Eur Spine J DOI 10.1007/s00586-014-3455-z ORIGINAL ARTICLE Modic (endplate) changes in the lumbar spine: bone micro-architecture and remodelling Egon Perilli • Ian H. Parkinson • Le-Hoa Truong • Kuan C. Chong • Nicola L. Fazzalari • Orso L. Osti Received: 2 October 2013 / Revised: 1 July 2014 / Accepted: 2 July 2014 Ó The Author(s) 2014. This article is published with open access at Springerlink.com Abstract structure model index. Then, samples underwent histolog- Purpose In the literature, inter-vertebral MRI signal ical analysis, for determination of bone remodelling indi- intensity changes (Modic changes) were associated with ces: osteoid surface to bone surface ratio (OS/BS), eroded corresponding histological observations on endplate biop- surface to bone surface (ES/BS) and osteoid surface to sies. However, tissue-level studies were limited. No eroded surface ratio (OS/ES). quantitative histomorphometric study on bone biopsies has Results Micro-CT analysis revealed significantly higher yet been conducted for Modic changes. The aim of this BV/TV (up to 70 % increase, p \ 0.01) and Tb.Th (up to study was to characterise the bone micro-architectural ?57 %, p \ 0.01) in Modic 3 biopsies, compared to Modic parameters and bone remodelling indices associated with 1 and 2. Histological analysis showed significantly lower Modic changes. OS/BS in Modic 2 biopsies (more than 28 % decrease, Methods Forty patients suffering from disabling low back p \ 0.05) compared to 1 and 3. ES/BS progressively pain, undergoing elective spinal surgery, and exhibiting decreased from Modic 1 to 2 to 3, whereas OS/ES pro- Modic changes on MRI (Modic 1, n = 9; Modic 2, n = 25; gressively increased with significantly higher values in Modic 3, n = 6), had a transpedicular vertebral body Modic 3 (up to 159 % increase, p \ 0.05) than in Modic 1 biopsy taken of subchondral bone. Biopsies were first and 2. examined by micro-CT, for 3D morphometric analysis of Conclusions Significant differences were found in bone bone volume fraction (BV/TV), trabecular thickness micro-architectural parameters and remodelling indices (Tb.Th), trabecular separation, trabecular number, and among Modic types. Modic 1 biopsies had evidence of highest bone turnover, possibly due to an inflammatory process; Modic 2 biopsies were consistent with a reduced & E. Perilli ( ) bone formation/remodelling stage; Modic 3 biopsies sug- Medical Device Research Institute, School of Computer Science, Engineering and Mathematics, Flinders University, Sturt Rd., gested a more stable sclerotic phase, with significantly Bedford Park, South Australia 5042, Australia increased BV/TV and Tb.Th compared to Modic 1 and 2, e-mail: egon.perilli@flinders.edu.au linked to increased bone formation and reduced resorption. E. Perilli Á I. H. Parkinson Á L.-H. Truong Á K. C. Chong Á N. L. Fazzalari Keywords Modic changes Á Vertebra Á Micro-computed Bone and Joint Research Laboratory, SA Pathology and Hanson tomography Á Histology Á Micro-architecture Á Bone Institute, Adelaide, South Australia, Australia remodelling I. H. Parkinson Á L.-H. Truong Á N. L. Fazzalari Discipline of Anatomy and Pathology, The University of Adelaide, Adelaide, South Australia, Australia Introduction K. C. Chong Á O. L. Osti Low back pain linked to disc degeneration is one among Calvary Healthcare, North Adelaide Campus, 12 Walter St., North Adelaide, South Australia 5006, Australia the leading causes of functional incapacitation and chronic e-mail: [email protected] disability. Modic et al. [1, 2] in 1988 described and 123 Eur Spine J classified characteristic magnetic resonance imaging (MRI) consent form prior to surgery. The reading of MRI images changes in the vertebral body bone marrow and endplates, and grading of Modic changes was carried out indepen- adjacent to degenerating intervertebral discs. Depending on dently and blindly by two radiologists. In case of dis- the changes in T1 and T2 MRI signals, three types of agreement, a consensus was negotiated. The inter-observer lesions were identified (Modic changes): Modic type 1 and intra-observer reliability for the classification of changes (low T1 and high T2 signal) were attributed to Modic lesions was assessed by means of kappa statistics increased blood flow within the bone, type 2 changes (high (Cohen’s j)[11], showing very high agreement (j = 0.83 T1 and T2 signals) possibly reflecting fatty degeneration of and j = 0.92, respectively). the bone marrow. Type 3 changes (low T1 and T2 signals) The age of the 40 patients (17 women, 23 men) was were assumed to correlate with subchondral bone sclerosis 57 ± 11 years (mean ± standard deviation, SD), age range [1, 2]. Their occurrence appears to increase with age and is 35–75 years. The cases were subdivided as follows: Modic associated with low back pain [3, 4]. 1(n = 9), Modic 2 (n = 25), Modic 3 (n = 6). At time of Despite the increasing number of studies published in surgery, one transpedicular vertebral body subchondral the international literature on Modic changes, their aetiol- bone biopsy (up to 21 mm long, 3 mm in diameter) was ogy remains poorly understood [5–9]. Only a few papers harvested under biplanar (anterior-posterior and lateral) have attempted to examine excised vertebral tissue speci- image intensifier at the region corresponding to the Modic mens from patients with Modic changes [1–3]. In particu- lesion as seen on preoperative MRI (Fig. 1), using an 8G lar, to the best of the authors’ knowledge, no quantitative Jamshidi needle. The biopsies were always taken at the comparison among the three Modic types, in micro-archi- superior border of the vertebral body, due to the location of tectural and remodelling characteristics of the trabecular the pedicle (Fig. 1). If the lesion as seen on axial MRI was bone adjacent to the endplates, has yet been published. A asymmetric, the pedicle on the side of the lesion was cho- quantitative analysis of bone micro-architectural parame- sen, as confirmed by anterior-posterior fluoroscopy. The ters and bone remodelling indices would contribute to the bone biopsy sample was carefully retrieved from the needle characterisation of Modic changes, important for a better and preserved in 70 % alcohol. understanding of the pathological process involved. Thus, the aim of this study is to quantify and compare Bone specimen preparation amongst samples of different Modic types the three-dimen- sional (3D) bone micro-architecture by means of micro- All bone samples underwent serial dehydration by computed tomography (micro-CT), and bone remodelling increasing the alcohol concentration up to 100 % over a characteristics by means of histology, in vertebral bone period of 1 week, followed by acetone overnight. No biopsies taken from adults undergoing spinal surgery and decalcification was done. The biopsies were then infiltrated exhibiting Modic changes on preoperative MRI images. and embedded in methylmethacrylate [12]. Micro-CT Materials and methods A micro-CT scan of the embedded biopsies was carried out Forty patients undergoing lumbar spine surgery with pedi- to provide a 3D-analysis of the bone micro-architecture cle screw instrumentation and exhibiting Modic changes on using a desktop micro-CT system (SkyScan model 1076, preoperative lumbar MRI were selected for this study. In all patients the indication for surgery was disabling low back pain associated with marked disc degeneration at one or two levels. The inclusion criteria for patients were absence of history of bone metabolic diseases, of any anti-osteoporotic treatment, of previous spinal surgery, and of active or sus- pected ongoing infection. The size of the Modic lesions in terms of area in relation to the vertebra was assessed from the sagittal MRI images [10]. All the 40 lesions accepted for biopsy covered at least 19 % of the area (26 % on average). In those cases where inhomogeneities in the images were present (four cases), the largest homogeneous component in both T1 and T2 weighted signals was taken for classifica- Fig. 1 Intraoperative lateral fluoroscopy image (a), showing position tion. All patients were asked to read an information package of the biopsy needle (indicated by arrow) within the Modic lesion as and to sign an ethics committee approved specific informed seen on preoperative sagittal MRI (b) 123 Eur Spine J SkyScan NV, Kontich, Belgium) [13, 14]. The sample was Histology centred on the scanner bed and scanned with the following system settings: source voltage 74 kV, current 100 lA, Then, histological sections were prepared for analysis of tis- rotation step 0.88, rotation over 1808, 1 mm-thick alu- sue-level bone remodelling. All embedded bone specimens minium filter for beam hardening reduction, isotropic pixel were trimmed and sectioned with a microtome (Polycut-E, size 8.7 lm, exposure time 4.7 s, 4 frames averaging. The Leica SP 2600, Leica Microsystems, Wetzlar, Germany). cross-section images were then reconstructed using a fil- Sections, 5 lm thick, were stained using the von Kossa silver tered back-projection algorithm (NRecon software, V1.4.4, method and counterstained with hematoxylin and eosin SkyScan, Kontich, Belgium). Depending on the biopsy, a (H&E) to distinguish between the mineralized bone, the stack of up to 2,400 cross-section images was recon- cellular components of the marrow, and the osteoid. Histo- structed, with inter-slice distance of 1 pixel (8.7 lm), morphometric measurements were done over three sections corresponding to a maximum reconstructed length of taken through the biopsy, in order to be representative of the 21 mm, recreating the full length of the biopsy. The whole biopsy, using an optical microscope (Leica DFC 480) reconstructed axial cross-section images were of with an ocular-mounted 10 9 10 graticule at a magnification 600 9 600 pixel each, 8.7 lm pixel size, corresponding to of 9100. The histological measurements for quantifying the images of 5.2 9 5.2 mm, centred on the bone specimen, remodelling parameters were: percent osteoid surface to bone and saved in 256 grey-level image file format (.bmp).