The concentric laminar structures found in secondary bone were first recognized by Havers in 1691 (thus, “Haversian” systems and canals). A mature Haversian system structure in bone is also called an “osteon.” The structure of Haversian systems was subsequently confirmed and extended by many subsequent studies. However, these lead to two contrasting models on how the collagen fibrils were arranged within the individual lamellae (Gebhardt, 1906; Ruth, 1947; Rouiller et al., 1952; Engstrom and Engfeld, 1953; Frank et al., 1955; Smith, 1960; Ascenzi and Bonucci, 1968; Boyde and Hordell, 1969; Reid, 1986; Giraud-Guille, 1988; Carando et al., 1989; Boyde and Riggs, 1990; Marotti, 1993; Weiner et al., 1997, 1999; Ascenzi et al., 2004; Pazzaglia et al., 2011). In the first model, the lamellar arrangement has been explained by separate, concentrical layers with a parallel orientation of the collagen fibrils within the layer, but forming a different angle with the vascular canal axis in the next lamella; whereas in the second by an alternating sequence of dense and loose fibrils aggregation.
It is generally not questioned, if all the lamellar systems of secondary osteons have always had in their development a regular concentric pattern, perhaps because their initial osteonal design has been partially hidden by ongoing bone remodeling. Thus, most attention has been focused on the orientation of the collagen fibrils, whereas other parameters of lamellar organization (like thickness, completeness, or patterns) have generated less interest. Marotti (1993) observed that no appreciable differences in cortical bone structure were related to the age of the subject or to the radius of curvature of the lamellae within the osteon. In some studies, the thickness of the lamellae was measured (Reid, 1986; Weiner et al., 1999) with values reported in a range from 1.8 to 5 μm. Weiner et al. (1997, 1999) examined vitrified sections cut approximately in the plane of the lamellar boundary distinguished in each lamella five layers and measured the plywood-like structure of each sublayer. These authors assigned different thicknesses to each sublayer, and this reflected what was observed in baboon tibiae (Liu et al., 1999). It was assumed that the lamellae in the osteon were regular with minor deviances from a standard thickness and that they generally developed on the whole circumference and concentrically.
By applying a short and controlled etching with Na3PO4 to transverse sections of human cortical bone prepared for scanning electron microscopy (SEM) observation, it was possible to improve the resolution of the interface between neighboring lamellae and thus, the precision of measurements (Congiu and Pazzaglia, 2011). Therefore, we could test using morphometric methods previous assumptions concerning the patterns and the geometry of lamellar systems in osteons. The application to the study of lamellar organization of different methods (from histology to electron microscopy and diffractometry) has expanded our understanding of the osteonal lamellar system. However, there are few studies that attempted to correlate SEM morphology with the images obtained by polarized light microscopy (Reid, 1986). These observations, however, were carried out on different osteons. In this article, we were able to examine the same osteon using both SEM and polarized microscopy. As the fibrillar geometry of the lamellae is the result of a complex process that develops within a tunnel dug by the cutting cone and implies not only collagen synthesis and extrusion by isolated osteoblasts but rather coordination of the activity of a pool of osteogenic cells (Pazzaglia et al., 2011), this morphometric approach to characterize lamellar geometry provided indications on the synchronization, cycles, and behaviors of the osteoblast pool in general.
MATERIALS AND METHODS
Four tibias were obtained from male patients between 25 and 52 years of age who underwent an amputation below the knee because of severe leg traumatic injuries. All patients gave consent that a segment of the amputated tibia would be used for scientific purposes, and the protocol was approved by the Ethics Committee of Brescia Spedali Civili.
Two segments of the tibia about 1 cm of thickness were obtained between the one-third distal and the two-third proximal portion of the bone. These cylinders of diaphysis were cleaned of soft tissues and fixed in neutral formaldehyde (10%) for a week. One of the samples was then placed in hydrogen peroxide (40%) at room temperature for 4 weeks to remove all remaining soft tissues without physical manipulation. Three parallelepipeds were cut from the annular segment using a low-speed bone saw with diamond-coated wafering blades (Remet, Casalecchio di Reno, Bologna, Italy), corresponding, respectively, to the antero-medial, lateral, and posterior sector of the tibia. Four slices of about 1 mm in thickness cut in a plane perpendicular to the long axis of the tibia were obtained from each of the blocks. The sections were ground to about 400 μm in thickness with an automated grinder (Remet, Casalecchio di Reno, Bologna, Italy) and further reduced by manual grinding to about 300 μm. The sections were then polished, washed repeatedly in distilled water, and further cleaned in an ultrasonic bath. To enhance the lamellar pattern of the osteons, the cut surface of the slices was exposed for 1 min to a 6% Na3PO4 solution (pH = 9.1) at room temperature. The sections were then dehydrated in ascending concentrations of ethanol and critical-point dried in CO2. The sections were secured on stubs with conducting tape, coated with a thin layer of gold in a vacuum sputter-coater (Emitech) and examined with a Philips XL 30 scanning electron microscope in back-scattered electron imaging (BEI) and direct modes.
The bone specimens of the second ring (one for each tibia) were ground to about 40 μm as described earlier. The sections were then stored in 2% formaldehyde until further processing. These slices were observed wet with an Olympus BX51 microscope under polarized light and 10 osteons from each slice that could be easily traced by their position on the map of the specimen were photographed. The same specimens were then subjected to Na3PO4 etching and prepared for SEM as described earlier. The same osteons that were identified and photographed under polarizing light were also observed by SEM in both the BEI and the direct modes.
SEM digital images of the transverse surfaces etched with Na3PO4 were acquired in the direct mode with the electron beam perpendicular to the plane of the section, at 20 kV and 10 mm of working distance. The same magnification of 350× was used for each of all examined sections. We selected for morphometry all the osteons satisfying the following conditions:
(a)a central canal perpendicular to the section or forming an angle =70 degrees; this angle was calculated measuring the short cathetus of the right triangle (central canal axis) with the perpendicular (long cathetus = section thickness) (Fig. 1);
(b)a complete lamellar pattern without marginal notching;
(c)a central canal perimeter <400 μm in diameter.
In total, 82 osteons were evaluated. The SEM digital images of the single osteons were processed using Cell (Soft Imaging System GmbH, Munster, Germany). The following parameters were measured:
major and minor perpendicular diameters of the osteon intersecting within the central canal (μm);
outer perimeter of the osteon (μm);
inner perimeter of the osteon or central canal perimeter (μm);
total osteonal area (μm2);
central canal area (μm2);
lamellar bone area, given by the difference between the total osteonal area and the central canal area (μm2).
Each half of the major and minor diameter was numbered clockwise a1, b1, a2, and b2. They have been numbered starting from the diameter whose direction was better approximated on the right-top corner of the microscopic field to the one intercepting the external osteon perimeter, the central canal perimeter, and the grooves corresponding to the lamellar interfaces. The number and thickness of the lamellae were consequently evaluated by the segments intercepted on each hemidiameter (Fig. 2). Starting from hemidiameter a1 and proceeding clockwise, the difference between the number of lamellae in the four hemidiameters sequence was calculated. The sum of the absolute values of these differences was assumed as an indication of the frequency of the regular circumferential lamellae versus those whose extension was limited to a sector of the circumference or formed a coil (lamellar completeness index).
The single lamellar thickness was measured on the segments intercepted by the lamellar interface grooves on the four hemidiameters. The osteon mean lamellar thickness was calculated dividing the sum of all the single thicknesses by the total number of lamellae intercepted on the four hemidiameters. Descriptive statistics was used to describe the mean lamellar thickness of the whole population of 82 osteons examined.
The distributions of the mean number of lamellae, mean lamellar density and mean lamellar thickness were examined for osteonal matrix area classes by Pearson chi square test (Tables 1–3). The selected 82 osteons were divided into four matrix area classes. The limits of each class have been selected in such a way to have about 20 osteons in each class (19 + 22 + 21 + 20 = 82).
Table 1. Distribution of the mean number of lamellae for osteonal matrix area classes (Pearson chi square test)
Table 2. Distribution of the mean lamellar density for osteonal matrix area classes (pearson chi square test)
Table 3. Distribution of the mean lamellar thickness for osteonal matrix area classes (Pearson chi square test)
The measured parameters were examined independently by two observers: interobserver and intraobserver precision in linear measurements and in the count of lamellae was expressed as the coefficient of variation of repeated measurements given on a percentage basis. The interobserver precision was 1.29% and 7.21%, respectively, for the two observers, and for repeated measurements and counting, the intraobserver precision was 1.08%–6.22% and 1.49%–8.08%, respectively.
Etching with Na3PO4 enhanced the lamellar pattern of the osteons and more clearly delineated the boundary lines between lamellae and the reversal line at the periphery. They were clearly evident in both the direct and the backscattered modes. However, the resolution of the canalicular radial pattern was better in the backscattered mode (Fig. 3A,B). At higher magnification, the reversal line of the osteon and the interlamellar lines appeared as superficial grooves devoid of or with a lower density of collagen fibrils. The reversal line at the osteon periphery had a similar appearance except that it was larger and irregular (Fig. 4A). The lamellar surface cut by the saw and milled by the grinding procedures exposed the tip of cut off collagen fibrils. They were densely packed together with no detectable differences of the packing density between the adjacent lamellae. The cut off fibrils appeared oriented with an angle approximately perpendicular to the cut plane, while those on the bottom of the grooves laid parallel to the section plane (Fig. 4B).
The lamellar systems showed a general concentrical design, however, a more detailed analysis illustrated that three main patterns could be distinguished (Fig. 5). The first was circular lamellae, which closed on themselves forming a ring. The second pattern was spiral lamellae overlapping and turning 360 degrees, forming a coil. The third pattern was lamellae whose extension were limited to an arc of the circumference creating a crescent-moon-shaped pattern.
The three patterns could all be present in the same osteon, however, the most frequently observed were the circular and the crescent-moon-shaped patterns (94.6% and 5.6%, respectively). The coil pattern was rare (0.2%) and was never observed to extend beyond 540 degrees (a complete turn and half).
The mean lamellar thickness in the population of 82 osteons was 9.0 ± 2.13 μm with a normal distribution of the values, but on the right side of the curve there were very thick lamellae corresponding to almost the whole matrix area of the osteon. The thickest lamella was 66.3 μm (Fig. 6), and the thinnest was 1.9 μm.
The mean completeness index of the osteons was 1.77 ± 1.17. The mean number of lamellae for osteon was 8.38 ± 2.15, and this parameter had a direct correlation with the osteonal bone area. No significant correlation was observed between mean lamellar thickness and matrix area classes (Table 3).
The comparison of the same osteon in SEM and polarized light was carried out selecting two or three rectangular boxes (the same in the SEM and polarized light images). The polarized light box was rotated 180 degrees anticlockwise around the long right side of the box (an osteon radius), to match the corresponding right side of the SEM box (Fig. 7A,B). The pairing of the grooves and lamellae (in SEM) with bright and dark bands (in polarized light) could not rely on the thickness of the bands, because the latter had no clear-cut outlines and their blurred borders were further conditioned by the focal plane of the acquired image in polarized light. However, considering the sequence of the bright bands (the better demarcated) on the box paired margin between selected reference points, it was possible to document a good match between the bright bands and the grooves of the SEM images (Fig. 7C,D,E).
Occasionally large bright bands composed by thinner, closely packed bands, could be observed and in this case they corresponded to the whole thickness of a lamella between two sequential grooves in SEM (Fig. 7C).
Most of the osteons examined in polarized light microscopy corresponded to the alternate type with the sequence of bright and dark bands, while extinct osteons, as defined by Ascenzi and Bonucci (1968), were not observed.
Any morphologic and morphometric study of Haversian lamellar bone is affected by the bias of some artifacts. These include the angle of the section plane relative to the central canal axis, the biological matrix damage by sawing and grinding/polishing, the chemical etching to clean the surface to be examined and dehydration and drying for the standard SEM preparation.
The projection error in measurement of surfaces, perimeters, and diameters were limited in selecting osteons whose central canal was perpendicular or with an angle = 70 degrees to the section plane. The other two limiting criteria are used to analyze complete osteons and to exclude cutting cones and early forming osteons.
Some type of etching must be applied to remove the debris and the mixture of organic/inorganic residue left on the surface by grinding and polishing. This was necessary to reveal the lamellar interface and improve the definition of the image. Acid etching has been used to prepare cortical bone specimens for SEM observation (Marotti, 1993). However, little attention has been paid to the effects that type, strength, and time of exposure of the etching agent can have on the morphology of the examined surface. With a short etching with Na3PO4, which produced a thin layer of decalcification on the surface and exposed the tip of the cut collagen fibrils, a clean and well-contrasted cut surface of the cortical bone was obtained. This is evident in that both the reversal and the interlamellar lines were enhanced as thin grooves (Congiu and Pazzaglia, 2011). These findings are consistent with the prior observations that the cement lines of secondary osteons are relatively hypermineralized or alternatively collagen deficient with respect to the surrounding bone (Skedros at al., 2005). On the other hand, using a weak acid etching in wet conditions, we observed that it was possible to produce, instead of the Na3PO4 grooves, crests, as a result of the shrinkage of the superficial matrix layer of the cut lamella (Congiu and Pazzaglia, 2011). It has been observed that etching with strong acid or basic solution, although for a short time, produced surface corrosion and coagulation of the collagen fibrils with a loss of quality and definition of the interlamellar lines (Marchese et al., 2011). The scratches left by grinding also produced grooves on the surface, but the latter were straight and without any relationship with the osteonal geometry. Therefore, they were easily distinguishable from the interlamellar lines.
As the first histological studies, the lamella has been recognized as the structural key-stone of the secondary osteon (Gebhardt, 1906; Ruth, 1947; Rouiller et al., 1952; Frank et al., 1955; Smith, 1960; Ascenzi and Bonucci, 1968; Reid, 1986; Giraud-Guille, 1988; Marotti, 1993). From the beginning, the attention has been focused on the question of how the collagen fibrils were oriented inside the lamella and how they differed among neighboring lamellae. Other aspects, like the regular concentrical pattern or the thickness, have generated less interest. In the very early reports based on light microscopy, the osteonal lamellar thickness was not measured. More recently, using SEM, Reid (1986) reported an osteonal and interstitial half lamellar separations in the adult from 2.06 to 3.09 μm with statistically significant differences in age of the subjects and between interstitial and osteonal lamellae. Marotti (1993) observed a thickness of 1.8–2.2 μm for dense lamellae and 3.3–4.2 μm for loose lamellae, and Weiner et al. (1999) observed a lamellar thickness between 3 and 5 μm. In this study, a mean lamellar thickness of 9.0 ± 2.13 μm was measured, and this is greater than previous determinations and with a wider range of variation. This was well illustrated by the very thick lamellae that could almost fill the whole matrix area of the osteon. The discrepancy of previous measurements with our data can be explained by the lower definition of the lamellae, because no enhancement of the interlamellar lines was carried out and the method of measurement did not assess the thickness variation of the lamella along the circumference.
We observed a direct correlation between the number of lamellae and the osteonal bone area. However, the lamellar thickness distribution in the population of examined osteons was normal and had a random distribution for osteonal size classes.
The organization of the secondary osteon is commonly depicted using a “plywood model” (Giraud-Guille, 1988; Raspanti et al., 1996; Weiner et al., 1997), which is composed of a sequence of units (the lamellae) whose collagen fibrils have a sufficiently definite parallel arrangement, but whose plane can rotate within the lamella. The alternate bright and dark bands in polarized light have been explained through the change of direction from one fibrils layer to the next one in alternate osteons (Ascenzi and Bonucci, 1968). However, an exclusively longitudinally oriented collagen with little evidence of division into lamellae has also been described. This structural layout has, therefore, been correlated to extinct osteons (Reid, 1986). This pattern was also confirmed by SEM studies of decalcified human osteons dissected and manipulated using the method suggested by Frasca et al. (1976, 1977).
The lamellar organization of the secondary osteons, to the best of our knowledge, has been assumed to be formed by complete, circular, and concentric lamellae, whereas this study describes spiral- and crescent-moon-shaped patterns. These particular aspects do not seem to fit the concept of osteon morphotype, which is based on polarized light microscopy image analysis observations that correlated the orientation of collagen fibrils in the lamella with the tension/compression load in the osteons of the long bones diaphysis (Martin et al., 1996; Bigley et al., 2006; Beraudi et al., 2010; Skedros et al., 2011). This study introduces the concept of a greater variability of the osteonal system geometry. The adopted morphometric methods suggest a more reliable calculation of the lamellar number, density, thickness, and patterns of the lamellar organization.
The panoramic view (at low microscopic enlargement) of the cortex osteons suggests a well-defined geometrical design with a circular, concentric pattern, however, when single osteons are observed at higher magnification the spiral- and the crescent-moon-shaped lamellae introduced a certain degree of irregularity. If it is assumed that the cellular activity of the osteoblasts in the osteon infilling is executed by a mediator mechanism bridging individual activity to the whole osteon morphology, the ordered fibrillar organization of secondary osteon lamellae can be more easily explained by the activity of a coordinated pool of osteoblasts.
We documented through SEM and graded osmic maceration technique that on the growing bone surface the osteoblasts formed a network of cytoplasmic processes where individual boundaries of cells were not distinguishable (Pazzaglia et al., 2010, 2011), supporting the hypothesis of a functional pool. Earlier studies done mostly with transmission electron microscopy (TEM) proposed that a continuous osteocytes network acted as a functional syncytium (Palumbo et al., 2004).
Both the circular- and the crescent-moon-shaped patterns can represent the common and regular pattern according to which lamellae are structured by the osteoblasts pool and the measured variation can be the expression of a “biological plasticity” adapting to the local conditions.
The prevailing circular pattern of the lamellae is executed by the coordinated activity of the osteoblasts pool; however, the regularity of the osteon infilling is also determined by the shape of the cutting cone prepared by the osteoclasts, which in some osteons it is perfectly circular but in most of them is oval or even more irregular. Therefore, the biological variations of the osteon infilling process contribute to the advancement of the lamellar system toward a rounded or slightly oval profile of the central canal in the completely structured osteon. The coil-types lamellae were rare and may be determined by particular conditions modulating the pool of osteoblasts.
In the study of Robling and Stout (1999), the authors reported drifting osteon and explained the lateral shift of the lamellae with strain gradients acting perpendicularly to the axis of the osteonal canal. However, this situation may not be applicable to the crescent-moon-shaped lamellae. This was because they appeared intercalated with circular lamellae or distributed randomly along the circumference and do not fit with the definition of a drifting osteon (an Haversian system in which there is a continuous resorption on one side and continuous formation on the other).
In all studies (with the exception of those utilizing diffractographic methods), the delimitation of the osteon and the lamellae is based on appreciable zonal changes of the intercellular bone matrix (the grooves) named in a different way, like reversal lines, resting lines, interlamellar lines, and hypermineralized lines (Schaffler et al., 1987; Burr et al., 1988; Bain et al., 1990; Nyssen-Behets et al., 1994; Nanci, 1999). As Na3PO4 etching produces on the exposed bone a thin superficial layer of decalcification, the appearance of the grooves is consistent with the explanation given by Skedros et al. (2005) that the lines have a higher hydroxyapatite concentration, because collagen fibrils density is lower.
The ordered layout of the fibrils in the lamellae could be explained by the synchronous activity of a pool of osteoblasts (Pazzaglia et al, 2010, 2011). Assuming that this activity is phasic, the grooves (or lines) would correspond to an arrest of the matrix deposition by the synchronized cells of the pool. According to this interpretation, all the morphometric variations observed could be explained in terms of length of time of the appositional phase (the thickness of the circular lamellae), extension of the pool recruitment (the crescent-moon-shaped lamellae) and staggered activation of the osteoblasts pool in the coil lamellae when occurring (Fig. 8). The resting line evidenced by SEM, whatever enhancement technique is used, should represent the synchronous arrest of apposition by the osteoblasts of the row (pool). It seems reasonable to hypothesize that, if a significant change of the fibrils orientation must occur, it should take place when the pool of osteoblasts starts a new phase of apposition (the next lamella).
If, in the same osteons, the sequence of lamellae viewed through SEM was matched against the bands in polarized light, we could observe a sufficiently good correspondence between the bright bands and the grooves where the fibrils lay parallel to the plane of the section and the dark bands with the lamella between two grooves where the fibrils form an angle with the plane of the section. These findings in part validated the classic representation of the fibrils organization given by Ascenzi and Bonucci (1968) and most recently reconsidered by the studies analyzing the cortical osteons morphotypes (Boyde and Riggs, 1990; Martin et al., 1996; Beckstrom et al., 2010; Skedros et al., 2011). However, this model appears somewhat dogmatic, because the fibrils in each lamella are supposed to be strictly oriented along a vector which should change only in the passage from one lamella to the next. Examining the comparative SEM—polarized light figures of the lamellae, it was possible to observe in the same lamella also additional birefringence bands without a correspondence with a resting line. Therefore, within a phase of apposition, the direction on which osteoblasts orient the fibrils is not rigidly constant, but the angle can change gradually as apposition advances. These observations agree with the model of lamellar units as proposed by Weiner et al. (1997, 1999) and Wagermaier et al. (2006). They distinguished five sublayers of the collagen fibrils array characterized by different thickness and angles of the fibrils orientation.
The classic model of the osteon development is based on the assumption of an ordered concentrical apposition for constant intervals of time. Marotti (1996) proposed a theory of osteocyte differentiation from the osteoblasts during the osteon formation which postulates that, when a previously formed osteocyte becomes embedded within a lamella, it sends inhibitory signals through its dendritic processes to the neighboring osteoblasts which reduce their individual apposition rates. However, this contrasts with the regular alignment of the osteoblasts on the surface of the lamella and the ordered layout of the fibrils within the lamella. The size variations and the different lamellar patterns we have documented would suggest a less schematic model which is not based on the individual osteoblast apposition rate, but rather on the synchronous activity of a pool of osteoblasts. The latter hypothesis fits well with the results of Kerschnitzki et al. (2011) in that, after comparing the spatial arrangement of primary (microlamellar) and lamellar bone, they observed that the formation of a highly oriented collagen matrix requires an alignment of osteoblasts, whereby they can build a wide layer of matrix with a long range order on a solid substrate (scaffold).
A morphometric analysis of all the osteons of a section of diaphyses is not possible because of the remodeling and the superimposition of new osteons to the older. For this reason, in this study, only completely structured and un-notched osteons (which had not yet undergone remodeling) were selected. The observed lamellar pattern is consistent with a model where, in the canal dug by the osteoclasts of the cutting head, a osteoblasts pool differentiate from the perivascular cells of the cutting cone vascular loop and are committed to bone matrix layers apposition, whose fibrils are oriented by the processes network below the osteoblasts (Nanci, 1999; Pazzaglia et al., 2010, 2011). Although the osteoclasts are free to drift on the bone surface during resorption, the osteoblasts are bound to a fixed position by their processes that penetrate into the canalicula. They can move along a radial trajectory lengthening the processes as the underlying lamella grows, but they do not have the chance of lateral displacement. The osteoblasts of the pool are packed one near the other either during the active apposition or in the resting phase (Pazzaglia et al., 2012). The assumption that this activity should be phasic can explain the observed lamellar patterns.
The study was carried out with a scanning electron microscope of “Centre Great Instruments” of Insubria University.