The advent of magnetic resonance imaging (MRI) has enabled comprehensive structural assessment of the spine, providing new insights into the structural features associated with spinal pathology and non-specific low back pain. Recently, there has been particular interest in lumbar vertebral body marrow and endplate lesions, termed “Modic change” [1]. Three Modic changes have been classified, each with different histopathologic correlates. Modic type 1 change represents bone marrow oedema and inflammation; Modic type 2 change represents marrow ischemia and the conversion of normal red haemopoietic bone marrow into yellow fatty marrow [1]; and Modic type 3 change is rare and represents of subchondral bone sclerosis [2].

Although the histopathologic correlates of Modic change are becoming better elucidated, the clinical significance of Modic change remains the subject of investigation. Modic type 1 lesions in the lumbar spine are strongly associated with low back pain [3–7] and instability [3, 8]. With the passage of time, a large proportion of Modic type 1 change converts to type 2 [1, 4, 7, 9]. Modic type 2 change is less strongly associated with low back pain [3–6]. Nevertheless, Modic type 2 change is more common than type 1 [1, 2, 5, 10–12], and more frequent among individuals with degenerative disc disease [3, 13, 14]. Since Modic type 2 change is characterized histologically by fatty infiltration [1], it is possible that this lesion is related to adiposity. Although not widely examined, BMI has not been shown to be associated with Modic change [12]. Nevertheless, BMI cannot discriminate adipose from non-adipose tissue. No study has examined whether body composition and fat distribution are associated with Modic change.

Although past studies have demonstrated that Modic type 2 change occurs at sites of degenerative disc disease [1, 2, 10, 11], the definition of degenerative disc disease is broad and incorporates a multitude of radiological features including disc protrusion, herniation and reduced intervertebral disc height. No study has examined the associations between Modic change and individual radiological features of disc degeneration, such as intervertebral disc height.

The aim of this cross-sectional study was to determine whether body composition is a risk factor for Modic change in community-based adults and whether Modic change is associated with intervertebral disc height. We hypothesized that i) adiposity is a risk factor for Modic change; and ii) Modic change is associated with reduced intervertebral disc height.

This study of community-based adults demonstrated that Modic type 2 change is by far the most prevalent Modic change, and is associated with reduced intervertebral disc height of the lumbar spine. While increased fat mass was associated with a higher risk of Modic type 2 change, increased fat free mass tended to protect against this lesion. Moreover, distribution of body fat was important; with a gynoid pattern protecting against Modic type 2 change, and an android pattern of fat distribution increasing the risk of this lesion. Such findings suggest that Modic type 2 change, which histologically represents fat replacement of marrow, might have a metabolic component in its aetiology.

This study supports previous reports of a higher prevalence of Modic type 2 change at the L4-5 and L5-S1 level in the general community [5, 12, 17]. Moreover, past studies have demonstrated that Modic change occurs at sites of degenerative disc disease [1, 2, 10, 11]. Nevertheless, the definition of degenerative disc disease is broad and incorporates a multitude of radiological features including disc protrusion, herniation and reduced intervertebral disc height. No study has examined the associations between Modic change and individual radiological features of disc degeneration. We have used intervertebral disc height as a measure of disc degeneration. We recently demonstrated a negative dose–response relationship between increasing severity of disc degeneration with a reduction in intervertebral disc height [18]. The utility of intervertebral disc height as a quantitative and continuous measure enabled us to sensitively examine the relationships between intervertebral disc height and Modic lesions. For instance, we have shown that a reduction in intervertebral disc height is associated with Modic type 2 change in the adjacent vertebrae of the low lumbar spine, as well as the total and average intervertebral disc height of the lumbar spine. For every 1 mm reduction in the average disc height of the lumbar spine, there was a 1.6 times (95 % CI 1.1 to 2.3) increased risk of Modic change in the lumbar spine. Since the intervertebral disc abutting the vertebral endplates is comprised of avascular fibrocartilage, it is possible that the histological features of fatty replacement of the bone marrow that characterizes progressive Modic change [1, 2] impedes nutritional support to the intervertebral disc with resultant disc degeneration and loss of disc height. Alternatively, reduced intervertebral disc height may attenuate the cushioning effect and load redistribution to adjacent vertebrae, resulting in Modic change. The cross-sectional design of this study cannot however determine whether reduced intervertebral disc height is a cause or result of Modic change and longitudinal studies are required to explore a causal relationship.

Although not widely examined, BMI has not been shown to be associated with Modic change [12]. Nevertheless, BMI cannot discriminate adipose from non-adipose tissue. No previous study has examined whether body composition and fat distribution are associated with spinal structural change. In the current study, we observed an increased risk of Modic type 2 change in relation to greater fat mass but a reduced risk for this lesion associated with greater fat free mass. Such findings demonstrate the potential importance of body composition in the pathogenesis of structural abnormalities in the low back. Additionally, this study has also shown that the distribution of body fat is a determinant of Modic type 2 change. Whereas a gynoid pattern of fat distribution protected against the presence of Modic change, an android pattern increased the risk of Modic change in the lumbar spine. Android fat accumulates in the upper and central body, particularly the abdomen, while gynoid fat accumulates in the lower body, particularly the hips and thighs.

Compared with the gynoid pattern, the android pattern of fat distribution is considered to be more metabolically active and has been associated with increased risk of cardiovascular disease [19, 20], abnormal lipid profiles and diastolic blood pressure [21]. It is also becoming increasingly recognized that the android pattern of fat distribution is an important determinant of the burden of musculoskeletal diseases and a recent study has shown that android fat is strongly associated with foot pain and disability [22]. Moreover, android fat has been associated with structural abnormalities at the knee, such as cartilage defects [23] and reduced cartilage volume [24]. Whereas non-specific back pathology has typically been considered to have a strong biomechanical component to its aetiology, these data provide evidence that Modic change may be related to metabolic processes. Nevertheless, we cannot exclude a biomechanical contribution to Modic change. It is possible that excess mass in the trunk increases axial loads, leading to structural abnormalities such as Modic change. Nevertheless, this biomechanical mechanism would not explain the protective association between gynoid fat distribution and Modic change.

The mechanism accounting for the relationship of body composition and fat distribution with Modic change is however, speculative. Histologically, Modic type 2 change is representative of fatty replacement of the bone marrow [1], and may therefore be closely linked to adiposity. Although previously considered a passive store of energy, adipose tissue is now recognized as a highly active metabolic structure that contributes to systemic inflammation, secreting a host of pro-inflammatory cytokines including tumor necrosis factor (TNF) and interleukin-6 (IL-6) [25]. Both TNF-alpha and IL-6 have been shown to be predominantly derived from the adipose bed located within the trunk and abdomen [26]. TNF immunoreactive cells have been shown to be present in greater number in people with Modic change, and particularly Modic type 1 change [27]. Whereas Modic type 1 change is considered inflammatory in nature and a source of low back pain, the natural history of Modic change in the lumbar spine suggests that the majority of type 1 lesions progress to type 2 [3–7]. Since body composition and fat distribution data was collected approximately 4 to 6 years prior to MRI assessment, it is possible that the systemic inflammation imparted by adiposity resulted in type 1 Modic change that, with the passage of time, progressed to Modic type 2 change seen in this study. Longitudinal studies may help to better elicit the natural history between body composition, fat distribution and the natural history of Modic change.

Our cross-sectional study had a modest sample size and mainly comprised females (73.7 %). A larger study with a more even gender distribution is required to substantiate the generalizability of our findings. Moreover, longitudinal studies will help to determine any causal relationships between body composition, fat distribution and Modic change, as well as the interaction between Modic change and intervertebral disc degeneration. Fifty-seven of 72 participants had body composition and fat distribution measures performed. The baseline characteristics of the 15 subjects without these data were not different from the 57 with these data available, other than people who did not have absorptiometry (n = 15) being significantly more obese (BMI 33.5 kgm⁻² versus 28.6 kgm⁻², p = 0.04). The loss of more obese people not having absorptiometry would likely have reduced the chance of showing significant relationships between measures of body composition and fat distribution and MRI abnormalities. Additionally, measures of body composition and fat distribution were taken 4 to 6 years prior to MRI assessment of the lumbar spine. Since we have shown that the fat mass index was associated with an increased risk of Modic type 2 lesions, any weight loss that may have occurred from recruiting in part from weight loss clinics, would have underestimated associations. This is a strength of the study since it implies that an android pattern of fat distribution precedes Modic change. Reverse causality whereby Modic change predated an android pattern of fat distribution is less likely in a community-based population not recruited on the basis of established back pain and disability that could have imparted activity restriction and a subsequent change in body composition. Indeed, a large population study of people aged 15 to 98 years demonstrated that the natural history of fat mass is to increase with ageing, while fat free mass tends to reduce once reaching its peak between the ages of 35 to 54 years [28]. Such findings may account for the growing prevalence of low back pain seen with ageing [29]. Moreover, Modic change in this study was predominantly type 2, with only one participant having a type 1 change. Whether body composition and fat distribution are associated with Modic type 1 lesions requires investigation and may be best explored by selectively recruiting people with low back pain at the time of MRI. Finally, although the disc height at the L3/4 level was not significantly associated with Modic, this may be a reflection of the low prevalence of Modic lesions at this level. Indeed, Modic lesions were more common at L4 (19.3 %) and L5 (21.1 %), and less common at L3 (5.3 %). The lower prevalence at L3 has likely reduced the ability of this study to demonstrate associations at this level. Future studies would benefit from a larger sample size to address such issues.

This study has demonstrated Modic type 2 change, which is associated with reduced intervertebral disc height of the lumbar spine, is positively associated with fat mass while fat free mass tended to protect against Modic change. Moreover, distribution of body fat was important and whereas a gynoid pattern protected against Modic change, an android pattern of fat distribution increased the risk for Modic change. Such findings suggest that Modic type 2 change, which histologically represents fatty replacement of bone marrow, may have a metabolic component in its aetiology.