Modic changes and tumor necrosis factor inhibition: Is the door shut?

First described in 1988, Modic changes are MRI findings in the lumbar spine that are present in a substantial fraction – approximately 43% – of patients with back pain, and are thought to reflect pathobiological states characterized by combinations of inflammation, bone turnover, and fibrosis.^{1 2 3} Three subtypes of Modic changes are delineated by their characteristic appearance on MRI, but multiple types may coexist in any individual patient, and lesions may progress between types over time. Type 1 changes are hypointense on T1-weighted sequences and hyperintense on T2-weighted sequences, type 2 changes are hyperintense on T1-weighted sequences and hyper-or isointense on T2-weighted sequences, and type 3 changes are hypointense on both T1- and T2-weighted images.^1,2 Similar MRI changes have been described in other locations and other diseases, where they have received different names. For example, in axial spondyloarthritis, bone marrow lesions that are hypointense on T1-weighted sequences and hyperintense on T2-weighted (equivalent to Modic type 1 changes) are called bone marrow edema and are considered to represent active inflammatory lesions.⁴

Histopathological studies of Modic changes have been sparse.² Type 1 Modic changes may reflect an active bone turnover state accompanied by increased pain signaling and nerve fiber density in the vertebra.⁵ As a result, type 1 changes represent a clinically important target for potential intervention.⁵ Controversy has surrounded the primary stimulus that leads to Modic changes and the optimal management of patients who exhibit these findings. Some evidence has supported a potential role for an infectious catalyst,⁶ but trials of antibiotics have yielded mixed results.⁷ Many researchers believe that Modic changes begin with increased load on the vertebra initializing a sequence of endplate damage, disc herniation into the endplate, a repair response to the herniation, and changes in tissue-level cytokine expression including upregulation of inflammatory mediators including tumor necrosis factor (TNF).⁵

Several key findings have led to the hypothesis that Modic type 1 changes are largely inflammatory in nature and may respond to anti-inflammatory treatment, including TNF inhibitors. TNF-immunoreactive cells are more densely populated in endplates with type 1 compared to type 2 Modic changes,⁵ C-reactive protein levels are higher in patients with type 1 compared to type 2 Modic changes and low back pain,⁸ and response to intradiscal steroid injection was found to be superior in patients with type 1 compared to type 2 Modic changes.⁹ A trial of infliximab in disc herniation-induced sciatica suggested a signal for symptomatic benefit in a subgroup of patients who had Modic changes at the level of the herniation.¹⁰ As individualized treatment for symptomatic degenerative disc disease remains an important and elusive target, there has therefore been significant interest in the therapeutic potential for anti-inflammatory treatment in patients with type 1 Modic changes.

The results of the BackToBasic randomized controlled trial, which was carefully designed to address this hypothesis, are reported by Gjefsen and colleagues in this issue of Arthritis and Rheumatology¹¹. The trial’s protocol was published previously.¹² The study recruited 129 adult participants with chronic low back pain and Modic type 1 changes in the lumbar spine between 2018 and 2022. The trial specified minimum scores for either pain or disability as additional inclusion criteria. Patients with specific spinal diagnoses such as axial spondyloarthritis were excluded. Participants were randomized to four blinded intravenous infusions over 14 weeks of either infliximab 5 mg/kg or placebo. The primary outcome was change in the Oswestry Disability Index (ODI) from baseline to 5 months, with secondary outcomes including change in low back pain intensity and additional measures of disability and health-related quality of life. The ODI, a validated and frequently used self-report measure, consists of 10 questions about potential pain-related functional limitations, each scored from 0 to 5, and the final score is reported as a percentage from 0 to 100.¹³

The BackToBasic trial was powered to detect a 10-point difference in ODI change scores between the infliximab and placebo groups, a relatively large difference that is comparable to the standard deviation of the ODI in the study sample. While both groups improved over five months, the trial result was negative with respect to the primary outcome, with an estimated between-group difference of 1.3 points at five months on the ODI (95% confidence interval, −2.1 to 4.6). Indeed, in Figure 2, we see that the difference between groups at 5 months was small. In exploratory analyses, the investigators identified that the difference in ODI between the trial arms widened at 9 months to an estimated difference of 4.2 (95% CI, 0.8 to 7.6) ODI points. This is a modest difference (about 40% of a standard deviation) and smaller than most proposed minimal clinically important differences for the ODI (5 to 17 points on an absolute scale).¹⁴ The authors have appropriately reported the primary outcome without undue speculation regarding the nine-month finding. However, on broader visual inspection of Figure 2, we observe that the difference between the treatment and placebo groups narrowed at 5 months compared to prior months and again widened at 9 months, suggesting the null primary result may be partly attributable to chance.

In addition to the patient-reported outcomes reported in this manuscript, repeat imaging data, as described in the BackToBasic protocol, would be of interest. Bone marrow edema in axial spondyloarthritis improves with TNF inhibition;¹⁵ demonstrating lack of improvement in type 1 Modic changes in the BackToBasic trial would provide additional evidence against the role of anti-TNF therapies in this subset of patients with low back pain.

These considerations aside, the very small net improvement in ODI at five months due to infliximab in the BackToBasic trial may not be surprising, given the important pathobiological differences between axial spondylarthritis (in which bone marrow edema at similar locations is seen, and for which infliximab is quite effective¹⁶) and degenerative disc disease. In the former, immune cell activity and inflammation are primary actors in the production of back pain, stiffness and disability, and TNF inhibitors and other immunomodulators result in dramatic clinical improvement in many patients.¹⁷ By contrast, in degenerative disc disease, inflammation is thought to be a downstream effect of a primary mechanical disturbance, which itself may provide a nociceptive stimulus.¹⁸ Treating the downstream inflammatory component of the disorder may not reverse the pathological mechanical processes that ultimately drive pain and disability. Furthermore, low back pain in the context of degenerative disc disease is frequently multifactorial, resulting from structural factors (including increased load and damage to richly innervated bone, nerve injury and impingement, and ligamentous strain), inflammation arising from these structural factors, and pain centralization.¹⁸ A treatment strategy that addresses each of these mechanisms successfully would be exceptional among medical therapeutics; thus, it is not surprising that the infliximab treatment effect was quite modest in this trial.

Prior research has yielded mixed results with respect to the role for systemic immunomodulation in osteoarthritis and degenerative disc disease. Methotrexate was recently shown to reduce pain in randomized controlled trials of hand osteoarthritis (OA) (with synovitis on MRI)¹⁹ and knee OA (without the requirement for synovitis on imaging).²⁰ A secondary analysis of a large randomized controlled trial found that the anti-IL-1 beta antibody canakinumab was associated with reduced hazard for hip and knee replacement, regardless of prior history of inflammatory arthritis or gout.²¹ However, TNF inhibition, in particular, has not shown efficacy with respect to clinical outcomes in hand OA.²²

Overall, the results of the BackToBasic trial advance the discussion regarding the distinction between primary versus secondary inflammation in musculoskeletal disease, in particular helping to address the question of whether treatments that are successful in the former will be equally effective in the latter. Based on the results from this well-executed study, we can conclude that TNF inhibition is unlikely to have a clinically important effect on clinical outcomes in low back pain with Modic changes. However, it is difficult to shut the door tightly on the use of targeted anti-inflammatory therapies for degenerative disc disease. For example, the between-group difference of 0.4 SD at nine months in an exploratory analysis raises the question of whether potent immunomodulatory therapy may exert a delayed effect on pain and function. This observation should be addressed in another trial. Further studies are also needed to better understand the processes driving type 1 Modic changes at the cellular and molecular level. While the role for targeted immunomodulation in secondary inflammatory pathology (including degenerative low back pain with Modic changes) requires further study, we suggest that expectations for such treatment should remain guarded because of the primary role of other, mechanical processes.