Musculoskeletal Imaging. Christian W. A. Pfirrmann, MD Donald Resnick, MD

Musculoskeletal Imaging Christian W. A. Pfirrmann, MD Donald Resnick, MD Index terms: Schmorl nodes, 32.78, 33.78 Spine, diseases, 32.4961, 32.4963, 32.77, 32.78, 33.4961, 33.4963, 33.77, 33.78 Spine, intervertebral disks, 32.77, 32.78, 33.77, 33.78 Spine, radiography, 32.11, 33.11 Radiology 2001; 219:368 374 Abbreviations: DVJ discovertebral junction VEP vertebral endplate Schmorl Nodes of the Thoracic and Lumbar Spine: Radiographic-Pathologic Study of Prevalence, Characterization, and Correlation with Degenerative s of 1,650 Spinal Levels in 100 Cadavers 1 1 From the Department of Radiology, Veterans Affairs San Diego Healthcare System, 3350 La Jolla Village Dr, San Diego, CA 92161. From the 2000 RSNA scientific assembly. Received June 21, 2000; revision requested August 18; revision received September 29; accepted October 11. Supported by Veterans Affairs grant SA-360 and the Swiss National Science Foundation. Address correspondence to D.R. (e-mail: dresnick@ucsd.edu). RSNA, 2001 PURPOSE: To investigate the frequency and characteristics of Schmorl nodes in an elderly population and to correlate these findings with degenerative spinal changes. MATERIALS AND METHODS: Cadaveric thoracic and lumbar spines were removed at autopsy (mean age at death, 68.2 years; range, 43 93 years). Parasagittal sections of approximately 5-mm thickness were obtained and radiographed. At each of 3,300 endplates from T1 to L5, the presence of Schmorl nodes was noted. Vertebral endplate contour was analyzed, and abnormalities of the discovertebral junction were noted. The height of each interspace was measured, and the presence or absence of vacuum phenomena and spondylosis was recorded. RESULTS: Schmorl nodes were found in 58 specimens and were multiple in 41. Of 3,300 vertebral endplates, 225 revealed Schmorl nodes: 88 cranial and 137 caudal. More than 182 were between T7 and L2. Schmorl nodes correlated with disk space loss (P.001) but not with evidence of advanced disk degeneration: marked disk space loss (P.53), vacuum phenomena (P.82), or discogenic sclerosis or erosion (P.35). Schmorl nodes were associated with claw (P.001) but not traction (P.72) osteophytes. Straight (P.001) and fractured (P.001) vertebral endplates were associated with Schmorl nodes. CONCLUSION: Schmorl nodes are common in the spines in an elderly population, with a frequency similar to that in a younger population. Schmorl nodes are associated with moderate but not advanced degenerative changes. Geometric observations regarding the vertebral endplates support the concept that Schmorl nodes are caused by an abnormality of the discovertebral junction. Author contributions: Guarantors of integrity of entire study, C.W.A.P., D.R.; study concepts and design, C.W.A.P., D.R.; literature research, C.W.A.P.; clinical studies, C.W.A.P.; data acquisition and analysis/interpretation, C.W.A.P.; statistical analysis, C.W.A.P.; manuscript preparation, C.W.A.P.; manuscript definition of intellectual content, C.W.A.P., D.R.; manuscript editing, revision/review, and final version approval, D.R. Schmorl nodes represent displacements of intervertebral disk tissue into the vertebral body (1). Both Schmorl nodes and degenerative disk disease are common in the human spine. Schmorl nodes are a common but not obligate manifestation of Scheuermann disease, which is known to enhance premature disk degeneration (2 4). Furthermore, besides Scheuermann disease, various causes of Schmorl node formation have been emphasized (5). To our knowledge, the relationship of Schmorl nodes and degenerative changes of the thoracic and lumbar spine has never been emphasized in the literature, despite the large number of investigations devoted to degenerative spinal diseases and their major socioeconomic effect. Furthermore, there is a high variability in the reported prevalence of 368

TABLE 1 Association of Schmorl Nodes with Degenerative s of the Spine Schmorl Nodes* Degenerative s Node Absent Node Present Total P Value Narrowing of interspace 2,390/649 (21.4) 115/110 (48.9) 2,505/759 (23.3).001 Intervertebral collapse 2,977/62 (2.0) 219/6 (2.7) 3,296/68 (2.0).53 Vacuum phenomena 2,878/174 (5.7) 213/12 (5.3) 3,091/186 (5.7).82 DVJ changes 3,009/55 (1.8) 219/6 (2.7) 3,228/61 (1.9).35 Osteophytes 1,350/1,719 (56.0) 76/149 (66.2) 1,426/1,868 (56.7).003 Osteophytes 3 mm 2,821/434 (13.3) 186/39 (17.3) 2,821/473 (14.4).19 Claw osteophytes 2,650/421 (13.7) 169/56 (24.9) 2,819/477 (14.5).001 Traction osteophytes 1,921/1,150 (37.4) 138/87 (38.7) 2,056/1,237 (37.6).72 Diffuse idiopathic 2,886/185 (6.0) 219/6 (2.7) 3,105/191 (5.8).04 skeletal hyperostosis * Data are the number of VEPs without degenerative change/number of VEPs with nodes. Data in parentheses are the percentage of VEPs with the change. Difference is significant. Reduction of the intervertebral height of more than 75%. DVJ changes include discogenic sclerosis and erosive changes. Schmorl nodes (1,6 8); the prevalence in the spines of elderly persons is reported to be far lower than that of young persons (6). The ability to detect cartilaginous nodes on conventional routine radiographs, however, depends on several factors, including the size of the patient, the quality of the radiographs, the location and size of the cartilaginous nodes, and the degree of surrounding bone sclerosis. The purpose of this investigation was to determine the frequency and characteristics of Schmorl nodes in an elderly population and any relationship of such nodes to degenerative changes of the spine. MATERIALS AND METHODS Cadavers This study group was derived from a review of radiographs of 128 cadaveric spines; specimens with obvious bone destruction related to malignant tumors and specimens in which the spine was incompletely removed were excluded. The remaining 100 cadavers, which were used for this investigation, were patients whose mean age at the time of death was 68.2 years (range, 43 93 years). The anterior aspect of the thoracic and lumbar portions of the vertebral column was removed in one piece by using a saw. These specimens contained the vertebral bodies and intervertebral disks from the top of T1 to the bottom of L5 and, in most cases, the superoanterior aspect of the sacrum. The specimens were deep-frozen at 40 C (Bio-Freezer; Forma Scientific, Marietta, Ohio) and subsequently cut into parasagittal slabs of approximately 5-mm thickness with a band saw. Lowkilovoltage contact radiographs of each section were obtained with a radiographic unit (Faxitron series; Hewlett Packard, Mc- Minnville, Ore) with the use of fine-grain film (Kodak, Rochester, NY). X-ray beam factors included 45 kvp and a focus-tofilm distance of 20.3 cm. Exposure was adjusted with the use of an automatic exposure control. Photographs of selected slabs were obtained to demonstrate gross anatomy. Analysis of Contact Slab Radiographs Figure 1. Definition of the forms of the VEP. Diagrams show the morphologic criteria used to define VEP variations. The lower part of the vertebral body, as seen in a sagittal plane, is demonstrated. The left side of the diagrams represents the anterior part of the vertebral body. a, A Schmorl node (arrow) is defined as a focal indentation of the VEP. b, The normal concave form of the VEP is shown. c, The cupid s bow contour (arrowheads) has a smooth concavity in the posterior portion of the VEP. d, A straight VEP is present when a line drawn from the anterior edge to the posterior edge of the vertebral body is in contact with the central portion of the VEP. e, A fractured VEP (arrow) is shown. In each spine, each of the vertebral endplates (VEPs) from the inferior aspect of T1 to the inferior aspect of L5 was analyzed by one experienced radiologist (C.W.A.P., 3 years of experience in the musculoskeletal subspecialty). All contact-slab radiographs of one whole spine were made available at analysis, and each level of each spine was evaluated consecutively. The presence of Schmorl nodes was recorded. A Schmorl node was defined as a focal indentation of the VEP (Fig 1a). The location of the Schmorl node (anterior, middle, or posterior third of the VEP) was recorded, and the size (maximum height and maximum sagittal diameter of the defect in the VEP) was measured. The volume of the Schmorl node was calculated according to the following equation for the volume of the half of a sphere: Volume ( 1 2 maximum sagittal diameter) 2 maximum height (4 /6). Abnormalities of the discovertebral junction (DVJ), defined as discogenic sclerosis and erosive changes (irregular appearance of the endplate with thinning or focal loss of visualization of the subchondral cortical plate), were also recorded as present or absent. The height of each interspace was measured at its largest distance on the midsagittal section. In the presence of a Schmorl node, the height of the interspace was measured to the intact endplate. The presence of vacuum phenomena in the intervertebral disk and vertebral osteophytes at each level was recorded. Spondylosis was noted at every level. Large osteophytes were defined as outgrowths with an anteroposterior diameter was greater than 3 mm. Claw-type osteophytes were defined as excrescences arising from the DVJ that were triangular and curved at their tip; traction osteophytes were defined as linear osseous Volume 219 Number 2 Spinal Schmorl Nodes: Radiographic-Pathologic Study 369

Figure 4. (a) Sagittal slab contact radiograph at the L3-4 level in a 74-year-old man shows a Schmorl node (black arrow) in the distal VEP of L3 and shows a vacuum phenomenon in the intervertebral disk (white arrow). (b) Gross specimen of the same slab shows the Schmorl node, with displacement of the intervertebral disk (white arrow) in the VEP of L3. Cleft formation (black arrow) in the intervertebral disk corresponds to the site of the vacuum phenomenon. Figure 2. Graph shows the distribution of Schmorl nodes for each vertebral level from T1 to L5. Numbers on the x axis are the number of nodes. Figure 5. Sagittal slab contact radiograph of the L2-3 interspace in a 72-year-old man shows a Schmorl node (black arrow) in the lower endplate of L2 with traction osteophyte formation (white arrow). Figure 3. Sagittal slab contact radiograph of the T11 to L1 levels of the spine in a 61-yearold man shows Schmorl nodes (straight arrows) at the T11-12 level, with a straight configuration of the VEP and moderate disk space loss. A cupid s bow contour (curved arrows) is at the T12-L1 level. 370 Radiology May 2001 plates extending in a horizontal direction throughout their length (9). Diffuse idiopathic skeletal hyperostosis was recorded according to the following criteria: (a) the presence of flowing calcification and ossification along the anterolateral aspect of at least four contiguous vertebral bodies, with or without associated localized pointed excrescences of the vertebral body, and relative preservation of intervertebral disk height in the involved vertebral segment and (b) the absence of extensive radiographic changes of degenerative disk disease, including vacuum phenomena and vertebral body marginal sclerosis (10). Distinct variations of the sagittal contour of the VEP were recorded (Fig 1). A concave contour was considered to be the normal appearance of the VEP (Fig 1b). The cupid s bow contour was defined as a smooth concave curvature with its center located at the posterior portion of the VEP and with a steeper slope posteriorly than anteriorly (Fig 1c) (11). A straight endplate was recorded when the central portion of the VEP contacted a line drawn from the anterior to the posterior edge of the vertebral body on the midsagittal slab (Fig 1d). A fractured endplate was defined as an impression of the subchondral bone (Fig 1e), with disruption of the VEP and an abnormal angulation of at least 50% of the anteroposterior diameter of the vertebral body that was visible on more than two sections. Statistics The frequency of Schmorl nodes and degenerative changes, as well as the geometric findings of the DVJ, were compared by using a 2 test. The size and volume of the Schmorl node were related to the presence of degenerative changes by using the Mann-Whitney U test. The significance level was P less than.01. RESULTS Schmorl nodes were found in 58 (58%) of 100 specimens and were multiple in 41 specimens (mean, 3.9 nodes; range, 1 13 nodes). In total, 225 Schmorl nodes were found; 88 nodes were located in the cranial endplates and 137 in the caudal endplates. Schmorl nodes were evident in both endplates about a disk in 102 sites. The locations of the Schmorl nodes are shown in Figure 2. Of all 182 Schmorl nodes, 147 (81%) were found between T7 and L2 (Fig 2). Twelve (5%) of 225 Pfirrmann and Resnick

TABLE 2 Relationship of Diameter and Volume of Schmorl Nodes with Degenerative s of the Spine Figure 6. Sagittal slab contact radiograph of the L2-3 interspace in a 59-year-old man shows a Schmorl node (black arrow) in the lower endplate of L2 and claw osteophyte formation (white arrows). Degenerative s Absent* Diameter Present* P Value Absent Volume Present P Value Narrowing of interspace 6.1 2.6 5.9 2.9.312 85.6 117.0 86.5 131.4.404 Intervertebral collapse 6.0 2.7 7.5 3.8.360 82.6 111.5 211.1 356.5.741 Vacuum phenomena 6.0 2.8 5.4 1.6.884 87.0 126.7 69.2 55.0.976 DVJ changes 5.9 2.6 8.0 5.5.453 89.8 103.4 315.9 396.1.269 Osteophytes 5.9 2.6 6.0 2.8.928 80.5 109.7 88.8 130.9.896 Osteophytes 3 mm 6.0 2.3 5.8 2.3.952 88.9 132.7 72.4 67.5.780 Claw osteophytes 6.1 2.7 5.8 2.7.366 88.9 125.6 7.76 119.5.334 Traction osteophytes 5.8 2.6 6.3 2.8.298 77.1 111.5 100.2 141.0.197 * Data are the mean (in millimeters) plus or minus the SD. Data are the mean (in cubic millimeters) plus or minus the SD. Reduction of the intervertebral height of more than 75%. DVJ changes include discogenic sclerosis and erosive changes. TABLE 3 Association of Schmorl Nodes and Geometry of the VEP Geometric Characteristics Node Absent Schmorl Nodes* Node Present Total P Value Concave endplate 443/2,620 (85.5) 141/84 (37.3) 584/2,704 (82.2).001 Cupid s bow 2,929/218 (6.9) 134/7 (5.0) 3,147/141 (4.3).34 Straight endplate 2,783/280 (9.1) 100/125 (55.6) 2,883/405 (12.3).001 Fractured endplate 3,034/29 (0.9) 216/9 (4.2) 3,250/38 (1.2).001 * Data are the number of VEPs without the geometric characteristic/number of VEPs with the characteristic. Data in parentheses are the percentage of VEPs with the characteristic. Difference is significant. Negative association. TABLE 4 Reported Frequency of Schmorl Nodes Figure 7. Graph shows the distribution of Schmorl nodes and cupid s bow contour from T1 to L5. Black bars represent cupid s bows contours. Striped bars represent Schmorl nodes. Numbers on the x axis are the number of nodes. Schmorl nodes were located in the anterior third, 81 (36%) in the middle third, and 132 (59%) in the posterior third of the VEP. The mean diameter of the Schmorl nodes was 6 mm (range, 2 15 mm), and the mean height was 3.3 mm (range, 1 9 mm). The mean volume of the Schmorl nodes was 86 mm 3 (range, 2 923 mm 3 ). The correlation of the frequency of Schmorl nodes and degenerative changes of the spine is summarized in Table 1. Schmorl nodes correlated with the presence of disk space loss (P.001) (Fig 3). Investigation* Frequency (%) Method Present study 58 68.2 100 Slab contact radiograph Schmorl and Junghans, 38 NA NA Inspection of whole 1971 (1) specimens However, there was no correlation of such nodes with signs of advanced disk degeneration, that is, intervertebral collapse (P.53), vacuum phenomena (P.82) (Fig 4), and DVJ changes (P.35). Schmorl nodes and spondylosis were significantly associated (P.003). There Mean Patient Age (y) No. of Specimens or Patients Hilton et al, 1976 (12) 76 13 96 50 Slab radiographs Hansson et al, 1983 (8) 66 58 36 Inspection of slices of specimens Frymoyer et al, 1984 (13) 0.3 3.4 18 55 292 Lumbar radiographs Hamanishi et al, 1994 (6) Symptomatic patients 19 1 82 400 Magnetic resonance (MR) imaging Asymptomatic patients 9 1 82 100 MR imaging Stäbler et al, 1997 (7) 38 53 372 MR imaging Note. NA not available. * Numbers in parentheses are reference numbers. was no association of Schmorl nodes with osteophytes larger than 3 mm in an anteroposterior diameter (P.19). There also was no correlation of Schmorl nodes with traction osteophytes (P.72) (Fig 5), but such nodes correlated with claw osteophytes (P.001) (Fig 6). The pres- Volume 219 Number 2 Spinal Schmorl Nodes: Radiographic-Pathologic Study 371

nodes to a cupid s bow contour of the VEP in four cases (Fig 8). In three cases, Schmorl nodes in multiple consecutive levels were located in one straight vertical line. DISCUSSION Figure 8. (a) Anteroposterior and (b) lateral specimen radiographs of the T11-L5 segment in a 60-year-old man show the transition of Schmorl nodes (white arrows) in the VEPs of T11 to L2 to a cupid s bow contour (black arrows) of the VEP of L3 and L4. ence of diffuse idiopathic skeletal hyperostosis did not significantly correlate with the presence of Schmorl nodes (P.04). The correlation of the size of the Schmorl nodes and degenerative findings is displayed in Table 2. The size and volume of the Schmorl node did not correlate with narrowing of the intervertebral disks, vacuum phenomena, DVJ changes, or the presence of spondylosis. However, at levels containing both Schmorl nodes and DVJ changes or collapse of the intervertebral disk, nodes were considerably larger versus those at levels without changes 372 Radiology May 2001 or collapse (DVJ changes, 315.9 vs 89.8 mm3; collapse, 211.1 vs 82.6 mm3). The frequency of geometric shapes of the VEP is shown in Table 3. Straight (P.001) (Fig 3) and fractured (P.001) VEPs were associated with Schmorl nodes. No correlation of Schmorl nodes with a cupid s bow contour was noted (P.34). Most of the Schmorl nodes were concentrated about the T11 to L1 levels. There was a similar concentration of the cupid s bow contour more distally about the L2 to L4 levels (Fig 7). At analysis of continuous levels, a transition was seen from posteriorly located Schmorl Fifty-eight (58%) of 100 specimens revealed Schmorl nodes. This frequency is consistent with the upper spectrum of those previously reported (Table 4). These differences in prevalence are most likely related to the varying methods of investigation that were used. Conventional radiography depicts fewer Schmorl nodes than cross-sectional imaging methods or examinations based on inspection of specimens (14). Recently developed Schmorl nodes may not be seen on conventional radiographs due to the absence of surrounding sclerosis (15). Slab contact radiographs obtained with the use of finegrain film allow detailed assessment of the presence of Schmorl nodes, and, when nodes are present, they allow measurement of the size of Schmorl nodes and the height of the intervertebral disk space without any magnification. Subtle bone changes accompanying those nodes also can be depicted. Almost all Schmorl nodes were found in the lower thoracic spine or the thoracolumbar junction (Fig 2). This is in agreement with results in most of the previously reported investigations (1, 6 8). In our study, almost two-thirds of Schmorl nodes were located in the posterior part of the VEP, and one-third were in the middle part; anterior nodes were rare. To our knowledge, no similar data regarding the localization of Schmorl nodes can be found in the literature. However, traumatic Schmorl nodes occur predominantly in the posterior VEP and in the lower thoracic spine and thoracolumbar junction, sites that are particularly susceptible to injury (15 17). This distribution supports the finding that trauma is one of the causative factors for the development of Schmorl nodes. We found the frequency of Schmorl nodes to be 58% in our elderly population (mean age, 68.2 years). This is in contrast to the results of Hamanishi et al (6), who, in studying Schmorl nodes, reported frequencies of 57% in the 2nd decade of life and 5% in the 6th decade of life by using MR imaging as the vehicle of investigation. This discrepancy is probably due to the observation that Schmorl nodes tend to be smaller and have more surrounding sclerosis in the spines of elpfirrmann and Resnick

derly persons, a finding perhaps related to healing (1), and are less likely to have reactive concomitant bone marrow changes that facilitate their detection with MR imaging (18). In our series, the Schmorl nodes had a mean anteroposterior diameter of 6 mm (range, 2 15 mm) and a mean height of 3.3 mm (range, 1 9 mm). There was no correlation between the size and volume of the Schmorl nodes and the degree of disk degeneration. Stäbler et al (7) found a mean diameter of 8.2 mm (range, 4 20 mm) by using MR imaging. They found that patients with back pain tended to have larger Schmorl nodes that were larger than those of asymptomatic patients. In the past, many theories regarding the cause of Schmorl nodes have been emphasized. Schmorl nodes occur when the cartilaginous endplate of the vertebral body has been disrupted (14). Such disruption can be produced by an intrinsic abnormality of the plate itself or by alterations in the subchondral bone of the vertebral body (5). Weak areas in the cartilaginous plate include indentation sites left during the regression of the chorda dorsalis (notochord), ossification gaps, and previous vascular channels (19,20). The subchondral bone may be weakened by numerous local and systemic processes, such as osteomalacia, Paget disease, hyperparathyroidism, infection, neoplasm, trauma, and Scheuermann disease (1,5,21). Osteoporosis has been emphasized as a cause of Schmorl nodes, but this correlation has not been proved (8,22). Whatever the cause of the damage to the cartilaginous endplate, to the subchondral bone of the vertebral body, or to both structures, a weakened area is created that is thought to be unable to resist the expansive pressure of the adjacent nucleus pulposus (1). Our data indicate that degenerative changes accompanying Schmorl nodes are moderate in extent. No evidence of advanced degenerative disease of the disk or DVJ, such as intervertebral collapse, discogenic sclerosis or erosion, and vacuum phenomena, were associated with Schmorl nodes. This indicates that Schmorl nodes are probably not an important factor in the development of degenerative spinal disease. In our study, the association of Schmorl nodes with a straight VEP contour was significant, in contrast to a negative association with normal concave endplates. The importance of this observation is not clear. Schmorl nodes appear to occur as a reaction of the VEP to vertical loading. Compared with the normal concave VEP, a straight VEP seems to be more susceptible to the formation of Schmorl nodes owing to the expansive pressure of the nucleus pulposus (23). This is probably explained with pressure per surface ratio, which is lower in a concave VEP because its surface is larger than that of a straight VEP. Another possible explanation is that there is more space for the nucleus pulposus and, therefore, there is less pressure in intervertebral disks with concave VEPs because the volume of an intervertebral space with concave VEPs is larger than that associated with a straight VEP. Controversy exists regarding the clinical importance of Schmorl nodes. Most consider them to be asymptomatic (13), since Schmorl nodes are a frequent finding in persons without back pain (24). However, Hamanishi and co-workers (6) compared the findings of MR examinations of the lumbar spine in 400 patients with low back pain with those of a control group of 106 patients and found a significantly higher frequency of Schmorl nodes in the symptomatic group (19%) in comparison with the control group (9%). Schmorl nodes that show enhanced signal intensity after intravenous administration of a gadolinium-based contrast agent and those accompanied by bone marrow changes are found more frequently in patients with back pain than in asymptomatic patients (7). In an autopsy study (15) of the spines in 70 patients who had died in motor vehicle accidents, 10% had acute Schmorl nodes. Acute or chronic trauma due to excessive axial loading may cause Schmorl nodes that initially are symptomatic (15). There are some weaknesses in our study. Our analysis was based on observations made in cadavers of an elderly population at a single time (ie, the time of patient death), without serial radiographic examination; we are unable to establish whether Schmorl nodes or degenerative disk changes and geometric variations of the VEP occurred first. Medical records were not available, and the limited information that was available prevents us from commenting on the clinical importance of the radiographic observations. Because our specimens did not contain the posterior elements of the spine, these were not included in the analysis. Only one observer analyzed the contiguous sagittal specimen radiographs. Since he was not blinded to the purpose of our investigation, some bias may have been introduced. Gross pathologic or histologic observations were not used in our analysis. There is also some strength in our study design, however. Slab radiographs were used to analyze the frequency and characteristics of Schmorl nodes; observations based on slab radiographs are more accurate than those based on conventional radiographs. Furthermore, to our knowledge, the correlation of the frequency of such nodes with degenerative changes of the intervertebral disk and with various contours of the VEP has not been accomplished to date. In conclusion, Schmorl nodes are a common finding in the spines of the elderly, with a frequency similar to that reported for a younger population. They are associated with moderate degenerative changes of the spine. Straight or fractured VEPs appear more likely to be associated with Schmorl nodes, when compared with VEPs that have a normal concave configuration. The absence of advanced degenerative disk disease in association with Schmorl nodes and the geometric observations regarding the VEP support the concept that Schmorl nodes are caused by an abnormality of the DVJ rather than by discogenic factors. References 1. Schmorl G, Junghans H. The human spine in health and disease. New York, NY: Grune & Stratton, 1971; 158 172. 2. Paajanen H, Alanen A, Erkintalo M, Salminen JJ, Katevuo K. Disc degeneration in Scheuermann disease. Skeletal Radiol 1989; 18:523 526. 3. Stoddard A, Osborn JF. Scheuermann s disease or spinal osteochondrosis: its frequency and relationship with spondylosis. J Bone Joint Surg Br 1979; 61:56 58. 4. Cleveland RH, Delong GR. The relationship of juvenile lumbar disc disease and Scheuermann s disease. Pediatr Radiol 1981; 10:161 164. 5. Resnick D, Niwayama G. Intravertebral disk herniations: cartilaginous (Schmorl s) nodes. Radiology 1978; 126:57 65. 6. Hamanishi C, Kawabata T, Yosii T, Tanaka S. Schmorl s nodes on magnetic resonance imaging: their incidence and clinical relevance. Spine 1994; 19:450 453. 7. Stabler A, Bellan M, Weiss M, Gartner C, Brossmann J, Reiser MF. MR imaging of enhancing intraosseous disk herniation (Schmorl s nodes). AJR Am J Roentgenol 1997; 168:933 938. 8. Hansson T, Roos B. The amount of bone mineral and Schmorl s nodes in lumbar vertebrae. Spine 1983; 8:266 271. 9. Macnab I. The traction spur: an indicator of segmental instability. J Bone Joint Surg Am 1971; 53:663 670. 10. Resnick D, Niwayama G. Radiographic and pathologic features of spinal involvement in diffuse idiopathic skeletal hyperostosis (DISH). Radiology 1976; 119:559 568. 11. Chan KK, Sartoris DJ, Haghighi P, et al. Cupid s bow contour of the vertebral body: evaluation of pathogenesis with bone den- Volume 219 Number 2 Spinal Schmorl Nodes: Radiographic-Pathologic Study 373

sitometry and imaging-histopathologic correlation. Radiology 1997; 202:253 256. 12. Hilton RC, Ball J, Benn RT. Vertebral endplate lesions (Schmorl s nodes) in the dorsolumbar spine. Ann Rheum Dis 1976; 35:127 132. 13. Frymoyer JW, Newberg A, Pope MH, Wilder DG, Clements J, MacPherson B. Spine radiographs in patients with lowback pain: an epidemiological study in men. J Bone Joint Surg Am 1984; 66: 1048 1055. 14. Schmorl G. Über knorpelknötchen an den wirbelbandscheiben. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 1928; 38:265 279. 15. Fahey V, Opeskin K, Silberstein M, Anderson R, Briggs C. The pathogenesis of Schmorl s nodes in relation to acute trauma: an autopsy study. Spine 1998; 23: 2272 2275. 16. Denis F. The three column spine and its significance in the classification of acute thoracolumbar spinal injuries. Spine 1983; 8:817 831. 17. Petersilge CA, Pathria MN, Emery SE, Masaryk TJ. Thoracolumbar burst fractures: evaluation with MR imaging. Radiology 1995; 194:49 54. 18. Seymour R, Williams LA, Rees JI, Lyons K, Lloyd DC. Magnetic resonance imaging of acute intraosseous disc herniation. Clin Radiol 1998; 53:363 368. 19. Coventry MB, Ghormley RK, Kernohan JW. The intervertebral disc: its microscopic anatomy and pathology. III. Pathological changes in the intervertebral disc. J Bone Joint Surg 1945; 27:460 474. 20. Hassler O. The human intervertebral disc: a micro-angiographical study on its vascular supply at various ages. Acta Orthop Scand 1969; 40:765 772. 21. Resnick D, Niwayama G. Subchondral resorption of bone in renal osteodystrophy. Radiology 1976; 118:315 321. 22. Boukhris R, Becker KL. Schmorl s nodes and osteoporosis. Clin Orthop 1974; 0:275 280. 23. Hansson TH, Keller TS, Spengler DM. Mechanical behavior of the human lumbar spine. II. Fatigue strength during dynamic compressive loading. J Orthop Res 1987; 5:479 487. 24. Jensen MC, Brant-Zawadzki MN, Obuchowski N, Modic MT, Malkasian D, Ross JS. Magnetic resonance imaging of the lumbar spine in people without back pain. N Engl J Med 1994; 331:69 73. 374 Radiology May 2001 Pfirrmann and Resnick