Spinopelvic alignment predicts disc calcification, displacement, and Modic changes: Evidence of an evolutionary etiology for clinically‐relevant spinal phenotypes

Abstract Lumbar disc‐displacement, Modic changes (MCs), and UTE Disc Sign (UDS) on MRI are clinically relevant spinal phenotypes that can lead to sciatica/LBP. Not all degenerated discs result in disc‐displacement, MCs and UDS, suggesting varied etiologies. Spinopelvic parameters have been implicated in various spinal disorders. Pelvic incidence (PI) is “fixed parameter” since skeletal maturity. No study has addressed disc‐displacement, MCs and UDS in context of spinopelvic parameters. Therefore, the aim of study was to determine if spinopelvic parameters are associated and predict clinically‐relevant MRI‐phenotypes. One hundred and eight population‐based subjects (mean age: 52.3 years) were recruited. Spondylolisthesis and scoliosis individuals were excluded. Lumbar lordosis (LL), PI, sacral slope (SS), and pelvic tilt (PT) were assessed on lateral plain radiographs. Disc degeneration was assessed and summated, and presence or not of disc‐displacement and MCs were noted on T2W MRI. UDS was detected on UTE. Following exclusion criteria, 95 subjects were assessed. Disc‐displacement (82.1%), MCs (52.6%), and UDS (37.9%) were associated with lower PI, SS, LL, and LL/PI index. On multivariate analyses, lower PI was significantly related to development of these MRI phenotypes (adjusted OR range:0.95‐0.92; P < .05), with critical PI value of 42° or lower exhibiting fourfold increase risk of combined phenotypes (P = .020). Of UDS discs, 39.3% had adjacent MCs and 83.6% had disc‐displacement. 87.5% of MC had directly adjacent UDS. The first study to note that PI may “predict” the development of disc‐displacement, MCs and UDS, suggesting potential sub‐variants and mechanistic susceptibility that may be grounded in spinopelvic evolution. An “evolutionary etiological pathway” of spinal phenotype development is proposed.


| INTRODUCTION
Low back pain (LBP) is the leading disability worldwide and a tremendous socioeconomic burden. 1,2 Mismatches between conventional T2-weighted (T2W) MRI-defined disc degeneration and LBP exist. 3,4 Spinal phenotypes consistently associated with, and predictive for, LBP and/or sciatica, are lumbar disc displacement and Modic changes (MCs; ie, subchondral vertebral bone marrow lesions). 4,5 Vertebral endplate structural defects, such as Schmorl's nodes, may often be benign. 6 Lifestyle factors, excessive loading, and genetics among others may contribute to the development of such spinal phenotypes. 5,7,8 However, such risk factors are often not consistently replicated, suggesting that other factors can be more influential and/or part of the spectrum. 9,10 The "UTE Disc Sign" (UDS) on ultra-short time-to-echo (UTE) MRI by Pang et al 11 was recently reported, and found to be strongly associated with disc degeneration severity, lumbar disc displacement, MCs, LBP, and disability. The UDS was noted to may represent disc calcification, which usually embody active inflammation and can "stiffen" the disc material; thereby, affecting the kinematics of the disc and motion segment. 12 Such calcification has been found in degenerative and scoliotic discs, suggesting that abnormal mechanical loading may be a likely mechanism. 13 Therefore, the UDS may potentially have a role in the initiation and propagation of lumbar disc displacement and/or instability, causing disc and endplate disruptions/damage that can potentially lead to MCs; thereby, increasing the risk for LBP.
Balanced sagittal alignment of the spine is vital for spinal function and is crucial to maintain upright postures. 14,15 The neutral upright sagittal alignment is achieved only when the spine and pelvis are in sync. 16,17 The harmonious connection of the pelvis with the spine, also known as "spinopelvic balance," contributes considerably to overall sagittal balance. The three main defining spinopelvic parameters are pelvic incidence (PI), pelvic tilt (PT), and sacral slope (SS) (Figure 1). 18 PI is a "fixed" parameter, akin to one's genetic make-up, at skeletal maturity 19 and may vary between individuals. 14 PI is essentially an individual's genetic blue-print. This is based on concurrent spinopelvic adaptations as a consequence of human evolutionary requirements for a well-balanced upright spine in the progression toward permanent bipedalism. 20 Increased PI is an adaptation for naturally selecting a nearly perfect pelvis that can support the upright trunk and at the same time provide energy-efficient bipedalism. 21 PT and SS are usually determined by pelvic orientation and can vary. 17 PI, a summation of the PT and SS, is arguably the most studied spinopelvic parameter 22 and may highly be associated with pain and disability. 15,17,23 PI directly correlates with lumbar lordosis (LL), 24 which may be affected by degenerative changes. 17 PI is generally thought to be unaffected by lumbar degenerative changes. 17 Association of increased PI with spondylolisthesis 25 and facet joint degeneration 14,16 has been reported by several studies. Previously, a decreased PI was noted with degenerative disc disease, lumbar disc displacement, and chronic LBP. 17,26,27 Therefore, spinopelvic balance disruption can alter biomechanical stresses at the lumbo-sacral junction and also in the compensation mechanisms used to maintain an adequate posture, 28 leading to aforementioned spinal and disc changes as well as LBP. However, not all disc degeneration on MRI may lead to lumbar disc displacement and/or MCs, and that potential sub-variants in the etiologies of these phenotypes may exist. 5 Since the advent of MRI, a discrepancy still remains between various spinal phenotypes and their potential etiological factors. [29][30][31] Therefore, it is very plausible that spinopelvic parameters may be the "missing link" between clinically-relevant phenotypes, such as lumbar disc displacement, MCs, and UDS, and their development. Therefore, such spinopelvic parameters warrant further investigation to address this constellation of MRI findings. As such, this study aimed to determine if spinopelvic parameters, in particular PI (fixed entity), are associated and can "predict" the development of clinically-relevant MRI phenotypes of lumbar disc displacement, MCs, and UDS.

| Study population
One hundred and eight Southern Chinese volunteers with age range of 22 to 67 years (mean age: 52 years; 50% males) were recruited from a population study to be part of a new cohort addressing novel imaging. The recruitment parameters have been reported elsewhere. 7,8,32 In short and following institutional review board approval, 108 subjects were randomly recruited, irrespective of pain profile. No subjects underwent previous spine surgery. The sample size was determined based on funding constraints; however, currently represents the first study to date that have undergone T2-and T1-weighted as well as UTE MRI along with standing lateral plain radiographs. All subjects were enrolled consecutively and informed consent was obtained. Of these 108 subjects, 11 subjects were excluded for the following study because they exhibited spondylolisthesis, scoliosis, trauma, or active infections on MRI that were also confirmed on plain radiographs, and/or the femoral head was not visible to facilitate PI assessment. Two subjects were also excluded because, due to scheduling conflicts, did not undergo plain radiograph assessment. As such, 95 subjects were included in the following cross-sectional study.

| Radiographic measurements
Lateral plain standing radiographs of the lumbar spine (L1-S1), pelvis and proximal femur were assessed. The volunteers were standing erect with arms raised and slightly fisted hands resting on their clavicles. The film focus distance was 180 cm and other exposure factors were 88 kVp and 32 mAS. These radiographic acquisition parameters were kept consistent in all the individuals, irrespective of their body mass index (BMI). All radiographs were acquired digitally. LL and spinopelvic parameters (ie, PI, PT, and SS) were measured based upon the lateral radiographs, and are defined as illustrated in Figure 1. 15,17 The LL/PI index was also tabulated. 33 A medical doctor experienced in image assessment and blinded to MRI findings assessed all plain radiographs (UZ). All radiological parameters on 25 randomly selected radiographs were reassessed after 3 weeks to obtain the intraobserver reliability estimate. Based on Cronbach's alpha reliability assessment, 34

| MRI Assessment
All subjects underwent MRI of L1-S1 via a 3 T MRI scanner (Achieva, Philips Healthcare, Best, The Netherlands). Sagittal T2W MRI were acquired using a standard spin-echo imaging sequence with the fol- Based on sagittal T2W MRI, the Pfirrmann et al 35 grading system was used to assess disc degenerative scores. Grades 4 and 5 were regarded as "black discs." A cumulative "disc degeneration score" (potential range: 0-25) 36 was obtained from a summation of individual discs scored from L1 to S1 via the Pfirrmann et al 35 method. Lumbar disc displacement (disc bulges, protrusion, and extrusion were grouped together), represented by the displacement of the annular fibers beyond the vertebral margin were assessed. MCs 10,37 were noted, regardless of the Type due to the study sample size. One may deduce that the majority of the MCs were Type II, as noted in our previous studies of this ethnic group population. 37 In fact, we have found Modic Type I and Type II to be associated with pain/disability in past studies. 10,54 Sagittal UTE MRI was used to detect UDS (ie, hypointense disc band). 11 The location of the UDS in relation to MCs and lumbar disc displacement was also noted. If UDS and MCs overlapped or not in the same location, this observation was noted. MRI phenotypes were assessed by trained raters (U.Z., J.P.Y.C., and D.S.). All raters were kept blinded to radiographic findings. The inter-and intrarater reliability of T2W MRI and UTE MRI phenotypes were excellent (k = 0.91, k = 1.00, respectively) and previously reported. 11,32 Age (years), sex-type (males vs females), body weight (kilograms), body height (meters), and BMI (kg/m 2 ) were noted for every individual.

| Statistical analyses
SPSS v24 (Chicago, Illinois) was used to perform the statistical analyses. Descriptive and frequency analyses were performed of the data set, noting percent (%), and mean ± SD (SD) values. The spinopelvic parameters were found to be parametric. Univariate analyses consisted of independent-samples t-test and chi-square or Fisher's Exact Test where appropriate. Pearson correlation analysis (r) was obtained.
Multivariate analyses consisted of binary logistic regression, whereby odds ratios (OR) and 95% confidence intervals (CI) were evaluated.
Due to the sample size, a limit was placed as to how many co-variates could be used in a model. For the context of this study and based on prior knowledge, age, sex-type, and BMI were the subject characteristics noted in each model, along with the PI and the PI-LL index since the PI parameter is the only spinopelvic measure that is "fixed" in individuals and nonmodifiable to assess "prediction." The cumulative disc degeneration score was also used for model adjustment in an attempt to control for potential disc degeneration severity effects upon the development of lumbar disc displacement, MCs, UDS and a combination of all three. Factors related to black discs were also assessed. Statistical significance was established at P < .05.

| DISCUSSION
This is the first study that has assessed the predictive role of PI in lumbar disc displacement, MCs, and UDS. Our findings indicate that low PI "predicts" these clinically relevant spinal phenotypes, increasing the risk of development up to 8% for "each" decreased degree of angulation. This finding is irrespective of age, sex-type, BMI, and disc degeneration severity. In fact, in the context of these subject demographics, PI was a greater and more significant predictive factor. A critical PI value was noted; whereby, a PI of 42 or lower exhibited a fourfold increased risk in the development of these phenotypes, this 42 was determined from ROC analyses. Several studies have investigated spinopelvic alignment in patients having certain signs of disc degeneration, 38 lumbar disc displacement, 39 and spondylolisthesis. 17 However, and in particular, excluding cases of spondylolisthesis and having a more robust definition of disc degeneration in our study, we noted the novel finding that PI was significantly related to MCs, UDS as well as lumbar disc displacement and the combination of the three phenotypes. Furthermore, age and BMI were not associated with any of the pelvic parameters, which were in agreement with many previous reports. [40][41][42] Lumbar disc displacement subjects tended to have lower PI, SS, and LL, irrespective of demographics and degree of disc degeneration.
These findings are in accordance with some previous reports. 17,39 A lower PI with smaller SS may represent a more "vertical sacrum." Moreover, a straighter spine with a less pronounced LL may exhibit greater compressive forces on the discs that may fail to adapt, which may accelerate the degeneration of the disc; thereby, producing structural failure that can lead to lumbar disc displacement and other phenotypes ( Figure 4). 39 In our study, the overlap of the UDS with MCs is a substantial step forward in understanding the significance of these disc signs. 11,12 This pattern of association indicated that altered disc integrity brought upon by disc calcification can act as a focal stressor and result in a reactive response toward the adjacent endplate that may, in some individuals, lead to MCs. Interestingly, the most convincing association was seen in individuals having UDS, lumbar disc displacement and MCs with low PI. A lower PI would alter the transmission of T A B L E 1 Correlation analyses between subject demographics and cumulative degenerative disc score of L1-S1 on magnetic resonance imaging (MRI) to that of spinopelvic parameters Altered mechanical loading may increase levels of Collagen type X, 44 which implies a positive role in cartilage calcification. 45 The calcification process causes the gradual loss of cartilaginous endplate and reduction of disc nutritional pathways, further initiating degeneration. 46 The degenerated discs are closely related to reduced mechanical pressure and are characterized by increased porosity and thinning of the endplates, making them vulnerable to damage. 47 The damaged endplate removes the barrier between the discs and subchondral bone marrow, encouraging "cross-talk" and the cascade leading to MCs. 48

| Evolutionary etiology pathway of spinal phenotypes
Individuals with less structural adaptation to bipedalism, such as a lower PI and vertical pelvis, are prone to sub-optimal biomechanics and altered disc-endplate stressors ( Figure 4). 13 primates that possess a very short (anterior-posterior diameter) pelvic ring. 23 In our "earliest upright ancestors," bipedal walking was facilitated by some fundamental pelvic alterations. 49 The straight alignment with small spinal curves and small PI has been identified in Neanderthal lineage hominins 20,50 ; interestingly, a similar morphology was seen in almost 7.8% of healthy adult modern humans. 50 Despite evolution, variations in spinopelvic parameters in contemporary hominids may exist, contributing to their "personalized" evolutionary PI profile.
The role of genetics to the development of spinal degenerative phenotypes has been widely studied; however, heritability estimates vary from 26% to 77%. 9,51,52 Considering the potential possibility of genetic predisposition to PI, we can assume that genetics may play its part in the development of disc degenerative features by determining the PI and shape of the pelvis; thereby, degenerative changes of the disc and endplate region are induced rather than sole contribution from the disc degenerative process. As such, we propose an "Evolutionary Etiology" pathway with respect to clinically-relevant anterior column phenotypes (Figure 4).

| Clinical impact
Our findings further broaden the understanding of spine degenera-

| CONCLUSIONS
Though there are several previous studies who have seen the association of various spinal phenotypes on MRI and radiographic parameters of spinopelvic alignment 53,54 but our study is the "first" to identify that a low PI may "predict" the development of clinically-relevant spinal phenotypes of lumbar disc displacement, MCs and UDS, irrespective of subject demographics and disc degeneration severity.
A PI critical value of 42 or lower was found to have a fourfold increase in the development of these phenotypes. The association of PI with UDS is novel, stressing that altered biomechanics may induce disc calcification that can affect disc kinematics and the endplate.
Such calcification can create stress concentrations leading to the initiation of annular fissures and cracks in the endplate, increasing the risk F I G U R E 4 The Evolutionary Etiology pathway of the development of various spinal phenotypes of MCs. Our study notes evidence for the "Evolutionary Etiology" pathway related to anterior column spinal phenotypes that may lead to LBP/sciatica ( Figure 4). Understanding spinopelvic parameters and to potentially obtain one's "evolutionary spinopelvic imprint" with novel technology can lead to more personalized approaches to spine care and improved patient outcomes.

ACKNOWLEDGMENTS
This work was supported by grants from the Hong Kong Theme-Based Research Scheme (T12-708/12N) and the Hong Kong Research Grants Council (777111).