Introduction

The vertebral endplate is located between adjacent intervertebral discs and vertebral trabeculae and serves as passing way for substance exchange and mechanical transmission [1,2,3]. Many degenerative spinal disorders, such as lumbar disc herniation (LDH), degenerative spondylolisthesis and nonspecific low back pain (LBP), were found to be associated with intervertebral disc degeneration (IVDD) [4,5,6]. Vertebral endplate defects have various defect types and are commonly seen in disc degeneration disorders and have a close relationship with IVDD [7]. Endplate defects are clinically relevant and associated with LBP [8, 9] and influence clinical recovery after spine surgeries [10].

Osteoporosis is a condition of low BMD and microarchitectural degeneration of bone resulting in an increased risk of fracture [11]. Osteoporosis causes severe health problems. The prevalence of osteoporosis among the Chinese population is approximately 13% and increases with age [12]. Previous articles focused on the relationship between osteoporosis and IVDD. A study analyzed data from 168 postmenopausal women and found that higher lumbar bone mineral density (BMD) and Z score were related to more severe lumbar IVDD [13]. However, the positive relationship between osteoporosis and IVDD was not confirmed and remained controversial in other studies [14, 15].

The vertebral endplate has a porous microstructure. Vertebral trabeculae were microstructurally heterogeneous and reduced with age [16]. The density and trabecular architecture of the endplate region showed few differences between women and men, and the density of the endplate was associated with vertebral fracture [17]. The condition of bone quality influences the structure of endplates and may play roles in further disc degeneration. An animal experiment study based on rhesus monkeys found that osteoporosis decreased vascularization in the endplates and caused endplate calcification [18]. In a rat model combined with ovariectomy (OVX) and muscle section, researchers found that endplates exhibited abrasion defects and IVDD at specific times [19]. Micro-CT analysis of cadaveric lumbar spines showed that the endplate became more porous with age and IVDD [20]. Endplate remodeling was accompanied by increased porous microstructure in the OVX model and that the change affected matrix metabolism in the intervertebral disc [21]. Drug therapies to increase BMD have been developed since the 1940s, and common drugs include estrogen, calcitonin, bisphosphonates, teriparatide and Wnt signaling inhibitors [22]. Some papers reported that vitamin D, calcium and bisphosphonates could affect cartilage metabolism, and whether attempts to increase BMD would influence endplates and further lead to IVD is another doubt [23]. To the best of our knowledge, there is a lack of clinical evidence about the association between endplate defects and BMD and osteoporosis based on a large population in lumbar degenerative disc disease (LDD).

The present study was designed to explore the association between endplate defects and lumbar BMD conditions, and osteoporosis in patients with LDD. Furthermore, we examined the relationship between the lumbar BMD and IVDD and compared the prevalence of defect subtypes between the upper and lower lumbar spines at different T values.

Methods

Study population

This study was a retrospective study. The study included 325 Chinese adult subjects living in North China and was approved by the ethical committee of the medical center. The inclusion criteria required the patients to be diagnosed with LDD (nonspecific LBP, LDH, lumbar spinal stenosis and degenerative spondylolisthesis). The time range of the participants was from 2020 to 2021. The study excluded patients with abnormal renal function, parathyroid disease, spinal tumors, spinal infection, isthmic spondylolisthesis, congenital scoliosis, and trauma-induced lumbar fracture. Endplate defect evaluation, BMD measurement and serum biochemical index detection were performed when the patients were at the outpatient visit or the patients were admitted to the medical center before surgery.

Endplate defects and disc degeneration evaluation

The morphological characteristics of the endplates were evaluated based on lumbar magnetic resonance imaging (MRI). Sagittal MRIs of endplates from the L1 inferior endplate (IEP) to the S1 superior endplate (SEP) were obtained using a 1.5 T MR scanner. The detailed methods and standards of endplate defect evaluation have been well elaborated in previous articles [6, 9]. The endplates with defects were categorized into rim defects, local defects and erosive defects in this study. If more than one defect type was present in an endplate, the main type was recorded. To analyze the potential confounders of endplate defects, we divided the patients into a higher defect rate group and a control group. The incidence of lumbar endplate defects was approximately fifty percent in preliminary observations. Fifty percent was chosen as the cutting point. We defined a higher defect rate as the occurrence rate of lumbar endplate defects (from L1 IEP to S1 SEP) in a patient was more than fifty percent. Two well-trained researchers evaluated the defect types of each endplate. If their evaluation of the same endplate was different, an experienced surgeon would decide the final type without knowing the BMD results. MRIs from 100 patients were randomly selected at an interval of 2 weeks, and an observer performed the second assessment. Intraobserver variability and interobserver variability were analyzed. Disc degeneration was assessed using the Pfirrmann grading system (grades I to V). Intervertebral discs from L1/L2 to L5/S1 of each patient were evaluated.

Dual energy X-ray absorptiometry (DXA)

DXA is the gold standard for measuring BMD [24]. BMD was measured using a Discovery A DXA scanner (Hologic, Bedford MA). BMD at the lumbar vertebrae and bilateral hips was measured. The T score and Z score indicated the standard deviation of the mean BMD in young, healthy adults and age-matched adults of the same sex, respectively. Total lumbar BMD was calculated based on data from the L1 vertebra to the L4 vertebra. BMD at hips included areas such as the neck, troch and other parts. Osteoporosis was diagnosed by a physician from the BMD examination department based on the overall condition of BMD from the lumbar vertebrae and bilateral hips. The clinical standard was as follows: 1. Children, males beyond fifty years old and premenopausal females referred to Z values; if the represented Z values were beyond −2.0, the patients were diagnosed with low bone mass. 2. Other populations referred to T values; if represented T values were beyond −2.5, the patients were diagnosed with osteoporosis.

Serum biochemical index detection

Fasting venous blood was collected in the morning for serum separation. Serum levels of 25-hydroxy vitamin D (25(OH)D), calcium (Ca) and phosphorus (P) were detected.

Statistical analysis

The data were analyzed using SPSS 25.0 software. Normality and homogeneity of variance were checked, and the results were expressed as the mean ± standard deviation. Intraobserver variability and interobserver variability of evaluation endplate defects were analyzed using kappa coefficients. Comparisons of means and incidence rates were made using t tests and Chi-square, respectively. Particular correlation analysis was used to explore the associations between incidence rates of endplate defects, disc degeneration and other variables. Unstandardized beta coefficients for the association of the incidence rates of endplate defects with lumbar bone quality values, disc degeneration and other variables were assessed using linear regression analysis. Adjustment was then made for age, sex, BMI, serum levels of Ca and P in multiple linear regression analysis. Constitution of endplate morphology subtypes in the upper and lower lumbar regions in patients with different lumbar T values is shown in pie charts. A heatmap was generated to visualize the results of linear regression analysis. Statistical significance was found if the p-value ≤ 0.05.

Results

Patient characteristics

A total of 325 Chinese adult subjects with 3250 lumbar endplates and 1625 intervertebral discs were analyzed. The age of the patients varied from 18 to 85 yr, with an average of 58 yr. Most lumbar endplates had defects (1941/3250, 59.72%). A total of 84 patients were diagnosed with osteoporosis. Table 1 shows the clinical characteristics and biochemical profile characteristics of patients in higher defect rate group (n = 188) and control group (n = 137). The mean age in the higher defect rate group (61.90 ± 9.03 yr) was significantly higher than that in the control group (52.64 ± 12.08 yr). They also had more postmenopausal status in females (89 vs. 45; p < 0.001), lower serum Ca levels (2.27 ± 0.12 vs. 2.31 ± 0.12; p = 0.002) and lower serum P levels (1.19 ± 0.17 vs. 1.24 ± 0.17; p = 0.013) than the control group. No significant difference was seen in sex, BMI, diabetes or serum 25(OH)D levels.

Table 1 Baseline demographic and biochemical profile characteristics
Full size table

Bone mineral density in patients with higher defect rates and controls

The BMD, T value and Z value of the lumbar vertebrae and bilateral femur necks were compared between the higher endplate defect rate group and the control group (Table 2). The Z value of vertebrae in the higher defect rate group was significantly higher than that in the controls (0.27 ± 1.92 vs. −0.29 ± 1.25; p = 0.002). The lumbar T value followed the same trend without a significant difference. The BMD, T values and Z values of the femur necks were lower than those of the lumbar vertebrae. Significantly lower BMD and T values of the bilateral femur necks were seen in the higher lumbar endplate defect rate group. The percentage of osteoporosis diagnosis in the higher endplate defect rate group was 30.85% (58/188), which was significantly higher than the 18.98% (26/137) in the control group (p = 0.016).

Table 2 Bone mineral density-related values of the lumbar vertebrae and bilateral femur necks in two groups
Full size table

Endplate defect subtypes and disc degeneration analysis

Disc degeneration grades in the lumbar T ≤ −2.0 group (n = 89) and T > −2.0 group (n = 236) are shown in Table 3. The lower lumbar spine (L4/5 and L5/S1) had more severe disc degeneration than the upper lumbar spine (L1/2 to L3/4). Even though disc degeneration grades in the T ≤ −2.0 group were higher at some segments (L2/3 to L5/S1), no significant difference was found between the two groups.

Table 3 Comparison of Pfirrmann grades of each segment between the groups with different lumbar T values
Full size table

The intraobserver variability and interobserver variability in the evaluation endplate defects at each endplate level are shown in Table 4. The kappa coefficient from the interobserver values was > 0.8. The kappa coefficient from intraobserver values ranged from 0.870 to 0.958. The constituents of endplate defect subtypes in the upper lumbar spine (L1 IEP to L4 SEP) and lower lumbar spine (L4 IEP to S1 SEP) were compared and are presented in Fig. 1. The constituents of normal endplates and defect subtypes were similar in the lumbar T ≤ −2.0 group and T > −2.0 group. Lower lumbar spine had higher rates of abnormal endplates (66.29% vs. 54.68% in the T ≤ −2.0 group; 66.00% vs. 55.79% in T > −2.0 group). Erosive changes were more common in lower lumbar spine in both groups (28.93% vs. 3.93% in T ≤ −2.0 group; 25.53% vs. 6.57% in the T > −2.0 group).

Table 4 Intraobserver variability and interobserver variability in the evaluation endplate defects at each endplate level
Full size table
Fig. 1
figure 1

Comparison of the constitution of endplate morphology subtypes in the upper and lower lumbar between the two groups

Full size image

Association between the occurrence of defect endplates and lumbar bone mineral density values

Particular correlation analyses between endplate defects and lumbar BMD-related values and disc degeneration are summarized in Table 5. The analysis was adjusted for potential confounders including age, sex and BMI. A positive correlation was found between the incidence rate of endplate defects and lumbar BMD (p < 0.01), T value (p < 0.01) and Z value (p < 0.001). The occurrence of endplate defects was negatively correlated with Ca and P levels (p < 0.01). Disc degeneration was only significantly correlated with incidence rates of endplate defects (p < 0.001) and lumbar Z values (p < 0.05). The partial correlation coefficient between endplate defects and disc degeneration was 0.566.

Table 5 Partial correlation coefficient between incidence rates of endplate defects, disc degeneration and other variables
Full size table

The associations between endplate defect subtypes and lumbar BMD-related values were further analyzed, and adjustments were made for confounders including age, sex, BMI, serum Ca, P and 25(OH)D levels (Fig. 2). Endplate defects were significantly associated with age in most conditions. In overall, endplate defects were significantly associated with Z values, not BMD and T values before adjustment. The relationship became significant for BMD and T and Z values when adjusted for confounders. Local defects in the lower lumbar spine were negatively associated with the Z value (p < 0.05). A significant negative association was found between the rim subtype in the upper lumbar spine and serum Ca and P levels (p < 0.05). The association remained significant between endplate defects and disc degeneration in most conditions after adjustment.

Fig. 2
figure 2

Heatmap of unstandardized beta coefficients for the association of incidence rates of endplate defects with lumbar BMD-related variables and disc degeneration, *Unstandardized beta coefficients for all endplate defect rates (dependent variable) for lumbar BMD-related variables and disc degeneration (independent variable). **Statistically significant values with P < 0.05 are represented in color. The analyses of lumbar BMD, T value, Z value and disc degeneration were adjusted for age, sex, BMI and serum levels of Ca and P

Full size image

Discussion

In this paper, we retrospectively evaluated the occurrence of lumbar endplate defects in LDD patients of a large population. The evaluation included BMD-related values from various bone areas to avoid mismatching between lumbar BMD and osteoporosis diagnosis. The analysis also made adjustments for potential confounders to reduce errors. Few clinical studies have revealed the association between endplate abnormalities and BMD conditions, and the relevant studies were based on animal experiments [18, 19, 21]. The present findings indicated that lumbar endplate defects were associated with BMD conditions in patients with LDD.

Osteoporosis is age-related disease. Increasing age and female sex are risk factors for osteoporosis [25]. In this article, 25.85% (84/325) of patients were diagnosed with osteoporosis, the average age of osteoporosis patients was 64.24 ± 7.53 years old, and 55.82 ± 11.69 years old in the other patients (p < 0.001). Postmenopausal females were more frequently seen in osteoporosis patients (54/61 vs. 80/161; p < 0.001). Endplate defects and disc degeneration are common phenomena in degenerative spines and have been proven to be associated with aging [6, 9]. Hormonal factors, calcium and vitamin D levels are associated with osteoporotic fracture [26]. A lack of 25(OH)D levels (< 10 ng/mL) was found to be associated with LBP and disc degeneration in postmenopausal women [27]. However, few patients had low 25(OH)D levels (< 10 ng/mL), and the relationship between serum 25(OH)D levels and endplate defects and disc degeneration was not significant in this study. Serum Ca and P levels were in the normal range in most patients, and they had little variation in the participants. Some researchers thought serum calcium levels negatively correlated with lumbar BMD in older adults [28]. Serum calcium levels influence BMD at specific skeletal sites [29]. We found that serum Ca and P levels were lower in the higher endplate defect rate group (p < 0.05), but lumbar BMD was not significantly higher compared to the other group (p = 0.393). Excluding the influence of age and other confounders is important to analyze the relationship between BMD and endplate defects. The occurrence of endplate defects was significantly correlated with lumbar BMD-related values in partial correlation analysis (from r = 0.151 to r = 0.196, p < 0.01). Only lumbar Z values were significantly associated with the overall incidence rate of endplate defects in the regression analysis before adjustment. The association with lumbar BMD and T value became significant when adjusted for age, sex, BMI and serum levels of Ca and P.

Whether osteoporosis is a causative or protective factor for IVDD has been controversial, which suggests that the mechanism of osteoporosis on intervertebral disc components is still unclear [21, 30]. Interestingly, lumbar BMD values were not consistent with BMD in other bone areas. BMD, T values and Z values at the bilateral femur necks were lower than the lumbar BMD values. This may explain the mismatch between lumbar BMD-related values and the diagnosis of osteoporosis. It was not surprising to find that even though osteoporosis patients were more common in the higher defect rate group (30.85%) than in the control group (18.98%), the lumbar BMD-values were positively associated with the occurrence of endplate defects in further analysis.

Degenerative disc diseases were more common in the lower lumbar segments. An article compared patients with upper- and lower-disc dislodgement and found that patients were 10 years younger in the lower lumbar dislodgement group [31]. Endplate abnormalities were more prevalent in the lower lumbar spine (L4 to S1) than in the upper lumbar spine (L1 to L3) in LBP patients [32]. The upper and lower lumbar vertebrae bear different mechanical effects. Mechanical loading can not only directly injure structures such as endplates and discs, but also alter cell homeostasis through signaling pathways related to matrix metalloproteinases and cytokines [33]. This may explain the different prevalence and characteristics of normal endplates and defect endplates in the upper and lower lumbar spine. Disc degeneration was also more severe (Table 3), and defective endplates were more common in the lower lumbar spine under different BMD conditions (Fig. 1).

The endplate passage pathway is a main method of nutrient transport to intervertebral discs and is considered a potential therapeutic target to cure IVDD [34, 35]. Some previous studies used an OVX model to mimic bone loss conditions and found that OVX-induced vertebral osteoporosis had adverse effects on intervertebral discs [18, 21]. However, the real conditions in patients with degenerative spinal disorders are much more complex, and the concept of osteoporosis in animal experiments is not equal to the clinical diagnosis. A clinical article measured hip areal BMD with DXA, and reported no association between hip BMD and lumbar disc degeneration [15]. Some researchers used microcomputed tomography to obtain BMD from 137 cadaveric lumbar vertebrae and they excluded osteophytes and endplates. The association between greater BMD and more severe disc degeneration became significant when they excluded posterior elements of the vertebra [14]. Whether lumbar BMD correlates with IVDD and how BMD influences IVDD in patients with LDD are not clear. Unlike previous studies, the present study had a larger population and was closer to clinical reality. We collected BMD at various bone areas using DXA and focused on the association between lumbar BMD-related values and endplate defects, and IVDD. The findings were consistent with previous articles. IVDD tended to be more severe at lower T values, even though this trend was not significant (Table 3). More severe IVDD was positively associated with greater Z values (Table 5). The close relationship between endplate defects and IVDD has been confirmed in many studies and in this article. Recently, high-quality animal research used mutated sp7 and cathepsin K zebrafish and found a strong correlation between high BMD and IVDD, but low BMD does not protect against IVDD [36]. Based on previous and present studies, we hypothesized that endplate defects were influenced by lumbar BMD and eventually led to IVD in patients with LDD. This procedure associated with multiple factors.

This study has limitations. First, it was not age- or sex-matched between the two groups, even though this was adjusted in the statistical analysis. Second, we were not able to quantify the area size of each defect endplate using routine MRI. The prevalence of defect endplates was expressed by the occurrence rates. Further studies that only include specific demographic characteristics with a large population are needed to provide better insights.

Conclusion

To our knowledge, this is the first study to describe the association between endplate defects and lumbar BMD-related values and osteoporosis adjusted for age, sex and other confounders. Endplate defects are associated with lumbar BMD conditions, with the overall trend that higher occurrence rates are related to greater lumbar BMD in patients with LDD. In addition, this association varies for different defect subtypes and segments. Endplate defects should be taken into consideration in theory that osteoporosis treatment alleviates IVD.