Introduction
Ankylosing spondylitis (AS) is a chronic inflammatory rheumatic disease of the axial skeleton, which is characterized by low back pain and stiffness for more than 3 months that improves with exercise, but is not relieved by rest. Other important symptoms are restriction of motion of lumbar spine and limitation of chest expansion. Typical radiological features of AS are sacroiliitis on X-ray and bridging syndesmophytes of the spine, which usually take many years to develop, and besides that, not all AS patients develop these syndesmophytes. The first symptoms usually arise at an age younger than 30 years. Men are more often affected than women, and 90–95 % of the AS patients are HLA-B27 positive. The prevalence of AS ranges between 0.1 and 1.4 % in Europe and is partly dependent on the prevalence of HLA-B27, which differs in populations with different ethnic backgrounds [
1‐
3].
Osteoporosis in terms of decreased bone mineral density (BMD) is a common complication in AS patients. The prevalence ranges from 19 to 62 % [
4‐
6]. The decrease of BMD can be found both in the hip as well as in the spine and depends on disease duration and presence of syndesmophytes of the spine [
7,
8]. Besides osteoporosis, also vertebral fractures are an important complication of long-standing AS. The prevalence described of known vertebral fractures ranges between 1 and 19 %; however, usually most of them remain unrecognized [
4,
9‐
12]. In many cases, neither the patient nor the physician is aware of this increased fracture risk among the relatively young and predominantly male AS population which can result in delay of diagnosis and complications.
Osteoporosis in AS with a long disease duration has been well known for quite some time now [
9,
13‐
17]. Currently, the treatment options for AS have improved since the use of TNF-blocking agents, and the diagnosis is made at an earlier stage of the disease. The knowledge of the prevalence of decreased BMD in this early group, however, is limited despite the fact that a decrease of BMD increases the fracture risk [
8,
10,
18,
19]. For this reason, we decided to review the literature in order to investigate the prevalence of decreased BMD in patients with AS with a relatively short disease duration (<10 years). Secondly, we focused at the clinical relevance of low BMD in this early population, especially in relation with the risk of vertebral fractures, since vertebral fractures are a major and serious cause of morbidity in long-standing AS patients.
Methods
All patients included in the studies we used for this review had to fulfill the diagnosis of ankylosing spondylitis according to the modified New York criteria [
2]. BMD was defined according to the established criteria of the World Health Organization (WHO), which is based on data of postmenopausal women as no criteria of osteopenia and osteoporosis are available for males. Normal bone density was defined as
T score ≥ −1.0, osteopenia as −2.5 <
T score < −1.0, and osteoporosis as
T score ≤ −2.5 [
20,
21]. The
T score corresponds to the number of standard deviations (SD) from any result of the peak bone mass. We defined “decreased or low BMD” in this review as
T score ≤ −1.0 SD, including both osteopenia and osteoporosis.
For this review, disease duration was defined as “time since diagnosis” according to the modified New York criteria [
2]. However, in most articles, the definition of disease duration was not available or not clearly described, and when described, in most cases “time since diagnosis” was used as disease duration. Therefore, we decided that disease duration (defined as “time since diagnosis”) of less than 10 years was the inclusion criterion for our search, and not symptom duration (defined as “time since first symptoms”).
The diagnosis of AS, disease duration of less than 10 years, and the outcome BMD with prevalence numbers were our main inclusion criteria. For this review, we only used BMD data of the articles that were acquired by dual X-ray absorptiometry (DXA), and all other measurements performed were left out. Furthermore, we only used measurements of the lumbar spine and femoral neck. Other measurements will be briefly mentioned.
The databases used were PubMed, MEDLINE, Embase, and Google Scholar with the following free search terms: “ankylosing spondylitis,” “osteoporosis,” and “bone mineral density.” The search in PubMed was also performed with these free terms as MeSH terms. In our search, we always used the Boolean operator “AND.” Through Web of Science, we looked at which other articles had quoted our main articles. To expand our search even more, we also used the references mentioned by the articles and reviews we found. Case reports were excluded because they have primarily a signaling function, and extrapolation from their results is not valid.
Statistics
An analysis was made of the overall prevalence of decreased BMD and for the particular location lumbar spine or femoral neck. This was done by adding up the number of patients having a decreased BMD at that particular location. Subsequently, the total number of patients found with decreased BMD was divided by the total number of patients involved in those articles.
Discussion
The literature search for the prevalence of decreased BMD and osteoporosis in AS within the first decade after diagnosis revealed a high mean prevalence of decreased BMD between 51 % of the femoral neck and 54 % of the lumbar spine. Moreover, the prevalence of osteoporosis was 16 % for lumbar spine and 13 % for femoral neck. This corresponds with prevalence numbers found in long-standing AS patients. The variation in prevalence in the studies reviewed is probably mainly caused by differences in the presence or exclusion of cases with risk factors for low BMD between the included patients and the heterogeneity of the populations. However, the fact that the various studies performed within different populations show an almost equally raised prevalence of low BMD strengthens the conclusion that high prevalence of low BMD is already present at the early stages of AS. Obviously, this might be clinically relevant since low BMD is related to an elevated fracture risk in primary osteoporosis in postmenopausal women and men. In line with that, Cooper et al. and Vosse et al. have shown that vertebral fracture risk is elevated in patients with AS [
31,
46].
Some limitations of this review must be mentioned. Our choice of BMD outcome resulted in exclusion of several articles. We choose to only include studies which gave WHO defined T scores as well as a prevalence to be able to correctly compare and summon BMD outcomes. Also, for decreased BMD, we used the threshold of a T score < −1.0, as defined by the WHO. In ISCD 2007, decreased BMD for males younger than 50 years was described as a T score < −2.0. However, as all studies used osteopenia (−2.5 < T score < −1.0) or osteoporosis (T score ≤ −2.5) as BMD outcome measure, we used the osteopenia cutoff point as “low BMD.”
The seven studies used for this systematic review were conducted in different countries—Korea [
26,
27], France [
22], Germany [
7], Turkey [
24], India [
45], and The Netherlands [
19]. These countries differ in background risks of osteoporosis due to racial differences, dietary habits, and sun exposure. The differences in prevalence of decreased BMD between areas in a healthy population were shown by Lunt et al. They found that BMD of the spine appears to vary in 16 European populations, from 0.934 to 1.0670 g/cm
2 in women and from 1.047 to 1.262 g/cm
2 in men [
47‐
49]. Furthermore, the background prevalence of AS differs among several countries, roughly correlating with the prevalence of HLA-B27 [
1,
3]. Beside the genetic factors predisposing to AS, the prevalence of decreased BMD in (early) AS is also influenced by differences in environmental (risk) factors among countries, i.e., vitamin D deficiency, smoking, low body mass index, use of corticosteroids, and anti-TNF alpha therapy [
50‐
54]. Unfortunately, four of the studies reviewed excluded patients with common risk factors for osteoporosis, while the others did not. Besides that, the average participant number of the studies presented was 69 patients which is a relatively small number to extrapolate to a population prevalence [
7,
19,
22,
24,
26,
27,
45].
Next to general risk factors for low BMD, also disease-specific risk factors like high disease activity and inflammation are of influence on BMD [
9,
13,
55]. Some authors suggest that inflammation can explain the major part of the etiology of decreased BMD with cytokines as interleukin-1 and TNF alpha as primary mediators [
18,
23,
56‐
58], stating that low BMD in AS patients with inflammatory back pain is related to inflammation, whereas Aydin et al. as well as others found that bone loss was related to low serum sex hormone levels in AS [
14,
24]. However, Bronson et al. were unable to find a correlation between testosterone levels and decreased BMD [
15]. Remarkably, Karberg et al. found that more patients with syndesmophytes had low BMD than those without and therefore suggested that bone loss and bone growth occur parallel, also in early stages of AS [
7]. This could not be confirmed by the study of van der Weijden et al. because very few syndesmophytes were found in the early stage of disease in combination with a high prevalence of low BMD (46 %) [
19]. In this study, high disease severity indicated by impaired physical function (high BASMI and BASFI levels) and high CRP levels appeared to be strongly associated with low BMD AS within 8 years of onset. So, also in early AS, there seems to be an important role for inflammation [
19,
58].
A point worth mentioning regarding this review is the definition of disease duration. In this review, disease duration was defined as the “time since diagnosis.” Aydin et al. and van der Weijden et al. both mentioned in their articles the duration from the “moment of diagnosis” and the duration from the “moment of first symptoms” (both <10 years) [
19,
24]. Karberg et al., however, defined disease duration as the “time since first symptoms” (also <10 years) [
7]. Because the definition of Karberg et al. would only have shortened the disease duration (the mean delay between onset of first symptoms and diagnosis can run up to 8–10 years), this study was considered to be eligible for this review [
59‐
61]. And above that, the results from this study showed already a high prevalence of decreased BMD, which only indicates more that decreased BMD is already present in the (very) early stages of AS.
Another point worth mentioning are the difficulties of diagnosing osteoporosis in AS studies in general. This problem is due to the fact that DXA of lumbar spine has a low sensitivity in detecting decreased BMD in AS patients with a long disease duration due to occurrence of bridging syndesmophytes and ligamentous ossification that might increase the density of the axial skeleton [
8,
13,
28,
29]. The studies used in our review reported different percentages of patients with syndesmophytes varying between 7 and 55 % [
7,
19,
27,
45], whereas some others excluded patients with syndesmophytes [
22], and others did not address the issue of syndesmophytes at all [
24,
26]. The presence of syndesmophytes could have falsely increased the BMD of lumbar spine in these studies, but we found comparable prevalence numbers of low BMD in the hip and spine, so probably the relatively low numbers of syndesmophytes in this early group did not influence the data considerably. But if it did, the real BMD would have been even lower.
Considering the high prevalence of decreased BMD in patients with early AS, the question arises whether or not patients should be treated for this condition. The answer to this question depends on the clinical relevance of this observation. Low BMD becomes clinically relevant when it leads to an increased risk of vertebral and nonvertebral fractures since these fractures are a serious cause of morbidity and reduced quality of life [
62‐
64]. The link between decreased BMD and fracture risk is well known for the general healthy population [
65]. In later stages of AS, the risk of vertebral fractures is also significantly increased, and low BMD is an important risk factor for that complication [
9,
13‐
17,
33,
46]. Unfortunately, there is still no consensus whether or not decreased BMD in patients with early AS causes an increased fracture risk [
9]. Very few and small studies have been performed in early AS. Jun et al. and van der Weijden et al. found a positive correlation between low BMD and vertebral fractures [
25,
66]. Jun et al. found a relation between fractures and femoral BMD, whereas van der Weijden et al. found a relation with BMD of the lumbar spine. Nevertheless, other studies found no correlation with low BMD at all [
10,
13].