Early Failure of Short-Segment Pedicle Instrumentation for Thoracolumbar Fractures

Transcription

1 ( irpyright 993 by ii:, Journal of hone and Joint Surge rv, Incorporated Early Failure of Short-Segment Pedicle Instrumentation for Thoracolumbar Fractures A PRELIMINARY REPORT* BY ROBERT F. M1.AIN. M.D.t. E1)WARI) SPARLING. M.D4. AND DANIEL R. BENSON, M.D.t, SACRAMENTO, CALIFORNIA In vestigatioli J)erfOrfllt (I at i/to!)epartrnent of Orthopaedic Surgery, University of California, Davis, Sacramento ABSTRACT: The results after treatment of fiftytwo lumbar and thoracolumbar fractures with Cotrel- Dubousset instrumentation were reviewed as part of an ongoing study. Nineteen patients (average duration of follow-up, fifteen months) had been managed with short-segment pedicle-screw instrumentation. This preliminary report outlines the complications and pitfalls identified during the initial healing phase in this subgroup of patients. There were no neurological or vascular injuries due to placement of the pedicle screws, but ten patients had some form of failure of the fixation during the early period of healing. Failure of the fixation was manifested in three ways: progressive kyphosis secondary to the bending of screws (six patients), kyphosis secondary to osseous collapse or vertebral translation without bending of the hardware (three patients), and segmental kyphosis after a caudad screw in the lumbar construct broke (one patient, who had had a combined instrumentation for multiple fractures). Untreated anterior instability, and pre-stressing of the screws when the rods were contoured in situ, resulted in a high rate of failure. The high rate of failure of the hardware associated with this fixation construct suggests that posterior screw fixation alone may not be adequate when Cotrel- Dubousset instrumentation is used for short-segment lumbar arthrodeses. Bent screws or measurable kyphosis did not always herald a clinical failure, but patients who had progressive kyphosis of more than 10 degrees had substantially more pain than did those who had little or no progression. The results reported here are preliminary, and speculation as to the importance of these findings and as to the long-term outcome in these patients would be premature. Nonetheless, these early results suggest that surgeons should exercise cantion when applying short-segment pedicle-screw constructs to treat unstable thoracolumbar fractures. *No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study. tdepartment of Orthopaedic Surgery. University of California, Davis, 223OStockton Boulevard, Sacramento, California Please address requests for reprints to Dr. McLain. tdepartment of Orthopaedics. University ofarkansas,4301 West Markham, Little Rock. Arkansas The use of Cotrel-Dubousset instrumentation for the treatment of fractures involving the thoracolumbar spine is gaining popularity. In particular, short-segment pedicle-instrumentation constructs have been widely endorsed at international symposia2. Unfortunately, to our knowledge, no long-term follow-up reports are available to validate the safety and efficacy of these constructs in the treatment of fractures. Our preliminary follow-up study of patients managed with short-segment pedicle instrumentation revealed an alarming rate of early failure of the screw fixation as well as of progressive kyphosis. This prompted us to report our experience with this technique before completing our long-term follow-up study. Materials and Methods More than 800 spinal fractures were treated at the University of California, Davis, Medical Center between November 1987 and October Approximately 20 per cent of the fractures were treated operatively because of instability, actual or impending neurological compromise, or polytrauma necessitating early mobilization of the patient. We reviewed the charts, radiographs, and clinical records of fifty-two patients who had been managed consecutively with Cotrel-Dubousset instrumentation for thoracolumbar fractures and fracturedislocations. Nineteen of the fifty-two patients had been managed with a short-segment pedicle-instrumentation construct. The duration of follow-up in this subgroup averaged fifteen months (range, five to twenty-eight months) at the time of this preliminary study. The primary indications for application of the shortsegment pedicle-instrumentation construct were an Unstable fracture for which instrumentation was needed to restore mechanical stability or to protect neurological elements;a fracture at the thoracolumbar or lumbar level for which instrumentation was needed caudad to the level of injury; and a young, active patient in whom an extensive lumbar arthrodesis would prove detrimental over time. Patients who had thoracic injuries, a complete lesion of the cord, or pre-existing degenerative disease and were not likely to benefit from attempts to spare lumbar intervertebral motion were managed with alternative techniques and were not included in the study. I 62 THE JOURNAL OF BONE AND JOINT SURGERY

2 EARLY FAILURE OF SHORT-SEGMENT PEDICLE INSTRUMENTATION FOR THORACOLUMBAR FRACTURES 163 In eleven patients, the instrumentation spanned three vertebrae, one cephalad and one caudad to the fracture. In two patients who had contiguous fractures, four vertebrae were arthrodesed. In the six remaining patients, the fixation was stopped immediately caudad to the fracture to spare lumbar intervertebral motion, but it was extended more cephalad into the thoracic spine, so that it spanned five vertebrae. In one of the six patients, a long instrumentation construct was applied to stabilize both a thoracic fracture and a lumbar fracture; the lumbar fixation consisted of pedicle screws immediately cephalad and caudad to the fracture, and the patient was included in this subgroup. The instrumentation construct most often used in these patients was characterized by the application of pedicle screws immediately cephalad and caudad to the fracture site, so that the arthrodesis involved a short segment. Screws were forty or forty-five millimeters long, depending on the level and size of the vertebra. Six-millimeter-diameter screws were used at the eleventh and twelfth thoracic levels, and seven-millimeterdiameter screws were used at the first lumbar level and caudally. In all of the patients, only one segment was included in the instrumentation caudad to the fracture, Fi;. Radiograph on which three forces acting on the short-segment construct to produce deformity are illustrated. Axial loading of the ccphalad vertebral body results in settling of the vertebra on the screw (upper arrow); in this patient. the vertebral body settled until the screws engaged the superior end-plate. Translational forces result in a quadrilateral shift. with or without loosening of screws, resulting in anterior spondylolisthesis as the intact vertebra above slides forward on the fractured end-plate below (center arrow). Anterior instability. combined with the intact posterior column. results in a bending moment as axial load is applied (lower arrow). This bending moment is focused at the bone-screw interface, resulting in a bent screw (seen here) or loosening of the screw. I Short-segment Cotrel-Dubousset fixation was used in fifteen male and four female patients; the ages ranged from fifteen to fifty-eight years (mean, twenty-nine years). All had sustained the fractures as a result of major trauma. Twelve of the patients had been injured in a motor-vehicle accident; two, in an automobilepedestrian accident; and five, as a result of a fall from a height. Thirteen patients had a single-level burst fracture; five, a fracture-dislocation; and one, multiple lumbar burst fractures at non-contiguous levels. The levels of the fractures ranged from the twelfth thoracic to the fifth lumbar vertebra: the twelfth thoracic level was involved once; the first lumbar level, nine times; the seeond lumbar level, four times; the third lumbar level, three times; and the fourth and fifth lumbar levels, once each. The patient who had multiple burst fractures at non-contiguous levels had involvement of both the seeond and the fourth lumbar vertebrae. Seven patients had associated fractures of the spine at other levels. FIG. 2 Case 16. Intraoperative radiograph showing a burst fracture at the second lumbar level, stabilized with Cotrel-Dubousset short-segment pedicle instrumentation. in situ contouring of the rods reduced the residual kyphosis to 4 degrees but resulted in bending of both cephalad screws and possibly one of the caudad screws. Anterior strutgrafting was subsequently performed. which prevented progressive bending of the screws or kyphosis. VOL. 75-A, NO. 2, FEBRUARY 1993

3 I 64 R. F. M lain. EDWARD SPARLING. AND D. R. BENSON Ft;. 3-A FIG. 3-B Figs. 3-A and 3-B: Case 10. Fig. 3-A: Initial radiograph. made with the patient weight-bearing. showing a burst fracture at the first lumbar level treated with shortsegment pedicle instrumentation and in situ bending of the rods. The kyphosis measures 8 degrees. The cephalad screws were noted to bend during the contouring, and the caudad screws also show a slight deformation. Fig. 3-B: Radiograph made six months postoperatively. showing progressive deformity with 24 degrees of kyphosis. All screws had bent progressively. but there is no evidence that they pulled out. At eighteen months. the patient had a 28-degree kyphosis. which was painless but clinically apparent. minimizing the number of lumbar levels that were arthrodesed. Two patients had an anterior vertebrectomy, decompression of the cauda equina, and reconstruction with a strut graft at the time of the short-segment pedide instrumentation. The instrumentation was applied bilaterally, and cross-links (transverse traction devices) were used cephalad and caudad to the fracture to augment torsional rigidity. All patients were managed with immobilization in a custom-molded thoracolumbosacral brace for Six months postoperatively. The patients were followed with serial physical examinations and radiographs. With the use of lateral radiographs centered over the fracture level, kyphosis or lordosis was measured from the superior end-plate of the intact vertebra cephalad to the fracture to the infenor end-plate of the vertebra caudad to to the fracture. Scoliosis was similarly measured on posteroanterior radiographs. with use of the Cobb method. Progressive deformity was measured as the change in sagittal alignment of the spine from the initial postoperative radiographs, made with the patient weight-bearing, to the most recent follow-up radiographs. Progression was considered to be absent, minor, or major. Minor progression was defined as kyphosis measuring 5 to 10 degrees more at the time of follow-up than it had on the initial postoperative radiographs. Major progression was defined as an increase of 10 degrees or more compared with the measurement on the initial postoperative radiographs. Tomograms were made when the findings on the plain radiographs were equivocal and when there was clinical evidence of pseudarthrosis. A solid fusion without progressive deformity or failure of the implant was considered to be a successful result of the instrumentation. Failure or bending of the implant, or development of a major kyphosis before fusion occurred, were considered failures of fixation regardless of the duration of follow-up. Parameters of long-term outcome, such as neurological recovery or functional result, were not considered in this review. No valid statistical analysis of the results could be performed due to the small number of patients. Results All of the patients were followed until either the fixation failed or fusion was obtained. The duration of follow-up averaged fifteen months (range, four to twenty-eight months). Nine of the patients had solid fusion without progressive deformity (average duration THE JOURNAL OF BONE AND JOINT SURGERY

4 EARLY FAILURE OF SHORT-SEGMENT PEDICLE INSTRUMENTATION FOR THORACOLUMBAR FRACTURES 165 TABLE RESULTS OF SHORT-SEGMENT PEDICLE INSTRUMENTATION IN NINETEEN PATIENTS I Case Sex. Age (Yrs) Duration Stability of Kyphosis (Degrees) Level of Level of of Failure of Anterior Latest Time to Injury Arthrodesis Follow-up Implant Column Progression Follow-up Failure* (Most) (Mo.s) Pain at Latest 1 M. 19 L3 L2-L M.26 Li TW-L M.20 L3 L2-L M.35 LI T1O-L M. 15 L2 L1-L I 6 F.21 L2 L1-L F.15 Li T1I-L II F.24 LI Tl0-L (a M.32 LI T11-L M.18 LI T12-L I I M. 35 L3 L2-L M.3I T12 Til-LI M.47 LI T10-L I II I I 14 M.46 L2 L1-L M.32 L4 L3-L F.26 L2 L1-L M, 18 L5 L4-S M.27 LI T10-L M.58 Li T10-L *The interval from the operative fixation to the time of documented failure of a screw or of clinical kyphosis. to = no pain: I = mild. intermittent pain without a need formedication and nolimitation ofactivity: 2 moderate pain. necessitating occasional medication or limitation of activity: and 3 = severe, constant pain. Back Follow-upt Lower Limb of follow-up, fifteen months), and ten patients had a measurable loss of reduction (more than 5 degrees). Stability of the anterior column played a pivotal role in the success or failure of the procedure. None of the six patients who had an intact or restored anterior column had measurable progression. These patients (two of whom had been managed with anterior strut grafts, two of whom had had flexion-distraction injuries, and two of whom had had fracture-dislocations without collapse of the anterior column) had, on the average, less than 1 degree of angular change and no measurable progressive kyphosis. The thirteen patients who had residual instability of the anterior column had an average of 10 degrees of progression. The progressive deformity developed during the first six postoperative months in ten of these thirteen patients: two of the ten had minor progression, and eight had major progression (more than 10 degrees). Three of the eight patients had more than 15 degrees of progression. Two patients who had major progression were subsequently managed with additional procedures - one, for anterior collapse and neurological deterioration, and one, for pain and failure of the fixation - and, at the time of writing, late decompression was being considered for two others because of poor neurological recovery and persistent pain. The primary cause of the progressive deformity was failure of the fixation construct. There were three modes of failure: bending or breakage of screws, loosening or pull-out of screws, or translation of vertebrae that had been included in the instrumentation. In some patients, more than one of these modes of failure was involved (Fig. 1). Bent screws were the most common source of failure. In many patients, the bending was subtle and of doubtful clinical importance, but six patients had Severely bent screws, leading to a major kyphosis. Screws bent or broke at the junction of the tapered hub and the threaded shaft. When screws both cephalad and caudad to the fracture site deformed, marked kyphosis ensued. Screws pulled out in only one patient, who also had translational deformity and failure of other screws. In this patient, the pull-out was secondary to the translation and not a primary mode of failure. There was a rapid rate of failure - that is, in the early postoperative period - with no additional failures after six months. In all, sixty-seven pedicle screws were placed immediately cephalad or caudad to the fracture in the nineteen patients. Nine (29 per cent) of the thirtyone screws that had been placed cephalad to the fracture were bent at the time of the most recent follow-up: thirteen (36 per cent) of the thirty-six screws that had been placed immediately caudad to the fracture bent. and two of them broke. Bent or broken screws were found in ten patients at the time of follow-up. In six of the eight patients who had major progression of deformity, bending of screws was the primary mode of failure. Pre-stressing of the pedicle screws may have contnbuted to their failure in three patients. The rod had been contoured in situ, to restore lordosis, in two patients. The screws bent in situ in both patients, who subsequently had progressive failure as the spinal column settled into kyphosis (Figs. 2, 3-A, and 3-B). The third patient had been managed with axial distraction VOL. 75-A, NO. 2, FEBRUARY 1993

5 166 R. F. M LAIN. EDWARD SPARLING. AND D. R. BENSON through the pedicle screws to restore vertebral height and reduce compromise of the canal by ligamentotaxis. Although the rods were not contoured, the cephalad screws were bent during the distraction, and progressive kyphosis subsequently developed. The patients who had major progressive kyphosis had more symptoms of pain and more severe symptoms than those who had little or no progression (Table I). Of the patients who had 10 degrees of progressive deformity or more. three were asymptomatic. three had mild or moderate symptoms. and two had severe pain. Of the patients who had less than 10 degrees of progression, eight were asymptomatic, three had mild symptoms, and none had moderate or severe pain. Discussion There is little information in the orthopaedic literature regarding the use of Cotrel-Dubousset instrumentation in the treatment of fractures. Clinical studies that have specifically addressed the use of Cotrel-Dubousset instrumentation have mostly focused on longer constructs. McBride reported excellent results after the use of Cotrel-Duhousset instrumentation for thoracolumbar fractures, but the constructs in his study spanned at least three normal vertebrae cephalad and two caudad to the fracture site, according to traditional principles established for the use of Harrington rods. The clinical outcomes of the use of short-segment pedicle fixation for spinal fractures apparently have not been analyzed adequately. At the Sixth International Congress on Cotrel-Duhousset Instrumentation, McKinley et al. reported that seven patients who had been managed with short-segment Cotrel-Dubousset constructs had frequent breakage of the screws and an average loss of correction of 18 degrees. At the same meeting, Devito and Tsahakis advocated short-segment Cotrel- Dubousset instrumentation for the fixation of fractures as being safe, effective, and free of complications. Gullet et al. also reported that short Cotrel-Dubousset fixation produced favorable results, in thirty-nine patients, but they presented no data. Such reports encourage the use Of the System but offer no guidelines for the selection of patients, postoperative care. or technical application. No discussion of complications has yet appeared in the peer-reviewed literature. to our knowledge. The literature suggests that short-segment pedicle Cotrel-Dubousset instrumentation should not fail. Gurr et al. reported that the system provided torsional. flexural. and compressive rigidity comparable with that provided by longer Cotrel-Dubousset constructs and superior to that of traditional Harrington or Luque constructs#{176}7.they attributed this rigidity to the rigid screwrod junction provided by the Cotrel-Dubousset system. Their biomechanical analysis involved an acute model, however. and the constructs were not tested to failure. Citing the studies of Ashman et al. and of Johnston et al., Gurr et al. concluded that it was not feasible to reproduce, in the laboratory, cyclical loading comparable with that affecting patients during the first four postoperative months. From this acute model, they could not anticipate material fatigue or biological degradation factors that can affect intermediate or long-term stability. Several authors have reported good results with the use of pedicle screws in clinical situations. McAfee et al. carried out a survivorship analysis of instrumentation systems involving pedicle screws in 120 consecutively managed patients; the prevalence of problems with the screws was only 4 per cent, and the over-all prevalence of complications related to the instrumentation was 10 per cent. However, 93 per cent of these patients were managed for conditions involving sagittal, rather than axial, instability. Only eleven were managed for a traumatic injury, and only nine had a burst fracture. Matsuzaki et al. reported a 6 per cent prevalence of broken screws in fifty-seven patients who had been managed for degenerative disc disease, spondylolisthesis, and stenosis; none of these patients had axial instability. In a survivorship analysis of VSP instrumentation used to treat twenty-one burst fractures, Ebelke et al. identified a 6 per cent prevalence of broken screws, with an over-all 33 per cent prevalence of problems related to the hardware. They emphasized that all of the failures occurred in patients who had not been managed with anterior bone-grafting, and that transpedicle bonegrafting resulted in 100 per cent survival of the constructs. Their study demonstrated that disruption of the anterior spinal column results in a higher prevalence of failed fixation, and that not all hardware-related failures are the result of broken screws. In our experience, the most problems with fixation and failure of hardware occurred when the injury was at the first or second lumbar level. This observation is consistent with the higher prevalence of injury at these levels, but it may also reflect a greater degree of instability when there is a fracture at either of these 1evels. Certainly, neurological structures are at greater risk when fractures are at the cephalad lumbar levels, and the patient may be able to tolerate less deformity or translation here than in the caudad levels of the lumbar spine. Mechanical stresses on the constructs may also be greater than in the cephalad aspect of the lumbar Spine, where the iliolumbar ligaments and powerful paraspinous muscles may lend some support. We do not know for certain why the screws failed at such a high rate in our study. Failure of fixation due to a screw pulling out or due to vertebral settling would not be anticipated in this population of young, healthy patients. Instead, dense, healthy bone would be expected to transmit more of the stresses and bending moments to the bone-screw interface, concentrating forces at the hub of the screw. In any construct that does not support the anterior column of the spine, this would result in a powerful bending moment acting on the screw at the THE JOURNAL OF BONE AND JOINT SURGERY

6 EARLY FAILURE OF SHORT-SEGMENT PEDICLE INSTRUMENTATION FOR THORACOLtJMBAR FRACTURES I 67 point at which it enters the lamina. Whereas longer constructs can take advantage of contact forces between the lamina and the fixation rod, no such contact can be obtained with the use of short-segment constructs without excessive contouring of the rod. Without three-point contact. bending moments are resisted only by the intrinsic stiffness of the screw and rod. The weak link is the screw as it enters the bone. and this is where the bending occurs. Pre-stressing of the screws, through forced compression. distraction, or in situ bending, appears to be inadvisable, as it weakens the screw at the hub-lamina interface. Screws may then fatigue earlier, when the fracture is unstable and there is the least axial support, and multiple screws may fail. leading to the most severe deformities seen in this series. This preliminary review showed that the failure of screws is a common complication of the Cotrel- Dubousset system when a short-segment pedicle construct is used. Leaving the caudad pedicle screw of the Cotrel-Dubousset system isolated and unprotected appears unwise, particularly when the anterior column of the Spine has been compromised. References I. Ashman. R. B.; Birch, J. G.; Bone, L. B.; Corin, J. D.; Herring, J. A.; Johnston, C. E., III; Ritterbush, J. F.; and Roach, J. W.: Mechanical testing of spinal instrumentation. Cliii. Orthop., 227: i Cobb, J. R: Outline for the study of scoliosis. In Instructional Course Lectures, The American.4cade,nv of Ort/iopaedic Sttrgeoii.s Vol. 5. pp Ann Arbor.J. W. Edwards Devito, D. P., and Tsahakis, P. J.: Cotrel-Dubousset instrumentation in traumatic spine injuries. In Proceedings oft/it Sixth /ntermttio,ial Congress ott Cotrel-Dubousset instrumentation. pp. 4i -46. Montpellier. Sauramps Medical. i Ebelke, D. K.; Asher, M. A.; Neff, J. R.; and Krake, D. P.: Survivorship analysis of VSP spine instrumentation in the treatment of thoracolumbar and lumbar burst fractures. Spine, 16 (Supplement 8): Gillet, P.; Meyer, R.; Fatemi, F.; and Lemaire, R.: Short segment internal fixation using CD instrumentation with pedicular screws: biomechanical testing. In Proceedings of the Sixth international ( omigress on ( otrel-dubousset instrunientation. pp Monlpellier. Sauramps Medical Gurr, K. R.; McAfee, P. C.; and Shih, C.-M.: Biomechanical analysis of posterior instrumentation systems after decompressive laminectomy. An unstable calf-spine model. J. Bone andioint Surg.. 70-A: , June Gurr, K. R.; McAfee, P. C.; and Shih, C.-M.: Biomechanical analysis of anterior and posterior instrumentation systems after corpectomv. A calf-spine model.]. Bone and]oint Surg.. 70-A: I 182-i 191. Sept Johnston, C. E II; Ashman, R. B.; Sherman, M. C.; Eberle, C. F.; Herndon, W. A.; Sullivan, J. A.; King, A. G. S.; and Burke, S. W.: Mechanical consequences of rod contouring and residual scoliosis in sublaminar segmental instrumentation. J. Orthop. Res.. 5: McAfee, P. C.; Weiland, D. J.; and Carlow, J. J.: Survivorship analysis of pedicle spinal instrumentation. Spine. 16 (Supplement 8): McBride, 6. 6.: Cotrei-Duhousset rods in spinal fractures. Paraplegia. 27: McKinley, L. M.; Obenchain, T. G.; and Roth, K. R.: Loss of correction: late kyphosis in short segment pedicle fixation in cases of posterior transpeduncular decompression. In Proceediizgs of the Sixth imuernatiomial Congres.s on Cotrcl-D,tbou.s set imisirumnentaljon. pp Montpellier. Sauramps Medical Matsuzaki, H.; Tokuhashi, Y.; Matsumoto, F.; Hoshino, M.; Kiuchi, T.; and Torlyama, S.: Problems and solutions of pedicle screw plate fixation of lumbar spine. Spine. 15: Steffee, A D.; Biscup, R. S.; and Sitkowski, D. J.: Segmental spine plates with pedicle screw fixation. A new internal fixation device for disorders of the lumbar and thoracolumbar spine. Cliii. Orthop., 203: Suk, S. I.; Shin, B. J.; Lee, C. S.; and Lee, M. C.: CD pedicle screws in the treatment of unstable lumbar fractures. In Proceedings oft/ic Fifth International C ongress on Cotrel-Dubousset instriunentation, p. 93. Montpellier. Sauramps Medical VOL. 75-A. NO. 2. FEBRUARY 1993