Development of Cortical Bone Geometry in the Human Femoral and Tibial Diaphysis

Authors

  • James H. Gosman,

    Corresponding author
    • Department of Anthropology, The Ohio State University, Columbus, Ohio
    Search for more papers by this author
    • James H. Gosman and Timothy M. Ryan contributed equally to this work.

  • Zachariah R. Hubbell,

    1. Department of Anthropology, The Ohio State University, Columbus, Ohio
    Search for more papers by this author
  • Colin N. Shaw,

    1. McDonald Institute for Archaeological Research, Cambridge University, Cambridge, UK
    Search for more papers by this author
  • Timothy M. Ryan

    1. Department of Anthropology, Pennsylvania State University, University Park, Pennsylvania
    2. Center for Quantitative X-Ray Imaging, EMS Energy Institute, Pennsylvania State University, University Park, Pennsylvania
    Search for more papers by this author
    • James H. Gosman and Timothy M. Ryan contributed equally to this work.


Correspondence to: James H. Gosman, Department of Anthropology, The Ohio State University, 4052 Smith Laboratory, 174 W. 18th Street, Columbus, OH 43210. Fax: 419-841-1018. E-mail: jgosman@buckeye-express.com

ABSTRACT

Ontogenetic growth processes in human long bones are key elements, determining the variability of adult bone structure. This study seeks to identify and describe the interaction between ontogenetic growth periods and changes in femoral and tibial diaphyseal shape. Femora and tibiae (n = 46) ranging developmentally from neonate to skeletally mature were obtained from the Norris Farms No. 36 archeological skeletal series. High-resolution X-ray computed tomography scans were collected. Whole-diaphysis cortical bone drift patterns and relative bone envelope modeling activity across ages were assessed in five cross-sections per bone (total bone length: 20%, 35%, 50%, 65%, and 80%) by measuring the distance from the section centroid to the endosteal and periosteal margins in eight sectors using ImageJ. Pearson correlations were performed to document and interpret the relationship between the cross-sectional shape (Imax/Imin), total subperiosteal area, cortical area, and medullary cavity area for each slice location and age for both the femur and the tibia. Differences in cross-sectional shape between age groups at each cross-sectional position were assessed using nonparametric Mann-Whitney U tests. The data reveal that the femoral and tibial midshaft shape are relatively conserved throughout growth; yet, conversely, the proximal and distal femoral diaphysis and proximal tibial diaphysis appear more sensitive to developmentally induced changes in mechanical loading. Two time periods of accelerated change are identified: early childhood and prepuberty/adolescence. Anat Rec, 296:774–787, 2013. © 2013 Wiley Periodicals, Inc.

Growth-related changes in human long bone morphology and biomechanical features are key elements for understanding the variability and functional significance of adult bone structure (Smith and Buschang, 2004; Ruff, 2005). The heterogeneity of long bone diaphyseal shape and size variation occurs during ontogeny and is modulated by mechanical, nutritional, hormonal, and genetic patterning (Gosman et al., 2011). Modeling processes on the periosteal and endosteal surfaces are important contributors to this site-specific variability and provide a means of understanding developmental changes in long bone diaphyseal shape (Goldman et al., 2009; Gosman et al., 2011). Here, we advance the study of bone growth processes by documenting the interactions between ontogenetic growth periods and changes in femoral and tibial cross-sectional shape along the diaphysis.

Bone modeling is a tissue level process essential to skeletal size and shape changes, occurring predominantly during growth and development. Enlow (1963) and later Frost (1982) are credited with describing this growth-related pattern of independent bone formation and resorption on different, but frequently opposite bone surfaces (e.g., periosteal formation and endosteal resorption). Cortical drift, the changing spatial pattern of bone distribution on the periosteal or endosteal surfaces, is an outcome of the modeling process (Bromage, 1989; Mowbray, 2005). It is characterized by growth-related alterations in whole-bone shape and curvature, as well as heterogeneity within and among cross-sectional slices. The cross-sectional shapes of the femoral and tibial midshaft are altered from a relatively circular shape in early childhood, to a less uniform structure (strengthened anterolateral (AL)/posteromedial (PM) in the femur; anterior/posterior in the tibia) in early puberty (Hubbell et al., 2011). Goldman et al. (2009) suggest that the histological manifestations of drift patterns, identified as surface-specific bone resorption and formation, are evident at an earlier age than the measurable changes in the biomechanical properties of the cortical cross-sectional geometry. These patterns would indicate that early onset tissue microstructural properties interact with mechanical forces (i.e., body mass and locomotion) to establish adults' long bone geometric shape and biomechanical competence (Ural and Vashishth, 2006; Tommasini et al., 2007; Carpenter and Carter, 2008).

The ontogenetic pattern of surface-specific cortical bone geometric changes at defined anatomical locations has been well described in longitudinal studies of metacarpal radiographs (Garn, 1970), archaeological samples using computed tomography (CT) scans (midshaft femur) (Ruff et al., 1994), and recent clinical studies using peripheral quantitative CT (e.g., midshaft femur, 60% tibia; Macdonald et al., 2006). The accepted growth pattern based on a large volume of literature is net periosteal apposition and endosteal resorption throughout growth, with a relative endosteal contraction in postmenarchal girls. Several recent studies have questioned the evidence for endosteal deposition/contraction in adolescent girls, suggesting that further study may be needed (Kontulainen et al., 2005; Goldman et al., 2009).

Recent studies have demonstrated the importance of the relationship between fetal, childhood, and adolescent bone accrual and later bone loss, and the early development and later loss of biomechanical competence (Filardi et al., 2004; Oliver et al., 2007). The role of mechanical stimuli to femoral and tibial cross-sectional geometric properties in subadults has been demonstrated using archaeological samples (Ruff et al., 1994; Sumner and Andriacchi, 1996; Cowgill et al., 2010) as well as in modern clinical contexts (Macdonald et al., 2006; Kontulainen et al., 2007; Högler et al., 2008). The commonality of these studies using cross-sectional geometry is that, in general, they focus on a single location on the long bone for analysis—by convention the midshaft (Ruff, 1987; Wescott, 2005; Shaw and Ryan, 2012). We argue that the quantification of adult skeletal morphology will be enhanced through investigation of bone growth dynamics along the full long bone diaphysis, allowing a more complete view of bone development encompassing bone surface- and location-specific size and shape changes.

The importance of a whole-bone perspective on diaphyseal growth dynamics is supported by the published data on the distal femoral diaphysis and metaphysis, documenting the development of features such as the femoral bicondylar angle (Shefelbine et al., 2002) and sex differences in human lower limb structure (Ruff, 1987). Recent histological (Goldman et al., 2009), imaging (Hubbell et al., 2011; Cambra-Moo et al., 2012), and morphometric analyses (Morimoto et al., 2011) indicate growth-related regional heterogeneity in the micro- and macrostructural organization of cortical bone. These same authors demonstrate anatomical specificity in changes to bone shape and size in regions beyond the midshaft. A spatially enhanced perspective on bone development provides an opportunity to understand the processes involved in cortical bone structural differentiation and to elaborate on the patterns of diaphyseal bone adaptation during growth.

The goal of this study is to document the ontogenetic patterns of change in cortical bone cross-sectional geometry in the tibial and femoral diaphyses. Biomechanically relevant ontogenetic patterns of bone shape and size are quantified along the entire diaphysis in an envelope-specific sector analysis of the endosteal, endocortical, and periosteal surfaces in an effort to provide new data applicable to the study of long bone development. The goal of this study is not to merely create a detailed atlas of bone growth patterns, but to add to the understanding of the developmental origins of those patterns.

MATERIALS AND METHODS

The femur and tibia from a sample of 46 individuals ranging in age at death from neonate to 18 years were obtained from the Norris Farms #36 skeletal series, a collection of skeletons recovered from an Oneota Native American cemetery dating to approximately A.D. 1300 (Santure et al., 1990). Age estimation for all individuals in the Norris Farms No. 36 skeletal series was performed on the basis of dental development and epiphyseal closure (Milner and Smith, 1990). The sample was organized into five age groups for the ontogenetic analyses (Table 1).

Table 1. Composition of the skeletal sample used divided into age groups
Age groupAge rangen
10–1.921
22–4.911
35–8.94
49–13.94
514–17.96

Each bone was imaged using the OMNI-X High Resolution X-ray CT (HRCT) scanner at the Penn State Center for Quantitative Imaging, with voxel sizes ranging from 0.013 to 0.094 mm depending on the size of the specimen. Each specimen was mounted in foam inside a thin-walled plastic tube. Transverse cross-sections were collected for the entire bone with energy settings of 180 kV/0.11 mA, 2,800 projections, and a Feldkamp reconstruction algorithm. Image data were reconstructed as 1,024 × 1,024 16-bit TIFF images. The resulting data sets included between 860 and 3,066 slices for each bone.

Five cross-sections from each bone were used for the analyses of cross-sectional geometry and cortical drift patterns. The sections were located in the diaphysis at 20%, 35%, 50%, 65%, and 80% of total bone length beginning distally by convention (Fig. 1). The section at 80% of total bone length intersected the lesser trochanter in the majority of femora in the sample. At this location, much of the cortical wall thins and is replaced by trabecular bone, rendering such slices unsuitable for cross-sectional analysis of cortical bone properties. Consequently, those 80% length femoral slices intersecting the lesser trochanter were manually shifted several slices distally to the most proximal subtrochanteric region of the diaphysis. The mean adjusted slice position was 78% of full length; no slice was lower than 74%.

Figure 1.

Location of cross-sections along the diaphysis for femur and tibia.

Most individuals in the sample lacked fused epiphyses which made direct measurement of total bone length, and the subsequent identification of each section location (total bone length: 20%, 35%, 50%, 65%, and 80%), difficult. For these cases, a technique was developed to estimate the appropriate cross-sectional locations on each bone that matched the section locations as a percentage of total bone length. These adjusted diaphyseal cross-section locations were determined by measuring the length of the diaphysis plus metaphyses on all skeletal elements in the sample that included epiphyses. These measured diaphyseal/metaphyseal lengths were then used to calculate the median value for each slice location relative to the total length of the diaphysis plus the proximal and distal metaphyses. These adjusted section locations were then used for all bones in the sample lacking fused epiphyses. This protocol resulted in the following adjusted whole-bone slice locations for bones lacking epiphyses: femur (20% of total bone length = 15.60% of diaphyseal/metaphyseal length, 35% = 31.92%, 50% = 48.25%, 65% = 64.57%, and 80% = 80.90%) and the tibia (20% = 16.39%, 35% = 33.13%, 50% = 49.85%, 65% = 66.51%, and 80% = 83.17%). The adjustment of each cross-section location on elements lacking epiphyses ensured that all quantitative analyses were performed on homologous cross-sections whether the bones were complete or not.

The cross-sectional images from each bone were imported into ImageJ (http://imagej.nih.gov/ij/) for quantification of the second moments of inertia using the BoneJ plugin (Doube et al., 2010). The cortical bone of each cross-section was segmented using the iterative thresholding algorithm of Ridler and Calvard (1978) and Trussell (1979) as implemented in ImageJ. The maximum (Imax) and minimum (Imin) second moments of inertia and cortical bone area (CA) were calculated for each slice. Total subperiosteal area (TA) was calculated by filling the medullary cavity of the thresholded cross-section using the “Fill Holes” command in ImageJ to make a solid cross-section. The area of this filled cross-section was calculated and represents TA. The medullary area (MA) was derived from the CA and TA (MA = TA − CA). Cross-sectional shape was characterized using the ratio of Imax/Imin. Whole-diaphysis cortical drift patterns across ages were assessed by taking radial measurements of bone cross-sections from the center of each of five diaphyseal regions. The distance from the section centroid out to the endosteal and periosteal margins was measured on the thresholded cross-sectional images at the center of eight sectors around the cross-section: anterior (A), anteromedial (AM), medial (M), PM, posterior (P), posterolateral (PL), lateral (L), and AL by following the methodology outlined in Goldman et al. (2009) (Fig. 2). The difference between the periosteal and the endosteal measurements at each sector was defined as the cortical thickness at that sector position.

Figure 2.

Radial sectors used on each cross-section were used to measure distance from centroid to endosteal and periosteal surfaces. A, anterior; AM, anteromedial; M, medial; PM, posteromedial; P, posterior; PL, posterolateral; L, lateral; AL, anterolateral.

Pearson correlations were performed to test the relationship between the cross-sectional shape (Imax/Imin), TA, CA, and MA for each slice location and age for both the femur and the tibia. Differences in cross-sectional shape among age groups at each cross-sectional position along the diaphysis (20%, 35%, 50%, 65%, and 80%) were assessed using nonparametric Mann-Whitney U tests. Differences in cortical thickness at each of the eight sector locations (anterior, AM, medial, PM, posterior, PL, lateral, and AL) on each cross-section were compared between successive ages. In addition to comparisons of thickness changes at each sector location, length differences for the periosteal and endosteal rays measured at each sector position were compared between successive age groups. Statistical tests were evaluated with and without the Bonferroni correction to compensate for multiple comparisons.

RESULTS

Diaphyseal Cross-Sectional Shape

Cross-sectional morphometric data for the femur and tibia across all five section locations are summarized in Table 2 and shown in Figs. 3 and 4. The ontogenetic pattern of Imax/Imin values for the femoral diaphysis indicates that the cross-sectional shape undergoes regionally specific changes across age groups (Fig. 5). A general pattern of increasing circularity (Imax/Imin close to 1) is observed with age at the 20% cross-section. The proximal aspect of the femur (65%–80%) becomes relatively more asymmetric in cross-section with increasing age. Statistically significant relationships between diaphyseal shape and age were found only at 20% and 65%, following the Bonferroni correction (Table 3). Results of pairwise comparisons of femoral cross-sectional shape at each diaphyseal section between each age group are summarized in Table 4. There are few significant differences in femoral shape at any section position between successive age groups after the Bonferroni correction.

Table 2. Summary statistics (mean and standard deviation) for Imax/Imin
 FemurTibia
Age group20%35%50%65%80%20%35%50%65%80%
12.261.441.291.321.331.401.241.321.321.45
0.340.210.150.160.170.180.120.120.110.16
22.231.431.291.441.291.251.381.431.421.55
0.360.150.130.190.190.140.150.210.190.11
32.011.281.191.601.481.301.621.691.601.66
0.260.080.080.260.270.110.250.360.220.11
41.501.121.111.791.561.261.722.021.962.02
0.180.020.040.250.310.080.070.180.240.12
51.521.291.371.671.551.181.832.302.372.20
0.180.170.250.220.380.100.230.490.360.26
Figure 3.

Representative age group cross-sections at each femoral section location.

Table 3. Pearson correlation results for Imax/Imin and agea
 FemurTibia
 20%35%50%65%80%20%35%50%65%80%
  1. a

    Comparisons with significant P-values following the Bonferroni correction are indicated in bold font (corrected P = 0.005).

r−0.665−0.3580.0380.5990.365−0.4170.7920.8250.8800.859
n47474747474747474747
P-value<0.001<0.01NS<0.001<0.01<0.01<0.001<0.001<0.001<0.001
Figure 4.

Representative age group cross-sections at each tibial section location.

Table 4. Results of Mann-Whitney U tests comparing Imax/Imin between age groups at each section level in the femura
Age groups1234
  1. a

    Numbers in cells denote sections at which shape differences are significant at P < 0.05. No comparisons were significant following the Bonferroni correction.

2NS   
365%NS  
420%, 35%, 50%, 65%20%, 35%, 50%, 65%20%, 35% 
520%, 65%20%20%35%, 50%
Figure 5.

Age group mean Imax/Imin values versus percent bone length for femur and tibia.

The age-specific distribution of Imax/Imin values indicates that tibial cross-sectional shape follows a statistically significant pattern of morphological change, from more circular to more A-P strengthened, with increasing age. This is most obvious in the mid-distal to proximal portion of the diaphysis (locations, 35%–80%) (Table 2 and Fig. 5). Tibial diaphysis shape is essentially circular in individuals younger than 5 years of age (Table 3 and Fig. 6). The Imax/Imin indices increase markedly in all cross-sections from approximately 9 years of age onward, with the exception of the 20% distal-most cross-section which remains close to circular. Results of pairwise comparisons of tibial cross-sectional shape at each diaphyseal cross-section confirm the general pattern of gradual shape change with age although there are few significant differences following the Bonferroni correction (Table 5).

Table 5. Results of Mann-Whitney U tests comparing Imax/Imin between age groups at each section level in the tibiaa
Age groups1234
  1. a

    Numbers in cells denote sections at which shape differences are significant at P < 0.05. Comparisons that were significant following the Bonferroni correction are shown in bold font (P < 0.0005).

220%, 35%, 80%   
335%, 50%, 65%, 80%NS  
435%, 50%, 65%, 80%35%, 50%, 65%, 80%T80 
520%, 35%, 50%, 65%, 80%35%, 50%, 65%, 80%65%, 80%NS
Figure 6.

Imax/Imin versus age at each section location.

Subperiosteal Area

TA, CA, and medullary cavity area (MA) are all significantly correlated with age at all cross-sections from both femur and tibia (Table 6 and Fig. 7). All three area measures generally show a gradual increase with age with similar LOESS regression curves early in development, indicating roughly equal rates of periosteal and endosteal expansion in most sections of both the femur and the tibia. CA in most sections is higher than MA except in the proximal and distal most sections of each bone. In several locations (femur: 20%, 35%, 80%; tibia: 65%, 80%), LOESS regression lines suggest a slight increase of the rate of periosteal expansion and a consequent slowing of the rate of endosteal resorption after 10 years of age. This shift in periosteal and endosteal growth does not occur uniformly across all sections of the femoral or tibial diaphyses. The tibial 20% and 65% sections have similar CA and MA growth rates even after 10 years of age.

Table 6. Pearson correlations between CA, MA, and TA and age at each section for both femur and tibiaa
Section (%)VariableFemurP-valueTibiaP-value
  1. a

    All comparisons are significant following the Bonferroni correction at P < 0.0017.

20CA0.819<0.0010.955<0.001
 MA0.581<0.0010.832<0.001
 TA0.892<0.0010.921<0.001
35CA0.944<0.0010.947<0.001
 MA0.760<0.0010.802<0.001
 TA0.918<0.0010.932<0.001
50CA0.932<0.0010.955<0.001
 MA0.705<0.0010.837<0.001
 TA0.942<0.0010.943<0.001
65CA0.930<0.0010.955<0.001
 MA0.736<0.0010.863<0.001
 TA0.949<0.0010.942<0.001
80CA0.857<0.0010.907<0.001
 MA0.886<0.0010.851<0.001
 TA0.909<0.0010.949<0.001
Figure 7.

Plots of cortical area, MA, and TA versus age at each cross-section for all individuals in the sample. LOESS lines are plotted for each variable.

Cortical Surface and Thickness Changes

Radial graphs representing the mean distance of the endosteal and periosteal bone surfaces from the section centroids for the different age groups, sorted by cross-section location, are shown in Fig. 8. This visualization of quantitative data displays the patterns of ontogenetic change in cortical shape, size, and thickness. The age-related pattern of change for the femoral diaphysis is generally characterized by an alteration of the cross-sectional shape from an essentially circular cross-section in young individuals to a more mediolaterally directed cross-section proximally and an AL-/PM-directed shape in the middle third of the diaphysis. These changes are combined with expansion of the bone external diameters both proximally and distally, and a reconfiguration of the relative cortical thickness along the shaft (proximal > distal). The tibial diaphyseal cross-sectional shape develops from a primarily round cross-section in the youngest age group to a more anteroposteriorly directed cross-section in older individuals. These changes are most evident in the more proximal sections down to 35% of bone length.

Figure 8.

Filled radial plots of periosteal and endosteal surfaces at each section for the femur and tibia.

The age-related patterns of change in periosteal/endosteal bone formation and bone resorption are quantified as percentage changes between age groups in a series of bar graphs (Figs. 9 and 10). Sector-specific surface differentiation for the femur is especially accentuated for all slices between age groups of 1–2 and 4–5. The tibial data indicate that changes in the periosteal and endosteal surfaces are most distinct between age groups 1 and 2, very low from age groups of 3–4 and then higher again from age groups of 4–5. These patterns of change in both femur and tibia are also evident from statistical analyses of periosteal and endosteal measurements at each sector location (Figs. 9 and 10). In both femur and tibia, the most noteworthy changes in measured distance from the centroid occur between age groups of 1–2 and 4–5 at nearly all section locations. However, only changes in the periosteal border between age groups of 1 and 2 are significant following the Bonferroni correction. In the femur, there are very few significant length differences at any section between age groups of 2 and 3 or between age groups of 3 and 4 (Fig. 9). The tibia, by contrast, has significant changes in the antero-posterior axis in the mid-diaphyseal region (35%, 50%, and 65%) at all ages although none are significant following the Bonferroni correction (Fig. 10).

Figure 9.

Percent change in periosteal (black) and endosteal (gray) surfaces at each sector location on each section for the femur. Symbols above columns indicate significant differences between age groups based on Mann-Whitney U tests with the Bonferroni corrections (P < 0.00007): *, periosteal; #, endosteal; †, cortical thickness.

Figure 10.

Percent change in periosteal (black) and endosteal (gray) surfaces at each sector location on each section for the tibia. Symbols above columns indicate significant differences between age groups based on Mann-Whitney U tests with the Bonferroni corrections (P < 0.00007): *, periosteal; #, endosteal; †, cortical thickness.

Cortical bone thickness changes in the femur display an asymmetric pattern across the diaphysis with increasing age (Fig. 9). The most significant changes in thickness are between age groups of 1 and 2, in nearly all sectors including the midshaft. The Bonferroni correction results in very few significant differences between age groups. The general pattern of change in the femur is comparable with that in tibia in which thickness also increases occur at an early age, namely between age groups of 1 and 2 at most section locations (Fig. 10). The Bonferroni correction eliminates most of the significance between age groups.

DISCUSSION

Cross-Sectional Shape Change

The composite results of this study provide structurally and functionally relevant quantitative data on the developmental patterns of human long bone form/function interactions along the whole diaphysis and within individual cross-sections. The results produce further insight into the development of long bone diaphyses and extend our understanding of bone growth in the human lower limb. These data demonstrate a developmental pattern for both the femoral and the tibial diaphyses that can be characterized as a trajectory from a relatively circular cross-sectional shape along the entire diaphysis in young individuals to a more highly variable, asymmetric cross-sectional shape in late adolescence. Both bones have unique developmental trajectories with significant variation along the diaphysis. Components of this dynamic developmental process include reshaping of the proximal, midshaft, and distal external diameters; reconfiguration of the relative cortical thickness (proximal > distal); and changes in the cross-sectional shape (Imax/Imin). This study highlights the dynamic continuity and change in ontogenetic morphology of the entire diaphysis in response to a genetic patterning framework, developmental factors, and biomechanical forces.

The regions of the lower limb with the most significant shape change during ontogeny appear to be the distal most aspect of the femur (location, 20%) and the proximal half of the tibia (50%–80%). The distal-most section of the femur analyzed here changes from a generally asymmetric cross-sectional shape to a more circular shape during ontogeny. This shape change in the femur reflects cortical bone expansion in the antero-posterior axis most likely related to developmental kinematics and bending forces about the knee during walking and running (Pauwels, 1980). The midshaft regions of the femoral diaphysis retain a generally circular cross-sectional shape although some sections along the shaft appear relatively asymmetrical in shape (Ruff et al., 1994; Cowgill et al., 2010).

The tibial diaphysis, distinct from the femur, changes during ontogeny from an almost uniformly rounded shape along the whole diaphysis to a very highly asymmetric cross-section by young adulthood. This pattern is especially evident in the proximal half of the tibia in which the Imax/Imin ratio changes between 1 and 1.5 (i.e., relatively rounded) to values close to 3.0 in some individuals. This growth trajectory reflects the progressive development of the characteristic triangular cross-sectional shape of the tibia with strong growth in the antero-posterior plane. The distal most section of the tibia analyzed here (20%) is highly conserved, and retains a relatively circular shape throughout ontogeny.

Cortical Surfaces

Long bone diaphyses have four surfaces on which modeling and remodeling occur, namely periosteal, intracortical, endosteal, and trabecular (Gosman et al., 2011). The modeling process, which predominates during growth and development, involves the independent actions of bone formation and bone resorption on different surfaces, whereas the remodeling process involves the sequentially synchronized and coupled actions of bone resorption and bone formation on the same surface. Here, we show the age-associated regional effects of the modeling process involving periosteal bone formation and endosteal formation and/or resorption on the ontogenetic pattern of change in cortical thickness, drift, and cross-sectional shape. TA, MA, and cortical area (Fig. 7) follow trends previously described in the literature, namely net periosteal apposition and net endosteal resorption (Ruff et al., 1994; Goldman et al., 2009; Cambra-Moo et al., 2012). The data presented here, however, do not show the decrease in TA, CA, and MA in late adolescence found by Goldman et al. (Goldman et al., 2009). As an additional contrast, our data do not demonstrate the medullary contraction from net endosteal bone formation in late adolescence that many studies have reported (Garn, 1970; Macdonald et al., 2006; Chevalley et al., 2008). It is possible that the sex-pooled characteristics of our sample mask this effect. However, the absence of a sex difference in the relative degree of endosteal apposition has also been noted by recent authors (Kontulainen et al., 2005; Goldman et al., 2009)—this remains an area for further investigation.

The ontogenetic patterns of cortical drift are responsible for the region- and sector-specific changes in diaphyseal geometric shape. Our data indicate that the childhood drift pattern for the femur is variable, but generally posteriorly directed. The adult femoral configuration of A-P or AL-PL Imax directionality is reached by early adolescence. This is in line with histological data and thought to be, in part, derived from adductor and hamstring muscle attachment forces at the emerging linea aspera as well as locomotion-derived biomechanical strains (Mittlmeier et al., 1994; Epker and Frost, 2005; Goldman et al., 2009).

The age-related drift pattern for the tibia appears earlier in development compared to that of the femur. It also differs from the pattern in the femur in that the Imax of the tibia is more stereotypically in the A-P direction. Carpenter and Carter (2008) have developed a computer simulation model describing age-associated tibial morphology and functional adaptation, which accounts for the development of the triangular cross-sectional shape of this bone. In their model, pressure on the periosteal surface can impede bone formation and induce resorption, and periosteal tension strains perpendicular to bone surfaces can impede bone resorption and stimulate formation. The authors indicate that bone cross-sectional size (cortical area) may be related to the more distant mechanical environment (i.e., body mass and muscle force-induced strain), and the bone cross-sectional shape is influenced by local periosteal loads. Our age-related shape changes appear congruous with these researchers' simulations (Fig. 5).

Two periods of significant change in cross-sectional size and shape are noted from our results. The first occurs between our age group 1 (0–1.9 years) and age group 2 (2–4.9 years). During this time period, the size of both the femoral and the tibial cross-sections and the shape of the tibial diaphyseal cross-sections are undergoing change. The timing of these changes is further evidence (Ruff, 2003; Ryan and Krovitz, 2006; Gosman and Ketcham, 2009) that the acquisition of bipedal walking, and the associated changes in load type and magnitude on the lower limb, have significant impacts on the development of the bone morphology in early childhood. The second period of notable size and shape change is between age groups 4 (9–13.9 years) and 5 (14–17.9 years), reflecting, in part, the prepubertal (juvenile) hormonal and body mass-related increase in periosteal expansion and concurrent endosteal reshaping. These changes continue throughout adolescence into early skeletal maturity.

Significance

The significance of the ontogenetic patterns found in this study may be evaluated by situating them within two complementary frameworks: (1) a biomechanical/behavioral perspective and (2) an evolutionarily informed view of skeletal development focusing on the juvenile transition period of human life history. The complex linkages between adult cortical bone morphology and biomechanical forces are well established in the bone science and anthropological literature (Martin et al., 1998; Ruff and Hayes, 2005; Ruff et al., 2006; Shaw and Stock, 2009). However, relatively few researchers have focused on the importance of ontogenetic patterns of change of lower limb long bone cortical geometry, and how these patterns may be used to reconstruct behavioral differences and/or the attainment of locomotor skills (Sumner and Andriacchi, 1996; Cowgill et al., 2010; Hubbell et al., 2011). These concerns were articulated by Ruff et al. (2006) in their discussion of the greater effects of mechanical loading on cortical bone geometry in younger subadults, and how the time frame of later adolescence to early adulthood may be optimal for comparative research on life-style differences between populations.

Experimental models and anthropological observational research have studied the effects of strain distribution as well as possible confounding genetic influences on cortical cross-sectional geometry. Many researchers have pointed out the need for caution in inferring specific behavior, including locomotor skills, from midshaft cross-sectional geometry properties alone (Demes et al., 2001; Lieberman et al., 2004; Wallace et al., 2010). Our data generated through the whole diaphyseal perspective offer an opportunity to contribute to this discourse on the relationship of developmental changes in bone morphology to adult functional adaptation. In this study, we demonstrate that the midshaft cortical bone shape of the femur remains relatively constant throughout development. That is, this region is the most conserved and canalized during ontogeny compared to the higher degree of developmental variability in the proximal and distal regions of the femoral diaphysis. The tibia displays a different pattern with high cross-sectional shape variability proximally and less change distally during development.

From these observations, two ideas emerge that may contribute to enhancing the usefulness of cross-sectional geometry in bioanthropological research. First, these data have the potential to provide increased resolution for population-level comparisons and general behavioral inferences from midshaft cortical bone cross-sectional geometry when they are supported by high-quality data (including accounting for possible genetic differences). The concept of a modeling threshold provides an explanatory framework for utilizing these midshaft data in the interpretation of physical activity and mobility. According to Frost (1997), developmental bone strength and mass begin to increase when bone strains exceed a threshold range. When strains remain below this threshold, modeling ceases. This response has anatomical specificity in regards to bony element, section, and surface. The midshaft of the tibia and femur may represent “high threshold” regions requiring relatively more strain to trigger a modeling response. When this concept is applied to our Norris Farm sample of sedentary farmers, one could argue that the midshaft strains remain relative low in the biomechanically adapted long bones. This would allow the hip and knee joint effects to predominate, with little change evident in the mechanically adequate midshaft. Although our data indicate a relative conservation of midshaft cortical geometry throughout development, we argue that this enhances the biomechanical/behavioral/cortical-shape interpretation so well demonstrated in the Anthropology literature. This “enhancement” is perhaps best summarized as a midshaft cortical shape response to biomechanical forces in a region with a high modeling threshold, and thus adding additional significance (when present) to midshaft cortical geometry comparisons associated with behavioral differences. Second, use of cross-sectional analyses in diaphyseal regions other than the midshaft (e.g., distal femur and proximal tibia), especially when the midshaft data are inconclusive, may expand the potential of this analytical tool for biomechanical analysis (Ruff, 1987; Morimoto et al., 2011). Comparative ontogenetic projects, including hunter-gatherer and agricultural groups, are currently underway, examining both of these perspectives.

One of the striking aspects of the ontogenetic patterns demonstrated by our data is the acceleration of the rate and variability of regionally specific cortical shape and size change beginning in the late childhood period (age group 4) of growth and development—the juvenile transition (Bogin, 1999). This is a watershed life-history stage of human development with broad behavioral, morphological, and physiological implications. The skeletally relevant aspects of this ontogenetic phenotypic switch include hormonal change (adrenarche), an increase in body mass, neuromuscular adaptation characterized by an increase in the mechanical loading of the lower limb bones by muscle forces, and an increase in biological plasticity related to gene–environmental interactions (Del Giudice et al., 2009). The comprehensive ontogenetic whole-diaphyseal data set used in this analysis provides a portal for future research of these types of considerations, which place skeletal development within a larger evolutionary informed framework. This framework includes the examination of modulation in developmental bone (re)modeling processes by changes in mechanical load, inflammation, immune system, reactive oxygen species, nutrition, and hormonal status.

Limitations

This study has several limitations that constrain the extent of the conclusions that can be drawn. First, the cross-sectional and sex-pooled characteristics of the skeletal sample allow only associations to be developed; these data cannot establish causal relationships. The section- and sector-specific bone envelope change data are likely influenced to some degree by the inherent developmental variability of the sample. Second, the age group of 3–5 years is represented by a relatively small sample size, necessitating caution in the interpretation of these data. Taken together, however, the overall trends in cortical drift and modeling are consistent with the published data using other methods (Goldman et al., 2009; Cambra-Moo et al., 2012).

In spite of these limitations, the approach employed here has several important strengths, and the results provide novel insight into long bone growth during ontogeny. Quantitative data along the diaphysis and within specific cross-sections can account, in part, for the complexity of influences on regional cross-sectional geometry in ways that studies focused only on the midshaft cannot (Ruff, 1987; Wallace et al., 2010). This ex vivo whole-diaphyseal perspective supports the established concepts that developmental heterogeneity in bone morphology is a reflection of dynamically changing patterns of growth (Carlson et al., 2008a,b), spatially specific bone–muscle interactions (Carpenter and Carter, 2008), complicated locomotor biomechanical forces (Lieberman et al., 2004), and anatomically differentiated (re)modeling rates (Bass et al., 2002).

The whole-diaphysis data presented here are consistent with those from previously published studies using cross-sectional geometric analyses from selected regions of the diaphysis (Ruff et al., 1994; Sumner and Andriacchi, 1996; Goldman et al., 2009; Cowgill et al., 2010). Morimoto et al. (2011) used morphometric mapping techniques to analyze ontogenetic femoral diaphyseal shape variation in wild and captive chimpanzees. They concluded from their data that activity inferences may be best derived from morphological variation in the distal diaphysis (femur) and not the midshaft. This echoes Ruff's (1987) assertion that differences in activity patterns may be interpreted by cross-sectional geometry from the region encompassing the distal half of the femur to the proximal half of the tibia.

CONCLUSIONS

In this study, we use high-resolution CT imaging to examine the regional variability of bone surfaces in an ontogenetic series of human femora and tibiae. This study combines geometric and bone surface-specific data to address the question of whether or not cortical cross-sectional geometry demonstrates a preferential periosteal, endocortical, and endosteal distribution during growth and development. These data have allowed us to identify and quantify shape, size, and surface changes within- and among-age classes. Four main patterns emerge from the study. The first two, medio-lateral proximal femur and antero-posterior tibial expansions are consistent with the previous studies, pointing to hip breadth effects and locomotor loads, respectively. The other two, distal femur antero-posterior expansion and lack of change in the midshaft femur, are consistent with a regional heterogeneity in the sensitivity to mechanical loads. Developmental cortical surface differentiation defined as periosteal and endosteal bone formation/resorption is age and location specific. Two time periods of accelerated change have been identified: early childhood and prepuberty/adolescence.

The significance of this investigation into the developmental trajectory of cortical bone geometry is cross-disciplinary, contributing new data in regards to ontogenetic bone accrual and structural characteristics important in considerations of fracture risk, early gain and later bone loss, and concerns of bone functional adaptation including assessment of physical activity/behavior in past human groups. Continued investigation of the age-related anatomical variation from a surface- and region-specific perspective is expected to further define the spatial characteristics and regional significance of diaphyseal cortical bone accrual during growth as well as the potential factors that influence the shift toward mature cortical bone structural configuration. The data from this study will contribute to unraveling the combined effects of genetic patterning, linear growth, and epigenetic factors including the mechanical/locomotor influences on developmental and adult bone morphology.

ACKNOWLEDGEMENTS

The authors thank George Milner at Pennsylvania State University and Terrance Martin at the Illinois State Museum for their willingness to loan specimens for scanning. The authors thank Clark S. Larsen, whose helpful comments improved the manuscript. The authors also thank T. Stecko, L. Souza, and S. Sukhdeo for assistance with scan data collection and image processing.

Ancillary