Loss of Hip BMD in Older Men: The Osteoporotic Fractures in Men (MrOS) Study

INTRODUCTION

Fractures are a major health problem for the older population in the United States, even in older men, as about one third of hip fractures occur in men.(1) BMD is a major determinant of fracture risk in older men.(2,3) For example, in the Osteoporotic Fractures in Men (MrOS) Study, each SD decrease (SD = 0.137 g/cm²) in total hip BMD was associated with a 3.2-fold increase in the risk of incident hip fracture.(2)

Loss of hip BMD has been shown to increase linearly with age in women(4–9); many reports for men have been published.(10–25) However, the reports in men are limited to small studies of <350 older men(10–17,20,23,25); studies with few participants over age 80(14,18–25); studies that used differing absorptiometry machines between visits to determine change(10,12,16); and studies that observed change over <2 yr.(18) Other studies have focused on particular predictors of BMD loss, such as renal function or medication use, and have not focused on the patterns of BMD loss seen with age.(26–32) No previous study has formally examined heterogeneity in rates of change or evaluated whether BMD loss accelerates in a nonlinear manner with age in older men using longitudinal data.

The aim of these analyses was to describe hip bone loss (both in BMD and BMC) in older men; to test whether BMD is lost with age in a nonlinear manner; and to describe basic differences between men who maintain bone, men who lose bone at an expected rate, and men who have accelerated BMD loss.

MATERIALS AND METHODS

RESULTS

During an average of 4.6 yr of follow-up, average crude change in BMD at the FN was −1.72% (−0.013 g/cm²; range, −0.228 to 0.080 g/cm²) and average change at the total hip was −1.82% (−0.017 g/cm²; range, −0.184 to 0.068 g/cm²; Table 1). Change in BMC was similar to change in BMD: average BMC change at the FN was −1.43% (−0.059 g; range −0.887 to 1.28 g), whereas average BMC change at the total hip was −1.06% (−0.399 g; range, –5.751 to 6.093 g). FN area slightly increased with age, 0.44% during follow-up (0.024 cm²; range, −0.168 to 0.825 cm²). After adjustment for baseline weight, the estimated average hip BMD loss was somewhat reduced in magnitude.

Older age at baseline was strongly associated with increasing BMD loss at the FN. The β-coefficient for the age² (quadratic) term in clinical center–adjusted models was strongly statistically significant for the FN (β = 0.00007) and the total hip (β = 0.00013; p < 0.001 for both). The shape of the curve described by the model indicated that bone loss accelerated with each additional year of age in a nonlinear manner. Although the β-coefficient appeared small in magnitude, there were substantial differences in average bone loss between men at younger ages (∼65 yr) compared with men at very old ages (≥85 yr). For example, over an average of 4.6 yr, men 65 yr of age had an estimated loss of 0.008 g/cm² at the FN; men 75 yr of age had 0.014 g/cm²; and men 85 yr of age had 0.021 g/cm². Overall, men 85 yr of age at baseline were estimated to lose >2.5 times as much FN BMD during follow-up than men 65 yr of age at baseline, regardless of the baseline level of BMD. These changes are shown in Fig. 1.

In general, the higher the value of baseline BMD, the smaller the magnitude of loss in hip BMD and BMC (Fig. 2). This was more pronounced for change in BMC than for change in BMD for both the FN and total hip. Smaller FN area at baseline was modestly associated with greater loss of FN BMD over follow-up. Conversely, larger FN area at baseline was associated with greater loss in FN BMC over follow-up.

Although the average rate of loss was modest, there was variability in the rate of change in hip BMD in this cohort of older men. At the FN, 1149 (24.3%) participants maintained or increased BMD during follow-up, 2982 (63.2%) had “expected loss,” and 12.5% experienced “accelerated loss” (N = 589; Table 2). The mean change in FN BMD in the three groups over follow-up was increase of 0.012 g/cm² (+1.5%) in the maintained group; loss of 0.015 g/cm² (−2.0%) in the expected loss group; and a loss of 0.050 g/cm² (−6.6%) in the accelerated loss group. These differences are shown in Fig. 3. Similar variability in rate of loss was present in all age groups. Whereas older men (≥80 yr of age) tended to lose bone more quickly, 15.3% (N = 103) were classified in the maintained group. Younger men (65–70 yr of age) tended to lose bone more slowly; however, 7.6% (N = 116) of younger men were classified in the rapid loss group.

Most of the variation in BMD or BMC was caused by between-person variability: only 1.3–3.0% of the total variation in each of the BMD models was caused by within-person variation. The within-person variation for BMC was somewhat higher: 2.1–6.3% of the variation in each of the BMC models was caused by within-person variation. The largest within-person variation was seen for the FN area, as 14.1–15.2% of the FN variation was caused by within-person variation.

Maintenance of BMD was associated with younger age, greater baseline weight and height, higher total body fat and lean mass, higher BMI, and higher FN BMD and BMC at baseline (p ≤ 0.001 for all; Table 2). Men who reported white race were somewhat more likely to have expected or accelerated loss of BMD during follow-up compared with other races (p = 0.043). Neither fracture history, smoking status, nor alcohol use at baseline was significantly different between the categories of BMD change. FN area was associated with the bone loss category. Men who were classified into higher bone loss categories had smaller FN area at baseline (p = 0.003) and a greater increase in FN area over time (p = 0.020).

A cross-sectional decrease of 0.008 g/cm² in baseline BMD (Fig. 1), equivalent to the estimated amount of FN BMD lost for a 65-yr-old man during average follow-up, is associated with a small increase in the risk of hip fracture of 9% (RH per each 0.008 g/cm² cross-sectional decrease in femoral neck BMD: 1.09; 95% CI: 1.07, 1.11) over an average follow-up of 6.5 yr. A difference in baseline BMD of 0.021 g/cm², the average amount of FN BMD lost for a 85-yr-old man during follow-up, is associated with a larger increase in the risk of hip fracture, an increase of 25% (RH per each 0.021 g/cm² cross-sectional decrease in FN BMD: 1.25; 95% CI: 1.18–1.31) over an average follow-up of 6.5 yr. Use of models that incorporated BMD as a time-dependent covariate yielded similar results.

DISCUSSION

Although the average loss of BMD is modest in magnitude in healthy, ambulatory older men, BMD loss accelerates with each additional year of age; men 85 yr of age lose BMD more than twice as fast as men 65 yr of age. Several baseline characteristics, including lower baseline BMD and BMC and decreased body weight were associated with higher bone loss during follow-up.

The amount of BMD lost during follow-up in this cohort may be sufficient to considerably influence fracture risk, as seen from the results of cross-sectional differences in BMD in this cohort. A difference of 0.021 g/cm² of FN BMD, the expected amount of bone loss that occurred in men 85 yr of age at baseline over 4.6 yr of observation, translates to a 25% increased risk of hip fracture over follow-up for the whole cohort of 6.5 yr. Thus, the changes in BMD in this cohort, especially among the very old, may translate to significant increases in fracture risk. We also found that bone loss accelerated with aging, and this result may provide at least partial explanation for the exponential increase in the risk of fracture that occurs with aging.(37) Consistently, some(16,24) but not all others report(13) men with the lowest BMD at baseline lost bone most rapidly, suggesting that those already at highest risk of fracture are at an even higher risk as age advances and bone loss continues to increase.

There was considerable variation in the amount of BMD loss over time. Although older men tended to lose more bone, there was variability in the rates of loss at all ages. The majority of the variation in the amount of BMD loss over time was due to between-subject variation, which suggests that factors that differ between participants, rather than measurement error, explains the variation in BMD loss in this cohort. This also implies that the mechanisms underlying bone loss differ among older men, and we were able to identify some factors that were associated with higher loss: age, baseline BMD and lower weight. Adjustment for weight at baseline somewhat reduced the adjusted absolute and percent change for all hip locations, consistent with previous data from this(38) and other cohorts.(10,12,24) Several factors, including smoking, fracture history, and alcohol use, likely risk factors for fracture in older people,(39,40) were not associated with accelerated BMD loss. Because both a single measure of BMD(2,3,41) and change in BMD(42) are strongly associated with fracture risk, understanding the source of variation in change in BMD is important for future research.

A subset of men had more rapid loss of BMD with aging. As expected, men with accelerated bone loss were, on average, older than men who maintained BMD. However, the age range of men in the accelerated loss group was wide, as even men in their 60s and 70s were classified as having accelerated loss (Fig. 3). On the other hand, there was a narrower range of age for men who maintained BMD; the age of men in this group appeared to cluster around the younger ages, and only a few men in this cohort maintained BMD into their 80s.

Changes in measured FN bone area may be caused by either true changes in bone size (from the postulated phenomenon of periosteal apposition(19,43)) or artifactual changes caused by measurement error. In these analyses, the factor with the most with-in person variability was FN area, which suggests that the FN area measure has a fair amount of measurement error. Increasing bone width (and thus area) over time without change in BMC would result in a decline in measured BMD. In this cohort, there was evidence for increasing FN area over time. Whereas it is not possible to determine whether the changes in bone size seen in MrOS are caused by periosteal apposition or measurement error or a combination of these factors, it is likely that at least some of the BMD loss observed in the cohort is caused, in part, by increasing FN area rather than loss of mineral per se. However, in these analyses, the amounts of BMC lost over time are similar in magnitude to the amount of BMD lost over time, suggesting that decreased BMC, rather than increasing bone area alone, primarily explains the decline of BMD observed. Other imaging modalities, such as central or pQCT, should be used to better understand volumetric and compartmental changes in bone over time in older adults.

Changes in BMD, BMC, and FN area are reported as absolute or percent change during follow-up. Because the rate of bone loss increases nonlinearly as age increases, the rate of bone loss and percent change variables were not annualized across all years of follow-up. Annualized rates would overestimate bone loss at younger ages and underestimate bone loss during later years; this would be a particular problem at older ages as the bone loss per each extra year of life increases nonlinearly with age. Inability to report these results as annualized changes limits the comparison of these data to those from other cohorts. Studies from other cohorts have reported a range of decline in BMD for men of similar ages. Estimates of annual rates of change in BMD at the FN range from −0.19%/yr(14) to −0.40%/yr(18) in men ≥50 yr of age. One study reported extremely high rates of FN bone loss in older men (> −2%/yr); the higher rates observed in this outlying study may have been caused by the use of data from two different models of DXA scanners to ascertain change (Holgoic 2000 versus 4500)(16) To compare results from these other studies with those from MrOS, we estimated the amount of bone loss that would have been observed during the MrOS follow-up time for each cohort that reported an annualized rate of FN bone loss; data from the outlying study were ignored. Had the same follow-up time (average, 4.6 yr) been observed in these other cohorts as in MrOS, the range of decline would have been between −0.87% and −1.84% in the other studies over the follow-up period seen in MrOS. Our estimate of FN bone loss during follow-up (−1.72%) falls within this range.

This study has several strengths. MrOS is a very large, well-characterized cohort with repeat measures of BMD and BMC completed on the same absorptiometry machines over time. Rigorous cross-sectional and longitudinal quality assurance procedures were used for measures of bone mass, content, and area. The use of mixed effects models allows for multiple data points per participant and different intervals between visits and accounts for between- and within-subject variability. However, a few limitations should be noted. The amount of change observed in this cohort is very small and may approach the limits of precision of the DXA scanners. Additionally, the MrOS participants were quite healthy and ambulatory at baseline, and participants who returned for subsequent visits and DXA scans were healthier than participants who did not return to the clinics. Thus, application of these results to other populations, such as the infirm or institutionalized, may be limited. It is likely that the estimates of BMD loss from this cohort underestimate the “true” amounts of bone loss that would be observed in the general older population, because men with the most bone loss are perhaps the least likely to be scanned at follow-up visits.

In conclusion, hip bone loss, on average, is modest in magnitude in older men, although there is variation in the amount of BMD loss over time. Loss of BMD accelerates with increasing age and is the most rapid in those who have the lowest BMD at baseline. These decrements in BMD over time are likely to be associated with an increased hip fracture risk, and our findings provide insight into the increase in fracture risk that occurs with age. Future research should aim to further identify and characterize the subset of men who lose bone most rapidly and to understand the mechanism of loss.