Successful Short-Segment Instrumentation and Fusion for Thoracolumbar Spine Fractures

Transcription

1 Successful Short-Segment Instrumentation and Fusion for Thoracolumbar Spine Fractures A Consecutive Year Series Jeffrey W. Parker, MD, Joel R. Lane, MD, Eldin E. Karaikovic, MD, PhD, and Robert W. Gaines, MD SPINE Volume 25, Number 9, pp , Lippincott Williams & Wilkins, Inc. Study Design. A retrospective review of all the surgically managed spinal fractures at the University of Missouri Medical Center during the year period from January 1989 to July 1993 was performed. Of the 51 surgically managed patients, 46 were instrumented by short-segment technique (attachment of one level above the fracture to one level below the fracture). The other 5 patients in this consecutive series had multiple trauma. These patients were included in the review because this was a consecutive series. However, they were grouped separately because they were instrumented by long-segment technique because of their multiple organ system injuries. Objectives. The choice of the anterior or posterior approach for short-segment instrumentation was based on the Load-Sharing Classification published in a 1994 issue of Spine. The purpose of this review was to demonstrate that grading comminution by use of the Load-Sharing Classification for approach selection and the choice of patients with isolated fractures who are cooperative with spinal bracing for 4 months provide the keys to successful short-segment treatment of isolated spinal fractures. Summary of Background Data. The current literature implies that the use of pedicle screws for short-segment instrumentation of spinal fracture is dangerous and inappropriate because of the high screw fracture rate. Methods. Charts, operative notes, preoperative and postoperative radiographs, computed tomography scans, and follow-up records of all patients were reviewed carefully from the time of surgery until final follow-up assessment. The Load-Sharing Classification had been used prospectively for all patients before their surgery to determine the approach for short-segment instrumentation. Denis Pain Scale and Work Scales were obtained during follow-up evaluation for all patients. Results. All patients were observed over 40 months except for 1 patient who died of unrelated causes after 35 months. The mean follow-up period was 66 months (5 1 2 years). No patient was lost to follow-up evaluation. Prospective application of the Load-Sharing Classification to the patients injury and restriction of the short-segment approach to cooperative patients with isolated spinal fractures (excluding multisystem trauma patients) allowed 45 of 46 patients instrumented by the short-segment technique to proceed to successful healing in virtual anatomic alignment. From the Department of Orthopaedic Surgery, University of Missouri, Columbia, Missouri. Acknowledgment date: April 17, First revision date: August 3, Acceptance date: July 2, Device status category: 11. Conflict of interest category: 12. Conclusions. The Load-Sharing Classification is a straightforward way to describe the amount of bony comminution in a spinal fracture. When applied to patients with isolated spine fractures who are cooperative with 3 to 4 months of spinal bracing, it can help the surgeon select short-segment pedicle-screw-based fixation using the posterior approach for less comminuted injuries and the anterior approach for those more comminuted. The choice of which fracture dislocations should be strut grafted anteriorly and which need only posterior shortsegment pedicle-screw-based instrumentation also can be made using the Load-Sharing Classification. [Key words: short-segment instrumentation, fracture, thoracolumbar spine] Spine 2000;25: The development of pedicle screw-based posterior spinal instrumentation systems and successful anterior spinal implants have brought short-segment instrumentation (attachment of one normal vertebra above an injury to one healthy vertebra below an injury) into successful clinical practice. Early reports encouraged the use of these systems but offered only limited guidelines regarding selection of patients for their use. 6,14,24,32 Failure to support the anterior spinal column after posterior correction and instrumentation with pediclescrew-based implants has led to the failure of these implants by breakage, bending, or loosening in many patients. The critical period appears to be the 6 months after the procedure ,31,33 Loss of correction and failure of implants were more common in spine fractures repaired with pedicle screws 1,2,6,9, 14,24,25,31 than in studies that used anterior strut grafting and anterior instrumentation. 18 The mean loss of kyphosis correction ranged from 3 to 12 in the reported pedicle screw studies, whereas the mean loss of correction in the Kaneda studies was only 1. In addition, the failure rate of posterior instrumentation ranged from 9 to 54% in the pedicle screw studies, whereas it was 6% in the Kaneda studies. The authors early experience with short-segment pedicle-screw-based plating for spine fractures indicated that the occurrence of screw fracture was based solely on fracture selection criteria. 17,23 A comprehensive retrospective review of a consecutive series of 28 surgically managed patients allowed the authors to develop a new classification of spinal fractures the Load-Sharing Classification. 23 (Figure 1). The classification was developed by the authors after they recognized that by quantifying the comminution of 1157

2 1158 Spine Volume 25 Number Materials and Methods Figure 1. The Load-Sharing Classification of spinal fractures. the most injured vertebral body (regardless of the mechanism of injury, and without being column-specific regarding the comminution) before surgery, they could predict, with great accuracy, the occurrence of postoperative pedicle screw fractures in spine fractures treated by short-segment instrumentation. All currently used classifications, 10,12,15,16,21,28 including that of the authors, 23 accept the fact that imaging techniques provide only a static view of spinal displacement. Unidentified ligamentous ruptures; spontaneously reduced thoracolumbar subluxations, or even dislocations; and inability to demonstrate the maximal displacement of any given injury by available imaging techniques are limitations of all classifications. The use of the Load-Sharing Classification has made short-segment instrumentation and fusion the authors preferred treatment method for young active people, eliminating pedicle screw fractures from their clinical spine fracture practice. The Load-Sharing Classification was evaluated and found to be reliable and easy to use. 27 The study presented here validates the successful clinical application of the classification and secondarily establishes short-segment instrumentation as a high-quality, low-morbidity, injury-specific treatment technique for isolated fractures in patients cooperative with 3 to 4 months of postoperative bracing. This is a retrospective review of 51 consecutive thoracolumbar fractures managed surgically between January 1989 and July During this time, 46 patients underwent surgery by short-segment instrumentation (Table 1). Five seriously injured patients with multiple trauma had long-segment instrumentation (4 with Luque sublaminar wires and 1 with variable screws and plates [VSP]). All operations were performed by one surgical team under the supervision of the senior author (R.W.G.) at the University of Missouri Hospital and Clinics. The average follow-up period of all 51 patients was 66 months. All the patients were observed to solid radiographic union of their reconstruction. By assiduous communication, none of the patients was lost to follow-up assessment. Except for one patient, who died of unrelated causes 35 months after surgery, each of the 46 short-segment patients was observed more than 46 months (Table 2). The 46 short-segment instrumented patients included 29 patients treated with VSP screws and plates using a posterior approach 19 ; 16 patients treated surgically with the Kaneda device 18,20 using an anterior short-segment approach, and 1 patient whose procedure involved an Isola posterior pedicle screw system (Table 3). Among the 46 short-segment patients, 39 had fractures without translational displacement (burst and Chance fractures). They were treated with VSP plates and screws (23 patients) or the Kaneda device (16 patients) (Table 4). The other 7 of the 46 short-segment patients had fracture dislocations (injuries with translational displacement): Six were treated with VSP (2 T11 T12, 2 T12 L1, and 2 L2 L3 fractures), and 1 was treated with Isola instrumentation (L3 L4). The choice of approach for short-segment instrumentation was prospectively based on the Load-Sharing Classification (Figure 1). Prestressing of the screws through forced compression, distraction, or in situ bending was not used in fracture reduction. During hospitalization all the patients had high-thigh compression hose or pneumatic compression boots (when they became available). Compression hose were recommended until patients were ambulating on a regular basis (usually 4 to 6 weeks after surgery). The patients on whom VSP was used wore a thoracolumbosacral orthosis (TLSO) for 3 months after their reconstruction, whereas the patients with the Kaneda device were braced for 4 months. The single patient on whom the Isola was used wore a TLSO for 6 months after a staged short-segment reconstruction for a very severe fracture dislocation. The singlestage patients (anterior or posterior) were braced and ambulated within a few days after surgery. The three patients who had fracture dislocations with posterior short-segment instrumentation and staged anterior strut grafting had 1 month of bed rest after their reconstruction before ambulation. Statistical analysis was not performed because of the limited number of patients, the multiple variables involved, and the very low complication rate. Indications for Surgical Stabilization. Indications for surgical instrumentation in this entire series of patients included the presence of any one or more of the following:

3 Short-Segment Instrumentation and Fusion Parker et al 1159 Table 1. Demographic, Fracture, and Surgical Treatment Information Regarding Patients in this Series Patient No. Sex/Age (yr) Date of Surgery Level of Injury Type of Fracture (3- Column Classification) Load- Sharing Score Level of Fusion Implant Pre-Op Deformity ( ) Post-Op Correction ( ) Loss of Correction ( ) EBL (ml) F/U Month Complications 1 M/39 5/89 L1 Burst 2,2,3,7 T12 L2 Kaneda 24 K * M/23 5/89 L2 /L3 Fx/dislocation 3,3,2,8 L2 L4 VSP M/23 7/89 T12 Chance 1,1,1,3 T11 L1 VSP 14 K F/15 10/89 L2 Burst 2,2,2,6 L1 L3 VSP 16 K M/24 12/89 L3 Burst 2,2,2,6 L2 L4 VSP 11 K M/19 1/90 L3 Burst 2,2,2,6 L2 L4 VSP 10 K F/32 1/90 T12 Chance 1,1,1,3 T11 L1 VSP 6 L O M/20 7/90 T11 T12 Fx/dislocation 3,3,3,9 T11 L1 VSP 20 K Pseudoarthrosis 9 M/20 10/90 L2 Chance 1,1,1,3 L1 L4 VSP 3 K M/37 1/91 L1 Burst 2,2,3,7 T12 L2 Kaneda 16 K M/23 2/91 L1 Chance 1,1,1,3 T12 L2 VSP 4 K F/31 3/91 L3 Burst 3,3,2,8 L2 L4 Kaneda 8 K F/39 5/91 L1 Burst 2,2,2,6 T12 L2 VSP 14 K F/70 5/91 L1 Burst 3,3,3,9 T11 L1 Kaneda 47 K M/64 5/91 L1 Burst 2,3,2,7 T12 L2 Kaneda 16 K Delayed wound infection 2 yrs Post-Op 16 M/51 7/91 T12 L1 Fx/dislocation 3,3,1,7 T12 L2 VSP 10 K Screw backed out (see Figure 8E) 17 F/16 7/91 L1 Burst 2,1,2,5 T12 L2 VSP 27 K M/53 8/91 T12 Burst 3,2,3,8 T11 L1 Kaneda 35 K M/63 9/91 L1 Burst 2,2,3,7 T12 L2 Kaneda M/20 10/91 L2 L3 Fx/dislocation 3,2,3,8 L2 L4 VSP 10 K Fibrous union 21 F/33 10/91 T10 T11 Fx/dislocation 1,1,2,4 T9 T12 VSP 27 K F/47 12/91 L1 L2 Burst and ant. 1,1,1,3 T11 L2 VSP 10 K comp. fx 23 M/26 12/91 T11 T12 Ant. comp. 1,2,2,5 T10 L1 VSP 2 K and burst fx 24 F/29 1/92 T12 Burst 3,2,2,7 T11 L1 Kaneda 26 K F/15 3/92 L2 Chance 1,1,1,3 L1 L2 VSP 2 K M/15 3/92 T10 Chance 1,1,1,3 T9 T11 VSP 20 K M/36 4/92 L1 Burst 3,3,2,8 T12 L2 Kaneda 20 K F/16 6/92 L1 Chance 2,2,2,6 T12 L2 VSP M/35 7/92 T12 L1 2-level with Multilevel T10 L2 Kaneda 43 K disc injury 30 F/46 7/92 T12 L1 Fx/dislocation 2,2,1,5 T12 L2 VSP Delayed union 31 M/17 8/92 L5 Burst 2,2,1,5 L4 S1 VSP 22 L F/58 8/92 L1 Burst 1,1,2,4 T12 L2 VSP 11 K F/20 8/92 L2 Burst 2,3,2,7 L1 L3 Kaneda M/32 8/92 L1 Chance 1,1,2,4 T12 L2 VSP 18 K F/42 8/92 T12 Burst 1,1,1,3 T11 L1 VSP 15 K M/26 8/92 T12 Burst 3,2,3,8 T11 L1 Kaneda 36 K F/44 11/92 T12 Burst 1,1,2,4 T11 L1 VSP 26 K * F/15 11/92 L3 L4 Fx/dislocation 3,3,3,9 L3 L5 Isola 30 K M/28 11/92 L2 Burst 3,2,2,7 L1 L3 Kaneda 9 K M/28 1/93 L1 Burst 3,3,2,8 T12 L2 Kaneda 14 K F/18 3/93 L1 Burst 2,2,3,7 T12 L2 Kaneda 15 K M/35 4/93 L1 Burst 3,3,3,9 T12 L2 Kaneda 17 K Implant fracture, but bony union/ deceased from gastric carcinoma 43 M/30 6/93 L1 Chance 1,1,1,3 T12 L1 VSP 13 K M/32 7/93 T12 Burst 1,1,2,4 T10 L2 VSP 7 K M/61 7/93 L1 Burst 2,2,2,6 T12 L2 VSP 13 K * F/36 7/93 T11 T12 Fx/dislocation 3,3,3,9 T10 L2 VSP 27 K * Two-stage reconstruction for a fracture-dislocation. Anticipated second-stage procedure not performed. Scores for the individual components of the Load-Sharing Classification and their totals. Pre-Op preoperative; Post-Op postoperative; EBL estimated blood loss (ml); F/U follow-up; Fx fracture; VSP variable screw placement; K kyphosis; L lordosis; ant. comp. anterior compression. presence of neurologic involvement caused by the fracture; evidence of injury to all three columns of the spine; and/or translational displacement at the fracture site (fracture dislocation). Selection of Patients for Long-Segment Instrumentation. Five patients underwent surgery during this consecutive series using long-segment instrumentation (attachment of two or three healthy vertebra above the injury to two or three healthy

4 1160 Spine Volume 25 Number Table 2. Length of Follow-up Length (yr) No. of Patients vertebra below the injury). Four of these long-segment instrumentations were performed using the Luque rectangle and sublaminar wires. The fifth was instrumented using VSP screws and plates to attach two levels above to two levels below the injury. Four of the patients with long-segment instrumentation were multiple trauma victims, near death, with injuries in three or more organ systems in addition to spine fractures with severe translational displacement and immediate and complete paraplegia at the level of the injury. All of them were in intensive care unit (ICU) more than 1 week after surgery. The fifth patient was a well-known unreliable chronic alcoholic with incomplete neurologic defect. The common factor in the selection of patients for long-segment instrumentation was an inability accurately to predict their postoperative compliance with spinal bracing. All of them healed promptly without implant failure. They did not improve neurologically. Selection of the Approach for Short-Segment Instrumentation. Selection of the anterior or posterior approach for short-segment instrumentation was based on application of the Load-Sharing Classification to the patient s injury. 23 The three components of this classification were selected to quantify the immediate postoperative load transfer capacity of the most injured vertebral body itself after spinal fracture correction as well as instrumentation and fusion of one level above and below the fracture with pedicle screws (Figure 1). Three radiographically determined components are assessed to accumulate the point total (score) for any given fracture. The amount of the entire vertebral body involvement (comminution) is the first component of this classification. Injury to one third or less of the vertebral body receives 1 point. Injury of one third to two thirds of the body receives 2 points, and comminution involving more than two thirds of the vertebral body receives 3 points. Quantification of the amount that bony fracture fragments are displaced from the usual vertebral body outline seen on the preoperative axial computed tomography (CT) scans creates the second component of the Load-Sharing Classification. Displacement of the fragments 0 to 1 mm receives 1 point, whereas displacement of 2 mm or more over 50% or less of the vertebral body receives 2 points. Displacement of 2 mm or more over 50% of the body or more receives 3 points. Table 3. Fractures and Fracture Dislocations Treated With Short Segment Instrumentation and Fusion Variable VSP Kaneda Isola No. of patients Age (yr)* Load-Sharing Score 23* EBL (ml)* Bracewear (months)* * Mean values. VSP variable screw placement; EBL estimated blood loss. The third component of this classification is the amount of kyphosis correction necessary to restore physiologic sagittal plane alignment at the level of the injury. Correction of 0 to 3 of correction receives 1 point, whereas 4 to 9 of correction receives 2 points, and 10 or more receives 3 points. This measurement obviously depends on 1) the amount of traumatic kyphosis, and 2) the vertebral level at which the injury exists. Much greater correction is necessary at L4 than at T8. Therefore, any fracture, regardless of mechanism, can be graded at 3 to 9 points. Translational Displacement. To complete the assessment of any fracture, the only remaining task is to identify the presence or absence of translational displacement (Figure 2). If present, translational displacement indicates serious multiple spinal ligament disruption, which the authors use in their classification to define a fracture dislocation. Sometimes the amount of displacement is grotesque and easy to recognize by physical examination. However, in many cases, the displacement is subtle. Careful clinical assessment (local swelling, a palpable defect in supra and interspinous ligaments) and evaluation of pedicle position and/or spinous process malalignment (rotatory displacement) on anteroposterior radiographs can indicate this type of injury. Translational displacement commonly exists without rotational displacement. Of course, the radiographic appearance may be influenced by spontaneous reduction of a fracture on a backboard or radiograph table. Therefore, attention to subtle translation is extremely important. Axial CT scans may show various degrees of facet joint subluxation. The existence of translational displacement (indicating multiple spinal ligament and/or capsular ligament injuries) creates a much more compelling tendency toward operative treatment than a fracture without translation. 22 During the study period, the current authors operated on all the patients who presented with translational displacement. None were treated nonoperatively. However, even with fracture dislocations having severe comminution of the vertebral body (point total of 7 or more), the authors used staged shortsegment posterior fixation and planned an anterior strut graft for their reliable patients (wearing a brace for 3 months). Unreliable patients received long-segment posterior instrumentation alone. Treatment of Patients with Nonfracture Dislocation. Using the Load-Sharing Classification, patients with fractures totaling 6 points or less underwent surgery using a posterior approach with short-segment pedicle-screw-based fixation using VSP plates and screws. 19 Patients without translation who had point totals of 7 or more were treated surgically using anterior short-segment instrumentation with the Kaneda device. 20 The three-column classification 4,5 and the spinal injury level of the these patients are given in Table 4. Treatment of Patients With Fracture Dislocation. Seven patients had translational displacement (i.e., fracture dislocations). They were treated surgically first using the posterior approach, regardless of the severity of vertebral body injury, because it generally is simpler to reduce translational displacement using a posterior approach. One patient (case 30) with a load-sharing score of 5 points was treated only with posterior instrumentation, and healing occurred anatomically. The six remaining patients with fracture dislocation all had

5 Short-Segment Instrumentation and Fusion Parker et al 1161 Table 4. Short-Segment Nonfracture-Dislocation Cases Treated With VSP or Kaneda Device VSP Kaneda Load Sharing Scores 23 Load Sharing Scores 23 Score Patient No. 3-Column Classification 5 Score Patient No. 3-Column Classification 5 T10 11 flex/dist 1 T12 burst T10 Chance L1 burst T12 burst L2 burst T12 Chance L3 burst L1 burst 5 Multilevel 1 Multilevel 1 Total 23 L1 Chance 4 L2 burst 1 Total 16 Total 16 L2 Chance 2 L3 burst 2 L5 burst 1 VSP variable screw placement. Total 23 point totals of 7 or more. Five of these were reduced initially with VSP plates, anticipating an anterior strut graft across the vertebral body in a staged short-segment reconstructive procedure. The sixth patient with fracture dislocation who had a Load Sharing Classification score greater than 7 was instrumented with Isola. Three of these six patients (cases 2, 38, and 46) did have the staged anterior strut graft. The other three patients (cases 8, 16, and 20), for various reasons, did not have the graft performed. Spinal Canal Decompression. If patients had clinical evidence of neurologic damage to the spinal cord or nerve roots, a decompression procedure was performed concomitantly with spinal instrumentation. If the fracture was managed surgically using the posterior, pedicle-screw-based approach, a transpedicular approach was used for decompression. 17,34 If the anterior approach was used for short-segment instrumentation, then anterior decompression was performed. 18 The case numbers, neurologic outcomes, and techniques of decompression are shown in Table 5. Results There were no early wound problems in the entire series of patients. One delayed wound infection developed 2 years after the initial surgery. There were no deaths and no serious postoperative problems in the convalescence of any of the patients. There were no deep venous thromboses or pulmonary embolism. Simple Burst and Chance Fractures Treated With Short-Segment Instrumentation and Fusion Of the 39 burst and Chance fractures (none with translation), 23 patients with point totals of 3, 4, 5, or 6 were treated using the posterior approach with VSP instrumentation. All healed with virtually anatomic alignment. Table 5. Neurologic Status in Patients With Short-Segment Instrumentation and Fusion Patient No. Vertebral Level Frankel Classification Admit Latest F/U Decompression Figure 2. Anteroposterior, posteroanterior, and lateral views of translational displacement in a thoracolumbar fracture. The drawing on the left represents an intact spine, followed by drawings of translation from gross to subtle. 8 L5 D NML TPD 10 L1 C C L1 corpectomy 28 L5 D NML TPD 30 L5 D NML TPD 31 L5 D NML TPD 34 L2 C D Exploration cord 46 L1 C C TPD NML normal exam motor and sensory; TPD transpedicular decompression; F/U follow-up.

6 1162 Spine Volume 25 Number Figure 3. A, Typical flexion distraction injury involves only the top half of the vertebral body (2 points), minimal fragment displacement on the computed tomography (CT) scan, (1 point), and more than 10 correction of kyphosis to restore normal sagittal plane alignment at T12 (3 points). Thus, the point total is 6 points. B, Axial CT scan and sagittal reconstruction are required in assessing point totals. C, The plain radiographs 7 months after surgery show short-segment reconstruction and healing in normal alignment. There were no implant failures. Average preoperative kyphotic deformity was 11 ; the mean intraoperative correction was 12 ; and the loss of correction observed at the last follow-up evaluation was 4 (well within measurement error). There were no broken screws or plates (Figures 3 and 4). The remaining 16 nonfracture dislocations with point totals of 7, 8, or 9 were treated with the Kaneda device. All 16 patients with the Kaneda device healed, and 15 had trivial loss of correction (within the range of measurement error). The average preoperative kyphotic deformity was 23 ; the mean intraoperative correction was 16 ; and the loss of correction seen on the last follow-up evaluation was 4 (Figures 5 and 6). One patient with the Kaneda device showed implant fracture and a 32 loss of correction before healing. These data were not included in the averaged data because this case was an outlier. Tomograms showed union of the iliac crest strut graft and this patient s implants have not been revised. He had a history of sociopathic personality disorder before surgery, did not comply with postoperative bracing, and should not have been chosen for short-segment instrumentation. Fracture Dislocation Treated With Short-Segment Instrumentation and Fusion In this study, there were seven patients with fracture dislocation (injuries with translation) (patients 2, 8, 16, 20, 30, 38, and 46). Patient 30 had a point total of 5. Repaired posteriorly with VSP plates, he healed anatomically without implant-related complications. The remaining six patients had point totals of 7, 8, or 9 in addition to translation. All were treated surgically using a posterior approach first: five (patients 2, 8, 16, 20, and 46) with VSP, and one (patient 38) with Isola because the authors thought it was simpler to reduce the translation posteriorly than anteriorly. Three of these six total high-point patients (cases 2, 38, and 46) had the staged anterior strut graft. All healed anatomically without implant-related complications (Figures 7A to 7F).

7 Short-Segment Instrumentation and Fusion Parker et al 1163 Figure 4. A, Mild burst fracture involves only the top third of the body (1 point), only mild ( 2 mm) fragment displacement over less than 50% of the body (2 points), and correction of 4 to 9 for restoration of normal alignment (2 points). Thus the point total is 5 points, and the fracture is handled successfully by posterior instrumentation. Follow-up films at 11 months and 18 months (after implant removal) document the fusion mass and proper healing. B, Anteroposterior films of the same fracture show preoperative, postoperative, and follow-up radiographs after implant removal. Posterolateral decompression was used to relieve canal encroachment. The other three total high-point patients did not undergo the anticipated strut grafting because of clinical factors individual to each patient. Although each had a fine clinical outcome, and none had implant fracture, all had more loss of correction than the patients with strut grafts, and one had clinically silent backing out of the screws during the postoperative follow-up period (Figures 8A to 8F). Follow-Up Assessment The most recent follow-up assessment was attempted April to June At the most recent follow-up evaluation, three patients had died secondarily to unrelated causes, and five could not be contacted by telephone. All eight of these patients, who were lost to recent follow-up assessment, had been followed for at least 48 months before loss of contact. The remaining 38 short-segment patients were assessed during a follow-up examination or by a telephone conversation using the pain scale of Denis. 5 Of these, 17 patients had no pain; 9 patients had occasional slight pain with no need for medication; 10 patients had moderate pain with a need for occasional medication but no interruption of work or major change in activities of daily living; 1 patient had moderate to severe pain with a need for frequent medication and occasional absence from work or a major change in activities of daily living; and 1 patient had constant or severe incapacitating pain and a chronic need for medication. Postoperative work status was assessed as follows: return to previous activity (25 patients), return to less strenuous work (8 patients), disability after the injury (3 patients), unemployed before the injury (1 patient), and retired before the injury (1 patient). Neurologic Recovery Most of the patients in this study who were partially handicapped neurologically gained one Frankel grade of neurologic recovery as a consequence of decompression, and no patient had iatrogenic neurologic injury (Table 5). Discussion The data from this study document the successful clinical application of short-segment instrumentation for young reliable patients who will tolerate postoperative bracing. All 46 consecutively treated patients with isolated thoracolumbar fractures successfully underwent shortsegment spinal instrumentation, with implant fractures in 1 of 46 patients (2.1%). The authors recent clinical experience and functional data on the patients in this

8 1164 Spine Volume 25 Number Figure 5. A, Typical severe burst fracture involves entire body (3 points). B, Same fracture has more than a 2-mm displacement of fragments beyond 50% of the entire body circumference (3 points), but requires only 4 to 5 of kyphosis correction to restore anatomic sagittal plane (2 points). Thus, the fracture total is 8 points. C and D, The fracture is handled anteriorly by vertebrectomy and an autologous iliac crest strut graft reconstruction with the Kaneda device. study suggest that patients perform better with shorter fusions. This observation is supported by experimental data suggesting that immobilizing long segments of the spine increases the load and motion not only at the immediately adjacent segment, but also at all the distal segments. 29 The authors believe that their success arises from two simple factors: exclusion of patients who were uncontrollable (by virtue of illness or personality) and quantification of comminution by the Load-Sharing Classification for a proper selection of the operative approach for short-segment reconstruction. Five long-segment patients were deliberately included in this report because, as in many other areas of orthopedics, the authors think that patient selection is a fundamental component in the clinical success of a new technique (short-segment instrumentation). They believe that their inability to predict cooperation with postoperative bracing in this particular group of very seriously injured patients represents a very important risk factor

9 Short-Segment Instrumentation and Fusion Parker et al 1165 Figure 6. A, A 21-year-old woman with total body involvement (3 points) with widespread fragments on CT (B) (3 points) that needs more than 10 sagittal plane correction because the injury is at L3 (3 points). The point total is 9. Thus the patient is treated surgically from the anterior approach with Kaneda device and strut graft. Proper healing with no loss of reduction is illustrated at 9 month (C) and 25-month follow-up radiographs (D). predisposing multiple trauma patients to short-segment construct failure. Perhaps this particular group of patients will never become candidates for short-segment instrumentation. The principle of load-sharing came from the authors previously published experimental work. 11 This former study showed that the load transfer across an experimentally repaired spinal fracture site itself (without regard

10 1166 Spine Volume 25 Number Figure 7. A, B, and C, L3 severe fracture dislocation involves the entire body (3 points), has displacement of fragments wider than 2 mm over the entire body (3 points) but requires only 4 to 9 of correction to restore the sagittal plane (2 points). Translational displacement defines this injury as a fracture dislocation, although the displacement is only moderate. D, Initial treatment involves posterior instrumentation, posterolateral decompression with suture of dural laceration and containment of displaced roots, and posterolateral fusion with an autologous iliac crest bone graft. E, F, Second-stage fibular anterior strut autologous graft is carried out 1 week later to support the posterior construct mechanically. Full neurologic recovery and fracture healing is obvious 8 months after surgery.

11 Short-Segment Instrumentation and Fusion Parker et al 1167 Figure 8. A, B, A 44-year-old 114-kg man sustained this relatively innocent-appearing L1 fracture dislocation with subtle but definite translation. The entire body (both endplates) is damaged (3 points). C, D, Fragments are displaced more than 2 mm anteriorly and posteriorly (3 points), but very little sagittal plane correction is necessary (3 ) (1 point). Thus, the point total is 7, indicating a need for anterior short-segment instrumentation. E, The fracture dislocation is initially stabilized posteriorly to realign a translational displacement. Realignment is easier to achieve, particularly if grotesque, from a posterior than from an anterior approach. Postoperative reduction looks so good that the anterior procedure is postponed because the patient s pulmonary function is moderately reduced and he wants to avoid the surgical morbidity. F, Progressive collapse at the fracture site without implant failure is seen. This indicates the value of the classification in predicting the fate of short-segment constructs that neglect load-sharing. The patient is clinically fine.

12 1168 Spine Volume 25 Number for column involvement) was far more important in clinically successful spine reconstruction than the type of implants used to repair the fracture. Subsequent additional work by Carson et al 3,7 reinforced and reiterated the fundamental nature of this load-sharing concept in avoiding spinal implant failure. Before the development of the Load-Sharing Classification, the most widely used fracture classification in the United States was the three-column classification described by Denis 4,5 and McAfee et al 22 after routine CT scanning of fresh spine fractures had been introduced. When CT scanning became commonplace, identification of middle-column injury became simple, and many American surgeons rapidly followed Denis and McAfee et al s suggestion to use the presence of middle-column injury as the sole indication for surgical treatment. The current authors clinical and laboratory experience along with a large body of pre-ct clinical information suggested that isolated middle-column injury is related more directly to the presence of neurologic injury in patients with spinal fracture than to the load transfer properties of the spine. Their experience also demonstrated that many patients with middle-column involvement but no neurologic deficit and minimal or modest comminution of the entire body can very easily be managed nonoperatively. They concluded that the threecolumn classification suggested operative treatment for many fractures that could be handled nonoperatively, ignoring the comminution of the whole vertebral body while placing undue biomechanical emphasis on the middle column. Once short-segment instrumentation became surgically possible, the liabilities of the three-column classification quickly became evident to the current authors and others. 8,17,25,26 Patients who underwent surgery using short-segment pedicle-screw-based instrumentation for middle-column injury showed implant fracture early in their clinical course when total body comminution was ignored and the focus centered only on injury to the middle column. The article by McLain et al 25 shows this best. None of the patients in McLain et al s 25 series who had minimally comminuted injuries or a strut graft experienced postoperative collapse or implant problems regardless of the time from injury. In contrast, patients with higher degrees of vertebral body comminution had average of 10 correction loss, resulting in more symptoms of pain and more severe symptoms, which required reoperation in some cases. 25 The current authors emphasis on discriminating fractures according to (even very subtle) translational displacement is underscored by the recent report by Gertzbein, 13 who clearly showed that the spinal injuries most prone to neurologic deterioration after admission to the hospital, but before operative treatment, were those showing even mild translational displacement. It must be emphasized that, unlike other spine fracture classifications, the authors do not use their classification, by itself, to make decisions regarding surgical or nonsurgical care. Because the Load-Sharing Classification only quantifies comminution, and does not identify ligamentous disruption (a very important component of operative decision making), the authors never make operative/ nonoperative decisions based solely on its use. Only after a thorough physical; neurologic and spinal examination; and thorough patient evaluation regarding prior activities, social and educational background, and future plans do they review the patient s radiographs and CT scan to determine the risks and benefits of operative versus nonoperative care. If operative care is considered, they determine whether the patient is reliable enough for short-segment instrumentation and fusion. Used in this way, the Load-Sharing Classification is a helpful adjunctive tool that can complement but not replace other classifications. Although the authors have used and respect all previous spine fracture classifications, 5,10,12,16,21,28 they believe that the Load-Sharing Classification s focus on comminution, irrespective of column specificity or mechanism of injury, as well as its simple construction and formulation can direct the clinician toward successful anterior or posterior short-segment instrumentation for patients with isolated spine fractures or fracture dislocations more simply than previous classifications. Their data support their successful application of this approach. The authors further believe, but cannot yet prove, that the Load-Sharing Classification, by quantifying comminution, also can quite likely predict the anatomic and perhaps also the clinical results of nonoperative spinal fracture treatment. Sarmiento et al 30 has clearly shown that the structural results (shortening and deformity) from nonoperative treatment of the intertrochanteric femur, the tibial plateau, and both bone forearm fractures depends entirely on 1) initial fracture displacement and 2) comminution. Because the three components of the Load-Sharing Classification deal directly with comminution, the current authors plan to study the ability of the Load-Sharing Classification to predict posttraumatic kyphosis prospectively, and thereby the functional outcome in nonoperatively treated patients. Conclusions The current study demonstrates the use and clinical success of short-segment (one above attached to one below) instrumentation and fusion for spinal fractures using the Load-Sharing Classification for selection of the surgical approach. A low load-sharing score of 6 points or less indicates adequate sharing of load through the injured vertebral body itself along with the implant to permit only posterior pedicle-screw-based short-segment instrumentation and fusion. A load-sharing score of 7 points or more indicates poor transfer of load through the most injured vertebral body and points to the necessity for anterior instrumentation and strut grafting. High loadsharing scores for fracture dislocations define the need

13 Short-Segment Instrumentation and Fusion Parker et al 1169 for anterior strut grafting (autologous iliac crest and/or fibula) after posterior reduction when pedicle-screwbased short-segment instrumentation and fusion is performed. Key Points A general assessment of comminution (The Load-Sharing Classification) is the most successful way to predict clinically successful short-segment thoracolumbar spinal fracture repair. Fractures with mild comminution can be successfully repaired from the posterior approach with pedicle screw-based implants. Severely comminuted fractures must be repaired by an anterior approach with vertebrectomy and strut grafting. Bracing for 4 to 6 months postoperatively is a necessary part of successful short-segment spine fracture reconstruction. Fracture-dislocations (injuries with translation) are best initially instrumented short segment from the posterior approach. If the Load-Sharing point total is 7 or higher, then a vertebrectomy and anterior strut graft are applied later. Long-segment fracture repair is used for patients with unpredictable postoperative compliance. Fracture assessment alone (radiograph and CT scan) is never used alone to decide on fracture treatment. Patient-specific comorbidities are too important to ignore. References 1. Benson DR, Burkus JK, Montesano PX, Sutherland TB, McLain RF. Unstable thoracolumbar and lumbar burst fractures treated with the AO fixateur interne. J Spinal Disord 1992;5: Carl AL, Tromanhauser SG, Roger DJ. Pedicle screw instrumentation for thoracolumbar burst fractures and fracture dislocations. Spine 1992;17:S Carson WL, Duffield RC, Arendt M, Ridgely BJ, Gaines RW. Internal forces and moments in transpedicular spine instrumentation, the effect of pedicle screw angle and transfixation, the 4R 4bar linkage concept. Spine 1990;15: Denis F. The three-column spine and its significance in the classification of acute thoracolumbar spine injuries. Spine 1983;8: Denis F. Spinal stability as defined by the three-column spine concept in acute spinal trauma. Clin Orthop 1984;189: Devito DP, Tsahakis PJ. Cotrel-Dubousset Instrumentation in Traumatic Spine Injuries. Proceedings of the Sixth International Congress on Cotrel- Dubousset Instrumentation. Montpellier: Sauramps Medical, 1989: Duffield RC, Carson WL, Chen LY, Voth B. Longitudinal element size effect on load sharing, internal loads, and fatigue life of trilevel spinal implant constructs. Spine 1993;18: Ebelke DK, Asher MA, Neff JR, Kraker DP. Survivorship analysis of VSP spine instrumentation in the treatment of thoracolumbar and lumbar burst fractures. Spine 1991;16: Esses SI, Botsford DJ, Kostuik JP. Evaluation of surgical treatment for burst fractures. Spine 1990;15: Ferguson RL, Allen BL Jr. A mechanistic classification of thoracolumbar spine fractures. Clin Orthop 1984;189: Gaines RW, Carson WL, Satterlee CC, Groh GI. Experimental evaluation of seven different spinal fracture devices using nonfailure stability testing: The loadsharing and unstable mechanism concepts. Spine 1991;16: Gertzbein SD. Classification of thoracic and lumbar fractures. Spine 1994; 19: Gertzbein SD. Neurologic deterioration in patients with thoracic and lumbar fractures after admission to the hospital. Spine 1994;19: Gillet P, Meyer R, Fatemi F, Lemaire R. Short-Segment Internal Fixation Using CD Instrumentation With Pedicular Screws: Biomechanical Testing. Proceedings of the Sixth International Congress on Cotrel-Dubousset Instrumentation. Montpellier: Sauramps Medical, 1989: Harms JJ. Classification of fractures of the thoracic and lumbar vertebrae: (Klassifikation der BWS- und LWS-Frakturen). Fortschr Med 1987;105: Holdsworth F. Fractures, dislocations, and fracture dislocations of the spine. J Bone Joint Surg [Am] 1970;52: Holt BT, McCormack T, Gaines RW. Short-segment fusion: Anterior or posterior approach? The load-sharing classification of spine fractures. Spine. State Art Rev 1993;7: Kaneda K, Taneichi H, Abumi K, Hashimoto T, Satoh S, Fujiya M. Anterior decompression and stabilization with the Kaneda device for thoracolumbar burst fractures associated with neurological deficits. J Bone Joint Surg. [Am] 1997;79: Karaikovic EE, Gaines RW. Short-segment fixation using VSP plates and pedicle screws for trauma. In: Brown CW, McCarthy RE, eds. Spinal Instrumentation Techniques. Rosemont, IL: Scoliosis Research Society, Karaikovic EE, Kaneda K, Akbarnia BA, Gaines RW. Kaneda instrumentation for spinal fractures. In: Bridwell KH, DeWald RL, eds. The Textbook of Spinal Surgery. 2nd ed. Philadelphia: Lippincott-Raven, 1997: Magerl F, Aebi M, Gertzbein SD, Harms J, Nazarian S. A comprehensive classification of thoracic and lumbar injuries. Eur Spine J 1994;3: McAfee PC, Yuan HA, Fredrickson BE, Lubicky JP. The value of computed tomography in thoracolumbar fractures. J Bone Joint Surg [Am] 1983;65: McCormack T, Karaikovic E, Gaines RW. The load-sharing classification of spine fractures. Spine 1994;19(15): McKinley LM, Obenchain TG, Roth KR. Loss of Correction: Late Kyphosis in Short-Segment Pedicle Fixation in Cases of Posterior Transpedicular Decompression. Proceedings of the Sixth International Congress on Cotrel-Dubousset Instrumentation. Montpellier: Sauramps Medical, 1989: McLain RF, Sparling E, Benson DR. Early failure of short-segment pedicle instrumentation for thoracolumbar fractures. J Bone Joint Surg [Am] 1993;75: McNamara MJ, Stephens GC, Spengler DM. Transpedicular short-segment fusions for treatment of lumbar burst fractures. J Spinal Disord 1992;5: Moore KD, Gaines RW. Intraobserver reproducibility and interobserver reliability of the load sharing classification of spine fractures. Spine 1998 (submitted for publication). 28. Morsher E. Classification of spinal column injuries (Classification von Wirbelsäulenverletzungen). Orthopäde 1980;9: Nagata H, Schendel MJ, Transfeldt EE, Lewis JL. The effects of immobilization of long segments of the spine on the adjacent and distal facet force and lumbosacral motion. Spine 1993;18: Sarmiento A, McKellop HA, Llinas A, Park SH, Stetson W, Rao R. Effect of loading and fracture motions on diaphyseal tibial fractures. J Orthop Res 1996; 14: Sasso RC, Cotler HB, Reuben JD. Posterior fixation of thoracic and lumbar spine fractures using DC plates and pedicle screws. Spine 1991;16:S Steffee AD, Biscup RS, Sitkowski DJ. Segmental spine plates with pedicle screw fixation: A new internal fixation device for disorders of the lumbar and thoracolumbar spine. Clin Othop 1986;203: Stephens GC, Devito DP, McNamara MJ. Segmental fixation of lumbar burst fractures with Cotrel-Dubousset instrumentation. J Spinal Disord 1992;5: Viale GL, Silvestro C, Francaviglia N, et al. Transpedicular decompression and stabilization of burst fractures of the lumbar spine. Surg Neurol 1993;40: Address reprint requests to Robert W. Gaines, MD University of Missouri Department of Orthopaedic Surgery One Hospital Drive Columbia, MO 65212