A. Resch et al., RISK OF VERTEBRAL FRACTURES IN MEN - RELATIONSHIP TO MINERAL DENSITY OF THE VERTEBRAL BODY, American journal of roentgenology, 164(6), 1995, pp. 1447-1450
OBJECTIVE. Vertebral fracture is the most frequent manifestation of os
teoporosis in women. Because there is a lack of information about bone
density and the occurrence of fractures in men with osteoporosis, we
evaluated the relationship between vertebral fractures and spinal bone
mineral density (BMD) to determine if there is a threshold of BMD bel
ow which fractures are likely to occur. SUBJECTS AND METHODS. Radiogra
phs of the spine and BMD measurements of the lumbar spine as measured
by quantitative CT were obtained in 201 men 21-86 years old (mean age,
68 +/- 4 years) who were referred consecutively for osteoporosis scre
ening. Radiographs were interpreted for the presence or absence of ver
tebral fractures. The probability of fractures was determined after cl
assifying the patients into subgroups according to their quantitative
CT values, The relationship of spinal bone mass to spinal fracture was
examined by both logistic regression and receiver operating character
istic (ROC) analysis. RESULTS. Seventy-one patients were classified as
having definite vertebral compression fractures. Spinal BMD was 132 /- 34 mg/ml for the 130 men without Vertebral fractures and 75 +/- 22
mg/ml for the 71 men with vertebral fractures (p < .001), The number o
f fractures per patient and the BMD were negatively correlated (r = -0
.71, p < 0.0001), When a BMD of 100 mg/ml was given as a fracture thre
shold, 15% of the patients without fractures were below this value and
14% of the patients with fractures were above this threshold, Quantif
ying the overlap between values from patients with and without fractur
es by ROC analysis, the value of 100 mg/ml gave a sensitivity of 86%.
Logistic regression showed 105 mg/ml as the most discriminate value, r
esulting in a sensitivity of 90%. Logistic regression analysis of the
predicted fracture probability also indicated that age does not signif
icantly influence the regression curve. CONCLUSION. We found that dire
ct quantitative CT measurement of the BMD of the vertebral body is a h
ighly efficient approach to distinguish men without vertebral fracture
s from those with fractures, Thus, a fracture-threshold concept could
provide a quantitative criterion to identify men at high risk for vert
ebral fractures.