Synchrotron-Tomography for Evaluation of Bone Tissue Regeneration using rapidly Resorbable Bone Substitute Materials

Alexander RACK, Forschungszentrum Karlsruhe – ANKA, Eggenstein-Leopoldshafen, Germany

Christine KNABE, Michael STILLER, Christian KOCH, Hannah SELIGMANN, Charité Berlin – Experimental Dentistry (CBF), Germany

Simon ZABLER, Hahn-Meitner-Institut Berlin – Department SF3, Germany Gerd WEIDEMANN, Jürgen GOEBBELS, Bundesanstalt für Materialforschung und

–prüfung – Division VIII.3, Berlin, Germany

Abstract. This article presents a novel approach to evaluate two-dimensional histomorphometric studies of biodegradable ceramic particles by means of element-sensitive, three-dimensional and non-destructive synchrotron-microtomography (SCT). An in vivo animal study was performed in which bone substitute materials (352i, GB9/25) were implanted in the sheep mandible to support the bone regeneration. After 12 and 24 weeks of implantation samples were prepared and investigated using SCT and subsequent 3D image analysis as well as histological evaluation. A comparison of corresponding tomographical and histological slices delivers information about the newly formed bone and its stage of development. Additionally SCT gives insights into the structural changes of the bony tissue in a given defect and the local biodegradation of the bone substitute material in a three-dimensional manner.

1. Introduction

The use of oral implants has become a common treatment to replace missing or lost teeth [1, 2]. However, resorption of the alveolar ridge after tooth extraction frequently mandates site development by augmentation before implants can be placed [1, 3, 4]. Therefore, augmentation of the alveolar ridge before implant placement is frequently performed in implant dentistry [3, 4, 5, 6, 7]. The current gold standard for bone reconstruction in implant dentistry is the use of autogenous bone grafts [5, 6, 7]. Among the various techniques to reconstruct or enlarge a deficient alveolar ridge, the concept of guided bone regeneration (GBR) [5] has become a predictable and well-documented surgical approach [8]. The need for localized ridge augmentation prior to the placement of dental implants has been one of the clinical indications for GBR [5]. Using synthetic biodegradable bone substitutes as a membrane-supporting device would simplify GBR, since it avoids second-site surgery [9]. Relatively rapid biodegradation of synthetic biodegradable bone substitutes is desirable, especially prior to dental implant placement, because ideally new bone should form leaving no residual particles that may interfere with preparation of the implant bed at surgery [7]. For example bioactive calcium phosphate ceramics and bioactive glasses are candidate biomaterials which qualify as bone substitutes for this kind of application, since they are widely used in orthopaedics [10, 11]. An ideal bone replacement material should

ECNDT 2006 - Th.1.2.2

1

serve as a temporary scaffold for bone remodelling and thus resorb rapidly while being replaced by new bone tissue. In order to determine in vivo the biodegradation characteristics of bone substitute materials animal studies are performed where the resulting specimens are typically visualised using light microscopy and immunohistochemical analysis, delivering two-dimensional information about the newly formed bone and the included ceramic particles with a contrast related to the chemistry of the different material phases [31]. In this article additional information about the specimens is collected with the help of three-dimensional synchrotron-microtomography. By using monochromatic synchrotron-radiation here the resulting volume images allow to distinguish different material phases via the given density contrast [13, 14]. Algorithms originating from stochastic geometry are then used to separate voxels in the images which belong to one material phase [15, 16]. These separated images are the basis for high quality three-dimensional visualisations [17] as well as a quantitative analysis [25].

2. Methods

2.1 Synchrotron-Microtomography

The Bundesanstalt für Materialforschung und –prüfung (Federal Institute for Materials Research and Testing, Germany) runs in cooperation with the Hahn-Meitner-Institut Berlin (Department Materials Research – SF3, Germany) a synchrotron-tomography setup at Berlin’s storage ring BESSY II (Berlin electron storage ring company for synchrotron radiation, Germany) at the so-called BAMline [18, 19]. Here the synchrotron-radiation originating from a superconducting 7T wavelength shifter is monochromatised with a double multilayer monochromator, leading to a energy bandwidth of around 1% and a photon flux density of up to 5 x 1010 s-1mm-2: suitable to obtain low-noise images with exposure times in the range of seconds. Besides the high precision sample manipulator the important device for the microtomography is a detector-microscope designed to obtain highly resolved radiographic images, see fig. 1.

Figure 1: Left - tomographic setup at the BAMline, right: sketch of the x-ray detector.

2

The design of this detector follows mostly the concepts of U. Bonse and F. Busch [13]. For the experiments a nominal 10 µm thick Gadox (Gd2O2S) scintillating powder screen was used because of its high stopping power and excellent photonyield. For the microscope objective a Rodenstock “XR-Heliflex” (focal length f=100 mm, NA=0.15) was chosen in combination with a Nikon “Nikkor 180/2.8 ED” (f=180 mm) objective, resulting in a magnification factor of 1.8. In the parallel section of the optical path a diaphragm is placed to adjust the ratio of absorbed X-ray photons (Gadox) vs. detected luminescence photons (CCD) and to reduce the depth of focus. The optics image the luminescent screen onto the back-illuminated CCD chip of a Princeton Instruments VersArray: 2048B (14bit dynamics @ 1 MHz readout speed, 2048 x 2048 pixels each sized 13.5 µm). The resulting effective pixelsize is 7 µm which leads to a 14 x 14 mm2 field of view, large enough to investigate the usually 25 x 12 x 5 mm3 sized specimen. In the radiographic projection the resolution was estimated to about 20 µm with the help of a copper edge (10% of the modulation transfer function [20]). For the reconstruction of the tomographic images the filtered backprojection algorithm is used [21] via the ESRF software package PyHST [22].

For new bone substitute materials (GB14, GB9, and GB9/25 [12]) an animal study is performed in which these novel rapidly resorbable ceramics are implanted in the sheep mandible [26]. The aim of this study is to compare these materials with tricalcium phosphate and 352i [27, 28]. The specimens taken from sheep jawbones 12 and 24 weeks after implantation were embedded in resin and then imaged at the BAMline, two tomographic volume scans per sample which are combined into one volume after reconstruction. Afterwards tissue sections were cut longitudinally using a sawing microtome and prepared for the immunohistochemical analysis [31]. The histological images shown later in this article were then taken with a light microscope. Important for the interpretation is that the contrast in absorption tomography is related to the density of the material while the contrast in histological images depends on the local chemistry of the specimen.

Figure 2: Left – tomographic slice of sheep jawbone with ceramic particles (352i – light grey, bone in grey, the remaining area is the resin with some porosity) three months after implantation, right: segmented ceramic

particles – morpholigical information combined with density information.

1 mm

3

2.2 3D Image Analysis

SCT has already proven to be a suitable method for the investigation of biocomaptible materials [23]. The key question of this subparagraph is how to derive quantitative results and high quality visualisations with respect to the different material phases contrasted in the grey-scaled 3D images. A typical slice of a tomographic volume acquired is pictured in fig. 2 (left) – ceramic particles can be recognized in light grey, the bone is grey and the remaining information is the resin with a couple of pores. Ceramic particles are separated first due to their high contrast and well-defined morphology into Boolean images (all voxels belonging to identified ceramic image sections are carrying the value one, the remaining voxels are set to zero) – fig. 2 (right). Here, a region growing method is applied in combination with a threshold hysteresis [16]. The optimal parameters for detection are tuned manually, remaining noise is removed with the help of closure and open procedures [15]. These Boolean images can be used to determine the size of the particles: with the help of a pre-defined connectivity [24], contiguous voxel groups are interpreted as individual ceramic particles, plain voxel counting then delivers their volume. Additionally, the Boolean data can be used to increase the visualisation quality: a masking of the original grey-scale volume with the Boolean image delivers 3D data sets which contain a combination of the morphological information and the density information. By masking the original input data with the inverted Boolean image of the ceramic particles the latter are deleted in the data sets, making it more easy to separate the bone information into Boolean images as well. Both data sets can be rendered simultaneously using different color palettes with Volume Graphics VGStudioMax™ [17], leading to extraordinary good visualisations: see figure 3. Here, the structural changes of the newly formed bone can be investigated qualitatively by the physician in order to plan surgical actions.

Figure 3: Volume rendering of a segmented data set, showing ceramic particles (352i, white) and detected

bone (red), compare with figure 2. Newly formed bone can be identified between the particles.

1 mm

4

3. Results

This paragraph focuses on two biocompatible materials: 352i and GB9/25. The aim is to demonstrate the strength of our method via the analysis of selected specimen while from the statistical point of view the number of samples investigated does not allow further conclusions. So here the basic results in agreement with statistical relevant analysis [26] are extended by the three-dimensional aspect and density contrast of synchrotron-microtomography.

3.1 GB9/25

Two specimen were imaged: one 12 and one prepared 24 weeks after implantation of GB9/25 particles in a given defect within a sheep’s jawbone, the corresponding volume rendering can be seen in figure 4. The left rendering is a combination of the Boolean ceramic particle information (blue) and the surrounding bony tissue. Parts of the bony tissue are set transparent in order to give insights into the bony volume where some of the particles are located. One can recognize that the other particles are located outside the tissue, mainly within the cavity in the middle of the specimen. Nevertheless, the majority of the particles are already decomposed. 24 weeks after implantation no more particles can be detected in the tomographic volume images – figure 4 (right). So the ceramics dissolve leaving no residual particles while the new bone is formed.!

Figure 4: GB9/25 – left 12 weeks after implantation, ceramic particles (blue)

and surrounding bone, right: 24 weeks after implantation, no more ceramics left.

In order to validate our results a slice of the 24 weeks specimen was cut and prepared via the immunohistochemical approach. The colored slice was imaged using a light microscope, see figure 5 (left). Here, the bone can be recognized by the grey-blue color, the remaining part of the image is the resin used for fixing the sample. The volume data set of the 24 weeks sample was compared manually slice by slice with that histological image until the corresponding section was found – see figure 5 (right). No major difference between the detected bone structure in the tomographic image and the histological image can be seen. Due to the density contrast of the SCT one can say that here the bone is fully evolved as all bony tissue detected in the chemical contrasted histological image is visible in the tomographic slice as well.

1 mm

1 mm

5

Figure 5: GB9/25 24 weeks after implantation – left: histological prepared slice (bone in grey-blue, rest resin), right: corresponding tomographic slice.

3.2 352i

The opposite behaviour of a bone substitute material is shown by 352i. The sample investigated 12 weeks after implantation is displayed in fig. 3 (and fig. 6 – right). Compared to GB9/25 (see fig. 4 – left) nearly all the ceramic volume still remains while only a little amount of newly formed bone can be detected between the 352i particles. So here the particles do not disappear and the residual probably has a strong influence on the formation of new bone. Again, to validate and also to extend the results derived from SCT a histological slice was prepared, see fig. 6 (left).

Figure 6: Histological image of bone substitute particles (352i, black) in sheep sinus (bone in pink) – corresponding to the sample in fig. 3 and volume rendering of a comparable section from the 3D data set (right).

1 mm

1 mm

6

For the comparison the segmented volume image of bone and ceramic particles was cut and orientated so that the slice corresponding to the histological image is visible – fig. 6 (right). Obviously the amount of detected bone differs strongly between the two images. While taking a closer look one recognizes that the bone volume in the lower right of the images is more or less identical for the tomography as well as histology – this is bone tissue which was already there before the implantation (the surface of the defect). Nearly all bone tissue which is visible in the histology but not detected by the tomography is located between the 352i particles – newly formed bone tissue. In the early stages of the bone formation the tissue is rather in a state of low mineralisation than hard tissue, so the material is less dense than fully evolved bone [29, 30]. Therefore it can not be seen in tomography image due to the density contrast of this method while it can be detected in the histological slices – the density is to low. So by comparing GB9/25 with 352i with the help of tomographic and histological images we can see first a large difference in the velocity of the biodegradation between the two materials. For the fast decomposing GB9/25 fully evolved bone can be proven after 24 weeks while for the slow decomposing 352i only bone tissue with a low degree of mineralisation after 12 weeks is detectable. Therefore the conclusion can be drawn that the decomposition velocity of the bone substitute material has a significant influence in the velocity and quality of the formation of new bone tissue in the defects.

4. Discussion

In this article we presented a novel approach to evaluate histological images by means of synchrotron-microtomography. Histological scans only deliver a two-dimensional information, always bearing the risk that the slice chosen is not representative for the whole volume of the specimen. SCT can be used to gather the required volume information in order to prove the drawn conclusion right or wrong. The beamtime at storage rings like BESSY, the ESRF or ANKA is limited and expensive. Therefore SCT is not the method of choice to scan hundreds of samples for a statistical verification, it is only suitable to investigate a limited number of representative samples in order to verify results. But SCT is not only a verification tool, in the previous paragraph we have shown that a comparison between histological and tomographical slices can deliver important additional information about the stage of the development (degree of mineralisation) of the newly formed bone.

5. Acknowledgments

This work was funded by the German Research Foundation (DFG Grant KN 377/3-1). Special thanks to Heinrich Riesemeier for the experimental support and Diane Eichert for discussions.

7

References

[1] U.C. Belser, R. Mericske-Stern, J.P. Bernard, T.D. Taylor. Prosthetic management of the partially dentate patient with fixed implant restorations. Clin. Oral Implants Res. 2000, 11, Suppl 1, 126-145.

[2] M.M.Bornstein, A. Lussi, B. Schmid, U.C. Belser, D. Buser. Early loading of nonsubmerged titanium implants with a sandblasted and acid-etched (SLA) surface: 3-year results of a prospective study in partially edentulous patients. Int. J. Oral. Maxillofac Implants 2003, 18, 659-666.

[3] S. Winkler. Implant site development and alveolar bone resorption patterns. J. Oral Implantol. 2002, 28, 226-229.

[4] S.D. Ganz, M. Valen. Predictable synthetic bone grafting procedures for implant reconstruction: part two. J. Oral Implantol. 2002, 28, 178-183.

[5] D. Buser, K. Dula, H.P. Hirt, H. Berthold. Localized ridge augmentation using guided bone regeneration. In: D, Buser, C. Dahlin, R.K. Schenk (editors). Guided bone regeneration in implant dentistry. Chicago, USA; Quintessenz (1994), 189-233.

[6] D. Buser, K. Dula, H.P. Hirt, R. Schenk. Lateral ridge augmentation using autografts and barrier membranes: a clinical study with 40 partially edentulous patients. J. Oral Maxillofa. Surg. 1996, 54, 420-432.

[7] T. von Arx, D.L. Cochran, R.K. Schenk, D. Buser. Evaluation of a prototype trilayer membrane (PTLM) for lateral ridge augmentation: an experimental study in the canine mandible. Int. J. Oral Maxillofac. Surg. 2002, 31, 190-199.

[8] D. Buser, K. Dula, D. Hess, H.P. Hirt, U.C. Belser. Localized ridge augmentation with autografts and barrier membranes. Periodontol 2000 1999, 19, 151-163.

[9] D. Buser, B. Hoffmann, J.P. Bernard, A. Lussi, D. Mettler, R.K. Schenk. Evaluation of filling materials in membrane-protected bone defects. A comparative histomorphometric study in the mandible of miniature pigs. Clin. Oral Impl. Res. 1998, 9, 137-150.

[10] P. Ducheyne. Bioactive calcium phosphate ceramics and glasses. In: L. Sedel, M. Cabanela (editors). Hip surgery: New Materials and Developments. London, UK ; Martin Dunitz (1998), 75-82.

[11] J.O. Hollinger, J. Brekke, E. Gruskin, D. Lee. Role of bone substitutes. Clin. Orthop. 1996, 324, 55-65. [12] G. Berger, R. Gildenhaar, U. Ploska. Rapid resorbable materials based on a new phase: Ca2KNa(PO4)2.

in: J. Wilson, L.L. Hench, D.C. Greenspan (editors). Bioceramics 8. Oxford, UK; Butterworth-Heinemann (1995), 453-456.

[13] U. Bonse, F. Busch. X-ray computed microtomography (µCT) using synchrotron radiation (SR). Prog. Biophys. Molec. Biol. 65, 133-169 (1996).

[14] L. Grodzins. Critical absorption tomography of small samples. Proposed applications of synchrotron radiation to computerized tomography II. Nucl. Instr. & Meth. Phys. Res. 206, issue 3, 547-552 (1983).

[15] J. Ohser, F. Mücklich, Statistical Analysis of Microstructures in Materials Science, Wiley & Sons (2000). [16] G. Lohmann, Volumetric Image Analysis, Wiley-Teubner (1998). [17] C. Reinhart, C. Poliwoda, T. Günther, W. Roemer, S. Maass, C. Gosch. Modern voxel based data and

geometry analysis software tools for industrial CT. Proc. 16th World Conference on Nondestructive Testing (WCNDT2004), Montreal, Canada (2004).

[18] W. Görner, M.P. Hentschel, B.R. Müller, H. Riesemeier, M. Krumrey, G. Ulm, W. Diete, U. Klein, R. Frahm. BAMline: the first hard X-ray beamline at BESSY II. Nucl. Instr. & Meth. Phys. Res. A 467-468, 703-706 (2001).

[19] G. Weidemann, J. Goebbels, Th. Wolk and H. Riesemeier. First Computed Tomography Experiments at BAMline. BESSY Annual Report 2001, 249-250 (2001).

[20] H.H. Hopkins. The frequency response of a defocused optical system. Proc. R. Soc. London, Ser. A 231, 91-103 (1955).

[21] A.C. Kak, M. Slaney. Principles of Computerized Tomographic Imaging. IEEE Press, New York, 1988. [22] A. Mirone, http://www.esrf.fr/UsersAndScience/Experiments/TBS/SciSoft/ (last visit may

2006). [23] B. Müller, P. Thurner, F. Beckmann, T. Weitkamp, C. Rau, R. Bernhardt, E. Karamuk, L. Eckert, S.

Buchloh, E. Wintermantel, D. Scharnweber and H. Worch. Nondestructive three-dimensional evaluation of biocompatible materials by microtomography using synchrotron radiation. Proc. SPIE 4503, 178-188 (2002).

[24] J. Ohser, B. Steinbach and C. Lang. Efficient texture analysis of binary images. J. Microscopy 192, 20-28 (1998).

8

[25] Modular Algorithms for Volume Images – MAVI (Fraunhofer ITWM) http://www.itwm.fraunhofer.de/mab/projects/MAVI/index_en.php (last visit May 2006).

[26] R. Gildenhaar, E. Lehmann, G. Berger, M. Stiller, Ch. Koch, P. Ducheyne A. Rack, C. Knabe. A comparative study of the Biodegradability of Calcium-Alkali-Orthophosphate Ceramics in vitro and in vivo. 19th International Symposium on Ceramics in Medicine 2006 – Bioceramics 19.

[27] G. Cordioli, C. Mazzocco, E. Schepers, E. Brugnolo, Z. Majzoub. Maxillary sinus floor augmentation using bioactive glass granules and autogenous bone with simultaneous implant placement. Clinical and histological findings. Clin. Oral Implants Res. 12, 270-278 (2001).

[28] J. Wiltfang, H.A. Merten, K.A. Schlegel, S. Schultze-Mosgau, F.R. Kloss, S. Rupprecht, P. Kessler. Degradation characteristics of alpha and beta tri-calcium-phosphate (TCP) in minipigs. J. Biomed. Mater Res. 63, 115-121 (2002).

[29] J.E. Aubin. Osteogenic Cell Differentiation. in Bone Engineering, p. 19-30, edited by J.E. Davies, em squared inc. Publisher, Canada (2002).

[30] J. Sodek and S. Cheifetz. Molecular Regulation of Osteogenis. in Bone Engineering, p. 31-43, edited by J.E. Davies, em squared inc. Publisher, Canada (2002).

[31] C. Knabe, Ch. Koch, A. Rack, M. Stiller. Histological and Immunohistochemical Study of Biopsies Samples after Sinus Floor Augmentation Using Tricalcium Phopshate Particles with Varying Porosity. 31st Annual Meeting of the Society for Biomaterials U.S.A., 2006, Transactions p. 371.

9