Key points
- •
Modic changes are subchondral and vertebral bone marrow changes in conjunction with intervertebral disc degeneration, and they are classified into three categories (types 1, 2, and 3) based on magnetic resonance imaging appearance.
- •
Modic changes, especially Modic type 1 change, are associated with low back pain but they are also prevalent among people without low back pain.
- •
The precise pathology of Modic changes is unclear but their development requires structural disc damage and inflammatory response to it. Bacterial etiology behind Modic changes is possible but lacks solid evidence.
- •
To date, there are no definite guidelines on how to treat patients with low back pain and Modic changes.
Acknowledgment
Special thanks to Professor Frances Williams for helping to write the chapter and Anne Kukkonen for performing lumbar MRIs in the Northern Finland Birth Cohorts.
Introduction
Subchondral vertebral bone marrow changes were first described by Assheuer et al. [ ] and de Roos et al. [ ] in 1987 to separate degenerative marrow change from tumor or infection. In 1988, Modic et al. [ ] classified subchondral bone marrow changes associated with disc degeneration into two categories (type 1 and type 2) using magnetic resonance imaging (MRI) appearance and histologic samples, and afterward a third type (type 3) was identified using MRI [ ]. Later, these changes have been named generally as Modic changes (MCs). In the literature, MCs are also called as vertebral bone marrow change, vertebral endplate signal change, or vertebral endplate degenerative change (see Chapter 1, Chapter 10 ). MCs are changes of subchondral vertebral bone marrow ascending from the endplate toward the center of the vertebral body, and there are often no visible changes seen in the endplate.
The classification of MCs is traditionally based on bone marrow signal intensity on T1- (T1w) and T2-weighted (T2w) MRI to type 1 (MC1), type 2 (MC2), and Modic type 3 change (MC3). Subsequently, other sequences, such as fat-suppressed fluid-sensitive MRI (STIR) sequences, have also been utilized [ ]. The relationship between the three Modic types is discussed later (see Manifestations and Natural History).
Modic changes were previously not considered as a specific low back pain (LBP)-related imaging phenotype but, along with an increased understanding of MCs, this view has changed—especially in the case of MC1 [ ]. However, contradictory findings have been published also in recent years [ , ]. Many phenotypes have been described for lumbar MCs ( Table 11.1 ). In this chapter, we will discuss pathophysiology, natural history, and management of MCs with specific reference to its clinical and imaging phenotypes ( Table 11.2 ).
| Phenotype | Description | Author/Reference |
|---|---|---|
| Type 1 change (MC1) | Hypointense on T1-weighted (T1w) images and hyperintense on T2-weighted (T2w) images | [ ] |
| Type 2 change (MC2) | Hyperintense on both T1w and T2w | |
| Type 3 change (MC3) | Hypointense on both T1w and T2w | |
| Mixed types 1 & 2 (MC1/2) | MC1 and MC2 at the same locus | [ , ] |
| Mixed types 2 & 3 (MC2/3) | MC2 and MC3 at the same locus | |
| Endplate localization | Anterior Middle Posterior Left Left-paracentral Central Right-paracentral Right | [ ] |
| Intravertebral location | Central Anterior left Anterior right Posterior left Posterior right | [ ] |
| Maximum height ∗ | No MC Endplate only <25% of vertebral body volume 25-50% of vertebral body volume >50% of vertebral body volume | [ ] |
| Intravertebral volume | Normal Endplate only <25% of vertebral body volume 25-50% of vertebral body volume >50% of vertebral body volume | [ ] |
| Endplate area | No MC <25% of the endplate area 25-50% of the endplate area >50% of the endplate area | [ ] |
| Transverse size | No MC Small (MC in 2-3 slices 1 ) Moderate (MC in 4-5 slices 2 ) Large (MC in 6-8 slices 3 ) Very large (MC in ≥9 slices 4 ) | [ ] |
| Anteroposterior (AP) Diameter | Anterior Central Posterior Small: MC in one portion Moderate: MC in two portions Large: MC in all three portions | [ ] |
| Maximum height ∗ | No MC Endplate only <25% of vertebral body volume 25-50% of vertebral body volume >50% of vertebral body volume | [ ] |
| Intravertebral location | Anterior Midpoint Posterior Left lateral Right lateral | [ ] |
| Horizontal size | Summary score from nine affected intravertebral locations: Anterior left Anterior midpoint Anterior right Midpoint left Midpoint Midpoint right Posterior left Posterior midpoint Posterior right | [ ] |
| MC grading score | Grade A: <25% of vertebral body height Grade B: 25%-50% of vertebral body height Grade C: >50% of vertebral body height | [ , ] |
| Shape contouring of MC | Quantitative size: Region of MC | [ ] |
| Bone marrow composition | Bone marrow edema and fatty marrow Fat fraction of MC area | [ ] |
| Diffusion-weighted imaging | Claw sign Apparent diffusion coefficient | [ ] [ ] |
∗ Evaluated from the sagittal image in which MC had the largest extension into the vertebra
1 Corresponding <25% of the transverse diameter of the vertebral body
2 Corresponding <50% of the transverse diameter of the vertebral body
3 Corresponding <75% of the transverse diameter of the vertebral body
4 Corresponding >75% of the transverse diameter of the vertebral body
| Classifier | Subclassifier | Phenotype |
|---|---|---|
| Imaging (MRI) | Modic change (MC) | MC type |
| MC location | ||
| MC height | ||
| MC width | ||
| Disc | Disc height | |
| Disc signal intensity | ||
| Disc displacement | ||
| Endplate | Endplate defect | |
| Schmorl’s node | ||
| Clinical | Low back pain | Inflammatory pain pattern (MC1) |
| Treatment | Nonoperative | Good/poor responder |
| Operative | Good/poor responder | |
| Outlook | Imaging | UTE a , [ , ] |
| Treatment | Antibiotics [ ]/bisphosphonate [ ] |
Pathogenesis
The development of MCs in conjunction with intervertebral disc degeneration has been recognized. Modic changes were even originally described adjacent to a degenerated disc [ ], and they have been suggested to be a reactive response to disc degeneration and endplate injury [ , ].
Although the true nature and pathogenesis of MCs are still uncertain, MCs share several etiological factors and phenotypic characteristics with bone marrow changes at other sites of the body. For instance, bone marrow edema in conjunction with other joints, primarily hip and knee joint, has similar phenotypic characteristics with MC1 in MRI [ ]. Bone marrow edema can be found in multiple health conditions including knee osteoarthritis, spondylarthritis, and chronic tendinitis [ , ]. Histologically, the site of both bone marrow edema and MC1 can contain vascularization, infiltration of lymphocytes, and/or fibrosis, and these conditions can cooccur [ ].
Histologically, bone marrow demonstrating MC1 contains disrupted and fissured endplate, vascularized fibrous granulation tissue, and high bone turnover; MC2 consists of fatty regeneration of the bone marrow and reduced bone formation, whereas bone marrow demonstrating MC3 comprises bone sclerosis [ , ].
Overall, MCs can be viewed as a specific phenotype within the realm of endplate changes (ECs). The distinctive feature of MCs is the formation of intravertebral marrow edema. Other causes of ECs include infection, trauma, and Schmorl’s nodes [ ]. Previous spine injury, intervention, or systemic disease can cause injury or changes to the functional spine segment that can predispose or cause disc injury, ECs, and MCs. If available, such patient information should be included for both the MRI radiologist and the clinician ( Figs. 11.1–11.4 ) .
Genetic disposition and MCs
Research has demonstrated a close association between the genome and back pain [ ]. In the United Kingdom, twin study genetic heritability was found to be one of the major risk factors for reported episodes of severe LBP. Several gene variants have been linked with an increased genetic disposition to disc degeneration and LBP.
The association between gene variants and MCs has been investigated across studies. Karppinen et al. [ ] did in 2008 show that genetic variations in the interleukin-1A and matrix metalloproteinase-3 gene together were significantly associated with MC2. In 2019 a genome-wide meta-analysis identified a genetic locus on chromosome 9 associated with MCs [ ]. Rajasekaran et al. [ ] performed a proteomic analysis of intervertebral disc tissue to investigate the biological changes of MCs at the molecular level. They found 45 proteins specific for MC patients and in particular 14 host defense response proteins with altered pathways. Another study using proteomic analysis comparing endplate avulsion and MCs found significant alteration of nine proteins compared to patients without MCs [ ]. In 2022 a study by Vigeland et al. [ ] found that the expression of 37 genes was associated with STIR signal volume in patients with MC1. Gene sets related to interferon signaling, mitochondrial metabolism, and defense response to virus were significantly upregulated in all analyses. These results indicate that inflammation and immunological defense are important factors in MC biology in patients with LBP.
In summary, studies indicate that there is a genetic component to MC that includes altered signaling, metabolism, and host defense response.
Structural disc damage and inflammatory response
Structural disc damage through endplate damage or disc herniation induces the development of MCs [ ], and MCs are rarely seen at structurally intact disc levels (see Chapter 1 ) [ , ]. When the endplate damages, the disc depressurizes, and this alters the mechanical environment of the intervertebral disc and causes progressive structural change to the motion segment [ ]. Endplate defects are associated with MC development fairly consistently across studies [ ]. Extracellular matrix changes have been observed in cartilage endplate with MCs as shown in a study investigating 40 patients with MCs and 20 patients without, where expression of a disintegrin-like and metalloprotease with thrombospondin motifs 5 (ADAMTS-5), catabolic enzyme, and tumor necrosis factor-α (TNF-α) was significantly upregulated in cartilage endplates of those having MCs [ ].
Disc herniation is a potential predisposing factor to MCs [ , ], as it can also generate endplate junction injury. Disc herniation occurs more frequently as a result of endplate junction failure rather than annulus fibrosus rupture as previously thought [ , ]. Moreover, studies have shown that extruded discs have been shown to comprise cartilage endplate and are significantly associated with MCs [ , ].
Structural disc damage leads to change in diffusion and thereby metabolism of the intervertebral disc cells with increased production of proinflammatory cytokines, leading to an inflammatory response. The cytokines could migrate from the intervertebral disc and endplate into the vertebral bone marrow [ , ]. Structural damage in a depressurized disc could induce a persistent inflammation stimulus and attempts to heal, leading to a so-called frustrated healing response and, consequently, MC development [ ]. To support this, crosstalk between bone marrow with MCs and the adjacent disc has been discovered at a cellular level [ ]. Low-grade inflammation, determined by increased serum high-sensitivity C-reactive protein (hs-CRP) concentration, has been found to associate with MC1 [ ] but this finding has not been validated. Additionally, findings on the association of genetic variations in the interleukin-1 (IL-1) cluster and the matrix metalloproteinase-3 gene with MC support the role of innate immunity in MCs [ , ]. A study explored serum biomarkers among patients with chronic LBP and MCs and found several suppressed biomarkers among patients with MCs, and to date no serum biomarkers have been identified able to distinguish Modic types [ ].
Modic changes are associated with increased expression of TNF-α [ ] and TNF-immunoreactive cells in the endplate [ ]. Increased oxidative and nitrosative stress factors, leading to the production of proinflammatory cytokines such as TNF-α, interleukin-8 (IL-8), and prostaglandin E2 (PGE2), have also been linked to MC1 [ ]. In addition, macrophage migration inhibitory factor was found to be highly expressed in the cartilage endplate chondrocytes among subjects with MC1 suggesting, again, increased proinflammatory cytokine secretion [ ].
As MCs are lesions of the vertebral bone marrow, the role of osteoclasts has also been studied. Theoretically, metabolic pathways of bone remodeling could be important in MC. Messenger RNA of cytokines related to osteoclast proliferation and differentiation, such as macrophage colony-stimulating factor-1 (M–CSF–1), receptor activator nuclear factor kappa-β ligand, and osteoclast-associated receptor, were significantly more expressed in the herniated disc tissue samples with MCs at the operating level than in tissue samples without MCs, suggesting osteoclast involvement in the development of MCs [ ]. In addition, upregulation of pro-osteoclastic CSF-1 and peroxisome proliferator-activated receptor-γ and downregulation of interleukin-4 (IL-4) and osteoprotegerin have been seen at disc levels with MC2 [ ]. Increased sclerosis has been reported to be present on radiographs and computed tomography in vertebrae with MCs [ , ].
There are histologic studies that support inflammatory response to herniated tissue material. Herniated tissue upregulates vascular endothelial growth factor, which further stimulates angiogenesis [ , ]. In addition, herniated tissue shows an abundance of macrophages [ ]. When the disc degenerates due to structural changes, compressive stress loads are highly altered in the disc. This leads to neovascularization and neoinnervation that increase the metabolics, especially in the endplate and outer annulus fibrosus [ ]. Both angiogenesis and macrophage infiltration strengthen the inflammatory response. The derangement of annulus fibrosus and/or endplate enables a convenient route for cytokines to migrate into the vertebral bone.
Atypical mechanical loading
Atypical mechanical loading to vertebrae can be due to asymmetrical loading or increased shear stress of the motion segment (see Chapter 2 ). Asymmetrical loading is seen in scoliosis. Wu et al. [ ] studied patients with degenerative lumbar scoliosis and found MCs to be more prevalent among patients with scoliosis than without. Furthermore, usually, MCs are located on the concave side of the apex vertebrae where the mechanical load is greater ( Fig. 11.5 ) [ ].
Increased shear stress in the lumbar structures and vertebrae is the result of structural changes leading to an altered mechanical environment in the disc [ , ] (see Chapter 8 ). Increased shear stress leads to minor fracture of the endplate [ ] and, as a consequence, enhances ascending inflammatory response from the disc into the vertebral bone marrow, leading to MC development [ ]. Indeed, MC1s have been found to demonstrate fissuring and disruption of the endplate [ , ]. Increased shear stress could also influence the shape of the endplate. Irregular endplates were associated with greater translational motion and higher prevalence of MCs compared to concave or flat endplates [ ]. In addition, longitudinal follow-up of surgical fusion studies gives an interesting insight into the role of instability and the effect of stabilization on MCs (see below Surgical options) ( Fig. 11.6 ) [ , ].
Bacterial infection
Low-virulent bacterial infection of the intervertebral disc has also been suggested to induce MC development and is currently the topic of much research. When the disc herniates due either to annulus fibrosus rupture or endplate junction failure, an inflammatory response is initiated. Disc herniation, and therefore structural disc damage, enables a route for bacteria from the circulatory system to enter the disc through the ruptured structure [ ]. As the disc has normally no blood vessels, anaerobic bacteria could thrive in the very low oxygen environment of the inner disc [ ]. As a result, MCs could be a tissue reaction of this process, appearing as inflammation and edema in the subchondral and vertebral bone marrow [ ].
Stirling et al. [ ] were the first to report anaerobic bacterial growth from disc material. They studied controls and patients who had undergone microdiscectomy due to severe sciatica and were able to culture bacteria from tissue samples among 53% of the patients in contrast to none in the control group. Cutibacterium acnes ( C. acnes ) formerly known as Propionibacterium acnes ( P. acnes ), a skin bacteria associated with acnes, was the main bacterium in the disc materials [ ]. Bacterial DNA was also reported in 2 of 10 patients operated for disc herniation or postdiscectomy syndrome [ ]. Supportive to bacterial etiology in MC development, 80% of the patients with positive anaerobic cultures developed new MCs at the herniated level compared to 44% of the patients with negative cultures in a study of 61 patients with a disc herniation [ ].
However, the sterility in the studies and possible contamination of the cultures have raised criticism. Ben-Galim et al. [ ] used strictly sterile conditions and found only four aerobic cultures with coagulase-negative staphylococci out of 120 cultures, with no C. acnes growth in any of the cultures. They also questioned positive cultures found in their study to be suggestive of contamination. In contrast, Rollason et al. [ ] found the majority of isolates to consist of two subtypes of C. acnes that are not so abundant in the skin, which could support an active role of C. acnes in lumbar disc herniation. Moreover, in the study by Zhou et al. [ ] with disc and muscle samples, no C. acnes was found in the control muscle samples. When anterior and posterior approaches were compared in spine surgery, the posterior approach had a higher prevalence of C. acnes , supporting possible contamination as a source of the bacterium [ ].
In a study using fluorescence in situ hybridization, the authors reported C. acnes with host inflammatory cells in only 7/51 (14%) of lumbar disc herniation patients and 0/14 (0%) of controls, whereas PCR analysis demonstrated much higher positive rates [ ], indicating that the PCR technique may itself be prone to contamination and that the rate of infection in people with MCs is relatively low.
Urquhart et al. [ ] conducted a systematic review in 2015 on studies that examined the relationship between bacteria and back pain or Modic change. Eleven studies were included with nine of those studies evaluating the presence of bacteria in spinal disc material. Seven out of nine studies found C. acnes positive samples. Overall, the study found moderate evidence for a relationship between bacteria and MC1.
Fritzell et al. [ ] performed a multicenter study on adults with lumbar disc herniation/LBP (median age 43) and control patients with scoliosis (median age 17) that underwent spine surgery at seven hospitals. Samples were cultured and compared from skin, surgical wound, discs, and vertebrae. The C. acnes genetic relatedness of isolates was compared using single-nucleotide polymorphism analysis. Disc samples were analyzed using 16S rRNA-based PCR sequencing. Bacterial growth was shown in 34/40 (85%) disc herniation patients, compared with 17/20 (85%) scoliosis patients. In the disc herniation group, 29/40 (72%) patients had C. acnes positive samples, compared to 14/20 (70%) in the scoliosis group. Bacterial findings and Modic changes were not associated.
Dudli et al. [ ] injected C. acnes into rat tail discs collected from symptomatic human discs with MC1. The result was a proliferation of C. acnes , upregulation of IL-1 and interleukin-6 (IL-6), and MC1-like changes in the bone marrow. Similar results have been observed among rabbits, where injection of C. acnes resulted in MC 1-like changes whereas injection of Staphylococcus aureus developed proper discitis [ ].
In summary, bacteria have been identified frequently in disc material taken at surgery, with the most predominant bacterium species being C. acnes [ , ] but the relevance of these findings has yet to be determined. MCs can be induced artificially by an external stimulus including chymopapain and C. acnes injection into the disc [ ]. However, this does not necessarily prove C. acnes to be the causative agent of MCs. It is possible that a subgroup of MCs is induced by C. acnes , similar to the subgroup of MCs caused by endplate junction failure in disc herniation.
In a recent review, C. acnes was rationalized to cause true infections rather than only contamination [ ]. Nevertheless, contradictory results have been reported [ ], and the role of bacterial infection in the development of MCs remains controversial [ ].
Iatrogenic induced changes
Discography and discectomy surgery for lumbar disc herniation have been shown to cause MCs on follow-up MRI. Both interventions cause a disruption of anatomical structures and can depressurize the disc leading to altered biomechanical loading. In particular, discectomy often includes direct endplate injury from the surgical tools that can cause an influx of disc material into the vertebra and induce an autoimmune response including upregulation of cytokines and edema. From a microbiological viewpoint, surgical procedures can serve as a direct pathway through the skin from which C. acnes can enter the disc and cause an anaerobe infection.
Pain generators
The pain from MCs can result from the endplate and adjacent bone marrow, which contains more nerve endings than the outer annulus fibrosus [ ]. Furthermore, endplates adjacent to MC1 are more innervated than bone marrow without MCs [ , ]. Taken together, this suggests that MCs are true pain generators in LBP. The latest edition of the International Classification of Diseases (ICD-11) has seen the addition of Vertebrogenic low back pain (low back vertebral endplate pain, DM54.51) as a subclassification of patients with LBP and MCs. This underlines the clinical importance of MCs as a relevant and common disease entity causing LBP and related disability.
Manifestations and natural history
MC1 is associated with LBP, and MCs are more prevalent in patients with LBP than in the normal population [ , ] although controversial findings have been shown [ ]. In patients with LBP, MCs have high specificity but low sensitivity [ ]. Nevertheless, MCs are prevalent in the normal population [ , , ], and the presence of MCs does not automatically explain LBP in an individual.
The inflammatory background of MC1 is thought to cause more pain as this type is considered to be biologically the most active of the three categories. MCs are predominantly located in the lower lumbar spine (L4/L5 and L5/S1), and a presence at those levels is considered more strongly associated with LBP compared to those located in the upper lumbar spine [ , ]. Changes in the upper lumbar spine are suggested to be congenital in nature, whereas changes in the lower lumbar spine result more likely from disc herniation or structural changes over time [ ].
Although the regression of MCs is possible [ , , ], they are usually considered to be quite sustainable, and therefore symptoms may be long lasting. Part of the patients can have prolonged, severe LBP with limitations in daily activities [ , , ], back-related disability [ , ], and a poor response to traditional low back treatments [ ]. Also, the presence of MC1 has been reported to be a poor predictor of recovery in patients with LBP [ , ]. There are, however, studies that show no association of MCs with LBP [ , ], and MCs have been found among asymptomatic subjects as well [ ]. A recent large population-based cohort study, the Wakayama spine study, found MC1/2 in the lumbar spine significantly associated with LBP [ ].
From a long-term perspective (>10 years), MCs in themselves have not been associated with a worse trajectory for LBP patients in regards to pain and disability [ , ]. Nevertheless, especially MC1s are related to LBP quite consistently, and this has also been acknowledged widely [ ] bearing in mind controversial findings [ ]. One should be cognizant of MCs (especially MC1) when a patient has chronic, possibly unusually severe, LBP with minimal or abnormal improvement to treatments.
The prevalence of MCs varies with respect to age, study population, and region in the spine. In general, the prevalence is higher in the lumbar than in the cervical spine and among low back or neck pain patients than in general or asymptomatic population [ ]. The prevalence rates in the cervical spine range from 5% to 41% among asymptomatic subjects and neck pain patients, respectively [ , ]. Similar to other degenerative findings in the spine, the prevalence of MCs increases also markedly with age. In patients below the age of 25 years, the prevalence is low (<5%) but from the mid-20s to middle age, there is an increase in the prevalence of both MC1 and MC2 [ ] to a prevalence of 35% for having one or the other type. In the general population, the prevalence of MCs has been reported to be 0.5% among 13-year-old adolescents [ ], 1.4% among 21-year-olds [ ], 22% among middle-aged subjects with a mean age of 49 years [ ], and up to 64% among subjects with a mean age of 64 years [ ]. Considering the incidence of MCs, the prevalence of MCs increased from 39% to 49% during a 4-year follow-up [ ] while an incidence of 22% was found in a 10-year follow-up [ ]. In a patient with chronic LBP and lumbar disc herniation, the prevalence of MCs has been found to be consistently high—around 45%, indicating both a relationship with pain and pathophysiology [ , ].
The natural history of Modic type proceeds from type 1 toward type 3, although MC3 is quite uncommon [ ]. As the process is dynamic, different types can occur at the same time. When there is more than one type at the same bone marrow locus, mixed type classification is used, these being usually Modic type 1/2 and 2/3 ( Figs. 11.7 and 11.8 ). [ , , ]. The larger the MCs, the more likely they are stable over time [ , ]. This is similar to other bone marrow edema conditions, such as a bone bruise or spondyloarthritis, as edema can turn into fatty bone marrow and, eventually, to bone sclerosis [ , ]. Lumbar MCs can also disappear totally, more likely when they are small [ , , ]. This course is also comparable to bone marrow edema conditions at other sites of the body, as these can self-limit or recover [ ].
