Anterior thoracic scoliosis constructs

The Spine Journal 3 (2003) 213–219

Anterior thoracic scoliosis constructs: effect of rod diameter and intervertebral cages on multi-segmental construct stability David W. Polly, Jr., MDa,b,*, Bryan W. Cunningham, MScc, Timothy R. Kuklo, MDa,b, Lawrence G. Lenke, MDd, Itaru Oda, MDc, Teresa M. Schroeder, BS, MBAa,b, William R. Klemme, MDa,b a

Orthopaedic Surgery Service, Walter Reed Army Medical Center, 6900 Georgia Avenue, Washington, DC 20307, USA b Department of Surgery, Uniformed Services University of the Health Sciences, Bethesda, MD, USA c Orthopaedic Biomechanics Laboratory, Union Memorial Hospital, 201 E. University Parkway, Baltimore, MD 21218, USA d Department of Orthopaedic Surgery, Washington University, 1 Barnes Plaza, West Pavilion, Suite 11300, St. Louis, MO 63110, USA Received 19 November 2001; accepted 23 September 2002

Abstract

Background context: Many studies have reported on the use of anterior instrumentation for thoracolumbar scoliosis and more recently thoracic scoliosis. However, the optimal construct design remains an issue of debate. Purpose: To optimize construct design and enhance implant survival until a successful spinal arthrodesis is achieved. Study design: This study evaluated the effect of rod diameter and intervertebral cages on construct stiffness and rod strain using a long-segment, anterior thoracic scoliosis model with varying levels of intervertebral reconstruction. Methods: Sixteen fresh-frozen calf spine specimens (T1 to L1) were divided into two groups based on rod diameter reconstruction (4 mm and 5 mm). Testing included axial compression, anterior flexion, extension and lateral bending with variations in the number and level of intervertebral cage reconstructions: apical disc (one), end discs (two), apical and end discs (three), all seven levels (seven). Multisegmental construct stiffness and rod strain were determined and normalized to the intact specimen for analysis. Results: The seven-level intervertebral cage construct showed significantly greater stiffness in axial compression for both the 4-mm (366% increased stiffness) and 5-mm (607% increased stiffness) rod groups (p.001). The remaining constructs were not significantly different from each other (p.05). In flexion, similar results were obtained for the 4-mm construct (p.001) but not the 5-mm construct, because the reconstruction-alone, one-, two- and three-cage constructs were all significantly stiffer than the intact specimen (p.05). Multisegmental construct stiffness under extension loading, as well as right and left lateral bending, also exhibited significant differences between the seven-level interbody cage reconstructions and the remaining constructs. Apical rod strain for both the 4-mm-rod and 5-mm-rod groups were significantly higher for the two cage constructs (a cage at either end but not the apex where the strain gauges were located) as compared with the other con-

FDA device/drug status: anterior spinal instrumentation system, approved for this application. Author DWP acknowledges a financial relationship (grant research support from DePuy AcroMed) that may indirectly relate to the subject of this research. One or more of the authors of this paper are employees of the United States government. This work was performed as part of their official duties. As such this work cannot be copyrighted but is considered to be in the public domain. The opinions contained herein are the personal views of the authors and are not to be construed as official or as representing the Department of Defense. * Corresponding author. Chief, Department of Orthopaedic Surgery and Rehabilitation, Walter Reed Army Medical Center, Washington, DC 20307, USA. Tel.:(202)782-5440; fax (202)782-6845. E-mail address: [email protected] (D.W. Polly) 1529-9430/03/$ – see front matter © 2003 Elsevier Inc. All rights reserved. PII: S1529-9430(02)00 5 5 5 - 7

214

D.W. Polly et al. / The Spine Journal 3 (2003) 213–219

structs (p.05). These differences were more pronounced in the 4-mm-rod group. Similar results were obtained in anterior flexion, extension and lateral bending. Conclusions: Intervertebral cages at every level significantly improved construct stiffness compared with increasing rod diameter alone. Moreover, cages markedly decreased rod strain, and when structural interbody supports were not used, axial compression created the greatest rod strain. © 2003 Elsevier Inc. All right reserved. Keywords:

Anterior spinal fusion; Scoliosis; Rod diameter; Intervetebral cages

Introduction

Materials and methods

The goal of surgical treatment of scoliosis is to achieve a stable, balanced spine, centered over the pelvis, fusing as few motion segments as possible, and at the same time, minimizing the morbidity of the operative procedure. The optimal method to achieve this goal remains an issue of debate. Posterior procedures with multiple-anchor, multiple-rod systems can achieve excellent coronal and sagittal correction [1]. However, this often requires extension of the fusion into the lumbar spine, potentially leading to an increase in late onset back pain [2]. Anterior correction saves distal motion segments but carries the morbidity of the anterior approach [3]. The optimal anterior thoracic scoliosis construct has not yet been devised. Dwyer and Schafer’s [4] initial concept generated tremendous interest; however, the cable screw system was formidably challenged in the thoracolumbar spine. Zielke et al. [5] improved the instrumentation by using a flexible small threaded rod system designed to permit gradual curve correction. However, the system was kyphogenic and resulted in a high rate of rod breakage. Harms popularized the use of an anterior small rod system for anterior correction of thoracic curves [3,6]. However, initial clinical studies using a 3.2-mm threaded rod system led to rod failures. The next major change was the development of an anterior solid rod system, such as TSRH (Medtronic Sofamor Danek, Memphis, TN) [7]. With a more rigid rod, better sagittal plane control could be achieved. However, there has been continued debate about the optimal rod diameter and construct stiffness. The strategies to optimize anterior construct design are 1) increase rod diameter; 2) use a two-rod system; 3) use interbody load sharing. Some studies have shown that the presence of an anterior column graft has a far greater impact on construct stiffness and decreasing rod strain than would increasing rod diameter alone [8]. Other biomechanical studies suggest that rod strain should be maximal at the apex of the deformity [9; Choma TJ, Chwirut D, Polly DW, unpublished data, 1998]. The purpose of the current study was to evaluate the effect of rod diameter and the use of structural interbody support (Harms cages) on construct stiffness and rod strain in long-segment scoliosis constructs. Increased construct stiffness has been shown to improve fusion rate and the strength of the fusion mass [10,11]. Therefore, optimizing implant performance by minimizing rod strain at the critical stress point should prolong implant survival.

This study was conducted at the Orthopaedic Biomechanics Laboratory, Union Memorial Hospital, Baltimore, Maryland. Experimental model A thoracolumbar calf spine served as the experimental model in this study. This model has been extensively used in the past for spinal biomechanics and has a predictable track record. For long-segment testing, we decided that this model provides the most analogous bone–screw interface. Moreover, there is consistency between specimens, and the morphologic dimensions are appropriate. Transducer measurements: surface strain gauges and extensometer Data acquisition was performed using both surface strain gauges and an extensometer. Two uniaxial strain gauges (CEA-06-125UN-350; Measurements Group, Inc., Raleigh, NC) affixed to the anterior spinal rods provided measurement of hardware strain in response to the applied loads, for each experimental setup (Fig. 1). Strain data were acquired using a multichannel signal-conditioning amplifier (2100 System; Measurements Group, Inc., Raleigh, NC) interfaced with an IBM personal computer. For all reconstruction groups, rod strain was acquired from two surface-strain gauges located on opposite sides of the rod (anterior and posterior, 180 degrees apart) at the apical intervertebral disc level. An extensometer (Model 623,25C-20, 50.0-mm gauge length; MTS Systems, Inc., Minneapolis, MN) positioned on the anterior surface of the spine and spanning the operative motion segments was used to quantify segmental displacement. The extensometer gauge length was established using the zero pin, and the needle edges were then placed into the center of each body before initiating the test. Subsequent computational analysis for all transducer output was performed with Lotus 123 (Lotus Development Corporation, Cambridge, MA). Specimen preparation and biomechanical testing Sixteen T1 through L1 fresh-frozen calf spine specimens were used in this investigation and divided into two groups based on reconstruction hardware: Group 1, 4-mm-diameter rod, and Group 2, 5-mm-diameter rod. In preparation for testing, the thoracolumbar spine specimens were thawed to

D.W. Polly et al. / The Spine Journal 3 (2003) 213–219

215

Fig. 1. Spinal reconstructions. The intact spine was the first setup, and all testing results were normalized to this condition. The following reconstruction/stabilization procedures were then performed: (A) discectomy alone, (B) single apical disc, (C) end vertebral discs, (D) apical and end vertebral discs and (E) all seven levels.

room temperature and cleaned of all residual musculature, with care taken to preserve pertinent ligamentous structures. The distal (T13–L1) and proximal (T1–T3) ends of specimens were embedded in rectangular steel tubing foundations using four, four-point compression screws for fixation, thereby leaving the operative segments (T4 through T11) unembedded. Biomechanical analysis was performed using a servohydraulic MTS 858 Bionix testing device (MTS Systems, Inc., Minneapolis, MN) configured with LVDT/ RVDT displacement transducers and a 3.3-kilopound, 250 Newton-meter axial-torsional load cell. Load displacement data acquisition were performed through an MTS interface cable to a high-speed analog to digital DAS16G Metrabyte board (Laboratory Technologies, Taunton, MA) interfaced with an IBM 486 PS/2. Biomechanical testing parameters of the specimens quantified segmental stiffness (kN/M or Nm/ Deg) under axial compression (300 Newtons), flexion/extension (3 Newton-meter) and lateral bending (3 Newton-meter) nondestructive testing modes. As determined by a gravity-centered weight, the middle spinal column at the apical level served as the axis of rotation for flexion/extension and lateral bending testing modes. All bending moments were applied using a pure moment about the predefined axis and repeated over a period of five ramp cycles at a rate of 20% full scale/second. Data from the fourth cycle were used for subsequent computational analyses. Multisegmental spinal construct stiffness was calculated as the peak applied load (Newton or Newton-meter) divided by the corresponding segmental displacement (millimeters or

degrees). The peak stiffness computations represent one method for quantifying the overall rigidity of the reconstructions. However, it must be recognized that the nonlinear elastic behavior of the spinal segments can be further characterized into neutral zones and range of motion, which were not included in the current study. Reconstruction groups The following reconstruction groups were conducted for the 4-mm-diameter rod (Group 1; Fig. 1):[FIG 1] native intact spine, multilevel discectomy (T4–T5 through T10–T11) with rod/screw reconstruction, single apical cage (T7–T8), end vertebral cages (T4–T5 and T10–T11), apical and endvertebral cages (T4–T5, T7–T8 and T10–T11), cages at all levels (T4–T5 through T10–T11). The same reconstruction groups were conducted for the 5-mm-diameter rod (Group 2). All reconstruction groups except the intact spine were evaluated in combination with anterior reconstruction using the DePuy Motech MossMiami TM Monaxial Bone Screw (6.0 mm  40 mm screws, stainless steel) and 4-mm or 5-mm rods (Depuy AcroMed Corporation, Raynham, MA). Instrumentation was applied according to the manufacturer’s recommendations. Importantly, before tightening the screw/rod connections, a 75-Newton axial compressive preload was applied across the reconstructed interspace at each intervertebral level using a precalibrated compression beam deflection instrument. Two types of titanium mesh cages

216

D.W. Polly et al. / The Spine Journal 3 (2003) 213–219

were used in the current study: 1) two titanium mesh round cages (10-mm diameter, 6.5-mm height) placed at the T4– T9 level, and 2) two titanium mesh round cages (14-mm diameter, 10-mm height) placed from T9 through T11. Data and statistical analysis Multisegmental construct stiffness was calculated as the peak applied load (Newton or Newton-meter) divided by the corresponding segmental displacement (millimeters or degrees). Rod strain (microstrain) was measured at the peak load during the fourth loading cycle. In the most common loading situation, the rods may be subjected to the two force components: axial and bending forces. The rod surface strain can be described by a combination of the two measured strains resulting from axial and bending forces. Thus, the stress imposed by the axial load and bending stress can be calculated as axial stresselastic modulus50% sum of the surface strains, and bending stresselastic modulus50% difference of the surface strains. All data were normalized to the intact specimen and expressed as a percentage change (ratio) from the intact condition (mean1 standard deviation, n8/reconstruction). Statistical analyses were performed using one-way analysis of variance (ANOVA) and StudentNewman-Keuls post hoc multiple comparison procedure.

groups were significantly higher for the two cage reconstructions compared with the other constructs, which were not significantly different from each other (4-mm rod, oneway ANOVA F 11.62, p.001; 5-mm rod, one-way ANOVA F 6.40, p.001; Fig. 3). This difference in rod strain was more pronounced in the 4-mm group compared with the 5-mm group. Anterior flexion

Results

In similarity to axial compression, multisegmental construct stiffness under flexural loading conditions exhibited significant differences between the seven-level interbody cage reconstructions and remaining constructs (4-mm rod, oneway ANOVA F 27.93, p.001; Fig. 4). Additionally, for the 4-mm-rod group, the three cage constructs were significantly stiffer than the intact, reconstruction-alone, one- and two-cage constructs (p.05). However, the 5-mm-rod group indicated the reconstruction-alone, one-, two- and three-cage constructs as significantly stiffer than the intact condition (p.05). For the 4-mm rod, apical rod strain was significantly higher and lower in the two- and seven-cage constructs, respectively, compared with the remaining reconstructions (one-way ANOVA F 13.80, p.001; Fig. 3). However, no significant differences were observed in rod strain for the 5-mm-rod group (one-way ANOVA F 1.14, p .358).

Axial compression

Extension

Axial compressive loading indicated significance when comparing the seven-level interbody cage reconstruction stiffness to the remaining constructs, for both the 4-mmdiameter rod and 5-mm-diameter rod groups (4-mm rod, one-way ANOVA F 36.08, p.001; 5-mm rod, one-way ANOVA F 94.73, p.001; Fig. 2). However, the remaining constructs were not statistically different from each other (p.05). Apical rod strain for both the 4-mm and 5-mm rod

Multisegmental construct stiffness under extension loading conditions exhibited significant differences between the seven-level interbody cage reconstructions and remaining constructs in almost all cases (4-mm rod, one-way ANOVA F8.69, p.001; 5-mm rod, one-way ANOVA F10.63, p.001). However, the three-cage reconstruction (4-mm rod), although not different from the seven-cage construct, was stiffer than the intact, reconstruction-alone and one-cage

Fig. 2. Axial compressive stiffness: 4-mm and 5-mm rods. *Indicates significance from all other spinal construct groups at p.05. Error bars represent 1 standard deviation.

D.W. Polly et al. / The Spine Journal 3 (2003) 213–219

217

Fig. 3. Axial compressive rod strain: 4-mm and 5-mm rods. *Indicates significance from all other spinal construct groups at p.05. Error bars represent 1 standard deviation.

constructs. In the case of the 5-mm rod, all reconstructions were significantly different from the intact spine (p.05). Once again, apical rod strain for the 4-mm-rod and 5-mmrod groups was higher for the two-cage reconstructions compared with the other constructs (4-mm rod, one-way ANOVA F23.58, p.001; 5-mm rod, one-way ANOVA F4.64, p.005). The 4-mm-rod group indicated all constructs, except the two cages were significantly less than the reconstruction alone (p.05). However, the 5-mm group did not show the same trends. Lateral bending Lateral bending in the direction of the anterior spinal rod (ie, unloading the interbody cages) demonstrated the sevencage constructs, again, as significantly stiffer than almost all other reconstructions (4-mm rod, one-way ANOVA F 16.08, p.001). In the case of the 4-mm-rod group, no other statistically significant findings were observed. However, in

the case of the 5-mm-rod group, all reconstructions were significantly stiffer than the intact condition (p.05). Left lateral bending, the direction away from the anterior spinal rod (ie, loading the interbody cages), exhibited the sevencage constructs as significantly different from all others (4-mm rod, one-way ANOVA F12.55, p.001; 5-mm rod, one-way ANOVA F26.59, p.001). The two- and threecage reconstructions were stiffer than the intact condition for the 4-mm-rod group (p.05). Additionally, all reconstruction conditions for the 5-mm-rod group were significantly higher in stiffness level than the intact condition (p.05). Discussion Treatment of scoliosis requires a balancing of risks and benefits. The choice of surgical approach and fusion levels also affects the risk–benefit profile, because extension of the fusion into the lumbar spine increases the risk of back

Fig. 4. Anterior flexural stiffness: 4-mm rod, * and # indicate significance from all other spinal construct groups and from each other at p.05; 5-mm rod, * indicates significance from all other groups and # indicates significance from intact condition, p.05. Error bars represent 1 standard deviation.

218

D.W. Polly et al. / The Spine Journal 3 (2003) 213–219

pain and possibly accelerates disc degeneration [5]. The primary advantage of the anterior approach is that it provides the ability to save distal levels. Model selection for analysis of long-segment instrumentation is problematic. Options include human cadaveric tissue, animal cadaveric tissue and polyethylene cylinder models. Each model has advantages and disadvantages. Animal cadaveric spines have very consistent bone density and reasonable stiffness properties but lack human sagittal contour and have anatomic variation from the human spine. For this study, we chose to use calf spines, because we think that the bone–implant interface is a critical weak point in many anterior scoliosis constructs. This can be seen in many cases of anterior instrumentation where there has been partial pullout of the proximal screw. We wanted to obviate the bone– screw interface and look at the variables of rod diameter and structural interbody support. In our study we were able to test the intact spine stiffness, but once the discectomies were performed, the spine was so unstable that testing was impossible. Performing anterior instrumentation without structural interbody supports improved the stiffness of the spine in all testing modes compared with the spine after discectomies but did not return it to intact stiffness in axial compression (0.312 for 4-mm rod, 0.730 for 5-mm rod) or for flexion with a 4-mm rod (0.977 4-mm, 2.365 5-mm-greater than intact spine). In lateral bending (both right and left), the reconstructed spine was stiffer than the intact spine for both 4- and 5-mm rods. The most critical loading mode postoperatively is not yet well delineated. Some authorities suggest that flexion and lateral bending are perhaps most critical in the lumbar spine. Our study suggests that axial compression may be the critical loading mode (ie, the mode that creates the greatest rod strain). The effect of increasing rod diameter was not nearly as great an effect as using structural interbody support at each level. Using cages at every level significantly improved the construct stiffness in every testing modality. The greatest change was in axial compression. With the 4-mm rod and seven cages, the stiffness increased 366%. With the 5-mm rod and seven cages, it increased 607%. This strategy was clearly more efficacious in this testing mode than the dualrod system (which achieved only about 100% stiffness in axial compression, according to Kaneda and Shono [12] and Shono et al [13]). The alternate strategy of using a dual-rod system, however, has advantages and disadvantages, because it yielded a stiffness increase of 900% for flexion and extension and a 200% increase for rotational stiffness. The disadvantages of the dual-rod system include a higher profile, especially of the more anterior screw head, and no real increase in axial stiffness [12,13]. We presumed that maximal rod strain would be at the midpoint or apex. This assumption is based on data from Choma et al. [Choma TJ, Chwirut D, Polly DW, unpublished data, 1998], which showed that rod strain was maximal in a long-segment model at the apex. Anterior con-

structs, by their very nature with screws at every level, will concentrate rod strain even more. Any residual deformity of 16 degrees will also increase rod strain [9]. So we believe that apical rod strain, especially in axial compression, is the current biomechanical weak link in anterior thoracic scoliosis construct design. This study shows that an apical cage markedly reduced apical rod strain, whereas cages remote from the apex did not have a significant effect on apical rod strain. Thus, the optimal strategy for minimizing apical rod strain is to place a cage at that level. The optimal strategy for improving overall construct stiffness is to place a cage at each level. The use of cages has a greater impact than increasing rod diameter. Using cages at every level increased the stiffness more than increasing the rod diameter increased the construct stiffness. This may allow use of smaller, or less stiff, rods supplemented by structural interbody support. This could be particularly advantageous in thoracoscopic or minimally invasive applications. Although we have demonstrated a strategy for improving construct stiffness that decreases rod strain, there remains the question of the impact of this stiffness on uninstrumented adjacent segments. The appropriate concern is that a rapid transition in stiffness may lead to more rapid breakdown in the adjacent segment. From this study, we have no new insight into this long-term question. Perhaps optimizing stiffness at the site of maximum rod strain with a tapering out at either end may eventually prove useful. A better solution eventually would be to have the implant stiffness degrade over time once the fusion is solid. Future considerations for anterior instrumentation strategies possibly include improving the screw–bone interface with bioactive bone cements or with screw augmentation [14– 16]; altering screw orientation [17] and/or using smaller, flexible rods in conjunction with structural interbody supports at each level. Conclusions Use of cages at every level in a multilevel reconstruction has a much greater effect on stiffness than increasing rod diameter from 4 mm to 5 mm. Cages significantly decrease rod strain at the level used, and axial compression creates the greatest rod strain at the apex when structural interbody supports are not used. References [1] Lenke LG, Bridwell KH, Blanke K, Baldus C, Weston J. Radiographic results of arthrodesis with Cotrel-Dubosset instrumentation for the treatment of adolescent idiopathic scoliosis: a five to ten year follow-up study. J Bone Joint Surg [Am] 1998;80:807–14. [2] Ginsburg HH, Goldstein L, Haake PW, et al. Longitudinal study of back pain in postoperative idiopathic scoliosis: long term follow-up, phase IV. Presented at the Scoliosis Research Society annual meeting; September 13–16, 1995; Asheville, NC. [3] Betz RR, Harms J, Clements DH, et al. Anterior instrumentation for thoracic idiopathic scoliosis. Sem Spine Surg 1997;9:141–9.

D.W. Polly et al. / The Spine Journal 3 (2003) 213–219 [4] Dwyer AF, Schafer MF. Anterior approach to scoliosis. Results of treatment in fifty-one cases. J Bone Joint Surg [Br] 1974;56:218–24. [5] Zielke K, Stunkat R, Dequesne J, Beaujean V. Ventrale derotation spondylodese. Orthop Prax 1975;11:562–5. [6] Hammerberg KW, Zielke K. VDS instrumentation for idiopathic thoracic curvatures. Presented at the American Academy of Orthopaedic Surgeons annual meeting; January 1985; Las Vegas, NV. [7] Turi M, Johnston CE, Richards BS. Anterior correction of idiopathic scoliosis using TSRH instrumentation. Spine 1993;18:417–22. [8] Polly DW, Klemme WR, Cunningham BW, Burnette JB, Haggerty CJ, Oda I. The biomechanical significance of anterior column support in a simulated single level spinal fusion. J Spinal Disord 2000;13:58–62. [9] Ashman RB, Birch JG, Bone LB, et al. Mechanical testing of spinal instrumentation. Clin Orthop 1988;227:113–34. [10] Kanayama M, Cunningham BW, Sefter JC, et al. Does spinal instrumentation influence the healing process of posterolateral spinal fusion? An in vivo animal model. Spine 1999:24:1058–65. [11] McAfee PC, Farey ID, Sutterlin CE, Gurr KR, Warden KE, Cunningham BW. Device related osteoporosis with spinal instrumentation. A canine model. Spine 1989;14:919–26.

Thirty-Five Years Ago in Spine . . .

[12] Kaneda K, Shono Y. Kaneda anterior multisegmental instrumentation: two-rod system for the treatment of thoracolumbar and lumbar scoliotic curvatures. In: Bridwell KH, DeWald RL, editors. The textbook of spinal surgery, 2nd ed. Philadelphia: Lippincott-Raven, 1997:641–63. [13] Shono Y, Kaneda K, Yamamoto I. A biomechanical analysis of Zielke, Kaneda and Cotrel-Dubosset instrumentation in thoracolumbar scoliosis: a calf spine model. Spine 1991;16:1305–11. [14] Lieberman IH, Khazim R, Woodside T. Anterior vertebral body screw pullout testing: a comparison of Zielke, Kaneda, Universal Spine System and Universal Spine System with pullout-resistant nut. Spine 1998;23: 908–10. [15] Lotz JC, Hu SS, Chiu DFM, Yu M, Colliou O, Poser RD. Carbonated apatite cement augmentation of pedicle screw fixation in the lumbar spine. Spine 1997;22:2716–23. [16] Moore DC, Graziano RS, Maitra LA, Farjo SA, Goldstein SA. Restoration of pedicle screw fixation with an in situ setting calcium phosphate cement. Spine 1997;22:1696–705. [17] Horton WC, Blackstock SF, Norman JT, Hill CS, Feiertag MA, Hutton WC. Strength of anterior vertebral body screws. Spine 1996;21:439–44.

discograms in asymptomatic subjects. In the same year, another study concluded that discography was valuable and associated with few complications [2]. Holt’s study was critically reviewed by Simmons et al. in 1988 [3].

References

One of the most commonly quoted articles in the spine literature was published by E.J. Holt, Jr., in 1968 [1]. Based on study of prisoners who had volunteered to undergo discography, Holt reported 37% false-positive

219

[1] Holt EJ. The question of lumbar discography. J Bone Joint Surg 1968;50A:720–6. [2] Wiley JJ, Macnab J, Wortzman G. Lumbar discography and its clinical application. Can J Surg 1968;11:280–9. [3] Simmons JW, Aprill CN, Dwyer AP, Brodsky AE. A reassessment of Holt’s data on “the question of lumbar discography.” Clin Orthop 1988;237:120–4.