Femoral cortical bone in a Portuguese reference skeletal collection Massa óssea cortical do fémur numa coleção esquelética de referência portuguesa

This study aims to investigate patterns of femoral cortical bone fragility with age (at death) and to evaluate its associations with sex and bone mineral density. Radiogrammetric parameters of the femur and bone mineral density at the proximal femur were assessed in an adult sample (N=98) from the Coimbra Identified Skeletal Collection (Portugal). Diaphysis total width (DTW), femoral cortical index (FEMCI) and bone mineral density (BMD) are significantly higher in males, while medullary width (MW) is not statistically different between sexes. Cortical bone parameters of Resumo Neste trabalho, pretende-se investigar a fragilidade óssea cortical no fémur com a idade (à morte) e a sua associação ao sexo e à densidade mineral óssea. Os parâmetros radiogramétricos do fémur e a densidade mineral óssea no fémur proximal foram avaliados numa amostra de indivíduos adultos (N=98) da Coleção de Esqueletos Identificados da Universidade de Coimbra (Portugal). A largura total da diáfise (LTD), o índice cortical do fémur (FEMCI) e a densidade mineral óssea (DMO) são significativamente maiores nos homens, enquanto a largura medular (LM) não é estatisticamente di1 Research Centre for Anthropology and Health – Department of Life Sciences, University of Coimbra, Portugal. 2 Interdisciplinary Center for Archaeology and Evolution of Human Behavior – University of Algarve, Portugal. 3 Laboratory of Forensic Anthropology – Department of Life Sciences, University of Coimbra, Portugal. 4 Centre for Functional Ecology – Department of Life Sciences, University of Coimbra, Portugal. a orcid.org/0000-0002-0480-209X b orcid.org/0000-0003-2998-371X * Corresponding author: fcurate@uc.pt; franciscocurate@gmail.com DOI: http ://doi.org/10.14195/2182-7982_34_5 Artigo recebido a 14 de outubro de 2016 e aceite a 19 de junho 2017 Fr an ci sc o Cu ra te , E ug én ia C un ha 92


ANTROPOLOGIA P O R T U G U E S A
Abstract This study aims to investigate patterns of femoral cortical bone fragility with age (at death) and to evaluate its associations with sex and bone mineral density. Radiogrammetric parameters of the femur and bone mineral density at the proximal femur were assessed in an adult sample (N=98) from the Coimbra Identified Skeletal Collection (Portugal). Diaphysis total width (DTW), femoral cortical index (FEMCI) and bone mineral density (BMD) are significantly higher in males, while medullary width (MW) is not statistically different between sexes. Cortical bone parameters of Resumo Neste trabalho, pretende-se investigar a fragilidade óssea cortical no fémur com a idade (à morte) e a sua associação ao sexo e à densidade mineral óssea. Os parâmetros radiogramétricos do fémur e a densidade mineral óssea no fémur proximal foram avaliados numa amostra de indivíduos adultos (N=98) da Coleção de Esqueletos Identificados da Universidade de Coimbra (Portugal). A largura total da diáfise (LTD), o índice cortical do fémur (FEMCI) e a densidade mineral óssea (DMO) são significativamente maiores nos homens, enquanto a largura medular (LM) não é estatisticamente di-

Introduction
Osteoporosis and bone fragility are major public health problems facing postmenopausal women and aging individuals from both sexes. The clinical impact of osteoporosis stems from the complications associated to it, namely hip, distal radius, proximal humerus and vertebral compression fractures (Sattui and Saag, 2014). A growing body of litera-ture has detailed the long history of bone fragility in past groups, recounting the diachronic fluctuations in its epidemiological patterns, etiological agents and societal implications (Brickley and Ives, 2008;Curate, 2014a;2014b).
Radiogrammetry and dual x-ray absorptiometry (DXA) are undoubtedly the most common methods to study bone loss in past populations (Brickley, 2002;Brickley and Agarwal, 2003;Curate, 2014a). the femoral diaphysis are associated with age only in women, whereas BMD decreases with age in both sexes. The evaluation of femoral cortical bone reveals sex-specific trajectories of endosteal bone loss and periosteal apposition, stemming from sexual differences in the rate and pattern of bone loss, and in bone size. In females, endocortical bone loss rises with age, particularly in peri-and postmenopausal years, decelerating later in life. Concomitantly, accretion of bone in the subperiosteal surface persists throughout adulthood -partially offsetting bone fragility in women. Strength in the femoral mid-diaphysis appears to be preserved throughout most of the life course in both sexes.
The perception of osteoporosis as a disease of trabecular bone loss is fundamentally incorrect (Seeman, 2013). Trabecular bone exposure to faster remodeling and bone loss relates to its higher surface area; conversely, trabecular bone represents just 20% of the overall bone matrix volume. Therefore, the originally slower loss of cortical bone (~80% of the skeleton) causes a comparable amount of bone loss throughout the first years after menopause, being responsible for greater bone loss after 60 years of age (Seeman, 2013;Zebaze et al., 2010). Also, cortical bone is especially important to bone stability and strength -influencing resistance to external loads and the occurrence of fractures (Holzer et al., 2009).
This study focuses on the assessment of cortical bone at the femoral diaphysis in an adult sample from the Coimbra Identified Skeletal Collection (CISC). Radiogrammetry at the femur mid-diaphysis allows for comparisons with the second metacarpal, accounting for differences in functional stress -e.g., weight bearing and tension -associated with each bone. Also, given the relative size of the bones, femoral measurement error is lower. Finally, it conveys a valuable contrast with the proximal femur, rich in trabecular bone and a classic site for osteoporotic fractures. This paper aims to evaluate and interpret patterns of cortical bone fragility -with a particular emphasis on endosteal bone loss and periosteal apposition -with aging, and to explore its associations with sex and bone mineral density (BMD) at the proximal femur. Scientific advancement can be significantly supported by sharing both methods and raw data resources. As such, another objective is to divulge a database of femoral radiogrammetric parameters based in a reference skeletal population.

Materials and Methods
The CISC comprises 505 skeletons from the Cemitério Municipal da Conchada (cemetery at Coimbra, Portugal). All individuals died between 1904 and 1936, i.e., before the massive introduction of medical therapies against bone loss. Biographical information for each individual is available, including sex and age at death (Cunha and Wasterlain, 2007). The observed sample includes 98 Portuguese nationals from both sexes, born between 1831 and 1914, and that died between 1910 and 1936. Recorded ages at death range from 21 to 89 years old. Individuals were predominantly non-specialized manual workers with low socioeconomic status. A purposive sampling strategy was adopted, with equal numbers of females and males. Also, only individuals without macroscopical post-depositional alterations and/or blatant pathological conditions were included in the sample. Anteroposterior radiographs of the midshaft area of the left femur of each individual were taken using a mammogram film with an exposure time of mAseg 80-50, exposure of Kv 30-35 and focal distance of 1.0 m. Maximum length of the femur, as defined by Martin and Saller (1957), was determined. Measurements of diaphysis total width (DTW) and medullary width (MW) were taken following a standardized guide (Ives and Brickley, 2004; raw data available in Data S1). Radiogrammetry was used to establish cortical index (FEMCI) in the femoral mid-shaft (Mays et al., 1998). The revised femoral cortical thicknesses index (RFCI) and adjusted medullary width index (AMWI) were also computed (Glencross and Agarwal, 2011). Indexes were obtained from the raw measurements (in mm) as follows: ; . Bone mineral density was evaluated in the femoral neck through DXA. The femora were scanned using a Hologic QDR-4500A densitometer. Each femur was placed in standard anteroposterior position, with the diaphysis parallel to the central axis of the scanner, in a low--density container with dehydrated rice (10 cm depth). The absence of soft tissues and bone marrow in historical skeletal remains hampers DXA measurements; as such, a water bath or rice are usually used as soft-tissue proxies (Brickley and Agarwal, 2003;Curate, 2014a).
Descriptive statistics, viz. group means, standard deviation (SD) and 95% confidence intervals (95% CI) for the mean were estimated for each continuous variable. Normality was assessed through the skewness and kurtosis of the distributions (Kline, 2010). Homoscedasticity was evaluated with the Levene's test. A student's ttest (independent samples) was applied to evaluate the null hypothesis that the means of two groups were equal. Linear Pearson correlation was used to evaluate a possible linear relationship between the cortical parameters of the femur with recorded age at death and bone mineral density at the neck of the femur. Local polynomial regression fitting smoothing (LOESS) was employed to explore nonlinear empirical relationships between variables. LOESS offers a graphical summary of the relationship between variables by fitting a function of the independent variables locally (Cleveland, 1979). Twenty femora were measured in different days to check repeatability of the cortical measurements (DTW and MW). Measurement error was evaluated with the technical error of measurement (TEM), the relative technical error of measurement (rTEM) (Ulijaszek and Kerr, 1999) and the reliability coefficient (R c ) (Ward and Jamison, 1991). All statistical and graphical analyses were performed with R programming language (R Development Core Team, 2016;Chang and Wickham, 2016).

Results
The detection of the endosteal margin is sometimes complex rendering radiogrammetric measurements challenging (Schäfer et al., 2008). Notwithstanding, the results suggest that cortical measurements (DTW and MW) were performed within adequate levels of measurement error (Table 1).

Discussion
Sexual dissimilarity in femoral cortical parameters, particularly DTW and FEMCI, but also femoral length, revised femoral cortical thickness index and adjusted medullary width index, stems from sex-specific variations in bone size, and rate and pattern of bone loss (Samuel et al., 2009). Generally, males have larger bones, with puberty enacting a major role in skeletal size determination (Seeman, 2003;Samuel et al., 2009). In women, post-pubertal estrogen production supposedly inhibits periosteal bone formation and, consequently, bounds bone diameter; while pubertal androgen in men intensifies periosteal bone formation and bone diameter. Also, during growth men undergo a long-standing period of bone gain, resulting in the increase of bone cortical thickness (Seeman, 2003). On the other hand, bone loss accelerates in periand post-menopausal women as estrogen withdrawal increases the rate of bone remodeling, and more bone is resorbed and less is formed at each basic multicellular unit (Seeman, 2008). The differences between the sexes in cortical bone loss have been detailed both in modern (e.g., Virtamä and Helelä, 1969;Maggio et al., 1995;Ginsburg et al., 2001) and historical populations (e.g., Carlson et al., 1976;Mays et al., 1998;Drusini et al., 2000;Ives, 2007;Glencross and Agarwal, 2011;Umbelino et al., 2016). In the same skeletal collection but with a different sample, radiogram-      metry of the second metacarpal also exposed significant differences between sexes (Curate et al., 2015).
Cortical bone changes were significantly related with age only in females, with results showing signs of gendered trajectories of age-related cortical bone loss. In the second metacarpal, cortical index and medullary width were significantly associated with age in both sexes (Curate et al., 2015). Femoral medullary cavity enlarges in the course of aging as a result of imbalance of endosteal bone formation and resorption that leads to endocortical bone loss (Jergas, 2008). Although bone remodeling at the endosteal envelope is thought to increase mildly in aging men, it drastically increases in perimenopausal and early postmenopausal women, slowing with further aging (Clarke, 2008). As such, the "rate" of endosteal cortical bone loss is faster in women when compared to men -considering both MW and AMWI. Medullary width in women generally shows a slow and gradual increase with age, but the LOESS regression curve suggests an acceleration of endosteal bone loss starting around 50 years of age -it is important to note that the mean age of menopause in historical populations most likely ranged from 45 to 50 years (Pavelka and Fedigan, 1991). Remarkably, MW increases 15.1% between the fifth and sixth decades, and only 4.5% between the sixth and the seventh decades of life. Bone loss decelerates in the three to five years ensuing menopause, although it endures at a faster rate than before menopause (Seeman, 2008). Amongst men, medullary expansion is virtually absent until much later in life, with an apparent MW increase only after the middle of the sixth decade.
Age-continuous apposition of bone on the periosteal surface was originally validated by Smith and Walker (1964) and has been considered as an adaptive response to preserve resistance to bending (Lazenby, 1990;Allen et al., 2004;Szulc et al., 2006;Seeman, 2008;Peck and Stout, 2009). Diaphysis total width at the mediolateral axis -which can be regarded as a surrogate of periosteal apposition -increases moderately with age in women. In contradiction with some cross-sectional studies, in men DTW did not increase significantly with age (Virtamä and Helelä, 1969;Lazenby, 1990;Feik et al., 2000;Mays, 2001;Peck and Stout, 2009), and periosteal apposition (as a decennial percentage) was greater in women. Nevertheless, other epidemiological studies describe a greater percentage increase in periosteal apposition in women (Garn et al., 1967;1972;Kaptoge et al., 2003). In a prehistoric Mississippian sample, van Gerven et al. (1969) reported female gains and male losses in periosteal diameter; while Wallace et al. (2014) observed (in a sample of Inuit foragers) that periosteal area did not increase with age in either sex.
Dynamics of periosteal apposition in the femoral and second metacarpal diaphyses are apparently different: the external dimensions of the second metacarpal do not increase with age in males or females (Curate et al., 2015). The femur is a weight-bearing bone, subjected to increased biomechanical loading, which can explain some of the observed dissimilarities.
The underlying causes of periosteal apposition are intricate to delimit but it is probable that the extent of mechanical compensation depends on initial bone size, with smaller bones -women in the studied sample are generally smaller than men -showing more periosteal apposition. Sex-specific disparities in periosteal apposition have been partially attributed to sexual bone size differences, with slender diaphyses requiring greater rates of periosteal bone gain over time (Jepsen and Andarawis-Puri, 2012). Also, loading differences possibly contribute to individual dissimilarities in periosteal apposition at the lower limbs; e.g., long-term immobilization seems to inhibit periosteal bone formation (Schäfer et al., 2008). Mechanical stimulation is usually much greater in men than women (Vanderschueren et al., 2006). Certainly, most men in the CISC sample were manual workers engaged in highly demanding physical occupations -but, as a rule, women also experienced a physically active lifestyle, involving strenuous workloads (Cunha and Umbelino, 1995).
Cortical bone resorption at the endosteal and periosteal sites react differently to distinct metabolic stimuli (Grampp et al., 1997). The endosteal surface exhibits a higher remodeling activity, probably as a consequence of greater biomechanical strains or cytokine exposure from the contiguous bone marrow (Clarke, 2008). Also, periosteal cells appear to differ from endosteal cells; each cell population responds differently both qualitatively and quantitatively to a wide variety of hormones and growth factors (Allen et al., 2004). The femoral cortical index refers to a dimensionless parameter that is the ratio of medullary cavity width to bone diameter (Shepherd et al., 2005), and tentatively reflects the conjoint endosteal and periosteal remodeling activity. FEMCI declines slightly with age in females (but not in males) as a result of the uncoupling in bone deposition and resorption that occurs throughout aging -a small increase in periosteal bone formation in women is exceeded by a greater intensification in endosteal resorption with subsequent cortical thickness reduction. LOESS regression indicates that FEMCI decline in females occurs only after the seventh decade, suggesting that the femur maintains its strength until later in life. In males, strength seems preserved throughout the life course. Cortical index -as observed in the second metacarpal -usually declines from younger to older age groups in both females and males (Rewekant, 2001;Ives, 2007;Beauchesne and Agarwal, 2014;Mays, 2015) or just in females (Mays, 1996) from archeological samples. Cortical bone in the femur seems to decrease only in females (Erick-sen, 1982;Mays et al., 1998), but a negative association of relative cortical area of the femur with age in both sexes has also been observed (Doyle et al., 2011). Interestingly, the revised femoral cortical thickness index (a ratio adjustment intended to remove size information) (Glencross and Agarwal, 2011) does not show an association with age in any of the sexual groups.
For females, correlations between cortical parameters at the femoral diaphysis and BMD measured at the ROI "neck" are significant (except for DTW with BMDneck ), but the association is moderate. In males, cortical bone in the femur is not associated with BMD neck . Female results are similar to the associations between cortical bone and BMD stated by Mays et al. (1998), with a weaker association in males. Ives and Brickley (2005) reported a non-significant association between BMD neck and metacarpal cortical index in a pooled sample from both sexes. The results support the concept that bone loss and mass are not homogeneous within and among skeletal elements of the same individual, both because biomechanical factors and differences in macroscopical bone composition (Bonnick, 2010).

Conclusions
Femoral cortical bone loss during aging does not follow a linear course, with sex-specific patterns of endosteal bone loss and periosteal apposition. In women, endocortical bone loss increases with age, especially in presumed perimenopausal and early postmenopausal women, slowing later in life. Concurrently, accretion of bone in the femoral outer diameter throughout adult life continues -helping to preserve bone strength until the sixth decade. In men, medullary expansion and periosteal apposition are fundamentally nonexistent during adult life. Strength in the femoral diaphysis seems to be preserved throughout life. The observed correlations between femoral cortical bone and BMD at the femoral neck are suggestive of skeletal heterogeneity in bone loss and mass both between bones and within the same bone.
Radiogrammetry of the femur offers additional viewpoints to the study of bone loss in historical populations, and this study reiterates that different methods offer unique insights about bone remodeling and maintenance (Brickley and Agarwal, 2003). Notwithstanding, this study presents some limitations, including the cross-sectional nature of the data, reliance on mediolateral axis measurements only and sample size.