Fax: (250) 960-5545
Circumferential variation in human second metacarpal cortical thickness: Sex, age, and mechanical factors
Article first published online: 30 APR 2002
Copyright © 2002 Wiley-Liss, Inc.
The Anatomical Record
Volume 267, Issue 2, pages 154–158, 1 June 2002
How to Cite
Lazenby, R. A. (2002), Circumferential variation in human second metacarpal cortical thickness: Sex, age, and mechanical factors. Anat. Rec., 267: 154–158. doi: 10.1002/ar.10099
- Issue published online: 30 APR 2002
- Article first published online: 30 APR 2002
- Manuscript Accepted: 13 MAR 2002
- Manuscript Received: 30 OCT 2001
- NSERC. Grant Number: OPG 0183660
- bone cortex;
Variation in cortical thickness (CT) in four quadrants of the human second metacarpal was investigated in a sample (100 males and 72 females, skeletal age 20 to 50+ years) from a 19th-century cemetery. Both left and right elements were studied (total N = 344). Multivariate analysis of covariance (MANCOVA) (for age, sex, and side, controlling for absolute size) was used to test the hypothesis of equality of thickness in the dorsal, palmar, medial, and lateral quadrants. Differences in regional CT posits localized regulation of resorption and formation adapting bone shape to functional loads, with implications for activity-modulation of skeletal senescence. The palmar cortex was found to be uniformly thicker in both sexes and both sides, and at all ages (young, middle, and old adult); the medial, lateral, and dorsal cortices did not differ significantly. Patterns of age-related loss occurring preferentially at the endocortical surface differed between men and women, with women showing significant declines across all age groups for all quadrants, and males only small decrements after middle age. The greater CT in the palmar quadrant corresponds to the region of maximum compressive strain in the second metacarpal for functions involving full flexion (grasping). Although the palmar cortex is thicker at all ages, women lose mass in that quadrant at the same rate as in other quadrants, suggesting that function does not offer protection against endocrinologically-mediated depletion of bone mass (postmenopausal osteopenia). Anat Rec 267:154–158, 2002. © 2002 Wiley-Liss, Inc.
The characterization of bone mass and age-progressive bone loss in the human second metacarpal using radiogrammetric methods has shown itself to be a simple, cost-effective method in a variety of investigative contexts, including the study of bone growth (Himes and Huang, 1993) and aging (Aguado et al., 1997), the functional analysis of asymmetry and lateral hand dominance (Roy et al., 1994), and as a reliable indicator of fracture risk in osteoporosis (Meema and Meindok, 1992). Recently, Fox et al. (1995) reported that cortical thickness (CT) within individual second metacarpals is asymmetrical, with the radial (lateral) cortex being thicker than the ulnar (medial) cortex in both the left and right hands, in both males and females. Sex differences were noted, in that bone loss occurred only at the radial cortex in men, but in both radial and ulnar cortices in women. It was also observed that loss progressed more rapidly with age from the radial cortex in both sexes (although more precipitously in females). These findings imply localized mediation of bone mass accretion and depletion. Fox and colleagues (1995) suggested that differential muscle attachment of the dorsal and palmar interossei contributed to differential CT in the MC II midshaft. For example, the second palmar and the second dorsal interossei attach to the ulnar surface, but only one head of the first dorsal interosseous attaches radially. In addition, they argued that distinct patterns of aging and disuse for extensor/abductor vs. flexor/adductor muscles could account for their observations. Their cross-sectional study, using anterior-posterior hand-wrist radiographs collected as a component of the Baltimore Longitudinal Study on Aging (BLSA), limited the analysis to the medial and lateral cortices. However, in a study of functional asymmetry using radiographs sampled from the same BLSA series, Roy et al. (1994) had previously noted that functional loading of the second metacarpal is predominantly flexion, in which case an analysis of the dorsal and palmar cortices would be more informative vis à vis the relation of mechanical loading and the mediation of bone mass, asymmetry, and senescence.
As noted, data for the dorsal and palmar cortices are unavailable from hand-wrist radiographs taken on living individuals, although newer noninvasive technologies, such as peripheral quantitative computed tomography (pQCT), should rectify this situation. However, at present many researchers with interest in the skeletal biology of past and present peoples have limited access to such technology, particularly in nonclinical fields (such as anthropology). Fortunately, data from all four anatomical quadrants can be obtained from radiogrammetric analysis of skeletal samples, unconstrained as to positioning in either the A-P or M-L planes. This study provides an analysis of age, sex, and side variation in dorsal, palmar, medial (radial), and lateral (ulnar) CTs from radiographs of a large 19th-century skeletal sample from southwestern Ontario, Canada. As such, it extends our understanding of localized skeletal aging within a class of data (i.e., radiogrammetric) comparable to extant clinical and anthropological studies, which is particularly relevant as our appreciation of the antiquity of osteopenic bone loss becomes more refined (Mays, 2000).
MATERIALS AND METHODS
The paired left and right metacarpals (total N = 344) from a total of 172 individuals (100 males and 72 females) were used for the analysis. The sample was partitioned into three nominal age cohorts: 1) young: 23 M, 23 F; 2) middle: 43 M, 29 F; and 3) old: 34 M, 20 F. This partitioning followed age and sex determination of individual skeletons using standard cranial and pelvic markers and a multifactorial approach (Saunders et al., 1992). The three nominal categories equate to interval estimates of ca. 20–35 years, 36–50 years, and 50+ years, corresponding to the periods of attainment of peak bone mass, onset of age-related bone loss, and the advancement of postmenopausal osteopenia. This sample (notably the females within it) has been shown to be more skeletally robust than a typical sample of modern adult cadavers (Lazenby, 1994). This difference reflects the more diverse and active lifestyle characteristic of “pioneers” compared to increasingly urbanized 20th-century men and women, and facilitates the discrimination of local, functional effects from systemic (e.g., endocrinological) components. Certain risk factors for bone loss associated with modern “westernized” populations, including sedentary behavior and iatrogenic complications of prescribed medications known to affect bone physiology (e.g., anti-inflammatory glucocorticoids), have little or no influence on the analysis, although other relevant factors such as smoking, alcohol consumption, and diet cannot be discounted.
Radiogrammetric data were collected from perpendicular anteroposterior (AP) and mediolateral (ML) views taken with a Philips Diagnost 66 (Philips Medical Systems, Eindhoven, The Netherlands), triple phase machine using a Lanex fine screen Kodak cassette (Eastman Kodak Co., Rochester, NY). Tube–film distance was held constant at 94 cm, with exposure set at 45 kV and 2.5 mA/sec. The midshaft of each metacarpal was located at one-half interarticular length and marked with an indelible pen on the bone surface. For views of the dorsal and palmar cortices, bones were placed with the medial surface adjacent to the film cassette; for views of the medial and lateral cortices, the posterior surface contacted the cassette. Small amounts of plasticine were used as necessary to replicate positioning among elements. A rectangular frame with wire filaments situated over the midshaft mark aided in identification of the midshaft site for subsequent measurement. Linear radiographic magnification was determined from a comparison of external widths from radiographs vs. those measured directly from the bone, and was found to be minimal (≤3.0%). Outer (periosteal) and inner (endosteal) diameters, as well as their respective CTs, were measured to 0.05 mm using a Helios dial caliper. Replicate measurements on a random sample of 10 individuals showed intraobservor error to be ca. 2.0% for periosteal diameters and ca. 4.0% for medullary diameters; the larger error in the latter instance reflects greater inconsistency in defining the endosteal border.
Four dependent variables—the dorsal, palmar, medial, and lateral CTs—were analyzed. The lateral and palmar CTs were directly measured from radiographs. The corresponding medial and dorsal thicknesses were determined algebraically—for example, as the difference of the ML periosteal diameter minus the sum of lateral CTs and endosteal diameter. This technique ensures that the sum of the cortices and the medullary cavity do not exceed total width. A matched-pairs test for a random sample of 30 images comparing observed vs. calculated medial and dorsal widths was nonsignificant (e.g., medial cortex, mean difference = 0.02 mm; t = 0.859; P <.80). Data were analyzed by multivariate analysis of covariance (MANCOVA), with age, sex, and side as the main effects, and with means adjusted by the covariate interarticular length (IAL), which is the length of the bone from the head to the deepest point of the articular depression for the trapezoid. All else being equal, larger bones tend to have larger CT values as an allometric consequence of size. As males are on average larger than females (Fairbairn, 1997), men are expected to have larger cortices than women. The covariate IAL, as a proxy for body size, accommodates this allometric effect such that variability in CT adjusted by IAL is more readily interpretable as a function of sex-specific patterns in mechanical loading (physical activity) and skeletal senescence independent of size. The correlation coefficients and regression slopes of each dependent measure with IAL were all significant and nondivergent, satisfying assumptions of a linear relationship and homogeneity of regression hyperplanes (Weinfurt, 1995).
MANCOVA has advantages over a univariate model involving sequential tests (Weinfurt, 1995). Primarily, it considers the intercorrelation among dependent variables; thus the significance of a main effect on, e.g., lateral CTs, is informed by associated values for dorsal, palmar, and medial CTs. It has also been argued that MANCOVA controls experiment-wise type 1 error-rate at the nominal level specified (e.g., 0.05) thereby avoiding inflation of alpha associated with multiple independent tests (e.g., the Bonferroni inequality). Weinfurt (1995) has noted, however, that this is the case only when the null hypothesis is true—in this case, that there are no significant differences in CT among the four locations. Thus, inflation of alpha remains an issue if the MANCOVA is significant. In this study, significant main effects were subsequently investigated independently using Tukey's honest significant difference (HSD) for unequal N, with alpha apportioned at 0.01 over the four dependent variables to maintain the experiment-wise error rate of 0.05 (Tabachnick and Fidell, 1989). All analyses were carried out using Statistica 5.1 (Statsoft, 1997).
“Sex,” “cohort,” and “side” were all significant main effects underlying variation in the CTs for the second metacarpal; in addition, there was a significant sex × cohort interaction (Table 1). Cohort accounted for the largest proportion of variance (15.1%) in the CTs, followed by sex (13.3%) and then side (6.2%). (Note that, unlike R2 values in a multiple regression, Eta2 values are non-additive.) As is evident in Table 2, for all quadrants, males had greater CTs than females even after covariate adjustment for body size. The cohort effect contrasts old vs. either young or middle cohorts; there were no significant young vs. middle contrasts, although loss of CT over all cortices was evident in females for these two groups (Fig. 1). Finally, the “side” effect was restricted solely to the palmar cortex. Among quadrants, palmar CTs were significantly greater than any other, a result which held whether comparing the entire dataset (ANOVA F = 25.425, df 3, 1,371, P < 0.0000), or within groups. Figure 1 depicts the decline in thickness for males and females (sides combined). Males tend to lose relatively little mass, and then only for lateral and medial cortices, between young and middle adult ages. Loss in the dorsal and palmar quadrants occurs only after ca. age 50; indeed, there is a modest accretion in thickness between young and middle ages. In females, bone loss occurs in all regions from young adulthood, with an increase seen in older ages.
|Factor||Rao's R||Eta2||df 1||df 2||P-level|
|Sex × cohort*||4.178||0.095||8||656||0.000|
|Sex × side||0.131||0.002||4||328||0.971|
|Cohort × side||0.309||0.007||8||656||0.963|
|Sex × cohort × side||0.788||0.019||8||656||0.613|
|Cohort: Y - M||.219||.060||.272||.623|
|Cohort: Y - O||.000||.000||.000||.000|
|Cohort: M - O||.000||.000||.000||.009|
This study failed to duplicate the lateral–medial dichotomy for CTs reported by Fox et al. (1995). Within sexes, the lateral, medial, and dorsal cortices are statistically uniform in thickness and pattern of change with age; the apparent distinction for medial CTs in “old” females (Fig. 1) is a function of the seemingly anomalous rate of loss for this region between young and middle age. As with Fox et al.'s study, differences in regional CTs cannot be explained by expansion at the periosteal surface, as no significant differences in either ML or AP periosteal diameters were observed in either males or females, or left or right sides. (However, it is important to note that a lack of statistical significance for age-progressive periosteal expansion does not imply a lack of biological significance, as each unit area of periosteal bone contributes to strength in an amount proportional to the square of its distance from the neutral axis (Lazenby, 1990).) Indeed, the reduction in bone mass in this element is a feature of the endocortical surface (Maggio, 1997), which is also the locus of the minimum effective strain in bending.
What is evident in the present study is a significant contrast between the palmar cortex and all other cortices. This result is in agreement with Kimura's (1990) study of regional CTs and mineral density, which also found CTs to be greater in the palmar (volar) cortex of the second metacarpal. Kimura concluded that “the second metacarpal has more magnitude of bending rigidity in the dorsovolar direction than in the radioulnar, and more intensity against an external force on the volar side…” (Kimura, 1990, p. 43). This line of reasoning is consonant with Roy et al.'s (1994) contention that the second metacarpal's dominant loading in flexion generates greater strain in the dorsopalmar plane. Experimental studies of long bone strain in various avian and mammalian models has shown that superposition of axial compressive and bending strains results in much larger strains in the concave (i.e., palmar) than in the convex (i.e., dorsal) cortex (Martin et al., 1998). However, the second metacarpal is more properly modeled as a cantilever rather than a column, fixed proximally and intimately with the trapezoid, with two degrees of freedom at the metacarpophalangeal (MCP) joint (Chao et al., 1989). In this model, deflection will be greatest distally, with resultant strain increasing toward the base. The dorsal cortex will be loaded primarily in tension, and the palmar cortex in compression.
The fact that the palmar cortex is thicker in both left and right, and male and female elements suggests that strain deformation is eccentric around the longitudinal neutral axis of the metacarpal (i.e., compressive and tensile strains are indeed unequal, with the former greater than the latter). This implies the presence of an axial compressive component, possibly generated by resistance of the extensor tendon mechanism in flexion. While axial loading of the second metacarpal may be low, the interossei (especially the palmar) are flexors of the MCP, assisted following initiation by the flexor digitorum profundus and flexor digitorum superficialis. While these extrinsic muscles generate greater force than the interossei (Chao et al., 1989), they do not have tendinous insertions on the MC II, but cross the MCP and proximal interphalangeal (PIP) joints to insert at the base of the distal phalange, and the anterior surface of the middle phalange in the case of flexor digitorum superficialis. Additionally, even though it is vested solely in soft tissue, the index lumbrical also aids in flexion at the second metacarpal MCP, and extension at the interphalangeal joints.
Fox et al. (1995) argued that differential attachment and action of the dorsal and palmar interossei generated a greater radial cortical mass initially, and its greater rate of loss subsequently. This argument was founded on the observation that the radial side of the metacarpal serves as attachment for the “small head” of the first dorsal interosseous muscle, and that the ulnar side was the site for the second dorsal interosseous and first palmar interosseous muscles. A corollary of their model argued for differential aging in the extensor/abductors (radial) vs. flexor/adductors (ulnar) muscles, with a greater “age regression” in the former and thus a relatively greater “disuse” impetus for radial cortical loss. However, in a sample of 46 males and females aged 21–58 years, Chao et al. (1989) found that for rays II–V, abduction strength was consistently greater than adduction strength. Indeed, adduction of the second metacarpal is constrained by the proximity of MC III, and abduction by the deep transverse metacarpal ligament. Consequently, the principal deformation of the second metacarpal occurs through flexion and extension of the wrist, and flexion and pronation of the digits at the metacarpophalangeal joint. (Pronation is principally a motion of rays IV and V, visible in the dorsum of the hand in fist clenching). While flexion is involved in the range of hand functions, from pinching/precision to grasping/power grips, it is the latter which generate maximum mechanical loading, and hence bone strain (Chao et al., 1989). Consequently, the differential distribution of cortex toward the palmar surface suggests adaptation of the second metacarpal to functions involving power grips and full flexion, and in particular to the compressive strains generated at the palmar surface with the aim of reducing those strains to biomechanically acceptable levels. In this regard it is noteworthy that in the females in the present study, the dorsal cortex was the thinnest of the four, a result also noted by Kimura (1990).
In addition to the regional pattern of CT variation, it is also of note that age changes in thickness vary by region and sex. Fox et al. (1995) noted that the radial cortex thinned preferentially with age, a pattern not seen in the present study (if we accept the relatively flat young–middle transition for medial thickness as possibly anomalous). Clearly, men and women differ in rates of loss at all locations (Fig. 1). Of greater interest is that while mass is relatively conserved in the dorsopalmar plane among men, such is not the case for women. It is unclear to what degree this sex difference in bone loss may be reflective of differences in either timing or pattern of “unloading” among women in this sample. The fact that the sample derives from a pioneering, 19th-century peri-urban population argues for a lifetime of comparatively strenuous physical activity (certainly in relation to contemporary standards). Previous work has shown that women in this sample have more robust second metacarpals, in terms of overall size, than a modern sample from Britain (Lazenby, 1994). Thus, the uniform loss of cortex in females may indicate a failure of activity-induced mechanical loading to preserve bone mass in the face of systemic endocrinological factors predisposing to a deficit of resorption over formation (i.e., post-menopausal osteopenia). Such a conclusion, however, seems inconsistent with current evidence reporting positive correlations with lifetime physical activity and upper limb and second metacarpal bone density (Osei-Hyiaman et al., 1999; Chilibeck et al., 2000).
In conclusion, this study has shown that regional differences in CT do exist at the same anatomical location (e.g., the midshaft) within single skeletal elements, which can be correlated to differentially distributed functional loading and mechanical strain. In the second metacarpal, the primary axis of functional loading occurs in a dorsopalmar orientation, with the greatest osteogenic potential occurring at the palmar cortex in response to elevated strain in full flexion associated with grasping functions. Such a result is of great import in comparative studies (e.g., bilateral asymmetry, sexual dimorphism, and interspecific analyses), but is perhaps less significant in, e.g., prospective studies of fracture risk, in which the significant indicator may be those regions of cortex that are relatively unprotected by mechanical loading (Meema and Meindok, 1992).
- 1997. Cortical bone resorption in osteoporosis. Calcif Tissue Int 60: 323–326. , , , .
- 1989. Biomechanics of the hand. Singapore: World Scientific. 191 p , , , .
- 2000. Effect of physical activity on bone mineral density assessed by limb dominance across the lifespan. Am J Hum Biol 12: 663–637. , , , , .
- 1997. Allometry for sexual size dimorphism: pattern and process in the coevolution of body size in males and females. Annu Rev Ecol Syst 28: 659–687. .
- 1995. Radial and ulnar cortical thickness of the second metacarpal. J Bone Miner Res 10: 1930–1934. , , , .
- 1993. Menarcheal status and bone development in Guatemalan girls. Am J Hum Biol 5: 515–521. , .
- 1990. Bilateral asymmetry of cross-sectional radiogrametric and microdensitometric measurements in the second metacarpal. Okajimas Folia Anat Jpn 67: 39–46. .
- 1990. Continuing periosteal apposition. I. Documentation, hypotheses and interpretation. Am J Phys Anthropol 82: 451–472. .
- 1994. Identification of sex from metacarpals: effects of side asymmetry. J Forensic Sci 39: 1188–1194. .
- 1997. Age-related cortical bone loss at the second metacorpal. Calcif Tissue Int 60: 94–97. , , , , , , , , .
- 1998. Skeletal tissue mechanics. New York: Springer. 392 p , , .
- 2000. Age-dependent cortical bone loss in women from 18th and early 19th century London. Am J Phys Anthropol 112: 349–361. .
- 1992. Advantages of peripheral radiogrammetry over dual-photon absorptiometry of the spine in the assessment of prevalence of osteoporotic vertebral fractures in women. J Bone Miner Res 7: 897–903. , .
- 1999. Influence of grip strength on metacarpal bone mineral density in postmenopausal Japanese women: a cross-sectional study. Calcif Tissue Int 64: 263–266. , , , , .
- 1994. Hand dominance and bilateral asymmetry in the structure of the second metacarpal. Am J Phys Anthropol 94: 203–211. , , .
- 1992. A test of several methods of skeletal age estimation using a documented archaeological sample. Can J Forens Sci Soc 25: 97–118. , , , , .
- Statsoft. 1997. Statistica. Tulsa: Statsoft.
- 1989. Using multivariate statistics. 2nd ed. New York: Harper Collins. 746 p , .
- 1995. Multivariate analysis of variance. In: GrimmLG, YarnoldPR, editors. Reading and understanding multivariate statistics. Washington: American Psychological Association. p 245–276. .