Discussion
The findings of this study show significant differences in 3D bone micro-architectural parameters as measured by micro-CT and in bone remodelling indices as measured by histology among Modic types. It can be postulated, as suggested by Modic et al. [
1,
2] that different Modic types represent different stages of the same pathological process, linked to the adjacent disc.
Significant increases in BV/TV and Tb.Th were found in Modic 3 compared to Modic 1 and 2, whereas the other investigated micro-architectural parameters, Tb.Sp, Tb.N, and SMI were not statistically different amongst the three Modic types. These results suggest that the increase in BV/TV seen in Modic type 3 was mainly due to thickening of the trabeculae. For SMI a non-significant decrease was found, which is a trend towards a more plate-like structure in Modic 3, compared to Modic 1 and 2 [
23,
25]. This lack of statistical significance could be simply due to the small number of Modic 3 samples.
Histological analysis showed significant increase in the bone formation to bone erosion ratio (OS/ES) in Modic 3 samples, compared to Modic 1 and 2. This finding is consistent with both reduced bone resorption in Modic 3 samples compared to Modic 1, indicated by the significantly decreased ES/BS, as well as with the increased bone formation, with OS/BS significantly higher in both Modic 3 and in Modic 1 compared to 2.
Taken together, Modic 1 and 2 biopsies differed mainly in bone remodelling indices. On the other hand, bone samples in Modic 3 showed significant differences in both micro-architecture and also remodelling indices, compared to the other two Modic types. Modic 3 changes were consistent with a more stable sclerotic phase of the process, showing significantly higher BV/TV and thicker trabeculae compared to Modic 1 and 2, with reduced bone resorption and increased bone formation activity.
In our series, no obvious difference was observed in the overall size of different types of Modic changes as seen on MRI. Specimens were taken of same length due to the standard size of the biopsy needle (8G Jamshidi needle), with the biopsy always contained within the Modic lesions as confirmed by fluoroscopy.
Modic et al. [
1] in 1988 postulated that the different Modic types may represent different stages of the same pathological process, linked to the adjacent disc, reflecting a spectrum of marrow changes: Modic 1 changes indicating an ongoing degenerative process, disruption and fissuring of the endplates, with marrow replacement by small blood vessels; Modic 2 indicating an ongoing degenerative process, disruption of endplates, with bone marrow replaced by yellow marrow (fat); and Modic 3 changes reflecting bone sclerosis seen on plain radiographs, an indication of woven bone in the vertebral body, rather than a reflection of the marrow element.
The present study done on excised bone biopsies, examining bone micro-architecture parameters and remodelling indices, supports the hypotheses of different pathological stages postulated by Modic et al.. Modic 1 associated changes appeared to be consistent with a more active phase of the pathology, showing high bone formation ratio and erosion ratio. Modic 2 biopsies were more suggestive of a reduced bone formation/remodelling stage, as indicated by the remodelling indices. Modic 3 changes were consistent with the bone sclerosis assumed by Modic et al. with significantly higher BV/TV and thicker trabeculae, and high bone formation/erosion ratio, as well as reduced resorption. This might also explain the characteristic appearance of Modic type 3 in MRI images (low T1 and low T2 signals), which is likely a reflection of decreased or absent marrow and/or fat in these MRI images, due to the increased amount of bone in Modic 3 as found by micro-CT in this study, indicated by the increased BV/TV (up to +70 %) and Tb.Th (up to +57 %).
In a previous study, Kuisma et al. [
34] compared clinical MRI and clinical CT images taken from patients exhibiting Modic changes. Their study showed that in Modic type 3, the low T1 and T2 signals seen in MRI, corresponded to endplate bone sclerosis as indicated in the CT images, with increased Hounsfield units (HU) (increased bone mineral density), which is in agreement with our results using micro-CT. It is worth mentioning that the spatial resolution of clinical CT, as used in their study (slice thickness 1.25 mm and 0.625 mm interval, in-plane resolution not indicated) enables measurement of mineralized tissue in terms of HU, but does not allow characterization of the trabecular bone micro-architecture [
26], in particular in the vertebrae, where the trabeculae can be as thin as 80 µm [
27]. In other words, it does not allow to identify whether the changes in HU, were due to more, or thicker, trabeculae, or same number of trabeculae but more mineralized, or a combination of the above. Moreover, no bone biopsies were taken in that study, to assess the bone micro-architecture and bone remodelling. Micro-CT is a relatively recent, non-destructive high-resolution imaging technique, validated by 2D histology and physical 3D phantoms [
18,
25,
26,
28], capable of 2D and 3D quantitative characterisation of excised bone segments at a detection level not achievable with conventional MRI or CT scanners. It has a spatial resolution (10 µm range), which is 10–100 times better than conventional clinical CT or MRI scanners, and can accurately differentiate between the trabecular and cortical bone compartments and non-osseous tissues, as opposed to conventional clinical CT and MRI imaging [
14,
26,
27,
29]. Although it has been used in spine studies previously, it has never been used to study in Modic changes before [
30‐
32]. It is typically used for ex vivo scanning of excised human and animal bones [
14,
29] or in vivo scanning of small animals, at micrometer resolution [
33]. For clinical CT scans on human patients in vivo, the improvement in spatial resolution (decrease of pixel size) is currently limited by the corresponding increase in radiation dose [
28]. However, within the restrictions of the radiation dose, it can be expected that thanks to technological developments and improvements in spatial resolution, clinical CT systems might be increasingly used in future for studying Modic changes in vivo, as done by Kuisma et al. [
34]. Recently, pre-clinical High-Resolution peripheral Quantitative CT (HR-pQCT) systems have become available, for in vivo scanning of peripheral parts of the human body, such as the wrist and ankle [
35] with a pixel size (82 μm) approximating that of micro-CT.
In a previous study, Modic et al. examined excised endplate tissue using histological analysis, however, the analysis was only qualitative (descriptive), not quantitative, limited to Modic type 1 and type 2 patients with only three patients per group, and with no Modic type 3 patients [
1]. In the present study, excised bone biopsies from 40 patients were examined, with all the three Modic groups represented. To the best of the author’s knowledge, this is the first report on bone micro-architecture and remodelling indices in Modic changes done on excised bone biopsies. Furthermore, the specimens were subjected to micro-CT analysis, which enabled visualisation and quantification of the bone micro-architecture in 3D, as well as to histology, which enabled an assessment of the remodelling indices. This represents both the novelty as also the strength of this study.
A limitation of this study was the sample size (40 patients in total), in particular for the occurrence in the Modic 3 group (6 patients), compared to the Modic 1 group (9 patients) and Modic 2 groups (25 patients). The patients examined were those eligible for this study undergoing surgery at the hospital and exhibiting Modic changes on preoperative MRI, over a 24 months period. Despite the sample size, significant differences between the groups in micro-architecture and remodelling indices were detected. Moreover, the overall prevalence rate of patients in each group was found to be consistent with that reported in the literature, with Modic 2 being the most common followed by Modic 1, whereas Modic 3 appears to be the least common [
3,
36,
37]. Some studies even reported no Modic type 3 patients in their series [
7,
38].
In longitudinal studies reported in the literature, spontaneous conversions of Modic types have been found within a time span of 1-year, with Modic 1 to 2 being he most common, including conversions from type 2 to 1, and from type 1 to normal [
3]. Modic type 1 represents the major transition point from normality, and as such, appears to be a main focus of research [
3]. The exact mechanism causing Modic changes is still not clear; it could be a response to early inflammatory stages of the degenerating disc [
9]. In the present study, Modic 1 samples had the highest bone turnover with increased OS/BS and ES/BS, possibly due to an inflammatory reaction increasing both osteoblastic and osteoclastic activity. A systematic review of the literature and of mechanisms of conversion between Modic changes was done by Zhang et al. [
3]. According to current literature, the two major causes for the pathogenesis of Modic changes are biomechanical and biochemical [
3]. With regards to the biomechanical cause, micro-fractures and fissures in the endplates have been reported, in relation to human disc degeneration [
1]. As such, Modic changes might be due to altered mechanical stress, resulting in histological changes, leading to signal intensity changes on MRI, namely the Modic changes. Biochemical causes on the other hand, imply that Modic changes may result from an inflammatory reaction to toxic mediators or infection from the degenerating disc [
39]. An intervertebral disc herniation could be the entry point of bacteria into the disc, and the Modic changes could be therefore the result of anaerobic bacteria leading to oedema and inflammation surrounding the extruded intra-vertebral nuclear material [
9,
40]. A recent study has suggested antibiotic treatment as an effective option for a specific subgroup of patients with Modic type 1 changes [
9]. Significant controversy has followed, with a number of authors raising concerns about the methodology and conclusion of that study [
41‐
52]. Within the restrictions of our specimen size, and our study being a bone micro-architecture and remodelling study, not aimed at investigating bacterial infection, our histological slides of the bone biopsies (H&E stained) did not show associated oedema and no cellular infiltrate to indicate overt inflammation. However, the material analysed in the present study included bone biopsies only, and not intervertebral disc which may contain bacteria and show inflammation. This remains a controversial topic, in need of further research as suggested by numerous letters to this journal [
41‐
52]. Hence, evidence supporting the biomechanical and biochemical hypotheses is still not sufficient, and it remains unclear how they may play a synergistic role in the development of Modic changes in the human body [
3].
The trabecular BV/TV in this study was not significantly related to the age of the patients (
R
2 = 0.04,
p = 0.24, age range 35–75 years). This finding in Modic changes is different from that for normal (non-pathological) vertebral bone reported in the literature, where, in bone extracted from similar anatomical locations within the vertebrae and examined with histology and micro-CT, the trabecular BV/TV shows significant decrease with age (e.g.,
R
2 = 0.61,
p < 0.05, age range 23–95 years; or
R
2 = 0.61,
p < 0.05, age range 57–98 years) [
30,
53]. The absence of correlation with age in Modic changes in our study is a novel finding, and is likely linked to the increased remodelling activity found in the bone. It is worth noting that this result is similar to observations in bones of patients with osteoarthritis (e.g., in the proximal femur), for which typical age-related changes found in normal patients are altered, showing no age-dependent decrease in BV/TV [
17,
54,
55]. We acknowledge that having a comparison with specimens from a control group exhibiting no Modic changes in our study may have strengthened our findings.
In conclusion, this is the first quantitative study done on excised bone biopsies, examining quantitative bone histomorphometric changes among the three different Modic types, using micro-CT and histology. The findings have shown significant differences in bone micro-architectural parameters and bone remodelling indices among Modic types. This suggests that different Modic types might represent different stages of the same pathological process, linked to the adjacent disc. Modic 1 specimens have highest bone turnover, with increased OS/BS and ES/BS, possibly due to an inflammatory reaction; Modic 2 changes are consistent with reduced bone formation/remodelling; Modic 3 changes suggest a more stable sclerotic phase, with significantly increased bone volume fraction and trabecular thickness when compared to Modic 1 and 2, linked to increased bone formation and reduced bone resorption. Our study has not revealed any specific active pathological condition behind Modic changes, which therefore, at this point in time, should be interpreted as being merely a reactive response to the degenerating intervertebral disc.