Is Bone Mineral Density Predictive of Fracture Risk Reduction?

Charles A. Cefalu


Curr Med Res Opin. 2004;20(3) 

In This Article

Bone Mineral Density Increase and Fracture Risk Reduction

The lower the BMD, the greater the fracture risk.[8] There is a consistent doubling of fracture risk for each SD reduction in BMD, irrespective of fracture type and BMD measurement site (Figure 1). However, an inverse relationship between fracture risk reduction and increasing BMD has not been demonstrated in clinical trials of osteoporosis therapies.

Relationship between BMD T-score (before treatment) and fracture risk. BMD = bone mineral density; SD = standard deviation

At issue is not whether increases in BMD are associated with vertebral fracture risk reduction. All major studies of antiresorptive therapy for post-menopausal osteoporosis have shown a positive relationship. However, there appear to be other factors that account for some of the variability between BMD increase and fracture risk reduction.

Two meta-analyses of antiresorptive therapy used regression analysis to estimate the overall degree of vertebral fracture risk reduction attributable to spine BMD increase.[9,10] Risk reduction was estimated to be 54% for an 8% BMD increase[10] and, more recently, 3% for a 1% BMD increase.[9] Thus, the earlier meta-analysis determined that changes in BMD appeared to explain most of the observed risk reduction,[10] whereas the more recent meta-analysis determined that it explained only a fraction of the observed risk reduction.[9] For the individual trials of antiresorptive therapy evaluated in the more recent meta-analysis, changes in BMD accounted for as little as 9% and as much as 52% of the observed risk reduction.[9] The discrepancy between the results of these meta-analyses may be a consequence of several factors. In the meta-analyses, confidence intervals were large, both in the analyzed studies and in the meta-analysis estimates. More important, the most recent meta-analysis included several additional major trials[11,12,13,14] and weighted trials according to size. In both meta-analyses, the BMD increase was assumed to be immediate, linear, and continuous in the analyzed trials. This, however, was not always the case (Figures 2-4); in one trial, BMD increase with antiresorptive therapy peaked at 1 year and was maintained over the second year but decreased thereafter (Figure 4).

Mean percentage change from baseline in lumbar spine BMD in the VERT-NA trial. (Reprinted with permission from Harris ST, Watts NB, Genant HK, et al. JAMA 1999.[11]) (BMD = bone mineral density; VERT-NA = Vertebral Efficacy With Risedronate Therapy-North American study

Mean percentage change from baseline in lumbar spine BMD in the MORE trial. (Reprinted with permission from Ettinger B, Black DM, Mitlak BH, et al. JAMA 1999.[13]) BMD = bone mineral density; MORE = Multiple Outcomes of Raloxifene Evaluation

Mean percentage change from baseline in lumbar spine BMD in the PROOF trial. (Adapted from Chesnut CH, Silverman S, Andriano K, et al. Am J Med 2000.[14]) BMD = bone mineral density; PROOF = Prevent Recurrence of Osteoporotic Fractures

Moreover, the relationship between BMD increase and fracture risk reduction is not proportional. Rather, there seems to be (1) a markedly reduced fracture risk, even with small increases in BMD; (2) a fracture risk reduction that may occur rapidly (e.g., within 1 year) and not increase despite subsequent and perhaps further BMD build; and (3) no dosage-related increase in antifracture efficacy despite dosage-related increases in BMD. Hypotheses offered by the authors suggest that other factors may be responsible for at least some of the variability between increases in BMD and the fracture risk reduction observed with osteoporosis therapies.

The bisphosphonates risedronate and alendronate are effective agents for reducing vertebral and nonvertebral fracture risk.[11,12,15,16,17,18] In major clinical trials, risedronate and alendronate reduced vertebral fracture risk significantly and to a similar degree, approximately 40-50% after 3 or 4 years, despite slightly greater increases in spine BMD with alendronate ( Table 1 ). Risedronate significantly reduced nonvertebral fracture risk, possibly as a consequence of the reduction in hip fracture risk in a broader range of women with low femoral neck BMD. Risedronate reduced hip fracture risk in those with pre-existing vertebral fracture (-60%, p = 0.003) and in those with osteoporosis with or without pre-existing fracture (-40%, p = 0.009).[15] Alendronate reduced hip fracture risk in those with pre-existing vertebral fracture (-51%, p = 0.047) [19]but not in those without pre-existing vertebral fracture (-21%, p = 0.44).[16]

Theoretically, if the increase in BMD by the bisphosphonates was solely responsible for the reduction in fracture risk, a 1-SD increase in BMD should produce an approximate 50% reduction in fracture risk. In reality, small increases in BMD are associated with markedly reduced fracture risk, as illustrated by the Fracture Intervention Trial (FIT) prevalent vertebral fracture study.[17] Baseline T-scores in the spine, hip, and forearm increased by approximately 0.5, 0.2, and 0.1 SDs, respectively, during treatment with alendronate for 3 years.[20] Assuming a proportional inverse relationship between BMD increase and fracture risk reduction, the expected fracture risk reductions would be 25, 10, and 5%.[20] In contrast, the observed reductions were 47% (radiographically documented) and 55% (clinically apparent) for the spine, 51% for the hip, and 48% for the forearm, which are more than double that expected for the spine and several-fold greater than that expected for the hip and forearm.

In addition to a markedly reduced fracture risk associated with a small increase in BMD, fracture risk reduction can occur rapidly (e.g., within 1 year) and does not increase despite subsequent or additional BMD build. This is illustrated by the results of the Vertebral Efficacy With Risedronate Therapy - North American (VERT-NA) trial.[11] In this trial, postmenopausal women with ≥ 2 vertebral fractures or at least 1 vertebral fracture and low spine BMD (T-score below -2) at baseline were randomly assigned to receive risedronate or placebo. During the first year of therapy with risedronate 5 mg, there was a rapid but relatively small increase in BMD (approximately 3% difference from placebo in change from baseline; Figure 2) and a large rapid reduction in vertebral fracture risk (65%; 95% confidence interval [CI], 38-81%; p < 0.001). Despite continued increase of BMD between years 1 and 3 of risedronate therapy, there was no further increase in the vertebral fracture risk reduction. Similar results were reported in the multinational VERT trial (VERT-MN), in which there was also an approximate 3% difference from placebo in BMD increase and a dramatic reduction in vertebral fracture risk during the first year (61%; 95% CI, 32-78%; p = 0.001), with no further increase in the vertebral fracture risk reduction despite continued BMD increase during the subsequent 2 years of therapy. Thus, in both VERT studies, the maximum antifracture response of risedronate was observed within the first year of therapy, before the full effect on BMD was realized. Cox analyses examining the relationship between vertebral fracture and lumbar spine BMD change were applied to pooled data from the VERT studies. Changes in BMD predicted only 11.6% (at 2 years)[21] and 17% (at 3 years)[22] of the vertebral fracture risk reduction, leaving most of the reduction unexplained. Data from four risedronate studies showed that lumbar spine BMD changes (4.4%) over 3 years in women with low BMD but no prevalent fractures were not large enough to explain the 75% reduction in vertebral fracture risk.[23]

The absence of a proportional relationship between BMD build and therapeutic benefit is also illustrated by a smaller double-blind, placebo-controlled study in postmenopausal women aged 45-80 years (N = 994), with osteoporosis defined as a lumbar spine BMD at least 2.5 SD below the young adult mean.[18] The women were randomized to alendronate (5, 10, or 20 mg per day for 2 years, followed by 5, 10, and 5 mg, respectively, for a third year) or to placebo. The 10-mg dose was significantly more effective than the 5-mg dose at increasing BMD (increases of approximately 8 and 5%, respectively), but the incidence of vertebral fracture was similar (2.9 and 2.8%, respectively). Thus, there was no dosage-related increase in antifracture efficacy despite a dosage-related increase in BMD.

The Multiple Outcomes of Raloxifene Evaluation (MORE) trial determined that raloxifene significantly reduces vertebral fracture risk but not nonvertebral fracture risk ( Table 1 ).[13] In this double-blind study, 6828 postmenopausal women with osteoporosis were randomized to placebo or to raloxifene 60 or 120 mg/day for 3 years. In the subgroup consisting mostly (approximately 90%) of women with prevalent vertebral fracture, there was a 50% reduction in the risk of new vertebral fracture in the raloxifene 60-mg/day arm. This is similar to the vertebral fracture risk reduction seen with bisphosphonate treatment in women with prevalent vertebral fracture. Raloxifene 60mg/day was associated with a rapid but small increase in BMD [13] (approximately 3% difference from placebo in change from baseline; Figure 3) and a large rapid reduction in clinical vertebral fracture risk (68%; 95% CI, 20-87%) [24]during the first year of therapy. No further increase in the vertebral fracture risk reduction was observed during the subsequent 2 years of therapy, which is another similarity to the bisphosphonates. However, in contrast to the bisphosphonates, there was little continued increase in BMD between years 1 and 3 of raloxifene therapy, and the total increase, approximately 2.6%, was less than half that of the bisphosphonates ( Table 1 ). Thus, after 3 years, raloxifene 60 mg/day achieved vertebral fracture risk reduction similar to the bisphosphonates, with approximately half the increase in BMD. This markedly reduced fracture risk with a small increase in BMD further supports the absence of a proportional relationship between BMD increase and fracture risk reduction. Indeed, logistic regression analysis of all the women treated with raloxifene in the MORE trial (60 or 120 mg/day) determined that the BMD changes associated with raloxifene treatment accounted for only 4% of the observed vertebral fracture risk reduction.[25]

The Women's Health Initiative trial, a large randomized trial in over 16 000 predominantly healthy postmenopausal women, was designed to evaluate the effect of estrogen-progestin (hormone) therapy (HT) on a number of chronic diseases of older women. In the overall population, HT reduced the risk of hip and clinical vertebral fractures by 35% after a mean follow-up of 5.6 years.[26] In a subgroup who had BMD measurements, only 4% of patients in the HT and 6% in the placebo group had total hip BMD T-scores < -2.5. BMD increases were 3.3 and 2.1% at the lumbar spine and hip, respectively, at 1 year and 4.5 and 3.6%, respectively, after 3 years. In patients with BMD measurements at 6 years, lumbar spine BMD was increased by 7.5%. BMD increases and fracture risk reductions through year 3 were similar to increases observed with the bisphosphonates.

The Prevent Recurrence of Osteoporotic Fractures (PROOF) trial is a 5-year dose-ranging trial evaluating salmon calcitonin nasal spray (100, 200, and 400 IU/day) in postmenopausal women with osteoporosis.[14] Only 200 IU/day, which is the recommended dosage regimen,[27] reduced the vertebral fracture risk by 33% (p = 0.03) in women with low bone mass and by 36% in the subset with prevalent vertebral fractures (p = 0.03). However, the 200-IU dosage did not significantly reduce the nonvertebral fracture risk; reduction in nonvertebral fracture risk was statistically significant only in the 100-IU/day group (p < 0.05) ( Table 1 ). Lumbar spine BMD increased from baseline in all dosage groups. However, when compared with placebo, all three dosages were significantly different for the first 2 years; only the 400 IU/day was significantly different from placebo during the third year. Therefore, although BMD increases were dose dependent, fracture risk reduction was not. Furthermore, in comparison with other antiresorptive therapies, calcitonin 200 IU/day produced only a very modest increase in spine BMD (< 1%) that peaked at 1 year (Figure 4).

Parathyroid hormone (PTH) or teriparatide,[28] is the only anabolic osteoporosis agent available for clinical use. In a placebo-controlled trial in postmenopausal women with osteoporosis, teriparatide produced a similar reduction in the incidence of vertebral fractures (65% with 20 µg/day and 69% with 40 µg/day) as seen with the bisphosphonates. However, increases in spine BMD were dose related. Furthermore, the BMD increase at the lumbar spine with the 20-µg/day teriparatide dose (9%) was greater than increases observed with the recommended doses of alendronate and risedronate (3-8%).[29]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.