Skip to main content
NCBI home page
Journal of Anatomy logoLink to Journal of Anatomy
. 2019 Aug 2;235(5):873–882. doi: 10.1111/joa.13056

Histovariability in human clavicular cortical bone microstructure and its mechanical implications

Matthew A Crane 1, Kyle M Kato 1, Biren A Patel 2,3, Adam K Huttenlocker 2,✉
PMCID: PMC6794202  PMID: 31373387

Abstract

The human clavicle (i.e. collarbone) is an unusual long bone due to its signature S‐shaped curve and variability in macrostructure observed between individuals. Because of the complex nature of how the upper limb moves, as well as due to its complex musculoskeletal arrangement, the biomechanics, in particular the mechanical loadings, of the clavicle are not fully understood. Given that bone remodeling can be influenced by bone stress, the histologic organization of Haversian bone offers a hypothesis of responses to force distributions experienced across a bone. Furthermore, circularly polarized light microscopy can be used to determine the orientation of collagen fibers, providing additional information on how bone matrix might organize to adapt to direction of external loads. We examined Haversian density and collagen fiber orientation, along with cross‐sectional geometry, to test whether the clavicle midshaft shows unique adaptation to atypical load‐bearing when compared with the sternal (medial) and acromial (lateral) shaft regions. Because fractures are most common at the midshaft, we predicted that the cortical bone structure would show both disparities in Haversian remodeling and nonrandomly oriented collagen fibers in the midshaft compared with the sternal and acromial regions. Human clavicles (n = 16) were sampled via thin‐sections at the sternal, middle, and acromial ends of the shaft, and paired sample t‐tests were employed to evaluate within‐individual differences in microstructural or geometric properties. We found that Haversian remodeling is slightly but significantly reduced in the middle of the bone. Analysis of collagen fiber orientation indicated nonrandom fiber orientations that are overbuilt for tensile loads or torsion but are poorly optimized for compressive loads throughout the clavicle. Geometric properties of percent bone area, polar second moment of area, and shape (I max /I min) confirmed the conclusions drawn by existing research on clavicle macrostructure. Our results highlight that mediolateral shape changes might be accompanied by slight changes in Haversian density, but bone matrix organization is predominantly adapted to resisting tensile strains or torsion throughout and may be a major factor in the risk of fracture when experiencing atypical compression.

Keywords: bone histology, clavicle, collagen fiber orientation, cross‐sectional geometry, fracture, Haversian density

Introduction

Students of human anatomy often find the clavicle (collarbone) to be one of the most unusual bones in the skeleton because of its S‐shaped morphology and variability in shape within and across populations. As it serves as an important component of the pectoral girdle, it must be both relatively mobile and serve as a mechanical link to transfer forces between the upper limb bones and the thoracic cage. The latter function is of particular interest to clinicians because of the unusually high prevalence of clavicle fractures in humans (Nordqvist & Petersson, 1994; Postacchini et al. 2002). Of all postcranial skeletal fractures reported, between 2.6 and 5% involve the clavicle, with an annual incidence of 71/100 000 in men and 30/100 000 in women (Smekal et al. 2009). Moreover, the clinical literature reports that the three most common causes of clavicular fractures are a direct shoulder injury, an abnormally high transfer of force resulting from falling on an extended hand or simply direct insult to the bone due to its exposed position on the upper part of the thorax (Stanley et al. 1988; Post, 1989; Duprey et al. 2010).

Like other parts of the postcranial skeleton, a number of different types of forces (e.g. arising from muscle contractions) can cause the clavicle to be loaded in different ways including compression, tension, bending, and torsion (Reilly & Burstein, 1975). From biomechanical experiments, for example to test for bone buckling failure, it has been shown that clavicle fractures occur most frequently when loaded in compression, as force is applied approximately along its longitudinal axis from the acromial to the sternal end (Stanley et al. 1988). These same experiments also showed that higher loads in other directions were much more likely to dislocate the shoulder instead, before any significant macroscopic fractures formed that would otherwise cause the bone to fail completely (Stanley et al. 1988). Additionally, most clinical reports note that the middle third of the clavicle, encompassing the majority of the midshaft, is broken most often with a frequency of 70–80% of reported cases, as compared with the fractures located closer to the sternal or acromial ends, which break less often (Andermahr et al. 2007).

Few studies have proposed an explanation as to why the clavicle fractures in the middle region more often than at either the sternal or acromial ends. Those that have offered a rationale suggest that regional fracture patterns may be related to differences in diaphyseal cortical bone cross‐sectional geometric properties and shape. Ratios of cortical bone and cancellous bone have also been explored recently to understand breakage patterns (Yamamura et al. 2018). Using computed tomography (CT) to obtain serial transverse slices of the clavicle, Harrington et al. (1993) demonstrated that the middle region has the lowest values for section modulus (I), an estimate of bending strength, along both principal axes (I max and I min), as well as for polar second moment of area (J), an estimate of torsional rigidity and twice bending strength (calculated as the sum of I max and I min). These authors also showed that the middle region was less elliptical in shape, as assessed by a smaller I max /I min ratio, suggesting a relatively ‘uniform flexural rigidity’ (Harrington et al. 1993, p. 421). In contrast, Harrington et al. (1993) found that the acromial end in particular, and to a lesser degree the sternal end, had higher values for J and more elliptical cross‐sectional shapes, indicating a greater adaptation to resist bending and torsion laterally (and medially) relative to the midshaft, and a more unidirectional habitual loading regimen.

Despite this limited macroscopic evidence, there has been relatively little research investigating the possibility that the prevalence of breakage of the clavicular shaft could also be influenced by bone microstructural organization and history of remodeling, rather than exclusively via beam theory. For example, analysis of Haversian density could be used to test hypotheses of force distribution across the clavicle. Haversian systems, also known as osteons, are responsible for bone turnover and remodeling, as primary bone is gradually replaced internally by secondary and multigenerational osteons. This turnover has several essential functions, such as providing a functional interface between vasculature and skeletal tissues, maintaining mineral homeostasis, and responding to fatigue microdamage (Jowsey, 1966; Currey, 2006). Micromechanic modeling presents Haversian systems as fiber‐reinforced composite materials interspersed as hollow cylinders throughout cortical bone (Hogan, 1992; Rouhi, 2012). This bears useful comparison with a number of common building materials and metals, with hollow areas serving to blunt growing microcracks and preserve the overall integrity of the structure (O'Brien et al. 2005). Such analogies are complicated, however, by the hierarchical organization of bone tissue and its material properties (Currey, 2006). Debate persists over the main function of Haversian systems, and whether they increase bone strength by preventing microcracks or decrease it by introducing more lamellar bone with high porosity and brittleness (Currey, 2006). Thus, the benefits of remodeling are complex and likely dynamic: removing dead tissue, improving vascular supply, aiding mineral homeostasis, patching microcracks, and adapting directionality of collagen fibers (Currey, 2006). It has been demonstrated, however, that stresses on bone produce microcracks that directly stimulate bone remodeling and secondary osteon formation (Lee et al. 2002). Thus, an examination of the density of Haversian systems, regardless of their definitive effect on overall bone integrity, provides a unique opportunity to understand loading across a bone, with high density Haversian systems possibly corresponding to greater or more frequent mechanical stress.

Collagen fiber orientation is also thought to be related to bone mechanical properties (Goldman et al. 2003). Under circularly polarized light (CPL), the general orientation of collagen fibers in transverse thin‐section can be easily visualized by light and dark regions corresponding to transverse and longitudinal orientations, respectively (Bromage et al. 2003). Existing literature using CPL microscopy suggests that transverse fibers are better adapted to resist compressive forces, whereas longitudinal fibers are better adapted to resist tensile forces – and that this directionality may arise throughout the remodeling process as a part of active bone adaptation (Riggs et al. 1993; Currey, 2006). Combinations of these may suggest adaptation to torsion. Note that here we use the term ‘adaptation’ loosely to convey a given bone's specialization to its functional context, both inherited and experienced, and not to imply evolutionary adaptation per se (Currey, 2003). Collagen fiber orientation maps have been used to visualize these hypothesized load‐bearing adaptations in cross‐section, most frequently quantified using a relative brightness index as a proxy for orientation (Carando et al. 1991; Goldman et al. 2003; Lee, 2004).

In this study, we examined clavicular Haversian density and collagen fiber orientation under the hypothesis that these two measures have the capacity to demonstrate Wolff's law of bone adaptation on a microstructural level (Frost, 2004). As such, Haversian density should provide an indication of the magnitude of the loading history at a specific location, and collagen fiber orientation should reveal how matrix organization may be overbuilt (or underbuilt) to accommodate the orientation of the loading history at a specific location. These two properties were compared at three locations to assess medial to lateral regional variation: sternal, midshaft, and acromial ends. Additionally, we reassessed clavicular cross‐sectional geometry in these same regions following the work of Harrington et al. (1993). Given the high prevalence of breakages at the midshaft (Andermahr et al. 2007), we predicted that this region would express indicators of high stress – using Haversian density as a proxy – and that nonrandom collagen fiber orientations throughout the shaft – using brightness index as a proxy – could be indicative of poor optimization for compressive (or impact) loading.

Materials and methods

A sample of adult human clavicles (n = 16) was acquired from the osteological collections of the Keck School of Medicine of the University of Southern California. Demographic information such as age and sex were not available for these individuals. None of the specimens exhibited any external signs of osteopathology. Photographs and μCT scans (100 μm voxel resolution) of each specimen were obtained for record‐keeping and digital preservation prior to destructive sampling, which required extracting blocks of bone from three designated regions of the shaft, including the medial shaft (i.e. sternal), midshaft, and lateral shaft (i.e. acromial) – evenly spaced at 30%, 50%, and 70% of the total bone length from medial to lateral (Fig. 1). The extracted blocks were dehydrated and cleared with Histo‐Clear, embedded in resin, and cross‐sections sampled perpendicular to the long‐axis of each of the individual blocks using a low‐speed diamond blade saw. The resulting 48 sections (3 × 16 specimens) were further prepared using standard bone histologic techniques (Lamm, 2013; Lee & Simons, 2015), which included mounting the sections on glass slides, followed by grinding and polishing to 100 ± 5 μm thickness (measured using a digital Mitutoyo micrometer). Slides were imaged by scanning on a Leica DM microscope configured for CPL imaging. imagej software (Rasband, 1997–2018) was used to analyze cross‐sectional geometry and histomorphology variables.

Figure 1.

Figure 1

Open in a new tab

Sampling methodology and representative thin section with nonpolarized and with circularly polarized light. Top image is an inferior‐view outline of a left human clavicle. Sampling for all bones was performed with cuts made perpendicular to diaphyseal surface at 30%, 50%, and 70% of bone length, measured from sternal to acromial ends. Clavicle length was assessed as the maximum possible tip‐to‐tip distance. Thin section outlines are shown, with the middle section magnified with nonpolarized light. The insert shows Haversian canals under circularly polarized light.

Cross‐sectional geometry

Using the BoneJ plugin (Doube et al. 2010) for imagej software, maximum and minimum second moments of area (I max and I min) were determined from each scanned thin‐section, from which polar second moment of area (J) and cross‐sectional shape (I max /I min) were then calculated. We also calculated percent bone area (%BA), which is a relative measure of the total amount of bone in the cross‐section divided by total area. For all three of these measures, any trabecular bone in the cross‐section was included to prevent the objective decision‐making involved with assigning arbitrary cutoffs from cortical bone. In general, however, the total amount of trabecular bone was minimal and had negligible effects on the statistical results. Because J will likely be absolutely larger in larger specimens or larger cross‐sections, J was scaled by clavicular length prior to making any comparisons; I max /I min and %BA are already scale‐free.

Haversian density

Procedures for analysis of Haversian density (HD) followed counting protocols detailed in Stout & Paine (1992). Briefly, Haversian systems with at least 90% of their concentric lamellae complete were counted as ‘intact’, with any interrupted systems counted as ‘fragmentary’. The reported results combined these two values under the established standard of determining total osteon density (Stout & Paine, 1992). To enable representative sampling, 24 different sectors of approximately equal area were sampled circularly around each cross‐section, with six sites sampled from four anatomic quadrants (superior, anterior, inferior, and posterior). Sampling sites were as large as could be permitted by the cortical bone area, up to 1 mm2 each. When adequate bone area was not present for six sites in a quadrant, due to thinning bone wall or poor bone preservation, as many sites were selected as possible while excluding endosteal resorption cavities. HD was then calculated as the number of osteons mm–2, averaged over each quadrant; HD was calculated for each diaphyseal region. Such an approach of using an index allows for a relative measure of the osteon density to compare across sections of different absolute size within (and between) specimens.

Collagen fiber orientation

Collagen fiber orientation analysis on each thin‐section was carried out using CPL microscopy following a protocol modified from Bromage et al. (2003), which was itself modified from Carando et al. (1991) and Goldman et al. (2003). Specifically, under CPL microscopy, the general orientation of collagen fibers in transverse‐section can be easily visualized by light and dark regions corresponding to transversely and longitudinally oriented collagen fiber bundles, respectively (Bromage et al. 2003). This pattern results from the birefringence of the bone matrix, coupled with the specimen thickness and angle of matrix fibers to the slow axis. Specifically, the anisotropic structure of longitudinally arranged matrix in cross‐section produces differential retardance of filtered light as it traverses the specimen, resulting in optical extinction in regions where matrix fibers are primarily longitudinal. Importantly, the circular effect introduced by an additional quarter‐wave retardation plate under the specimen in the CPL configuration corrects extinction artifacts that are sometimes introduced under traditional linear or elliptical polarization (e.g. Maltese cross pattern of osteons: see Bromage et al. 2003, figure 1). To ensure uniformity in comparison between samples of equal thickness, all manual and digital light settings were maintained constant during specimen imaging, configured using an LED light source (gain, 1.0; exposure, 11.9). The light levels of all images were also standardized through white balance relative to the first photograph in the series. Using imagej, CPL images of each thin‐section were converted to 8‐bit grayscale images such that pixel value histograms were arrayed in a range of values between 0 (black) and 255 (white). From this range, three generalized pixel color groups were created: (1) pixels with values of 0–28 were regarded as ‘background’ and discarded from analysis; (2) pixels with values of 29–141 were designated as ‘dark’ and represent a proxy for predominantly longitudinal (and tensile load bearing) fibers; and (3) pixels with values of 142–255 were designated as ‘light’ and represent a proxy for predominantly transverse (and compressive load‐bearing) fibers that are mostly tangential to the long‐axis of the bone (Fig. 2). Excluding pixels in the 0–28 bin was essential to eliminate background and empty spaces, preparation artifacts, and osteocyte lacunae, but may also exclude a few areas of strongly longitudinal fibers, thus producing a conservative estimate of ‘longitudinality’ in our sample. For analytical convenience, results are reported as a relative brightness index (BI), defined as the percentage of light (transverse) pixels out of the total of pixels analyzed in a cross‐section. Actual pixels counts were performed in imagej using the histogram function, followed by additional calculations in a spreadsheet. Figure 2 presents a collagen fiber orientation map using the 9‐bin protocol of Carando et al. (1991) in order to visually inspect fiber orientations, with darker colors representing a range of longitudinal matrix organization and increasingly lighter colors representing a corresponding range of more oblique to transverse fibers. Gray value ranges for the binned pixels are also defined in Fig. 2. BI was calculated for each diaphyseal region and, as noted above, such an index allows comparison of relative collagen fiber orientation across sections of different absolute size within (and between) specimens.

Figure 2.

Figure 2

Open in a new tab

Transverse (A) and longitudinal (B) thin‐sections of the same clavicle viewed under circularly polarized light. In transverse section, collagen fibers appear predominantly dark, as they are parallel to the viewing angle. In the longitudinal section, collagen fibers appear bright, as they run perpendicular to the viewing angle. (C) Pixel map showing the orientations of collagen fiber bundles expressed as binned gray values and depicted with a false color scheme (based on Goldman et al., 2003 and Lee, 2004). ‘Dark’ pixels (bins 1–4 with purple, blue, and green colors) correspond to longitudinal fibers that are oriented mostly parallel to the viewing angle. ‘Light’ pixels (bins 5–8 with brown, pink, gray, and white colors) correspond to fibers that are more oblique to transversely oriented. Bin 0 represents ‘masked’ regions that were omitted from analysis to avoid inclusion of background pixels.

Statistical analyses

Intra‐clavicle comparisons between the acromial, middle, and sternal ends were assessed using matched paired t‐test (two‐tailed) using JMP pro 13.0 software. Because three sets of comparisons were made within the same bone – acromial vs. middle, acromial vs. sternal, and middle vs. sternal – the chance of committing a Type 1 error is inflated. Therefore, Bonferroni‐adjusted alpha levels of 0.017 were used to assess significance conservatively (0.05/3). In addition, we provide box‐and‐whiskers plots to visualize variation in each variable across the three regions of the clavicle.

Results

Cross‐sectional geometry

Each cross‐sectional geometry variable varied widely across the sampled regions of the clavicle (Table 1; see Appendix for full dataset). Scaled J was significantly higher in the acromial region than in the middle region (P = 0.0005) and approached significance in comparison with the sternal region (P = 0.0179) (Fig. 3). The middle region has a mean value that is lower than either the of the two medial or lateral ends of the clavicle. Cross‐sectional shape (I max /I min) of the sternal region was significantly lower (i.e. more circular) than the middle (P = 0.0134) and acromial (P = 0.0004) regions; the latter is the most elliptical in shape. Percent bone area (%BA) was significantly lower at the sternal region of the diaphysis than at the middle (P = 0.0003) or acromial (P ≤ 0.0001) regions.

Table 1.

Results of paired t‐tests comparing regional differences in bone properties

Comparison Variable* Region Mean† P‐value t
Sternal‐Middle BA Sternal 66.96 0.0003 4.8936
Middle 74.566
J Sternal 10.408 0.1087 −1.7221
Middle 9.0214
I max /I min Sternal 1.4547 0.0134 2.8582
Middle 1.7487
HD Sternal 15.433 0.1748 1.4244
Middle 14.732
BI Sternal 0.0858 0.1845 −1.3911
Middle 0.0728
Sternal‐Acromial BA Sternal 66.96 <0.0001 6.0743
Acromial 77.696
J Sternal 10.408 0.0179 2.7099
Acromial 13.234
I max /I min Sternal 1.4547 0.0004 4.7619
Acromial 2.0032
HD Sternal 15.433 0.4811 −0.7226
Acromial 15.929
BI Sternal 0.0858 0.409 −0.8494
Acromial 0.0706
Middle‐Acromial BA Middle 74.566 0.0904 1.8292
Acromial 77.696
J Middle 9.0214 0.0005 4.5472
Acromial 13.234
I max /I min Middle 1.7487 0.0793 1.904
Acromial 2.0032
HD Middle 14.732 0.0139 −2.7854
Acromial 15.929
BI Middle 0.0728 0.8963 −0.1326
Acromial 0.0706
Open in a new tab

Bold values indicate significance at the alpha = 0.017 level.

*BA, %Bone Area; BI: brightness index; J, scaled polar second moment of area; I max /I min, shape index; HD, Haversian density. See text for more details.

†Mean values calculated from the raw data presented in Appendix.

Figure 3.

Figure 3

Open in a new tab

Box‐and‐whiskers plot of scaled polar second moment of area (J), cross‐sectional shape (I max /I min), and percent bone area (%BA) across three regions of the clavicle. ACR, acromial region (70% of bone length); MID, middle region (50% of bone length); STR, sternal region (30% of bone length). Horizontal line within each box illustrates the median of the distribution. Boxes envelop the interquartile range (50% of values) of the sample distribution, and whiskers encompass the range excluding outliers. Circles beyond whiskers indicate outliers. See Table 1 for results of paired t‐tests.

Haversian density

The acromial and middle regions of the clavicle showed the greatest disparity in HD (see Appendix for full dataset), with the midshaft having a mean HD of 14.7 osteons mm−2 (the lowest among the three regions), and the acromial end displaying a mean HD of 15.9 osteons mm−2 (the highest among the three regions). Paired t‐tests indicated that differences in relative HD across the three regions of the clavicle are significant only between the acromial and middle regions (P = 0.0139; Table 1; Fig. 4). Across the sample, HD of the sternal end is intermediate and not significantly different from either the acromial or middle region of the shaft.

Figure 4.

Figure 4

Open in a new tab

Box‐and‐whiskers plot of the Haversian density (HD) assessed across three regions of the clavicle. ACR, acromial region (70% of bone length); MID, middle region (50% of bone length); STR, sternal region (30% of bone length). Horizontal line within each box illustrates the median of the distribution. Boxes envelop the interquartile range (50% of values) of the sample distribution, and whiskers encompass the range excluding outliers. See Table 1 for results of paired t‐tests.

Collagen fiber orientation

The Brightness index (BI) is generally low in all regions with each cross‐sections appearing predominantly dark when viewed under CPL (mean brightness index of 0.071, 0.073, and 0.086 for acromial, midshaft, and sternal regions, respectively) (Table 1; see Appendix for full dataset). Thus, each region is predominantly composed of collagen fiber bundles that are longitudinally skewed. Although the data presented here show a slightly higher number of lighter pixels transitioning from the acromial to sternal end (i.e. progressively increasing BI), these regional BI differences are not statistically significant (Table 1; Fig. 5). The distributions were shifted greatly toward darker gray values in all regions, more so than toward the lighter end of the range (acromial mean skewness: 1.99; midshaft mean skewness: 2.13; sternal mean skewness: 2.03).

Figure 5.

Figure 5

Open in a new tab

Box‐and‐whiskers plot of the brightness indices (BI) across three regions of the clavicle. ACR, acromial region (70% of bone length); MID, middle region (50% of bone length); STR, sternal region (30% of bone length). Horizontal line within each box illustrates the median of the distribution. Boxes envelop the interquartile range (50% of values) of the sample distribution, and whiskers encompass the range excluding outliers. Circles beyond whiskers indicate outliers. See Table 1 for results of paired t‐tests.

Discussion

Analysis of shape and microstructural properties of the clavicle reveals features that agree with some previous studies, while providing further novel insights into fracture behavior. The complementary inclusion of cross‐sectional geometric analysis in our study gives context for the microanatomic properties investigated, providing a useful point of comparison with existing literature on the subject (e.g. Harrington et al. 1993). Specifically, the polar moment of inertia (J) provides a measure of bending and torsional strength. Stanley et al. (1988) initially presented a view of clavicle fractures based on slenderness ratio (length/least radius of gyration) as nearly entirely dependent on compressive forces leading to buckling. However, Harrington et al. (1993) revised this view with increased emphasis on the clavicle's S‐shaped macrostructure, which increases the torsional stresses acting on the middle region. Our results show that J at the acromial region was significantly higher than at the middle region and approached being significantly higher than at the sternal region. Harrington et al. (1993) measured this value in 10% increments along the length of the bone, and the 30%, 50%, and 70% measurement sites (defined in our study as regionally representative of sternal, middle, and acromial) exhibited the same trend observed here. This decrease in polar moment of inertia away from the acromial end supports the finding that clavicular shaft fractures are more prone to occur in the middle of the bone and near its sternal ends, because these regions are less adapted to great torsional loads.

Cross‐sectional shape based on measures of maximum and minimum second moment of area provides an estimate of how the bone is loaded (i.e. uniform or multidirectional); a shape index of I max /I min that is smaller and has values closer to 1.0 indicates a more circular shape (e.g. Patel et al. 2013). The results of our analysis show that the sternal end is markedly more circular than the middle or acromial regions. Biomechanically, this suggests the sternal end experiences greater or more frequent multidirectional mechanical loads, a likely scenario considering that the sternoclavicular joint is a relatively mobile synovial joint that can move in a wide range of circular motions (Inman et al. 1944). Harrington et al. (1993) also observed a similarly stark decrease in I max /I min moving from the acromial to sternal end, with a local minimum at around 30% of bone‐length, which is approximately the same location sampled in the present study representing our ‘sternal’ region. These results might suggest that the loading regimes of the clavicle become more directionally specialized moving from the sternal to the middle part of the bone, and ultimately to acromial end.

In regard to histologic properties, Haversian density varies across the length of the clavicle, though perhaps not as strongly as gross shape and cortical bone distribution changes (see above). The slight decrease in Haversian density of the midshaft supports our prediction that the midshaft should express disparities in osteonal densities and remodeling, and further suggests that this region may have experienced less rigorous loading – with the caveat that it is still unknown how Haversian density translates to bone strengthening (Riggs et al. 1993; Currey, 2006). Additional support for this finding comes from previous studies on clavicle Haversian density, finding near‐identical density values at the midshaft as in our study (Stout & Paine, 1992). However, our study adds the comparative value of the sternal and acromial ends. It is possible that mediolateral variations in habitual loading could be explained by nearby muscle and ligament attachments, especially at the acromial end with its unique shape adaptations. For example, the deltoid, pectoralis major, sternocleidomastoid, and sternohyoid attach to the acromial and sternal ends of the clavicle (Mays et al. 1999). Moreover, surgical exploration of the region by Abbott & Lucas (1954) found that the middle region of the clavicle could be fully resected without functional disability, remarking that the essential stabilizing ligaments and muscles were not found in this area.

In contrast to the differences observed in cross‐sectional geometry and Haversian density, collagen fiber orientation is remarkably uniform across the bone. The preferred arrangement of collagen fiber bundles appears to be longitudinal across all three sampled regions, likely encompassing the entire shaft. This arrangement indicates that the clavicle shaft is reasonably well adapted to longitudinally directed tensile stresses during normal activity. That is to say, the shaft is optimized for tensile stresses while being underbuilt for high compressive loads. This result offers additional support for biomechanical analyses conducted by Stanley et al. (1988) that have shown that compressive loads on the clavicle are more likely to result in fracture via buckling than other styles of loading.

The results of the present study confirm previous findings on clavicle cross‐sectional geometric properties, while providing a novel histologic perspective that offers further insights into clavicle (and, by extension, pectoral girdle) biomechanics. Through the added context of Haversian density measurements, our descriptive investigation supports the initial hypothesis that the middle of the clavicle experiences different loading regimens than the acromial end and may have a reduced need for bone remodeling. This could be related to a lower degree of bone strain during normal activity, especially from acting muscle forces. Moreover, collagen fiber orientation offers a perspective on existing compression studies by Stanley et al. (1988) to show that the clavicle's predominantly longitudinal arrangement of collagen fibers is likely well adapted to normal tensile stresses, and perhaps torsion due to the slightly oblique nature of the matrix in some areas, but is vulnerable to atypically high compression that could lead to fracture throughout the length of the shaft.

Future studies on the topic of clavicle biomechanics and fracture risk will also seek to better characterize our understanding of other microstructural and material properties. Subtle differences in circular organization of longitudinal versus transverse fibers around the clavicle (e.g. anterior and posterior regions of the shaft) could also contribute to the structural integrity and behavior of this bone, offering an avenue of future investigation towards discovering which types of forces and in what directions are involved in catastrophic bone failure. The current study sought to understand medial‐to‐lateral variations in bone modeling, remodeling and overall organization, but there are myriad other factors that may explain observed fracture patterns, including bone mineral density (Andermahr et al. 2007). A better understanding of clavicle loading and breakage patterns may contribute to our ability to prevent this common diagnosis and inform design of safety equipment such as seatbelts, athletic shoulder pads or bullet‐resistant vests (Throckmorton & Kuhn, 2007; Harris & Spears, 2010).

Author contributions

B.A.P. and A.K.H. conceived the study; M.A.C., K.M.K., and A.K.H. collected the data; M.A.C. analyzed the data; M.A.C. drafted the manuscript with input from B.A.P. and A.K.H.; all authors gave final approval for publication.

Acknowledgements

This research was funded through the generosity of the USC Provost Undergraduate Research Fund and the USC Rose Hills Summer Research Fellowship (to M.A.C.) and the USC Undergraduate Research Associates Program (to B.A.P.). We thank N. Schultz for providing helpful comments on an early draft of the manuscript. We also thank our handling editor, Edward Fenton, as well as Timothy Bromage and one anonymous reviewer for their constructive feedback.

List of measurements

Specimen Region Percent transverse (%) Percent longitudinal (%) Haversian density Scaled J I max /I min %BA
CLA01 Sternal 5.73 94.27 10.71 11.63 1.073 70.78
Middle 6.07 93.93 12.49 7.97 1.357 80.69
Acromial 7.42 92.58 15.19 12.33 2.035 79.57
CLA02 Sternal 4.48 95.52 15.85 9.06 1.202 55.21
Middle 6.49 93.51 12.72 4.56 1.382 76.74
Acromial 3.40 96.60 13.42 9.41 1.542 73.45
CLA03 Sternal 18.73 81.27 17.12 5.76 1.997 75.50
Middle 18.83 81.17 16.85 6.91 2.465 76.77
Acromial 3.77 96.23 21.70 9.62 2.622 85.36
CLA04 Sternal 10.13 89.87 14.20 8.67 1.457 61.30
Middle 14.92 85.08 10.18 10.30 1.364 63.53
Acromial 13.64 86.36 11.82 14.57 1.890 62.97
CLA05 Sternal 6.78 93.22 15.47 9.05 2.406 58.08
Middle 4.72 95.28 15.64 – – –
Acromial 2.20 97.80 15.75 9.08 1.711 72.16
CLA08 Sternal 6.96 93.04 15.29 8.79 1.932 78.14
Middle 5.74 94.26 14.84 8.67 2.316 78.78
Acromial 9.24 90.76 16.15 9.88 2.012 85.08
CLA09 Sternal 5.86 94.14 14.04 6.03 1.071 75.94
Middle 4.25 95.75 16.57 6.64 1.207 84.10
Acromial 5.27 94.73 16.88 9.20 1.224 80.60
CLA10 Sternal 15.58 84.42 19.60 10.39 1.078 56.45
Middle 9.65 90.35 19.25 9.96 1.851 60.66
Acromial 4.47 95.53 20.35 8.26 2.005 74.98
CLA102 Sternal 6.18 93.82 13.59 9.00 1.737 56.13
Middle 9.09 90.91 11.82 7.72 1.416 63.68
Acromial 5.03 94.97 10.62 10.63 1.719 78.22
CLA12 Sternal 14.68 85.32 16.63 12.73 1.503 71.63
Middle 3.52 96.48 18.66 7.86 1.358 80.30
Acromial 4.63 95.37 19.42 10.55 2.696 81.82
CLA18 Sternal 12.06 87.94 16.23 – – –
Middle 14.14 85.86 14.49 – – –
Acromial 10.47 89.53 16.80 12.14 1.858 54.27
CLA19 Sternal 4.35 95.65 17.51 9.85 2.147 73.86
Middle 1.09 98.91 15.01 7.18 2.525 82.24
Acromial 3.41 96.59 15.91 13.75 2.948 77.31
CLA22 Sternal 5.66 94.34 10.89 21.95 1.228 71.76
Middle 4.35 95.65 12.20 13.62 1.874 80.88
Acromial 5.10 94.90 16.06 24.15 1.083 77.64
CLA23 Sternal 4.02 95.98 19.29 6.92 1.243 38.60
Middle 2.03 97.97 16.16 7.00 1.328 55.07
Acromial 21.55 78.45 18.21 9.18 1.644 59.50
CLA34 Sternal 8.60 91.40 14.49 15.29 1.488 77.42
Middle 6.96 93.04 14.23 17.32 2.627 80.07
Acromial 7.56 92.44 12.74 28.83 2.534 86.07
CLA35 Sternal 7.44 92.56 16.00 9.65 1.211 74.71
Middle 4.65 95.35 14.64 10.58 1.413 80.40
Acromial 5.73 94.27 13.86 14.91 2.092 85.16
Open in a new tab

References

  1. Abbott LC, Lucas DB (1954) The function of the clavicle: its surgical significance. Ann Surg 140, 583. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Andermahr J, Jubel A, Elsner A, et al. (2007) Anatomy of the clavicle and the intramedullary nailing of midclavicular fractures. Clin Anat 20, 48–56. [DOI] [PubMed] [Google Scholar]
  3. Bromage TG, Goldman HM, McFarlin SC, et al. (2003) Circularly polarized light standards for investigations of collagen fiber orientation in bone. Anat Rec (Hoboken) 274, 157–168. [DOI] [PubMed] [Google Scholar]
  4. Carando S, Portigliatti‐Barbos M, Ascenzi A, et al. (1991) Macroscopic shape of, and lamellar distribution within, the upper limb shafts, allowing inferences about mechanical properties. Bone 12, 265–269. [DOI] [PubMed] [Google Scholar]
  5. Doube M, Kłosowski MM, Arganda-Carreras I, et al. (2010) BoneJ: free and extensible bone image analysis in ImageJ. Bone 47(6), 1076–1079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Currey JD (2003) The many adaptations of bone. J Biomech 36, 1487–1495. [DOI] [PubMed] [Google Scholar]
  7. Currey JD (2006) Bones: Structure and Mechanics. Princeton: Princeton University Press. [Google Scholar]
  8. Duprey S, Bruyere K, Verriest JP (2010) Clavicle fracture prediction: simulation of shoulder lateral impacts with geometrically personalized finite elements models. J Trauma Acute Care Surg 68, 177–182. [DOI] [PubMed] [Google Scholar]
  9. Frost HM (2004) A 2003 update of bone physiology and Wolff's law for clinicians. Angle Orthod 74, 3–15. [DOI] [PubMed] [Google Scholar]
  10. Goldman HM, Bromage TG, Thomas CDL, et al. (2003) Preferred collagen fiber orientation in the human mid‐shaft femur. Anat Rec (Hoboken) 272, 434–445. [DOI] [PubMed] [Google Scholar]
  11. Harrington MA, Keller TS, Seiler J, et al. (1993) Geometric properties and the predicted mechanical behavior of adult human clavicles. J Biomech 26, 417–426. [DOI] [PubMed] [Google Scholar]
  12. Harris DA, Spears IR (2010) The effect of rugby shoulder padding on peak impact force attenuation. Br J Sports Med 44, 200–203. [DOI] [PubMed] [Google Scholar]
  13. Hogan HA (1992) Micromechanics modeling of Haversian cortical bone properties. Journal of Biomechanics 25(5), 549–556. [DOI] [PubMed] [Google Scholar]
  14. Inman VT, dec M Saunders JB, Abbott LC (1944) Observations on the function of the shoulder joint. JBone Joint Surg 26, 1–30. [DOI] [PubMed] [Google Scholar]
  15. Jowsey J (1966) Studies of Haversian systems in man and some animals. J Anat 100(Pt 4), 857. [PMC free article] [PubMed] [Google Scholar]
  16. Lamm ET (2013) Preparation and sectioning of specimens In: Bone Histology of Fossil Tetrapods: Advancing Methods, Analysis, and Interpretation (eds Padian K, Lamm E‐T.), pp. 55–160. Berkeley: University of California Press. [Google Scholar]
  17. Lee AH (2004) Histological organization and its relationship to function in the femur of Alligator mississippiensis . J Anat 204, 197–207. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lee AH, Simons EL (2015) Wing bone laminarity is not an adaptation for torsional resistance in bats. PeerJ 3, e823. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lee TC, Staines A, Taylor D (2002) Bone adaptation to load: microdamage as a stimulus for bone remodelling. J Anat 201, 437–446. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Mays S, Steele J, Ford M (1999) Directional asymmetry in the human clavicle. Int J Osteoarchaeol 9, 18–28. [Google Scholar]
  21. Nordqvist A, Petersson C (1994) The incidence of fractures of the clavicle. Clin Orthop Relat Res 300, 127–132. [PubMed] [Google Scholar]
  22. O'Brien FJ, Taylor D, Lee TC (2005) The effect of bone microstructure on the initiation and growth of microcracks. J Orthop Res 23, 475–480. [DOI] [PubMed] [Google Scholar]
  23. Patel BA, Ruff CB, Simons ELR, et al. (2013) Humeral cross‐sectional shape in suspensory primates and sloths. Anat Rec 296, 545–556. [DOI] [PubMed] [Google Scholar]
  24. Post M (1989) Current concepts in the treatment of fractures of the clavicle. Clin Orthop Relat Res 245, 89–101. [PubMed] [Google Scholar]
  25. Postacchini F, Gumina S, De Santis P, et al. (2002) Epidemiology of clavicle fractures. J Shoulder Elbow Surg 11, 452–456. [DOI] [PubMed] [Google Scholar]
  26. Rasband WS (1997. ‐2018) ImageJ. Bethesda: U.S. National Institutes of Health; https://imagej.nih.gov/ij/. [Google Scholar]
  27. Reilly DT, Burstein AH (1975) The elastic and ultimate properties of compact bone tissue. J Biomech 8, 393–405. [DOI] [PubMed] [Google Scholar]
  28. Riggs CM, Vaughan LC, Evans GP, et al. (1993) Mechanical implications of collagen fibre orientation in cortical bone of the equine radius. Anat Embryol 187, 239–248. [DOI] [PubMed] [Google Scholar]
  29. Rouhi G (2012). Biomechanics of osteoporosis: the importance of bone resorption and remodeling processes. In Osteoporosis. InTech.com.
  30. Smekal V, Oberladstaetter J, Struve P, et al. (2009) Shaft fractures of the clavicle: current concepts. Arch Orthop Trauma Surg 129, 807–815. [DOI] [PubMed] [Google Scholar]
  31. Stanley D, Trowbridge EA, Norris SH (1988) The mechanism of clavicular fracture. A clinical and biomechanical analysis. J Bone Joint Surg Br 70, 461–464. [DOI] [PubMed] [Google Scholar]
  32. Stout SD, Paine RR (1992) Histological age estimation using rib and clavicle. Am J Phys Anthropol 87, 111–115. [DOI] [PubMed] [Google Scholar]
  33. Throckmorton T, Kuhn JE (2007) Fractures of the medial end of the clavicle. J Shoulder Elbow Surg 16, 49–54. [DOI] [PubMed] [Google Scholar]
  34. Yamamura S, Hayashi S, Li ZL, et al. (2018) Investigations of cortical and cancellous clavicle bone patterns reveal an explanation for the load transmission and the higher incidence of lateral clavicle fractures in the elderly: a CT‐based cadaveric study. Anat Sci Int 93, 479–486. [DOI] [PubMed] [Google Scholar]

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

RESOURCES