Despite decades of interdisciplinary research, ranging from epidemiology to molecular biology1,2, much remains unknown about the etiology and pathogenesis of disc degeneration. Traditionally, research into the pathogenesis of disc degeneration has focused primarily on the disc itself. The vertebral end plates, thin structures located at the cranial and caudal ends of the intervertebral disc, have long been neglected3 despite their direct interaction with the intervertebral discs and their possible involvement in the pathogenic pathway of disc degeneration.
The end plate, consisting of osseous and cartilaginous components, is essential to the health and function of the intervertebral disc. The end plate serves as a physical shield to prevent the nucleus pulposus from penetrating into the adjacent vertebral body4, while also acting as the gateway for nutrient transport between the vertebral marrow and the intervertebral disc5,6. In addition, it serves as a mechanical interface between stiff bone and resilient disc and it contributes to an even distribution of physical load in the vertebra-disc complex7.
Despite its importance, surprisingly little is known about the vertebral end plate. Literature on the morphometrics (measurements of size and shape) of the vertebral end plate is scarce. The shape, size, and concavity of the vertebral end plate have been measured in vivo, and the results suggest a role of end plate morphometrics in disc degeneration8,9 and disc herniation10. However, such studies sampled only one or several sections of the vertebral body with routine computed tomography (CT) or magnetic resonance imaging (MRI). As the morphology of the vertebral end plate varies considerably between subregions11, such measures are likely inadequate to fully characterize the end plate and its relation to disc pathology.
Most previous research on the end plate has actually focused on Schmorl nodes, the most common form of end plate lesion12. Schmorl nodes were often studied in the thoracic region, where they are more prevalent, and were less studied in the lumbar region, where their role in disc pathology remains controversial13-16. The inconsistent findings leading to that controversy may be due in part to the fact that the definition of disc degeneration varied among the studies. However, a more important factor may involve the ability to detect Schmorl nodes. There is a striking variation in the reported prevalence of Schmorl nodes, ranging from 9% to 75%, in previous studies13,15,17,18, with the prevalence measured in cadaveric spines being much higher than that measured with use of MRI. Thus, it appears that a substantial percentage of Schmorl nodes seen in cadaveric vertebrae are not detected on MRI. Consequently, data from cadaveric spines may be helpful to accurately identify end plate lesions and to determine their relation to disc degeneration.
The primary aims of the current study were to use cadaveric spines to determine the prevalence of end plate lesions, to characterize end plate morphology, and to explore the association of these factors with lumbar disc degeneration. We tested the hypothesis that smaller size and greater concavity of vertebral end plates are beneficial to the adjacent disc. We also tested the hypothesis that lesions on the vertebral end plates are detrimental, with larger lesions being associated with more severe disc degeneration. Since the main focus of our study involved the associations between the characteristics of the osseous end plate and the disc, we specified the end plates as cranial or caudal with respect to the intervertebral disc.
Subjects
We had access to an archive of 149 cadaveric lumbar spines from white male donors19. The donors were below the age of sixty-five years, had died in hospital wards, and had a short history of illness. Most of the donors had died from heart attacks or other cardiovascular problems. Exclusion criteria were prolonged hospitalization and death from cancer or infectious diseases. Age, body weight, and height had been obtained at the time of death. The families of eighty-six donors reported that the donors had had a history of back pain19. As one of the planned aims was to explore the association between end plate lesions and back pain, we included only subjects whose back pain history data were available. We included all available lumbosacral vertebrae and corresponding discography data in the present study. Although radiographs of all lumbar spines were available, some of the vertebrae had been lost during preservation. The study was approved by the Health Research Ethics Board at our institute.
Measurement of Intervertebral Disc Degeneration
Disc degeneration was evaluated with use of discography, which was performed after a routine autopsy examination of the lumbar spine19. Using a 20-gauge needle and finger pressure, 2 to 5 mL of BaSO4 was injected anteriorly into the disc center. In most cases, all five lumbar intervertebral discs (L1/2 to L5/S1) were examined. Anteroposterior and lateral radiographs were made after the injection of the BaSO4 contrast medium. A four-grade ordinal scale was used to rate the degree of disc degeneration as judged on the basis of anular disruption seen on the discograms. Disc degeneration was rated as none if the BaSO4 remained in the center of the disc, slight if it spread into the inner anulus fibrosus, moderate if it spread from the inner to the middle region of the anulus, and severe if it spread to the outer part of the anulus (Fig. 1). The intraobserver agreement of measurements made with use of this scale has been assessed previously, and the weighted kappa was found to be 0.8119.
Although both discography and MRI can be used to evaluate disc degeneration, discography is based on morphological changes within the disc, whereas MRI depends on changes in the water content of the disc. Although it is less commonly used today, discography is a traditional approach that is able to differentiate among successive stages of disc degeneration20, and it has been deemed comparable with MRI for the evaluation of disc degeneration21.
After discography, the specimen was steamed and the soft tissues around the vertebrae, including the end plate cartilage and the discs, were carefully peeled off the bones. The vertebrae were dried and then archived under room temperature and humidity, and the osseous vertebral end plates were studied.
Visual Assessments
In adults, a vertebral end plate consists of an epiphyseal rim, which is a ring of smooth bone at the periphery of the end plate, and a central portion22. The relatively smooth and solid epiphyseal rim anchors the anulus fibrosus, whereas the thin and porous central portion is covered by end plate cartilage and is adjacent to the nucleus pulposus12. The epiphyseal rim approximates the size of the anulus fibrosus, whereas the central portion of the end plate approximates the size of the nucleus pulposus.
End Plate Shape
All visual assessments of end plate shape were performed by one of the authors (Y.W.). The end plates were classified as concave, flat, or irregular according to the relationship between the central end plate and the epiphyseal rim. The apex of the concavity was further classified as absent, single, or double. Details of the end plate shape measurement method have been described previously23.
End Plate Lesions
The methodology for categorization of end plate lesions was developed by examining a sample of vertebrae. Schmorl nodes, characterized by a localized indentation in the central end plate with a smooth margin and an osseous wall12, commonly appeared in our sample. However, we also identified other forms of end plate pathology24, including what appeared to be trauma-related fissures and compressions, intensive calcium deposition, and some diffuse, shallow end plate “erosions.” For the purpose of the present study, these were grouped together with the Schmorl nodes as “end plate lesions.”
End plate lesions were rated visually, on the basis of the maximal diameter, as absent if the end plate had no lesion, small to moderate if the lesion was smaller than one-half of the anteroposterior diameter of the vertebral end plate, or large if it was one-half of the diameter or larger. A vernier caliper was used when necessary.
Since two vertebral end plates border each disc, measurements of end plate lesions from the paired cranial and caudal end plates were combined to investigate their association with disc degeneration. The end plate lesions for each disc were classified as absent if both end plates were intact, small to moderate if one end plate had small to moderate lesions and the other had no lesion, or large if at least one of them had large lesions or if both had small to moderate lesions.
The reliability of these visual evaluations was assessed by randomly selecting a sample of 200 vertebral end plates and reevaluating them one week later. The intrarater reliability was good or excellent (kappa = 0.78 for the end plate shape, 0.82 for the apex of the concavity, and 0.89 for the lesion size).
Digital Measurements of End Plate Morphometrics
Each vertebral end plate was scanned with use of a noncontact three-dimensional digitizer (VIVID 910; Konica Minolta Sensing Americas, Ramsey, New Jersey) to measure the surface geometry. The three-dimensional end plate images were processed and measured with use of the Polygon Editing Tool (version 2.21; Konica Minolta Sensing Americas). First, the sagittal and transverse diameters of the end plate were measured (in mm). The circularity (the ratio of the sagittal diameter to the transverse diameter) was calculated to quantify the cross-sectional shape of the end plate. The central portion of the end plate and the epiphyseal rim were then separated in the three-dimensional images of the end plate. Corresponding surface area measurements (i.e., measurements of the area of the three-dimensional surface) (in cm2) were obtained for the whole end plate, the central portion of the end plate, and the epiphyseal rim. The cross-sectional area of the end plate, defined as the planar area within the outermost rim of the end plate, was also measured. Finally, the three-dimensional end plate images were imported into three-dimensional graphing software (DPlot version 2.2.6.3; HydeSoft Computing, Vicksburg, Mississippi) to measure the mean depth (in mm) and volume (in mm3) of the end plate concavity. The details of this technique and the reliability of the resulting digital measurements have been reported previously23.
Statistical Analysis
Descriptive statistics were used to analyze end plate shape and lesions. As the prevalence of disc degeneration differs substantially between the upper and lower lumbar spine25, the lumbar discs were grouped into two groups, the upper (L1/2, L2/3, and L3/4) and lower (L4/5 and L5/S1) regions. Ordinal logistic regression was used to examine the associations between end plate morphometrics, lesions, and adjacent disc degeneration. First, univariate regression was used to analyze the cranial and caudal end plates separately, controlling for age and spinal level. Data from the cranial and caudal end plates were then combined and a purposeful procedure was used to construct a multivariate regression model, controlling for age, lumbar region (upper compared with lower), and body mass index (BMI). Data for selected quantitative measurements of the cranial and caudal end plates were averaged, and the combined end plate lesion grade was used to characterize end plate lesion size. If the data for one of the end plates were missing, the data for the other end plate were used in the analysis of adjacent disc degeneration. Statistical analyses were performed with use of Stata (version 9.2; StataCorp, College Station, Texas). The “cluster” command in Stata was used in all regression analyses to account for the fact that multiple vertebrae were analyzed in each lumbar spine.
Source of Funding
No funding was received in support of this study.
The study included 266 cadaveric lumbar vertebrae (L1 through L5), sixty-nine sacral vertebrae (S1), and 313 adjacent intervertebral discs (L1/2 through L5/S1) from seventy-six male human donors. Six hundred end plates (264 cranial end plates and 336 caudal end plates, including the cranial end plate of S1) were assessed radiographically and visually. Nine of these vertebral end plates were excluded from the morphometric assessment because of poor quality of the digital images, and quantitative measurements of the remaining 591 end plates were made. The mean donor age was 51.3 years (range, twenty-one to sixty-four years).
Prevalence of Findings
End plate lesions were found in fifty-five (72.4%) of the lumbar spines and in 197 (32.8%) of the end plates. Seventy-five (38%) of the lesions were rated as large and 122 (62%) were rated as small to moderate. The prevalence of lesions did not differ significantly between the cranial and caudal end plates (p = 0.87, chi-square test). Small or moderate end plate lesions were more common in the upper lumbar region (69.7%), whereas large lesions were more common (70.7%) in the lower lumbar region (p < 0.001, chi-square test).
Overall, 58.2% of the end plates were visually assessed as concave in shape (with 238 [39.7% of all end plates] having a single apex, 111 [18.5%] having two apices, and the remaining 251 [41.8%] having no apparent apex), 33.3% as flat, and 8.5% as irregular.
On the basis of discography, forty (12.8%) of the discs had no evident disc degeneration, 101 (32.3%) were rated as having slight degeneration, seventy-one (22.7%) had moderate degeneration, and 101 (32.3%) had severe degeneration.
The distribution of disc degeneration according to end plate lesion size is presented in Figure 2. Larger end plate lesions tended to be associated with more severe adjacent disc degeneration.
Association of End Plate Morphometrics and Lesions with Disc Degeneration
Univariate Regression (Table I)
Age (odds ratio [OR] = 2.1 per decade increase in age, 95% confidence interval [CI] = 1.7 to 2.6, p < 0.001) and lumbar region (OR = 3.8 for lower compared with upper, 95% CI = 2.5 to 5.7, p < 0.001), but not BMI (p = 0.33), explained a portion of the variance in the occurrence of disc degeneration. Age and lumbar region were therefore controlled for in all univariate regressions, and disc degeneration was the dependent variable.
Irregular end plate shape (OR = 2.6 to 2.7, p < 0.05), the presence of lesions (OR = 2.2 to 5.3, p = 0.00 to 0.01), and greater area of both the cranial and the caudal end plate (OR = 1.18 to 1.23 per cm2, p = 0.003 to 0.05) were associated with more adjacent disc degeneration. Smaller mean depth of the concavity (OR = 0.7 per mm, p = 0.02) and greater end plate circularity (OR = 1.07, p = 0.004) in the end plate cranial to the disc, but not in the caudal end plate, were also significantly associated with ore severe adjacent disc degeneration. Furthermore, greater central end plate area both cranially and caudally (OR = 1.23 to 1.32 per cm2, p < 0.05) was associated with adjacent disc degeneration, but the area of the epiphyseal rim was not (p = 0.23 to 0.41).
Multivariate Regression (Table II)
The visual assessments of end plate shape and apex indicated the approximate degree of concavity, which was further measured digitally. Only the digital measurements were considered in the multivariate model. Although the measurements of the mean depth and volume of the end plate concavity were highly correlated (r = 0.92), the volume measurement was not significantly associated with disc degeneration in the univariate model. Thus, only the mean depth measurement was used in the multivariate analysis. Similarly, the end plate surface area and cross-sectional area measurements were highly correlated (r = 0.95), and only the end plate cross-sectional area measurement, which was independent of concavity, was selected to indicate the size of the end plate. Therefore, the final model included end plate lesions and three end plate morphological measurements: the mean depth of the concavity, the cross-sectional area, and the circularity. Disc degeneration was the dependent variable in the final regression model.
After controlling for age, BMI, and lumbar region, end plate lesions were found to be significantly associated with disc degeneration, with greater lesion size being associated with more severe adjacent disc degeneration (OR = 2.31 for small to moderate lesions and 3.54 for large lesions, p < 0.001 for both).
Among the three analyzed measurements of end plate morphometrics, only the cross-sectional area was significantly associated with adjacent disc degeneration (OR = 1.2 per cm2, p = 0.027), with larger end plate size associated with more disc degeneration. There was a tendency for greater circularity and less concavity of the end plate to be associated with more adjacent disc degeneration, but neither effect reached significance. When end plates with lesions were excluded and the association between morphometrics and adjacent disc degeneration was analyzed in only the intact end plates, similar results were obtained: only end plate cross-sectional area was associated with adjacent disc degeneration (OR = 1.21 per cm2, p < 0.05).
Furthermore, when the area of the central portion of the end plate was used in place of the total end plate cross-sectional area, an association between greater size of the central end plate and more disc degeneration was still observed (OR = 1.22 per cm2, p = 0.018). However, when the epiphyseal rim area was used in the model, no association was found (OR = 1.06 per cm2, p = 0.40).
Our data revealed that the integrity of the vertebral end plate and that of the intervertebral disc were interdependent. End plate lesions were common in the lumbar spine, appearing in approximately one-third of end plates, and were associated with adjacent disc degeneration seen on discography, with greater lesion size associated with more severe adjacent disc degeneration. The findings also indicated an association between larger end plates and more adjacent disc degeneration. More specifically, a larger central end plate region, reflecting a larger nucleus pulposus, was associated with more adjacent disc degeneration. Greater end plate concavity and less circularity may play a marginal role in the pathogenesis of disc degeneration.
The current study has several strengths. First, the visual inspection of the vertebral end plates included the entire end plate and the lesion size was measured directly, making this method inherently superior to imaging evaluations. MRI is unlikely to detect all end plate lesions, as only a few sections of the vertebra are typically sampled. The observed prevalence of end plate lesions (using a broad definition) was higher in this study compared with previously reported MRI findings15-17, but it was in line with previous studies of cadavers from white adults of similar age, in which Schmorl nodes were observed in 48% to 75% of spines13,14,18,26. Second, discography provides a reliable and valid measurement of disc degeneration19,27 that is not influenced by features of osseous degeneration. Third, appropriate statistical models were used to examine the local interactions between end plates and adjacent discs. Earlier studies typically assessed Schmorl nodes and disc degeneration according to spinal level and then summed the scores for the entire spinal region studied16,28, which may have diluted the associations.
Echoing the results of a clinical CT study that suggested that larger end plate size was associated with disc herniation in men10, the findings of the present study support an association between greater end plate size and more adjacent disc degeneration. The mechanism underlying this association remains unexplained. On the basis of our earlier findings that greater cross-sectional disc size measured with MRI is a risk factor for disc degeneration9, we speculated that it is the larger size of the disc, rather than that of the end plate, that explains the association found in the present study. The measured cross-sectional area of the end plate reflects the “original” size of the corresponding intervertebral disc. Furthermore, we found that the size of the central region of the end plate, but not that of the epiphyseal rim, was associated with the adjacent disc degeneration. As the nucleus pulposus lies between the central regions of the end plates12, we postulate that a larger disc, especially one with a larger nucleus pulposus, is more susceptible to degeneration than a smaller disc. Although a larger central end plate region tends to have more marrow channels supplying nutrients to the disc29, a larger nucleus pulposus also has more cells demanding nutrients, and nutrient transport within the nucleus matrix may be impeded by a larger size. Thus, the association between larger end plates and more adjacent disc degeneration may be due to an overall decreased efficiency of nutrient supply to a large disc.
End plate shape, as judged from sagittal MRIs, has previously been reported to be associated with adjacent disc degeneration in surgically treated patients8, with flat and irregular end plates associated with more severe disc degeneration compared with concave end plates. Although our findings support an association of irregular end plate shape with adjacent disc degeneration, it remains unknown whether the irregular end plate shape is a cause or a consequence of the disc degeneration. End plate concavity was thought to represent an adaptation to age30 or axial compressive loading31. The disc was believed to be protected by sinking into the vertebral body30, with greater end plate concavity being associated with less disc degeneration8. The accurate digital measurements in the current study supported this association with respect to the cranial end plate, but not with respect to the caudal end plate. This discordant association of disc degeneration with the cranial and caudal end plates, which was also present in the association of greater end plate circularity with disc degeneration, is puzzling, but it may be due to the structural asymmetries between the end plates. For example, the cranial end plate is more concave23, is thicker, and has greater bone mineral density than the corresponding caudal end plate32. In contrast to the situation involving end plate fractures, which have been reported to be more common in the caudal end plate of a disc than in the cranial end plate33, the prevalence of lesions in the current study did not differ significantly between cranial and caudal end plates. This may be due to the definition of end plate lesions that we used, which likely included multiple end plate pathologies (fractures, Schmorl nodes, and others). The associations between specific types of end plate lesions and disc degeneration need further investigation.
This study also has some limitations. First, all of the donors were male, and most of them had been employed as laborers. Thus, findings regarding the prevalence of end plate lesions may not be generalizable to women or to individuals who are not laborers. Second, some vertebrae were missing from the spine archive. It is possible that the prevalence of end plate lesions was underestimated because of the missing vertebrae. Third, end plate lesions as defined in the present study included a variety of end plate pathologies. Thus, it is not appropriate to compare the current findings with those from focused studies of Schmorl nodes. It is also possible that different types of end plate lesions may play distinct roles in the pathogenesis of disc degeneration.
In summary, vertebral end plate lesions were common and were associated with adjacent disc degeneration, with greater end plate size associated with more severe adjacent disc degeneration. The findings strongly suggest that the integrity of the vertebral end plate is important in maintaining the health of the adjacent intervertebral disc. The morphometrics of the end plate, and particularly the area corresponding to the cross-sectional disc size, may play a modest role in the pathogenesis of disc degeneration. Evidence from the current cadaveric study suggests that the end plate may be involved in the pathway of the disc degeneration cascade. However, the cross-sectional nature of the study limits the ability to draw conclusions regarding causation.
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. One or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.