The objectives of this study were to determine if mineralization density demonstrated patterned variation around the cortex of the human femur, and to characterize this variability in mineralization density patterning within an adult population sample. The BSE-SEM methodology afforded high resolution, so that even small changes in mineralization density could be detected (Boyde & Jones, 1996). Significant age and sex differences were demonstrated, along with limited consistencies in the location of high and low average mineralization bone through the cortex. These were broadly similar to the observations of Portigliatti-Barbos et al. (1983) of mineralization patterning at the mid-shaft femur. However, the most interesting finding of this study may be the characterization of the degree of individual variation in mineralization density that cannot be explained by age or sex. After discussing each of the contributing factors analysed in explaining the variation (age, sex and location in the cortex), the significance of the additional variation will be addressed.
Age was a significant contributor to the variability identified in this sample. Mineralization density decreased with adult age, most evident between males of all age groups and between young and middle-aged females. This finding is consistent with previous studies that focused on age trends in mineralization among adults (e.g. Reid & Boyde, 1987; Simmons et al. 1991), and probably reflects accumulations of increasing numbers of remodelling events. Qualitative examination of the images in this study suggests that the high degree of mineralization in the young age individuals relates to the retention of large amounts of primary circumferential bone in this group, particularly on the medial aspect of the bone cortex. As this primary bone eventually becomes remodelled with increasing age, mineralization decreases.
Not all aspects of the cortex demonstrated such age-associated decreases. For instance, whereas posterior and medial aspects decreased in mineralization with age, the antero-lateral aspect did not. These regional differences may be a reflection of localized differences in modelling and remodelling rates through the lifespan (see discussion of location in cortex, below).
Sex differences were identified in the sample in both middle and older age groups. On average, middle age group males showed a higher degree of mineralization than females, but only when all circumferential rings and sectors were pooled. Some regions of the cortex (i.e. anteriorly) of females of this age group tended to have higher mineralization density than males. In the oldest age group, females showed significantly higher mineralization than males within each of the circumferential rings. Moreover, mineralization significantly decreases between middle and older age groups in males, but not in females.
Few studies have investigated sex differences in mineralization density with age. Boyde et al. (1995b) identified higher mineralization values in the iliac crests of females relative to males, but this study did not employ external standards and the tissue studied was taken at autopsy immediately deep to the iliac crest, and therefore not comparable with all other ‘histomorphometric’ studies of iliac crest biopsies, which are taken 1–2 cm below the crest. The crestal zone contains much residual calcified fibrocartilage and Sharpey fibre bone. Baud & Gossi (1980; as cited in Meunier & Boivin, 1997) suggested that females show decreased mineralization density in the iliac crest at an earlier age than males. Hobson (1998) demonstrated a small shift towards higher mineralization bone in the mandibles of males relative to females, whereas Kingsmill & Boyde (1998) found no gender differences in the mandible. It is difficult to compare these studies to one another, as they represent different skeletal elements, but Kingsmill & Boyde (1999) used the same methodology as the present study. They found that mineralization density in different skeletal elements within the same individual did not correlate with one another, suggesting that mineralization differences do not primarily reflect systemic factors. Rather, Kingsmill & Boyde (1998) found that site-specific loading experiences showed a strong relationship to mineralization variability. Examination of the localized sex differences in mineralization density identified in the present study may be examined in the future relative to lower limb functional loading differences between the sexes (e.g. Kerrigan et al. 2000).
We can also examine the results of the current study in light of their possible relationship to bone turnover differences between adult males and females, particularly at female peri- and post-menopausal ages. Some studies of histomorphometric indices have found lower bone turnover, especially lower bone formation, in adult females relative to adult males (Meunier et al. 1976; Dahl et al. 1988; Schnitzler et al. 1990). Schnitzler et al.'s (1990) study of iliac crest biopsies also demonstrated that males lacked erosion surface increases with age, suggesting a closer balance between resorption and formation. Studies of biochemical markers of bone turnover, specifically those of bone formation, are inconsistent. Some have reported higher levels of turnover in males relative to females (Duda et al. 1988; Vanderschueren et al. 1990; Resch et al. 1994; Henry & Eastell, 2000), whereas others report the opposite (Epstein et al. 1984). One study demonstrated higher levels of bone resorption markers in males relative to females, but the difference was not significant when body size was taken into account (Henry & Eastell, 2000).
Together, these studies tend to support an increase in bone turnover in males relative to females, suggesting more frequent and more efficient renewal of bone in males. Athough females may lose bone at an equal or higher rate than males, particularly along the endosteal surface of the bone, they do not replace it to the same extent as males. Consequently, females may accumulate areas of hypermineralized bone that escape the remodelling process (Schnitzler, 1993). Unfortunately, the automated methods employed in this study (in which rings and sectors used to divide the cortex were generated by the Optimas macro) limited our ability to interpret equivalent areas of bone tissue at the endosteal surface, or to account for tissue lost from endosteal resorption (for further discussion of this issue, see Goldman et al. 2003). Nevertheless, the present results demonstrate higher mineralization density bone in older females relative to males and highlight a potentially important difference in bone ageing that is rarely addressed in the bone biology literature.
In order to study this further, it is worthwhile to consider variation in mineralization density relative to that of porosity, as the distribution of both relate to variations in remodelling rate. Previous studies of this same research sample have demonstrated increases in total subperiosteal porosity with age, due primarily to increases in pore size, rather than pore number (Stein et al. 1999). Moreover, this increased porosity is seen at an earlier age in females than in males (Feik et al. 1997). The present study demonstrates increased proportions of bone with high mineralization density bone in older females relative to males of the same age group. Focusing on the relationship between the distributions of these two variables will provide added information on age and sex differences in remodelling rate.
On a related note, given that the bone cortex generally becomes more porous with age, a methodological concern of any study relying on edge-detection technology to differentiate bone from non-bone (as was done here) might be that the proportion of edges to bone would increase with age, thereby increasing error in older individuals. However, because increases in porosity appear to be primarily due to increases in pore size, rather than pore number, this is unlikely to be an issue. Moreover, the resolution of the BSE-SEM images, combined with the relatively low magnification of the images, renders any small increase in the proportion of edge to bone to be of little consequence.
Location in cortex
Location (ring and sector) was also significant in explaining the variability in degree of mineralization. Specifically, the periosteal ring was consistently the least mineralized relative to bone located in the mid-cortex and endosteal cortex. In all age groups, the antero-lateral aspect of the bone tended to be the least mineralized and the postero-medial aspect most mineralized, though these results were only significant in the periosteal and mid-cortex of the youngest age group. The significant differences in mineralization density between aspects of the cortex may reflect regional differences in bone turnover within the femur, at least within younger adults in the sample. This pattern is broadly consistent with that identified by Portigliatti-Barbos et al. (1983) in two male specimens, aged 43 and 46 years. These authors suggested a relationship between low mineralization in the anterior and lateral aspects of the bone with high remodelling rates induced by tensile stresses. Studies of other skeletal elements – for instance the human proximal femur (Loveridge et al. 2000), human mandible (Kingsmill & Boyde, 1998) and horse radius (Riggs et al. 1993), have also demonstrated consistent locational variation in mineralization density that has been attributed to regional differences in functional loading. Further study of loading at the human femoral mid-shaft should improve our ability to interpret the regional variability identified in this study.
Range of mineralization density variation
The range of mineralization densities characterizing this sample was similar to results reported in previous studies of adult humans (Baud & Gossi, 1980; Roschger et al. 1998; as cited in Meunier & Boivin, 1997). In the present study, coefficients of variation were shown to increase with age, particularly among females. This finding is supported by Sharpe (1979), but contradicts studies that have observed a decrease in variability in mineralization with age (i.e. Amprino & Engstrom, 1952; Reid & Boyde, 1987, in human ribs). The latter studies, however, included subadults in their samples as well.
The increase in variability in the middle and older age groups in this study may reflect increased heterogeneity in the degree to which different individuals of the same age turnover their bone. Whereas some individuals in the middle age group retain large portions of highly mineralized primary circumferential bone, others have extensively remodelled their cortices. The increased variability in degree of mineralization, particularly among older females, may point to an increase in areas of both hypo- and hyper-mineralization (Sharpe, 1979). Alternatively, as this sample is likely to contain both osteoporotic and ‘normal’ individuals, mineralization differences between normal and osteoporotic bone may account for some of this variability (Jowsey, 1964; Parfitt, 1993; Boyde et al. 1995b). Moreover, although most specimens in this sample were obtained at a date prior to extensive use of osteoporosis preventive treatments (e.g. bisphosphonates), the effect of osteoporosis treatment on mineralization amongst the elderly of this sample is unknown. To examine the relative changes in mineralization variability with age further, studies focusing on smaller regions of the cortex, at higher resolution, would clarify the relative abundance of hypo- and hyper-mineralized areas. In addition, separation of the oldest age group in this study into older and younger subgroups may help to identify any trends in mineralization within the elderly, such as those described in Simmons et al. (1991).
Implications of this research
This study clearly demonstrates that although degree of mineralization of the mid-shaft femur has a limited range within adult individuals, interindividual variation is important within a single age and sex group. Moreover, regional variability in mineralization even within a single bone element of a single individual is extensive. This result is consistent with other human studies (Frost, 1969), and even with studies utilizing animal models in which more control over the genetics, activity patterns and environment was possible (Iwaniec & Crenshaw, 1998). This study of mineralization density is also consistent with dynamic and static histomorphometric analyses illustrating variations in remodelling rates between regions of cross-sections of single elements (Jowsey, 1966; Martin et al. 1980; Raab et al. 1991; Bertelsen et al. 1995; Pfeiffer et al. 1995; Feik et al. 1997), and variability within a region of a given element cross-section (Pfeiffer et al. 1995).
The results of this study of mineralization density in many ways mirror the results of a similar study examining collagen fibre orientation variability utilizing this same material (Goldman et al. 2003). Analysis of that histocompositional variable demonstrated broad age and sex trends, along with some consistencies in patterning across the mid-shaft femur cortex. It also demonstrated overwhelming variability within age and sex groups that could not be explained assuming similar function of the lower limbs. An exploration of mechanical explanations was undertaken in that paper, the discussion of which could equally apply to the current study. Given the broad similarities observed in these two data sets (i.e. in overall variability and patterning), further consideration of the relationships between bone mineralization density, remodelling rates and preferred collagen fibre orientation may allow us to explain better the variability that has been characterized in these studies. Ongoing integrative research utilizing these same research samples will address these issues further.