience, expert opinion, and the method of treatment that is taught and practiced among practitioners in their specific region, language, and provider community.

Within the AOSKFT and within the broader spine trauma expert medical community there is general agreement that injuries exhibiting translation (AO type C injuries) and/or neurological impairment are likely to require surgical treatment, however, in the case of neurologically intact fractures of the vertebral body there is little consensus.^19–23 There are many reported protocols for non-operative management of these injuries^23–25 and an equal or greater variety and complexity of surgical treatment options^26–29 that are advocated.

Even among the members of the AOSKFT there are some who believe passionately that surgery provides the optimal outcome for the patient based on their experience, training, and local practice, while others support non-operative treatment with equal certainty. Without addressing the specifics of treatment standardization, even in the broad categories of Surgery or No Surgery, there is widespread disagreement with some surgeons treating ‘almost all these injuries’ surgically and others treating almost all non-surgically. The landscape related to treatment of neurologically intact thoracolumbar burst fractures perfectly fits the definition of Equipoise proposed by Freeman in 1987 where a “true state of equipoise exists when genuine uncertainty exists within the expert medical community—not necessarily on the part of the individual investigator—about the choice between two or more care options”. ³⁰

It is in this environment of widespread variability in treatment approach for seemingly similar injuries that the methodological puritans would state categorically that what is needed is a properly designed randomized controlled surgical trial (PRCT) and this is justified by the existence of widespread inconsistency and local area variation in use of specific treatment approaches; a condition that they would describe as clinical equipoise and would use to justify randomization. In reality, it is difficult to find individual surgeons who are both skilled and comfortable in surgical and non-surgical management and do not hold a strong internal preference for one treatment method over the other for specific cases. This and issues of standardization, lack of blinding, restrictive inclusion criteria, patient and provider preference makes a surgical randomized trial in TL burst fractures extremely difficult, if not practically impossible. ³¹ It is in this environment of a high degree of “communal equipoise” and concomitant high degree of individual provider certainty where a large scale prospective observational study can benefit from equipoise and create a study design superior to the PRCT in terms of its generalizability and assessment of effectiveness in the real world.^19,20

There can be a multitude of explanations for why there is vast variability in treatment, and it is the objective of this focus issue to utilize a novel conceptual framework and methodology (outlined in the first article in this focus issue) to try to understand wherein lies the breakdown in this seemingly sound scientific method of classification, scoring and subsequently making a consistent treatment recommendation.

Within the AOSKF Trauma (AOSKFT) we are fortunate to have 2 unique opportunities that at their nexus enable the investigations reported in this focus issue. The first is that we have within the AOSKFT a community of expert clinicians from around the world that make results internationally generalizable and reflect the polarized individual therapeutic preferences that surgeons display when treating TL Burst fractures. It is the members of the AOSKFT that have come together to form the 22 clinical experts who are described in the methodology paper.

The second opportunity is that we have collected radiographs including plain films, CT scans, and some MRI’s prospectively on 183 patients with thoracolumbar burst fractures. These images are from an observational clinical trial titled: Thoracolumbar burst fractures (AO Spine A3, A4) in neurologically intact patients: An observational, multicenter cohort study comparing surgical vs non-surgical treatment. (Spine TL A3/4 Study; ClinicalTrials.gov Identifier: NCT02827214). This prospective multicentre observational cohort study that relied on each recruiting surgeon to provide what they believed to be their optimal standard of care had radiographs and CT scans collected as one of the secondary outcomes to be analysed. The Spine TL A3/4 study results as of the date of this focus issue have not been published.

The data from the Spine TL A3/4 study has been under the stewardship of a CRO and on our request the CRO made available only the actual treatment that the 183 patients received from their treating surgeon. No outcome data was shared or discussed. It is the imaging of these 183 prospectively collected patients that form the substrate for the investigations in this issue. This is discussed in detail in the first article in this focus issue which provides a conceptual framework for these investigations.

Heterogeneity of the morphologic appearance of these spinal column injuries is a recognized clinical reality with some resulting from high energy injury, others from simple falls from standing height. With some reliability,^13,32–36 surgeons have been able to identify patterns from plain radiographs, computerized tomography scanning and other advanced imaging modalities such as magnetic resonance imaging. It is plausible that low reliability is the root of the difficulty leading to therapeutic inconsistency. Several articles in this focus issue^3,4 will consider the pivotal issue of reliable interpretation of fracture morphology.

The classification of these injuries based upon the morphological AO Spine Thoracolumbar Trauma classification may be inherently flawed in that the categories and subtypes may not be relevant or may not lend themselves to clear unambiguous categorization of these injuries. The varying severity of injury and the propensity to fail treatment through development of deformity, pain or neurology may not be inherently represented within the categories of the current classification. There may be improvements that could be applied to this classification that would heighten its accuracy reliability and improve consistency and clinical relevance and this is considered in several articles.^5–7

The process of utilizing a scoring system to transition between types of injuries as described in a classification and the specific treatment recommendation is often based upon expert opinion, occasionally supplemented by a Delphi process. This step could include other variables to assist and improve therapeutic consistency such as including ‘cut-off’ values of radiographic measures such as % body comminution or likelihood of PLC injury. This is explored in an article which attempts to formulate a predictive algorithm. ⁸

Finally, it may be that surgeons are fundamentally motivated by variables outside the categories of classifications and radiographic findings and through their clinical background, education, local practice and peer expert opinion have come to believe that their preferred treatment leads to optimal outcomes. This belief may be strengthened in an environment where most of these fractures heal over time regardless of treatment and patients with these injuries generally improve and exhibit a high degree of general satisfaction with treatment. Anecdotal or individual experiences may strengthen these beliefs and this is also explored in this focus issue. ⁹

We propose that the key to improving care for these patients is to truly understand the processes that take place from the interpretation of radiographic findings, through the assessment of the severity of various injuries, to inclusion within a classification category and finally to selecting a specific treatment.

The papers in this focus issue will attempt to determine the role of certain radiographic characteristics and classifications in guiding surgeons in their decision-making process and how these characteristics could be integrated into improved classifications, scoring systems and guidelines.