Abstract
Objectives
Tubules dominate the microstructure of dentin, and in crowns of human teeth they are surrounded by thick mineralized peritubular cuffs of high stiffness. Here we examine the three-dimensional (3D) arrangement of tubules in relation to enamel on the buccal and lingual aspects of intact premolars and molars. Specifically we investigate the angular orientation of tubules relative to the plane of the junction of dentin with enamel (DEJ) by means of wet, non-destructive and high-resolution phase-contrast (coherent) tomography.
Methods
Enamel capped dentin samples ( n = 16), cut from the buccal and lingual surfaces of upper and lower premolar and molar teeth, were imaged in water by high-resolution synchrotron-based phase-contrast X-ray radiography. Reconstructed 3D virtual images were co-aligned with respect to the DEJ plane. The average tubule orientation was determined at increasing distances from the DEJ, based on integrated projections onto orthogonal virtual planes. The angle and curl of the tubules were determined every 100 μm to a depth of 1.4 mm beneath the DEJ.
Results
Most tubules do not extend at right angles from the DEJ. Even when they do, tubules always change their orientations substantially within the first half-millimeter zone beneath the DEJ, both on the buccal and lingual aspects of premolar and molar teeth. Tubules also tend to curl and twist within this zone. Student t -tests indicate that lower teeth seem to have greater tilts in the tubule orientations relative to the DEJ normal with an average angle of 42° (±2.0°), whereas upper teeth exhibit a smaller change of orientation, with an average of 32° (±2.1°).
Significance
Tubules are a central characteristic of dentin, with important implications on how it is arranged and what the properties are. Knowing about the path that tubules follow is important for various reasons, ranging form improving control over restorative procedures to understanding or simulating the mechanical properties of teeth. At increasing depths of dentin beneath enamel, tubules are significantly tilted relative to the DEJ norm, which may be important to understand clinical challenges such as sensitivity, effectiveness of bonding techniques or prediction of possible paths for bacterial invasion. Our data show dissimilar average tubule angles of upper versus lower teeth with respect to the DEJ which presumably contributes to different shear responses of the tissue under function. The degree to which this may warrant improved restoratives or new adhesive techniques to enhance adhesive restorations merits further investigation.
1
Introduction
Dentin in human teeth has been extensively researched for more than 150 years and in recent decades, details of the microstructure have been augmented by extensive studies of its material characteristics. As a result, both elastic and plastic/failure attributes of this biological material are rather well understood in relation to the tissue organization (see overview in ). And yet, open questions remain as to what in the microstructure contributes most to the tissue properties, and in particular how the complex 3D design is able to withstand the trying conditions of the mouth for extensive periods of time. This is remarkable, because dentin is a variant of the mineralized collagen based tissues that does not remodel or self-fix, such that biological turnover is relatively insignificant. And in light of the fact that man-made restorations do not perform nearly as well as dentin in vivo (despite having excellent properties), it is prudent to assume that there are details of the 3D organization of features in dentin that we still do not fully understand. Dentin attributes are routinely considered with respect to tubules, and in this study, we report on the orientations of tubules relative to the DEJ plane on the buccal and lingual aspects of human premolar and molar teeth.
As a member of the bone family of materials and at the lowest organizational level dentin is composed of mineral, organic macromolecules (mostly collagen) and water. The mineral, a calcium phosphate salt (dahllite), accounts for about 65% of the weight. Similar to bone, the mineral appears in the form of small crystals of carbonated hydroxyl-apatite, sized ∼36 nm × 25 nm × 4 nm . The mineral embeds the organic macromolecules that account for about 20–25% of the weight and appear mostly as a felt of thin (∼100–200 nm thick) highly cross-linked type I collagen fibers (about 90%) . Other organic components include non-collagenous phosphorylated/non-phosphorylated proteins, small amounts of proteoglycans, neutral and acidic mucopolysaccharides and lipids.
At a higher organizational level, dentin is a porous material, densely perforated by more-or-less parallel tubules that lead inwards towards the pulp chamber. Despite being a thick sensitive biological tissue, dentin is a-cellular. The tubules however, often contain protrusions or extensions of living cells—the odontoblasts . These cells are responsible for excreting tissue products that assemble into dentin while mineralizing along a front surrounding the cellular protrusions and continuously moving inwards towards the pulp and down the roots. After tooth formation is complete, the very same odontoblasts continue to produce dentin at a slow but steady pace within the pulp chamber and root canals, forming a lower quality and less organized tissue. Dentin in vital teeth is thus in fact part of a dentin–pulp complex that is both sensitive and reactive to the tooth environment.
The tissue created by the odontoblasts is in effect a complex three-dimensional arrangement of several forms of dentin, most notably pertitubular dentin (PTD) and intertubular dentin: While the former (PTD) is known to be highly mineralized and contains no collagen, the latter is made of mineralized collagen fibers that presumably lie in incremental layers approximately parallel to the pulp wall, at right angles to the tubules. PTD lines many of the tubules ( Fig. 1 ), mainly in the crown and is formed at the time when the bulk of dentin is being laid down – during tooth formation . PTD differs from the mineral precipitates that may collect in tubules – particularly in older dentin, where tubules become occluded or sclerotic over time . It creates dense collars with internal diameters of about one micrometer for most of the tubule length. The PTD surrounded tubules extend up to several millimeters along their long axis as they follow an S shape path inwards towards the pulp. It is deep in the bulk of the crown that PTD is thickest (>1 μm). Near the dentino–enamel junction (DEJ) in the crown (or the cemento–dentinal junction in the root) in the outer regions of dentin, close to where tubules branch and originate (with diameters of less than 0.5 μm ) PTD is thin or missing. On the other end in deep dentin near the pulp, tubules become tightly packed as they converge together, with their internal diameters exceeding 2 μm .
The uniform and highly aligned distribution of tubules suggests that dentin should be a very anisotropic tissue, deforming differently when loaded along versus across tubule orientations. But curiously, little or negligible elastic anisotropy has been found . Dozens of experiments performed over many years and for a varying range of sample sizes and measurement techniques have shown that dentin exhibits only hints of anisotropy. And yet nanoindentation has revealed large differences in the stiffness of PTD as compared with intertubular dentin—the former amounting to 30 GPa while the latter measuring only 15–20 GPa . With such a ratio one might expect to find a lower modulus when loading dentin samples orthogonal to the tubule direction (effectively densifying the tubules that are embedded in less-stiff intertubular dentin), as compared with loading along the tubule orientations. This however is not seen, and Kinney et al. developed a micromechanics model showing that despite increasing thickness and density of the PTD, the tubular structures on average do not reinforce dentin and hence they do not impose mechanical anisotropy: dentin properties are thus governed by the intertubular dentin. Few studies however, such as the one by Palamara et al. , reported small differences in moduli along versus across the main tubule orientations when measured in compression ( E = 10.7 and 11.9 GPa, respectively). Using resonant ultrasound spectroscopy, Kinney et al. concluded that an orthogonal arrangement of collagen fibrils with respect to tubules renders human dentin approximately 10% anisotropic . They suggested that dentin has a Young’s modulus of 23.2 GPa along the tubules versus 25.0 GPa across the tubules. Arola and Reprogel used bending experiments to provide additional evidence to the existence of some elastic anisotropy related to tubule orientation: they reported flexural moduli of 15.5 GPa when bending bar samples where tubules run along the long bar axis, versus 18.7 GPa when bending samples with tubules orthogonal to the long bar direction. Reduced stiffness was found in the near-DEJ zone where collagen fibers run orthogonal to enamel and parallel to the tubules. In all these studies however, tubule orientations were determined or inferred from observations on the edges of the samples, but no direct link between tubule orientations and elastic properties was shown.
Unlike elastic properties, anisotropy with respect to tubule orientation has been rather clearly demonstrated for failure properties of human dentin. Rasmussen et al. demonstrated that the work to fracture across tubules was about half that of the work to fracture along the tubule orientation. A dependence of shear strength on dentin site and tubule orientation within the crown was clearly demonstrated by Watanabe et al. and tensile strength anisotropy was reported by Carvalho et al. and other workers. Arola and Reprogel demonstrated significantly reduced strains at fracture (about 0.005 m/m) along tubule orientations compared with higher strain values across tubule orientation (0.015 m/m) and reported increased flexural strength across the tubules (∼160 MPa)- 30% higher than the strength along the tubules. They also noted a decay of strength with increasing tubule density nearing the pulp. These and other studies of fracture toughness and fatigue have demonstrated that failure in dentin occurs much more easily across tubules than along their trajectory (see for example ), possibly due to increased brittleness when tubules are pulled or compressed along the long axis.
The plastic properties of dentin suggest that different tubule orientations in human dentin may lead to significant variations in the mechanical properties. These orientations may also have an indirect effect on the stiffness. Tubules beneath enamel are thought to extend orthogonal to the DEJ ‘plane’ and they then follow a somewhat winding S shaped path deep into the bulk . We propose that a better understanding of dentin as a structure and specifically its mechanical properties may arise from knowing more about the 3D course that the tubules occupy when passing through the tissue. In particular, large – mm sized test samples – may have unknown tubule orientations. Recent methods of non-destructive high-resolution and partially coherent X-ray imaging allow exploring tubule orientation variations deep in the dentin bulk. One excellent method of achieving this is by microtomography.
High-resolution X-ray microradiography and microtomography have become important investigation tools, useful for imaging minute details spanning less than 10 μm within small representative (millimeter sized) samples. Absorption is the primary interaction of interest, because attenuation of the X-ray signal is linked to important sample attributes: geometry, density and composition. Briefly, for each of many projection images taken from equally spaced angles around the sample (at least 180° for parallel beam geometry, usually obtained by rotating the sample) it is possible to obtain the ratio between sample attenuated intensities and non-attenuated intensities in images where the sample is absent. For tomography these ratios are used for numerical backprojection, so as to create 3D images of the volume which represent the sample structure. Additional details about tomography and methods of reconstruction may be found in textbooks (for example see ) as well as in many online references.
Inherent to high-resolution X-ray absorption imaging is the problem of low signal to noise ratios. Variations in the source intensity and stability, detector response limitations and faults, system vibrations and thermal fluctuations all contribute to lack of precision and a reduction in the clarity of features in each projected radiograph. This results in deteriorated 3D reconstructions and the emergence of artifacts. Furthermore, absorption imaging has limited use when we consider that most biomaterials are made of light elements that induce little or negligible attenuation contrast between adjacent materials or structures. The light elements are almost transparent to X-rays and they contribute significantly more to scattering than to absorption, which reduces the signal to noise ratio dramatically. The outcome of all this is that the small structural details that exist in dentin bring about similar and indistinguishable attenuation effects in most absorption radiographs, and consequently standard or ‘regular’ microtomography is of little use for the study of the thin dentinal tubules and their 3D orientations in dentin.
High energy hard X-ray sources such as those found in 3rd generation synchrotron (electron acceleration) facilities offer extremely high brilliance at a range of energies that is relevant for the study of mineralized tissues (20–40 keV) . Large propagation distances (tens of meters) between the X-ray source and detectors in these huge machines entail some laser-like attributes to the X-rays: they become partially coherent, such that constant phase relations are found over meaningful distances compared to the sample dimensions . This leads to the appearance of interference patterns/fringes in radiographs so that either constructive (brighter) or destructive (darker) spots/lines are seen. Such fringes arise due to the difference in interaction of the radiation with matter found on opposing sides of edges, voids and discontinuities. This difference results in an overlapping of mutually coherent X-ray fields on the detector, positioned far behind the sample, and this difference is a consequence of the beam passing through different path densities within the sample microstructure. Note that even between sample features composed of light elements and even if these features are of micron or sub-micron dimensions, significant path density differences arise, seen as edge enhancements and inhomogeneity accentuation in the radiographs.
In this paper we report on measurements and findings of tubule orientation variations that are seen in a zone of 1.4 mm beneath the DEJ on the buccal and lingual sides of molar and premolar teeth, using phase-contrast tomography. Phase-contrast imaging of this type is not a new concept: the ability to project and produce slightly enlarged phase-enhanced silhouettes of edges in objects was originally used for inline holography and electron microscopy, pioneered by Gabor and others in the late 1940s. However, with current X-ray instrumentation and when coupled with tomographic reconstruction, the technique brings about contrast enhancement even between very small or chemically similar material phases. As a result, it is possible to visualize the 3D structure and distribution of features sized much less than a micrometer so as to reveal sub-micron details . In particular dentin tubules, even when immersed in water, may be clearly visualized, as long as destructive and constructive interference patterns do not cancel each other out .
2
Materials and methods
2.1
Tooth sample preparation
Samples were cut from the mid-buccal and mid-lingual surfaces of 8 intact permanent teeth, extracted from young people in private clinics in the greater Berlin area (Germany) in accordance with conditions set by the ethical review board of the Berlin Charité dental school. Four molars (2 upper and 2 lower wisdom teeth) extracted from young adults and 4 premolars (2 upper and 2 lower) extracted from teenagers during orthodontic treatment were used. Samples approximately 2 mm × 2 mm × 3 mm were manually cut by standard dental high-speed rotatory instruments so as to include intact enamel on one side while the sample bulk was made of dentin, extending deep towards the pulp on the other side ( Fig. 2 ). Each cylinder-like sample was cut such that the DEJ was oriented approximately orthogonal to the sample length, spanned the sample width. With enamel clearly visible on one side, the samples were mounted upright in small Plexiglas vials with the dentin edge embedded in a thin layer of hand-mixed dental acrylic (BosworthTrim, The Bosworth Company, Durham, England) where they were kept in water. Samples were stored at 4 °C until imaged as described in the following section. Note that upper palatal samples are termed lingual for convenience throughout this paper.
2.2
Phase-contrast X-ray imaging of wet samples
The 16 samples (buccal + lingual specimen from each tooth) were mounted upright in the microtomography setup on the BAMline imaging setup of the Berlin electron storage ring company synchrotron facility (BESSY ) and 900 or 1200 radiographs (depending on sample diameter) were recorded at angular increments of 0.2° or 0.15° using an energy of 28 keV with a sample-to-detector distance of 430 mm . The projection images were normalized and reconstructed using Octopus V8.1 (XrayLAB, Ghent University, Belgium ) and then visualized, rotated and cropped (Amira 4.1, Visage Imaging GmbH, Germany) so as to align the DEJ perpendicular to the Z direction of a parallelepiped of dentin, capped with enamel ( Fig. 3 ).
2.3
Tubule orientation visualization and analysis
As discussed in Zabler et al. , tubules surrounded by either thin or thick PTD layers produce inverse (constructive or destructive) interference rings, a consequence of the ratio of tubule lumen thickness to PTD thickness. Often these interference patterns cancel out, revealing little or no detail ( Fig. 3 , right hand side). In any event, when tubule traces are seen, they appear as either bright silhouettes surrounding dark centers or as bright centers surrounded by dark or gray rims. Both forms are useful in order to estimate the average tubule orientation at various depths beneath the DEJ. To achieve this, consecutive subvolumes of dentin sized 400 μm × 400 μm × 100 μm were extracted from each 3D volume image at 100 μm intervals beneath the DEJ, such that the shorter side of each subvolume corresponded to the Z depth beneath the DEJ ( Fig. 4 ). The intensities within these subvolumes were summed and projected onto the two orthogonal parallelepiped sides of each subvolume (ImageJ stack projection function, ImageJ V1.38 , see Fig. 4 , left and right) resulting in 2 orthogonal and clear projections of the dominant tubule orientations at each depth beneath the DEJ. The average rotation angles ( α , β ) were then determined in each pair of orthogonal projections for each sample. These angles are of course arbitrary because the scanned volumes were rotated randomly along the sample enamel–dentin ( Z ) axis. The determined angles α and β were converted into absolute tubule tilt and curl angles ( φ , ) defined with respect to the direction perpendicular to the DEJ using:
and
We thus obtained the orientation of angles relative to the Z -axis from which the tubule tilt angles are given with respect to the DEJ plane normal, and we could also derive information about changes in tubule rotation/twist relative to their original trajectory near the DEJ.
2
Materials and methods
2.1
Tooth sample preparation
Samples were cut from the mid-buccal and mid-lingual surfaces of 8 intact permanent teeth, extracted from young people in private clinics in the greater Berlin area (Germany) in accordance with conditions set by the ethical review board of the Berlin Charité dental school. Four molars (2 upper and 2 lower wisdom teeth) extracted from young adults and 4 premolars (2 upper and 2 lower) extracted from teenagers during orthodontic treatment were used. Samples approximately 2 mm × 2 mm × 3 mm were manually cut by standard dental high-speed rotatory instruments so as to include intact enamel on one side while the sample bulk was made of dentin, extending deep towards the pulp on the other side ( Fig. 2 ). Each cylinder-like sample was cut such that the DEJ was oriented approximately orthogonal to the sample length, spanned the sample width. With enamel clearly visible on one side, the samples were mounted upright in small Plexiglas vials with the dentin edge embedded in a thin layer of hand-mixed dental acrylic (BosworthTrim, The Bosworth Company, Durham, England) where they were kept in water. Samples were stored at 4 °C until imaged as described in the following section. Note that upper palatal samples are termed lingual for convenience throughout this paper.
