Axial loading during MRI induces significant T2 value changes in vertebral endplates—a feasibility study on patients with low back pain

Low back pain (LBP) is an endemic condition with high impact for the individuals affected but also with tremendous socio-economic consequences [1]. LBP is closely associated with both degeneration of the intervertebral disc (IVD) and with vertebral endplate (EP) changes [2‐5]. The exact etiological factors for LBP are however still not elucidated and neither are the exact mechanisms initiating the degenerative cascade of the IVD. Further, the lack of a true “gold standard” diagnostic tool that unequivocally points out which spinal structures cause LBP prevents optimal success for used and suggested therapeutic strategies [6, 7].

The avascular IVD relies on passage through the EP for nutrient supply and metabolite removal, primarily regulated by diffusion and convection via the vertebral EP [3, 8]. Dynamic contrast-enhanced MRI (DCE-MRI) has revealed that EPs with signs of degeneration have altered contrast enhancement pattern compared to non-degenerated EPs [9‐11]. Hence, structural EP changes, like fissures or sclerosis, may alter its permeability and play a crucial role in the progression of IVD degeneration [2, 3, 8, 12]. The exact relationship between compromised EP function and IVD degeneration is not fully understood. To deepen the understanding of the pathophysiology behind IVD degeneration and, in extension, also the understanding of LBP, non-invasive diagnostic tools with ability to in detail characterize early biochemical EP changes are desirable.

Quantitative MRI methods, sensitive for alterations in composition between macromolecules, collagen, and water, reliably reflect changes in biochemical composition and structural integrity in degenerated IVDs [13‐16]. In spite of the ability of these methods to reveal early signs of IVD degeneration and thereby potentially find the window when biological targeted therapy might be favorable, quantitative MRI methods are still not used routinely in a clinical setting. Quantitative MRI studies of the EPs are, in contrast to the above discussed IVD studies, limited but have been reported to also reveal subtle deterioration of biochemical EP composition [17‐19].

Mechanical loading of the spine has been shown to affect quantitative MRI parameters in studies investigating the IVDs [15, 16, 20‐22]; however, similar studies of the EPs are lacking. A recent review regarding the immediate lumbar response to various loading conditions concluded that there is a gap in knowledge on how loading affects spinal imaging [23]. Imaging of the EP in a loaded state versus in an unloaded state might reveal differences in EP characteristics that may reflect altered EP functionality. It was hypothesized that by comparing quantitative MRI of EPs performed with conventional supine MRI (unloaded MRI) with axial loading during MRI (alMRI), dynamical properties of the EP can be displayed. The aim of the present study was to investigate if alMRI instantaneously can induce EP changes, measured with quantitative MRI.

This feasibility study, comparing T2 values of EPZs obtained by unloaded MRI and alMRI, showed that significantly higher T2 values were induced in the loading state. In abnormal EPZs, no significant induced difference was shown at a group level as opposed to in normal EPZs. Further, the known morphological and functional differences, like for example different stiffness and ability to resist load [26], between superior and inferior EPs were clearly displayed quantitatively.

Spinal load has been shown to affect quantitative MRI parameters of the IVD, which may provide detailed information of the IVD characteristics and its behavioral patterns [15, 16, 20‐22]. In this study, it was hypothesized that alMRI would also affect quantitative EP values. The present study seems to confirm this hypothesis by displaying significant induced changes in normal EPZs, however, not in abnormal EPZs. Hence, combining T2 mapping with alMRI adds another dimension of the EPZ characteristics compared with information obtained with solely unloaded MRI and may, thus, provide a clinical feasible, non-invasive method with potential to reveal biochemical behavioral patterns of the EPZs.

The cartilaginous EP consists of approximately 50–60% water [18], with a high degree of transport in the well-hydrated matrix. With degeneration, the permeability of the EP decreases due to occlusion of the vascular buds and increased ratio of calcification and collagen content [18, 27, 28] which restrict the nutritional transport over the EP [8, 9, 27]. EP damage can however paradoxically also result in increased permeability with loss of essential nutrients and enzymes [9]. For example, Rajesekaran et al. performed serial post contrast-enhanced MRI sequences in 54 individuals and reported either enhanced diffusion or delayed such as in EPs with degenerative signs (fissures, Schmorls, etc.), showing that type of EP injury results in different functional behavior [8, 9, 27]. In general, higher T2 values are interpreted as high hydration. However, this is a simplified model since T2 values reflect not only the composition between proteoglycans, collagen, and water but also depend on tissue anisotropy, i.e., matrix organization [14, 15, 21, 22, 29, 30]. Why certain EPZs respond to load in different ways are obviously complex and probably multifactorial. The differences between abnormal and normal EPZs seen in the current study show, however, that the EPZs display different characteristics, and the lack of response from alMRI in abnormal EPZs may be a reflection of impaired integrity.

Despite the fact that alMRI induced an increase in the EPZ T2 value on a group level, on an individual EPZ-level, a large variation in the induced changes were seen, from 187% increase to 55% decrease (Fig. 3). In abnormal EPZs, the range of induced changes was not as pronounced (89 and 26% respectively). This wide range of T2 values may in fact reflect large differences in the characteristics between individual EPZs, and the smaller range of values in abnormal EPZs may reflect a lower dynamical variation due to impaired EP function. Future studies are warranted to strengthen these results and to investigate the pathophysiological mechanisms behind this divergent behavior between normal and abnormal EPZs.

Mechanical stimuli can exert both anabolic and catabolic effects on the spine [17, 29‐31]. The effect that alMRI exert on each EP is not known, and it cannot be excluded that this instantaneous load do not impair EP transport at some levels. However, since alMRI aims to simulate the load under upright position, alMRI have been assumed to exert load within a physiological beneficial range [32]. Other potential factors that might influence the various alMRI-induced EPZ responses are the level of the EPZ in the lumbar spine, the lordosis and IVD angle induced, and the status of the EPZ and IVD. Recently, alMRI was shown to induce regional alterations in IVD T2 values depending on IVD level [20]. The present feasibility study was not powered to investigate such relations. Induced differences however seemed to decline in caudal direction of the lumbar spine, which speculatively might be explained by degenerative changes being more common caudally. Out of 16 degenerated IVDs, 12 were localized at L4-S1 level as were nine of the 12 abnormal EPZs.

The absolute T2 values obtained in the present study were slightly lower (22–74 ms unloaded MRI) than previously reported by Delucca et al. (60–100 ms) [28], a discrepancy likely due to either the previous cadaver study design or that outlining of only the cartilaginous EP was not possible in the current in vivo study. The significant difference in absolute T2 values between superior and inferior EPZs likely reflects what has been shown in biomechanical in vitro studies and animal in vivo with the inferior EP being stiffer, resist load better and with different transport kinetics compared with the superior EP [26, 30]. This phenomenon is reflected in clinical situations where for example superior EPs being more prone to burst fractures [33]. Also human in vivo studies show that contrast enhancement differ between superior and inferior vertebral EPs [10, 11]. To our knowledge, this is the first human in vivo study displaying quantitative differences, which may reflect morphological differences, between the superior and inferior EPZs. Thus, these results constitute ground data for future studies.

Data from a single time point cannot fully characterize the complex kinetics occurring in the EP and why comparing induced changes between unloaded MRI and alMRI provides another dimension, indicating dynamic EP characteristics. Such has been reported promising in the IVD [20], and this study shows that alMRI is feasible to induce changes also in the less hydrated EP. Despite large differences at individual levels, significant differences were found at group level and in addition with a strong test-retest repeatability, making the results of this feasibility study worth studying in large-scale studies.

The distinct responses induced with alMRI makes the combination of quantitative MRI and alMRI a promising method to assess behavioral EP characteristics clinically. In addition, alMRI offers a method to image the spine in a position inducing concordant pain in LBP patients [34]. To elucidate if the differences in EPZ characteristics between unloaded MRI and alMRI represent altered diffusion/perfusion, future studies are encouraged to include diffusion-weighted sequences. If this method can reveal also early signs of impaired EP integrity, before manifest IVD degeneration appears, it would provide important information that ultimately may be used for targeted therapy. To evaluate clinical application of the method, work has been initiated to elucidate if this method can depict pain predictors by comparing if the induced EP T2 value changes vary between a large-scale LBP cohort and healthy controls.

The small sample size is a limitation. In a previous alMRI study that investigated IVD changes [20] (50 IVDs), significant changes were reported, why the current sample size was considered appropriate to evaluate if it is feasible to induce EPZ T2 value changes with alMRI. However, considering the limited number (n = 12) of abnormal EPZs, it cannot be excluded that this study is underpowered for evaluation of behavioral pattern regarding abnormal EPZs. The so-called “magic angle effect,” appearing at approximately 55° from the main magnetic field [18], must be considered. However, the effect in the current study is regarded as non-existing at L1-L4 and only minor at L5-S1 since none of the EPZs at level L1-L4 were imaged between 50° and 60° with unloaded MRI and only one L5 EPZ was imaged at 58° and four S1 EPZs imaged at 53°–54°. With alMRI only, three EPZs were imaged between 50° and 60°.

Contribution of minor signal from adjacent tissue, for example the IVD, cannot be completely excluded due to the image resolution obtained with T2 mapping. However, this potential partial volume effect ought to be minor on the whole segmented EPZ volume and in addition negligible regarding induced EPZ changes since such effect ought to be similar on unloaded MRI and alMRI.

This study shows, for the first time, that alMRI induce changes in human EPZs characteristics in vivo. However, when analyzed on group level, the T2 value significantly increased in normal EPZs, with lack of such in abnormal EPZs. Combining T2 mapping with alMRI provides a clinical feasible, non-invasive method with potential to reveal biochemical behavioral EP patterns, thus adding another dimension of the EPs characteristics compared with information obtained with solely unloaded MRI. Thus, this study opens up an area for future research since imaging during loading conditions might increase the diagnostic specificity, providing new biomarkers, within spine imaging.