Differential Age-related Changes in Bone Geometry between the Humerus and the Femur in Healthy Men

Abstract

Muscle pull and weight-bearing are key mechanical determinants of bone geometry which is an important feature of bone strength that declines with adult aging. However, the relative importance of these determinants in young and old adults has not been evaluated systematically. To differentiate the influence of each type of mechanical loading we compared humeral and femoral bone shaft geometry and cross-sectional area (CSA) of the arm and thigh muscles in young and old men. Contiguous transverse MRI (Siemens 1.5T) scans of the arm and thigh were made in 10 young men (21.9 ± 1.0 years) and 10 old men (78.1 ± 4.9 years). Image analysis yielded total (TA), cortical (CA) and medullary (MA) CSA of the humeral and femoral shafts, as well as muscle CSA of the corresponding regions of the arm and thigh. Humeral CA was significantly greater in the young, whereas humeral and femoral MA were significantly greater in the older group. Significant correlations were found between arm muscle CSA and humeral CA (r = 0.73); between thigh muscle CSA and femoral CA (r = 0.69); and between body mass and femoral CA (r = 0.63) and TA (r = 0.55). Moderate correlations between muscle CSA and CA suggest that muscle pull is an important determinant of bone geometry. The significant difference observed between young and old in humeral, but not femoral CA, and the correlation between body mass and femoral, but not humeral cortical area, suggests that weight-bearing attenuates bone loss associated with adult aging.

Decreases in bone mineral density are known to occur with age and may result in osteoporosis [1]. In addition to changes in material properties such as bone mineral density, bone strength is also affected by bone shape or geometric properties [2]. Knowledge of bone shape is important as it has been postulated that mechanical loading may cause changes in bone strength through geometric adaptations, with little change in bone mineral density [3]. Bone geometry is altered throughout life. In general, total bone cross-sectional area increases with age due to subperiosteal apposition [4], while at the same time there is a progressive increase in the area of the medullary cavity due to endosteal resorption [5]. The area of cortical bone is ultimately determined by the rate at which these two processes (apposition and resorption) occur during the aging process. A net loss of cortical bone area leads to a loss in bone strength [6]. However, the shifting of the cortical mass further from the bone center through the process of periosteal apposition helps to attenuate the age-related decline in bone strength [7].

There is evidence that the magnitude of age-related changes in bone geometry differs among different long bones of the body and can vary within a bone along the proximal-distal axis length [5,7,8,9,10]. In a study of male cadavers it was found that there is a greater age-related decline in the cortical area of the humerus than the femur [9]. Using magnetic resonance imaging (MRI), we found that cortical areas were significantly less in the humerus and ulna, but not the radius, in older compared with younger men [8], and that cortical area of the tibia, but not the fibula, was reduced with increased age [5]. Others have reported that bone geometry is most conserved at sites that undergo the greatest levels of mechanical stress, either by weight-bearing, muscle strain or both [7]. Indeed, the relative preservation of cortical area in the femur than the humerus may relate to the weight bearing role and larger muscle mass of the lower limb, but these relationships have not been tested [9].

It has been found from analysis of skeletal remains (cadaver and archaeological samples) that cross-sectional geometric properties scale to bone length [9]. For example, femoral cross-sectional areas and second moments of area vary in approximate proportion to the square and fourth power of bone length, respectively [9]. These findings suggest that long bone design ensures near equal mechanical stress in bones of different length when loaded. Since bone geometry is scaled to bone length, it is likely that the former would also be associated with body height [11]. In most studies of aged skeletal remains, however, geometric measures were not normalized to bone length or body height, despite that seniors (ie. eighth decade of age) are generally shorter and lighter than their younger counterparts [12]. Hence, differences in height (and weight) among individuals could have an effect, unrelated to aging, on bone geometry.

Most imaging studies of age-related bone changes have used dual energy X-ray absorptiometry (DXA), a measure that well represents bone mineral density, but is not well suited to provide information on how it is distributed geometrically [13,14]. Few have used MRI in the study of age-related bone changes [5,8], but MRI produces images that provide detailed information regarding bone structure and geometry, and unlike DXA, MRI does not expose subjects to ionizing radiation [15]. Furthermore, loss of bone strength is often viewed as a problem primarily affecting women [16]; however, 20% of men over the age of 50 are estimated to be osteoporotic [17]. Additionally, 30% of all hip fractures worldwide occur in men [18] and hip fracture mortality is much higher in men than women [19]. Thus, although the relative impact of various factors causing changes to bone structure may differ between men and women, it is important that investigations are carried out independently in both sexes [4].

The primary purpose of this study was to compare the geometry of the femur and humerus in young and older men. Using MRI, we sought to determine the relationship between muscle size and bone geometry in a weight-bearing and non-weight-bearing bone, and how adult aging affects this relationship. We anticipated that femoral and humeral geometric measures (ie. cortical and total bone area) would be positively associated with age, height, weight, and muscle size across subjects, irrespective of age. In addition, secondary to the weight bearing role of the femur, we hypothesized that it will have fewer age-related differences in bone geometry than the humerus, despite equal relative reductions in muscle size in both limbs.

METHODS

Subjects

Ten young men (age 21.9 ± 1.0 years) and ten older men (age 78.1 ± 4.9 years) volunteered to participate in this study. The two groups were of similar mass and height (Table 1). All subjects were nonsmokers, healthy for their respective ages and none had any chronic or acute diseases or taking medications that would affect musculoskeletal health. The young men were recreationally active university students. The old men were living independently in the community, and were recruited from various local senior’s physical activity programs. The exercise programs consisted of ∼ 20–30 minutes of aerobics (walking) followed by 15 minutes of calisthenics and stretching, performed ∼ 2–3 times per week. These exercise programs were designed to maintain cardiovascular fitness, muscular endurance and flexibility for the older participants. This study was conducted in accordance with the guidelines for experimentation on human subjects established by the local university’s ethics review board and conformed to the Declaration of Helsinki. Subjects completed an MRI screening questionnaire to ensure the absence of metallic objects within the body (i.e. surgical pins, shrapnel etc) and informed, written consent was obtained from all subjects. Data were collected during a single visit to the imaging unit.

Table 1.

Subject Characteristics

	Young	Old
Age (years)	21.9 ± 1.0	78.1 ± 4.9
Height (cm)	179 ± 6	176 ± 6
Mass (kg)	80.1 ± 7.3	79.2 ± 13.4

MRI Image Acquisition

Images were acquired with a 1.5 T whole-body magnetic resonance imaging (MRI) system (Siemens Magnetom Vision System, Siemens Medical, Erlangen, Germany). In the supine position, subjects were slowly inserted into the center of the magnet bore. Images of the mid-third section of the femur and humerus, identified from initial sagittal and coronal scout scans of the full femur and humerus lengths, were acquired from the left thigh and left arm. All subjects demonstrated right-sided dominance in the arm and leg, which was determined by asking which hand or foot they used to manipulate an object or kick a ball. The non-dominant limb was imaged since subject differences in activity level of this limb are expected to be less than in the dominant limb. The femur length was defined as extending from the greater trochanter to the lateral femoral condyle. The humerus length was defined as extending from the head of the humerus to the medial epicondyle. A Siemens body array coil and a Siemens soft limb coil (Siemens Medical, Erlangen, Germany) were used for the thigh and arm, respectively. Forty contiguous transverse images (slice thickness = 6mm, matrix size = 256×256) perpendicular to the long axis of the bone were made of the mid-third section of both the arm and thigh using a T1 weighted spin-echo sequence (TE/TR = 14ms/1000ms). Of these images, the 10 centered on the mid-point of the bones were selected for further analysis. A representative scan of the arm for each age group is displayed in Figure 1.

Figure 1.

Axial Magnetic Resonance Images. Left arm of a 20 year old male (left) and the left arm of an 85 year old male (right) showing the largest area for the biceps brachii muscle (bar = 2 cm).

Image Analysis

Image analysis was completed by a single technician using “AnalyzeAVW” Version 3.0 software program (Biomedical Imaging Resource, Mayo Foundation, Rochester, MN, USA). This program allowed for the semi-automated calculation of cross-sectional areas and circumferences for each image. The geometric variables of total cross-sectional area (TA), cortical cross-sectional area (CA) and medullary cross-sectional area (MA) were assessed for the 10 middle images (the 5 images on either side of the mid-point of the bone length) of both the humerus and the femur. Additionally, arm and thigh muscle CSA were calculated. The respective areas from the 10 images were averaged to give the representative areas for each bone and muscle measurement in each subject.

Statistical Analysis

All statistical analyses were conducted with SPSS (version 10 for Windows). Individual humeral and femoral TA, CA and MA and arm and thigh muscle CSA were used to calculate group mean values (old and young) for each variable. Independent samples t-tests were applied to compare the young and old means. Differences were considered significant when p ≤ 0.05. Relationships between parameters (ie., CA versus muscle CSA) were assessed with Pearson correlation coefficients. Stepwise multiple linear regression analysis was applied to determine which variables of age, height, mass and muscle volume provided the best predictor of cortical area.

RESULTS

On average, the old group was 3 cm shorter than the young group, but this difference was not significant (p = 0.3, Table 1). The mean weight was similar in the old and young (p>0.05).

Bone Geometry and Muscle Cross-sectional Area

Mean absolute bone areas (mm²), and bone areas normalized to height (mm²/cm), for both groups are presented in Table 2. The old group was found to have significantly less humeral CA and more MA than the young (p<0.05). No significant differences were found between young and old humeral TA (p>0.05). In the femur, there were no age group differences in CA, but the old had significantly greater MA than the young. Mean femoral TA was significantly greater in the old compared to the young (p = 0.05).

Table 2.

Geometry of the femur and humerus in young and old men as measured by cross-sectional area

	Young (n = 10)		Old (n = 10)
Humerus	Absolute (mm2)	Normalized (mm2/cm)	Absolute (mm2)	Normalized (mm2/cm)
Cortical CSA	278.3 ±37.5	1.55 ± 0.21	226.1 ± 21.2 *	1.28 ± 0.12 *
Medullary CSA	133.8 ± 43.1	0.74 ± 0.23	190.3 ± 42.6 *	1.0 ± 0.21 *
Total CSA	412.1 ± 64.5	2.29 ± 0.34	416.4 ± 45.6	2.36 ± 0.22
Rel. Cort. CSA (%)	67.7 ± 7%		54.4 ± 6% *
Arm Muscle CSA	478.83 ± 71.2	2.71 ± 0.39	331 ± 44.9*	1.87 ± 0.25 *
Femur
Cortical CSA	538.0 ± 58.8	3.0 ± 0.27	502.2 ± 53.9	2.84 ± 0.26
Medullary CSA	155.2 ± 35.1	0.86 ± 0.2	226.9 ± 45.5 *	1.28 ± 0.23 *
Total CSA	693.2 ± 59.5	3.86 ± 0.28	729.1 ± 73.6*	4.13 ± 0.31*
Rel. Cort. CSA (%)	77.6 ± 5%		69.7 ± 5% *
Thigh Muscle CSA	1605.5 ± 228.5	8.95 ± 1.05	1098.3 ± 209.8*	6.2 ± 1.06*

Values are group means +/− standard deviations for bone variables. Normalized columns include bone geometry values normalized for subject height.

^*indicates significant difference compared to young group (p≤0.05). CSA = cross-sectional area; Rel. Cort. CSA = the % of total bone CSA that consists of cortical bone.

The relative area occupied by cortical bone (CA/TA x 100) in the humerus and femur was less in the old than the young (p<0.05, Table 2). Mean thigh and arm muscle CSA were both significantly smaller in the old than the young by 31.6% and 30.0%, respectively (p<0.05, Table 2).

Correlations

Humerus: In the humerus, no significant correlations were found between body mass or height and bone geometry (Table 3). There was a significant negative association between humeral CA and age (r = −0.68). Humeral CA and arm muscle CSA were also significantly related (r = 0.73). These correlations suggest that the smaller humeral CA in the older group can be partially attributed to their smaller arm muscles (ie., atrophy), but not to differences in height or weight. Stepwise multiple linear regression determined that for the humerus, cortical volume was best predicted by muscle volume (r = 0.73), and medullary volume by age + height (r = 0.70).

Table 3.

Correlations of humeral and femoral bone geometry with other variables

	Cortical CSA	Medullary CSA	Total Bone CSA
Humerus
Age	−0.68*	0.54*	0.00
Body Mass	0.162	0.28	0.38
Height	0.24	0.26	0.42
Arm Muscle V	0.73*	−0.34	0.22
Femur
Age	−0.35	0.67*	0.24
Body Mass	0.63*	0.01	0.55*
Height	0.64*	−0.03	0.52*
Arm Muscle V	0.69*	−0.60*	0.11

Values reported are Pearson correlation coefficients, CSA = cross-sectional area, V = volume.

^*p <0.05

Femur: Age was unrelated to femoral CA, but was associated with MA (r = 0.67). Similar to the arm, thigh muscle CSA correlated significantly with CA (r = 0.69) (Table 3). Additionally, height (0.64) and body mass (0.63) were significantly correlated with CA, unlike in the humerus. These comparisons suggest that femoral CA is determined by both muscle CSA and body size. Stepwise multiple linear regression determined that for the femur, cortical volume was best predicted by muscle volume + weight (r = 0.80), and medullary volume by age (r = 0.67) and total volume by weight (r = 0.55).

DISCUSSION

We observed that there are significant differences in limb bone geometry between young and old men, and that these differences were not the same in the femur and humerus. CA was less in the humerus of the old than the young men, whereas femoral CA was similar in both groups. The smaller humeral CA (−19%) in the old men likely resulted from ongoing endosteal resorption (i.e. increased MA) combined with little or no periosteal apposition (no increase in TA). The preserved femoral CA in the old group may reflect greater periosteal apposition, as indicated by the larger femoral TA in most of the subjects (ie., 8 of the 10 largest femoral TA were recorded in the old group). Humeral CA was most strongly associated with arm muscle CSA, and was unrelated to body mass or height. Femoral CA was most closely associated with thigh muscle CSA as well as body mass and height. Our results suggest that bone strength may be better preserved in the femur than the humerus due to the maintenance of femoral CA with adult aging. Additionally, whereas maintaining arm muscle mass is most important in preserving healthy humeral bone geometry, a combination of leg muscle mass and weight-bearing activity provide greater influence in maintaining femoral bone geometry.

Muscle size and weight-bearing are both key factors that impact bone geometry [5]. Weight-bearing exerts an influence on bone remodeling through cyclic impact activities (i.e. walking) which impart relatively high levels of strain on bone [20]. Additionally, strain induced by muscle contraction has been suggested to be a critical factor in determining bone strength because the largest mechanical forces placed on bones are due to muscle pull rather than weight-bearing per se [21]. Preservation of femoral CA and increased TA in the old group could be partially attributed to the effects of weight-bearing [5,22,23]. Body weight was the most closely related variable to femoral TA (r = 0.55), but no single variable was found to significantly predict humeral TA. Despite the observed maintenance of CA in the femur, MA was approximately 45% greater in the old group for both bones, which likely is an indication of endosteal resorption. These findings indicate that the CA maintenance observed in the femur may be due partially to periosteal apposition in response to weight-bearing. Because periosteal apposition is thought to help maintain bone strength, despite decreases in bone mineral density and the “hollowing” effects of endosteal resorption, our results suggest weight-bearing activity is important in preserving the strength of weight-bearing bones. Although our old subjects were more than 50 years older than the young, they were healthy independent seniors with no significant mobility impairments. Supporting our findings regarding upper and lower limb geometry differences are previous studies of bone density in cadaver and archaeological samples. These studies have suggested this functional difference (weight-bearing versus non-weight-bearing) could account for the greater age-related increase in porosity and loss of bone mineral content in the humerus than femur [9,24,25].

We did not observe any relative differences (differences between cortical and medullary area expressed as a percent of total area) between the humerus and femur within the young group, despite the influence of weight-bearing activity on the femur being present throughout healthy adulthood. Non-weight-bearing bones, including the humerus, maintain their structural properties in young adults due to systemic factors, particularly through hormonal and metabolic control [26], as well as the biomechanical strain placed upon bone by muscle [27]. With adult aging, these systemic factors are potentially modified and may no longer be sufficient to maintain optimal bone structure [21]. Thus mechanical strain through weight-bearing and muscle pull become more important factors in determining bone structure with age to compensate for metabolic and hormonal changes [5,15,28].

It is well known that loss of muscle mass and strength are closely associated with adult aging [29,30,31]. Although strength was not measured, is it reasonable that the older mean were weaker based on their smaller muscle CSA. Humeral and femoral CA, which are indicators of bone strength, correlated significantly with arm (r = 0.73) and thigh muscle CSA (r = 0.68), respectively. These relationships support previous findings that demonstrated significant correlations exist between MRI-derived arm muscle area and humeral CA in men in their 3^rd, 8^th, and 9^th decades (r = 0.52) [8]. Also, a significant correlation (r = 0.59) has also been reported between leg muscle cross-sectional area and tibial CA in children [32].

The correlations found between muscle CSA and CA in this study, indicate that muscle CSA is an important determinant of bone geometry, but the 31% and 32% differences observed between young and old in arm and thigh muscle area, respectively, are nearly equal. Therefore, age-related differences in muscle CSA cannot account completely for the preservation of femoral CA and smaller humeral CA in the older group. Rather, results from the present study seem to indicate that weight-bearing activity is responsible for preserved CA in the femur. Femoral CA is moderately correlated to both muscle CSA (r = 0.69) and body weight (r = 0.63). This finding underscores the importance of maintaining mobility in adult aging. If weight-bearing activity is reduced sufficiently, the CA of weight-bearing bones such as the femur may decrease, leading to a compromised geometrical bone structure, and thus an increased susceptibility to fracture.

The strength of a bone is affected by its structural properties, including cross-sectional geometry [9]. In our study, we found a significant negative correlation between age and humeral CA (r = −0.68), whereas no significant relationship was found between age and femoral CA. As mentioned previously, studies have shown greater increases in bone porosity and decreases in bone density (both markers of decreased bone strength) in the humerus versus the femur [24,9] Thus, our findings regarding bone geometry closely follow this pattern regarding the relative changes in predicted strength of the humerus and femur. The strength of a bone in compression (i.e. its resistance to axial loads) depends primarily on its cortical cross-sectional area [33]. Our findings of reduced humeral CA in the old group indicate a significant decrease in humeral axial strength whereas with no differences in femoral CA between groups would suggest that femoral axial strength is better preserved in old age. Additionally, Beck and colleagues found periosteal apposition protects the strength of the femoral neck which would otherwise be reduced by bone mineral loss through an increase in bone cross-sectional moment of inertia [34]. In our study periosteal apposition in the femur likely increased the moment of inertia in the femur, which could maintain bone strength in the old group despite the increased femoral medullary cavity and porosity versus the young [34]. Our results indicate that with adult aging the femur undergoes geometrical change that attenuates increases in fracture-risk versus the humerus due to a better preservation of cortical bone and the increase in TA. This conclusion has some applicability in that older men could mitigate the loss of bone strength by strengthening the muscles associated with non-weight-bearing bones [35]. This could be accomplished through resistance training, and it could help reduce the risk of fracture in these bones (i.e. the humerus).

Finally, the effects of interstitial fluid pressure should be considered when comparing bone geometry changes between upper and lower limb bones. Previous studies have shown increases in intramedullary pressure lead to greater amounts of appositional bone growth [36]. Additionally, there is a fluid pressure gradient that increases from the superior to inferior aspects of the body created by gravity [37], and thus the femur experiences greater intramedullary pressure relative to the humerus. This pressure difference could help explain why femoral cortical area is better maintained in comparison to the humerus. The mechanism behind the effect of fluid pressure on bone is related to the pressure sensitivity of osteoblasts and osteoclasts, and through fluid pressure, which creates additional shear stress on bone leading to the observed osteogenic response [37].

One potential limitation of the present study is the lack of quantitative data pertaining to physical activity history in both the young and old participants. However, activity differences were minimized somewhat by testing the non-dominant limb and recruiting active older adults. Physically compromised older adults likely would exhibit larger differences in bone geometry compared with younger men than what we report. Thus, one value of these results is to direct future similar studies in other groups to identify important differential age-related alterations in limb bones that could affect physical function and health. Currently very few studies have explored these factors and relationships.

In conclusion, the observed age-related differences in bone geometry were greater in the humerus than the femur. The CA of femur was similar between age groups whereas it was significantly smaller in the humerus of the old men. Our results suggest that muscle mass is a modest predictor of both upper and lower limb CA, and that muscle strain is a primary source of mechanical loading. In contrast, the combination of muscle mass and body weight is the most influential determinant of CA in the femur. These findings can likely be attributed to the difference in weight-bearing stress placed upon the two bones during activities of daily living in active ambulatory seniors.