Skip to main content
NCBI home page
Journal of Anatomy logoLink to Journal of Anatomy
. 2015 Dec 28;228(5):733–745. doi: 10.1111/joa.12433

Spatial variation in osteon population density at the human femoral midshaft: histomorphometric adaptations to habitual load environment

Timothy P Gocha 1,✉, Amanda M Agnew 1,2
PMCID: PMC4831343  PMID: 26708961

Abstract

Intracortical remodeling, and the osteons it produces, is one aspect of the bone microstructure that is influenced by and, in turn, can influence its mechanical properties. Previous research examining the spatial distribution of intracortical remodeling density across the femoral midshaft has been limited to either considering only small regions of the cortex or, when looking at the entirety of the cortex, considering only a single individual. This study examined the spatial distribution of all remodeling events (intact osteons, fragmentary osteons, and resorptive bays) across the entirety of the femoral midshaft in a sample of 30 modern cadaveric donors. The sample consisted of 15 males and 15 females, aged 21–97 years at time of death. Using geographic information systems software, the femoral cortex was subdivided radially into thirds and circumferentially into octants, and the spatial location of all remodeling events was marked. Density maps and calculation of osteon population density in cortical regions of interest revealed that remodeling density is typically highest in the periosteal third of the bone, particularly in the lateral and anterolateral regions of the cortex. Due to modeling drift, this area of the midshaft femur has some of the youngest primary tissue, which consequently reveals that the lateral and anterolateral regions of the femoral midshaft have higher remodeling rates than elsewhere in the cortex. This is likely the result of tension/shear forces and/or greater strain magnitudes acting upon the anterolateral femur, which results in a greater amount of microdamage in need of repair than is seen in the medial and posterior regions of the femoral midshaft, which are more subject to compressive forces and/or lesser strain magnitudes.

Keywords: biomechanical loading, bone microstructure, bone remodeling, geographic information systems, human femur, osteon population density

Introduction

The spatial organization of bone microstructure is influenced by and also influences the way bone responds to the biomechanical forces placed upon it. Of particular interest to some researchers are microstructural properties at the midshaft of the human femur, due to its importance in obligate bipedalism. Numerous microstructural variables of human bone have been quantified and their spatial distribution examined in the human femoral midshaft, including distribution of collagen fiber orientation (Goldman et al. 2003a, 2005), mineralization density (Goldman et al. 2003b, 2005), and intracortical porosity (Thomas et al. 2005). However, despite secondary osteons being considered the fundamental unit of compact bone (Enlow, 1962), the spatial distribution of these remodeling events has only been examined in non‐human taxa (e.g. McFarlin et al. 2008; Zedda et al. 2015), and there has not yet been any systematic treatise of this variable in the human femoral midshaft.

Skeletal remodeling involves removal of old bone tissue and replacement with newly formed bone. It is a complex sequence of events that aids the body in altering and maintaining a proper mineral balance, provides a mechanism for repair of microdamage, and helps the skeleton to better adapt functionally to its mechanical environment (Martin et al. 1998). Remodeling can be categorized as either targeted or nontargeted (Martin et al. 1998). As the name implies, targeted remodeling is caused by a specific local signaling event that begins the activation process of remodeling. The two most common such local events are microdamage and/or osteocyte apoptosis (Allen & Burr, 2014). Because nontargeted remodeling is not location‐specific, this stochastic remodeling is associated with metabolic purposes, such as aiding mineral homeostasis (Martin, 2002). Burr (2002) has suggested that 30% of all remodeling is targeted to repair microdamage. However, continued evidence that osteocyte apoptosis is not just one way to initiate remodeling, but perhaps the most common way, suggests that targeted remodeling is perhaps responsible for the majority of remodeling rather than a mere 30% (Allen & Burr, 2014). Further, theoretical models produced by Martin (2002) are consistent with the suggestion that, in the absence of trauma and disease, virtually all intracortical remodeling may be targeted at repairing microdamage.

An understanding of the spatial distribution of remodeling may then be useful to elucidate biological mechanisms of adaptation to applied forces in the femur. While intracortical remodeling in the human femoral midshaft has been studied numerous times by anthropologists, the goal of such research has usually been estimation of age at death. These research efforts only sampled small areas of the cortex and typically conflated data from separate regions of interest (ROIs) sampled (e.g. Kerley, 1965; Ericksen, 1991). An exception is work by Drapeau & Streeter (2006), who examined remodeling in several ROIs from the femur in relation to biomechanical loading and reported that remodeling is not uniformly distributed within the femoral midshaft. Portigliatti Barbos et al. (1983) observed osteon distribution across the entirety of the femoral cortex – the only such study to date – and found the greatest osteon density in the anterolateral cortex. However, that research was extremely limited in scope because the authors studied only three complete cross‐sections from a single individual.

The goal of the current study was to determine the distribution and spatial variation of all intracortical remodeling events (intact osteons, fragmentary osteons, and resorptive bays) across the entirety of the femoral midshaft. Osteon population density (OPD), a term coined by Frost (1987a,b), quantifies the total number of remodeling events per square millimeter. OPD in this study was measured in both sexes and across the adult human lifespan in a modern sample of cadaveric donors. Using geographic information systems (GIS) software, complete cross‐sections from the femoral midshaft were divided radially into periosteal, middle, and endosteal thirds, circumferentially divided into octants, as well as each octant separated into radial thirds. All remodeling events were annotated in GIS and their location within the cortex was mapped to appreciate spatial distribution of osteons and their density. Based on limited previous research (Portigliatti Barbos et al. 1983), we hypothesized that remodeling density would be greatest in the anterolateral regions of the human femoral midshaft.

Materials and methods

Sample materials

The skeletal material used in this study includes 30 complete cross‐sections taken from the femoral midshaft of modern human cadaveric donors. Twenty‐five of the samples came from the Ericksen femur collection, collected by Dr. M. F. Ericksen from 1972 to 1989 from dissecting room cadavers at George Washington University Medical School (Ericksen, 1991). The collection is now curated by Dr. Sam D. Stout at The Ohio State University. An additional five samples came from the Skeletal Biology Research Lab (SBRL) collection, which contains skeletal material extracted from cadavers received through the Division of Anatomy's Whole Body Donation Program at The Ohio State University. Criteria for including a femoral sample in this study were: (i) the complete cross‐section was present and undamaged; (ii) the sample was from an individual of known demographics (age, sex, and ancestry); (iii) orientation of the sample could be determined with certainty; and (iv) the sample displayed no gross evidence of disease, healed trauma or osteoporosis as indicated by excessive intracortical porosity.

The sample included 15 males and 15 females, ranging from 21 to 97 years (overall sample mean = 58.9 years, SD = 22.1 years). Both males and females had a nearly equal distribution across the adult age spectrum (male sample mean = 56.1 years, SD = 22.7 years; female sample mean = 61.8 years, SD = 21.9 years), with two males and one female aged 20–29 years, two males and females each per decade aged 30–89 years, and one male and two females aged 90–99 years. Causes of death for the sample, when they were known, included both acute/traumatic causes (e.g. suicide) as well as chronic conditions (e.g. cancer and heart disease). Of the individuals in this study, one was black and 29 were white.

Sample preparation and image acquisition

Samples were extracted as ~ 5 cm blocks of bone from the midshaft of the femur and were marked to maintain orientation. Bone blocks from the Ericksen collection were cleaned and degreased by unspecified means (Ericksen, 1991). Bone samples from the SBRL collection were macerated in a solution of warm water and Biz detergent, after which any remaining adherent soft tissue was manually removed and the medullary cavity thoroughly rinsed of any non‐mineralized material. Thick sections, approximately 1 mm thick and oriented transverse to the long axis of the bone, were isolated from the midshaft blocks using a Buehler Isomet 1000 Precision saw (Lake Bluff, IL, USA). Thin sections approximately 100 μm thick were produced by grinding thick sections on a Buehler variable speed grinder equipped with diamond grinding discs. Thin sections were cleaned with distilled water and, once dry, mounted using Permount mounting medium and coverslipped. Orientation was maintained during all slidemaking procedures.

Thin sections were imaged at 40× magnification (10× oculars, 4× objective) under linearly polarized light using an Olympus DP73 digital camera affixed to an Olympus BX63 microscope (Center Valley, PA, USA) equipped with an ultrasonic motorized stage. Images were acquired using olympus cellsens dimension software (v.1.7), which automatically assembled separate frames into a seamless montage representing the entire cross‐section of a sample. Resulting images were saved in a tagged image file format (TIFF) for use in other computer software programs. Polarized light was used instead of standard brightfield microscopy for static imaging because it allows better visualization of bone lamellar structure, which is of great utility in recognizing intact and fragmentary osteons by virtue of their concentric osteonal lamellae (Bromage et al. 2003; Skedros et al. 2009).

Image processing and data collection

Cross‐sectional TIFF images were imported into a new map document in arcgis (v. 10.1, ESRI) to serve as a baseline image from which all measurements and variables would be notated within the ArcMap interface of arcgis. GIS software is designed to store, manipulate, and analyze spatial data that are typically related to physical locations on the Earth's surface (Steinberg & Steinberg, 2006). However, recent research has demonstrated that data related to skeletal biology can be analyzed using GIS software because these data, although not tied to a particular location on earth, are inherently spatial. Rose et al. (2012) demonstrated that GIS could be used to study osteon morphotypes to investigate biomechanical loads experienced by human metatarsals during bipedal gait, and Cambra‐Moo et al. (2012) established its utility to map long bone compartmentalization during ontogeny.

Within arcgis, polygon feature classes were added as layers to the arcgis map for each sample, representing cortical, medullary, trabecular, and abnormally large resorption spaces (Fig. 1). Periosteal, endosteal, trabecular, and large porous space borders were manually defined within arcgis by an experienced observer (T.P.G.).

Figure 1.

Figure 1

Open in a new tab

Polygon features representing for an example femoral midshaft in digital space within arcgis. Anterior is at the top and medial to the right.

Trabeculae were defined by identifying struts or spicules of bone extending into the medullary area that interrupted continuity of the endosteal border and were usually devoid of discernable remodeling events (Agnew & Stout, 2012). Abnormally large resorption spaces were defined as any intracortical pore with an area > 0.1164 mm2. This threshold value was selected based on data presented by Bell et al. (2001), who found that pores greater than this size are unlikely to be Haversian canals or resorptive bays. Bell et al. report that pores of this size are often the result of several individual remodeling events coalescing to a point where their Haversian canals meet and form a structure termed a ‘super‐osteon’, although pores larger than this value also represent areas of age‐related intracortical bone loss. Since the goal of this research was to examine the number of remodeling events per square millimeter, it was essential to remove such large pores, whose inclusion would artificially inflate measures of cortical area.

The momentmacro (v. 1.3) plugin (Ruff) for imagej (v. 1.46, NIH) was used to determine the mathematical centroid for the cross‐section, defined as the intersection between principal bending axes. A spoke axis was then overlaid on the cross‐section, such that a line representing 0–180° ran along the anteroposterior axis of the section, with the posterior end of the spoke bisecting the linea aspera (Fig. 2, dashed line). These axes digitally subdivided the cortex into various ROIs, which were assessed as three segmentation methods: (i) radially the cortex was divided into three ROIs: periosteal, middle, and endosteal thirds, each of equal width; (ii) circumferentially the cortex was divided into eight ROIs: anterior, anteromedial, medial, posteromedial, posterior, posterolateral, lateral, and anterolateral cortical octants; and (iii) a combination of radial and circumferential segmentation schemes resulted in 24 ROIs, with each octant subdivided into periosteal, middle, and endosteal thirds (Fig. 2). Similar segmentation of long bone cortices has been employed by Goldman et al. (2003a,b, 2005), Thomas et al. (2005), and McFarlin et al. (2008).

Figure 2.

Figure 2

Open in a new tab

The femoral cortex subdivided circumferentially into anterior (A), anteromedial (AM), medial (M), posteromedial (PM), posterior (P), posterolateral (PL), lateral (L), and anterolateral (AL) octants. The cortex is subdivided radially into periosteal, middle, and endosteal thirds. The dashed line running anteroposteriorly, which passes through the mathematically defined centroid and bisects the linea aspera, was used as the axis origin.

All secondary remodeling events for each femoral cross‐section were manually identified by an experienced observer (T.P.G.) and annotated digitally with point features in arcgis; a total of 230 870 remodeling events were identified from the 30 samples. Whenever histomorphological delineation was unclear on the static image, original samples were viewed directly under a microscope so that areas of uncertainty could be thoroughly investigated. This allowed viewing under various magnifications, fine‐tune scrolling in the z‐plane to view three‐dimensionality of structures, as well as switching between polarized and standard brightfield illumination for better visualization of reversal lines, also known as cement lines.

Due to the time‐consuming nature of annotating all remodeling events in each femoral cross‐section, these features were simply recorded as remodeling events, rather than specifically coding each type (e.g. intact or fragmentary osteon). However, a set of standardized definitions of histomorphometric variables put forth by Heinrich et al. (2012) was used to determine which features to include as remodeling events. As per these definitions, an intact secondary osteon was defined as a remodeling event that was either completely or partially bounded by a reversal line, but had an intact Haversian canal. Fragmentary secondary remodeling events were those in which the Haversian canal had been encroached upon and was either partially or completely removed by subsequent remodeling events. In such instances, fragmentary osteons were further distinguished from interstitial lamellae by their concentric lamellar structure as well as by the osteocytic lacunae embedded in them that follow the concentric orientation of the lamellae. Resorptive bays represent the cutting cone of a basic multicellular unit (BMU) and therefore also constituted a remodeling event, even though they were not technically intact or fragmentary osteons. Further details of all steps outlined here are provided in Gocha (2014).

A subset of 12 femoral cross‐sections was selected for measures of intra‐ and interobserver error, with at least one femur per decade of adult life included in this study. Because quantification of all remodeling events across the femoral cortex is extremely time‐consuming, a smaller ROI was used to assess quantification of OPD for observer error, specifically the periosteal third of the anterior octant on each of the 12 cross‐sections. For intraobserver error, a period of at least 2 months elapsed between the original and repeated measures. For interobserver error, a qualified observer who had significant experience with skeletal histology and identification of remodeling events was chosen. Both observers used the same histomorphological definitions (Heinrich et al. 2012) to inform their decisions on which features to quantify.

Data analysis

The number of remodeling events per square millimeter was used to calculate osteon population density (OPD) for each cortical ROI for the three segmentation methods tested – radial thirds, circumferential octants, and octants by thirds. Due to the small sample size, Shapiro–Wilk tests were used to evaluate normality of data distribution for all tests requiring this assumption. To test for differences in OPD for males and females, multivariate analyses of covariance (mancova) tests were used. Differences in distribution of cortical remodeling were examined semi‐qualitatively through remodeling density maps in arcgis and quantitatively examined through one‐way ancova analyses. Measures of intra‐ and interobserver error were analyzed by paired sample t‐tests. All statistical tests were performed in spss® v. 21.0 (IBM Corp.). To minimize the chance of a Type I error, all statistical tests were performed with a significance level set at α = 0.01.

Results

OPD values from all ROIs for each of the three cortical segmentation methods were normally distributed (0.06 < all P‐values < 0.99) for both males and females. The mancova tests for differences between males and females treated each of the ROIs in a segmentation method as dependent variables, sex was the fixed factor, and – because age greatly affects femoral OPD (Kerley, 1965; Gocha, 2014) – age was treated as a covariate. The OPD from each ROI for all segmentation methods did not differ significantly between sexes (0.06 < all P‐values < 0.97). Further, an independent samples t‐test verified there was no significant difference in age distribution of male and female samples (P = 0.49). Therefore, data from both sexes were combined for further quantitative analyses.

Density and distribution of intracortical remodeling events were examined semi‐qualitatively by producing both absolute and relative density maps within arcgis. To generate density maps, arcgis uses each pixel of the base image as a starting point and calculates the number of remodeling events fall within a given area around that pixel. For this research, an area of 1 mm2 was used so that all OPD calculations would be given as the number of remodeling events per square millimeter. The OPD value associated with each pixel can then be color‐coded within arcgis, allowing remodeling density to be visualized in a meaningful manner.

Absolute density maps were generated by breaking OPD values into 12 classes, each representing an OPD range of five remodeling events per square millimeter, covering an OPD range up to 59.99. Figure 3 depicts absolute density maps for one individual per decade of adult life included in this study to provide a succinct visualization of how remodeling density increases over the adult age spectrum, as evidenced by an increase in warm‐toned pixels with increasing age. These absolute density maps depict OPD for each sample along the same fixed, or absolute scale.

Figure 3.

Figure 3

Open in a new tab

Absolute density maps of femoral OPD for a representative sub‐sample: (a) 21‐year‐old male, (b) 35‐year‐old female, (c) 44‐year‐old female, (d) 57‐year‐old female, (e) 69‐year‐old female, (f) 76‐year‐old male, (g) 84‐year‐old male, and (h) 92‐year‐old female. All samples are oriented with anterior at the top and medial to the right.

Relative density maps depict OPD along a relative scale, which is unique to each sample. In this research, the highest OPD value found in a sample was assigned a red color, the lowest OPD value a blue color, and all OPD values in between are stretched along a color palate determined by arcgis. This type of density map is beneficial for investigating patterns that might not be as easily discernable in absolute density maps, for instance, which areas of the cortex are typically most heavily remodeled. Such an approach corrects for the fact that OPD in the most heavily remodeled area in one sample may be very different from the OPD value in the most heavily remodeled area in a different sample. Figure 4 depicts relative density maps for one individual per decade of adult life included in this study. Figures 3 and 4 demonstrate there is a great deal of variation in distribution of remodeling events across the depth and circumferentially around the femoral cortex. Beginning in the middle of the fourth decade (mid‐30s) and continuing through adulthood, there is a marked tendency for remodeling density to be highest in the lateral and anterolateral regions of the cortex, as evidenced by a distinct localization of warm‐toned pixels in these areas.

Figure 4.

Figure 4

Open in a new tab

Relative density maps of OPD. Red indicates the highest observed OPD in a sample, blue the lowest OPD in a sample, with all values between stretched along the color palate: (a) 21‐year‐old male, high OPD 25.0; (b) 35‐year‐old female, high OPD 32.0; (c) 44‐year‐old female, high OPD 42.0; (d) 57‐year‐old female, high OPD 38.0; (e) 69‐year‐old female, high OPD 45.0; (f) 76‐year‐old male, high OPD 48.0; (g) 84‐year‐old male, high OPD 48.0; (h) 92‐year‐old female, high OPD 52.0. Lowest observed OPD for each sample was 0.0. All samples are oriented with anterior at the top and medial to the right.

Remodeling distribution was analyzed quantitatively by comparing mean OPD value for each cortical ROI within a segmentation method (Figs 5, 6, 7). Figure 5 demonstrates that when remodeling density was considered across the depth of the cortex, OPD values were, on average, highest in the periosteal third, intermediate in the middle third, and lowest in the endosteal third of the cortex. Figure 6 demonstrates that when remodeling density was considered circumferentially around the cortex, the highest OPD values were found in the lateral and anterolateral octants, while the lowest OPD values were found in the anteromedial, posteromedial, and posterior octants. Figure 7 demonstrates that when remodeling density was considered both circumferentially and radially across the depth of the cortex, OPD values were highest in the periosteal third, intermediate in the middle third, and lowest in the endosteal third for the posterior, posterolateral, lateral, and anterolateral regions of the femoral cortex. A different pattern is noted in the anterior and medial regions of the femoral cortex, where the highest average OPD values are found in the middle third of the cortex and varies by region as to whether the periosteal or endosteal thirds has the intermediate or lowest OPD values.

Figure 5.

Figure 5

Open in a new tab

Bar chart of mean OPD values for each ROI in the cortical thirds segmentation method.

Figure 6.

Figure 6

Open in a new tab

Bar chart of mean OPD values for each ROI in the cortical octants segmentation method.

Figure 7.

Figure 7

Open in a new tab

Bar chart of mean OPD values for each ROI in the cortical octants by thirds segmentation method.

With density maps and bar charts demonstrating that remodeling events are not uniformly distributed throughout the femoral cortex, ancova was used to test whether OPD values between ROIs from each segmentation method differed significantly. For each ancova analysis, the fixed factors were ROIs, the dependent variable was OPD, and the covariate was age. Table 1 presents the results of ancova analyses and shows a statistically significant difference in OPD between ROIs for each segmentation method.

Table 1.

The ancova analyses testing for differences in OPD between ROIs per segmentation method

Segmentation method df F P‐value
Cortical thirds 2, 86 9.793 <0.0001
Cortical octants 7, 231 21.663 <0.0001
Cortical octants by thirds 23, 695 17.752 <0.0001
Open in a new tab

Data for measures of intra‐ and interobserver error (Table 2) were found to be distributed normally (all P‐values ≥ 0.25) and were thus tested for significant differences with two‐tailed paired t‐tests. Table 3 presents the results of the paired t‐tests, which confirmed that the OPD values resulting from remodeling event counts were not significantly different between observers or within one observer. Although the results of the t‐tests near statistical significance, the differences in OPD calculation between observers and within one observer are minimal; a difference of one or two osteons per square millimeter is insignificant (both practically and statistically) when considering the overall variation of OPD across the femoral cortex. Furthermore, the average percent error between OPD values calculated for measures of intra‐ and interobserver error were found to be <5% between observer counts.

Table 2.

OPD data for measures of intra‐ and interobserver error

Subject Age Intraobserver 1 Intraobserver 2 Interobserver
OPD OPD OPD
A 29 12.04 12.08 11.46
B 32 12.57 12.06 11.46
C 35 20.06 19.95 19.41
D 44 15.30 15.82 14.70
E 45 12.28 13.67 12.94
F 57 15.09 15.64 14.61
G 61 20.82 20.48 19.89
H 69 28.10 29.49 26.62
I 76 27.30 28.00 27.00
J 81 21.98 24.71 20.79
K 84 26.73 26.62 27.05
L 92 35.71 36.02 32.54
Open in a new tab

Table 3.

Paired t‐tests for intra‐ and interobserver error

Paired samples test
Paired differences t df P‐value
Mean SD SE mean 95% CI of difference
Lower Upper
Pair 1 Intraobserver 1 – Intraobserver 2 −0.55 0.92 0.27 −1.13 0.04 −2.07 11 0.06
Pair 2 Intraobserver 1 – Interobserver 0.79 0.96 0.28 0.18 1.40 2.85 11 0.02
Open in a new tab

CI, confidence interval.

Discussion

In this study, we examined spatial distribution of secondary remodeling events at the femoral midshaft in a sample of modern human cadavers aged 21–97 years. Complete cross‐sections were digitally mapped in geographic information systems software (arcgis), and the cortex subdivided with three different segmentation methods. Radially, the cortex was divided into periosteal, middle, and endosteal thirds, each of equal width; circumferentially, the cortex was divided into octants; the cortex was segmented by an overlay of the first two methods, where octants were considered by radial thirds. All remodeling events (i.e. intact osteons, fragmentary osteons, and resorptive bays) were identified by an experienced bone microscopist (T.P.G.) and digitally annotated in arcgis, and remodeling density (OPD) calculated based on areas for all ROIs for each segmentation method. Distribution of the user‐defined intracortical remodeling events was also mapped across each cross‐section through density maps in arcgis. Both remodeling density maps and comparisons of mean OPD values by ROI confirmed the hypothesis that there is a great deal of heterogeneity in spatial distribution of secondary remodeling events in the human femoral midshaft. More specifically, we found a distinct pattern where remodeling density tended to be highest in the periosteal third of the bone and, beyond the middle of the fourth decade of life, almost invariably highest in the lateral and anterolateral regions of the femoral cortex. On average, lowest remodeling densities were noted in the endosteal third of the bone and circumferentially in the posterior, posteromedial, and anteromedial regions of the femoral cortex.

One potential explanation for higher remodeling density in the anterolateral cortex of the femoral midshaft could be that this area of the cortex has a higher effective age of the adult compacta (i.e. older primary bone) and therefore has more remodeling due to greater time available for osteons to accumulate. Indeed, McFarlin et al. (2008) report higher proportions of osteonal bone in older areas of long bone cortices of various catarrhine primates. In light of what is known about femoral modeling drift in humans, however, this is a poor explanation of the pattern found in this study.

Modeling drift manifests in long bone diaphyses through formation modeling (apposition) occurring on one periosteal and one endosteal surface; at the same time, resorptive modeling is active on the opposite periosteal and endosteal surfaces (Allen & Burr, 2014). In early childhood, the human femoral midshaft appears to undergo a drift pattern that is somewhat variable, though generally directed in the posterior direction (Goldman et al. 2009; Gosman et al. 2013). By late childhood, modeling drift in the femur shifts direction to primarily lateral and/or anterolateral and continues in this direction through skeletal maturation (Fig. 8; Goldman et al. 2009; Maggiano et al. 2011). Thus through development, some areas of the cortex that existed in early life are completely removed, and entirely new areas of bone are deposited. This results in an adult cortex that contains a collection of lamellae of different ages (Robling & Stout, 2008). For the femoral midshaft, this effectively means that areas of the cortex exhibiting the youngest primary lamellar tissue age will be located endosteally on the medial cortex, histomorphologically known as the endosteal lamellar pocket (ELP; Maggiano et al. 2011), and periosteally on the lateral and anterolateral cortex.

Figure 8.

Figure 8

Open in a new tab

Schematic diagram of modeling drift at the femoral midshaft. After Maggiano et al. (2011).

Considering the highest remodeling density exists in an area of the cortex that is among the youngest primary tissue in the femur, we conclude that the anterolateral region of the femoral midshaft has a higher remodeling rate than elsewhere in the cortex. This finding complements the conclusions of Goldman et al. (2003a, 2005), who report the anterolateral portion of the femoral midshaft has a significantly lower mineralization density than elsewhere in the cortex. Since newly formed bone is less mineralized than older bone, regions of the cortex with higher remodeling rates tend to have lower mineralization density (Boivin & Meunier, 2002).

If not a product of tissue age, the spatial variation in remodeling density found in this study is best understood in the context of what skeletal remodeling is trying to achieve. Skeletal remodeling is generally recognized as having three purposes: (i) maintenance of mineral homeostasis, (ii) repair of microdamage, and (iii) adaptation to the mechanical environment (Burr, 2002), with the latter two being biomechanical in nature. To aid in maintenance of mineral homeostasis, bone remodeling could theoretically act upon or within any bone envelope, though ideally this would not alter the structural integrity of bone. In reality, however, remodeling for the purpose of mineral homeostasis is primarily active in cancellous bone tissue of the axial skeleton (Bronner, 1992; Parfitt, 2002). Further, it has been demonstrated that mechanical factors, not mineral homeostasis, are primarily responsible for differences in distribution of osteonal bone in primates (Schaffler & Burr, 1984). Therefore, the pattern of higher remodeling density in the anterolateral femoral cortex – which is present in nearly all samples older than 35 years, regardless of sex, femoral size, or femoral cortex shape – is likely a product of mechanical loading associated with obligate bipedalism in humans. Biomechanical loading in the femoral diaphysis involves a complex combination of bending and axial compression (Lieberman et al. 2004); however, as in vivo strain data for the human femur during gait are lacking, exactly how loading manifests as microstructure throughout the cortex remains an open question.

One line of inquiry that offers some insight as to what strain modes act on cortical bone tissue is examination of type I collagen fiber orientation (CFO). Ascenzi & Bonucci's (1967, 1968) pioneering research demonstrated that in response to compressive forces, collagen fibers in bone are preferentially deposited in a more transverse orientation, whereas in response to tensile forces, fibers are preferentially laid down in a more longitudinal orientation. In the first study to observe CFO across the entirety of the femoral midshaft, Portigliatti Barbos et al. (1983) examined three complete cross‐sections from a single 46‐year‐old male and concluded that longitudinally oriented collagen fibers were predominantly found in the anterolateral femoral cortex, and transversely oriented collagen fibers in the posterior and medial regions of the femoral midshaft. Years later, Goldman et al. (2005) examined CFO at the femoral midshaft in a much larger sample (n = 37) and reported that, despite a high degree of individual variation, in general the anterolateral femoral cortex had higher proportions of longitudinally oriented collagen fibers, whereas the posteromedial region of the cortex tended to have high proportions of transversely oriented collagen fibers. From this evidence, the anterolateral region of the femoral midshaft seems to be subject more to tensile strains and the posteromedial femoral midshaft more to compressive strains, although there is some overlap between these strain modes, in addition to shear forces acting throughout the human femoral midshaft (Skedros, 2012).

Finite element modeling of strain distribution in the human femur suggests a similar pattern. Duda et al. (1998) examined a femur model that included all thigh muscles as well as joint contact forces and found that tensile strains exceed compressive strains in the anterior and lateral aspects of the midshaft during gait. Similarly, Polgár et al. (2003) tested two different physiological loading scenarios using a muscle standardized femur model and found, in both scenarios, that tensile forces were more localized to the lateral side of the femur, whereas compressive forces acted more on the medial side of the femur. These finite element models, however, do not take into account variations in CFO which, as previously stated, aid cortical bone in responding to different types of loading. Finite element models that do take CFO into account have recently been proposed for the proximal human femur (Ascenzi et al. 2013) but are still lacking for the midshaft of the human femur. Therefore, while the results of the finite element models considered here and studies of CFO are imperfect and indirect measures of biomechanical forces acting on the human femoral midshaft, we conclude that the preponderance of evidence suggests that the increased osteon densities that we found in the lateral and anterolateral regions of the cortex reflect the combination of tension and shear strains rather than compressive strains.

Bone is an anisotropic material and therefore responds differently depending on the direction of force being applied. Generally speaking, bone is weaker in tension than in compression (Evans, 1957). This is not only true for gross failure but also at the microscopic level. There is evidence that microdamage not only appears in cortical bone subject to tension/shear before compression (Skedros et al. 2011; Skedros, 2012), but also accumulates more rapidly in tensile cortices than in compressive ones during fatigue loading (Ebacher et al. 2007). Furthermore, it is now accepted that different types of microdamage are produced by varied strain modes in cortical bone. Diffuse microdamage, which appears as a spread mesh of submicroscopic cracks, is associated with areas of tension, whereas sharply defined microscopic linear microfractures are more typically the result of compression (Boyce et al. 1998; Karim & Vashishth, 2012).

In 1960, Frost suggested that microdamage might disrupt canalicular networks of osteocytes and thereby initiate remodeling. However, it was not until relatively recently that the importance of osteocytes in skeletal homeostasis, through functions of mechanosensation, mechanotransduction, and as a catalyst for intracortical remodeling, was fully appreciated (Bonewald, 2011). As Allen & Burr (2014) point out, osteocyte apoptosis is now considered the critical event to signal initiation of intracortical remodeling. If diffuse damage were to disrupt the canalicular networks of osteocytes, as was previously thought, it might be responsible for higher remodeling rates found in the anterolateral cortex of the midshaft femur. However, Seref‐Ferlengez et al. (2014) have now provided empirical evidence of what some researchers suggested years ago (e.g. Boyde, 2003) – that diffuse damage can be repaired in vivo by a yet‐to‐be understood mechanism other than intracortical remodeling.

This is not to suggest, however, that the increased remodeling rate we observed in the anterolateral cortex of the femoral midshaft is not due, at least in part, to reparative remodeling. The fact that diffuse damage can be repaired by a mechanism other than intracortical remodeling does not necessarily mean that it is never repaired in such a way. Diab & Vashishth (2007) report that there is fourfold more diffuse damage in areas of the cortex with higher remodeling. The authors also have demonstrated that as chronological age increases, the ability of bones to develop diffuse damage decreases. Instead, the prevalent form of microdamage observed in older individuals is linear microfractures, which are still thought to be repaired primarily through intracortical remodeling. Therefore, it is our suggestion that the pattern of higher remodeling density and rates we have found in the anterolateral cortex of the femur are due, at least in part, to location‐specific strain modes of tension/shear that result in a greater incidence of microdamage, which over time is more likely to manifest as linear microfractures that require intracortical remodeling to repair. However, repair of microdamage is not the only impetus for intracortical remodeling. Skedros (2012) notes that remodeling can probably be activated by high local strain magnitudes sensed by osteocytes, even when traditional microdamage has not occurred, which would constitute intracortical remodeling as a mechanism for bone adapting to its mechanical environment.

To understand how intracortical remodeling can help a bone adapt to its mechanical environment, it is important to understand how osteons affect the strength and toughness of bone. Evans & Bang (1967) showed that osteons reduce the tensile strength and elastic modulus of bone, likely because cement lines of osteons are considered sites of weakness where failure can occur. Lacking these inherent sites of weakness, primary bone was determined to be stronger than osteonal bone (Currey, 1975). Although osteons appear to decrease the strength of bone, however, they appear to increase the toughness of the cortex, as Yeni et al. (1997) report that increased osteon density increases both shear and tensile toughness. Further, both Evans & Vincintelli (1974) and Lanyon et al. (1982) demonstrate that the presence of osteons in cortical bone increases fatigue life of that cortex, not because the presence of osteons helps to prevent microdamage, but because it seems to aid in reducing severity. This is because osteonal barriers (i.e. cement lines) help to attenuate energy of microdamage, thereby restricting propagation (Gibson et al. 2006). Indeed, Mullins et al. (2006) have demonstrated that the severity of microdamage, as measured by crack length, is much greater in areas of interstitial bone and much less severe in areas of osteonal bone. Therefore, higher remodeling density found in the anterolateral femoral cortex in this study might be evidence of an adaptive toughening mechanism in response to habitual tension/shear loading in this region to reduce the severity of microdamage experienced.

In addition to strain mode, strain magnitude has also been demonstrated to influence regional variation in osteon density. For example, Skedros et al. (2009) found a significantly higher OPD in the cranial cortex of sheep tibiae, which is primarily loaded in tension, than in the caudal cortex which is primarily loaded in compression; however, a relatively uniform pattern in CFO was also found in this bone, indicating the pattern was not due solely to strain mode. Instead, Skedros and colleagues attributed the higher remodeling density to the greater strain magnitude experienced by the cranial cortex of sheep tibiae during accelerated gaits. Bouvier & Hylander (1981) also point out that in addition to strain mode, it is important to consider strain magnitude. In their study on macaque monkeys, the authors report finding more remodeling events in the mandibular cortex of subjects whose diet resulted in higher strain magnitudes during mastication.

Unfortunately, there is a paucity of reliable data on strain magnitude in the human femoral midshaft during its various phases of locomotion. Lacking in vivo strain gage data, it is difficult to say with certainty whether the anterolateral femoral cortex experiences greater strain magnitudes than elsewhere in the cortex. While finite element modeling could provide some insight on the subject, the varied assumptions, soft tissue modeling conditions, and differences in phases of human bipedalism have yielded results that vary widely with regard to hypothesized strain magnitudes around the femoral cortex. In the absence of this data, our understanding of femoral loading is incomplete and therefore the interpretations of our data are preliminary; there may be higher strain magnitudes acting upon the anterolateral femoral cortex that result in increased remodeling at that location. However, considering the patterns of collagen fiber orientation discussed earlier (Portigliatti Barbos et al. 1983; Goldman et al. 2005), in addition to the consistent patterns of strain mode from finite element modeling, we contend that higher remodeling density in the anterolateral femur is also, at least in part, due to tension/shear strains acting more upon this region of the cortex, resulting in microdamage in need of repair.

Our finding of higher remodeling density in tensile areas of compact bone agrees with the only previous study to consider intracortical remodeling across the entirety of the human femoral cortex, in which Portigliatti Barbos et al. (1983) concluded, based on sections from a single individual, that bone remodeling was most active in areas capable of supporting tensile stress. It does not, however, agree with other previously published reports. Mason et al. (1995) report varied results for remodeling density and its association with strain mode. In sections from 50 and 65% of bone length in equine radii (measured from distal), the density of secondary osteons was significantly higher in areas of compression, although in sections at 35% of bone length, the pattern was reversed and higher remodeling density was in fact found in the tensile cortex. Looking at the equine femoral diaphysis, Zedda et al. (2015) report that areas under compressive strains have significantly higher remodeling density than areas under tensile stress. Mayya et al. (2013) go so far as to conclude that intracortical remodeling in mammalian bone is associated only with high compressive strains; as their research examined only one goat femur, such a broad conclusion is scientifically errant. The differences found between our data and those of Mason et al. (1995) and Zedda et al. (2015) may exist, in part, because of taxonomic differences in their study material and ours. Horse primary compact tissue is plexiform in nature (Hillier & Bell, 2007), whereas primary compact bone in humans is circumferential lamellae, and to some extent this affects the way compact tissue responds to the loads placed on it (Norman et al. 1995).

Differences in our findings and others may also be due to the way in which secondary remodeling events were quantified. Zedda et al. (2015) counted only intact osteons from the middle third of the equine femoral cortex, ignoring the periosteal and endosteal thirds, and therefore did not account for all possible spatial variation in secondary remodeling events across the cortex. Mason et al. (1995) examined the entire cortex of the equine radius and report counting intact and fragmentary secondary osteons, though their definition of fragmentary osteons still required the presence of an intact Haversian canal. Therefore, fragmentary osteons by the definition utilized here (having only a partial Haversian canal or lacking a canal altogether) were not counted by Mason et al. (1995). Quantification of secondary remodeling events in such a manner will not accurately depict remodeling density, because areas of the cortex with higher remodeling rates (and thus higher OPD) will necessarily create more osteon fragments that may be missing Haversian canals; ergo, remodeling density will be underreported if fragmentary osteons without Haversian canals are omitted. Based on the standardized definitions adopted in this research (Heinrich et al. 2012), any remodeling event that has an intact Haversian canal as viewed in a two‐dimensional cross‐section should still be considered an intact secondary osteon, as it is still a viable, independent structure transmitting vasculature; only when the Haversian canal has been encroached upon should an osteon be considered fragmentary. Identifying all intact, fragmentary, and resorptive bays as was done in this research is indeed time‐consuming but is necessary for an accurate representation of intracortical remodeling densities.

Concluding remarks

A thorough understanding of the spatial distribution of secondary remodeling at the femoral midshaft is important not only to understand the reparative response to loads placed upon it, but also to appreciate how the cortex adapts to these biomechanical demands. The results of this study demonstrate that the density of secondary remodeling is highest in the anterolateral cortex, specifically in the periosteal third, of the human femoral cortex. These locally higher remodeling densities are thought to be the result of more tensile strains and/or higher strain magnitude placed upon the lateral/anterolateral femoral midshaft. However, patterns of secondary remodeling examined in this study tell only part of the story, as bone responds to biomechanical loading in more varied ways than secondary remodeling alone. An even more complete understanding of this relationship could be achieved through examination of remodeling along with other variables, such as osteon morphotypes and collagen fiber orientation, intracortical porosity, bone mineral content, and various cross‐sectional mechanical properties, which this study did not consider. Furthermore, due to the scarceness of reliable data on biomechanical loading at the human femoral midshaft, this study only considered two strain‐related stimuli (mode and magnitude), but there are several more (see Brown et al. 1990 and Skedros, 2012) that should be considered, where possible, in future studies. Nevertheless, these results provide important baseline information that was previously lacking on the patterns of secondary remodeling across the entirety of the femoral midshaft.

Further, this examination of osteon population density at the femoral midshaft also has practical applications in physical and forensic anthropology, as there is a robust relationship between increased osteon density and chronological age (Kerley, 1965; Robling & Stout, 2008). The pattern of spatial distribution in secondary remodeling found in this study can likely inform researchers interested in estimating age at death from human skeletal remains as to which regions of the femoral cortex display increased osteon population density that correlates best with chronological age. This practical application of our results and the relationship with other mechanical properties of the femoral midshaft, are topics currently under investigation.

Author contributions

T.P.G. conceived and designed the research, collected all data, analyzed and interpreted the findings, and drafted and revised the article. A.M.A. contributed to the design of the research, the analysis and interpretation of the findings, as well as the initial draft and revisions of the article.

Acknowledgements

We would like to thank all of the donors who generously donated their body for scientific research. Thanks to Drs Christian Crowder and Sam Stout for access to the Ericksen femur sample, and also to Victoria Dominguez and Dr. Randee Hunter for their help in sample preparation. Drs Paul Sciulli, Sam Stout, and Mark Hubbe also provided helpful insight into both the theoretical and practical aspects of this research, greatly improving it. Thanks also to Dr. A. R. S. Gocha for helpful editorial comments, as well as the two anonymous reviewers for their thoughtful and detailed suggestions.

References

  1. Agnew AM, Stout SD (2012) Reevaluating osteoporosis in human ribs: the role of intracortical porosity. Am J Phys Anthropol 148, 462–466. [DOI] [PubMed] [Google Scholar]
  2. Allen MR, Burr DB (2014) Bone modeling and remodeling In: Basic and Applied Bone Biology. (eds Burr DB, Allen MR.), pp. 75–90, New York: Academic Press. [Google Scholar]
  3. Ascenzi A, Bonucci E (1967) The tensile properties of single osteons. Anat Rec 158, 375–386. [DOI] [PubMed] [Google Scholar]
  4. Ascenzi A, Bonucci E (1968) The compressive properties of single osteons. Anat Rec 161, 377–392. [DOI] [PubMed] [Google Scholar]
  5. Ascenzi MG, Kawas NP, Lutz A, et al. (2013) Individual‐specific multi‐scale finite element simulation of cortical bone of human proximal femur. J Comput Phys 244, 298–311. [Google Scholar]
  6. Bell KL, Loveridge N, Reeve J, et al. (2001) Super‐osteons (remodeling clusters) in the cortex of the femoral shaft: influence of age and gender. Anat Rec 264, 374–386. [DOI] [PubMed] [Google Scholar]
  7. Boivin G, Meunier PJ (2002) Changes in bone remodeling rate influence the degree of mineralization of bone. Connect Tissue Res 43, 535–537. [DOI] [PubMed] [Google Scholar]
  8. Bonewald LF (2011) The amazing osteocyte. J Bone Miner Res 26, 229–238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bouvier M, Hylander WL (1981) Effect of bone strain on cortical bone structure in macaques (Macaca mulatta). J Morphol 167, 1–12. [DOI] [PubMed] [Google Scholar]
  10. Boyce TM, Fyhrie DP, Glotkowski MC, et al. (1998) Damage type and strain mode associations in human compact bone bending fatigue. J Orthop Res 16, 322–329. [DOI] [PubMed] [Google Scholar]
  11. Boyde A (2003) The real response of bone to exercise. J Anat 203, 173–189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Bromage TG, Goldman HM, McFarlin SC, et al. (2003) Circularly polarized light standards for investigations of collagen fiber orientation in bone. Anat Rec 274B, 157–168. [DOI] [PubMed] [Google Scholar]
  13. Bronner F (1992) Bone and calcium homeostasis. Neurotoxicology 13, 775–782. [PubMed] [Google Scholar]
  14. Brown TD, Pedersen DR, Gray ML, et al. (1990) Toward an identification of mechanical parameters initiating periosteal remodeling: a combined experimental and analytic approach. J Biomech 23, 893–905. [DOI] [PubMed] [Google Scholar]
  15. Burr DB (2002) Targeted and nontargeted remodeling. Bone 30, 2–4. [DOI] [PubMed] [Google Scholar]
  16. Cambra‐Moo O, Meneses CN, Barbero MÁR, et al. (2012) Mapping human long bone compartimentalisation during ontogeny: a new methodological approach. J Struct Biol 178, 338–349. [DOI] [PubMed] [Google Scholar]
  17. Currey JD (1975) The effects of strain rate, reconstruction, and mineral content on some mechanical properties of bovine bone. J Biomech 8, 81–86. [DOI] [PubMed] [Google Scholar]
  18. Diab T, Vashishth D (2007) Morphology, localization and accumulation of in vivo microdamage in human cortical bone. Bone 40, 612–618. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Drapeau MSM, Streeter MA (2006) Modeling and remodeling responses to normal loading in the human lower limb. Am J Phys Anthropol 129, 403–409. [DOI] [PubMed] [Google Scholar]
  20. Duda GN, Heller M, Albinger J, et al. (1998) Influence of muscle forces on femoral strain distribution. J Biomech 31, 841–846. [DOI] [PubMed] [Google Scholar]
  21. Ebacher V, Tang C, McKay H, et al. (2007) Strain redistribution and cracking behavior of human bone during bending. Bone 40, 1265–1275. [DOI] [PubMed] [Google Scholar]
  22. Enlow DH (1962) Functions of the Haversian system. Am J Anat 110, 269–305. [DOI] [PubMed] [Google Scholar]
  23. Ericksen MF (1991) Histologic estimation of age at death using the anterior cortex of the femur. Am J Phys Anthropol 84, 171–179. [DOI] [PubMed] [Google Scholar]
  24. Evans FG (1957) Stress and Strain in Bones. Springfield: Charles C Thomas. [Google Scholar]
  25. Evans FG, Bang S (1967) Differences and relationships between physical properties and the microscopic structure of human femoral, tibial, and fibular cortical bone. Am J Anat 120, 79–88. [Google Scholar]
  26. Evans FG, Vincintelli R (1974) Relations of the compressive properties of human cortical bone to histological structure and calcification. J Biomech 7, 1–10. [DOI] [PubMed] [Google Scholar]
  27. Frost HM (1960) Presence of microscopic cracks in vivo in bone. Henry Ford Hosp Med Bull 8, 25–35. [Google Scholar]
  28. Frost HM (1987a) Secondary osteon populations: an algorithm for determining mean tissue age. Yearb Phys Anthropol 30, 221–238. [Google Scholar]
  29. Frost HM (1987b) Secondary osteon populations: an algorithm for estimating the missing ones. Yearb Phys Anthropol 30, 239–254. [Google Scholar]
  30. Gibson VA, Stover SM, Gibeling JC, et al. (2006) Osteonal effects on elastic modulus and fatigue life in equine bone. J Biomech 39, 217–225. [DOI] [PubMed] [Google Scholar]
  31. Gocha TP (2014) Mapping spatial patterns in cortical remodeling from the femoral midshaft using geographic information systems software: implications for age estimation form adult human skeletal remains. Unpublished doctoral dissertation, The Ohio State University, Columbus. [Google Scholar]
  32. Goldman HM, Bromage TG, Thomas CDL, et al. (2003a) Preferred collagen fiber orientation in the human mid‐shaft femur. Anat Rec 272A, 434–445. [DOI] [PubMed] [Google Scholar]
  33. Goldman HM, Bromage TG, Boyde A, et al. (2003b) Intrapopulation variability in mineralization density at the human femoral mid‐shaft. J Anat 203, 243–255. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Goldman HM, Thomas CDL, Clement JG, et al. (2005) Relationships among microstructural properties of bone at the human midshaft femur. J Anat 206, 127–139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Goldman HM, McFarlin SC, Cooper DML, et al. (2009) Ontogenetic patterning of cortical bone microstructure and geometry at the human mid‐shaft femur. Anat Rec 292, 48–64. [DOI] [PubMed] [Google Scholar]
  36. Gosman JH, Hubbell ZR, Shaw CN, et al. (2013) Development of cortical bone geometry in the human femoral and tibial diaphysis. Anat Rec 296, 774–787. [DOI] [PubMed] [Google Scholar]
  37. Heinrich R, Crowder C, Pinto D, et al. (2012) Proposal and validation of definitions for intact and fragmented osteons. Am J Phys Anthropol 147(S54), 163. [Google Scholar]
  38. Hillier ML, Bell LS (2007) Differentiating human bone from animal bone: a review of histological methods. J Forensic Sci 52, 249–263. [DOI] [PubMed] [Google Scholar]
  39. Karim L, Vashishth D (2012) Bone microdamage and its contributions to fracture In: Skeletal Aging and Osteoporosis. (ed. Silva MJ.), pp. 87–104, Berlin: Springer‐Verlag. [Google Scholar]
  40. Kerley ER (1965) The microscopic determination of age in human bone. Am J Phys Anthropol 23, 149–164. [DOI] [PubMed] [Google Scholar]
  41. Lanyon LE, Goodship AE, Pye CJ, et al. (1982) Mechanically adaptive bone remodeling. J Biomech 15, 141–154. [DOI] [PubMed] [Google Scholar]
  42. Lieberman DE, Polk JD, Demes B (2004) Predicting long bone loading from cross‐sectional geometry. Am J Phys Anthropol 123, 156–171. [DOI] [PubMed] [Google Scholar]
  43. Maggiano IS, Maggiano CM, Tiesler V, et al. (2011) A distinct region of microarchitectural variation in femoral compact bone: histomorphometry of the endosteal lamellar pocket. Int J Osteoarchaeol 21, 743–750. [Google Scholar]
  44. Martin RB (2002) Is all cortical bone remodeling initiated by microdamage? Bone 30, 8–13. [DOI] [PubMed] [Google Scholar]
  45. Martin RB, Burr DB, Sharkey NA (1998) Skeletal Tissue Mechanics. New York: Springer. [Google Scholar]
  46. Mason MW, Skedros JG, Bloebaum RD (1995) Evidence of strain‐mode‐related cortical adaptation in the diaphysis of the horse radius. Bone 17, 229–237. [DOI] [PubMed] [Google Scholar]
  47. Mayya A, Banerjee A, Rajesh R (2013) Mammalian cortical bone in tension is non‐Haversian. Nat Sci Rep 3, 2533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. McFarlin SC, Terranova CJ, Zihlman AL, et al. (2008) Regional variability in secondary remodeling within long bone cortices of catarrhine primates: the influence of bone growth history. J Anat 213, 308–324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Mullins LP, Sassi V, McHugh PE, et al. (2006) Differences in crack resistance of interstitial, osteonal, and trabeculae bone tissue. Ann Biomed Eng 37, 2574–2582. [DOI] [PubMed] [Google Scholar]
  50. Norman TL, Vashishth D, Burr DB (1995) Fracture toughness of human bone under tension. J Biomech 28, 309–320. [DOI] [PubMed] [Google Scholar]
  51. Parfitt AM (2002) Misconceptions (2): turnover is always higher in cancellous than in cortical bone. Bone 30, 807–809. [DOI] [PubMed] [Google Scholar]
  52. Polgár K, Gill HS, Viceconti M, et al. (2003) Strain distribution within the human femur due to physiological and simplified loading: finite element analysis using the muscle standardized femur model. Proc Inst Mech Eng [H] 217, 173–189. [DOI] [PubMed] [Google Scholar]
  53. Portigliatti Barbos M, Bianco P, Ascenzi A (1983) Distribution of osteonic and interstitial components in the human femoral shaft with reference to structure, calcification and mechanical properties. Acta Anat 115, 178–186. [DOI] [PubMed] [Google Scholar]
  54. Robling A, Stout S (2008) Histomorphometry of human cortical bone: applications to age estimation In: Biological Anthropology of the Human Skeleton, 2nd edn (eds Katzenberg MA, Saunders SR.), pp. 149–182, New York: John Wiley & Sons. [Google Scholar]
  55. Rose DC, Agnew AM, Gocha TP, et al. (2012) The use of geographic information systems software for the spatial analysis of bone microstructure. Am J Phys Anthropol 148, 648–654. [DOI] [PubMed] [Google Scholar]
  56. Ruff CB. MomentMacro program for ImageJ [internet]. Available at: http://www.hopkinsmedicine.org/fae/mmacro.htm
  57. Schaffler MB, Burr DB (1984) Primate cortical bone microstructure: relationship to locomotion. Am J Phys Anthropol 65, 191–197. [DOI] [PubMed] [Google Scholar]
  58. Seref‐Ferlengez Z, Basta‐Pljakic J, Kennedy OD, et al. (2014) Structural and mechanical repair of diffuse damage in cortical bone in vivo . J Bone Miner Res 29, 2537–2544. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Skedros JG (2012) Interpreting load history in limb‐bone diaphyses: important considerations and their biomechanical foundations In: Bone Histology: An Anthropological Perspective. (eds Crowder CM, Stout SD.), pp. 153–220, Boca Raton: CRC Press Taylor & Francis Group. [Google Scholar]
  60. Skedros JG, Mendenhall SD, Kiser CJ, et al. (2009) Interpreting cortical bone adaptation and load history by quantifying osteon morphotypes in circularly polarized light images. Bone 44, 392–403. [DOI] [PubMed] [Google Scholar]
  61. Skedros JG, Sybrowsky CL, Anderson WE, et al. (2011) Relationships between in vivo microdamage and the remarkable regional material and strain heterogeneity of cortical bone of adult deer, elk, sheep, and horse calcanei. J Anat 219, 722–733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Steinberg SJ, Steinberg SL (2006) GIS: Geographic Information Systems for the Social Sciences: Investigating Space and Place. London: Sage Publications. [Google Scholar]
  63. Thomas CDL, Feik SA, Clement JG (2005) Regional variation of intracortical porosity in the midshaft of the human femur: age and sex differences. J Anat 206, 115–125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Yeni YN, Brown CU, Wang Z, et al. (1997) The influence of bone morphology on fracture toughness of the human femur and tibia. Bone 21, 453–459. [DOI] [PubMed] [Google Scholar]
  65. Zedda M, Lepore G, Biggio GP, et al. (2015) Morphology, morphometry, and spatial distribution of secondary osteons in equine femur. Anat Histol Embryol 44, 328–332. [DOI] [PubMed] [Google Scholar]

Articles from Journal of Anatomy are provided here courtesy of Anatomical Society of Great Britain and Ireland

RESOURCES