Prevalence of low trabecular bone score and its association with disease severity and activity in patients with axial spondyloarthritis

Axial spondyloarthritis (axSpA) increases the risk of osteoporosis and vertebral fractures. Bone mineral density (BMD) measured by dual X-ray absorptiometry (DXA) has limitations in axSpA patients. Trabecular bone score (TBS) indirectly assesses bone microarchitecture and can be used to predict fracture risk. However, few studies have investigated the role of TBS in axSpA patients. The objective of this study were to compare TBS between axSpA patients and 1:1 sex- and age-matched healthy volunteers and determine factors associated with low TBS in axSpA patients. A cross-sectional study was conducted in two tertiary-care hospitals. A total of 137 axSpA patients and healthy volunteers were enrolled. Demographics, disease characteristics, and risk factors for osteoporosis were recorded. TBS, BMD at the lumbar spine, hip, and vertebral fractures were assessed by DXA. Low TBS was defined as a TBS value < 1.230. Factors associated with low TBS were examined by logistic regression. Most patients were male (75.9%) and tested positive for HLA-B27 (88.3%). The mean (SD) age was 42.8 (12.0) years. The mean (SD) of TBS in the axSpA patients was lower than those in the healthy volunteers [1.402 (0.107) vs 1.440 (0.086), respectively; p = 0.002]. The mean (SD) of lumbar BMD in the axSpA patients was higher than in healthy volunteers [1.186 (0.212) vs 1.087 (0.124), p < 0.001], whereas the mean (SD) of femoral neck BMD in the axSpA group was lower than that in the healthy volunteers [0.867 (0.136) vs 0.904 (0.155), p = 0.038]. Disease severity as indicated by sacroiliac joint fusion and a high ASDAS score were associated with low TBS with the odds ratios (95% confidence interval) of 11.8 (1.2–115.4) and 5.2 (1.6–16.9), respectively. In conclusion, axSpA patients had a higher prevalence of low TBS than healthy volunteers. Sacroiliac joint fusion and a high ASDAS score were associated with low TBS.


Statistical analysis
Statistical analysis included computing the percentage frequencies for categorical variables and means (standard deviation, SD) or medians (interquartile ranges, IQR) for continuous variables.Comparisons of categorical variables were made using the chi-squared or Fisher's exact test, as appropriate.Continuous variables were tested for normality using a Shapiro-Wilk test and were compared using Student's t-test and one-way ANOVA or the Mann-Whitney U test and the Kruskal-Wallis test to compare two and three groups' means, respectively.Medians with IQR were used for non-normally distributed variables.Univariate and multiple logistic regression analyses were used to determine the relevant risk factors for low TBS.Only risk factors that had univariate associations of P-values < 0.20 were further considered in subsequent multivariate models.The lack of collinearity was confirmed by testing variance inflation factors.P-values < 0.05 were considered statistically significant.All statistical analyses were performed using SPSS software (IBM SPSS Statistics for Windows, Version 23.0.Armonk, NY: IBM Corp.).

Ethics approval and consent to participate
The ethical approval was approved by the Institutional Review Board of the Royal Thai Army Medical Department and the Human Research Ethics Committee of Khon Kaen University.The reference numbers are R178h/62 and HE631141, respectively.This research complied with the Declaration of Helsinki.The written informed consent was obtained from all participants and/or their legal guardians prior to entry to the study.

Results
As shown in Table 1, the study population recruited 137 axSpA patients (115 from Phramongkutklao Hospital and 22 from Khon Kaen University Hospital) and 137 matched healthy volunteers.The majority of participants were male (75.9%), and 34.3% of the women were post-menopausal.The mean (SD) of age and mean duration of disease were 42.8 (12.0) years and 86.8 (96.9) months, respectively.HLA-B27 was found in 88.3%, whereas the family history of axSpA was found in only 13.1%.Patients with axSpA had concomitant anterior uveitis and psoriasis (5.8% and 1.5%, respectively).In axSpA patients, alcohol consumption who used greater than or equal to 3 units/day was found in 19.7%, whereas patients with smoking including ex-smoker were found in 35.8%.
The mean LS-BMD was significantly higher in axSpA patients compared with the healthy volunteers.In contrast, the mean FN-BMD was significantly lower in axSpA patients than in the healthy volunteers.There was no significant difference in TH-BMD between the two groups.There were 13 patients (9.6%) within the axSpA group who met the criteria for osteoporosis or bone mass below the expected range for age.The prevalence of low BMD/osteoporosis at TH was significantly higher in axSpA patients than that in the healthy volunteers, but it was not significantly different between axSpA patients and the healthy volunteers at LS and FN (   www.nature.com/scientificreports/1.402 (0.107) and 1.444 (0.107), respectively (p = 0.002), and exhibiting a statistical power of 94.7%.The computed effect size was 0.391.The prevalence of degraded bone/low TBS in the axSpA patients was also significantly higher than that in the healthy volunteers (7.35% vs 1.46%, p = 0.010).Due to the limited number of patients in the low TBS and osteoporosis group, the clinical significance of these findings might be unclear, thus warranting a cautious interpretation.
As depicted in Table 2, axSpA patients with degraded bone/low TBS group had higher CRP levels, disease activity scores (including PGA, BASDAI, BASFI, BASMI, and ASDAS), and prevalence of grade 4 sacroiliitis and current NSAIDs use than those in normal and high TBS groups.In addition, axSpA patients with degraded bone/low TBS group had a significantly shorter disease duration when compared with other groups.We found that patients with axSpA who had never received anti-TNF or anti-OP medication had lower TBS than patients who had received those medications.(p = 0.037 and p = 0.028, respectively).Furthermore, axSpA patients with degraded bone/low TBS group had significantly lower BMD at LS, FN, and TH than those with intermediate/ normal TBS groups.The radiographic VFs were found in 10 patients (7.5%, 10/134).The prevalence of VFs was significantly higher in patients with degraded bone/low TBS group as compared with patients with intermediate and normal TBS groups (30.0%vs 6.67% vs 5.5%, p = 0.019, respectively).

Factors associated with the trabecular bone score
In the univariate linear regression, ex-or current smoker, alcohol intake ≥ 3 units/day, BMI, current NSAIDs use, all parameters of disease activity, presence of syndesmophyte, and grade 4 sacroiliitis were associated with low TBS.In multivariate linear regression analysis, TBS was negatively associated with female gender, ex-or current smoker, alcohol intake ≥ 3 units/day, ASDAS, and grade 4 sacroiliitis, whereas BMI was positively associated with TBS.
The univariate logistic regression model showed that higher disease severity measurements including PGA, ESR, hsCRP, BASDAI, BASMI, BASFI, ASDAS, grade 4 sacroiliitis, and the presence of syndesmophyte, alcohol intake ≥ 3 units/day, shorter disease duration, and PPI uses were significantly associated with degraded bone/ low TBS.The ASDAS and grade 4 sacroiliitis were respectively selected to represent current disease severity and damage in multivariate models.In multivariate logistic regression, ASDAS, and grade 4 sacroiliitis were independently associated with degraded bone/low TBS with the adjusted odds ratios (confidence interval: CI) of 5.228 (1.611-16.963)and 11.820 (1.211-115.412),respectively.Short disease duration was also found to be associated with degraded bone/low TBS; although the association was modest with an adjusted odds ratio (CI) of 0.982 (0.966-0.998), (Table 3).
Combining the results from TBS and BMD, axSpA patients who had both degraded bone/low TBS and osteoporosis/low bone mass at the femoral neck or total hip were at the highest risk of having VFs as compared to patients with normal TBS and FN-BMD group with the OR (95%CI) of 15.8 (0.88-285.48), p = 0.061 for both femoral and total hip site.In contrast, axSpA patients that had normal BMD and degraded bone/low TBS at the www.nature.com/scientificreports/lumbar spine had an increased risk of VFs as compared with those with normal BMD and normal TBS at the same site with the OR (95%CI) of 11.3 (1.58-81.23),p = 0.016), (Fig. 2).

Discussion
In this study, patients with axSpA had significantly lower TBS values than sex-age-matched healthy volunteers.The prevalence of degraded bone/low TBS was higher in axSpA patients when compared with healthy volunteers.VFs were more common in patients in the degraded bone/low TBS group than in the intermediate and normal groups.Female gender, smoking status, alcohol intake greater than or equal to 3 units/day, shorter disease duration, elevated hs-CRP, higher disease activities, current NSAIDs use, and not receiving anti-TNF or anti-OP drugs were all significantly associated with lower TBS in patients with axSpA.The findings from this study were consistent with previous studies [25][26][27][28] .High disease activity related to inflammation and new bone formation, impaired function, and spinal mobility as assessed by BASFI and BASMI were found to be associated with low TBS in axSpA patients.These findings were supported by an MRI study of the spine, which revealed that local inflammation, known as bone marrow edema, was detected in trabecular bone regions 29 as well as the presence of new bone formation on plain films, such as fusion of sacroiliac joints and syndesmophytes, were associated with low TBS in axSpA patients [30][31][32] .
A meta-analysis of seven longitudinal studies and one randomised controlled trial reported that anti-TNF treatment increased LS-BMD in AS patients by 5.1% and 8.6% after 1 and 2 years, respectively 33 .Another prospective study in 12 AS patients found that, despite a significant increase in LS-BMD, there was no significant improvement in TBS after 2 years of anti-TNF treatment 34 .This study found that axSpA patients who did not receive anti-TNF treatment had lower TBS than those who did and that low TBS was associated with high disease activity.These findings emphasized the significance of controlling disease activity with anti-TNF agents or other agents, as they could improve bone quality (TBS) and bone mass (LS-BMD) in axSpA patients.
In this study, alcohol and smoking were associated with degraded bone/low TBS in axSpA patients.Several studies in postmenopausal women found that alcohol consumption and smoking were strongly associated with low BMD 33,35 ; although, these associations had not previously been reported in axSpA patients 27,28 .Furthermore, those risk factors have been linked to increased radiographic progression in patients with axSpA 36,37 .As a result, physicians should encourage axSpA patients to discontinue the modified risk factors.
This study found that shorter disease duration was independently associated with degraded bone/low TBS; although, the association was modest.This finding supported the Danish Health Registries' findings that patients with AS had an increased risk of clinical vertebral and non-vertebral fractures, with the highest risk occurring within the first 2.5 years after diagnosis 38 .In addition, previous studies demonstrated that a decrease in LS-BMD was associated with active inflammatory lesions at the lumbar spine and sacroiliac joints in early AS patients 39,40 , Table 3. Linear regression for TBS and logistic regression for degraded bone/low TBS in patients with axSpA.*p < 0.05.# p < 0.20.† Adjusted for gender, ex-or current smoker, alcohol intake ≥ 3 units/day, BMI, current NSADIs use, current PPIs use, Hb, BASDAI, BASMI, BASFI, ASDAS, grade 4 sacroiliitis, presence of syndesmophyte.‡ Adjusted for alcohol intake ≥ 3 units/day, duration of disease, current PPIs use, BASDAI, BASMI, BASFI, ASDAS, grade 4 sacroiliitis, syndesmophyte.§ Statistical analysis cannot be conducted due to a zero value within one of the TBS groups.www.nature.com/scientificreports/and the 2-year follow-up of the DESIR cohort also showed that the prevalence of low BMD at the lumbar spine was 22.4% in patients with early axSpA 41 .Patients with a briefer disease duration may experience more severe symptoms and disease progression, leading them to seek medical attention at an earlier stage.The appropriate bone health assessment should be performed as part of an initial treatment plan in axSpA patients.AxSpA patients in the degraded/low TBS groups had a higher prevalence of VFs and a higher risk of developing new VFs when compared to those in the normal or intermediate TBS groups 27,28,[42][43][44] .Degraded bone/low TBS had higher sensitivity than FN-BMD and LS-BMD for detecting VFs in axSpA patients without losing specificity.Although the 2019 ACR recommendations suggested patients with axSpA should be screened for osteoporosis with a DXA scan of the spine and hip 10 , the results of previous studies together with findings from this study  suggested TBS provided additional benefit for vertebral fracture assessment than LS-BMD and FN-BMD alone since vertebral fractures in axSpA patients from this study occurred at a normal BMD.
The clinical application of TBS for fracture risk assessment in patients with axSpA is not well established.TBS could be used in two ways to assess fracture risk in axSpA patients.First, degraded bone/low TBS level could be a tool for fracture risk assessment in AS patients.The cut-off for degraded bone/low TBS used in this study was derived from a meta-analysis of 14 prospective population-based cohorts from countries around the world examining TBS in fracture risk prediction 19 .However, it is unclear whether this threshold could be used in axSpA patients of any ethnicity since ethnicity affects the fracture-predictive ability of TBS 45 .Secondly, the value of TBS could be incorporated in FRAX (fracture assessment tool) along with clinical risk factors and FN-BMD.Further research into the clinical use of FRAX and TBS is needed to determine the appropriate adjustment to assess fracture risk in axSpA patients.
There were some limitations in this study.First, this study did not examine some other risk factors for osteoporosis such as vitamin D level, serum calcium, and iPTH.Second, there were limited numbers of patients with low TBS and VFs and only a few patients had other major osteoporotic fractures.Third, as a cross-sectional study, this study data might not represent overall disease activities in patients with axSpA, although the results showed that all parameters of disease activity were consistent with the severity of the disease.Furthermore, only a few patients were menopause or had already received anti-osteoporotic treatment; although, after excluding post-menopausal status or anti-osteoporotic treatment, the results were indifferent.

Conclusion
Patients with axSpA have lower TBS than healthy volunteers.The proportion of axSpA patients with degraded bone/low TBS was higher than the proportion of healthy volunteers.TBS was found to be independently associated with gender, former or current smoker, ≥ 3 units/day alcohol consumption, low BMI, high disease activity including PGA, BASDAI, BASFI, BASMI, and ASDAS, and fusion of sacroiliac joints.Therefore, TBS is a good non-invasive tool for assessing fracture risk in axSpA patients.TBS can be used alone or in conjunction with BMD measurement to identify axSpA patients at high risk for vertebral fractures.

Figure 1 .
Figure1.The prevalence of vertebral fractures (two cases were not included in which the radiologist could not interpret vertebral fractures due to artifacts) (VFs) stratified by low bone mineral density (BMD) or osteoporosis (OP) at the lumbar spine (LS), total hip (TH), femoral neck (FN), and trabecular bone score (one patient, whose TBS data could not be analysed due to morbid obesity, was excluded) (TBS) risks in patients with axial spondyloarthritis (axSpA).*p < 0.05, compared with normal TBS group.

Figure 2 .
Figure 2. The prevalence of vertebral fractures, stratified by trabecular bone score (TBS) and osteoporosis/low bone mineral density (BMD) according to DXA.Asterisk: statistically significant when compared with patients with ax-SpA with normal BMD and normal TBS.The values displayed in the center of each bar represent the odds ratios for vertebral fractures when compared to individuals with both normal TBS and normal BMD at the lumbar spine, total hip, and femoral neck in patients with axSpA.

Table 2 .
Characteristics of patients with axial spondyloarthritis, stratified by trabecular bone score.AxSpA axial spondyloarthritis, TBS trabecular bone score, BASDAI The Bath Ankylosing Spondylitis Disease Activity Index, ASDAS The Ankylosing Spondylitis Disease Activity Score.*Statistically significant when comparing among the three groups (degraded bone group, partially degraded bone group, and normal bone group).† Statistically significant when comparing the degraded bone group with the normal bone group.$ Included individuals with Z-score below -2.0.# Not included 3 cases in whom the radiologist could not interpret VFs due to artifact.§ Statistically significant when comparing the degraded bone group with the partially degraded bone group.¶ Statistically significant when comparing the partially degraded bone group with the normal bone group.