Nfa. Peel et al., IMPACT OF ANOMALOUS VERTEBRAL SEGMENTATION ON MEASUREMENTS OF BONE-MINERAL DENSITY, Journal of bone and mineral research, 8(6), 1993, pp. 719-723
In anteroposterior (AP) bone mineral density (BMD) measurements of the
lumbar spine (LS), the presence of ribs is used to identify vertebra
T12. Similarly, in lateral LS-BMD measurements, the position of the il
iac crest is used to identify the lumbar vertebrae. The aim of this st
udy was to determine the impact of variations in spinal segmentation a
nd iliac crest position on BMD measurements. In 375 women (ages 50-85
years) radiographs were taken of the thoracic and lumbar spine, as wel
l as AP measurement of LS-BMD, by dual-energy x-ray absorptiometry (DX
A). In 121 subjects lateral decubitus LS-BMD was also measured. Anomal
ous spinal segmentation was found in 16.5%, and L1 would have been inc
orrectly identified on the AP-DXA image in 13%. The change in BMC and
BMD between adjacent vertebrae was greater in the upper than the lower
lumbar spine. Misidentification of L1 for T12 resulted in underestima
tion of the bone mineral content in grams (BMC) of L1 by a mean of 11.
5 +/- 14.4% (SD; range -33.5 to 33.5%). For the usual region of intere
st, L2-4, the BMC (g) was underestimated by 8.4 +/- 8.7% (range -1.5 t
o 29.2%), with the BMD (g/cm2) underestimated by 3.6 +/- 4.8% (range -
5.4 to 11.6%). The position of the iliac crest on the lateral decubitu
s DXA scans would have led to misidentification of either L2 or L4 for
L3 in 15 cases (12%). This resulted in the BMD of L3 being underestim
ated by 2.7 +/- 19.4% (range -242.4 to 34.6%). We conclude that (1) an
atomic variations in spinal segmentation are common; (2) the impact of
mididentification of vertebra L1 on the AP scan is considerable if on
ly one lumbar vertebra is studied but small if a larger area is analyz
ed and expressed as g/cm2; and (3) misidentification of L3 in the late
ral projection can result in a large error in the BMD measurement of L
3, although the mean error in terms of a population is small. Thus the
impact of misidentification of lumbar vertebrae does not result in im
portant overall changes when studying groups of subjects but on an ind
ividual basis could lead to incorrect diagnosis of osteopenia.