Effect of Different Types of Modic Changes on Segmental Motion and Disc Degeneration in Cervical Spine

Background: Few studies have reported the segmental motion characteristics of different types of Modic changes (MCs) in the cervical spine in contrast to the lumbar spine. Considering the close relationship between MCs and disc degeneration (DD), this study is designed to elucidate the relationship of MCs with DD and DD-based angular motion in the cervical spine. Methods: MRI of C2/3~C6/7 from 594 patients were reviewed and those with MCs were selected and evaluated. MCs were dened as type 0, I, II, and III, and the grade of DD was classied from A to E (recorded from 1 to 5 points) as previously reported. DD score of each segment (sDD), total sDD of the entire cervical spine (tDD), and VAS score (neck pain) of each patient with MC were also recorded. Cervical lordosis (CL), C7 slope (C7S), segmental angular motion (sROM) and total angular motion of the entire cervical spine (tROM) were calculated from X-ray images. Results: Totally 135 MC segments were evaluated. In the two groups of DD D and DD E, the sROM of MC I and III segments are smaller than that of MC 0 segments, and the sROM of MC II segments are similar to that of MC 0. In addition, MC I segments present smaller sROM than that of MC II in the group of DD E, and their sROM are larger than that of MC III in the group of DD D. No difference is found in tROM between the patients with MC I, II and III. Our results also show that MC I and III segments are associated with high sDD and tDD scores. Finally, the patients with MC I have larger CL and C7S than those with MC III, and there is no statistical difference in VAS scores of neck pain between the patients with MC I, II and III. Conclusions: In the cervical spine, both MC I and III indicate decreased segmental angular motion and high disc degeneration grade. Cervical MCs may not be a sign of unstable neck pain or an indication for interbody fusion.


Background
Modic changes (MCs), which were rstly reported by Modic et al. in 1988 [1], were de ned as signal intensity changes of vertebral endplates and subchondral bone observed from magnetic resonance imaging (MRI) of patients with spinal degenerative diseases [2]. Generally, MCs are classi ed into 3 types with different typically changed MRI signals: MC I, II, and III. Sometimes, authors also consider a normal endplate as MC type 0 or grade 0 [3,4]. In the recent few decades, several authors believe that MC I will cause segmental instability, which requires lumbar interbody fusion [5][6][7], and our previous study also has shown that MC II can increase translational segmental motion while MC III can reduce segmental mobility in the lumbar spine [3]. However, the effects of different types of MCs on the cervical segmental motion remain unclear. The results of the previous studies only indicate that the presence of MC affect segmental mobility, but the speci c role of MC I, II and III are still inconclusive [8,9]. Moreover, since segmental mobility is determined by both the endplate and intervertebral disc [2], it is necessary to exclude the in uence of the disc degeneration (DD) on the segmental mobility when discussing the kinematic characteristics of MCs [3]. Therefore, in this study, we hypothesize that MC I, II and III have different effects on segmental mobility on the basis of the same grade of DD, and they may have some relationship with DD grades.

Patient Population
Informed consent was obtained from all individual participants in the study, and the study was approved by the Ethical Committee of Huashan Hospital of Fudan University (No. 2017-056). 594 cervical degenerative patients (317 men and 277 women, 58.7y ± 10.99) were reviewed from September 2018 to August 2019. The inclusion criteria were de ned as patients who had neck pain with or without neurologic symptoms (i.e., radiculopathy or myelopathy). The exclusion criteria were trauma, infection, rheumatoid arthritis, spinal tumors, and history of cervical spine surgery. A total of 2975 segments from C2/3 to C6/C7 were retrospectively evaluated for all the patients in this study, and the patients with MC I, II and III were selected and evaluated for comparisons. Neck pain was record by using VAS (visual analogue scale) scores.

Image Assessment
All images were recorded by computer-based measurements, and all calculations were performed with Centricity Web 2.1 (GE Healthcare, Boston, Massachusetts, USA).
MRI was used for MC classi cation and DD grading. All MR images for MCs were reviewed twice by 2 blinded radiologists, and the intra-and inter-observer reliability of the ratings for MCs were assessed using the κ value and was assessed using 50 cases (250 intervertebral levels). MCs were classi ed into types 0, I, II, and III. Type 0, normal endplate; Type 1, hypointense signal on T1-weighted sequences and hyperintense signal on T2-weighted sequences; Type 2, hyperintense signal on T1 sequences and hyper-or iso-intense signal on T2 sequences; Type 3, hypointense signal on T1 and T2 sequences [3] (Fig. 1). Grading System for Cervical Intervertebral Disc Degeneration proposed by Hayashi et al [9] was used for DD (C2/3 to C6/7) grading (Table 1), and DD grade A to E was scored from 1 to 5 points. DD score of each segment (sDD) and total DD scores of the entire cervical spine from C2/3 to C6/7 (tDD) were both recorded. Angular motion range, which was measured from X-ray images in exion and extension positions, was selected for kinematical evaluation in this study ( Fig. 2a and b). It is not only because most of the authors consider angular motion change as the most important value affected by cervical MCs [8,10,11], but also due to the trivial translational change caused by cervical MCs when is compared to the angular motion change [8]. Brie y, vertebral bodies were marked at 4 points (anterior-inferior, anterior-superior, posterior-superior, and posterior-inferior) from C2 inferior endplate to C7 superior endplate, and segmental angular motion range (sROM) was calculated as the absolute value of the difference between the angle between adjacent vertebral bodies in exion and in extension in degrees. Similarly, total angular motion range (tROM) was recorded as the difference in C2-C7 angle (de ned as the angle formed by the inferior end plates of C2 and C7) during exion and extension ( Fig. 2a and b). In addition, cervical lordosis (CL) and C7 slope (C7S) were also measured for all the patients from X-ray images in neutral positions (Fig. 2c). CL was judged based on the C2-C7 Cobb angle, and C7S was de ned as the angle formed by the horizontal plane and the upper end plate of C7. C7S was used instead of T1 slope because of di culty in measuring T1 slope in many patients. The calculations of all patients were conducted and recorded by an experienced spine specialist twice and the nal values of angular motion, CL and C7S were the means of the two measurements. Pain killers were routinely used before patients underwent imaging studies to ensure clear and accurate images.

Statistical analysis
Statistical analyses were performed using SPSS (version 20; SPSS, Chicago, IL) computer software and values were expressed as mean ± standard deviation (SD). A P value of 0.05 was considered statistically signi cant. One-way ANOVA followed by LSD was applied to compare VAS, DD scores, cervical motions and sagittal parameters between different MC types; independent-sample t test was used to compare sROM between DD grade D and E. The reliability of diagnosis of MCs was assessed as follows: κ: 0-0.2 showed slight agreement, 0.21-0.4 fair agreement, 0.41-0.6 moderate agreement, 0.61-0.8 substantial agreement, and 0.81-1 excellent agreement. No MC was observed on the segment with normal disc (DD grade A); MC I and III were not found on the segments with DD grade B either. Several groups had relatively small numbers: DD B-MC II (2 segments), DD C-MC I (1 segment), DD C-MC III (1 segment), DD D-MC III (5 segments), DD E-MC I (9 segments) and DD E-MC III (6 segments) Table 2 . The inter-observer agreement of MC was substantial with a κ value of 0.744. The intra-observer agreement of MC of the two radiologists was excellent with a κ value of 0.878 and 0.855.

sROM and tROM of different types of MCs
Because only 1 MC I and 1 MC III segment was observed in the group of DD C (Table 2), statistical analysis was conducted for the MC segments in the group of DD D and DD E (Fig. 3).
In the group of DD E, similar trend is observed. MC 0 segments (5.32°±3.8) have larger sROM than those with MC I (2.33°±1.04, P 0.01) and III(1.30°±1.27, P 0.01); Segments with MC II (5.04°±2.79) have larger sROM than those with MC III (P 0.01). However, we also nd MC II segments present larger sROM than MC I segments (P=0.012). No difference is found in sROM between the segments with MC 0 and II(P=0.574), and between the segments with MC I and III(P=0.408).
Results of the comparison in sROM between the groups of DD D and DD E show that, the segments with MC 0 and I in the group DD D have signi cantly larger sROM than those in the group of DD E (P=0.035 and 0.01).
No statistical difference is found in tROM between the patients with MC I, II and III (P=0.527).

DD and VAS scores of the segments and the patients with different types of MCs
As is shown in Fig. 4a, the segments with MCs are associated with signi cantly higher sDD scores than those without MCs (MC 0) (all P 0.05); the segments with MC I present higher sDD scores than those with MC II (P=0.015); the segments with MC III present higher sDD scores than MC II though statistical difference is not observed (P=0.052). According to Fig. 4b, the patients with MC I and III have higher tDD scores than those with MC II (all P 0.05), but no signi cant difference is found between MC I and MC III (P=0.114).
As is shown in Fig. 4c, no statistical difference is observed in the VAS scores of neck pain between the patients with MC I, II and III (P=0.356).

Sagittal parameters of the patients with different types of MCs
According to Fig. 5, the patients with MC I have larger CL and C7S than those with MC III (P=0.016 and 0.046 respectively).

Discussion
In the lumbar spine, each type of MC has a different effect on the segmental mobility, and a speci c type of MC may be the cause of low back pain and an indication of fusion [3,[5][6][7]12]. As more and more scholars begin to pay attention to cervical spine MC [8,9], it is also of important clinical signi cance to explore the in uence of different types of cervical spine MCs on segmental mobility. Considering that there is a close relationship between MC and DD, and the segmental mobility is determined by both endplate and intervertebral disc degeneration [3,12], this study aims to explore the kinematic characteristics of cervical spine MC based on the grading of DD, and to analyze the relationship between MC type and intervertebral disc degeneration grade.

Prevalence and distribution of the cervical MCs
The prevalence of cervical MCs is lower than those of the lumbar spine, resulting in fewer related studies. According to our results, 20% patients have cervical MCs and most of them are MC II, which is similar to previous reports (16.9%~19.2%) [8,9,13].
The reason for why more MCs are found on C6/7 than C4/5 in our study is interesting. Several authors have reported the effects of MCs on cervical sagittal balance and many of them suggested that patients with MCs have larger cervical curvature, larger T1 slope or C7S [21][22][23]. Higher T1 slope or C7S needs more CL to keep horizontal balance, which may cause more stress and hypermobility at the bottom of the cervical spine (C5/6 and C6/7), resulting in unusual mechanical stress and more endplate degeneration [21].

MC I indicate small sROM and advanced DD grade.
MC I are considered as the early phase of endplate degeneration, although only a few authors clearly reported the effects of MC I on segmental motion, they are still believed to cause segmental instability in the lumbar spine [5,[24][25][26][27]. However, in this study, we nd that the segments with MC I present lower sROM than those with MC 0 in the group of DD D and DD E, which is actually consistent with the conclusions of Hayashi et al [9]. Moreover, the VAS scores of neck pain in the patients with MC I are similar to that of patients with MC II and III (Fig. 4c), indicating that MC I may not be the cause of unstable neck pain.
According to previous reports, the reason why cervical MC I can cause segmental motion reduction is because MC I are always accompanied by advanced DD grades that can signi cantly decrease segmental motion [9,19]. Our results also show that the sDD scores of MC I are signi cantly higher than those of MC II and similar to those of MC III (Fig.4a), more importantly, we even nd that between the segments with the same DD grade, the segments with MC I present much smaller sROM than those with MC 0 and MC II, indicating that MC I itself has the effect of reducing sROM (Fig.3a).
As the CL, C7S and tDD of the patients with MC I are relatively large (Fig. 4b and 5), the reason why MC I is always accompanied by high-grade DD can be explained by cervical spine aging, because the increase in CL and C7S is considered to be a compensatory change in the aging process of the cervical spine [28].

MC II indicate unchanged sROM and it is the transition phase of endplate degradation between MC I and III
Though MC II are accompanied by lower sDD, tDD scores than those of MC I and MC III (Fig. 4a and 4b), we still consider MC II as the middle transition phase between MC I and III for the following reasons: (1) During the degeneration process from DD A to DD D, though sROM keeps decreasing, the difference is not actual signi cant, but when disc degenerate from DD D to DD E, sROM reduction becomes statistically signi cant [19]. This theory is also supported by our result which is that, when disc degenerate from grade D to E, sROM of MC 0 and I decrease signi cantly (Fig. 3a). As a result, in the segments with advanced DD (especially grade E), sROM are primarily determined by the status of endplate degeneration instead of the disc, which has been proven by us in the lumbar spine [3]. In this study, when MC I turn into MC II in DD E, sROM increase signi cantly (Fig. 3a), which is the same as that of MC II in the lumbar spine, that is, MC II is a transitional phase between MC I and III that can increase segmental mobility [3]; (2) Both MC I and II are unstable and they can convert to each other [2], as a result, though MC I have higher DD scores than those of MC II in our study, several of them may be actually converted from MC II, indicating that they may have longer degeneration process than expected. In addition, many segments with MC I-II mixed type are found in our study, which can also support this theory. 4. MC III indicate the nal phase of endplate degeneration process, which can signi cantly reduce sROM MC III is the nal and stable phase of endplate degeneration process and is wildly agreed to signi cantly decrease segmental angular and translational motion in the lumbar spine [2,3]. According to our results, similar results are found in the cervical spine, that is, MC III indicate small sROM and high-level DD grade. However, the patients with MC III have signi cantly smaller CL and C7S than those with MC I, indicating that high-grade DD may be due to the straight cervical spine rather than cervical spine aging [28].

Effect of MCs on tROM
The tROM of patients with MCs are affected by many factors. In our study, neck pain degree and disc degeneration level of the entire cervical spine are considered the most important.
Our results show that no statistical difference in tROM is found between the patients with MC I, II and III, and they are all smaller than the normal range [29]. Though pain killers were routinely used before patients underwent imaging studies, the effects of pain on decreasing cervical motion cannot be completely ruled out which may nally cause the similarity of tROM between the patients with MC I, II and III [ Fig. 4c]. In addition, we nd that the sDD scores of the segments with MCs are higher than that of the segments without MCs, and MC I and III have similar high sDD and tDD scores [ Fig. 4a and b], which indicate that the overall high degree of degeneration of the cervical spine limits the tROM of the patients with MCs.

Conclusions
In the cervical spine, the segments with MC I and III present smaller angular motion range than those with MC II and MC 0, though no difference in total cervical motion range are found between the patients with MC I, II and III. Both MC I and III indicate high disc degeneration degree but they are probably due to cervical aging and small cervical lordosis respectively. Finally, cervical MCs may not be a sign of unstable neck pain or an indication for interbody fusion.  sROM was de ned as the angle difference formed between yellow lines of each segment from extension(a) and exion(b), and tROM was de ned as the angle difference between C2-7 angle (α and β) difference from exion (a) and extension (b). CL and C7S were indicated as γ and θ in the neutral X-ray image (c). sDD scores(a), tDD scores(b) and VAS scores (Neck pain, c) for the patients with MC I, II and III. a: * indicates that the sDD score of MC 0 is signi cantly lower than that of any other MC type (P 0.01); # indicates that the sDD score of MC I is signi cantly higher than that of MC II (P=0.015); b: * indicates the tDD score of MC II is signi cantly lower than those of MC I and III (all P 0.05); c: No statistical difference is observed in the VAS scores of neck pain between the patients with MC I, II and III (P=0.356).