Ontogenetic Patterning of Cortical Bone Microstructure and Geometry at the Human Mid-Shaft Femur

Authors


Abstract

The bone growth process has long-lasting effects on adult bone structure and mechanical adaptation, yet the tissue level dynamics of growth are poorly studied. The specific aims of this study were to (1) quantify changes in bone size and shape through ontogeny, (2) describe the distribution of tissue types and patterns of cortical drift and expansion through ontogeny, and (3) explore relationships between cortical drift and ontogenetic variation geometric size and shape. The study utilized 14 juvenile (ages 2–19) mid-shaft femur blocks removed at autopsy from individuals who died suddenly. Eighty-μm-thick sections were imaged using polarized and brightfield microscopy. For descriptive purposes the sample was divided into five age groups. Features of collagen fiber matrix orientation, vascularity (e.g., pore orientation and density), and osteocyte lacunar density and shape were used to classify primary and secondary tissue types in LM images. This information, combined with evaluation of resorptive versus depositional bone surfaces, was used to identify cortical drift direction. A pattern of posterior and medial drift was identified at the mid-shaft femur in the toddler years. The drift pattern shifts antero-laterally in late childhood, predating the appearance of a more adult-like geometry. On the basis of the presence of transitional fibrolamellar bone complex, growth is more rapid during the toddler years and peri-puberty, and slower in early to late childhood and in later adolescence. Extensive variability in histological and geometric organization typifies the sample, particularly beginning in late childhood. The potential implications of this variability for adult fracture risk warrant further study. Anat Rec, 2009. © 2008 Wiley-Liss, Inc.

Throughout childhood, adolescence, and into early adulthood our bones undergo dramatic changes in size, shape, and organization through the processes of modeling and remodeling. An increasing volume of literature has focused on the importance of bone growth in fetal life, early childhood, and adolescence as a determinant of osteoporosis risk (Cooper et al., 2002; Duan et al., 2003). It appears that the bone that we acquire as children may have long-lasting effects on resultant adult bone structure and mechanical adaptation (Oliver et al., 2007). As such, an improved understanding of the normal human bone growth process will have implications for our understanding of bone loss during aging and in diseases such as osteoporosis. Knowledge gained from the study of normal subadult long bone samples is also needed to understand how various juvenile disease states may result in deviations from normal bone gain during childhood. Moreover, such information can help us better interpret functional adaptations of the skeleton—important in both anthropological and orthopaedic contexts.

Within the cortical bone of the adult human femoral diaphysis, extensive regional variability in microstructural organization has been reported. This includes the number and size of secondary osteons (Pfeiffer et al., 1995; Drusini, 1996; Pfeiffer, 1998; Pfeiffer et al., 2006; Chan et al., 2007), the distribution of collagen fiber orientation (Goldman et al., 2003b), mineralization density (Goldman et al., 2003a; Goldman et al., 2005), and porosity (Lazenby, 1986; Feik et al., 1997; Bousson et al., 2001; Thomas et al., 2005). Of particular interest to the present question concerning the relationship between bone growth and resultant adult structure, Goldman et al.'s studies noted the presence of areas of extensive periosteally and endosteally deposited primary bone in specific anatomical regions of the femoral mid-shaft (largely medially and anteriorly; present in some individuals well into their fifth decade). Given the high degree of mineralization density of these bone regions, it was suggested that they might represent remnants of bone laid down through the process of modeling during subadult growth periods (Goldman et al., 2005). Given the variable retention of these regions of primary lamellar bone deposition within the adult bone cortex, the authors questioned whether similar degrees of variability could be found in childhood bone development, and whether the regional distribution of this bone tissue could be explained by patterns of cortical drift during growth. Unfortunately, little data exists on tissue level changes in bone microstructural organization through ontogeny, limiting our ability to link such tissue-level variability in the adult femoral cortex to that established during the growth process. In this article, we report on such a study utilizing an ontogenetic sample of mid-shaft femora, in an effort to elucidate the effects of bone modeling on overall tissue level structure.

The characteristic size, shape, and curvature of long bone diaphyses is accomplished during ontogeny by modeling [(Frost, 1973) [originally termed “growth remodeling” by Enlow (1963; 1966)]. This process involves coordinated patterns of bone deposition and bone resorption on complementary bone surfaces (Enlow, 1962a; Enlow, 1963; Enlow et al., 1982; Enlow and Hans, 1996) during growth, and to a lesser extent throughout life (Seeman, 2003). Coordination of these modeling processes may result in “drift” of a whole bone cortex (see Fig. 1A), leading to changes in relative position within a cortex of bone tissue laid down during earlier growth stages (Enlow, 1963; Biewener and Bertram, 1993b). This process has been differentiated from that of bone remodeling (A.K.A. “secondary remodeling,” Enlow, 1976), in which bone resorption and bone deposition occur at the same site (with the formation of the Basic Multicellular Unit or BMU) (Frost, 1986), resulting in the replacement of older bone volumes with new bone material through the lifespan (forming Haversian systems or secondary osteons within intracortical areas). These two processes may or may not be linked to the same regulating mechanism (Martin, 2000); however, both are believed to be influenced by mechanical loading as well as phylogenetic (Cubo et al., 2005), metabolic (Lanyon, 1993), and hormonal (Parfitt, 1994; Lee and Lanyon, 2004) influences. Further, the distribution of secondary remodeling events in the adult is likely influenced by the patterns of modeling during growth, which itself determines the distribution of tissue types within the developing cortex (Enlow, 1962b, 1976; Goldman et al., 2005). Although the distributions of secondary bone (e.g. Haversian systems and their remnants) in the adult human (and to a lesser extent ontogenetically) have been the subject of some study (e.g., Chan et al., 2007) the distribution of primary tissues (those laid down at the endosteal and periosteal surfaces through the modeling process) is poorly known, leaving an important gap in our understanding of bone tissue level dynamics.

Figure 1.

(A) Specimens were oriented such that the linea aspera was positioned posteriorly, with a predominantly medial Sharpey-fiber orientation at the linea aspera. Depositional (+) versus resorptive (−) surfaces were differentiated according to characteristics described in the text in order to reconstruct a predominant drift direction (postero-medial in this case—large black arrows). In this brightfield transmitted light image of a three-year-old male cross-section, Sharpey's fibers can be visualized as darker streaks at the base of the linea aspera (small arrows). (B) Sharpey's fibers can be readily visualized in circularly polarized light—see arrows. SO, secondary osteon; RB, resorption bay. (C) Sixteen sectors were delineated at equal angles radiating from the centroid of the cross-section. Cortical widths were measured where eight of these radial sectors crossed the bone (yellow lines). Abbreviations for radial lines as follows: P, posterior; PM, posteromedial; M, medial; AM, anteromedial; A, anterior; AL, anterolateral; L, lateral; PL, posterolateral.

Studies of subadult human bone samples that have focused on tissue-level bone growth dynamics include those of iliac crest bone biopsies (Glorieux et al., 2000; Parfitt et al., 2000; Rauch et al., 2007), and archaeological (Pfeiffer, 2006) and autopsy (Epker and Frost, 1965; Reid and Boyde, 1987; Streeter, 2005) derived samples of ribs. However, a lack of sample availability has largely limited research on long-bone growth in humans to assesment of bone size, geometry, and density using in vivo methods such as radiography, dual energy X-ray absorptiometry (DXA), and quantitative computed tomography (QCT) (Petit et al., 2005; Wren and Gilsanz, 2006). These studies include examinations of mid-shaft femora from archaeological contexts (Ruff et al., 1994; Sumner and Andriacchi, 1996), as well as some recent clinical, pQCT studies of the tibia (Kontulainen et al., 2005; Macdonald et al., 2006). The few notable, descriptive histological works exist from the early 20th century (Foote, 1916; Demeter and Matyas, 1928; Amprino and Bairati, 1936) were generally limited by their small sample sizes and other methodological issues, and completed without the benefit of a modern understanding of bone growth mechanisms.

With such a paucity of data available on subadult age changes in cortical bone tissue organization and its variability owing to the modeling process during growth, we embarked on an ontogenetic study utilizing a rare sample of human mid-shaft femur cross-sections obtained from an autopsy collection of individuals of known age and sex. We aimed to (1) quantify changes in bone size and shape through ontogeny using measures of cross-sectional geometry, and (2) describe the distribution of tissue types and patterns of cortical drift and expansion through different stages of childhood growth, and (3) explore relationships between these processes of cortical drift and resultant ontogenetic variation geometric size and shape. This study will establish fundamental aspects of local bone growth dynamics, which may underlie geometric changes that have been extensively documented through human childhood growth at the femoral shaft in previous studies (Ruff et al., 1994; Sumner and Andriacchi, 1996; Ruff, 2003b).

MATERIALS AND METHODS

Materials

This study utilized human subadult specimens, derived from the “Melbourne Femur Collection”—a repository of human midshaft femur blocks removed at autopsy from the Victorian Institute of Forensic Medicine (VIFM), in collaboration with the University of Melbourne, School of Dental Science, Australia (Bertelsen et al., 1995; Feik et al., 1997). Samples were obtained from individuals who died suddenly, and information on age, sex, height, weight, and cause of death is available for each individual in the sample. The collection was obtained with authorization by the VIFM Ethics Committee, which operates under Australian NHMRC human research ethical guidelines. Drexel University IRB exempt approval has also been obtained (University Protocol no. 1439). A total of 14 individuals were included in this initial study, including nine males and five females, who were divided into five age groups: Toddler (2–4 years), Young child (5–8 years), Older child (9–11 years), Young adolescent (12–16 years), and Older adolescent (17–19 years), as described in Table 1.

Table 1. List of specimens and data on the orientation of the Imax axis (theta)
Age groupSpecimen #AgeSexHeight (cm)Weight (kg)ThetaLocationa
  • a

    Location is defined by where the theta angle falls relative to the sectors defined in Fig. 1c. Anteromedial-posterolateral (AM-PL) includes angles from +45° to +67.5°, Antero-posterior (A-P) includes angles from +67.6° to −67.6°, postero-medial - antero-lateral (PM - AL) includes angles of −67.5° to −45°.

Toddler852M971448.8AM - PL
962F971659.3AM - PL
243Mn/a1547.2AM - PL
1533M991776.6A - P
Young child1505M981679.3A - P
Older child959F1303580.7A - P
16510M13124−87.2A - P
11211M15442−81.5A - P
Young teen15214F15961−58.5PM - AL
2015Mn/an/a87.1A - P
12316M18666−86.9A - P
Older teen16717F15359−70.3A - P
3818F16253−88.1A - P
2119M17066−88.9A - P

Specimen Preparation

At the time of their collection, bone blocks were fixed in a 70% ethanol solution. For the purposes of this study, a thinner cross-sectioned block (each less than 0.5 cm in thickness) was removed from each sample, then cleaned, dehydrated, and embedded in poly methyl methacrylate (PMMA) according to procedures described and modified from Goldman et al. (1999). Using a series of graded carbide papers, the block was polished to a 1200 grit finish on one surface, then mounted with its polished face down to a plastic slide using Technovit 7200 VLC light cure adhesive. The block was then removed from the slide using a Buehler (Lake Bluff, Ill) Isomet 1000 saw, leaving behind a ∼300-μm thick section. The resulting section was ground down to an ∼80-μm section thickness, again using a series of graded carbide papers, polished with a 1-μm diamond suspension using a Buehler Ecomet III, and coverslipped for transmitted light microscopy imaging.

Imaging

Each prepared thin-section was imaged using transmitted light microscopy with (CPL) and without (LM) the presence of circularly polarized light filters. Images were collected using a Zeiss Axioplan 40 (Wexlar, Germany) transmitted light microscope fitted with an automated X, Y, and Z stage, with output via an Optronix digital CCD camera to MBF Bioscience's (South Burlington,Vt) Virtual Slice software program. Images were thus obtained in a montage across the whole cross section, resulting in a single high-resolution image of the entire cross-section (pixel size = 1.44 μm). Binary images were created by gray-level thresholding of transmitted light montages for cross-sectional geometric measurements.

Sample Orientation

Samples were collected by mortuary staff at autopsy from an approximate mid-shaft location. The blocks may have been removed from either limb, and no information on the supero-inferior/medio-lateral orientation of the specimen was maintained. Approximate posterior (vs. anterior) orientation can be ascertained by the presence of the linea aspera, however medio-lateral orientation is more problematic. To maintain consistency in our descriptions and analyses we decided to orient all of our images such that the linea aspera is located towards the bottom of the montaged image, with the most prominent Sharpey's fibers (SF) entering to the left side of the linea aspera (See Fig. 1A,B). Although it is impossible to say with certainty whether the left side of these images represents the medial or the lateral cortex, based on observations of Sharpey's fiber insertions and linea aspera shape from histological sections of adult mid-shaft femora (Goldman, 2001), and from cross-sectional images derived from MicroCT data sets of known orientation, archaeologically derived juvenile samples (Goldman and Robbins, unpublished data), we are confident that the left side of our images represents the medial cortex. For the purposes of this article, we describe medio-lateral differences based on this orientation scheme.

Geometric Properties

Each LM image was reduced to 25% of the original pixel number, and converted to a binary in Adobe Photoshop so that background pixels, representing regions outside of the bone cortex including any spaces in the bone (e.g., Haversian canals, resorption bays), were given a gray level value of 0 (black), and bone a gray level of 255 (white). The binary image was analyzed in ImageJ (http://rsb.info.nih.gov/ij/) using “MomentMacroJ,” a macro for calculating cross-sectional moments (made publicly available by Dr. Chris Ruff). The following variables were analyzed in this study: Total Subperiosteal Area (TSPA), Cortical Area (CA), and Medullary Area (MA); and theta [θ] (defined as the angle between the medio-lateral axis of the cross-section and the orientation of maximum bending rigidity [Imax]) (Ruff and Hayes, 1983). Cortical width was then measured along each of eight radial transects, positioned as shown in Fig. 1C.

Geometric measures and cortical widths were plotted against age in order to document changing geometric properties of the sample through ontogeny that may reflect cortical and medullary expansion or contraction. The purpose of this analysis was to compare the geometric properties of our sample relative to that of larger clinical and/or archaeological collections to determine if our sample was representative in its growth trajectories, and if any consistencies in the orientation of the axis of greatest bending rigidity could be detected within age groups.

Tissue types and Cortical Drift Patterning

Patterns of bone microscopic organization were classified qualitatively into different primary and secondary tissue types according to the criteria used by McFarlin (2006), following Francillon-Viellot et al. (1990) and Enlow (1963), with some modifications. According to this classification, tissue types are defined based on developmental origin, collagen fiber matrix orientation, vascularity (e.g., pore orientation and density), and osteocyte lacunar density and shape. Figure 2 provides examples and brief descriptions of several tissue types that have been reported in human and nonhuman primate juvenile bone. For each age group, we qualitatively describe distributions of primary periosteal bone tissues, primary endosteal bone tissues, and intracortically remodeled (secondary osteonal) bone. Regional variation in the degree of porosity and in pore orientation within regions of primary bone tissue is also reported, but not quantified.

Figure 2.

Illustrations of bone tissue types recognized in primate skeletons, according to the definitions of Francillon-Viellot et al. (1990) and Enlow (1963), as modified by McFarlin (2006). (A) Classification by collagen fiber matrix organization—imaged here in circularly polarized light (CPL). Lamellar bone is a slow-forming tissue type characterized by a highly ordered arrangement; the optical distinction of individual layers, or lamellae, is the result of regular changes in the preferred orientation of collagen fiber bundles. Woven bone is a rapidly forming tissue type, characterized by loosely packed collagen fibers coursing in all directions in a more or less random arrangement. (Parallel-fibered bone, intermediate in its collagen fiber organization, is not shown here.) Fibro-lamellar bone is classically deposited as an initial framework of fine cancellous trabeculae of woven bone; the intervening cancellous spaces are later in-filled by deposition of lamellar bone, forming primary osteons where vascular canals are present. Transitional fibro-lamellar tissue patterns are also observed, in which the initial cancellous scaffolding may be formed by a mixture of woven, parallel-fibered and/or lamellar tissue types. (B) Vascular canals may be incorporated into growing bone in a variety of orientations—imaged here in brightfield microscopy (LM). True plexiform bone, as shown here is not seen in primates; however, since some authors have used the term somewhat interchangeably with fibrolamellar bone it is included here for comparative purposes. (C) Two additional tissue types are shown here, imaged in CPL. Endosteal compacted coarse cancellous bone is an endosteally deposited tissue formed in bone areas originally comprised of coarse cancellous trabeculae. Intervening spaces between trabeculae are later in-filled (or “compacted”) with parallel-fibered or lamellar bone as these bone regions are incorporated into the developing cortex, giving this tissue type a distinctively convoluted appearance. Secondary osteons can also be observed in this image. Secondary osteonal (Haversian) bone is formed by the sequential resorption and deposition of bone at single sites within the cortex.

Cortical drift patterns were determined using microstructural characteristics clarified by Enlow (1963) including the distribution of primary bone tissue types, reversal cement lines indicating a change in growth direction, and/or features that indicate recently depositional/resting (a smooth surface with unremodeled primary bone tissue; or in faster growing periosteal tissues, an appearance of active incorporation of primary vascular channels at the periosteal surface) versus resorptive bone surfaces (a scalloped surface indicative of the presence of Howship's lacunae). To quantify surface modeling status (e.g., resorptive vs. depositional/resting), LM cross sections were traced using coded contour lines based on morphological features indicative of being either recently resorptive (erosion surface) versus depositional (or resting). An additional category of “trabecularized endocortical bone” was added to demarcate regions where subendosteal resorption was followed by minimal amounts of subsequent localized lamellar bone deposition, but that had likely been largely resorptive in their recent history. Periosteal surfaces in the region of the linea aspera were quantified separately under a category called “Sharpey's Fibers”. Perimeter measures for each outline were quantified, as was the ratio of erosion surface (ES = sum of resorptive and trabecularized endocortical bone surface) to total bone surface (BS). Further, for each of eight sectors in the cortex (demarcated with radial transacts, see geometry section above, Fig. 1), the predominant condition of the surface (resorptive vs. depositional/resting) was charted.

To demonstrate the patterning of cortical drift within each cross-section, boundary lines demarcating regions comprised predominantly of either primary bone of endosteal origin (blue) or periosteal origin (red) were delineated from bone areas comprised largely of secondarily remodeled bone (i.e., where ∼50% or more of the cortex was remodeled) (see Fig. 6). Further, since a continuous boundary between periosteally and endosteally deposited cortex was often difficult to trace in heavily remodeled samples, asterisks were used to mark those regions where enough interstitial bone remained to confidently identify the innermost distribution of periosteally deposited bone cortex (as viewed in circularly polarized light images). The resultant line of green asterisks (see Fig. 6) therefore represents the approximate periosteal-endosteal boundary, and reflects long-term drift direction in the sample. All tracings and overlays were drawn utilizing MBF Bioscience's (S. Burlington, VT) Stereoinvestigator software program, with data exported to Excel (Microsoft Corporation).

Analysis

Given the small and heterogeneous sample sizes within age groups, the current study sample was not appropriate for quantitative statistical analysis. Plots of geometric properties relative to age provide a useful method of characterizing growth trends, which can then be related to observations of age-related variability in histological features.

RESULTS

Geometry

Images of representative bone cross-sections within each age group are shown in Fig. 3. Raw data for theta measurements are provided in Table 1. Means and standard deviations by age group for other geometric measurements are provided in Table 2.

Figure 3.

Binarized versions of brightfield images demonstrating cross-sectional geometric appearance of representative individuals from each of the five subadult age groups.

Table 2. Descriptive statistics for cross - sectional geometry data
Age groupTA (mm2)MA (mm2)CA (mm2)Cortical Width (mm)
MeanStd. DevMeanStd. DevMeanStd. DevSectorMeanStd. Dev.
  1. TA, total subperiosteal area; MA, medullary area; CA, cortical area. Cortical Width measurements taken at 8 sectors around the cortex, automatically determined using a customized MATLAB program; see Figure 2 for sector definitions.

Toddler (n = 4)160.8714.6547.1611.18113.7112.12P4.390.65
PM3.800.50
M3.320.66
AM3.000.54
A2.700.35
AL2.880.42
L2.960.62
PL3.840.88
Young child (n = 1)172.33n/a59.79n/a112.54n/aP3.79n/a
PM3.46n/a
M3.20n/a
AM3.07n/a
A2.78n/a
AL2.78n/a
L3.06n/a
PL3.21n/a
Older child (n = 3)340.4498.32104.7314.82235.7185.54P5.091.55
PM5.101.29
M4.270.75
AM4.330.35
A4.591.51
AL5.032.26
L4.471.38
PL4.551.43
Young teen (n = 3)537.9796.78140.8569.41379.1234.70P8.380.99
PM7.131.14
M6.470.75
AM6.370.86
A5.281.51
AL6.760.83
L6.370.58
PL6.331.06
Older teen (n = 3)477.5869.25100.8315.74376.7454.25P9.220.95
PM7 030.59
M6.790.74
AM6.240.39
A5.611.15
AL6.830.36
L6.761.08
PL6.210.99

Figure 4 is a plot of mean TA, CA, and MA for each of the five age groups. Note that in the youngest individuals in the sample, TA and CA appear to increase in a similar pattern to MA, suggesting bone expansion is occurring at the same rate as endosteal expansion. By late childhood, however, the MA curve appears to diverge, as the relative increase in cortical area outpaces the resorption endosteally. Beginning in the young adolescent individuals, there is a trend towards net formation endosteally as MA begins to decline, while CA begins to level off. A slight decrease in areal properties of the bone is seen between early and late adolescence.

Figure 4.

Plot of mean total area (TA), cortical area (CA), and medullary area (MA) plotted relative to the five age groups defined in the text.

As demonstrated in Fig. 3, femora in the toddler group are not only smaller in size compared to older age classes, but are shaped differently as well, with a thickened posterior cortex. Figure 5 is a plot of regional variability in cortical widths with age, again demonstrating the thickened posterior, postero-medial, and postero-lateral cortices in the toddler group, as well as the young child individual. With continued growth, the cross-sectional dimensions increase because of net apposition of bone periosteally and net resorption endosteally. This results in a greatly expanded cortex relative to the toddler group, and an expanded medullary canal. By the adolescent years, periosteal expansion still typifies the sample, but medullary area is relatively reduced, suggestive of net endosteal deposition. The geometric shape of the cross section also appears more adult-like, based on the location of the axis of greatest bending rigidity (theta), which most often falls in an anterior-posterior (A-P) or antero-lateral to postero-medial (AL-PM) plane (rather than antero-medial to postero-lateral (AM-PL) as in the toddler samples) and the thicker antero-lateral cortex relative to younger age groups (Fig. 5).

Figure 5.

Plot of mean cortical width for each sector for each age group. See Fig. 2 for explanation of radial sector abbreviations.

Figure 6.

Cortical drift maps created using MBF Bioscience's Stereoinvestigator software to manually overlay tissue type boundaries on CPL images. The red and blue lines demarcate the boundary at which approximately half of the primary periosteal and endosteal cortices respectively are secondarily remodeled. The green asterisks follow the approximate boundary between endosteally and periosteally derived tissues. These maps support continued postero-medial drift in the younger two sample groups, and an antero-lateral drift in the older groups.

Histological Description

The following are qualitative observations of cortical bone tissue organization within each of the five subadult age categories. Refer to Fig. 2 for a description of each tissue type discussed.

Toddler (Age 2–4 years, N = 4)

The toddler age group is typified by a primary periosteal cortex that includes transitional fibrolamellar bone, and histological features that suggest a predominantly postero-medial cortical drift direction

All individuals exhibit transitional fibrolamellar bone in areas of periosteal bone apposition (medial and lateral surfaces) with vascularization via primary osteons and scattered simple primary vascular canals, both having a predominantly longitudinal orientation. Along the postero-medial aspect in particular, primary canals appear more circumferentially oriented, approaching the appearance of laminar bone (Fig. 71B).

Figure 7.

7-1 (A) CPL montage of a two-year-old male cross-section demonstrating postero-medial drift. (B) Postero-medial cortex demonstrating transitional fibrolamellar bone. Vascularity is increased in this area, with a laminar organization. RB, resorption bay; SO, secondary osteon. (C) The antero-lateral cortex is endosteally derived, resorptive at periosteal surface (arrows), and contains remnants of compacted coarse cancellous bone (CCC). 7-2 (A) CPL montage of a five-year-old male cross-section demonstrating periosteal expansion along with continued postero-medial drift. (B) The primary periosteal cortex, seen here along the postero-medial side, is moderately vascularized primary lamellar bone (PLB). (C) In many regions, such as this postero-lateral cortex, remnants of transitional fibrolamellar bone (arrows) can be observed in regions of older periosteal bone deposition. 7-3 (A) CPL montage of a nine-year-old female cross-section demonstrating periosteal expansion along with antero-lateral drift. (B) The primary periosteal cortex of this individual, shown here along its postero-medial aspect, is entirely lamellar with low vascularity. (C) The antero-lateral cortex of this same individual contains extensive periosteal primary lamellar bone deposition with numerous primary vascular canals. 7-4 (A) CPL montage of a 14-year-old female cross-section demonstrating continued periosteal expansion, along with antero-lateral drift. (B) The primary periosteal cortex is entirely lamellar; however, it is relatively less extensive and vascularized postero-medially Compared to antero-laterally (C). 7-5 (A) CPL montage of an 18-year-old female cross-section demonstrating continued periosteal expansion and antero-lateral drift, along with endosteal contraction. (B) The endosteal surface of this individual is smooth indicating it was either recently depositional or resting. (C) Evidence of extensive periosteal primary lamellar bone deposition can still be observed, particularly along the antero-lateral cortex, though pores appear to be less numerous.

Endosteally deposited bone surrounds the medullary cavity with the exception of the posterior and postero-medial cortex (Fig. 71A). Areas of more recent endosteal deposition (closest to the medullary cavity) are circumferential lamellar in organization with some limited regions of compacted coarse cancellous (CCC) bone deeper in the cortex. Vascular canals appear to be largely radially oriented (See Fig. 71C).

In the boundary region between endosteally and periosteally deposited cortex, the bone is heavily remodeled, including numerous Haversian systems, often with a characteristic drifting morphology (Robling and Stout, 1999). A concentration of large resorption bays can be observed near the endosteal–periosteal boundary. The posterior cortex is heavily remodeled, though remnants of periosteally deposited primary cortex can be seen interstially throughout its width. Conversely, the interstitial bone areas of the anterior cortex are largely endosteally deposited tissue (see Fig. 6A).

Distributions of surface modeling are reported in Table 3 and support the tissue distributions described above. Note that both two-year-old individuals in the sample exhibit scalloped (resorptive), periosteal surfaces anteriorly and to a large extent antero-laterally. By three years of age, however, the antero-lateral surface demonstrates periosteal deposition (Fig. 8A). Some additional variations observed in this age group are illustrated in Fig. 8B,C.

Figure 8.

(A) The antero-lateral cortex of this three-year-old male individual demonstrates a depositional periosteal surface. A reversal line can be seen in a deeper region of this cortex (arrows) suggesting that this had been a resorptive surface at some earlier stage of development (as it is in the 2 two-year-old individuals in the sample). Field width = 1.73 mm (B) CPL montage of a two-year-old male cross-section demonstrating more medially directed cortical drift relative to the other individuals in this age group. Extensive compacted coarse cancellous bone can be seen within the lateral cortex, the entire extent of which is endosteal in origin. (C) This three-year-old individual is quite different from the others in its age group, with a thickened cortex and periosteal deposition around its circumference. (D) Brightfield image from the lateral cortex of an 11-year-old male individual demonstrating extensive transitional fibrolamellar bone deposition within the primary periosteal cortex. Field width = 1.73 mm (E) CPL montage of a 10-year-old male individual demonstrating periosteal expansion and lacking evidence for antero-lateral drift. (F) CPL montage of a 17-year-old female cross-section demonstrating periosteal expansion, but again a lack of evidence for antero-lateral drift and a very different geometry from most, but not all, of the adolescent specimens. Scale bar for all CPL montage images = 2 mm.

Table 3. Surface modeling data
Age groupSpecimen#AgeSexTotal Periosteal BSTotal Endosteal BSTotal Periosteal ESTotal Endosteal ESTotal subendosteal trabecular surface% Periosteal ES/BS% Endosteal ES/BSPredominant surface characteristics by sector
 AALLPLPPMMAM
  1. Resorptive surfaces (with characteristic scalloped Howship's lacunae) and subendosteally trabecularized surfaces (see text for explanation) were traced both periosteally and endosteally as described in the text. ES, erosion surface; BS, bone surface (in μm). Note that the ES used in the % Endosteal ES/BS calculation includes the sum of endosteal resorptive surfaces and subendosteal trabecularized surfaces. A surface labeled as “+” is interpreted as depositional or resting based on the presence of a smooth surface and/or evidence of vascular budding. A surface labeled “−” is interpreted as either actively or recently resorptive, based on the presence of scalloped Howship's lacunae. A surface region labeled +/− contained regions of both despositional and resorptive surfaces. A surface labeled “SF”contained many Sharpey's fibers, is interpreted as a muscle attachment site, and its resoptive vs. depositional condition was not quantified.

Toddler852M46581.528092.412327.710964.60.026.539.0Per+SF+++
End+++
962F51071.031061.511809.014269.10.023.145.9Per+/−++SF+++/−
End++++/−+
243M46945.530265.26423.717171.50.013.756.7Per+++SF+++
          End+++/−+
1533M46475.121276.10.011442.68401.60.093.3Per++++SF+++
End+/−
Young child1505M47804.429951.26728.012414.70.014.141.4Per++/−+/−++/SF+++
END++++/−+
Olderchild959F63350.937317.20.011130.715133.80.070.4Per++++SF+++
End+/−++/−
16510M59234.539927.3474.55379.317898.10.858.3Per++++SF+++
End+++/−+
11211M78044.842686.03037.66772.035914.03,9100.0Per++++SF+++
End
Youngteen15214F79694.941745.20.0915.722428.80.055.9Per++++SF+++
End+/−++++/−
2015M82218.538026.00.04052.723593.40.072.7Per++++SF+++
End+
12316M92924.363081.94325.54142.058939.94.7100.0Per++++SF+/−++
End
Older teen16717F79558.344213.72482.35570.038643.73.1100.0Per++++/−SF+++
End
3818F77560.934876.10.00.00.00.00.0Per++++SF+++
End++++++++
2119M68703.839579.30.00.00.00.00.0Per++++SF+++
End++++++++

Young child (age 5–8; N = 1)

The young child age group is typified by a primary periosteal cortex that is lamellar, with a shift towards periosteal expansion around the entire cortex, and some continued posterior drift

In the single five-year-old male individual representing this age group (Fig. 72A), the primary bone of the outer periosteal cortex is comprised almost entirely of lamellar bone with predominantly longitudinally oriented primary vascular canals and primary osteons (Fig. 72B). Qualitative observations suggest an increase in primary canal number and/or size along the medial cortex relative to other primary bone areas. Transitional fibrolamellar bone was limited to interstitial regions representing areas of older deposited periosteal cortex (Fig. 72C).

The endosteal primary cortex is composed of circumferentially deposited lamellar bone, and is observed anteromedially, anteriorly, and antero-laterally. Similar to that seen in the toddler age group, the boundary region between endosteally and periosteally derived bone tissues is largely remodeled with numerous Haversian systems.

Details of resorptive versus depositional surface locations are reported in Table 3. These data, together with the descriptions above, support a shift towards periosteal expansion around the entire cortex, along with some continued posterior cortical drift.

Older child (age 9–11, N = 3)

The older child age group is highly variable in its microstructural organization and cortical drift. Although some individuals show characteristics similar to the young child group, others demonstrate changes that anticipate those seen in the young adolescents

The primary periosteal cortex is generally comprised of lamellar bone (Fig. 73B,C). Two individuals show local distributions of transitional fibrolamellar bone with predominantly longitudinal vascular canals, along the bone's medial side in one case (no. 165), and its lateral in the other (no. 112, Fig. 8D).

The primary endosteal cortex is also lamellar in organization, though remnants of CCC bone are present in one individual. Dense intracortical remodeling is largely limited to endosteally deposited regions, although scattered secondary osteons are also observed in periosteal cortex.

This age group is highly variable in tissue type distribution and cortical drift direction. In one specimen (Figs. 6D and 73A), the bone cortex is expanding (with signs of periosteal deposition/resting surfaces, see Table 3) in all directions, but evidence of recent antero-lateral drift orientation is indicated by the following observations: more extensive primary periosteal cortex antero-laterally with many obvious vascular canals (Fig. 73C) compared to medially, and endosteal bone deposition medially and resorption laterally. A second individual shows a similar drift pattern, but histologically contains evidence of recent, transitional fibrolamellar bone deposition, particularly along the lateral cortex (no. 112, Fig. 8D). The third individual also contains some regions of transitional fibrolamellar bone deposition, but shows little evidence of a current predominant drift orientation based on surface modeling morphology (no. 165, Fig. 8E). However, a recent history of postero-medial drift is indicated by the extensive endosteal bone deposition anteriorly, and thickened posterior and medial cortices.

Early adolescent (age 14–16, N = 3)

The early adolescent age group is typified by a lamellar primary periosteal cortex, and evidence of antero-lateral cortical drift

In all individuals, the primary periosteal cortex is entirely lamellar, typified by primary vascular canals and occasionally some primary osteons (Fig. 74B). It is entirely depositional/resting, indicative of continued periosteal expansion. The primary endosteal cortex is also lamellar, and thickest along the medial and posterior cortices. Intracortical regions of endosteal origin are largely secondarily remodeled, while regions of periosteal origin contain fewer Haversian systems. In two of three individuals, continued antero-lateral drift is suggested (Fig. 6D), as evidenced by an extensive and vascular primary periosteal cortex anteriorly and laterally (Fig. 74C), posterior and/or medial endosteal primary lamellar bone deposition, and trabecularized subendosteal bone laterally (Fig. 74A, Table 3).

Late adolescent (age 17–19, N = 3)

The late adolescent age group is typified by a primary periosteal cortex that consists of lamellar bone, and variable cortical drift patterns

All the individuals in this age group demonstrate a primary periosteal cortex that is entirely lamellar (Fig. 75C). In two of the three individuals, the endosteal surface is entirely depositional, with a rim of primary endosteal lamellar bone around its circumference and a lack of subendosteal trabecularization (Fig. 75A,B). The third individual has a more trabecularized subendosteal surface (Fig. 8F). The samples vary extensively in their overall cortical thickness, geometric shape and extent of intracortical remodeling. Only a single individual (no. 38) shows clear evidence of continued antero-lateral drift (thickened antero-lateral cortex and more extensive and porous primary periosteal bone present in that location).

DISCUSSION

The study of long bone growth in humans has generally been limited to data that can be obtained noninvasively; hence, tissue level details of human limb bone growth dynamics are largely unknown with the exception of some published studies from the early 20th century (Foote, 1916; Demeter and Matyas, 1928; Amprino and Bairati, 1936), and a few more recent anthropological studies that largely focused on age variability in osteonal number (Burton et al., 1989; Baltadjiev, 1995; Sawada et al., 2004). A strong need exists to study human diaphyseal bone histological variability in sub-adults, in light of a modern understanding of tissue level bone growth mechanisms (as established by Enlow in the 1960s). Here we have presented a novel study of bone growth dynamics at the human mid-shaft femur, combining both histological and geometric analyses of cortical bone organization.

Geometry

Ontogenetic variation in absolute bone size

Much of what we know about ontogenetic variation in bone size, and its reflection of surface growth phenomena is based on the series of analyses by Garn and coworkers (e.g. Frisancho et al., 1970; Garn, 1970). These studies were longitudinal in design and utilized metacarpal radiographs to measure total subperiosteal diameter, medullary cavity width, and cortical width. They demonstrated rapid increases in subperiosteal apposition in early childhood, a slowing of apposition in late childhood, followed by an adolescent growth spurt (earlier and of less magnitude in females). Subperiosteal apposition was shown to slow again post-puberty, but continue at a low rate throughout life (later studies, e.g., Martin and Atkinson, 1977; Ruff and Hayes, 1988; Feik et al., 2000 suggest this process is more efficient in males than females). The endosteal surface was characterized largely by net resorption through the growth period. Resorption slowed through the juvenile years, and then reversed to net apposition during puberty, particularly in females (Smithgall et al., 1966; Johnston and Watts, 1969; Frisancho et al., 1970). Through the aging process in adults these studies again demonstrated net resorption endosteally

Our cross-sectional ontogenetic sample largely follows the growth trajectories described above, as well as those reported by Ruff et al. (1994) in a CT-based study of archaeological femora. TA and CA curves parallel one another throughout the growth period, reflecting net periosteal apposition, while the slope of the MA curve suggests initial net resorption throughout childhood, followed by a net gain towards early adolescence. Moreover, periosteal apposition (as reflected by TA and CA) appears to slow in the early adolescent sample, as evidenced by the shallower slope of these curves. However, the current study reveals two notable differences from trends seen in other samples. First, because of the small body and bone size of the one young child in our sample, we see no increase in TA/CA or MA between the toddler and young child age classes. Second, we see a notable decrease, rather than an increase, in TA, CA, and MA in the late adolescent group relative to the early adolescent group. This decrease appears to be caused by the influence of one very large individual (no. 23, large both in height and weight) in the young adolescent group.

Many authors have reported a more pronounced contraction of the medullary cavity in females relative to males due to greater net endosteal deposition/ consolidation (Garn, 1970; Zamberlan et al., 1996; Bass et al., 1999; Neu et al., 2001; Kontulainen et al., 2005; Wang et al., 2005; Macdonald et al., 2006). The two oldest individuals (one male and one female) in the sample both show signs of endosteal deposition, and no obvious sex difference was observed. Further study of larger cadaveric samples will be required to address this important question in the future.

Ontogenetic variation in bone shape

Our study also demonstrated changes in geometric shape through ontogeny. Cortical width data demonstrate that the posterior aspect of the mid-shaft femur is thickened in the toddler/early childhood samples, likely reflective of the posterior cortical drift occurring through this period (as demonstrated by our tissue type analysis, below). The Imax axis in the toddler group tends be oriented in an AM to PL direction. In later childhood, cortical thickness is, on average, similar between cortices, reflective of a general pattern of periosteal apposition in all areas of the cross-section, while the Imax axis shifts to a more anterior-posterior (A-P) orientation. In the younger and older adolescent samples, the geometry of the cross-section often resembles that seen in the adult, with the Imax axis most often falling in the A-P plane, owing to the influence of the prominent linea aspera, or towards an AL-PM orientation. The AL-PM plane has been demonstrated to be the average location of the Imax axis in adult modern humans, according to several studies (Pauwels, 1980; Ruff et al., 1984; Ruff, 1987), including our own of this same population sample (Goldman, 2001).

Our results generally support those reported in previous studies of subadult femur geometric properties. Sumner (1984) demonstrated that at mid-shaft, the Imax axis rotated from a more consistent AM–PL orientation in the young child (age, 1–5, N = 4) group to an A-P orientation in the older child (age, 5-10, N = 5) to a AL–PM orientation in the peri-pubertal to adult age range (age, 10–19, N = 3). A similar shift in bone shape during early ontogeny at the subtrochanteric region of the femoral diaphysis was found by Wescott (2006). The study reports a shift in early childhood (age 0–5) from a more rounded morphology to one that is more mediolaterally broad, similar to that seen in adults. The relationship between these shape changes in early ontogeny and gait are explored further below in the “cortical expansion and cortical drift” section.

Histological Analysis

The study of primary tissue type organization in ontogenetic series can provide insight into growth processes including variation in bone depositional rates through ontogeny (Castanet, 1993; de Ricqlés, 1993) owing to intrinsic and environmental factors (McGuigan et al., 2002), and local mechanical regulators of bone development (Biewener and Bertram, 1993a; Mosley and Lanyon, 2002; Pearson and Lieberman, 2004). Many studies (mostly of dinosaur and avian species, but some mammalian studies as well) have utilized features of primary tissue type and organization to reconstruct the history of growth, and local variations in speed and direction of bone deposition and resorption through ontogeny (Enlow, 1962a; Newell-Morris and Sirianni, 1982; Herrmann and Danielmeyer, 1994; Klevezal, 1996; Chinsamy-Turan, 2005; Erickson, 2005). These studies frequently employ and/or test “Amprino's Rule,” after Amprino (1947) who originally suggested that variation in primary bone tissue types and their vascularity reflected variations in the rate of bone deposition during ontogeny, which in turn reflects different rates of body growth (Francillon-Viellot et al., 1990; de Ricqlés et al., 1991; de Margerie et al., 2002; de Margerie et al., 2004). In the discussion below, we put our results concerning tissue type variability during ontogeny into such a context.

Histological organization: primary periosteal cortex

The toddler age group is characterized by a primary periosteal cortex comprised of transitional fibrolamellar bone with predominantly longitudinal vascular canals. By five years of age, however, recently deposited primary periosteal bone is largely lamellar in organization, with remnants of transitional fibrolamellar bone found only in areas of older bone deposition deep within the cortex. A brief reappearance of transitional fibrolamellar bone deposition could be seen in two late childhood individuals. The primary periosteal cortex of adolescent individuals is again typified by lamellar bone

The transitional fibrolamellar bone identified here is similar to that demonstrated in other primates by McFarlin (2006) who used the term “transitional” to differentiate the tissue from classic fibrolamellar bone due to its somewhat intermediate structure—though similarly vascularized by primary osteons, its initial scaffolding may be formed by a mixture of woven, parallel-fibered and/or lamellar bone tissues. Classic fibrolamellar bone tissues are characterized by rapid depositional rates, typically greater than 5–10 μm/day (e.g., Newell-Morris and Sirianni, 1982; de Margerie et al., 2002). Although we do not have data on the specific growth rate of this transitional fibrolamellar tissue relative to classic fibrolamellar bone, it would be faster forming than lamellar bone. Therefore, its presence in recently deposited regions of the primary periosteal cortex of toddlers, as well its reappearance in some older childhood individuals (ages, 10 and 11), is indicative of periods of relatively rapid growth in these two age groups. Indeed other histological studies have noted this tissue type in various locations of the toddler and peri-pubescent skeleton (Amprino, 1947; Martin and Burr, 1989; Streeter, 2005; Pfeiffer, 2006), though these authors used the term “plexiform” bone to describe it.

This conclusion is supported by data on cross-sectional dimensions of the midshaft femur of human children. These studies indicate a growth spurt at 1-2 years of age, followed by a slowing of growth velocity by the beginning of the third year (Ruff, 2003a). Ruff associated this growth spurt with marked changes in biomechanical loading associated with the transition from crawling to walking during the toddler years. Longitudinal studies of human growth patterns also suggest an accelerated pattern of linear growth of the femur during the toddler years, particularly after age 1.5 (Gasser et al., 1991) followed by a decrease in growth velocity between age 4–9 years of age (Smith and Buschang, 2004). Growth velocities have also been shown to increase with the onset of puberty, reaching a peak in the femur at age 11.8 in females and 13.8 years in males (Smith and Buschang, 2004) (see also Marshall, 1977; Parfitt, 1994). The timing of the toddler growth spurt appears to correlate well with periods of changing gait in childhood, such as the onset of walking (Sinclair and Dangerfield, 1998) and attainment of adult-like gait closer to early childhood (around age, 4–5) (Sutherland, 1997).

Our qualitative observations suggest there is additional variability within primary periosteal tissue types in terms of pore size, number, and orientation between different bone cortices and age groups. On the basis of these observations, we hypothesize that more vascularized primary periosteal tissues are found in the direction of cortical drift. We did not quantify these differences within the current study, as our images did not allow for automated segmentation of pores. We are currently acquiring such quantitative data utilizing 2D microradiographs and 3D microCT scans which will help to address this hypothesis in future, and potentially provide additional evidence to support the drift patterns described in the current article (see cortical expansion and drift section below).

Histological organization: primary endosteal cortex

Even in the youngest individuals in this sample, most bone deposited at the endosteal surface is lamellar in organization, with radially oriented vascular canals. Endosteal bone is most extensive anteriorly and antero-laterally in the toddler and young child samples, while in the older child and adolescent samples it is either present around the entire cortex, or most abundantly along the postero-medial aspect. The distribution of these primary endosteal bone tissues likely reflects the direction of cortical drift in these different age groups

In the toddler, young child and some older child samples, remnants of CCC bone are also observed, possibly representing tissue formed towards the metaphysis but incorporated into the diaphysis as the bone elongated.

Beginning in the older child age group, and continuing through adolescence, many regions of the endosteal cortex demonstrate subendosteal trabecularization. Examination of the bone tissues just beneath the endosteal surface suggests that these cortices were largely resorptive for a period of time, but that some localized bone deposition was occurring on the surface as well, possibly reflecting localized remodeling rather than modeling activity. For analysis purposes, these regions were interpreted as recently resorptive. Additional information concerning mineralization density of the bone at the endosteal surface and just beneath, would help to confirm whether this interpretation is accurate (e.g., using quantitative backscattered electron microscopy or microradiography).

Histological organization: secondarily remodeled bone

Our results section includes qualitative observations concerning the distributions of secondary bone tissue within each of our age groups. These include variations in the distribution of secondary osteons in regions of periosteally or endosteally derived bone matrix, the clustering of forming osteons along periosteal-endosteal boundaries, and the presence of so-called “drifting” or “waltzing” osteons (Epker and Frost, 1965; Robling and Stout, 1999). Previous studies of adult femoral intracortical porosity (Lazenby, 1986; Feik et al., 1997; Thomas et al., 2005) and osteon population density (Pfeiffer et al., 1995) have noted patterned variations that reflect the geometric properties of the cross-section, and potentially mechanical loading demands. Other authors (Jordan et al., 2000; Bell et al., 2001) have described clusters (“super osteons”) of remodeling events occurring in close proximity in more periosteally located regions of the adult femoral mid-shaft cortex. Future quantitative studies of such variations in secondarily remodeled bone using juvenile samples will help elucidate how such patterned variation develops in the femoral cortex may reflect differing tissue age, local circumstances of bone growth, and mechanical loading history during ontogeny (Enlow, 1962b; Bouvier and Hylander, 1981; McFarlin et al., 2008). Given the attention that secondary osteons have received in the anthropological literature as a means of determining skeletal age, such studies would also be relevant for the development of improved skeletal aging techniques, both in juvenile and adult specimens

Cortical Expansion and Cortical Drift

Geometrically, our youngest aged individuals are typified by a relatively round mid-shaft cross-section, with a poorly developed linea aspera. Histologically, there is evidence of continued posterior, and slight medial, cortical drift (based on patterns of periosteal and endosteal deposition and resorption; and in the locations of periosteally and endosteally derived primary bone tissues). Such a drift pattern might be expected owing to continual tension exerted posteriorly through the attachment of the adductor and hamstring muscles at the developing linea aspera (Epker and Frost, 1966; Hoyte and Enlow, 1966). Interestingly, Mittlmeier et al. (1994) modeled the effects of various muscle activities on the formation of the linea aspera and found that the inclusion of the action of the adductor magnus muscle could result in normal linea aspera shape. This muscle attaches along the medial side of the linea aspera, along with other adductors, and may partially account for the tendency towards postero-medial drift.

To our knowledge, no other studies have specifically addressed modeling drifts at the human femoral mid-shaft early in ontogeny using histological organization. Early descriptions and illustrations by Foote (1916) and Demeter and Matyas (1928) suggest such a pattern based on tissue type organization, but do not explore drift specifically. Other regions of the diaphysis have been more well studied, at least from a geometric standpoint. For instance, the growth of the distal femoral diaphysis is believed to be heavily influenced by the extensive changes in bicondylar angle between birth (where it is 0°) and age four (reaching adult values of ∼6°–8° by age 8) (Tardieu and Damsin, 1997) as gait shifts from one characterized by wider spaced feet, more externally rotated hips and increased knee flexion seen in infants and young toddlers, to that of the mature gait seen by the age of five (Keen, 1993). To accomplish this increasing angularity, the femoral diaphysis, and the knee itself, must therefore undergo a medial shift (MacMahon et al., 1995; Shefelbine et al., 2002). The morphology of the mid-shaft femur may too be influenced by this increase in angularity characterizing the onset of walking, though to a lesser extent than distally.

By early childhood, our sample demonstrates evidence of a shift towards periosteal expansion around the cortex, and net resorption endosteally, supporting trends reported in the geometry literature (Ruff et al., 1994). Histologically, however, signs of continued posterior drift are still evident. By late childhood, continued periosteal deposition and net endosteal resorption results in an expanded bone and medullary cavity, again consistent with the geometric literature. This age group, however, is highly variable in drift pattern, with some individuals merely expanding radially, and others demonstrating a shift in cortical drift towards an antero-lateral direction. It is likely that this age group contains individuals that may have been entering the pubertal growth spurt; therefore, increased variability in this group is not surprising.

By early adolescence, our geometric analysis demonstrates that periosteal expansion has typified the sample, along with endosteal contraction/consolidation. Moreover, geometric shape of the cross section begins to appear more adult-like, with the orientation of Imax falling in an A-P or AL-PM direction, reflecting a greater amount of material located further from the neutral axis in this plane. A variety of mechanical explanations may account for this shifting geometric shape towards puberty. For instance, biomechanical loads may change once the valgus angle of the femur is established. In addition, the widening of the hip, particularly amongst females, and resultant changes to femoral neck length, may result in an overall increase in mechanical loading that accompanies increasing muscle mass during the growth spurt (Schonau, 1998).

Interestingly, our data suggest that changes in cortical drift patterning appear to predate changes in geometric properties of the cross section, so that patterns of deposition and resorption at bone surfaces through ontogeny greatly influence resultant adult geometric properties of the femur post-puberty. Possibly reflecting these early histological reflections of drift patterns, primary bone distributions that typify the young adult femur (Goldman et al., 2003a; 2003b; 2005) appear to have their origin from an early age, as do regional variations in the degree and distributions of secondary remodeling. On the basis of our study, we would hypothesize that geometric adaptation of the bone cortex responds to tissue level composition and organizational features established earlier in growth. As both geometric and microstructural properties of bone have been shown to interact to produce bone's overall mechanical properties (Tommasini et al., 2005; Ural and Vashishth, 2006; Tommasini et al., 2007; Tommasini et al., 2008) continued study of these relationships in juvenile specimens, including mechanical testing data, would greatly improve our understanding of the impact of growth phenomena on bone strength through the lifespan.

Limitations of the study

There are a number of limitations to this study. Those related to sample size and loss of orientation have been discussed previously. In addition, the study was limited largely to qualitative descriptions for our tissue type analysis, involving time-consuming manual discrimination, owing to the lack of automated means for differentiating the subtle differences between them. We did attempt to quantify certain parameters, such as resorptive versus depositional surfaces, but did not utilize a traditional, established approach. This would have required alternative histological preparation methods that preserve osteoid and/or cell structures to allow for the more definitive determination of whether bone was actively being laid down or removed at a surface (e.g. Rauch et al., 2006). Preparation needs for our tissue type analysis precluded the use of such methods. Future studies are planned utilizing quantitative backscattered electron microscopy and/or microradiography, which would allow for bone of more recent deposition to be distinguished based on relative mineralization density (Boyde et al., 1999; Goldman et al., 2003a; Roschger et al., 2008). Finally, in this study, we only report qualitative descriptions of pore distribution within primary bone tissue areas, and did not explore porosity variability at all within secondarily remodeled bone regions. Our study clearly demonstrates the need to study these variations further in a quantitative manner. As our imaging methods did not allow for automated detection of pore spaces in bone, additional methods will be employed in the future to allow for such analyses, including the use of microradiography and MicroCT

Implications of research

Our study provides a significant advance in our understanding of the complexities of tissue level growth dynamics in the cortical bone of the human femoral diaphysis. The sample provides a rare opportunity to study the effects of childhood bone modeling on bone microstructural organization that may have effects on bone's mechanical properties well into adulthood. We are able to document both consistencies and variability in drift patterns and tissue type distribution within age classes that have previously been unknown, and demonstrate their relationship to changing geometric properties of the mid-shaft femur through ontogeny. The hypotheses generated by this study will contribute to more focused analyses of tissue level details (e.g. porosity distributions within and between tissue types; secondary bone distributions) using larger samples in future. The trends highlighted in the current study provide important baseline information that can be used in future comparative studies of bone growth in normal and abnormal (e.g., nutritionally or diseased stressed) individuals, important in both archaeological and orthopedic contexts

Acknowledgements

We thank the mortuary staff and the staff of the Donor Tissue Bank at the VIFM for their assistance in the collection of the bone specimens; Mr. Damian Lauder, Mount Sinai School of Medicine for specimen embedding; and Ms. Karen Rudo and Ms. Barbara Robinson for assistance with specimen preparation and imaging. This study was inspired by the life's work of Dr. Don Enlow, and is an extension of the pilot work presented at the 2006 Donald H. Enlow International Research Symposium: An Integrative Approach to Skeletal Biology at the New York University College of Dentistry.

Ancillary