Subjects were 164 working men (machine drivers,
concrete workers and office workers),
aged 40-45 years,
and 26 young men who had not had long lasting low back pain,
aged 18-20 years.
Sagittal T2-weighted (T2WI) and PD-weighted (PDWI) MR images of the lumbar spine were obtained at 0.01T using a dual echo (GE and SE sequence) technique (DE2000/25-86) with 2 acquisitions.
The slice thickness was 7mm,
FOV 410x410mm and the pixel size 1.6 x 1.6 mm.
Lumbar discs from L2/L3 to L5/S1 were visually analyzed using reference images.
Changes in annulus fibrosus,
nucleus pulposus,
bony endplates and subchondral bone were separately graded in detail [2] according to criteria in table 1.: 1.
relative size and shape of the nuclear complex – central bright region of the disc (on T2WI) including nucleus pulposus and the inner annulus (N complex),
2.
decrease of signal intensity (SI) of the nucleus pulposus (NP DSI),
3.
inhomogeneity of the signal of the nucleus pulposus (NP inhom),
4.
presence and irregularity of the intranuclear cleft (INC),
5.
focal dark spot on INC (INC DOT),
6.
inhomogeneity (AF inhom),
7.
irregular inner border (AF border) and 8.
bulging or protrusion (Bulge) of the annulus fibrosus anteriorly (ant) or posteriorly (post),
on T2WI or PDWI,
9.
bony endplate lesions (EPL) and 10.
subchondral signal changes in bone marrow (Modic).
Most variables had 3-5 grades on an ordinal scale,
from normal to the most degenerated finding.
In case image quality was not good enough for evaluation of certain disc,
the observation was recorded as missing. Most of them were at L5/S1 disc level.
Three experienced radiologists evaluated all discs.
Interobserver agreement was assessed calculating weighted kappa-values between any two observers.
The evaluations of the most experienced observer were used in final analyses.
First,
prevalences of each graded variable were calculated separately at each disc level and compared between age groups.
Variables showing strongest differences in prevalences of degenerative findings between age groups were chosen for further analyses.
Two combinations of MRI variables (N complex,
NP DSI,
NP inhom and Bulge in table 2) and (NP DSI,
INC,
AF inhom and Bulge in table 3) were separately observed.
Prevalences of all possible combinations of different grades of variables were calculated at each disc level L2/L3-L5/S1,
separately for both age groups.
Comparison between young and middle-aged men was made.
A more detailed analysis was performed among the middle-aged men for 11 MRI variables: size and shape of the nuclear complex (N complex), signal intensity of nuclear complex (NP DSI),
inhomogeneity of nuclear complex (NP inhom),
regularity of intranuclear cleft (INC),
focal dark spot in INC (INC DOT),
inhomogeneity of annulus fibrosus anteriorly and posteriorly (AF inhom ant and post),
regularity of inner border of annulus fibrosus anteriorly and posteriorly (AF border ant and post),
defect in the bony end plate (EPL) and subchondral signal abnormality (Modic),
to find which of them have the strongest association with the other variables.
Correlation coefficients (Pearson’s r) were calculated first for each pair of degenerative variables separately at each disc level.
Then a factor analysis was performed with three factors’ solution to show which of the chosen variables best describe the degenerative process together.
Since the analysis found two important factors,
the final factor analysis was performed excluding the variables which showed the weakest association with any factor.