Statistics from the year 2005 indicated that the prevalence of spine fusion surgery increased by 73% nationally over a nine-year period (1997 to 2005) to 349,400 procedures per year1. With improvements in spinal care and the increasing number of patients undergoing these procedures, the numbers are likely much higher today. Despite improvements in fusion surgery, it is sometimes necessary to perform postoperative imaging with magnetic resonance imaging (MRI) and computed tomography (CT) to assess residual pain or radicular symptoms as well as to determine fusion status. Postoperative imaging of implants has historically been difficult as the implants typically produced high artifact.
Implants including pedicle screws and rods, which were once made exclusively of stainless steel, are also now manufactured with titanium, alloys, and polymers, which in turn provide less imaging artifact and image distortion. Titanium spinal implants produce less imaging artifact than stainless steel2-5, and it is widely accepted that metal implants elsewhere in the body have similar effects6-10. A recent evaluation has shown that cobalt-chromium rods are stronger than titanium when tested to failure with cyclical loading11. With use of this information, smaller-diameter rods and screw heads of similar strength to currently marketed titanium implants have been fabricated using cobalt-chromium alloys.
The objective of this study was to examine the imaging characteristics on MRI and CT of two spinal fixation systems placed in two human cadaver spine segments and determine if there were any differences in the clarity of imaging between the implant materials. The systems were composed of either titanium alloy (titanium; 5.5-mm spinal rods and 6.5-mm multiaxial screws) or a combination of titanium and cobalt chromium (titanium-cobalt; a 4.75-mm cobalt-chromium-molybdenum alloy rod and a 6.5-mm titanium screw with cobalt-chromium-molybdenum alloy multiaxial head system). Our aim was to evaluate the effects of these material combinations on the image clarity of specific spinal anatomical regions of interest using MRI and CT techniques in the posterior lumbar spine. We hypothesized there would be no reduction of either MRI or CT image quality between the materials on the basis of the clarity of regions of interest in comparative images.
This comparative study utilized two fresh-frozen female torsos with no evidence of previous surgery (age at death, fifty-one and fifty-four years). The specimens were stored to preserve the native structures and were allowed to thaw at room temperature prior to spine exposure (from L1 to S1) and implantation of the posterior spinal stabilization systems.
Construct Materials
The implant constructs tested consisted of two different combinations of metals. The titanium-cobalt system, CD HORIZON SOLERA (Medtronic Spinal and Biologics, Memphis, Tennessee), consisted of a 6.5-mm titanium alloy (Ti-6A1-4V) screw with a cobalt-chromium-molybdenum (Co-28Cr-6Mo) multiaxial head and 4.75-mm-diameter cobalt-chromium-molybdenum (Co-28Cr-6Mo) spinal rods. The titanium spinal system, CD HORIZON LEGACY (Medtronic Spinal and Biologics), was composed of 6.5-mm titanium alloy (Ti-6Al-4V) screws with multiaxial heads and 5.5-mm spinal rods. Screw lengths in both systems were 40 mm (L1 and L2) and 45 mm (L3 and L4).
Interventions and Studies
The senior author (T.R.T.) prepared each specimen to simulate a four-level fusion construct, allowing for multiple disc level and pedicle evaluations. Each respective specimen was sequentially instrumented with one of the two defined spinal stabilization systems. In the first specimen, lumbar pedicles (L1 through L4) were prepared by predrilling with a 5.0-mm bit and tapping appropriately for the screw diameter of the titanium-cobalt system. Alternatively, the second specimen was implanted with the titanium screw system. With use of a specialized intraoperative multidimensional imaging, computer-assisted spinal navigation platform, O-arm linked with StealthStation TREON (Medtronic, Littleton, Massachusetts), care was taken to ensure symmetric trajectories for the pedicle screws of each system. Postimplantation images were recorded with use of the O-arm and were used to confirm the accuracy of screw placement.
The implanted cadaveric specimens were imaged in the supine position and were mapped (i.e., imaging trays were marked to a specific reference point) to allow for the exact repositioning of the specimen after removal and replacement of the implant systems and to ensure the scans were acquired in an identical position. The implanted specimens were imaged with T1 and T2-weighted MRI on a 1.5-T scanner (Symphony, with software version 2004A; Siemens, Erlangen, Germany) as well as with a CT scanner (LightSpeed Plus, with software version 07MW11.10; General Electric, Milwaukee, Wisconsin) with axial and reformatted sagittal imaging (see Appendix). After both scans were completed in each of the specimens, the implant systems were removed and implanted in the other cadaver (Figs. 1-A and 1-B). This switching of implant systems between the two cadavers served as a control variable, eliminating differences that existed in the cadavers and equalizing the images for both systems in each specimen. The specimens were then reimaged as previously described; thus, each cadaver was imaged by MRI and CT twice—once each with each system implanted.
Image Analysis
Representative images for grading were compiled by the primary investigator. Images of each group were captured at the same imaging plane and/or location with use of identical imaging settings. All imaging was performed at a single-site facility and was captured digitally, replicating the clinical evaluation scenario. The level of the axial slice was through the midpoint of the pedicle and the middle of the disc space, and sagittal images were acquired in the midline (Fig. 2). Images acquired from L1-L2 through the L4-L5 segments were used for image analysis. Specifically, MRI and CT images at the L1 pedicle, L1-L2 disc space, L2 pedicle, L2-L3 disc space, L3 pedicle, L3-L4 disc space, and L4 pedicle were evaluated. An image quality score was obtained by summing the scores collected from nine physicians who independently reviewed each digitized image. A total of six spine surgeon reviewers, including three orthopaedic spine surgeons and three neurosurgeons, with a combined average of eleven years of experience, evaluated each image for clarity and scored the respective cells. Additionally, three radiologists specializing in neurological imaging reviewed the images using the same guidelines. Each physician grader, blinded to the level and system being graded, was instructed to view the image and grade according to their interpretation of the image and its ability to provide diagnostic clarity. No time constraints were placed on the graders.
Scores were calculated by assessing individual grids that were electronically placed onto the MRI and CT digital images with cells positioned over respective areas of interest (Figs. 3-A and 3-B). Specifically, regions of interest in the sagittal plane included the disc space, spinal cord and/or canal, dural sac, and exiting foramen, whereas the axial plane consisted of the spinal cord and epidural space. Both views evaluated the pedicle for screw breach. Each area of interest was graded for image clarity, and scores were totaled to provide an overall score for the spinal stabilization system. Each cell was scored from 1 to 5 points on the basis of the clarity and definition of the structures visualized within the cell or region of interest (Table I). Regions of interest consisted of four, six, or eight grid scores summed for an overall score per image, resulting in total scores from 4 to 40 points, provided that a numerical value was applied to each grid space (Fig. 4). The number of the grids was the result of the nature of the image such that MRIs were smaller than CT images, and the sagittal images were different because of orientation. This approach was taken to keep the grid size constant. Images were prepared so that all defining indications of implant material were removed prior to distribution of the image summaries to the participating physician reviewers.
Statistical Methods
Mean image scores (and standard deviations) for both titanium and titanium-cobalt systems were estimated, with use of a two-way analysis of variance, for each physician group. Pairwise comparisons of mean rating scores between physician groups were performed with use of the Tukey Honestly Significant Difference test. Intraclass correlations were calculated for each physician group with use of methodology described by Shrout and Fleiss12. The type-I error (alpha) for all analyses was set at 0.05.
Because the image grading results were skewed, a natural log transformation was utilized to center the score distribution for the titanium system. The titanium-cobalt system scores required a square root transformation to normalize the distribution. The Mann-Whitney U test was used to test the overall scores between the two systems.
Source of Funding
This study was financed in part by funds donated by Medtronic Spinal and Biologics to the Orthopaedic Research Foundation.
Overall System Scores
Each of the nine physician reviewers graded a set of 100 images (900 image scores in total), consisting of thirty-six CT and sixty-four MRI scans (split equally between the titanium and titanium-cobalt systems). The overall mean score results (and standard deviation) were 18.16 ± 9.89 (range, 4 to 40) for titanium (n = 450) and 17.45 ± 9.64 (range, 3 to 40) for titanium-cobalt (n = 450); the differences were not significant (p = 0.275). The cadaver had little effect on implant system as displayed by the similarity in overall imaging scores.
On the basis of the mean similarities, we sought to determine if there were any differences between images obtained via MRI compared with those obtained by CT in each specimen. We further explored any commonalities between physician groups.
Analysis of Scores by MRI Imaging
When titanium-cobalt images were compared with use of MRI only, specimen 1 had a mean score of 12.36 ± 6.57 (range, 4 to 38 points) and specimen 2 had a mean score of 12.24 ± 6.78 (range, 3 to 34 points); the difference was not significant (p = 0.883). Similar results were seen in the titanium system, with a mean score of 13.10 ± 6.82 (range, 4 to 38 points) for specimen 1 and 11.67 ± 5.00 (range, 4 to 26 points) for specimen 2; however, this resulted in a minimally significant difference (p = 0.044).
Analysis of Scores by CT Imaging
The scores for CT imaging of both specimens were not significantly different when the titanium-cobalt (p = 0.274) and titanium systems (p = 0.837) were compared. The mean score, with use of titanium-cobalt system, was 25.90 ± 6.28 (range, 12 to 40 points) for specimen 1 and 27.11 ± 7.46 (range, 10 to 40 points) for specimen 2. In the titanium system, the mean scores were 28.54 ± 5.81 (range, 8 to 40 points) and 28.33 ± 7.10 (range, 12 to 40 points), respectively.
Physician Group Comparison
Physician groups scored both systems similarly, with radiologists grading higher overall than neurosurgeons, followed by orthopaedic surgeons. In grading the titanium system, radiologists had significantly higher mean scores (19.95 ± 9.51; range, 6 to 40 points) than either neurosurgeons (17.46 ± 9.30; range, 4 to 40 points) (p = 0.023) or orthopaedic surgeons (17.07 ± 10.63; range, 4 to 40 points) (p = 0.014). The mean score for neurosurgeons was not significantly higher than that for orthopaedic surgeons (p = 0.738). In grading the titanium-cobalt system, radiologists scored higher (18.84 ± 9.78; range, 5 to 40 points) than neurosurgeons (17.23 ± 8.94; range, 4 to 40 points) (p = 0.139) and significantly higher (p = 0.027) than orthopaedic surgeons (16.28 ± 10.07; range, 3 to 40 points). Again, the mean score for neurosurgeons was not significantly higher (p = 0.386) than that for orthopaedic surgeons. Despite the differences observed in the mean scores by physician specialty, no significant differences were observed in the overall scores for image clarity between the systems and only a minimally significant difference was found between specimens in the titanium model with use of MRI.
Note: The authors thank the following individuals for their assistance with this research study: Dr. John Dietz, Dr. Bradley Hostetter, Dr. Saad Khairi, Dr. Jean-Pierre Mobasser, Dr. Eric Potts, Dr. Kent Remley, Dr. Shane Rose, Dr. Joseph Riina, and Dr. David Schwartz.