Is the presence of Modic changes associated with the outcomes of different treatments? A systematic critical review

Low back pain (LBP) is a large problem in the Western world. It has considerable financial consequences both for the individual person and for society. Currently, less than 15% of patients seeking care for LBP are thought to have specific causes of LBP [1], leaving the remaining 85% classified as having 'non-specific LBP'. Previous studies of various types of treatment show only little or no treatment effects in patients with non-specific LBP [2]. One explanation for this could be that non-specific LBP is a symptom caused by several different pathologies, thereby dividing patients into different pathological subgroups. If this were the case, it would not be surprising if the effect of treatment were often unimpressive, as these unidentified subgroups would be treated without any evidence-based rationale. Therefore, more knowledge on different methods of subgrouping of patients in relation to indications for various treatment approaches would be very helpful, as it would improve the possibilities for a more targeted treatment approach.

One such possible subgroup, Modic changes (MCs), have recently been associated with LBP [3]. MCs appear to be a stage of the disc degeneration process [4‐7]. They were first defined in the literature by Modic et al. [8] who described two types (type I and II) of signal changes visible on magnetic resonance imaging (MRI). This research group also made a histological examination of the findings which revealed fissured endplates and vascular granulation tissue adjacent to the endplate in type I, and disruption of the endplates as well as fatty degeneration of the adjacent bone marrow in type II. Modic et al. also described a third type as corresponding to sclerosis seen on radiographs [9]. Although it is uncertain whether 'normal' disc degeneration is associated with LBP, an association between MCs and LBP was found in a systematic review by Jensen et al. who reported a positive association in seven out of ten studies with odds ratios from 2.0 to 19.9 [10].

There are two main theories as to why MCs develop - a biomechanical theory and an infection theory. The first theory is that MCs are caused by mechanical stress [11]. The degeneration of the disc leads to changes in the mechanical conditions in and around the disc [4, 12]. Improper loading and shear forces then cause micro-fractures of the endplate resulting in inflammation in the vertebral endplate and the adjacent bone marrow [13, 14]. Jensen et al. found that persons from the Danish general population with disc degeneration, bulges or herniations had twice the odds of new endplate changes (MCs) over four years compared with persons with normal disc contours or no degeneration [15], and Albert et al. found that among patients with disc herniation at baseline 17% had developed new MCs type I at 14-months follow-up at the same vertebral level as the previous herniated disc [16].

The second theory is that the inflammation and oedema in the vertebral endplate are caused by a bacterial infection in the associated disc [11]. Following a disc herniation, new capillarisation and inflammation occur which are thought to be a 'port' for anaerobic bacteria to enter the disc. Sterling et al. [17] found the presence of low virulent bacteria in 53% of the disc material harvested from surgery on herniated discs whereas Carricajo et al. [18] found only 7%, and argued that the presence of bacteria was due to contamination of the disc samples. The debate about the presence of bacteria in MCs type I is ongoing as Wedderkopp et al. [19] found no trace of anaerobic bacteria in 24 biopsies taken from vertebrae affected by MCs type I. However, Ohtori et al. [20] found that 4 patients out of 71 with MCs type I at baseline developed clinical symptoms of pyogenic spondylitis over a two-year period, and that 3 of those patients had a pyogenic infection confirmed with a biopsy.

If MCs constitute a specific subgroup of LBP, one would expect different outcomes with different treatments for this condition. However, this would depend on the etiology of this pathology, which remains contentious. If the mechanical theory is correct, one would expect alleviation of symptoms with rest, because immobility might be necessary to heal any micro-fractures, whereas vigorous weight bearing exercise might prevent micro-fracture healing. Similarly, a good outcome might also be expected from fusion surgery, because fusion may neutralize the biomechanical dysfunctions in the vertebral segment [21]. Perhaps for this reason, some surgeons are of the opinion that the presence of MCs is a good indication for fusion surgery [22]. In contrast, if the bacterial theory is correct, the outcome should be favourable with antibiotic treatment [23].

In other words, the presence of MCs may have an impact on outcome, either positively or negatively, for various types of treatments. In order to synthesise the evidence, we performed a systematic critical literature review. The objective was to investigate if the presence of MCs at baseline is associated with outcomes from different kinds of treatments for LBP.

The studies identified in this review were too few and too heterogeneous to reach a definitive conclusion as to whether MCs can be used to guide optimal treatment in patients with LBP. In addition, only 1 of the 6 studies was considered to be of good quality. The results of that study, which investigated the effect of exercise, were not particularly helpful, as they failed to differentiate MCs into sub-categories. Obviously, this is important, if various stages of MCs require different types of treatments.

The other studies, of varying quality that never exceeded fair, provided a confusing picture. Results varied for different types of treatments, at different times of follow-up and for different outcomes, in a manner that could not easily be interpreted. The only treatment investigated in two separate studies was lumbar intradiscal steroid injections but the results did not concur. Furthermore, results could not be obtained from enough studies to make a systematically reporting of the standardised effect size or another uniform measure meaningful.

For those reasons, it was also not possible to interpret the findings relative to the main theories on the etiology of MCs. There was not enough weight of evidence in favour of the biomechanical theory and we found no studies that could directly cast any light on the bacterial theory.

A potential weakness of this study is that it is possible that we missed some studies on this subject, even though the search was comprehensive and included several languages other than English. Also, some relevant studies may have been overlooked, as the initial screening of abstracts and titles was undertaken by only one of the authors. Furthermore, as we were unable to find any broadly accepted quality check-lists for this type of study, we designed our own. The presence of other quality criteria could have resulted in another judgment of the quality of this study, although it is unlikely this would have changed the interpretation of the results.

The weaknesses identified in this review make it relevant to comment on the need for future studies to respect certain methodological criteria. Two types of study designs would be suitable. 1) The one arm prospective outcome study with internal control groups i.e. the presence/absence of MCs. 2) A better design is the randomised controlled trial (RCT). RCTs should be conducted and reported according to general recommendations [29]. In order to study the predictive value of MCs it would be necessary to define its various types (such as type I, type II and mixed types). Also, obviously, the normal steps to avoid selection bias and bias in data interpretation must be taken.

In conclusion, the studies in this review were too few, too heterogeneous and often lacking in adequate methodological rigour, to make a definitive conclusion as to if and how MCs are an indication for specific therapies for LBP. Therefore, although MCs may be associated with pain, it remains unclear if MCs are of clinical importance for prescribing treatment for a patient with LBP and more high quality research on the topic is needed. It also seems necessary to differentiate the types of MCs in any future analysis of treatment effects involving patients with MCs.