Low bone mineral density predicts the formation of new syndesmophytes in patients with axial spondyloarthritis

Axial spondyloarthritis (axSpA) is a chronic inflammatory disease that predominantly affects the sacroiliac joints and the spine. axSpA includes the subtypes ankylosing spondylitis (AS) and non-radiographic axSpA (nr-axSpA). These separate entities are discriminated by the structural damage to the sacroiliac joints visible on conventional X-ray images [1].

For most patients with SpA, the burden of disease results from a combination of inflammation and structural bone damage [2]. Radiographic damage in the spine presents as syndesmophyte formation leading to bridging of the intervertebral spaces. Structural damage not only affects patients by causing disability and permanent loss of function, but also has secondary effects, with ankylosis changing the balance of loads and forces on the skeletal system, leading to muscle stiffness and accelerated degenerative spine disease [3]. As structural damage contributes to impairment of spinal mobility and function, the retardation of spinal radiographic progression should be an important treatment goal [4].

Spinal progression varies widely among axSpA patients, and previous studies examined predictors influencing the heterogeneous formation of syndesmophytes within these patients. The strongest predictor of radiographic spinal progression is the presence of syndesmophytes at baseline [5, 6]. Furthermore, increased levels of acute phase reactants and smoking are independent predictors of radiographic spinal progression in early axSpA patients [6].

In a recent study, persisting high disease activity according to the Ankylosing Spondylitis Disease Activity Score (ASDAS) was found to be associated with accelerated radiographic spinal progression in early axSpA patients [7]. This advocates the early use of anti-inflammatory treatment in patients with early and active disease, in the hope that decreasing the disease activity will also slow down the radiographic progression. Identification of the predictors of spinal progression at baseline is important for clinical decision-making on aggressive anti-inflammatory treatment.

Chronic inflammation of the spine leads not only to new bone formation in axial joints and vertebral spaces, but also to bone resorption leading to osteoporosis, which is increased in axSpA [8]. It has been established that the generalised bone loss may be due to systemic inflammation and disease activity [9], and as disease activity in AS contributes to the rate of bone loss, osteoporosis is considered to be a manifestation of the disease itself, rather than a comorbidity [10].

Low bone mass in axSpA is a result of increased bone resorption through differentiation and activation of osteoclasts caused by inflammation. Therefore, bone mass changes reflect the severity of persistent inflammation, rather than a time-specific inflammatory state. Trabecular bone loss has been clearly and repeatedly demonstrated in the spines of patients with axSpA [11]. In a recent study, trabecular bone microarchitecture was found to be associated with spinal structural damage, as well as systemic inflammatory markers [12]. Furthermore, inflammation on spinal MRI is related to low bone mass in patients with nr-axSpA [13]. These studies show the site-specific relationship between low bone mass and inflammation in the spine. Earlier studies also reported the association between low bone mass in AS and spinal structural damage [14–16]; however, despite the identification of this relationship, it is not yet known whether the presence of low bone mass can independently predict radiographic spinal progression in axSpA patients.

The aims of the present study were therefore to evaluate the association between low bone mass and the formation of new syndesmophytes, and to investigate whether low bone mass independently predicts radiographic progression in axSpA patients.

New syndesmophytes had developed in 22 (21%) patients at the 2-year follow-up. The patient group with new syndesmophytes had a higher percentage of current smokers at baseline and patients with a longer symptom duration than the patient group without new syndesmophytes (p = 0.038 and p = 0.035, respectively). Patients with new syndesmophytes had a higher frequency of syndesmophytes at baseline and a higher baseline mSASSS score than those who did not develop new syndesmophytes (p = 0.001 and p = 0.015, respectively). Low BMD in any site (lumbar spine, femoral neck or total hip) at baseline was more frequent in patients with new syndesmophytes (p = 0.047). There were no significant differences in disease activity scores, inflammatory markers and treatment agents.

Table 3 presents the mean number of syndesmophytes at baseline and 2-year follow-up for the total group and the groups with or without low BMD at baseline. The number of syndesmophytes significantly increased in the period from baseline to 2 years in the total patient group (0.4 over 2 years) and both subgroups dichotomised according to low BMD. In the group of patients without low BMD at baseline, the mean increase was 0.3 over the 2 years, while in those with low BMD at baseline it was 1.2 (Fig. 1a). New syndesmophytes after 2 years occurred in 15% of patients without low BMD and in 37% of patients with low BMD (p = 0.047; Fig. 1b).

Group	Baseline syndesmophytes	2-year syndesmophytes	p value
Total patients (n = 119)	2.3 ± 4.8	2.7 ± 5.4	< 0.001
Normal BMD (n = 100)	2.1 ± 4.7	2.4 ± 5.0	< 0.001
Low BMD (n = 19)	3.0 ± 5.6	4.2 ± 7.4	< 0.001

Values presented as mean ± standard deviation

BMD bone mineral density

Variable	Univariable analysis			Multivariable analysisa
	OR	95% CI	p value	OR	95% CI	p value
Male sex	0.5	0.2–1.5	0.223
Age (years)	1.1	1.0–1.1	0.019
Symptom duration (years)	1.1	1.0–1.1	0.064
BMI (kg/m2)			0.817
< 18.5	1.0	0.2–5.3	0.964
18.5–22.9	1.0 (reference)
23.0–24.9	1.4	0.5–4.0	0.487
≥ 25.0	1.5	0.6–4.2	0.408
Current smoking	4.1	1.7–9.6	0.001	3.0	1.1–7.7	0.025
HLA B27-positive	1.3	0.3–6.7	0.743
Peripheral arthritis	0.5	0.2–1.6	0.269
Radiographic sacroiliitis	4.1	1.2–14.7	0.029
Presence of syndesmophytes	5.7	2.4–13.8	< 0.001	4.6	1.8–11.9	0.002
BASFI score, per 1 point	1.0	0.8–1.2	0.678
BASDAI score ≥ 4	0.9	0.7–1.1	0.328
ASDAS-CRP			0.313
Low (< 1.3)	1.0 (Reference)
Moderate (< 2.1)	1.2	0.4–3.6	0.769
High (≤ 3.5)	0.3	0.2–1.3	0.135
Very high (> 3.5)	0.9	0.2–3.2	0.839
Increased ESR (≥ 20 mm/h)	1.1	0.5–2.4	0.842
Increased CRP (≥ 5 mg/L)	1.4	0.6–3.3	0.431
Patients on NSAIDs	0.7	0.2–3.1	0.716
Patients on sulfasalazine	0.7	0.3–1.9	0.530
Patients on TNF inhibitors	1.6	0.7–3.7	0.298
Patients on calcium	3.0	0.8–11.3	0.097
Patients on vitamin D	2.1	0.4–10.1	0.336
Low BMD, any site (Z score ≤ − 2.0)	3.0	1.1–8.3	0.033	3.6	1.2–11.2	0.028

mSASSS Modified Stoke Ankylosing Spondylitis Spinal Score, OR odds ratio, CI confidence interval, BMI body mass index, BASFI Bath Ankylosing Spondylitis Functional Index, BASDAI Bath Ankylosing Spondylitis Disease Activity Index, ASDAS Ankylosing Spondylitis Disease Activity Score, CRP C-reactive protein, ESR erythrocyte sedimentation rate, NSAID non-steroidal anti-inflammatory drug, TNF tumour necrosis factor, BMD bone mineral density

^aAdjusted for age, smoking, radiographic sacroiliitis, presence of syndesmophytes and presence of low BMD at any site at baseline

In this longitudinal observational study, we investigated the formation of new syndesmophytes in the spines of patients with axSpA over a period of 2 years, and identified predictors of new syndesmophyte formation and spinal radiographic progression. About 20% of the patients developed a new syndesmophyte over 2 years. The presence of a syndesmophyte at baseline and low BMD were predictors of the formation of new syndesmophytes and significant mSASSS progression.

Abnormal bone metabolism in axSpA is characterised by pathological new bone formation in the cortical zone of the vertebrae and the loss of trabecular bone from the centres of the vertebral bodies. Osteoproliferation leads to syndesmophytes, while the loss of trabecular bone leads to low BMD [26].

This study is the first to demonstrate that low BMD predicts radiographic progression in axSpA. The main determinants of low BMD in axSpA patients are systemic inflammation and bone-specific inflammation [27]. The inflammatory process is associated with altered systemic bone remodelling, increased bone resorption and impaired bone formation resulting from the effects of inflammatory mediators on the differentiation and activity of osteoclasts and osteoblasts. Proinflammatory cytokines can influence osteoclastogenesis and osteoblastic activity [28]. Thus, the presence of low BMD in axSpA is considered to be a result of altered bone remodelling caused by persistent inflammation. In the present study, serum levels of ESR correlated with BMD values of the lumbar spine, femoral neck and total hip, and Z scores of the femoral neck and total hip. Additionally, CRP levels correlated with the Z score at the lumbar spine and femoral neck (data not shown).

Bone loss resulting from chronic inflammation and the associated changes in bone microarchitecture have been proposed as a potential driving mechanism for the ankylosing process [29]. The inflammatory process induces bone loss, which affects the microarchitecture in the trabecular bone, thereby leading to instability. Reduced bone strength triggers a stabilising anabolic effort that results in bone formation. Trabecular and cortical compartments appear to have different reactions to inflammation; in axSpA, inflammation has a direct effect on the trabecular bone of the vertebrae, but not on the cortical bone [8]. As persistent inflammation in the trabecular bone of the vertebral bodies may prevent the anabolic response from correcting the instability, new bone formation in the cortical bone of the vertebrae may be increased [30]. This would result in the formation of syndesmophytes: compensatory stability for the spine but with a loss of normal mobility [31].

Another explanation is that low BMD may represent the presence of a repairing area that was affected by active inflammation in the past. There is increasing evidence that new bone formation in axSpA is the consequence of previous inflammation in the subchondral bone marrow, with the appearance of granulated repair tissue occurring as a mandatory intermediate step, with this then being followed by new bone formation. It has been proposed that the best current treatment for the prevention of bone formation is the early and effective suppression of bony inflammation [32]. Additionally, if it is possible to detect the presence of repair tissue, the risk of new syndesmophyte formation could also be predicted. The presence of a low BMD means that the repair process may become apparent, to compensate for bone loss resulting from the inflammatory process. Therefore, the presence of low BMD may represent areas with ongoing repair affected by inflammation, such as fatty lesions on MRI.

In the current study, the presence of baseline syndesmophytes in axSpA was found to be the strongest predictor for the formation of new syndesmophytes. This finding is consistent with those of earlier studies. Similar results with respect to spinal radiographic progression have been found in early axSpA patients, as well as in AS patients [5, 33, 34]. However, a substantial proportion of patients with baseline syndesmophytes do not show progression over 2 years. Furthermore, it is arguable whether the presence of baseline syndesmophytes should be considered a true predictor, because the patients already exhibited the features that the model was designed to predict [5].

Smoking is reported to be associated with spinal radiographic progression in early axSpA [6] and the radiographic severity of AS [35], although the exact mechanisms for the influence of smoking on radiographic progression are not known. In the present study, smoking was associated with new syndesmophyte formation only in the univariable analysis, whereas it was independently associated with significant mSASSS progression over 2 years after adjustment for confounding factors. Only the status of smoking at baseline was included in the analysis, and the relatively small number of patients may have influenced the inconsistent results with respect to smoking. The influence of smoking on new syndesmophyte formation should be studied in a larger cohort.

The presence of radiographic sacroiliitis was significantly associated with spinal radiographic progression over 2 years, but this significant association was not present in the multivariable analysis. This finding is consistent with a previous result in early axSpA patients [6]. In the present study, 90% of patients with low BMD had radiographic sacroiliitis. This high proportion of radiographic sacroiliitis in the patients with low BMD could have affected the result. The exact effect of radiographic sacroiliitis on spinal progression should be clarified in a large patient cohort including non-radiographic axSpA patients with low BMD.

Although we did not observe a significant result in this study, systemic inflammatory markers were reported as predictors of radiographic progression in patients with early axSpA in the German Spondyloarthritis Inception Cohort (GESPIC) [6]. This previous study included early axSpA patients with short symptom duration (mean 4.2 years) and low mSASSS (mean 4.25) [6]. In this study, the mean symptom duration and mSASSS were 8.6 years and 7.6 points. These differences suggest that our study patients had more severe and longstanding symptoms and were advanced patients. In a prospective observation of a cohort with longstanding AS, inflammatory markers did not emerge as independent predictors, as per the results of the present study [5]. These discordant results could possibly be explained by the differences in disease duration and structural damage severity. Furthermore, we only analysed the associations between baseline ESR and CRP measurements and new syndesmophyte formation, not time-averaged inflammatory markers. Baseline inflammatory markers may be less reflective of the status of persistent systemic inflammation than time-averaged values. Therefore, our results for the predictive role of systemic inflammation should be interpreted with caution.

Inflammation plays a key role in bone loss in axSpA, and anti-inflammatory drugs are expected to have a beneficial effect on bone through both the increased mobility related to pain relief and the direct effects on bone [28]. The best treatment for the prevention of bone formation/progression is currently the early and effective suppression of bony inflammation [32]. Our results suggest that successful anti-inflammatory treatment reduces inflammation and allows the bone metabolism to normalise, thereby taking away the compensatory anabolic response that leads to new bone formation in the cortical bone of the spine. Although we did not find a beneficial effect of NSAIDs or TNF inhibitors in this 2-year follow-up study, recent long-term follow-up data show that TNF inhibitors suppress radiographic spinal progression [36, 37]. Thus, if patients are treated for a longer time with a TNF inhibitor (preventing new occurrences of the sequence of inflammation, repair and new bone formation), or if they are treated early in the course of their disease, such a treatment seems to be effective in retarding the process of new bone formation [32]. Taken together, active anti-inflammatory treatment is crucial for the prevention of spinal ankylosis in young axSpA patients, especially those with low BMD.

This study has some limitations. First, the number of axSpA patients was relatively small; therefore, the regression analysis could be underpowered. In our patients, those with low BMD received more calcium agents and vitamin D treatments, and these agents could have affected the bone metabolism. Although calcium and vitamin D intake were not significant in the regression analysis, their effects on spinal progression need to be clarified in a large cohort including more patients with low BMD. The BMD in axSpA can be affected by the presence of syndesmophytes or other structural lesions such as an ankylosed posterior arch and periosteal bone formation. Therefore, the BMD of the lumbar spine in patients with syndesmophytes could have been overestimated. Lastly, it is known that there is an association between serum levels of sex hormones and BMD in AS [38], but these were not measured in this study.

The presence of low BMD and syndesmophytes at baseline were independently associated with the formation of new syndesmophytes in young axSpA patients. Current smoking, syndesmophytes and low BMD at baseline were also associated with significant mSASSS progression. Effective anti-inflammatory treatment may modify radiographic spinal progression in young axSpA patients with low BMD. Our findings require confirmation in other large cohorts of axSpA patients.