Effect of Aging and Degeneration on Disc Volume and Shape: A Quantitative Study in Asymptomatic Volunteers

Transcription

1 Effect of Aging and Degeneration on Disc Volume and Shape: A Quantitative Study in Asymptomatic Volunteers Christian W.A. Pfirrmann, 1 Alexander Metzdorf, 2 Achim Elfering, 3 Juerg Hodler, 1 Norbert Boos 2 1 Department of Radiology, Orthopedic University Hospital, Balgrist, Switzerland 2 Centre for Spinal Surgery, University of Zurich, University Hospital Balgrist, Forchstrasse 340, CH-8008 Zurich, Switzerland 3 Department of Psychology, University of Berne, Berne, Switzerland Received 1 February 2005; accepted 22 September 2005 Published online 11 April 2006 in Wiley InterScience ( DOI /jor ABSTRACT: Debate continues on the effect of disc degeneration and aging on disc volume and shape. So far, no quantitative in vivo MRI data is available on the factors influencing disc volume and shape. The objective of this MRI study was to quantitatively investigate changes in disc height, volume, and shape as a result of aging and/or degeneration omitting pathologic (i.e., painful) disc alterations. Seventy asymptomatic volunteers (20 78 years) were investigated with sagittal T1- and T2- weighted MR-images encompassing the whole lumbar spine. Disc height was determined by the Dabbs method and the Farfan index. Disc volume was calculated by the Cavalieri method. For the disc shape the disc convexity index was calculated by the ratio of central disc height and mean anterior/posterior disc height. Disc height, disc volume, and the disc convexity index measurements were corrected for disc level and the individuals age, weight, height, and sex in a multilevel regression analysis. Multilevel regression analysis showed that disc volume was negatively influenced by disc degeneration ( p < 0.001) and positively correlated with body height ( p < 0.001) and age ( p < 0.01). Mean disc height and the disc convexity index were negatively influenced by disc degeneration but not by gender, weight, and height. Disc height was positively correlated with age ( p < 0.01). From the results of this study, it can be concluded that disc degeneration generally results in a decrease of disc height and volume as well as a less convex disc shape. In the absence of disc degeneration, however, age tends to result in an inverse relationship on disc height, volume, and shape. ß 2006 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 24: , 2006 Keywords: aging; degeneration; disc volume; disc shape; quantitative MRI; intervertebral disc INTRODUCTION Although it is the general view that a decrease in disc height is an intrinsic feature of disc degeneration, there is no quantitative data in the literature confirming this view. 1,2 On the contrary, some studies question this relationship. Nachemson et al. 3 did not find a relationship of disc space narrowing and aging or degeneration in a small series of cadaver spines. Twomey and Taylor 4,5 reported that average disc height is maintained or even increased in old age as the disc sinks into the vertebrae. Berlemann et al. 6 investigated the relationship between height, shape, and histological changes in early degeneration in the L4/5 and Correspondence to: Norbert Boos (Telephone: ; Fax: ; norbert.boos@balgrist.ch) ß 2006 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. L5/S1 disc level. They concluded that early histologic changes do not significantly influence intervertebral disc height. The drawback of the aforementioned studies was that the data was based on cadaveric spines which do not reflect the actual in vivo situation. 7 While Brinckmann et al. 8,9 provided extensive quantitative data on the influences of age and work load on disc height as well as on vertebral and discal shape using distortion-compensated roentgen analysis (DCRA), a simultaneous assessment of the extent of disc degeneration was not possible by radiography. An assessment by MRI overcomes this disadvantage and the measurements are not prone to radiographic distorsion. Several studies so far used MRI to investigate the relationship of disc height and disc degeneration Frobin et al. 10 and Luoma et al. 12 found that disc height decrease appears not to be associated with early degeneration. However, no quantitative data on the effect of disc degeneration on disc volume and shape is 1086 JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006

2 EFFECT OF AGING ON DISC VOLUME AND SHAPE 1087 available. Quantitative data on the morphologic parameters of disc degeneration are essential to better understand the mechanisms involved in the degenerative cascade which ultimately leads to a destruction of the motion segment. So far, MRI has not yet been used to quantitatively investigate the changes in lumbar intervertebral disc volume and shape which occur with aging and/or degeneration. The objective of this MRI study was therefore to quantitatively investigate changes in disc height, volume, and shape as a result of aging and/or degeneration omitting pathologic (i.e., painful) disc alterations. METHODS Study Population The rationale for using an asymptomatic population was based on the general notion that it is very difficult to distinguish physiologic changes of aging and those resulting from pathologic (i.e., painful) degeneration. 13 We therefore rather focused on a well-defined asymptomatic population to be able to provide baseline data on the normal aging process (i.e., painless degeneration). For the purpose of this study, we use the term disc degeneration without implying that disc degeneration is synonymous with a painful disc. The study protocol was approved by the institutional review board. The subjects were recruited from patients referred for nonspinal musculoskeletal disorders and from hospital staff. Inclusion criteria were: 1) no back pain within the last 5 years, and 2) willingness to undergo an MRI scan of the lumbar spine. Volunteers were excluded if they ever were absent from work because of low back pain or if they had to consult a physician because of low back pain. Individuals were also excluded in the presence of spinal fractures, diagnosed osteoporosis, or serious illness (e.g., tumor, infection). According to these inclusion and exclusion criteria, 70 individuals were recruited (10 volunteers for each decade, 2nd to 8th decade). The 37 men and 33 women had an average age of 48 years (20 78 years). MRI Protocol Each volunteer underwent an MR examination encompassing the entire lumbar spine. MRI was always performed in the afternoon in order to account for the known influence of diurnal variations on disc height and volume. 14,15 All individuals reported that they had continued with their regular activities of daily living for at least 6 hours without lying down prior to the scan. MRI of the lumbar spine was performed on a 1.0-T scanner (Siemens Impact Expert; Siemens Medical Systems, Erlangen, Germany) using a dedicated receive only spine coil. The imaging protocol included sagittal T1-weighted (TR/TE: 700/12 ms) and T2-weighted turbo spin echo images (5,000/130) with the following parameters: matrix, ; field of view, cm; section thickness, 4 mm; interslice gap, 0.8 mm. The images encompassed the complete vertebral bodies and intervertebral discs of the lumbar spine. Image Analysis Measurements of the intervertebral discs were performed on the midsagittal slice using a graphics tablet (Intuos A5, Wacom Technology, Vancouver, WA) with a spatial resolution of 2,540 lines per inch. Disc height (Fig. 1) was calculated using the Dabbs 16 method (i.e., mean of anterior and posterior disc height) and the Farfan 17 index (sum of anterior and posterior disc height divided by disc diameter), respectively. The Farfan index was developed for an assessment of disc height using standard radiographs. The division of the average anterior and posterior disc height by disc diameter was necessary to account for a compensation of varying magnification factors on conventional radiographs and body height. We used the Farfan index to allow for a comparison with data based on radiographs. Disc convexity was calculated as the ratio between central disc height and the mean of the anterior and posterior disc height (Fig. 1). Values greater than 1.0 indicate a convex disc, values smaller than 1.0 a concave one. The measurements were performed on a midsagittal MR image of each disc. Disc volume was calculated by the point-counting (Cavalieri) method (Fig. 2) which is an accepted method 18 to estimate the volume of an arbitrary object based on sectional images obtained by Figure 1. Measurement of disc height (a, anterior disc height; c, central disc height; p, posterior disc height; d, diameter). The anterior and posterior landmarks for measuring disc height were defined as the smallest cranio-caudal thickness of the intervertebral disc, anterior and posterior, respectively. The landmarks for the central disc height measurements were defined as the largest cranio-caudal thickness in the disc center. The disc diameter was defined as the outer borders of the anulus fibrosus. DOI /jor JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006

3 1088 PFIRRMANN ET AL. CT or MRI A recent study has demonstrated that an edge-tracing method and the Cavalieri method reveal very comparable results but that the latter has the advantage of greater speed and, therefore, efficiency. 19 The point-counting method consists of overlaying, uniformly at random on the section (sagittal image), a transparent film containing a systematic array of test points, and counting the number of occasions on which a grid point lies within the feature. 18 The random overlay avoids a possible bias due to the parallel alignment of the endplates. A square grid with a distance of 3 mm between the test points was used for this analysis. The use of 4-mm slices with a 0.8-mm interslice gap in combination with a 3-mm grid results in a voxel size of mm which may be responsible for only a negligible imprecise volume assessment using the Cavalieri method. 19 The crosssectional area was then calculated by counting the number of grid points. In a next step, disc volume was calculated by multiplying the grid covered disc area by the slice thickness and by accounting for the magnification factor and interslice gap (Fig. 2). All sagittal volume segments were summed to calculate the total disc volume. These measurements were done for each lumbar disc of the whole lumbar spine. Disc degeneration was assessed on the midsagittal T2-weighted MR images (Fig. 3) and graded based on the decrease of signal intensity and the loss of distinction between annulus and nucleus For the purpose of this study, disc height was not considered as a criterion for the grading of disc degeneration because of the obvious bias with regard to the volume assessments (Fig. 3). All images were read by two observers blind to the age of the individual. After the initial independent review was concluded, disagreements on the ratings were resolved in conference (i.e., open discussion). In none of the cases, the raters were unable to achieve concordance. The resulting consensus assessment was used for further statistical analysis. Figure 2. Determination of the disc volume by the Cavalieri point counting method. 18 The Cavaleri method consists of overlying a transparent film with a square grid (3 mm) on the target feature (i.e., intervertebral disc) and counting the number of occasions on which a grid point lies within the target feature. The volume of each disc slice is calculated by multiplying the grid covered disc area with the slice thickness (accounting for the magnification factor). All sagittal volume segments were summed to calculate the total disc volume. Data Analysis The reliability of the measurements (interrater error) were assessed by calculating coefficients of variation, Pearson correlation, and intraclass correlation coefficients, respectively. The potential influences of disc degeneration, body weight, height, age, gender, and disc level on disc height (Dabbs method, Farfan index), disc volume, and disc convexity index were explored using a multilevel regression approach. 28 A problem of the analysis of disc data is that the particular measurements within a person (five discs per individual) are interrelated. One way of statistical control is to introduce n-1 person variables as covariates into a regression analysis. This type of control ignores individual differences and has several other disadvantages. 28 Another approach is to aggregate the measures within persons (for instance sum or mean values). However, this approach will reduce the power of the analysis. It implicates the loss of a whole case when a particular measure is missing, and it loses all information about differences within persons. A solution to these problems is the multilevel approach, that has been developed in the context of educational research and is more and more common in areas of medical research. 29 The multilevel approach is an extension of the multiple regression analysis, and allows analyses of data with a hierarchical structure, like discs within individuals. Multilevel analyses provide intercept and slope parameters that are similar to the nonstandardized coefficients in an ordinary multiple regressions analysis. However, these parameters are adjusted for the amount of dependency that is due to the hierarchical structure in the data, i.e., discs within persons tend to be more similar than between persons. The normal distribution of all independent variables was controlled using a Kolmogoroff- Smirnov test. Standard statistical software (SPSS 12.0) and multileveling software MLwiN 30 was used for data analysis. The level of significance was set at p < 0.05, two-tailed. JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006 DOI /jor

4 EFFECT OF AGING ON DISC VOLUME AND SHAPE 1089 Figure 3. Classification of disc degeneration (modified from Pearce et al. 25,26 and Pfirrmann et al. 27 ). RESULTS The data analyses included a total of 350 intervertebral discs from 70 volunteers. The population had a mean height of 171 cm (range, cm) and average body weight of 71.1 kg (range, kg). The mean body mass index was 24.3 kg/m2 (range, kg/m2). There were 175 (50.0%) normal discs, whereas 93 (26.6%) exhibited mild, 67 (19.1%) moderate, and 15 (4.3%) severe disc degeneration, respectively. The descriptive sample statistics are reported in Table 1. Quality Control of Data Assessment The assessment of intra- and interrater reliability of the measurements (coefficients of variation, Pearson correlation, and intraclass correlation coefficient) is detailed in Table 2. All measurements had an error of less than 4%. Disc Height Multilevel regression analysis (Table 3) showed that disc height exhibits a significant ( p < 0.001) positive correlation with the disc level (i.e., disc height in L5/S1 is in general higher than in L1/2). Disc degeneration resulted in an overall decrease of disc height ( p < 0.01). Increasing age was positively correlated ( p < 0.01) with disc height (i.e., the older the subject, the higher the disc). There was no influence of weight, height, and gender on disc height. Based on multilevel regression analysis, the Farfan index exhibited similar results with regard to disc level ( p < 0.001) and disc degeneration ( p < 0.001). Although, the influence DOI /jor JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006

5 1090 PFIRRMANN ET AL. Table 1. Descriptive Sample Statistics a Variable Disc Height (Dabbs) Disc Height (Farfan) Disc Volume Disc Convexity Unit mm Index cm 3 Index Mean 8.90 (9.39/8.36) (49.79/50.97) 7.07 (8.10/5.92) 1.30 (1.28/1.33) Standard error of mean 0.11 (0.15/0.18) 0.60 (0.78/0.94) 0.10 (0.15/0.10) 0.02 (0.03/0.04) Median 8.75 (9.35/8.20) (49.64/50.14) 6.84 (7.94/5.84) 1.23 (1.22/1.26) Standard deviation 2.20 (2.06/2.25) (10.62/12.06) 2.02 (2.01/1.31) 0.40 (0.34/0.47) Minimum 3.20 (3.20/3.60) (21.99/24.49) 2.27 (2.27/3.17) 0.28 (0.57/0.28) Maximum (14.25/16.75) (75.76/ (14.15/9.50) 3.63 (2.55/3.63) Percentile 25th 7.25 (7.90/6.85) (42.82/43.26) 5.61 (6.73/5.11) 1.04 (1.04/1.05) 50th 8.75 (9.35/8.20) (49.64/50.14) 6.84 (7.94/5.84) 1.23 (1.22/1.26) 75th (10.90)/9.80) (56.74/57.97) 8.39 (9.50/6.84) 1.53 (1.52/1.57) a Values for men/women in parentheses. of age was also positively correlated with disc height, it did not reach statistical significance. Disc Volume Multilevel regression analysis (Table 3) showed that disc degeneration is a significant predictor of disc volume ( p < 0.001). Gender, weight, and disc level did not have a significant influence, but there was a strong positive correlation with age ( p < 0.01), and body height ( p < 0.001). This demonstrates that on the single observation level, disc volume is decreasing with increasing disc degeneration, and that differences between participants in disc volume are related to differences in age and height. Disc Convexity Index Multilevel regression analysis (Table 3) showed that disc level ( p < 0.001) and disc degeneration ( p < 0.001) were negatively related to the disc convexity index (i.e., disc were less convex in lower lumber disc levels and in the presence of disc degeneration). DISCUSSION To the best of our knowledge, this is the first study investigating the influence of aging and/or degeneration on disc volume and shape using quantitative in vivo MRI data of an asymptomatic population. Since it is difficult to distinguish a painful degenerated (i.e., pathologic) disc from a disc with physiologic changes of aging, we used a well-defined asymptomatic population to be able to describe the process of normal aging of the lumbar intervertebral discs. While MRI has become the standard noninvasive tool to assess degeneration of the intervertebral disc, debate continues on the specific criteria and potential quantification. Osti and Fraser 31 suggested using a semiquantitative assessment of signal intensity. Frobin et al. 10 used endplate Table 2. Mean Intra- and Interobserver Errors (Coefficients of Variation) and Correspondence of Measures (Pearson Correlation and Intraclass Correlation) Errors Disc Volume Disc Convexity Index Disc Height (Dabbs) Intraobserver Coefficient of variation 3.12 (2.10) 3.82 (2.51) 2.62 (1.76) Mean (standard deviation) Pearson correlation 0.90* 0.96* 0.98* Interobserver Coefficient of variation 3.68 (3.06) 3.97 (2.94) 2.24 (1.68) Mean (standard deviation) Intraclass correlation 0.94* 0.96* 0.98* *p < 0.001, two-tailed. JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006 DOI /jor

6 EFFECT OF AGING ON DISC VOLUME AND SHAPE 1091 Table 3. Prediction of Disc Height, Disc Volume, and Disc Shape in Multilevel Regression Analyses a Disc Height (Dabbs) Disc Height (Farfan) Disc Volume Disc Convexity Predictor variable Parameter Standard error Parameter Standard error Parameter Standard error Parameter Standard error Data level 1 (disc) Disc level (upper vs. lower) *** *** *** Degeneration ** *** *** *** Data level 2 (individual) Gender Age b ** ** Weight b Height b *** Variance Components VAR c (%Expl.) VAR c (%Expl.) VAR c (%Expl.) VAR c (%Expl.) Data level 1 (disc) Data level 2 (individual) IPC VAR, variance on data level; %Expl., percent of explained variance on data level; IPC, intraparticipant correlation, i.e., proportion of the between participants variance on the total variance. IPC measures the extent to which the scores of discs from one participant resemble each other, compared with those discs from different participants. a Parameter: 0 fixed parameter estimates (based on iterative generalized least squares algorithm, IGLS); SE ¼ standard error with significance levels of the Wald-Test (* p < 0.05; ** p < 0.01; *** p < 0.001; two-tailed). Gender (0 ¼ male; 1 ¼ female). Examplar guidance of regression results for disc height (Dabbs): 29% of variation between discs is due to personbound factors that are the same for each disc of an individual (level 2). Both predictor variables that vary on disc-level (level 1) are significant predictors of disc height. First, disc level is a significant (p < 0.001) predictor of disc height (i.e., from L1/2 with each lower level the estimated height increased with 0.63 mm). Higher disc degeneration predicted lower disc height (p < 0.01). With each grade of degeneration, the disc height was estimated to decrease by the amount of 0.43 mm. Among predictor variables that were the same for all discs of an individual (level 2), age was positively correlated (p < 0.01) with disc height (i.e., the older the subject the higher the disc). There was no influence of weight, height, and gender on disc height. b Predictor variables centered by group mean. c As estimated from a variance components model. DOI /jor JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006

7 1092 PFIRRMANN ET AL. abnormalities and presence of disc herniation as additional criteria. Luoma et al. 12 used quantitative signal intensity measurements of the intervertebral disc which raises some methodological concerns. 32,33 Pearce and colleagues 25,26 have suggested a grading system which originated from a comparison of MR images of intervertebral disc with their gross morphologic appearance. In continuing the work by Pearce, 25,26 Pfirrmann et al. developed a comprehensive MRI grading system and tested its reproducibility. 27 However, this grading system includes disc height as a discriminating factor and cannot directly be applied to the current study because disc height is a dependent variable. The advantage of using a multilevel approach (i.e., multiple regression analysis allowing for data analysis with a hierarchical structure) for statistical analysis is demonstrated by our results exploring the effect of disc degeneration and aging on disc volume and shape. It resolves some discrepancies reported in the literature. While the association of disc degeneration and age is well documented in the literature, debate continues on the association of disc degeneration and changes in disc height and shape. Vernon-Roberts and Pirie 2 reported that the disc height decreases with age and that this is responsible for the wellknown loss of body height with age. However, Nachemson and coworkers 3 could not find a correlation between mechanical properties and aging as well as between disc space narrowing and intervertebral disc changes. Berlemann et al. 6 concluded from their cadaveric study that early histologic changes do not significantly influence intervertebral disc height. Twomey and Taylor, 4,5 on the other hand, stated that the true average height of the disc increases with age. They found that vertebral bone density significantly decreases with age with a decline of the horizontal trabeculae. As a result, the vertebra becomes concave because the disc is stiffer than the osteopenic end-plate. Twomey and Taylor conclude that, in old age, major changes occur in the vertebral body and not in the disc. We have found that disc volume predominantly depends on disc degeneration and age, while disc shape seems to depend on disc degeneration but not on age (Table 3). The most striking finding was that disc degeneration is negatively correlated with all morphologic parameters, while age exhibits an inverse (i.e., positive) correlation with disc height and volume. Further analysis demonstrated that degeneration was much stronger associated with decreasing disc heights in older individuals than in younger individuals. Our findings, therefore, explain the conflicting data reported in the literature 3 5 and demonstrate how important it is to simultaneously consider confounding variables in a statistical model. The vast majority of the data in the literature is derived from cadaver studies. 3 6 These studies are potentially hampered by postmortem changes with regard to disc height and volume, respectively. As shown by Johnstone et al., 7 the decrease of hydrostatic pressure after death increases the fluid content due to the swelling pressure which results in an increase in disc height and volume. This finding explains why our absolute values on disc height are somewhat smaller than those reported by Berlemann et al. 6 In a study comparing standard radiography with MR images of the same individual, Frobin et al. 10 found that the first sign of disc degeneration (defined as a moderate loss of nucleus signal) precedes disc height decrease. In more advanced stages, disc degeneration was associated with diminished disc height in their study. However, they highlighted that disc height decrease appears to be only Figure 4. Sagittal T2-weighted MR image of a 75 year old lady with severe disc degeneration and almost resorption of the disc illustrating the relationship of disc degeneration and disc narrowing. JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006 DOI /jor

8 EFFECT OF AGING ON DISC VOLUME AND SHAPE 1093 loosely correlated with disc degeneration. Luoma et al. 12 investigated the correlation of disc height and disc degeneration in individuals with different occupations (i.e., machine operators, construction carpenters, and office workers and students). The authors reported that relative signal intensity was lower in the middle-aged men than in the young men, indicating age-related disc degeneration. Although a general positive association between disc narrowing and decreased relative signal intensity was found, it was concluded that the relationship of disc narrowing, signal intensity, and age is not simple. In this study, we have found that disc height, volume, and shape is negatively correlated with disc degeneration but not independent of age, which exhibits a positive correlation. The reason for the inverse dependency on age becomes more plausible in the older age group. While disc degeneration generally results in decreased disc height, volume, and disc convexity index (Fig. 4), the disc can buckle into the adjacent vertebra in the absence of severe disc degeneration (Fig. 5). A similar observation was reported by Roberts et al. 37 They have found that the central height of L1 to L4 nondegenerated discs is higher in older than in younger individuals. Dai 38 used a finite element model to explore this observation in more detail. He concluded that when a disc is of normal height or without degeneration, the adjacent vertebral body is more likely to be deformed for patients with osteoporosis. However, the confounding effect of osteoporosis has not been explored in this study because the presented results were somewhat unexpected. We have excluded patients with known osteoporosis or osteoporotic fractures. The study setup, therefore, did not foresee the inclusion of a bone density measurement. In retrospect, this is a drawback because our results now may indicate such an effect of osteoporosis, and it is not possible to retrospectively add a bone density assessment. However, this effect deserves further attention and investigation when explaining the positive relationship of age on disc height and volume. We conclude from our study that disc degeneration generally results in a decrease of disc height and volume, as well as a less convex disc shape. In the absence of disc degeneration, however, age tends to result in an inverse relationship on disc height, volume, and shape. ACKNOWLEDGMENTS No benefits of any form have been received or will be received by the authors from a commercial party related directly or indirectly to the subject of this article. REFERENCES Figure 5. Sagittal T2-weighted MR image of a 77- year-old lady with mild disc degeneration and substantial buckling of the intervertebral discs into the adjacent vertebral body. 1. Buckwalter JA Spine update: aging and degeneration of the human intervertebral disc. Spine 20: Vernon-Roberts B, Pirie CJ Degenerative changes in the intervertebral discs of the lumbar spine and their sequelae. Rheumatol Rehab 16: Nachemson A, Schultz A, Berkson M Mechanical properties of human lumbar spine motion segments. Influences of age, sex, disc level and degeneration. Spine 4: Twomey L, Taylor J Age changes in lumbar intervertebral discs. Acta Orthop Scand 56: Twomey LT, Taylor JR Age changes in lumbar vertebrae and intervertebral discs. Clin Orthop 224: Berlemann U, Gries NC, Moore RJ The relationship between height, shape and histological changes in early degeneration of the lower lumbar discs. Eur Spine J 7: Johnstone B, Urban JP, Roberts S, et al The fluid content of the human intervertebral disc. Comparison DOI /jor JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006

9 1094 PFIRRMANN ET AL. between fluid content and swelling pressure profiles of discs removed at surgery and those taken postmortem. Spine 17: Brinckmann P, Frobin W, Biggemann M, et al Quantification of overload injuries to thoracolumbar vertebrae and discs in persons exposed to heavy physical exertions or vibration at the work-place. Part I: the shape of vertebrae and intervertebral discs study of a young, healthy population and a middle-aged control group. Clin Biomech 9(Suppl): Brinckmann P, Frobin W, Biggemann M, et al Quantification of overload injuries to thoracolumbar vertebrae and discs in persons exposed to heavy physical exertions or vibration at the workplace. Part II: occurrence and magnitude of overload injury in exposed cohorts. Clin Biomech 13:S1 S Frobin W, Brinckmann P, Kramer M, et al Height of lumbar discs measured from radiographs compared with degeneration and height classified from MR images. Eur Radiol 11: Goh S, Tan C, Price RI, et al Influence of age and gender on thoracic vertebral body shape and disc degeneration: an MR investigation of 169 cases. J Anat 197: Luoma K, Vehmas T, Riihimaki H, et al Disc height and signal intensity of the nucleus pulposus on magnetic resonance imaging as indicators of lumbar disc degeneration. Spine 26: Fraser RD, Osti OL, Vernon-Roberts B Intervertebral disc degeneration. Eur Spine J 2: Boos N, Wallin A, Boesch C, et al A new MR image analysis method for the measurement of disc height variations. Spine 21: Botsford DJ, Esses SI, Ogilvie-Harris DJ In vivo diurnal variation in intervertebral disc volume and morphology. Spine 19: Dabbs VM, Dabbs LG Correlation between disc height narrowing and low-back pain. Spine 15: Farfan HF, Cossette JW, Robertson GH The effects of torsion on the lumbar intervertebral joints: the role of torsion in the production of disc degeneration. J Bone Joint Surg 52A: Roberts N, Cruz-Orive LM, Reid NM, et al Unbiased estimation of human body composition by the Cavalieri method using magnetic resonance imaging. J Microsc 171: Cotter D, Miszkiel K, Al-Sarraj S, et al The assessment of postmortem brain volume; a comparison of stereological and planimetric methodologies. Neuroradiology 41: Diab KM, Ollmar S, Sevastik JA, et al Volumetric determination of normal and scoliotic vertebral bodies. Eur Spine J 7: Gundersen HJ, Jensen EB The efficiency of systematic sampling in stereology and its prediction. J Microsc 147: Lund H, Christensen L, Savnik A, et al Volume estimation of extensor muscles of the lower leg based on MR imaging. Eur Radiol 12: McNulty V, Cruz-Orive LM, Roberts N, et al Estimation of brain compartment volume from MR Cavalieri slices. J Comput Assist Tomogr 24: Roberts N, Puddephat MJ, McNulty V The benefit of stereology for quantitative radiology. Br J Radiol 73: Eyre D, Benya P, Buckwalter J, et al Intervertebral disk: basic science perspectives. In: Frymoyer JW, Gordon SL, editors. New perspectives on low back pain. Park Ridge, IL: American Academy of Orthopaedic Surgeons; pp Pearce RH, Thompson JP, Bebault GM Magnetic resonance imaging reflects the chemical changes of aging degeneration in the human intervertebral disk. J Rheumatol Suppl 27: Pfirrmann CWA, Metzdorf A, Zanetti M MR classification of lumbar intervertebral disc degeneration. Spine 26: Hox JJ Applied multilevel analysis. Amsterdam: TT Publikaties. 29. Langford IH, Bentham G, McDonald AL Multi-level modelling of geographically aggregated health data: a case study on malignant melanoma mortality and UV exposure in the European Community. Stat Med 17: Rasbash J, Browne W, Goldstein H, et al MLMMP, Institute of Education, University of London: a user s guide to MLwiN. London: Multilevel Models Project, Institute of Education, University of London. 31. Osti OL, Fraser RD MRI and discography of annular tears and intervertebral disc degeneration. A prospective clinical comparison. J Bone Joint Surg [Br] 74: Boos N, Boesch C Imaging corner: quantitative magnetic resonance imaging of the lumbar spine. Potential for investigations of water content and biochemical composition. Spine 20: Modic MT, Herfkens RJ Devil s advocate. Intervertebral disk: normal age-related changes in MR signal intensity. Radiology 177: Miller JAA, Schmatz C, Schultz AB Lumbar disc degeneration: correlation with age, sex, and spine level in 600 autopsy specimens. Spine 13: Powell MC, Wilson M, Szypryt P, et al Prevalence of lumbar disc degeneration observed by magnetic resonance in symptomless women. Lancet 2: Tertti M, Paajanen H, Laato M, et al Disc degeneration in magnetic resonance imaging. A comparative biochemical, histologic, and radiologic study in cadaver spines. Spine 16: Roberts N, Gratin C, Whitehouse GH MRI analysis of lumbar intervertebral disc height in young and older populations. J Magn Reson Imaging 7: Dai L The relationship between vertebral body deformity and disc degeneration in lumbar spine of the senile. Eur Spine J 7: JOURNAL OF ORTHOPAEDIC RESEARCH MAY 2006 DOI /jor