- Email: [email protected]
Vertebroplasty increases compression of adjacent IVDs and vertebrae in osteoporotic spines
The Spine Journal 13 (2013) 1872–1880
Basic Science
Vertebroplasty increases compression of adjacent IVDs and vertebrae in osteoporotic spines Srinidhi Nagaraja, PhDa,*, Hassan K. Awada, BSa, Maureen L. Dreher, PhDa, Shikha Gupta, PhDa, Scott W. Miller, PhDb a
Division of Solid and Fluid Mechanics, Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, U.S. Food and Drug Administration, 10903 New Hampshire Ave, Silver Spring, MD 20993, USA b Division of Biostatistics, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, U.S. Food and Drug Administration, 10903 New Hampshire Ave, Silver Spring, MD 20993, USA Received 30 August 2012; revised 13 March 2013; accepted 1 June 2013
Abstract
BACKGROUND CONTEXT: Approximately 25% of vertebroplasty patients experience subsequent fractures within 1 year of treatment, and vertebrae adjacent to the cemented level are up to three times more likely to fracture than those further away. The increased risk of adjacent fractures postaugmentation raises concerns that treatment of osteoporotic compression fractures with vertebroplasty may negatively impact spine biomechanics. PURPOSE: To quantify the biomechanical effects of vertebroplasty on adjacent intervertebral discs (IVDs) and vertebral bodies (VBs). STUDY DESIGN: A biomechanics study was conducted using cadaveric thoracolumbar spinal columns from elderly women (age range, 51–98 years). METHODS: Five level motion segments (T11–L3) were assigned to a vertebroplasty treated or untreated control group (n510/group) such that bone mineral density (BMD), trabecular architecture, and age were similar between groups. Compression fractures were created in the L1 vertebra of all specimens, and polymethylmethacrylate bone cement was injected into the fractured vertebra of vertebroplasty specimens. All spine segments underwent cyclic axial compression for 115,000 cycles. Microcomputed tomography imaging was performed before and after cyclic loading to quantify compression in adjacent VBs and IVDs. RESULTS: Cyclic loading increased strains 3% on average in the vertebroplasty group when compared with controls after 115,000 cycles. This global strain manifested locally as approximately fourfold more compression in the superior VB (T12) and two- to fourfold higher axial and circumferential deformations in the superior IVD (T12–L1) of vertebroplasty-treated specimens when compared with untreated controls. Low BMD and high cement fill were significant factors that explained the increased strain in the vertebroplasty-treated group. CONCLUSIONS: These data indicate that vertebroplasty alters spine biomechanics resulting in increased compression of adjacent VB and IVD in severely osteoporotic women and may be the basis for clinical reports of adjacent fractures after vertebroplasty. Published by Elsevier Inc.
Keywords:
Vertebroplasty; Osteoporosis; Biomechanics; Vertebral compression fracture; Adjacent segment disease
FDA device/drug status: Approved for this indication (Bone Cement [Stryker]). Author disclosures: SN: Nothing to disclose. HKA: Nothing to disclose. MLD: Nothing to disclose. SG: Nothing to disclose. SWM: Nothing to disclose. * Corresponding author. U.S. Food and Drug Administration, Center for Devices and Radiological Health, Division of Solid and Fluid Mechanics, 10903 New Hampshire Ave., Building 62, Room 2110, Silver Spring, MD 20993-0002, USA. Tel.: (301) 796-0396; fax: (301) 796-9932. E-mail address: [email protected] (S. Nagaraja) 1529-9430/$ - see front matter Published by Elsevier Inc. http://dx.doi.org/10.1016/j.spinee.2013.06.007
Introduction Osteoporosis-related vertebral compression fractures (VCFs) are a concern for elderly women as they affect 25% of all postmenopausal women in the United States and 40% of women older than 80 years [1,2]. In fact, VCFs in women are three times more frequent than in men of similar ages [3]. A common surgical treatment for VCFs is percutaneous vertebroplasty, a minimally invasive
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
procedure in which polymethylmethacrylate bone cement is injected into the fractured vertebra to stabilize the fracture and relieve associated pain. However, clinical studies report that approximately 25% of the patients experience subsequent fractures within 1 year of the treatment [4,5]. Moreover, vertebrae adjacent to the cemented level represent most (50–67%) of these subsequent VCFs [4,6,7] and are up to three times more likely to fracture than those further away [8–10]. Thus, the potential risk of additional VCFs after procedure may outweigh the benefits of vertebroplasty. However, the causality between vertebroplasty and new fractures has not been established, and there is no consensus on the root cause of this phenomenon. Although vertebroplasty may play a role in adjacent VCFs, many of these patients already have vertebrae weakened by osteoporosis (as evidenced by the initial fracture) and are predisposed to fractures in the same region of the spine. Prior investigations of additional VCFs have been primarily limited to retrospective analyses of clinical data after secondary fractures have occurred. These studies are unable to directly determine the biomechanical effects of vertebroplasty on surrounding spinal tissues, especially because VCFs in adjacent and nonadjacent levels can also be influenced by procedural factors such as the volume and distribution of cement inside the vertebral body (VB) [11,12]. Nonclinical studies have hypothesized that additional fractures are because of the bone cement, which has a stiffness 7 to 10 times greater than osteoporotic vertebral bone [13]. The treated vertebra produces a ‘‘stress riser’’ effect where increased stresses are transferred to intervertebral discs (IVDs) and vertebrae adjacent to the original fracture [14,15]. This mechanistic explanation is plausible; however, other biomechanical studies indicated that vertebroplasty restores compressive loads in the neural arch, IVDs, and VB to prefracture levels [16,17]. For example, Kayanja et al. [18] found that vertebroplasty restored stiffness and adjacent vertebral cortex strains to normal levels and concluded that adjacent level fractures after vertebroplasty are because of severe loading, not vertebroplasty. Therefore, the biomechanical consequences of vertebroplasty on osteopenic/osteoporotic spine remain unclear. The objective of this research study was to elucidate the biomechanical consequences of vertebroplasty on adjacent IVDs and vertebrae.
Materials and methods Specimen preparation Caucasian female cadaveric spines (51–98 years) were acquired from Maryland State Anatomy Board (Baltimore, MD, USA) or National Disease Research Interchange (Philadelphia, PA, USA). Twenty donors were chosen based on radiographs and gross examination that confirmed the absence of device implantation, scoliosis, and vertebral
1873
fractures in the thoracolumbar region of each spine. Dual-energy X-ray absorptiometry (DXA; Hologic, Bedford, MA, USA) scans in the lumbar vertebrae indicated T-scores in the osteopenic/osteoporosis range (0.6 to 4.0) for these donors. Spine segments containing T11–L3 vertebral levels were dissected and cleaned while preserving the structural integrity of the spine section leaving the vertebrae, IVDs, pedicles, ligaments, and bony structures intact. The T11 and L3 vertebral levels of each sample were fixed in rapidly curing epoxy (3M, St. Paul, MN, USA) to provide flat uniform surfaces for mechanical testing. The samples were wrapped in 0.9% saline-soaked gauze and stored at 20 C. Fracture creation A wedge fracture was made in the L1 vertebra as it has the highest fracture incidence in the female lumbar spine [19]. A thin transverse cut was created through the anterior half of the L1 VB in accordance with previously published methods [20]. Compression bending was applied to create the wedge fracture. A custom jig was used to load specimens at a rate of 10 mm/min in a mechanical testing system (MTS Corporation, Eden Prairie, MN, USA). Loading was stopped when compression reached half of the premeasured height of the L1 VB (Fig. 1). Although this method does not precisely simulate the morphology of clinically observed compression fractures, it consistently created anterior wedge fractures in the L1 VB by weakening the anterior cortex and internal trabecular microstructure. Specimen groups Microcomputed tomography (micro-CT; Scanco Medical, Wayne, PA, USA) was used to assess trabecular microarchitecture and mineralization in each donor after creating the wedge fracture. Micro-CT scans (voxel size of 51.4 mm) were used to quantify trabecular microstructure in the superior adjacent VB (SAVB, ie, T12) and inferior adjacent VB (IAVB, ie, L2). Parameters such as bone volume fraction, trabecular thickness, trabecular spacing, trabecular number, connectivity density, structure model index, and degree of anisotropy were calculated [21,22]. Using micro-CT and DXA data, specimens were assigned to an experimental group that underwent vertebroplasty and a control group that did not (n510/group) such that trabecular microarchitecture, bone mineral density (BMD), and age were statistically similar between groups (Table). Vertebroplasty procedure For specimens assigned to the vertebroplasty group, an 11-gauge needle was advanced into the L1 fractured VB using a unilateral transpedicular approach (Fig. 2). Polymethylmethacrylate bone cement (Stryker, Kalamazoo, MI, USA) was injected into the anterior three-fourths of the VB based on lateral projections [10,23,24]. To prevent leakage of bone cement, parafilm was tightly wrapped
1874
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
with previous observations of spinal compressive loads up to 2.5 times body weight during brisk walking [29]. Cyclic loading was stopped at 115,000 cycles (approximately 16 hours) to prevent significant degradation or changes in spinal range of motion [30]. Mechanical testing was conducted at room temperature, and specimens were wrapped in saline-soaked gauze with fluid sprayed intermittently to maintain hydration throughout testing. Force-displacement data were continuously recorded throughout ramp and every 1,000 cycles during cyclic loading. Specimen strain was defined as the displacement divided by the original specimen height. Specimen stiffness was calculated in the linear region of the final 0.25% strain during ramp. Compression in adjacent VBs and IVDs
Fig. 1. Representative image of a cadaver specimen in wedge fracture jig. A transverse cut was initially created in the L1 vertebra to facilitate wedge fracture. Compression bending was applied to specimens through loading off axis.
around the wedge fracture. Fluoroscopy during the procedure and micro-CT imaging after the procedure were used to confirm containment of bone cement within the VB, without leakage into the adjacent IVD or outside the VB. The injected volume (5.760.5 mL) and cement fill (2263%) were similar to values reported clinically [25– 27]. After augmentation, cement was allowed to cure for approximately 4 hours at room temperature. Specimens were then stored at 20 C for up to 2 weeks before testing.
All specimens were reimaged with micro-CT after cyclic loading to assess changes in VB height and IVD thickness from precyclic loading (Fig. 3). Permanent axial compression in adjacent VBs and IVDs was calculated based on height change from premechanical and postmechanical testing scans. A software algorithm determined VB and IVD height by spatially calculating the distance between superior and inferior surfaces. Vertebral body compression was defined as the difference in height at the center of the VB between pretest and post-test micro-CT images. Intervertebral disc thickness was compared spatially in four distinct equally sized regions along the anterior-posterior axis (anterior, anterior middle, middle posterior, and posterior). Intervertebral disc compression was defined as the change in thickness between pretest and post-test images. One superior IVD in the vertebroplasty group could not be fully reconstructed and was excluded from analysis. Percent change in adjacent IVD circumference was also quantified based on differences in the average perimeter in pretesting and post-testing micro-CT scans. Contours were drawn along the perimeter on four different coronal slices that spanned a centrally located region in superior and inferior adjacent IVDs. Intervertebral disc perimeter was calculated as the average length of the four contours.
Cyclic loading All spine segments were axially compressed at 10 N/s to 685 N followed by cyclic compression between 685 and 1,370 N at 2 Hz. Based on Centers for Disease Control and Prevention health statistics, 685 N is the 50th percentile body weight for 70- to 80-year-old American females [28]. In addition, the chosen cyclic loads are in good agreement
Statistical analysis All statistical analyses were conducted using SAS, version 9.2 (SAS Institute, Cary, NC, USA). Statistical tests were two sided with a .05 significance level. Values reported were mean6 standard deviation. Demographic characteristics of the
Table Summary of age, bone density, and trabecular microarchitecture parameters for control and vertebroplasty groups Group
Age (y)
BMD (g/cm2)
T-score
BVF (%)
Tb.Th (mm)
Conn.D (1/mm3)
SMI
Tb.N (1/mm)
Tb.Sp (mm)
DA
Control Vertebroplasty
77614 7869
0.7560.11 0.7460.12
2.760.9 2.961.1
8.361.1 8.061.3
0.11360.005 0.11360.006
4.260.8 4.061.3
2.260.2 2.360.2
1.0460.13 1.0260.10
0.9860.13 0.9960.09
1.2760.08 1.2760.06
BMD, bone mineral density; BVF, bone volume fraction; Tb.Th, trabecular thickness; Tb.N, trabecular number; Tb.Sp, trabecular spacing; Conn.D, connectivity density; SMI, structure model index; DA, degree of anisotropy. Donors were assigned into control and vertebroplasty groups to have statistically similar (pO.5) bone parameters (mean6standard deviation).
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
1875
Fig. 2. Fluoroscopy images of vertebroplasty procedure on a representative spine section. L1 level was injected with bone cement using a unilateral transpedicular approach. (Left) The needle was inserted into the midline of the vertebral body in the lateral view. About 5 mL of polymethylmethacrylate bone cement was injected into the anterior three-fourths without cement leakage as determined by (Middle) anterior-posterior and (Right) medial-lateral views.
samples were compared between the vertebroplasty and untreated control groups using two-sample independent t tests (similar results were obtained when using Wilcoxon rank sum test, results not shown). The time history of the cyclic data was analyzed using a mixed-effects repeated measures analysis of variance (ANOVA). In this analysis, the effect of vertebroplasty
on strain was assessed while controlling for the effect of cycles and a random effect to adjust for between-subject variability. A quadratic fit of cycles was used to model the data because of the curved responses over time. The effect of covariates (BMD, bone volume fraction, T-score, and age) was assessed via two-way ANOVAs of the strain values at 115,000 cycles using treatment, covariate, and
Fig. 3. Representative three-dimensional height map of an (Top Left and Top Right) adjacent VB and (Bottom Left and Bottom Right) IVD, (Top Left and Bottom Left) before and (Top Right and Bottom Right) after cyclic loading. The VB height was calculated as the maximum superior-inferior distance in the center of the VB (denoted in red). The IVD thickness was divided spatially into four regions indicated with dotted lines. Percent compression was then calculated from the difference in height before and after cyclic loading normalized by pretesting height. VB, vertebral body; IVD, intervertebral disc.
1876
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
Fig. 4. Strain versus cycles for control and vertebroplasty groups. Each point represents the mean strain at a given cycle. Vertebroplasty-treated samples compressed more than untreated controls during cyclic testing (p!.001).
a treatment-by-covariate interaction. For the vertebroplasty group, simple regression analyses were performed on strain at 115,000 cycles with BMD (linear and quadratic fit) and cement fill (linear). In addition, multiple regression analysis was performed on strain with both BMD (linear and quadratic fits) and cement volume (linear fit) as factors. A random-effects general linear model was used to assess compression in adjacent vertebrae and IVDs; treatment, level, region (if applicable), and their interaction were treated as fixed effects, and subject was treated as random.
cycles (Fig. 5). When analyzed separately, vertebroplastytreated specimens with lower BMD values had increased compressive strain relative to samples with higher BMD values (p5.004); untreated controls did not have this response (p5.98). Similarly, subgroup analysis for T-scores indicated a significant difference for vertebroplastytreated subjects (p5.005) and no effect for untreated controls (p5.69). For the vertebroplasty group, BMD (R250.64) and cement fill (R250.39) were significantly correlated (p#.05) with strain at 115,000 cycles (data not shown). Multiple regression analysis using both BMD and cement fill as factors resulted in an adjusted R2 value of 0.80 at 115,000 cycles (Fig. 6).
Results Spine segment biomechanics Stiffness during monotonic loading was significantly higher (p5.02) in the vertebroplasty group (461.4694.5 N/mm) compared with the nontreated controls (358.66 61.3 N/mm). This increased stiffness resulted in consistently lower strains in the vertebroplasty group during the ramp phase. At the end of the ramp, total compressive strain in vertebroplasty-treated samples (2.260.5%) was significantly less than strain in the control group (3.36 0.7%, p!.001). The strain increased in both groups with cycling loading; however, by 10,000 cycles the vertebroplasty group had compressed more relative to nontreated controls (Fig. 4). At the end of cyclic compression, strain in the vertebroplasty group (14.162.7%) was higher (p5 .07) than strain in the control group (12.161.5%). Cyclic strains (strain without ramp included) were constantly higher in the vertebroplasty group (p!.001) and by 115,000 cycles were significantly greater for vertebroplasty specimens (11.262.5%) as compared to nontreated controls (8.061.7%, p5.005), translating into approximately a 3 mm difference in displacement between groups. Bone mineral density was a significant factor in the difference between groups (p5.03) for strain at 115,000
Adjacent VB and IVD compression Compression in the SAVB was four times greater on average in the vertebroplasty group (4.567.3%) than the
Fig. 5. Strain versus BMD for control and vertebroplasty groups. Each point represents the cyclic strain of a spine section after 115,000 cycles of loading. BMD was a significant covariate that explained differences between groups (p!.01). BMD, bone mineral density.
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
1877
Fig. 6. Strain versus BMD and cement fill for vertebroplasty group. Each black circle represents the cyclic strain of a spine section after 115,000 cycles of loading. Regression analysis determined that BMD and cement fill were significant factors (p!.01) in strain values for the vertebroplasty group (R250.80). BMD, bone mineral density.
control group (1.160.9%) (p5.05) (Fig. 7). In the IAVB, there was no difference in compression (p5.93) between the vertebroplasty group (0.661.5%) and the control group (0.460.8%). Superior adjacent VB deformations were higher compared with the IAVB in the vertebroplasty group (p5.03) but not in the control group (p5.69). No covariates (BMD, T-score, and age) were found to be a factor in the increased compression observed in SAVB for the vertebroplasty group. Overall ANOVA indicated a significant effect of treatment (vertebroplasty vs. control, p!.01) and adjacent level (superior vs. inferior, p5.02) on IVD compression. In the vertebroplasty group, superior adjacent IVDs (T12–L1) experienced approximately 13% higher axial compression (p!.001) in the anterior region (23.4613.9%) when
compared with controls (10.3610.2%) (Fig. 8). Similarly, vertebroplasty specimens exhibited significantly higher axial compression (18.5610.9%) in the anterior-middle region of the IVD when compared with controls (9.768.7%, p5.01). The increased compression for vertebroplastytreated spines in the anterior portions of the superior IVD correlated well with low BMD and T-scores (p!.03). However, deformations in the middle-posterior and posterior regions of the superior adjacent IVD were not different between groups. Circumference change in the superior adjacent IVD in the vertebroplasty group (2.561.8%) was fourfold greater on average than in controls (0.661.5%, p5.01). Axial compression in the inferior adjacent IVD (L1–L2) was similar between vertebroplasty (15–19%) and control groups (13–19%, pO.26) for all four regions (Fig. 9). There were also no differences in average circumference between the groups in the inferior adjacent IVD. In addition, BMD
Fig. 7. Compression in adjacent VBs. In the vertebroplasty group, compression in the superior adjacent VB was significantly higher than controls (*p5.05). There was no significant difference between groups in the inferior adjacent VB (p5.93). Superior adjacent VBs in the vertebroplasty group compressed more than inferior VBs (yp5.03). VB, vertebral body.
Fig. 8. Compression in superior adjacent IVD. The anterior and anteriormiddle regions compressed 13% and 9%, respectively, greater on average in vertebroplasty samples (*p!.001 and yp5.01). No significant differences were observed in middle-posterior and posterior regions of the IVD. IVD, intervertebral disc.
1878
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
Fig. 9. Compression in inferior adjacent IVD. No significant differences were observed in between vertebroplasty and control groups or between regions of the IVD. IVD, intervertebral disc.
and T-score were not significant factors in circumference change between the two groups (p$.53). Pairwise comparisons did not detect differences between superior and inferior IVDs for either group (.12#p#.29).
Discussion The proper treatment and management of VCFs still present important challenges in light of an increasingly elderly population. One of the limitations with vertebroplasty for treatment of VCFs is reports of increased incident fractures, especially in adjacent vertebrae. Although these subsequent fractures have been well documented, a clear biomechanical understanding of how dynamic loading of vertebrae augmented with bone cement increases fracture risk in adjacent vertebral levels is still to be determined. By carefully selecting and grouping our cadaver specimens to minimize variability, we were able to conduct a controlled study that specifically investigated the biomechanical consequences of vertebroplasty as a treatment for osteoporotic VCFs. The results indicated that the increased stiffness of treated spines reduced strain during monotonic loading, whereas repetitive loading was detrimental biomechanically. However, not all vertebroplasty-treated spines experienced high deformations. Vertebroplasty treatment in osteoporotic spines (BMD between 0.6 and 0.7 g/cm2 and T-scores between 3.0 and 4.0) had increased strains when compared with similar nonaugmented spines. Furthermore, vertebroplasty-treated cadavers with both reduced BMD and high levels of cement fill sustained the greatest spinal compression. Previous biomechanical studies investigating adjacent vertebral fractures have investigated postvertebroplasty changes in ultimate load or more locally in cortical shell strains from quasi-static loading to failure [31–33]. These articles demonstrated that cement augmentation reduced spinal strength and failure loads up to 48% as well as increased cortical strains in adjacent levels. A recent study found that vertebroplasty increased collapse of the trabecular region in the adjacent superior vertebra when subjected
to fatigue loading [34]. Our study expands on previous studies by examining how increased global strains with vertebroplasty manifested locally as permanent deformations in the center of adjacent vertebrae and spatially in IVDs under physiologically relevant cyclic loading conditions. We developed a novel specimen-specific technique that used repeated micro-CT measurements of each sample to determine that vertebroplasty-treated donors experienced higher compression in the superior adjacent levels of the spine segment. The SAVB was compressed on average over four times more in the vertebroplasty-treated group. In addition, the superior IVD (T12–L1) deformed twice as much axially in the anterior region and four times more circumferentially. These results support the theory that increased stiffness of the augmented vertebra may weaken the rest of the spine by altering load transfer to the adjacent VBs and IVDs [14,32]. Previous finite element studies reported 4% to 6% higher principal strains in adjacent vertebrae with vertebroplasty after static loading [35,36]. Although our results suggest that vertebroplasty results in substantially higher compression (twofold to fourfold) in the superior adjacent levels as compared with these studies, the effect may be a result of time-dependent cyclic loading. We hypothesize that the high stiffness of the bone cement prevented deformation in the treated vertebra, which increased stress on the adjacent superior IVD and VB. The magnitude of this increased stress is likely dependent on the quality of the adjacent VBs and IVDs. We suspect that donors with low BMD or T-scores in the vertebroplasty group experienced increased local stresses with repetitive loading. The increased stress gradually compressed the superior adjacent levels in a time-dependent manner for donors with low BMD or T-score. In addition, low BMD or T-scores correlated well with increased compression in the anterior region of the superior adjacent IVD. We believe that the significantly higher compression observed in the anterior of the IVD arose from placement of bone cement in the anterior portions of the treated vertebra, directly below the anterior region of the superior IVD. Over time, increased stress in the IVD and its deceased ability to properly distribute load to the adjacent VB may contribute to clinically reported incident fractures, particularly in adjacent vertebrae. Clinical studies predict an approximately equal distribution of superior and inferior adjacent vertebral fractures after vertebroplasty [37,38]. However, our results suggest that biomechanical differences may occur preferentially in the SAVB. These results are in agreement with previous mechanical testing studies that report at least a four times higher fracture rate of the SAVB [13,39]. Increased fractures in SAVB may be because of the smaller cross-sectional area of superior levels producing larger stresses when compared with IAVB. It is also possible that the discrepancy between clinical and bench testing may be a consequence of the mechanical test where specimens were loaded from the superior surface and fixed at the inferior surface.
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
Several limitations must be considered when interpreting these data. Our study did not account for the fracture healing response that would help stabilize the original fracture and potentially reduce the stress transfer to adjacent levels. This limitation associated with cadaver studies would be better addressed through animal models that allow for healing of the fractured vertebra. In addition, we focused on elderly Caucasian women as this demographic is susceptible to VCFs; thus, our findings may not be applicable to other populations. Finally, we recognize that multiple freeze-thaw cycles during preparation and imaging may have altered the biomechanical properties of the spine segments. We designed this study to have a total of three freeze-thaw cycles to minimize these effects. Furthermore, by exposing both groups to the same number of defrosting cycles, we believe that changes in spine biomechanics would be similar between the two groups, particularly with the similar cadaver population. In conclusion, we developed a specimen-specific technique to understand whether vertebroplasty alters spine biomechanics in vertebrae and IVDs adjacent to the augmented vertebra. By incorporating micro-CT imaging before and after cyclic mechanical testing, we were able to determine that vertebroplasty increased compression of superior adjacent vertebrae and IVDs up to fourfold in osteoporotic women. These findings may explain increased reports of adjacent fractures after vertebroplasty and can be used clinically when deciding on treatment plans. In particular, the strong correlation of increased strain in cadavers with low BMD and high cement fill may aid in improving surgical intervention strategies for women who require vertebroplasty to treat particularly painful fractures.
Acknowledgments This research study was funded by a grant through the US Food and Drug Administration’s Office of Women’s Health. The authors thank Anton Dmitriev, PhD, and William Pritchard, MD/PhD, for assistance with vertebroplasty procedures; David Baer, PhD, and Valerie Elliott for assistance with DXA measurements; and Ronald Wade and Anthony Pleasant for cadaver procurement.
References [1] Melton LJ. Epidemiology of spinal osteoporosis. Spine 1997;22: 2S–11S. [2] Stock H, Schneider A, Strauss E. Osteoporosis: a disease in men. Clin Orthop Relat Res 2004;425:143–51. [3] Kinoshita T, Ebara S, Kamimura M, et al. Nontraumatic lumbar vertebral compression fracture as a risk factor for femoral neck fractures in involutional osteoporotic patients. J Bone Miner Metab 1999;17: 201–5. [4] Chosa K, Naito A, Awai K. Newly developed compression fractures after percutaneous vertebroplasty: comparison with conservative treatment. Jpn J Radiol 2011;29:335–41.
1879
[5] Lin EP, Ekholm S, Hiwatashi A, Westesson PL. Vertebroplasty: cement leakage into the disc increases the risk of new fracture of adjacent vertebral body. AJNR Am J Neuroradiol 2004;25:175–80. [6] Kamano H, Hiwatashi A, Kobayashi N, et al. New vertebral compression fractures after prophylactic vertebroplasty in osteoporotic patients. AJR Am J Roentgenol 2011;197:451–6. [7] Trout AT, Kallmes DF. Does vertebroplasty cause incident vertebral fractures? A review of available data. AJNR Am J Neuroradiol 2006;27:1397–403. [8] Voormolen MHJ, Lohle PNM, Juttmann JR, et al. The risk of new osteoporotic vertebral compression fractures in the year after percutaneous vertebroplasty. J Vasc Interv Radiol 2006;17:71–6. [9] Trout AT, Kallmes DF, Kaufmann TJ. New fractures after vertebroplasty: adjacent fractures occur significantly sooner. AJNR Am J Neuroradiol 2006;27:217–33. [10] Kim SH, Kang HS, Choi JA, Ahn JM. Risk factors of new compression fractures in adjacent vertebrae after percutaneous vertebroplasty. Acta Radiol 2004;45:440–5. [11] Han I-H, Chin D-K, Kuh S-U, et al. Magnetic resonance imaging findings of subsequent fractures after vertebroplasty. Neurosurgery 2009;64:740–4. [12] Molloy S, Mathis JM, Belkoff SM. The effect of vertebral body percentage fill on mechanical behavior during percutaneous vertebroplasty. Spine 2003;28:1549–54. [13] Boger A, Heini P, Markus W, Schneider E. Adjacent vertebral failure after vertebroplasty: a biomechanical study of low-modulus PMMA cement. Eur Spine J 2007;16:2118–25. [14] Baroud G, Heini P, Nemes J, et al. Biomechanical explanation of adjacent fractures following vertebroplasty. Radiology 2003;229: 607–8. [15] Baroud G, Bohner M. Biomechanical impact of vertebroplasty. Postoperative biomechanics of vertebroplasty. Joint Bone Spine 2006;73: 144–50. [16] Luo J, Skrzypiec DM, Pollintine P, et al. Mechanical efficacy of vertebroplasty: influence of cement type, BMD, fracture severity, and disc degeneration. Bone 2007;40:1110–9. [17] Farooq N, Park JC, Pollintine P, et al. Can vertebroplasty restore normal load-bearing to fractured vertebrae? Spine 2005;30:1723–30. [18] Kayanja MM, Evans K, Milks R, Lieberman IH. Adjacent level load transfer following vertebral augmentation in the cadaveric spine. Spine 2006;31:E790–7. [19] Samelson EJ, Hannan MT, Zhang Y, et al. Incidence and risk factors for vertebral fracture in women and men: 25-year follow-up results from the population-based Framingham study. J Bone Miner Res 2006;21:1207–14. [20] Kettler A, Schmoelz W, Shezifi Y, et al. Biomechanical performance of the new BeadEx implant in the treatment of osteoporotic vertebral body compression fractures: restoration and maintenance of height and stability. Clin Biomech 2006; 21:676–82. [21] Hildebrand T, Ruegsegger P. A new method for the modelindependent assessment of thickness in three-dimensional images. J Microsc 1997;185:67–75. [22] Hildebrand T, Laib A, Muller R, et al. Direct three-dimensional morphometric analysis of human cancellous bone: microstructural data from spine, femur, iliac crest, and calcaneus. J Bone Miner Res 1999;14:1167–74. [23] Kallmes DF, Comstock BA, Heagerty PJ, et al. A randomized trial of vertebroplasty for osteoporotic spinal fractures. N Engl J Med 2009;361:569–79. [24] Buchbinder R, Osborne RH, Kallmes D. Vertebroplasty appears no better than placebo for painful osteoporotic spinal fractures, and has potential to cause harm. Med J Aust 2009;191:476–7. [25] Diamond TH, Champion B, Clark WA. Management of acute osteoporotic vertebral fractures: a nonrandomized trial comparing percutaneous vertebroplasty with conservative therapy. Am J Med 2003;114: 257–65.
1880
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
[26] Grados F, Depriester C, Cayrolle G, et al. Long-term observations of vertebral osteoporotic fractures treated by percutaneous vertebroplasty. Rheumatology (Oxford) 2000;39:1410–4. [27] Kaufmann TJ, Trout AT, Kallmes DF. The effects of cement volume on clinical outcomes of percutaneous vertebroplasty. AJNR Am J Neuroradiol 2006;27:1933–7. [28] McDowell MA, Fryar CD, Ogden CL, Flegal KM. Anthropometric reference data for children and adults: United States, 2003-2006. National Health Statistics Reports 10. Atlanta, GA: Centers for Disease Control, 2008. [29] Cappozzo A. Compressive loads in the lumbar vertebral column during normal level walking. J Orthop Res 1984;1:292–301. [30] Wilke HJ, Jungkunz B, Wenger K, Claes LE. Spinal segment range of motion as a function of in vitro test conditions: effects of exposure period, accumulated cycles, angular-deformation rate, and moisture condition. Anat Rec 1998;251:15–9. [31] Fahim DK, Sun K, Tawackoli W, et al. Premature adjacent vertebral fracture after vertebroplasty: a biomechanical study. Neurosurgery 2011;69:733–44. [32] Berlemann U, Ferguson SJ, Nolte LP, Heini PF. Adjacent vertebral failure after vertebroplasty. A biomechanical investigation. J Bone Joint Surg Br 2002;84:748–52.
[33] Kayanja MM, Togawa D, Lieberman IH. Biomechanical changes after the augmentation of experimental osteoporotic vertebral compression fractures in the cadaveric thoracic spine. Spine J 2005;5:55–63. [34] Wang JL, Chiang CK, Kuo YW, et al. Mechanism of fractures of adjacent and augmented vertebrae following simulated vertebroplasty. J Biomech 2012;45:1372–8. [35] Baroud G, Nemes J, Heini P, Steffen T. Load shift of the intervertebral disc after a vertebroplasty: a finite-element study. Eur Spine J 2003;12:421–6. [36] Polikeit A, Nolte LP, Ferguson SJ. The effect of cement augmentation on the load transfer in an osteoporotic functional spinal unit: finiteelement analysis. Spine 2003;28:991–6. [37] Anselmetti GC, Manca A, Hirsch J, et al. Percutaneous vertebroplasty in osteoporotic patients: an institutional experience of 1,634 patients with long-term follow-up. J Vasc Interv Radiol 2011;22:1714–20. [38] Mazzantini M, Carpeggiani P, d’Ascanio A, et al. Long-term prospective study of osteoporotic patients treated with percutaneous vertebroplasty after fragility fractures. Osteoporos Int 2011;22:1599–607. [39] Nouda S, Tomita S, Kin A, et al. Adjacent vertebral body fracture following vertebroplasty with polymethylmethacrylate or calcium phosphate cement: biomechanical evaluation of the cadaveric spine. Spine 2009;34:2613–8.
Basic Science
Vertebroplasty increases compression of adjacent IVDs and vertebrae in osteoporotic spines Srinidhi Nagaraja, PhDa,*, Hassan K. Awada, BSa, Maureen L. Dreher, PhDa, Shikha Gupta, PhDa, Scott W. Miller, PhDb a
Division of Solid and Fluid Mechanics, Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, U.S. Food and Drug Administration, 10903 New Hampshire Ave, Silver Spring, MD 20993, USA b Division of Biostatistics, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, U.S. Food and Drug Administration, 10903 New Hampshire Ave, Silver Spring, MD 20993, USA Received 30 August 2012; revised 13 March 2013; accepted 1 June 2013
Abstract
BACKGROUND CONTEXT: Approximately 25% of vertebroplasty patients experience subsequent fractures within 1 year of treatment, and vertebrae adjacent to the cemented level are up to three times more likely to fracture than those further away. The increased risk of adjacent fractures postaugmentation raises concerns that treatment of osteoporotic compression fractures with vertebroplasty may negatively impact spine biomechanics. PURPOSE: To quantify the biomechanical effects of vertebroplasty on adjacent intervertebral discs (IVDs) and vertebral bodies (VBs). STUDY DESIGN: A biomechanics study was conducted using cadaveric thoracolumbar spinal columns from elderly women (age range, 51–98 years). METHODS: Five level motion segments (T11–L3) were assigned to a vertebroplasty treated or untreated control group (n510/group) such that bone mineral density (BMD), trabecular architecture, and age were similar between groups. Compression fractures were created in the L1 vertebra of all specimens, and polymethylmethacrylate bone cement was injected into the fractured vertebra of vertebroplasty specimens. All spine segments underwent cyclic axial compression for 115,000 cycles. Microcomputed tomography imaging was performed before and after cyclic loading to quantify compression in adjacent VBs and IVDs. RESULTS: Cyclic loading increased strains 3% on average in the vertebroplasty group when compared with controls after 115,000 cycles. This global strain manifested locally as approximately fourfold more compression in the superior VB (T12) and two- to fourfold higher axial and circumferential deformations in the superior IVD (T12–L1) of vertebroplasty-treated specimens when compared with untreated controls. Low BMD and high cement fill were significant factors that explained the increased strain in the vertebroplasty-treated group. CONCLUSIONS: These data indicate that vertebroplasty alters spine biomechanics resulting in increased compression of adjacent VB and IVD in severely osteoporotic women and may be the basis for clinical reports of adjacent fractures after vertebroplasty. Published by Elsevier Inc.
Keywords:
Vertebroplasty; Osteoporosis; Biomechanics; Vertebral compression fracture; Adjacent segment disease
FDA device/drug status: Approved for this indication (Bone Cement [Stryker]). Author disclosures: SN: Nothing to disclose. HKA: Nothing to disclose. MLD: Nothing to disclose. SG: Nothing to disclose. SWM: Nothing to disclose. * Corresponding author. U.S. Food and Drug Administration, Center for Devices and Radiological Health, Division of Solid and Fluid Mechanics, 10903 New Hampshire Ave., Building 62, Room 2110, Silver Spring, MD 20993-0002, USA. Tel.: (301) 796-0396; fax: (301) 796-9932. E-mail address: [email protected] (S. Nagaraja) 1529-9430/$ - see front matter Published by Elsevier Inc. http://dx.doi.org/10.1016/j.spinee.2013.06.007
Introduction Osteoporosis-related vertebral compression fractures (VCFs) are a concern for elderly women as they affect 25% of all postmenopausal women in the United States and 40% of women older than 80 years [1,2]. In fact, VCFs in women are three times more frequent than in men of similar ages [3]. A common surgical treatment for VCFs is percutaneous vertebroplasty, a minimally invasive
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
procedure in which polymethylmethacrylate bone cement is injected into the fractured vertebra to stabilize the fracture and relieve associated pain. However, clinical studies report that approximately 25% of the patients experience subsequent fractures within 1 year of the treatment [4,5]. Moreover, vertebrae adjacent to the cemented level represent most (50–67%) of these subsequent VCFs [4,6,7] and are up to three times more likely to fracture than those further away [8–10]. Thus, the potential risk of additional VCFs after procedure may outweigh the benefits of vertebroplasty. However, the causality between vertebroplasty and new fractures has not been established, and there is no consensus on the root cause of this phenomenon. Although vertebroplasty may play a role in adjacent VCFs, many of these patients already have vertebrae weakened by osteoporosis (as evidenced by the initial fracture) and are predisposed to fractures in the same region of the spine. Prior investigations of additional VCFs have been primarily limited to retrospective analyses of clinical data after secondary fractures have occurred. These studies are unable to directly determine the biomechanical effects of vertebroplasty on surrounding spinal tissues, especially because VCFs in adjacent and nonadjacent levels can also be influenced by procedural factors such as the volume and distribution of cement inside the vertebral body (VB) [11,12]. Nonclinical studies have hypothesized that additional fractures are because of the bone cement, which has a stiffness 7 to 10 times greater than osteoporotic vertebral bone [13]. The treated vertebra produces a ‘‘stress riser’’ effect where increased stresses are transferred to intervertebral discs (IVDs) and vertebrae adjacent to the original fracture [14,15]. This mechanistic explanation is plausible; however, other biomechanical studies indicated that vertebroplasty restores compressive loads in the neural arch, IVDs, and VB to prefracture levels [16,17]. For example, Kayanja et al. [18] found that vertebroplasty restored stiffness and adjacent vertebral cortex strains to normal levels and concluded that adjacent level fractures after vertebroplasty are because of severe loading, not vertebroplasty. Therefore, the biomechanical consequences of vertebroplasty on osteopenic/osteoporotic spine remain unclear. The objective of this research study was to elucidate the biomechanical consequences of vertebroplasty on adjacent IVDs and vertebrae.
Materials and methods Specimen preparation Caucasian female cadaveric spines (51–98 years) were acquired from Maryland State Anatomy Board (Baltimore, MD, USA) or National Disease Research Interchange (Philadelphia, PA, USA). Twenty donors were chosen based on radiographs and gross examination that confirmed the absence of device implantation, scoliosis, and vertebral
1873
fractures in the thoracolumbar region of each spine. Dual-energy X-ray absorptiometry (DXA; Hologic, Bedford, MA, USA) scans in the lumbar vertebrae indicated T-scores in the osteopenic/osteoporosis range (0.6 to 4.0) for these donors. Spine segments containing T11–L3 vertebral levels were dissected and cleaned while preserving the structural integrity of the spine section leaving the vertebrae, IVDs, pedicles, ligaments, and bony structures intact. The T11 and L3 vertebral levels of each sample were fixed in rapidly curing epoxy (3M, St. Paul, MN, USA) to provide flat uniform surfaces for mechanical testing. The samples were wrapped in 0.9% saline-soaked gauze and stored at 20 C. Fracture creation A wedge fracture was made in the L1 vertebra as it has the highest fracture incidence in the female lumbar spine [19]. A thin transverse cut was created through the anterior half of the L1 VB in accordance with previously published methods [20]. Compression bending was applied to create the wedge fracture. A custom jig was used to load specimens at a rate of 10 mm/min in a mechanical testing system (MTS Corporation, Eden Prairie, MN, USA). Loading was stopped when compression reached half of the premeasured height of the L1 VB (Fig. 1). Although this method does not precisely simulate the morphology of clinically observed compression fractures, it consistently created anterior wedge fractures in the L1 VB by weakening the anterior cortex and internal trabecular microstructure. Specimen groups Microcomputed tomography (micro-CT; Scanco Medical, Wayne, PA, USA) was used to assess trabecular microarchitecture and mineralization in each donor after creating the wedge fracture. Micro-CT scans (voxel size of 51.4 mm) were used to quantify trabecular microstructure in the superior adjacent VB (SAVB, ie, T12) and inferior adjacent VB (IAVB, ie, L2). Parameters such as bone volume fraction, trabecular thickness, trabecular spacing, trabecular number, connectivity density, structure model index, and degree of anisotropy were calculated [21,22]. Using micro-CT and DXA data, specimens were assigned to an experimental group that underwent vertebroplasty and a control group that did not (n510/group) such that trabecular microarchitecture, bone mineral density (BMD), and age were statistically similar between groups (Table). Vertebroplasty procedure For specimens assigned to the vertebroplasty group, an 11-gauge needle was advanced into the L1 fractured VB using a unilateral transpedicular approach (Fig. 2). Polymethylmethacrylate bone cement (Stryker, Kalamazoo, MI, USA) was injected into the anterior three-fourths of the VB based on lateral projections [10,23,24]. To prevent leakage of bone cement, parafilm was tightly wrapped
1874
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
with previous observations of spinal compressive loads up to 2.5 times body weight during brisk walking [29]. Cyclic loading was stopped at 115,000 cycles (approximately 16 hours) to prevent significant degradation or changes in spinal range of motion [30]. Mechanical testing was conducted at room temperature, and specimens were wrapped in saline-soaked gauze with fluid sprayed intermittently to maintain hydration throughout testing. Force-displacement data were continuously recorded throughout ramp and every 1,000 cycles during cyclic loading. Specimen strain was defined as the displacement divided by the original specimen height. Specimen stiffness was calculated in the linear region of the final 0.25% strain during ramp. Compression in adjacent VBs and IVDs
Fig. 1. Representative image of a cadaver specimen in wedge fracture jig. A transverse cut was initially created in the L1 vertebra to facilitate wedge fracture. Compression bending was applied to specimens through loading off axis.
around the wedge fracture. Fluoroscopy during the procedure and micro-CT imaging after the procedure were used to confirm containment of bone cement within the VB, without leakage into the adjacent IVD or outside the VB. The injected volume (5.760.5 mL) and cement fill (2263%) were similar to values reported clinically [25– 27]. After augmentation, cement was allowed to cure for approximately 4 hours at room temperature. Specimens were then stored at 20 C for up to 2 weeks before testing.
All specimens were reimaged with micro-CT after cyclic loading to assess changes in VB height and IVD thickness from precyclic loading (Fig. 3). Permanent axial compression in adjacent VBs and IVDs was calculated based on height change from premechanical and postmechanical testing scans. A software algorithm determined VB and IVD height by spatially calculating the distance between superior and inferior surfaces. Vertebral body compression was defined as the difference in height at the center of the VB between pretest and post-test micro-CT images. Intervertebral disc thickness was compared spatially in four distinct equally sized regions along the anterior-posterior axis (anterior, anterior middle, middle posterior, and posterior). Intervertebral disc compression was defined as the change in thickness between pretest and post-test images. One superior IVD in the vertebroplasty group could not be fully reconstructed and was excluded from analysis. Percent change in adjacent IVD circumference was also quantified based on differences in the average perimeter in pretesting and post-testing micro-CT scans. Contours were drawn along the perimeter on four different coronal slices that spanned a centrally located region in superior and inferior adjacent IVDs. Intervertebral disc perimeter was calculated as the average length of the four contours.
Cyclic loading All spine segments were axially compressed at 10 N/s to 685 N followed by cyclic compression between 685 and 1,370 N at 2 Hz. Based on Centers for Disease Control and Prevention health statistics, 685 N is the 50th percentile body weight for 70- to 80-year-old American females [28]. In addition, the chosen cyclic loads are in good agreement
Statistical analysis All statistical analyses were conducted using SAS, version 9.2 (SAS Institute, Cary, NC, USA). Statistical tests were two sided with a .05 significance level. Values reported were mean6 standard deviation. Demographic characteristics of the
Table Summary of age, bone density, and trabecular microarchitecture parameters for control and vertebroplasty groups Group
Age (y)
BMD (g/cm2)
T-score
BVF (%)
Tb.Th (mm)
Conn.D (1/mm3)
SMI
Tb.N (1/mm)
Tb.Sp (mm)
DA
Control Vertebroplasty
77614 7869
0.7560.11 0.7460.12
2.760.9 2.961.1
8.361.1 8.061.3
0.11360.005 0.11360.006
4.260.8 4.061.3
2.260.2 2.360.2
1.0460.13 1.0260.10
0.9860.13 0.9960.09
1.2760.08 1.2760.06
BMD, bone mineral density; BVF, bone volume fraction; Tb.Th, trabecular thickness; Tb.N, trabecular number; Tb.Sp, trabecular spacing; Conn.D, connectivity density; SMI, structure model index; DA, degree of anisotropy. Donors were assigned into control and vertebroplasty groups to have statistically similar (pO.5) bone parameters (mean6standard deviation).
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
1875
Fig. 2. Fluoroscopy images of vertebroplasty procedure on a representative spine section. L1 level was injected with bone cement using a unilateral transpedicular approach. (Left) The needle was inserted into the midline of the vertebral body in the lateral view. About 5 mL of polymethylmethacrylate bone cement was injected into the anterior three-fourths without cement leakage as determined by (Middle) anterior-posterior and (Right) medial-lateral views.
samples were compared between the vertebroplasty and untreated control groups using two-sample independent t tests (similar results were obtained when using Wilcoxon rank sum test, results not shown). The time history of the cyclic data was analyzed using a mixed-effects repeated measures analysis of variance (ANOVA). In this analysis, the effect of vertebroplasty
on strain was assessed while controlling for the effect of cycles and a random effect to adjust for between-subject variability. A quadratic fit of cycles was used to model the data because of the curved responses over time. The effect of covariates (BMD, bone volume fraction, T-score, and age) was assessed via two-way ANOVAs of the strain values at 115,000 cycles using treatment, covariate, and
Fig. 3. Representative three-dimensional height map of an (Top Left and Top Right) adjacent VB and (Bottom Left and Bottom Right) IVD, (Top Left and Bottom Left) before and (Top Right and Bottom Right) after cyclic loading. The VB height was calculated as the maximum superior-inferior distance in the center of the VB (denoted in red). The IVD thickness was divided spatially into four regions indicated with dotted lines. Percent compression was then calculated from the difference in height before and after cyclic loading normalized by pretesting height. VB, vertebral body; IVD, intervertebral disc.
1876
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
Fig. 4. Strain versus cycles for control and vertebroplasty groups. Each point represents the mean strain at a given cycle. Vertebroplasty-treated samples compressed more than untreated controls during cyclic testing (p!.001).
a treatment-by-covariate interaction. For the vertebroplasty group, simple regression analyses were performed on strain at 115,000 cycles with BMD (linear and quadratic fit) and cement fill (linear). In addition, multiple regression analysis was performed on strain with both BMD (linear and quadratic fits) and cement volume (linear fit) as factors. A random-effects general linear model was used to assess compression in adjacent vertebrae and IVDs; treatment, level, region (if applicable), and their interaction were treated as fixed effects, and subject was treated as random.
cycles (Fig. 5). When analyzed separately, vertebroplastytreated specimens with lower BMD values had increased compressive strain relative to samples with higher BMD values (p5.004); untreated controls did not have this response (p5.98). Similarly, subgroup analysis for T-scores indicated a significant difference for vertebroplastytreated subjects (p5.005) and no effect for untreated controls (p5.69). For the vertebroplasty group, BMD (R250.64) and cement fill (R250.39) were significantly correlated (p#.05) with strain at 115,000 cycles (data not shown). Multiple regression analysis using both BMD and cement fill as factors resulted in an adjusted R2 value of 0.80 at 115,000 cycles (Fig. 6).
Results Spine segment biomechanics Stiffness during monotonic loading was significantly higher (p5.02) in the vertebroplasty group (461.4694.5 N/mm) compared with the nontreated controls (358.66 61.3 N/mm). This increased stiffness resulted in consistently lower strains in the vertebroplasty group during the ramp phase. At the end of the ramp, total compressive strain in vertebroplasty-treated samples (2.260.5%) was significantly less than strain in the control group (3.36 0.7%, p!.001). The strain increased in both groups with cycling loading; however, by 10,000 cycles the vertebroplasty group had compressed more relative to nontreated controls (Fig. 4). At the end of cyclic compression, strain in the vertebroplasty group (14.162.7%) was higher (p5 .07) than strain in the control group (12.161.5%). Cyclic strains (strain without ramp included) were constantly higher in the vertebroplasty group (p!.001) and by 115,000 cycles were significantly greater for vertebroplasty specimens (11.262.5%) as compared to nontreated controls (8.061.7%, p5.005), translating into approximately a 3 mm difference in displacement between groups. Bone mineral density was a significant factor in the difference between groups (p5.03) for strain at 115,000
Adjacent VB and IVD compression Compression in the SAVB was four times greater on average in the vertebroplasty group (4.567.3%) than the
Fig. 5. Strain versus BMD for control and vertebroplasty groups. Each point represents the cyclic strain of a spine section after 115,000 cycles of loading. BMD was a significant covariate that explained differences between groups (p!.01). BMD, bone mineral density.
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
1877
Fig. 6. Strain versus BMD and cement fill for vertebroplasty group. Each black circle represents the cyclic strain of a spine section after 115,000 cycles of loading. Regression analysis determined that BMD and cement fill were significant factors (p!.01) in strain values for the vertebroplasty group (R250.80). BMD, bone mineral density.
control group (1.160.9%) (p5.05) (Fig. 7). In the IAVB, there was no difference in compression (p5.93) between the vertebroplasty group (0.661.5%) and the control group (0.460.8%). Superior adjacent VB deformations were higher compared with the IAVB in the vertebroplasty group (p5.03) but not in the control group (p5.69). No covariates (BMD, T-score, and age) were found to be a factor in the increased compression observed in SAVB for the vertebroplasty group. Overall ANOVA indicated a significant effect of treatment (vertebroplasty vs. control, p!.01) and adjacent level (superior vs. inferior, p5.02) on IVD compression. In the vertebroplasty group, superior adjacent IVDs (T12–L1) experienced approximately 13% higher axial compression (p!.001) in the anterior region (23.4613.9%) when
compared with controls (10.3610.2%) (Fig. 8). Similarly, vertebroplasty specimens exhibited significantly higher axial compression (18.5610.9%) in the anterior-middle region of the IVD when compared with controls (9.768.7%, p5.01). The increased compression for vertebroplastytreated spines in the anterior portions of the superior IVD correlated well with low BMD and T-scores (p!.03). However, deformations in the middle-posterior and posterior regions of the superior adjacent IVD were not different between groups. Circumference change in the superior adjacent IVD in the vertebroplasty group (2.561.8%) was fourfold greater on average than in controls (0.661.5%, p5.01). Axial compression in the inferior adjacent IVD (L1–L2) was similar between vertebroplasty (15–19%) and control groups (13–19%, pO.26) for all four regions (Fig. 9). There were also no differences in average circumference between the groups in the inferior adjacent IVD. In addition, BMD
Fig. 7. Compression in adjacent VBs. In the vertebroplasty group, compression in the superior adjacent VB was significantly higher than controls (*p5.05). There was no significant difference between groups in the inferior adjacent VB (p5.93). Superior adjacent VBs in the vertebroplasty group compressed more than inferior VBs (yp5.03). VB, vertebral body.
Fig. 8. Compression in superior adjacent IVD. The anterior and anteriormiddle regions compressed 13% and 9%, respectively, greater on average in vertebroplasty samples (*p!.001 and yp5.01). No significant differences were observed in middle-posterior and posterior regions of the IVD. IVD, intervertebral disc.
1878
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
Fig. 9. Compression in inferior adjacent IVD. No significant differences were observed in between vertebroplasty and control groups or between regions of the IVD. IVD, intervertebral disc.
and T-score were not significant factors in circumference change between the two groups (p$.53). Pairwise comparisons did not detect differences between superior and inferior IVDs for either group (.12#p#.29).
Discussion The proper treatment and management of VCFs still present important challenges in light of an increasingly elderly population. One of the limitations with vertebroplasty for treatment of VCFs is reports of increased incident fractures, especially in adjacent vertebrae. Although these subsequent fractures have been well documented, a clear biomechanical understanding of how dynamic loading of vertebrae augmented with bone cement increases fracture risk in adjacent vertebral levels is still to be determined. By carefully selecting and grouping our cadaver specimens to minimize variability, we were able to conduct a controlled study that specifically investigated the biomechanical consequences of vertebroplasty as a treatment for osteoporotic VCFs. The results indicated that the increased stiffness of treated spines reduced strain during monotonic loading, whereas repetitive loading was detrimental biomechanically. However, not all vertebroplasty-treated spines experienced high deformations. Vertebroplasty treatment in osteoporotic spines (BMD between 0.6 and 0.7 g/cm2 and T-scores between 3.0 and 4.0) had increased strains when compared with similar nonaugmented spines. Furthermore, vertebroplasty-treated cadavers with both reduced BMD and high levels of cement fill sustained the greatest spinal compression. Previous biomechanical studies investigating adjacent vertebral fractures have investigated postvertebroplasty changes in ultimate load or more locally in cortical shell strains from quasi-static loading to failure [31–33]. These articles demonstrated that cement augmentation reduced spinal strength and failure loads up to 48% as well as increased cortical strains in adjacent levels. A recent study found that vertebroplasty increased collapse of the trabecular region in the adjacent superior vertebra when subjected
to fatigue loading [34]. Our study expands on previous studies by examining how increased global strains with vertebroplasty manifested locally as permanent deformations in the center of adjacent vertebrae and spatially in IVDs under physiologically relevant cyclic loading conditions. We developed a novel specimen-specific technique that used repeated micro-CT measurements of each sample to determine that vertebroplasty-treated donors experienced higher compression in the superior adjacent levels of the spine segment. The SAVB was compressed on average over four times more in the vertebroplasty-treated group. In addition, the superior IVD (T12–L1) deformed twice as much axially in the anterior region and four times more circumferentially. These results support the theory that increased stiffness of the augmented vertebra may weaken the rest of the spine by altering load transfer to the adjacent VBs and IVDs [14,32]. Previous finite element studies reported 4% to 6% higher principal strains in adjacent vertebrae with vertebroplasty after static loading [35,36]. Although our results suggest that vertebroplasty results in substantially higher compression (twofold to fourfold) in the superior adjacent levels as compared with these studies, the effect may be a result of time-dependent cyclic loading. We hypothesize that the high stiffness of the bone cement prevented deformation in the treated vertebra, which increased stress on the adjacent superior IVD and VB. The magnitude of this increased stress is likely dependent on the quality of the adjacent VBs and IVDs. We suspect that donors with low BMD or T-scores in the vertebroplasty group experienced increased local stresses with repetitive loading. The increased stress gradually compressed the superior adjacent levels in a time-dependent manner for donors with low BMD or T-score. In addition, low BMD or T-scores correlated well with increased compression in the anterior region of the superior adjacent IVD. We believe that the significantly higher compression observed in the anterior of the IVD arose from placement of bone cement in the anterior portions of the treated vertebra, directly below the anterior region of the superior IVD. Over time, increased stress in the IVD and its deceased ability to properly distribute load to the adjacent VB may contribute to clinically reported incident fractures, particularly in adjacent vertebrae. Clinical studies predict an approximately equal distribution of superior and inferior adjacent vertebral fractures after vertebroplasty [37,38]. However, our results suggest that biomechanical differences may occur preferentially in the SAVB. These results are in agreement with previous mechanical testing studies that report at least a four times higher fracture rate of the SAVB [13,39]. Increased fractures in SAVB may be because of the smaller cross-sectional area of superior levels producing larger stresses when compared with IAVB. It is also possible that the discrepancy between clinical and bench testing may be a consequence of the mechanical test where specimens were loaded from the superior surface and fixed at the inferior surface.
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
Several limitations must be considered when interpreting these data. Our study did not account for the fracture healing response that would help stabilize the original fracture and potentially reduce the stress transfer to adjacent levels. This limitation associated with cadaver studies would be better addressed through animal models that allow for healing of the fractured vertebra. In addition, we focused on elderly Caucasian women as this demographic is susceptible to VCFs; thus, our findings may not be applicable to other populations. Finally, we recognize that multiple freeze-thaw cycles during preparation and imaging may have altered the biomechanical properties of the spine segments. We designed this study to have a total of three freeze-thaw cycles to minimize these effects. Furthermore, by exposing both groups to the same number of defrosting cycles, we believe that changes in spine biomechanics would be similar between the two groups, particularly with the similar cadaver population. In conclusion, we developed a specimen-specific technique to understand whether vertebroplasty alters spine biomechanics in vertebrae and IVDs adjacent to the augmented vertebra. By incorporating micro-CT imaging before and after cyclic mechanical testing, we were able to determine that vertebroplasty increased compression of superior adjacent vertebrae and IVDs up to fourfold in osteoporotic women. These findings may explain increased reports of adjacent fractures after vertebroplasty and can be used clinically when deciding on treatment plans. In particular, the strong correlation of increased strain in cadavers with low BMD and high cement fill may aid in improving surgical intervention strategies for women who require vertebroplasty to treat particularly painful fractures.
Acknowledgments This research study was funded by a grant through the US Food and Drug Administration’s Office of Women’s Health. The authors thank Anton Dmitriev, PhD, and William Pritchard, MD/PhD, for assistance with vertebroplasty procedures; David Baer, PhD, and Valerie Elliott for assistance with DXA measurements; and Ronald Wade and Anthony Pleasant for cadaver procurement.
References [1] Melton LJ. Epidemiology of spinal osteoporosis. Spine 1997;22: 2S–11S. [2] Stock H, Schneider A, Strauss E. Osteoporosis: a disease in men. Clin Orthop Relat Res 2004;425:143–51. [3] Kinoshita T, Ebara S, Kamimura M, et al. Nontraumatic lumbar vertebral compression fracture as a risk factor for femoral neck fractures in involutional osteoporotic patients. J Bone Miner Metab 1999;17: 201–5. [4] Chosa K, Naito A, Awai K. Newly developed compression fractures after percutaneous vertebroplasty: comparison with conservative treatment. Jpn J Radiol 2011;29:335–41.
1879
[5] Lin EP, Ekholm S, Hiwatashi A, Westesson PL. Vertebroplasty: cement leakage into the disc increases the risk of new fracture of adjacent vertebral body. AJNR Am J Neuroradiol 2004;25:175–80. [6] Kamano H, Hiwatashi A, Kobayashi N, et al. New vertebral compression fractures after prophylactic vertebroplasty in osteoporotic patients. AJR Am J Roentgenol 2011;197:451–6. [7] Trout AT, Kallmes DF. Does vertebroplasty cause incident vertebral fractures? A review of available data. AJNR Am J Neuroradiol 2006;27:1397–403. [8] Voormolen MHJ, Lohle PNM, Juttmann JR, et al. The risk of new osteoporotic vertebral compression fractures in the year after percutaneous vertebroplasty. J Vasc Interv Radiol 2006;17:71–6. [9] Trout AT, Kallmes DF, Kaufmann TJ. New fractures after vertebroplasty: adjacent fractures occur significantly sooner. AJNR Am J Neuroradiol 2006;27:217–33. [10] Kim SH, Kang HS, Choi JA, Ahn JM. Risk factors of new compression fractures in adjacent vertebrae after percutaneous vertebroplasty. Acta Radiol 2004;45:440–5. [11] Han I-H, Chin D-K, Kuh S-U, et al. Magnetic resonance imaging findings of subsequent fractures after vertebroplasty. Neurosurgery 2009;64:740–4. [12] Molloy S, Mathis JM, Belkoff SM. The effect of vertebral body percentage fill on mechanical behavior during percutaneous vertebroplasty. Spine 2003;28:1549–54. [13] Boger A, Heini P, Markus W, Schneider E. Adjacent vertebral failure after vertebroplasty: a biomechanical study of low-modulus PMMA cement. Eur Spine J 2007;16:2118–25. [14] Baroud G, Heini P, Nemes J, et al. Biomechanical explanation of adjacent fractures following vertebroplasty. Radiology 2003;229: 607–8. [15] Baroud G, Bohner M. Biomechanical impact of vertebroplasty. Postoperative biomechanics of vertebroplasty. Joint Bone Spine 2006;73: 144–50. [16] Luo J, Skrzypiec DM, Pollintine P, et al. Mechanical efficacy of vertebroplasty: influence of cement type, BMD, fracture severity, and disc degeneration. Bone 2007;40:1110–9. [17] Farooq N, Park JC, Pollintine P, et al. Can vertebroplasty restore normal load-bearing to fractured vertebrae? Spine 2005;30:1723–30. [18] Kayanja MM, Evans K, Milks R, Lieberman IH. Adjacent level load transfer following vertebral augmentation in the cadaveric spine. Spine 2006;31:E790–7. [19] Samelson EJ, Hannan MT, Zhang Y, et al. Incidence and risk factors for vertebral fracture in women and men: 25-year follow-up results from the population-based Framingham study. J Bone Miner Res 2006;21:1207–14. [20] Kettler A, Schmoelz W, Shezifi Y, et al. Biomechanical performance of the new BeadEx implant in the treatment of osteoporotic vertebral body compression fractures: restoration and maintenance of height and stability. Clin Biomech 2006; 21:676–82. [21] Hildebrand T, Ruegsegger P. A new method for the modelindependent assessment of thickness in three-dimensional images. J Microsc 1997;185:67–75. [22] Hildebrand T, Laib A, Muller R, et al. Direct three-dimensional morphometric analysis of human cancellous bone: microstructural data from spine, femur, iliac crest, and calcaneus. J Bone Miner Res 1999;14:1167–74. [23] Kallmes DF, Comstock BA, Heagerty PJ, et al. A randomized trial of vertebroplasty for osteoporotic spinal fractures. N Engl J Med 2009;361:569–79. [24] Buchbinder R, Osborne RH, Kallmes D. Vertebroplasty appears no better than placebo for painful osteoporotic spinal fractures, and has potential to cause harm. Med J Aust 2009;191:476–7. [25] Diamond TH, Champion B, Clark WA. Management of acute osteoporotic vertebral fractures: a nonrandomized trial comparing percutaneous vertebroplasty with conservative therapy. Am J Med 2003;114: 257–65.
1880
S. Nagaraja et al. / The Spine Journal 13 (2013) 1872–1880
[26] Grados F, Depriester C, Cayrolle G, et al. Long-term observations of vertebral osteoporotic fractures treated by percutaneous vertebroplasty. Rheumatology (Oxford) 2000;39:1410–4. [27] Kaufmann TJ, Trout AT, Kallmes DF. The effects of cement volume on clinical outcomes of percutaneous vertebroplasty. AJNR Am J Neuroradiol 2006;27:1933–7. [28] McDowell MA, Fryar CD, Ogden CL, Flegal KM. Anthropometric reference data for children and adults: United States, 2003-2006. National Health Statistics Reports 10. Atlanta, GA: Centers for Disease Control, 2008. [29] Cappozzo A. Compressive loads in the lumbar vertebral column during normal level walking. J Orthop Res 1984;1:292–301. [30] Wilke HJ, Jungkunz B, Wenger K, Claes LE. Spinal segment range of motion as a function of in vitro test conditions: effects of exposure period, accumulated cycles, angular-deformation rate, and moisture condition. Anat Rec 1998;251:15–9. [31] Fahim DK, Sun K, Tawackoli W, et al. Premature adjacent vertebral fracture after vertebroplasty: a biomechanical study. Neurosurgery 2011;69:733–44. [32] Berlemann U, Ferguson SJ, Nolte LP, Heini PF. Adjacent vertebral failure after vertebroplasty. A biomechanical investigation. J Bone Joint Surg Br 2002;84:748–52.
[33] Kayanja MM, Togawa D, Lieberman IH. Biomechanical changes after the augmentation of experimental osteoporotic vertebral compression fractures in the cadaveric thoracic spine. Spine J 2005;5:55–63. [34] Wang JL, Chiang CK, Kuo YW, et al. Mechanism of fractures of adjacent and augmented vertebrae following simulated vertebroplasty. J Biomech 2012;45:1372–8. [35] Baroud G, Nemes J, Heini P, Steffen T. Load shift of the intervertebral disc after a vertebroplasty: a finite-element study. Eur Spine J 2003;12:421–6. [36] Polikeit A, Nolte LP, Ferguson SJ. The effect of cement augmentation on the load transfer in an osteoporotic functional spinal unit: finiteelement analysis. Spine 2003;28:991–6. [37] Anselmetti GC, Manca A, Hirsch J, et al. Percutaneous vertebroplasty in osteoporotic patients: an institutional experience of 1,634 patients with long-term follow-up. J Vasc Interv Radiol 2011;22:1714–20. [38] Mazzantini M, Carpeggiani P, d’Ascanio A, et al. Long-term prospective study of osteoporotic patients treated with percutaneous vertebroplasty after fragility fractures. Osteoporos Int 2011;22:1599–607. [39] Nouda S, Tomita S, Kin A, et al. Adjacent vertebral body fracture following vertebroplasty with polymethylmethacrylate or calcium phosphate cement: biomechanical evaluation of the cadaveric spine. Spine 2009;34:2613–8.