Up to 84% of the adult population may experience lower back pain (LBP) at some point in their lifetime (Walker, 2000), which has resulted in an estimated aggregate healthcare cost of £3.2 billion to the United States alone (Zemedikun et al., 2024). Whilst a multitude of conservative (analgesia, physical therapy, bracing) and surgical (fusion, decompression) treatment options exist, barriers to the management of LBP arise from the multi-faceted nature of LBP pathogenesis and the difficulty in mechanistic diagnoses (Allegri et al., 2016).

Excessive tissue strains of the intervertebral discs (IVDs) and motion of the vertebral bodies are thought to hallmark LBP associated with disc degeneration (Murata et al., 1994), stenosis (Lee et al., 2020), or spondylolisthesis (Dickey et al., 2002; Panjabi, 2003). Spinal instability is defined as abnormal intersegmental motion in response to physiological loading (Pope and Panjabi, 1985), disrupting the ability to protect neurological structures from irritation. For example, excessive pre-operative intersegmental motion has been hypothesised to contribute to pseudarthrosis (Heggeness et al., 1993) and other post-operative complications such as recurrent stenosis (Johnsson et al., 1989), or IVD herniation after decompression surgery (Takenaka et al., 2016).

It is currently challenging to predict which patients would benefit most from conservative therapy, and if indicated for surgical management, whether the provision of stabilisation alongside decompression is required (Simmonds et al., 2015; Dang et al., 2020). This may be due to the knowledge that instability is traditionally assessed using standing radiographs in flexion-extension positions (Dupuis et al., 1985; Murata et al., 1994; Pitkänen et al., 2002; Iguchi et al., 2004). Whilst this technique has contributed to initial knowledge of changes to spinal anatomy and instability (Iguchi et al., 2004; Leone et al., 2007), the movement can be limited by patient compliance due to pain generated by the movement, and measurements are restricted to planar translations.

Therefore, the ability to characterise 3-dimensional spinal motion would enable clinicians to make informed decisions regarding the degree of intervention needed. Supine MRI is routinely acquired as part of pre-operative evaluation or assessment of LBP, enabling surgeons or practitioners to readily obtain metrics such as Pfirrmann grade (Pfirrmann et al., 2001), IVD height, and sagittal alignment. Despite this, such images can only offer a static snapshot of patient spinal anatomy, with little knowledge surrounding the biomechanical implications of anatomical variance across patients. In parallel, the quantification of in vivo IVD strains and 3D vertebral body translations using MRI under supine extension has recently been validated using digital volume correlation (DVC) (Tavana et al., 2023). Yet this information remains challenging and time-consuming to acquire in a clinical environment. Therefore, it is of clinical interest to be able to predict vertebral motion and IVD deformation from conventional pre-operative metrics, especially in movements such as spinal extension, of which is routinely used for various physiotherapy exercises (Erhard et al., 1994; Mann et al., 2023; Park et al., 2024) to theoretically reduce stress in the posterior annulus (Edmondston et al., 2000).

Prior to establishing such a relationship, it is first necessary to understand the baseline interaction between lumbar vertebral translations, IVD strains, and clinically-acquired parameters, i.e., identifying the association and its directionality within a healthy cohort (presenting no LBP). The identification of variables that are associated with spinal motion and IVD deformation in a pain-free population may form a point of reference for future studies investigating cohorts with spinal pathology and presenting LBP, and may contribute towards defining a mechanism whereby spinal morphology and instability are related.

Therefore, the aim of this study was to determine whether IVD strains and segmental kinematics during extension loading could be predicted with conventional clinical measures of spinal geometry and IVD morphology in a healthy cohort.

The present study demonstrates the existence of relationships between 3-dimensional internal IVD strains, segmental kinematics, and anatomical measures that can be made from standard clinical images within a cohort presenting no LBP. Understanding how sagittal alignment and IVD morphology affect spinal motion in such a population is necessary, as the parameters flagged in this study (e.g., segmental lordosis, sacral angle) may be indicative of instability outside of the healthy range reported here, thus could be used to optimise the choices of covariates when investigating a pathological population.

IVD strains and notable motion patterns were only associated with clinically measured variables at the L4-L5 and L5-S1 segment. This finding bears clinical relevance, as the L4-L5 and L5-S1 segments are the most common to require surgical intervention, with the highest incidence of pathological events such as herniation (Jordan et al., 2009), spondylolisthesis (Hey et al., 2024), and stenosis (Tomkins-Lane et al., 2014). Prior studies have highlighted the L5-S1 segment for its tendency to produce anterolisthesis in functional myelography, whilst other segments produce negligible motion (Johnsson et al., 1989). However, this finding may in part be related to the resolution constraints of the DVC, where higher lumbar levels generated motion that did not surpass the error threshold to a sufficient degree.

In L4-L5 and L5-S1, the presence of non-degenerated (Pfirrmann 2) IVDs were significantly associated with the increase in mean and peak minimum principal IVD strains during extension (Table 3). Disc degeneration has been shown to be associated with anterior migration of the nucleus in response to extension moments (Tsantrizos et al., 2005), which could alleviate posterior compressive stresses. Additionally, O’Connell et al. (2011) observed that in extension, non-degenerated IVDs were more likely to exhibit higher radial strains in the posterior annulus than degenerated IVDs using high-resolution MRI (O’Connell et al., 2011); this finding at the apparent scale could manifest as greater compression at the whole-IVD level observed in the present study, where individual lamellae cannot be resolved.

In the L4-L5 segment, lower anterior-posterior IVD height ratio and reduced segmental lordosis were associated with higher minimum principal IVD strains (Figure 4A). This suggests that those with reduced lordosis, or “flat back syndrome” (Lu and Chou, 2007), exhibit high compressive strains during extension, likely in the posterior annulus. Given that maximum principal strains – likely prominent in the anterior annulus – were not also significantly increased when angulation of the segment decreased, it could suggest a more anterior pivot point in the IVDs of those with straighter spines as a compensatory response to extension loading. It is hypothesised that some patients presenting LBP with reduced segmental lordosis do not exhibit this compensation response, and as a result may be less suited to extension-based physiotherapy. In some instances, a lumbar roll may be prescribed in supine or sitting movements (Erhard et al., 1994; Takasaki et al., 2018). Therefore, lumbar kinematic data upon the application of such equipment could be used as a benchmark in future investigations into patient suitability of extension-based physiotherapy.

Previous work has demonstrated that the variation in IVD strains and vertebral translations between individuals is significantly smaller than the variation between lumbar levels (Tavana et al., 2023), suggesting that a “healthy” range of strain and translation exists for individuals reporting no LBP. Spinal alignment metrics during extension loading have become important in predicting the risk of post-operative complications (Takenaka et al., 2016). In the present study, lordosis angle and anterior-posterior IVD height ratio was negatively associated with greater anterolisthesis at the L4-L5 segment (Figure 5) – in other words, straighter segments indicated greater forward “slippage” of the L4 vertebra. However, it has been shown that reduced lordotic angle paired with increased retrolisthesis of the superior vertebra significantly increased the risk of herniation after decompression surgery (Takenaka et al., 2016). The present study, taken together with Takenaka et al. (2016) findings, demonstrates that particular combinations of geometrical measures and identifiable movement patterns may be characteristic of the development of spinal pathology in certain cohorts. Thus, further work is needed to characterise the 3-dimensional internal IVD strains and vertebral kinematics during extension loading in non-healthy cohorts.

Clinically, anterolisthesis or retrolisthesis of one vertebra relative to another, accompanied by the loss of lumbar lordosis, is characteristic of degeneration-related LBP (Barrey et al., 2007). However, it is unclear as to whether lumbar lordosis or segmental angle contributes to the development of pathological events such as spondylolisthesis (Berlemann et al., 1999; Chen and Wei, 2009; Abu-Leil et al., 2016). For instance, unlike the present study, Lee et al. (2021) found no relation between IVD angle and instability in patients with spondylolisthesis (Lee et al., 2021). This highlights that relationships between clinical parameters and vertebral motion may be distinct in pathological versus control cohorts, although the use of 2-dimensional measures and the diagnosis of instability through flexion motions in Lee et al. (2021) may in part explain the discrepancy. Future work capturing MRIs in the flexed position is required for a comprehensive evaluation of the how spinal geometry can influence lumbar kinematics.

The inability to capture motion from extension to flexion may explain the somewhat surprising finding that disc degeneration was not associated with lumbar anterolisthesis. It is likely that the extension alone is insufficient to capture this motion (Boden and Wiesel, 1990), and furthermore, due to restabilisation of the joint as degeneration progresses (Yong Hing and Kirkaldy Willis, 1983), any subluxation observed on static radiographs may not correspond to the motion observed dynamically (Boden and Wiesel, 1990).

Sacral angle displayed significant relationships with both IVD minimum principal strain and anterolisthesis of the L5-S1 segment only (Tables 3, 4). In particular, a greater sacral angle was associated with higher compressive IVD strains, which is counterintuitive to reduced segmental lordosis angle being a parallel predictor. However, the relationship between sacral angle and lumbar lordosis is not strictly positive nor linear, and largely determined by the lordosis inflection point and apex (Roussouly et al., 2005). Based on this, participants in this study with high compressive strains at the L5-S1 IVD are more likely to exhibit “type 4” lordosis, defined by a more cranial lordosis apex (e.g., at the L3 vertebra) (Roussouly et al., 2005). It is interesting to note that spinal stenosis is linked to “type 4” lordosis (Roussouly et al., 2005). It could be hypothesised that flexion-extension movements with this posture could generate excessive IVD strains in the lower lumbar region, which over time could predispose the IVD to annular delamination and outwards bulging (Tavana et al., 2021).

A limitation of this study is that only 15 healthy participants were recruited, which means caution should be applied when interpreting findings in relation to spinal pathologies and clinical outcomes. However, as a proof-of-concept study, the small sample was sufficient to understand whether such correlations could be drawn between conventional clinical measures and biomechanical quantities, potentially landmarking a shift in the way pre-operative metrics are utilised and interpreted. Thus, future work which focusses on cohort expansion and the stratification of patients undergoing intervention for LBP-related pathologies is warranted to eventually lead to improved clinical outcomes.

Additionally, supine positioning causes altered sagittal alignment compared to standing (Tian et al., 2025). These associations are therefore only applicable for supine geometrical measures. However, it been demonstrated that anterior sagittal translation is significantly greater in weight-bearing standing radiographs relative to supine (Lowe et al., 1976), suggesting that effect sizes may be under-estimated in this study. Another limitation was that translation components were resolved relative to the global axis, which meant that it was not possible to report anterior-posterior and cranial-caudal translations respective to the orientation of the vertebral body. Again, vertebral translations may have therefore been under-estimated with the local axes orthogonal to the global co-ordinate system. However, the separation of regression models into respective lumbar levels may have somewhat standardised vertebral inclination and thus minimised this confounding effect.

Lastly, in some participants, Vx displacements did not exceed 10 times that of the precision estimate (Dall’Ara et al., 2017). Particular caution should be applied to the anterolisthesis cases, where propagation of error would see the precision estimate to be 0.330 mm, and that meaningful displacements were not observed at the L5 vertebra (Figure 3C). However, the translational value of having access to physiological measurements of internal kinematics may supersede error criteria which are typically based on in vitro data. Human in vivo imaging protocols are inherently limited in their ability to produce errors on a similar scale (an issue also faced by Oravec et al. (2025)), and thus, the spatial resolution of the displacement field should be compromised accordingly, but not so much that the subset size exceeds the dimensions of the tissue being studied (Dall’Ara et al., 2017). In this instance, the subset size was optimal at 16 voxels (4.64 mm) as larger subsets would not be able to discern axial strains within the IVD, where the average disc height recorded in this cohort was approximately double the subset size (9.6 mm).

This study demonstrates that measures of sagittal angle such as segmental lordosis and anterior: posterior IVD height ratio are positively associated with IVD compressive strains and anterior translation of the vertebra cranial to the segment. These associations were restricted to the lower lumbar levels. Results suggest that in the future, it may be possible to infer patient-specific spinal movement patterns from static radiographic measurements, which may assist in the streamlining of clinical decision making to improve quality of care in patients with LBP.