A three-dimensional microcomputed tomographic study of site-specific variation in trabecular microarchitecture in the human second metacarpal


Dr Richard A. Lazenby, Anthropology Program, University of Northern British Columbia, 3333 University Way, Prince George, BC Canada, V2N4Z9. T: 250 960 6696; F: 250 960 5545; E: lazenby@unbc.ca


Variation in trabecular microarchitecture is widely accepted as being regulated by both functional (mechanical loading) and genetic parameters, although the relative influence of each is unclear. Studies reporting inter-site differences in trabecular morphology (volume, number and structure) reveal a complex interaction at the gene–environment interface. We report inter- and intra-site variation in trabecular anatomy using a novel model of contralateral (left vs right) and ipsilateral (head vs base) comparisons for the human second metacarpal in a sample of n = 29 historically known 19th century EuroCanadians. Measures of bone volume fraction, structure model index, connectivity, trabecular number, spacing and thickness as well as degree of anisotropy were obtained from 5-mm volumes of interest using three-dimensional microcomputed tomography. We hypothesized that: (i) the more diverse loading environment of metacarpal heads should produce a more robust trabecular architecture than corresponding bases within sides and (ii) the ipsilateral differences between epiphyses will be larger on the right side than on the left side, as a function of handedness. Analysis of covariance (Side × Epiphysis) with Age as covariate revealed a clear dichotomy between labile and constrained architectures within and among anatomical sites. The predicted variation in loading was accommodated by changes in trabecular volume, whereas trabecular structure did not vary significantly by side or by epiphysis within sides. Age was a significant covariate only for females. We conclude that environmental and genetic regulation of bone adaptation may act through distinct pathways and local anatomies to ensure an integrated lattice of sufficient mass to meet normal functional demands.


Site-specific variation in cortical and trabecular bone properties has been documented across different levels of organization, including collagen fiber orientation (Skedros & Hunt, 2004), cortical porosity and Haversian remodeling (Thomas et al. 2006), bone mineral density (Nonaka et al. 2006) and trabecular microarchitecture (Sran et al. 2007). Such studies point to a correspondence between skeletal microanatomy and local experience of functional loading, in keeping with the basic tenets of ‘Wolff's Law’ (Ruff et al. 2006). For example, Lazenby et al. (2008a) found significant right-biased directional asymmetry in trabecular bone volume fraction, number and connectivity corresponding to handedness in the human second metacarpal distal epiphysis. Although mechanical function is considered the prime determinant of skeletal mass and form through growth and adulthood (Tanck et al. 2001; Ryan & Krovitz, 2006), the role of genetic regulation in skeletal biology is receiving greater scrutiny (Robling et al. 2006). For example, variation in trabecular bone volume fraction, connectivity and anisotropy has been shown to exist among different inbred mouse strains subjected to similar mechanical environments (Bouxsein et al. 2004).

Various approaches to the question of functional vs genetic regulation of trabecular bone properties have been investigated, including the documentation of site-specific variation within and among skeletal elements. Rupprecht et al. (2006) showed that the trabecular bone structure for three separate volumes of interest (VOI) within the human calcaneus changes independently with age. The transition from plate-like to rod-like structure and reduction in bone volume fraction appear highly localized within this element and within individuals, pointing to a primary role for mechanical loading. In an experimental model using three genetically heterogeneous mouse strains, Judex et al. (2004) found a high degree of site specificity in femoral cortical and trabecular bone mass, with large differences appearing in some regions and similar values in others (e.g. the metaphyseal trabecular volume was ca. 400% greater in one strain but the epiphyseal volume was almost identical). Judex et al. (2004) favored a multi-gene regulatory model to account for these regional differences, although they admit that this argument (p. 605) ‘seems overly complex given the great number of anatomical sites within and across bones’. Such studies point strongly to a prominent role for function as an arbiter of trabecular mass and form. However, Bouxsein et al. (2004) examined the trabecular architecture in two mouse strains and found a complex and heterogeneous association of bone properties to mapped quantitative genetic loci (19 quantitative trait loci (QTL) on 13 autosomes), a number of which were distinct from QTL previously identified for femoral and vertebral bone mineral density in the same strains. As QTL are DNA markers for the polygenic inheritance of phenotypic traits, the association of multiple QTL with trabecular variation supports the multi-gene regulatory model proposed by Judex et al. (2004).

It is clear that trabecular bone is continually altered with aging, mechanical loading and pathology against a specific genetic background, and thus the analysis of local architecture is essential in obtaining a fuller appreciation of the contribution of trabecular structure to mechanical competence, as well as its pathophysiology (Stauber et al. 2006). In the present study, we use an ipsi- and contralateral single element model (the human second metacarpal) to explore the relationship of site-specific variation in functional loading and the proximate phenotype (Bouxsein et al. 2004) of local trabecular architecture.

The second metacarpal is the largest of four short tubular bones defining the longitudinal and transverse arches of the palm and is of interest both phylogenetically in documenting the shift in hand use from locomotion to manipulation (Lazenby et al. 2008b) as well as clinically, given that nearly 10% of all skeletal fractures occur in the metacarpals and phalanges (Chin & Vedder, 2008).

Anatomically, metacarpals consist of a quadrilateral base, cortical shaft, neck and ‘cam-shaped’ head. Distally, deep transverse ligaments bind adjacent metacarpal heads and collateral ligaments cross the metacarpophalangeal (MCP) joint. Dorsal and volar ligamentous attachments also bind the bases of each metacarpal to each other and with the contiguous bones in the distal carpal row at the carpometacarpal (CMC) arthrosis. Although the fourth and fifth metacarpal CMC joints are capable, respectively, of ca. 15° and 25° of motion, the second and third CMC joints are rigidly bound, having little if any independent movement. The base of the second metacarpal is further stabilized in articulation with the trapezoid, trapezium, capitate and third metacarpal by insertions of the extensor carpi radialis longus and flexor carpi radialis tendons. The configuration of the human index metacarpal base is unique among hominoids (Tocheri et al. 2005, p. 582) in having more transversely oriented facets for articulation with the trapezium and capitate, allowing for ‘distribution of load between the second metacarpal and these two distal wrist bones’. In an experimental study of the effects of trapezoidectomy on index metacarpal stability, Wright et al. (2006) report that normal axial compressive loading across the second carpometacarpal arthrosis is ca. 125 N.

Distally, the multiaxial condyloid MCP joints are capable of flexion, extension, adduction and abduction (and, in combination, circumduction). At about 70° of MCP joint flexion, tautness in the collateral ligaments stabilizes the finger against radioulnar deviation, enhancing ‘power pinch and grip’ (Chin & Vedder, 2008, p. 2). Compressive stresses at the index MCP joint during strenuous manipulation could reach 3.0–4.5 N mm−2, similar to that of weight-bearing elements in the lower limb (Tamai et al. 1988).

Thus described, the functional anatomy of the proximal and distal second metacarpal arthroses suggests that mechanical loading will be both quantitatively and qualitatively different between the ipsilateral head and base. The high degree of anatomical constraint characteristic of the CMC suggests that loading between contralateral bases will vary primarily in magnitude; the loading variation between contralateral heads will reflect primarily quantitative differences due to handedness. As such, we hypothesize that: (i) metacarpal heads should have a more robust trabecular architecture than bases within sides reflecting the more diverse loading environment at the MCP joint and (ii) given the preponderance of right-handedness within human populations (Lazenby, 2002), these ipsilateral differences between epiphyses will be larger on the right side than on the left side.

Materials and methods

A complete description of the research design is given in Lazenby et al. (2008a). Briefly, 14 male and 15 female paired second metacarpals were selected from a well-studied 19th century EuroCanadian skeletal collection, having excellent preservation and showing no evidence of traumatic or physiological pathology. With the exception of five females, all are historically documented individuals. Females ranged in age from 17 to 67 years ( = 36.27; SD = 12.64) and males from 20 to 75 years ( = 50.07; SD = 20.44). The head and base of each metacarpal were removed from the diaphysis on a Struers Minitom® slow-speed saw, cleaned using ultrasonication and allowed to air-dry.

Microcomputed tomographic images of each epiphysis were acquired with a SkyScan 1072 cone-beam micro-CT scanner (Aartselaar, Belgium) at the University of Calgary 3D Morphometrics Laboratory (Fig. 1). The protocol employed x-ray tube settings of 100 kV and 98 µA, an exposure time of 5 s per image with three-frame averaging to improve signal-to-noise ratio and a rotation step of 0.90°. The nominal isotropic resolution was 19 µm. A 1-mm aluminum filter and beam-hardening correction algorithm were used to compensate for artifacts associated with the use of a polychromatic x-ray source. Serial 8-bit 1024 × 1024 pixel image sequences were reconstructed using a cone-beam algorithm (SkyScan Cone Recon® software).

Figure 1.

Three-dimensional volumetric reconstructions of the second metacarpal head (a) and base (b). The complex morphology of the base derives from its constrained articulation with three carpals and the third metacarpal.

The VOI (5 mm3) were sampled from each epiphysis, comprising 259 serial images. The position of the VOI was determined with regard to a scout image acquired at the time of scanning. A reference slice was positioned within the epiphysis at approximately the midpoint of the epiphyseal trabecular mass. A median filter was applied to the image stack to reduce image noise. Following filtration, the images were segmented using global thresholding defined with respect to the inter-peak minimum of grey-scale values for each VOI. Although the use of a global vs a local threshold has been the subject of some debate, recent studies have shown that, at scanning and reconstruction resolutions in the range employed in this study (ca. 20 µm), both approaches yielded comparable accuracy in characterizing trabecular microarchitecture (Kim et al. 2004; Waarsing et al. 2004).

Data were collected using SkyScan's proprietary software CTan®. Both volumetric [bone volume fraction (BV/TV), trabecular number (Tb.N) and trabecular separation (Tb.Sp)] and structural [structure model index (SMI), connectivity (Tb.Pf), thickness (Tb.Th) and degree of anisotropy (DA)] variables were assessed. The reader is referred to Lazenby et al. (2008a), Fajardo et al. (2002) and Ketcham & Ryan (2004) for a complete description of these variables, commonly measured in micro-CT studies of trabecular bone.

We assessed reproducibility using Dahlberg's method error statistic (Dahlberg, 1940):


where d is the difference between repeat observations and n the sample size, in this case five metacarpal heads. For the trabecular variables examined, method error ranged from 1.746% for BV/TV to 0.005 mm for Tb.Th and the correlation (r) between measures ranged from 0.735 for SMI to 0.962 for DA, indicating very good to excellent reproducibility for these measures.

The questions posed by this study are, simply put, in what respect and to what degree is site-specific variation in trabecular microarchitecture expressed between the second metacarpal epiphyses? The ‘global’ null hypothesis of an absence of within-element, between-site differences was tested using ancova by sex and side with epiphysis as factor and age as covariate, using Systat 11 with α = 0.05. The coefficient of variation was used to evaluate site-specific variability.


Tables 1 and 2 summarize the descriptive (mean ± SD) and ancova results, respectively. Compared with bases, metacarpal heads had a significantly higher bone volume fraction in both left and right sides, and in males and females alike. The elevated BV/TV in the head was associated with a significantly greater Tb.N and a concomitant reduction in Tb.Sp. Tb.Th did not differ among epiphyses. Values for Tb.Pf were invariably lower in the head, indicating a more well-connected lattice structure, although significantly so only for the female right metacarpal. No significant differences occurred in the head vs base comparison for SMI. Trabeculae in the metacarpal base were significantly more isotropic in orientation, with DA values closer to 0.0. Age was a significant covariate only for females, influencing the SMI in both hands and Tb.Pf and Tb.Th in the left hand. Figure 2 reports the absolute difference between ipsilateral epiphyseal means by side and sex. None of the differences were significant as determined by a t-test (not reported). Contrary to our second hypothesis, the head/base differences for the right metacarpal were not marked, favoring only bone volume fraction and connectivity in females (although age was a significant factor for several left metacarpal variables, including Tb.Pf). Figure 3 illustrates a typical pattern for all four epiphyses as presented in a series of 5-mm cubic VOIs from a 20-year-old male.

Table 1.  Descriptive statistics by sex, side and epiphysis
  1. BV/TV, bone volume fraction (%); Tb.Pf, connectivity (mm−1); SMI, structure model index; Tb.N, trabecular number (mm−1); Tb.Th, trabecular thickness (mm); Tb.Sp, trabecular separation (mm); DA, degree of anisotropy.

RightBV/TV19.10 (7.72)11.22 (5.06)19.32 (4.72)10.19 (2.91)
Tb.Pf 3.96 (3.35) 4.52 (1.67) 2.99 (1.95) 4.31 (0.91)
SMI 0.85 (0.42) 0.88 (0.18) 0.75 (0.28) 0.84 (0.13)
Tb.N 1.21 (0.33) 0.71 (0.23) 1.21 (0.24) 0.65 (0.16)
Tb.Th 0.16 (0.04) 0.15 (0.03) 0.16 (0.03) 0.16 (0.02)
Tb.Sp 0.57 (0.11) 0.87 (0.20) 0.61 (0.11) 0.96 (0.21)
DA 0.23 (0.08) 0.16 (0.07) 0.26 (0.05) 0.18 (0.05)
LeftBV/TV18.55 (7.75)10.72 (4.27)17.01 (4.83) 9.91 (3.21)
Tb.Pf 3.85 (3.25) 4.25 (2.00) 3.84 (1.99) 4.38 (1.06)
SMI 0.82 (0.39) 0.82 (0.23) 0.86 (0.26) 0.83 (0.12)
Tb.N 1.19 (0.32) 0.68 (0.20) 1.14 (0.23) 0.63 (0.14)
Tb.Th 0.15 (0.04) 0.16 (0.03) 0.15 (0.03) 0.16 (0.03)
Tb.Sp 0.59 (0.11) 0.98 (0.35) 0.62 (0.10) 0.98 (0.18)
DA 0.24 (0.09) 0.17 (0.08) 0.27 (0.11) 0.19 (0.05)
Table 2. ancova results for ipsilateral differences by side and sex
H – BEffect FP =Cov FP =H – BEffect FP =Cov FP =
  1. H – B, head – base; effect, epiphysis; covariate (Cov), age.

  2. BV/TV, bone volume fraction (%); Tb.Pf, connectivity (mm−1); SMI, structure model index; Tb.N, trabecular number (mm−1); Tb.Th, trabecular thickness (mm); Tb.Sp, trabecular separation (mm); DA, degree of anisotropy.

RightBV/TV 7.878 9.8220.0040.0320.859 9.12640.2430 0.7460.395
Tb.Pf–0.554 0.3020.5880.4890.491–1.319 5.9230.022 2.4080.132
SMI–0.025 0.0410.8400.9150.348–0.086 1.3410.257 5.0480.033
Tb.N 0.49721.33300.6570.425 0.55855.0340 0.2390.629
Tb.Th 0.002 0.0180.8952.5610.122 0.005 0.2860.597 3.8380.061
Tb.Sp–0.30829.01903.6600.067–0.35535.9610 2.1890.151
DA 0.070 6.3380.0190.0140.907 0.08321.5610 3.0880.090
LeftBV/TV 8.72510.5500.0030.0050.9437.10324.0090 2.8810.101
Tb.Pf–0.396 0.1490.7020.8330.370–0.537 1.0190.322 6.4800.017
SMI–0.005 0.0020.9631.6510.211 0.029 0.2150.64611.1250.002
Tb.N 0.50524.22300.7780.386 0.50451.3730 0.2470.623
Tb.Th–0.003 0.0560.8151.8510.186–0.007 0.5560.462 5.5150.026
Tb.Sp–0.39117.77703.0940.091–0.36450.1960 2.1510.154
DA 0.072 5.3530.0290.1910.666 0.082 7.1150.013 2.6200.117
Figure 2.

Absolute differences between ipsilateral epiphyseal means by side and sex. BV/TV, bone volume fraction; Tb.Pf, connectivity; SMI, structure model index; Tb.N, trabecular number; Tb.Th, trabecular thickness; Tb.Sp, trabecular separation; DA, degree of anisotropy.

Figure 3.

The typical pattern of findings for all four epiphyses as presented in a series of 5-mm cubic volumes of interest (VOI) from a 20-year-old male. Note variation in both volumetric measures and structural properties (e.g. plates vs rods).


Although the paradigm of adaptation of bone structure to its mechanical environment has held pride of place for over a century of research in skeletal biology (Frost, 2001), there is a rapidly growing body of evidence for a prominent role for genetic mitigation of normal (Marshall et al. 2008) and pathological bone response to mechanical loading (Zhang et al. 2008). There is also an increasing appreciation of the potential for modulation (up- or down-regulation) of genetic signals for bone formation and resorption by local mechanical loading history (Zhi et al. 2008).

In our study, we found a consistent pattern of labile and constrained morphology repeated in both left and right sides, and males and females alike. Variables related to mass (bone volume fraction, trabecular number and spacing) show a distinct difference between head and base, with the head having a larger bone volume fraction owing to greater trabecular number. The mechanical significance of BV/TV is well appreciated. Across four sites in the human appendicular and axial skeleton, Ulrich et al. (1999) found that BV/TV was the major contributor to mechanical aptitude, accounting for up to 82% of variance in trabecular strength. Van Lenthe et al. (2006) and others (Giesen et al. 2004) have found BV/TV to be highly predictive of bone stiffness, whereas Bevill et al. (2006) found that a decline in BV/TV was the most significant predictor of reduction in strength under large-deformation loads.

With respect to DA, in all cases the metacarpal head was found to be significantly more anisotropic relative to the base in our sample, indicating that distal epiphyseal trabecular orientation was adapted to a more uniform loading history. This was a somewhat surprising outcome, given that the MCP joint experiences more varied loading than occurs at the CMC articulation. One possible explanation is that the greater isotropy (absence of preferred orientation) found in the base is determined by proximal metacarpal articular shape. As noted above, the base contacts three elements in the distal carpal row, as well as the third metacarpal. Although movement is highly constrained among these joints, load transfer across the complex CMC may translate as isotropy within the trabecular mass beneath the external cortical shell.

Of equal interest are those features of trabecular architecture that did not differ between ipsilateral epiphyses in spite of the very different functional contexts to which they are exposed (left and right; male and female) and which may suggest mitigation through greater genetic constraint. With Tb.Pf in the female right metacarpal as the only exception, those variables indicative of structure (connectivity, SMI and trabecular thickness) did not differ significantly between the head and base (although the distal epiphysis tends to have a better connected, plate-like lattice; Fig. 3). The nearly uniform trabecular thickness among the ipsi- and contralateral epiphyses is consistent with the model of ‘constant trabecular size’ proposed by Swartz et al. (1998) in their theoretical and empirical study of inter-specific variation in cancellous bone architecture in mammals across five orders of magnitude of body size variation. They proposed that trabecular bone adapts to change in functional demand primarily through the addition of more trabeculae and that the demands of calcium metabolism place an upper threshold on trabecular size. Applied to the present study, this adaptive response would account for the greater trabecular number and BV/TV in the metacarpal head noted above. Bouxsein et al. (2004) observed no difference for trabecular thickness between C57BL/6J and C3H/HeJ mouse strains in their comparative study of trabecular microarchitecture, although other volume and structural measures differed significantly. Giesen et al. (2004) found no differences for trabecular thickness in specimens excised from the manidibular condyles of human dentate and edentate subjects, although the latter subjects (in which the level of loading was presumably lower given a lack of rooted teeth) did have a lower BV/TV and a more rod-like structure.

Not unexpectedly, for some variables age was a significant covariate within females, the sample of which includes a number of subjects who would be peri- or postmenopausal. Interestingly, age appears to impact only structural and not volumetric variables. This is perhaps understandable if, as we suggest, volumetric properties such as BV/TV and Tb.N are more responsive to mechanical regulation and would be maintained through functional loading. Indeed, this would account for the fact that the influence of age is felt more on the left rather than the right metacarpal given the human propensity for a right-hand bias (handedness).

Our argument that there are measurable differences between functionally labile and constrained trabecular architectures not only between but within anatomical sites begs the following question: why would structural properties such as trabecular connectivity and SMI appear to be relatively isolated from mechanical input in comparison to a variable such as bone volume fraction? This question is complicated by clinical and experimental studies that point to sensitivity of structural properties to interventions, either prophylactic and/or exercise (increased loading or unloading). For example, Fox et al. (2005) reported increases in bone volume fraction, connectivity and a shift to more plate-like structure in iliac crest biopsies in postmenopausal women following 18 months of parathyroid hormone 1–84 administration, whereas Chen et al. (2008) found that alfacalcidol, a vitamin D metabolite, inhibited bone loss and increased bone formation (bone volume fraction and trabecular thickness) in the lumbar vertebrae of bipedal and sedentary female rats. Examining the effect of unloading on trabecular and cortical architecture in male BALB and C3H mice strains, Squire et al. (2004) found that both strains showed a similar level of response to loss of mechanical loading, with reductions in metaphyseal and epiphyseal BV/TV, Tb.N and Tb.Th. Interestingly, these results for male mice differed from those found for females of the same strains subjected to the same protocols (Judex et al. 2004), suggesting a complex gender/gene/anatomical site interaction. Clinically, mechanical unloading due to spinal cord injury resulted in reduced bone volume and structure (thickness) (Modlesky et al. 2004), although notably Tb.Th declined only in the distal femur and not in the proximal tibia. Similarly, increased loading due to targeted exercise (gymnastics) increased apparent bone mineral content, bone volume and trabecular number but not trabecular thickness in the proximal tibia of female college athletes (Modlesky et al. 2008).

The degree to which the intervention research applies to the current study is somewhat of an open question, as our subjects were drawn from a mostly urban, non-sedentary 19th century population unencumbered with the ‘conveniences’ of modern living (motorized transport, processed foods, a diverse array of over-the-counter and prescribed medications). However, and importantly, the clinical and experimental data do suggest that the primary response to either drug or exercise intervention occurs through altered bone volume mediated through an increase/decrease in Tb.N and, to a lesser and more variable degree, through changes in Tb.Th and connectivity. This pattern broadly agrees with our results for the contra- and ipsilateral second metacarpal epiphyses. It seems likely therefore that a fundamental objective of adaptive bone regulation is to maintain the integrity of the lattice structure through conservation of connectivity, thickness and the SMI while maintaining responsiveness to fluctuations (increase or decrease) in loading via volume and number.

There are several limitations to the present study. Among these is our presumption that the right metacarpal would experience a more diverse loading history as a function of handedness. Although reasonable at the population level, in fact, we do not know the hand preference of these particular individuals, derived from an historic archaeological sample. Previous research with this sample clearly points to a strong directional asymmetry consistent with right-handedness in both trabecular (Lazenby et al. 2008a) and cortical (Lazenby, 1998) structure. We also do not know whether these individuals suffered any metabolic or endocrine disorders that may have affected bone physiology. However, as noted earlier, no manifestations of disorders known to be grossly or histologically evident in skeletal remains (e.g. hyperparathyroidism, vitamin D deficiency) were observed. It is also the case that our relatively small sample of males and females covers a broad age range. It would be of interest to know what effect, if any, a more comprehensive sampling of adults across the lifespan would be revealed with regard to age and trabecular physiology.


Characterization of trabecular microarchitecture in proximal and distal epiphyses in the human second metacarpal reveals patterns of variation partitioned between a functionally responsive and conserved morphology. The unique combination of anatomy at the CMC and MCP arthroses in a context of human functional laterality affords the opportunity to investigate the impact of varied loading environments in ipsilateral and contralateral epiphyses. As hypothesized, the second metacarpal head reveals a more robust architecture in keeping with a more diverse mechanical loading environment. However, this is true only with regard to volumetric measures, including bone volume fraction and trabecular number, and in terms of greater anisotropy for trabecular orientation. In contrast, structural measures, such as connectivity, thickness and relative ‘plateness’ appear to be relatively protected from variation in mechanical loading. The existence of functionally constrained properties within a diverse loading environment suggests mitigation of mechanical input through genetic regulation, although our study design does not permit a direct test of that hypothesis.


Support for the research was provided by grants from the Natural Sciences and Engineering Research Council (R.A.L. and B.H.), the Canadian Foundation for Innovation, Alberta Innovation and Science and the Alberta Heritage Fund, Genome Canada and Genome Alberta (B.H.).