Important variations in Z-score per vertebra, which is a common expres
sion of bone mineral density (BMD), are sometimes observed. The presen
t study evaluates the clinical significance of this heterogeneity. Nor
mal and osteoporotic subjects were defined by using strict criteria. F
or every scan, the minimal Z-score (the vertebra with the lowest Z-sco
re) and the delta Z (highest Z-score - lowest Z-score) was calculated.
Of the investigated subjects, 30% presented a delta Z greater than or
equal to 1. No significant correlation could be found between delta Z
and age, BMD, height and weight. There was no difference in delta Z b
etween scans of good, average or poor quality. Osteoporotic subjects h
ad significantly lowered BMD values, whether evaluated through Z-score
s for the L2-L4 site (P < 0.001; t = 3.71) or by minimal Z-score (P <
0.001; t = 3.97). Reproducibility calculated for the L2-L4 site on pha
ntoms as well as on patients was excellent (C.V. < 1%). When reproduci
bility was calculated on each vertebra in vitro or in vivo, an increas
e in variability was observed. These data show that marked heterogenei
ty in BMD per vertebra is not infrequent. In some subjects low BMD may
be measured at certain vertebrae but not at the total site. Our data
suggest that in those cases the lowest BMD should be considered. In fo
llow-up studies however, the BMD should be calculated on the L2-L4 seg
ment, since a loss of precision is observed when only one vertebra is
measured.