The relative importance of fat and lean tissue mass in determining bone mineral mass among postmenopausal women was examined in this 1-year longitudinal study. Fifty postmenopausal Caucasian women entered the study; 45 of them completed a 1-year follow-up. Dual-energy X-ray absorptiometry was employed for measuring total and regional bone mineral density (BMD) and bone mineral content (BMC), fat tissue mass (FTM), lean tissue mass (LTM), and body weight. Results from linear regression analysis using the cross-sectional data (n = 50) of the study indicated that LTM explained a larger percentage of variation in bone mineral mass than did FTM. FTM and LTM were found to be moderately correlated (r = 0.55); when FTM was entered in the same predicting regression models, LTM was a significant predictor (p < 0.05) of the total and regional BMC, but not BMD. The percent FTM (and inversely %LTM) was correlated with BMD and BMC, but significant correlation was primarily found only for total body BMD (or BMC). Weight was the best predictor of total body BMD and BMC. Longitudinally (n = 45), annual changes in both FTM and weight were significantly associated with annual changes in regional BMD after adjustment for initial bone mineral values (p < 0.05). We conclude that bone mineral mass is more closely related to LTM than to FTM, while annual changes in regional BMD are more closely correlated with changes in FTM in healthy postmenopausal women. Meanwhile, increased body weight is significantly associated with increased bone mineral mass.
A substantial body of evidence indicates that bone mineral mass is closely related to other body composition variables at different age stages.1–6 It is clear that greater understanding of the relationships between body composition and bone mineral mass and their underlying biological mechanisms is important for preventing and managing osteoporosis. Available data show that increased body weight contributes significantly to higher bone mineral mass in both pre- and postmenopausal women.1,2 However, results of earlier studies do not agree on whether fat tissue mass (FTM) or lean tissue mass (LTM) is the major determinant of bone mineral mass for different age groups of women.3–6 Several mechanisms have been proposed regarding the association between bone mineral mass and the body composition indices LTM and FTM. Besides increased mechanical load on bone associated with increased FTM or LTM as a mechanism for increasing bone mineral mass, increased FTM is also considered to be beneficial to bone mineral mass through promoting estrogen production.4,7 However, the positive relation between LTM and bone mineral mass is thought to indicate higher physical activity and/or potential genetic association between LTM and bone mineral mass.3 Since estrogen deficiency is a known cause of declining bone mass in postmenopausal women, FTM may become more important to bone metabolism at this period of life.6 The extent to which bone mineral mass in postmenopausal women is determined by FTM, however, is still controversial.3,5
Body composition varies with age and menopausal status.8 In healthy females, FTM usually increases,9,10 while LTM and bone tissue mass decrease8 after menopause. How the changes in FTM and LTM may influence the rate of bone loss is not well established. Limited data suggest that the postmenopausal rate of bone loss may be affected more by a change in FTM than by a change in LTM.11
In the present study, both cross-sectional and 1-year longitudinal data were used to examine the relations between bone mineral mass and other body composition variables in healthy postmenopausal women. Unique to our study is the investigation of both the cross-sectional and longitudinal association of body composition and bone mineral mass in the same population. Based on previous studies and proposed mechanisms for the relationship between elements of body composition and bone mineral mass, we hypothesized that LTM is the major determinant of bone mineral density (BMD) and bone mineral content (BMC), with FTM making further, but smaller, contributions in predicting BMD and BMC in postmenopausal women. After menopause, annual changes in bone mineral mass should be more closely related to change in FTM than to change in LTM.
MATERIALS AND METHODS
In the spring of 1992, an advertisement to recruit postmenopausal women as study participants was distributed on the campus of the University of Arizona and at women's health clubs and fitness centers in the Tucson area. Fifty Caucasian postmenopausal women who were at least 3 years past menopause (time since last menstruation) and younger than 65 years were included in this study. The subjects either had never received hormone replacement therapy (HRT) or had stopped taking HRT for at least 1-year prior to the time of the start of the study. All subjects were apparently healthy and not using medications known to interfere with changes in body composition. The project was approved by the University of Arizona Institutional Review Board for research on human subjects, and written informed consent was obtained from each participant. During the 1-year follow up, 4 people dropped out the study for personal reasons and 1 person suffered a spinal fracture before the end of the study, leaving 45 women in the longitudinal sample.
Dual-energy X-ray absorptiometry
BMD, BMC, FTM, and LTM were estimated using dual-energy X-ray absorptiometry (DXA) (DPX Lunar Corp., Madison, WI, U.S.A.)12 in the Body Composition Laboratory, Department of Exercise and Sport Science, University of Arizona. Each study participant was scanned twice (within a 1-week interval) at the beginning of the study and twice (within a 1-week interval) after 1 year of follow up. DXA-derived measurements included total body BMD (TBMD), total BMC (TBMC), LTM, FTM, percentage of FTM (%FTM), percentage of LTM (%LTM), and weight. Regional BMC and BMD at the lumbar spine (L2–L4) (LBMD, LBMC), femoral neck (NBMD, NBMC), femoral Ward's triangle (WBMD, WBMC), and femoral trochanter sites (FTBMD, FTBMC) were also measured by DXA. The scan modes were slow for femoral and spinal sites and medium for the total body measurements. Subjects were positioned according to the standards in the Operator Menus (Lunar Corp.). The precision (percent coefficient of variation %CV) based on the measurements taken 1 week apart (n = 50) in the measurements of BMD was 0.7, 1.3, 1.0, 2.8, and 1.8% for total body, lumbar spine, femoral neck, femoral Ward's triangle, and femoral trochanter, respectively. The precision for total BMC was 1.3%. The precision in FTM measurement was 2.2%, and the precision for LTM was 1.7%. The 1-year long-term in vivo precision of DXA derived measurements was 0.7, 2.2, 2.0, 3.7, and 2.1% for TBMD, LBMD, NBMD, WBMD, and FTBMD, respectively. The long-term in vivo precision of TBMC, FTM, and LTM was 2.0, 5.7, and 2.4%, correspondingly.
Anthropometry, physical activity, and dietary assessment
The subjects' height was measured to the nearest 0.1 cm with a wall-mounted stadiometer at the time of each bone scan. Each woman in the study completed a self-administrated physical activity questionnaire at the beginning and again at the end of the study. Mean nutrient intakes were assessed from 8-day dietary records with 4 days taken during the first month of the study and 4 days taken during the last month of the study. For each 4-day record, the 4 days were assigned randomly in a period of 1 month. However, 1 of the 4 days was required to be on the weekend in order to detect the variation between food intake during the week and on the weekend.13 A 1000 mg/day (Oscal 500) calcium supplement was provided to all the subjects during the study-year to ensure their adequate calcium intake. The detail of physical activity and nutrient intakes will be described elsewhere.
Although DXA and single photon absorptiometry (SPA) are very reliable methods of bone mineral measurement, system and operation errors may introduce artificial variation in the measurements. Estimating the percentage of technical variation in a measurement is important for studying biological changes of bone density and body composition over a period of 1 year. An analysis of variance procedure was used for evaluating the reliability of measurements taken during this longitudinal project, which incorporated the special feature of repeated measurements at both the beginning and the end of the study. The basic principle of this analysis is to determine the proportion of variability attributed to the factors of interest to the total variance in a measurement.14,15 In this particular study, two sources of variability were of interest to us: (1) the variation between subjects at baseline (or final) measurement and (2) the between-subject variation in differences between final and baseline measurements. The first source of variation was defined as “reliability at a point in time” (RPT) and the second as “reliability of 1 year change score” (RCS).
The average of the two baseline measurements in body composition and BMD was used to represent the initial values, and the averages of the two measurements done at the end of the study were used as final values. The annual change in body composition and BMD was defined as the difference between final and initial values. Univariate and multivariate linear regression analyses were employed to estimate the independent contribution of each individual body composition variable to bone mineral mass. Initial values of bone mineral measurements were included in each corresponding model for predicting change in bone mineral measurements to adjust for regression to the mean phenomenon. All of the statistical analysis was done using Statistical Package for the Social Sciences (SPSS).
Calcium intake in the study population was above recommended level (1500 mg/day) for women of this age group.16 History of taking hormone replacement therapy had no significant impact on either bone mineral values or rates of bone loss. The resulting analysis of variance for estimation of the reliability in measurements by DXA is presented in Table 1. If we define the total between-subject variation of a measurement as one, then the closer the value of RPT or RCS is to one, the higher the reliability (counting more biological variation and less artificial variation in total between-subject variation) that measurement has. The result indicated that the reliability of point in time (RPT) was generally higher than the reliability of a 1-year change score (RCS) for all the measurements. At least 91% of the total variation reflects interindividual biological variation for all the initial measurements. Biological variation counts more than 50% of the total intersubject variation in a 1-year change score of all the measurements except TBMD and WBMC.
Table Table 1. Reliability of Bone Mineral and Body Composition Measurementsby DXA
The clinical characteristics of the study population are shown in Table 2. The range in body mass index (BMI) values was quite large since obese women were not excluded from the sample. The bone mineral values of the current study population were comparable to the same age group of women in other studies.
Table Table 2. Descriptive Statistics for Bone Mineral and Body Composition Samplen = 50
Correlation analysis (Table 3) indicated that among FTM, %FTM, LTM, %LTM, weight, and height, weight was the best predictor for TBMD and TBMC and explained a significant amount of variation in regional bone mineral mass for this study population. Moreover, height contributed to a significant portion of the variation in total BMC, spinal BMC, and spinal BMD. Compared with FTM, LTM explained a relatively larger proportion of variation in most of the bone mineral measurements except for TBMD, as indicated by the squared Pearson correlation. %LTM was inversely correlated with BMD and BMC.
Table Table 3. Correlation Between Initial Bone Mineral Measurement and Initial Body Composition (n = 50)
LTM and FTM were found to be moderately correlated with each other (r = 0.55). Both were also highly correlated with weight (r = 0.75 and 0.96, respectively). The independent predicting roles of LTM and FTM were examined using both univariate (LTM or FTM alone models) and multivariate (both LTM and FTM models) regression analysis (Table 4). LTM was a significant predictor for all the BMC and BMD measurements in univariate regression models. After adjustment for FTM, LTM was still a significant predictor for all the BMCs and the WBMD. FTM had reduced significance in all the predicting models when LTM was also in the regression equation. The linear relation between TBMC and LTM is shown in Figure 1.
Table Table 4. Regression Analyses of Initial Bone Mineral Measurements on Initial Fat and Lean Tissue Mass
The annual changes were 0.2, 3.1, and 0.7% for LTM, FTM, and weight, respectively. Changes in BMD (ΔBMD) and BMC (ΔBMC) were not significantly correlated with initial body composition measurements. After adjustment for initial bone mineral value, change in FTM (ΔFTM) and change in weight (Δweight) were significant predictors of ΔBMD (p < 0.05) (Table 5). Change in LTM (ΔLTM) had no significant relation with changes in any bone mineral measurement. The linear relation between ΔFTM and ΔFTBMD is presented in Figure 2. Meanwhile, ΔFTM and Δweight were correlated highly with each other (r = 0.91).
Table Table 5. Regression of Change in Bone Mineral Measurements on Change in Body Composition After Adjustment of Initial Bone Mineral
It is well known that multiple genetic and environmental factors determine human bone mass. Relationships between body composition indices and bone mineral mass may be confounded by such other factors as estrogen level and calcium intake. Inadequate calcium intake may accelerate loss of bone, and a decline in the serum estrogen level is significantly associated with a higher rate of bone loss among postmenopausal women.17–20 Moreover, in addition to its impact on bone, a change in estrogen level may also affect LTM.8,20 We managed to limit the range of variation in estrogen level by including in the current study only women who were not taking HRT and were at least 3 years past menopause.
Although HRT may have a lasting effect on bone, we did not find a significant difference between past hormone users and nonhormone users in BMD (or BMC) and rate of bone loss. This may be explained by the fact that only 12 of 50 people in the study took short-term (average <1 year) HRT and they all stopped HRT at least 1 year before entering this study. Furthermore, we supplied 1000 mg/day calcium to all the subjects to ensure adequate calcium intake during the study.
Our results confirm the positive relation between body weight and BMD (and BMC) suggested by other studies.21–25 Meanwhile, we also found that weight reduction accelerated the rate of bone loss in postmenopausal women. The positive relationship between increased weight and increased BMD is probably the result of increased mechanical (gravitational) forces on bone by weight loading.26 Since LTM and FTM are the two major components of body weight, it is not surprising that FTM and LTM have both been found to be positively related to bone mineral mass in most previous studies3-6,23,27 as well as in this one. However, the independent effects and relative importance of FTM and LTM on total and regional BMD and BMC at different ages need to be investigated further. Early data from Reid et al.4,5 and Compston et al.3 suggest that FTM is superior to LTM in the prediction of BMD. However, LTM was found to be a better predictor of both total and regional BMD in premenopausal women regardless their physical activity level in a recent study, although the association of LTM with BMD was stronger in an exercise group than that in a nonexercise group.27
In another recent study by Aloia et al.,6 it was also found that both FTM and LTM were significant predictors of total bone mineral mass, but a larger proportion of variation in total bone mineral mass was explained by LTM than by FTM in both pre- and postmenopausal women. This finding led to the conclusion that LTM is the major determinant of total bone mineral mass and that FTM may have no significant physiological effect on bone. However, since FTM explained a greater percentage of the variation in total bone mineral mass among postmenopausal than among premenopausal women, an increased influence of FTM on bone in the postmenopausal years is suggested.
Our data from postmenopausal women agree with Aloia et al.6 In this study, although both LTM and FTM significantly predict total and most regional bone mass, LTM makes a greater contribution to both BMD and BMC, except for TBMD. However, the difference between the contribution of FTM and LTM to total bone mass is smaller in our study (33% in FTM vs. 43% in LTM) than in Aloia's study (21% in FTM vs. 47% in LTM). The difference between study results may be due in part to the fact that we did not exclude obese women from our sample while Aloia did. Also, the LTM was measured in the extremities by Aloia, while the present study used whole body values. It should be further noted that in our study LTM has a much higher correlation with BMC than with BMD, and after adjustment for FTM, LTM is still significantly associated with BMC, but not BMD. These results may reflect the fact that both LTM and BMC are more size-related and/or that there may be a closer genetic association between LTM and BMC. Although %FTM (and inversely %LTM) were correlated with BMD and BMC, significant correlation was primarily found only for total body BMD (or BMC). LTM and FTM are better predictors of both total and regional BMD and BMC.
BMC and BMD are two important quantity expressions of bone mineral measurement by densitometry. BMC is the bone mineral content within a given area and it is affected by the body frame size. BMD is the bone mineral area density calculated as BMC/area. Because BMD provides the best diagnostic sensitivity and the lowest precision error, Mazess et al.28 advocate BMD as the optimal expression for bone densitometry. However, the discussion of whether BMC, BMD, or even corrected BMD (volumetric density) should be used in osteoporosis research is continuing.27,29–31 In our study, both BMD and BMC were used to examine their different relationships to body composition.
Although increased LTM is a better predictor of higher BMD and BMC, changes in regional BMD are found to be more related to ΔFTM than to ΔLTM in our study. ΔFTM has been suggested by Houtkooper and coworkers13 as a significant predictor of ΔTBMD among premenopausal women. Reid et al.11 also found a significant relationship between ΔBMD and ΔFTM in postmenopausal women. ΔLTM is very small in the current sample, which could have an effect on the relationship between ΔLTM and ΔBMD. However, in Reid's study11 even when ΔLTM was larger than ΔFTM, it was still ΔFTM, not ΔLTM, that was the significant predictor of ΔBMD. Our study not only confirms their results but also suggests that in the same population, BMD (or BMC) and ΔBMD may be affected by different components (FTM vs. LTM) of body composition.
By investigating both cross-sectional and longitudinal associations of body composition and BMD in the same population, we reduced the effect of interpopulation variation on the study results. Our findings indicate that different mechanisms may predominate at different stages of life. The correlation between LTM and bone mineral mass is likely to reflect a genetic association between higher peak bone mass and higher lean tissue mass; a physically active lifestyle also induces increases in both bone and muscle mass. Since conversion of androgen to estrogen becomes a major source of estrogen only after menopause,7 the endocrine effect of FTM on bone may be minimal during growth. Our data and previous studies1,2 show a predominantly positive effect of body weight on bone. Although both increased FTM and increased LTM may increase weight loading on bone leading to higher bone mineral mass, LTM contributes a greater proportion of body weight during early adulthood. Therefore, LTM as the major component of weight plays a more important role in the increase and maintenance of bone mineral mass in premenopausal women. However, with increased age, and especially after menopause, FTM and weight increase while LTM decreases,10 changing the relative proportion of LTM and FTM to weight. Consequently, the association between FTM and bone mineral mass increases with age as more of the weight-loading on bone is attributable to FTM. At the same time, it is also likely that FTM assumes increased endocrine functions that influence bone mineral mass. However, both our findings and those of Aloia et al.6 indicate that despite the increase in the endocrine effects of fat tissue on bone after menopause, LTM is still the major determinant of BMC and BMD. Because of the important association of body weight and changes in body weight with bone mass and changes in bone mass, it is most likely that the major influence of LTM and FTM on bone is through weight loading, and the altered effect of FTM and LTM on bone following menopause may be explained by changes in body composition and response in the skeleton to changes in mechanical force.
Under most circumstances, the loss of bone mineral occurring in a year's time is only a small percentage of the total bone mineral mass. Therefore, the precision of techniques used to detect real change is crucial in determining the relationships between changes in BMD and other factors. We acknowledge that the difference in reliability of measurements may attenuate the significance of correlation between bone mineral mass and body composition to some degree. Longer follow-up time and larger sample size would enhance the statistical power of future studies on the relationship between bone mineral measurement and body composition. Nevertheless, the result of this study provides an important insight into the relative contribution of LTM and FTM to bone mass and rate of bone loss. Because of the significant correlation between Δweight and ΔBMD in the current study, we suggest a careful monitoring of the rate of bone loss for postmenopausal women who have experienced weight loss. More studies are needed to understand further the consequences and mechanism of bone loss after weight reduction and change in fat tissue mass.
In conclusion, our cross-sectional data indicated that LTM is a major determinant of bone mineral mass, especially for total and regional BMC. However, when approached longitudinally, change in body weight, especially change in FTM, is more closely associated with changes in regional bone mineral mass in healthy postmenopausal women.