The ability to more accurately predict future fracture risk in individual patients should assist health care providers as they decide among medical therapies for women with postmenopausal osteoporosis. Bone mineral density (BMD), age and prevalent fractures are key determinants of fracture risk; however, an assessment of vertebral fracture status is often not included in the clinical evaluation of postmenopausal women at risk for osteoporosis. To determine the impact of radiographic prevalent vertebral fractures (T4-L4) on vertebral fracture risk, independent of lumbar spine BMD, we examined data from 2651 postmenopausal women (1181 with prevalent vertebral fractures) from the placebo groups of the Fracture Prevention Trial (median observation 21 months) and the MORE trial (observation 2 years). Prevalent vertebral fracture status was defined as: 1) prevalent vertebral fracture (no/yes); 2) prevalent vertebral fracture number (0, 1, 2, ³ 3); 3) maximum semi-quantitative (SQ) deformity grade [mild (grade 1), moderate (grade 2), severe (grade 3)(Genant et al.1983)]; and 4) spinal deformity index (SDI) score. SDI is the sum of SQ scores for T4 to L4. A logistic regression model was used to assess the relationship between prevalent vertebral fracture status and fracture risk across baseline lumbar spine BMD. The relationship was modeled as a function of prevalent vertebral fracture status, baseline lumbar spine BMD and the interaction.

As expected, incident vertebral fracture risk increased as lumbar spine BMD decreased. Compared to patients without a prevalent vertebral fracture who had the same T-score, the risk of an incident vertebral fracture was approximately tripled in patients with at least 1 prevalent radiographic vertebral fracture. Greater numbers of prevalent vertebral fractures were associated with greater risks (2 fractures = ~ 5-fold increase, ³ 3 fractures = 7 to 8-fold increase). Compared to patients without a prevalent vertebral fracture who had the same T-score, the risk of an incident vertebral fracture was approximately tripled in patients with an SQ score of at least 1. Higher prevalent SQ scores were associated with greater risks (SQ score of 2 = ~ 5-fold increase, SQ score of 3 = 8 to 10-fold increase). Similarly, higher prevalent SDI scores were associated with greater risks (SDI score of 1 to 3 = ~ 3-fold increase, SDI score of 4 to 6 = 6 to 7-fold increase, SQ score of ³ 7 = 9 to 11-fold increase, Table). These findings indicate that at any given lumbar spine BMD, the incorporation of information regarding prevalent vertebral fracture status impacted the risk of incident vertebral fracture by up to 11-fold.

	Prevalent Vertebral Fracture		SDI Score
Lumbar Spine T-Score	No	Yes	0	1-3	4-6	³7
-2	2.1	10.9	2.2	7.0	17.0	25.2
-3	3.1	15.4	3.1	9.8	22.8	32.6
-4	4.5	21.3	4.4	13.5	29.7	41.0

Values refer to incident vertebral fracture risk (%) over a median observation period of 23 months.