Clinical Rheumatology

, Volume 25, Issue 6, pp 835–839

The relationship between severity and extent of spinal involvement and spinal mobility and physical functioning in patients with ankylosing spondylitis

Authors

    • Physical Medicine and Rehabilitation DepartmentIzmir Training and Research Hospital
  • Fazil Gelal
    • Radiology DepartmentAtaturk Training and Research Hospital
  • Rezzan Gunaydin
    • Physical Medicine and Rehabilitation DepartmentIzmir Training and Research Hospital
Original Article

DOI: 10.1007/s10067-005-0180-3

Cite this article as:
Kaya, T., Gelal, F. & Gunaydin, R. Clin Rheumatol (2006) 25: 835. doi:10.1007/s10067-005-0180-3

Abstract

The present study was undertaken to determine the relationship between spinal radiological changes of ankylosing spondylitis (AS), spinal mobility, and physical functioning. Thirty-one patients diagnosed as AS according to the modified New York criteria for AS were included in this study. Three radiographic scoring methods were used to assess spinal damage. Severity of spinal involvement was assessed by using Stoke Ankylosing Spondylitis Spine Score (SASSS) and Bath Ankylosing Spondylitis Radiographic Index–Spine (BASRI–S). To assess the extent of spinal involvement, the total number of vertebrae showing radiological findings attributable to AS [number of vertebrae involved (NoVI)] was calculated according to the AS grading system defined by Braun et al. Statistical analysis, consisting of bivariate correlation, Spearman correlation, and multiple linear regression analysis, was performed using Windows Statistical Package for the Social Sciences 13.0. NoVI was negatively correlated with modified Schober and lateral spinal flexion and was positively correlated with occiput-to-wall distance and BASMI. SASSS was negatively correlated with the modified Schober. BASRI–S was negatively correlated with the modified Schober and positively correlated with BASMI. When BASMI and Bath Ankylosing Spondylitis Functional Index were taken as dependent variables, only the NoVI was found to be associated with BASMI. In our data, the extent of spinal involvement (NoVI) showed a more significant correlation with spinal measurements such as modified Schober and BASMI as compared with the other radiologic scores (SASSS and BASRI–S). Furthermore, because only the NoVI was found to be associated with BASMI, we can conclude that the extent of spinal involvement, which also includes thoracic vertebrae, affects spinal measurements.

Keywords

Ankylosing spondylitisRadiologyScoringSpinal

Introduction

Structural damage and radiological status are important objective endpoints in the assessment of ankylosing spondylitis (AS) [1, 2]. The magnitude of the structural changes in the spinal structures visible on roentgenograms is relevant for the outcome and the assessment of disease activity [3]. The radiological scoring methods in AS evaluate lumbar and cervical spine and measure the magnitude of spinal involvement [47]. However, no method has so far been described to evaluate thoracic spine [3, 5, 7]. Braun et al. [8] have suggested radiological grading of AS by observing radiological changes of spinal structures and has included thoracic spine in this evaluation. This grading method relies on the number of segments and vertebrae involved, thereby evaluating the extent of spinal involvement. In this cross-sectional study, we aimed to determine the relationship between spinal radiological changes of AS, spinal mobility, and physical functioning. We also investigated whether the extent or the magnitude of spinal involvement may correlate with the spinal mobility and the physical functioning.

Materials and methods

Thirty-one patients diagnosed with AS according to the modified New York criteria for AS were included in this study [9]. Patients with hip joint involvement were excluded because it is known that this could affect spinal measurements. Subjects were recruited from our outpatient department.

To evaluate disease activity and functional status, two patient-reported outcome measurement indices [Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) and Bath Ankylosing Spondylitis Functional Index (BASFI)] were used [10, 11]. These indexes were shown to be valid and reliable for Turkish population [12, 13]. Bath Ankylosing Spondylitis Metrology Index (BASMI) [14] and Bath Ankylosing Spondylitis Patient Global Index [15] were also assessed. All spinal measurements including finger-to-floor distance, occiput-wall distance, modified Schober, and lateral spinal flexion were performed at the same time of the day (early afternoon) in all patients. Lateral spinal flexion measurement was performed using the method suggested by Calin [16]. Severity of current pain and night pain and severity of morning stiffness were assessed using a 100-mm visual analog scale.

Anteroposterior (AP) pelvis, AP, lateral lumbar and thoracic spine, oblique lumbar spine, and AP-lateral cervical spine roentgenograms were obtained from the 31 patients. X-ray examinations were performed in a 500-mA unit. The magnitude of spinal involvement was assessed using BASRI–S and Stoke Ankylosing Spondylitis Spine Score (SASSS). BASRI–S is a reliable method ranging between 0 and 12 for grading radiographic changes in patients with AS. The radiographs used in this method are AP pelvis film, AP and lateral lumbar spine films, and a lateral cervical spine film [6]. The cervical spine was defined as extending from the lower border of C1 to the upper border of C7. Lateral lumbar spine radiogram included Th12 vertebra and upper sacral vertebrae and was used in the assessment of SASSS. For SASSS, all corners of each vertebra between the lower border of Th12 and the upper border of S1 are examined and scored 1 for erosion, squaring, or sclerosis; 2 for syndesmophyte formation; and 3 for total bridging, giving a maximum possible score of 72 [5].

Braun et al. [8] have defined five radiological stages in AS, relying on the spinal and sacroiliac bony changes. Stage I includes those patients who have grade II or higher bilateral sacroiliitis. Stages II–V assess the extent of spinal involvement by taking the number of affected vertebrae into consideration. This grading system relies on the extent of spinal involvement; it determines the number or percent of vertebrae and number of segment(s) involved. This method also evaluates zygapophyseal joints [8]. One spine segment indicates involvement of cervical, thoracic, or lumbar spine. In the oblique X-rays of the lumbar spine, the presence of arthritis in the upper or lower articular facet of the same vertebra was considered as the involvement of the apophysis. In patients with AS, radiological spinal involvement manifested as erosion, sclerosis, joint space narrowing, blurring at joint margins, pseudowidening, complete and incomplete bony fusion, bony bridging, squaring of vertebrae, disk calcification, trolley track sign, spondylitis, spondylodiscitis, squaring of vertebra, syndesmophytes, ligament ossification, ankylosis, and zygapophyseal joint arthritis [3, 8]. In this method, the total number of vertebrae in all three segments (cervical, thoracic, and lumbar) bearing any of the above findings were assessed. In our study, the number of vertebrae involved (NoVI) was considered as the parameter predicting the extent of spinal involvement. Both anterior and posterior corners of all vertebrae were evaluated. Radiographic scoring was performed by the same radiologist.

Statistical analysis

The correlation between the scoring methods was performed by using Spearman correlation analyses. Bivariate correlations (controlling for disease duration) was conducted for each X-ray method (NoVI, BASRI–S, and SASSS) with BASFI, BASMI, and the individual measures of spinal mobility.

The regression of the contribution of the radiological damage to the BASFI and BASMI was performed by using multiple linear regression analyses. These analyses were corrected for disease activity (BASDAI) and age. All analyses were performed using Statistical Package for the Social Sciences for Windows 13.0. The p<0.05 was considered significant.

Results

The radiographs of the 36 patients were obtained. However, five patients were excluded from the study due to the inadequate quality of the radiograms. Out of 31 patients, 29 were male, and two were female. The patients aged 36.22±8.12 (mean±standard deviation) years (range: 20–63) and disease duration was 105.12±77.33 (mean±standard deviation) months (range: 12–300). Patient data are shown in Table 1. All radiologic scores showed positive correlation with the disease duration (Table 2).
Table 1

The mean and the range of patient data

 

Mean±SD (N=31)

Min.

Max.

Age (year)

36.22±8.12

20.00

48.00

Length (cm)

170.58±6.84

155.00

185.00

Weight (kg)

65.67±10.36

43.00

85.00

Dd (month)

105.12±77.33

12.00

300.00

ESR (mm/h)

30.06±23.31

1.00

84.00

CRP (mg/dl)

2.10±3.37

0.20

18.80

Chest expansion

2.99±1.91

0.50

8.50

Modified schober

2.49±1.58

0.50

7.00

Finger-to-floor distance

24.53±13.31

0.00

58.00

Occiput-to-wall distance

11.25±6.74

0.00

28.00

Lateral spinal flexion

1.40±1.11

0.30

5.20

VAS (current pain) (0–10)

5.32±3.31

0.40

10.00

VAS-Night (0–10)

6.31±3.57

0.00

10.00

VAS-MS (0–10)

6.36±2.68

1.20

10.00

SASSS (0–72)

19.32±18.21

0.00

62.00

NoVI (0–24)

12.41±8.46

0.00

24.00

BASRI–S (0–12)

7.74±2.97

3.00

12.00

BASMI (0–10)

5.38±2.02

1.00

10.00

BASDAI (0–10)

5.70±2.33

1.23

9.6

BASFI (0–10)

4.80±2.36

0.81

10.00

BAS-G (0–10)

7.27±2.46

1.75

10.00

SD Standard deviation, Min. minimum, Max. maximum, Dd disease duration, ESR erythrocyte sedimentation rate, CRP C-reactive protein, VAS visual analog scale, VAS-Night severity of night pain, VAS-MS VAS-morning stiffness (severity of morning stiffness), and NoVI number of vertebrae involved

Table 2

The relationship of disease duration to the radiological scores and patient reported outcomes

 

BASMI

BASFI

BAS-G

BASDAI

SASSS

BASRI–S

NoVI

Disease duration

r

0.130

−0.200

0.063

−0.135

0.398*

0.486**

0.423*

p

0.484

0.280

0.738

0.470

0.026

0.006

0.018

*Correlation is significant at the 0.05 level (2-tailed)

**Correlation is significant at the 0.01 level (2-tailed)

NoVI was negatively correlated with the modified Schober (r=−0.572, p=0.001) and lateral spinal flexion (r=−0.365, p=0.047) and was positively correlated with occiput-to-wall distance (r=0.373, p=0.043) and BASMI (r=0.487, p=0.006). SASSS was negatively correlated with the modified Schober (r=−0.367, p=0.046). BASRI–S was negatively correlated with the modified Schober (r=−0.484, p=0.007) and was positively correlated with BASMI (r=0.475, p=0.008) (Table 3). To identify percentage of patients with abnormalities in the different parts of the spine according to the three different scoring methods, cut off level >1 was chosen. According to this cut off level, the percentage of patient having cervical spine abnormality was 54.83 for NoVI and 58.06 for BASRI–S. The percentage of patient having thoracic spine abnormality was 74.19 for NoVI. The percentage of patient having lumbar spine abnormality was 90.32 for both all three radiographic scoring methods.
Table 3

The relationship of each radiological score to the BASFI, BASMI, and spinal measurements

Radiological scores

BASFI

Chest expansion

Modified schober

Finger-to-floor

Occiput-to-wall

Chin-manubrium sterni

Lateral spinal flexion

BASMI

NoVI

r

0.269

−0.305

−0.572*

0.194

0.373*

0.246

−0.365*

0.487*

p

0.151

0.101

0.001

0.305

0.043

0.190

0.047

0.006

BASRI–S

r

0.242

−0.192

−0.484*

0.140

0.340

0.210

−0.312

0.475*

p

0.198

0.308

0.007

0.460

0.066

0.264

0.094

0.008

SASSS

r

−0.008

−0.057

−0.367*

0.020

0.212

−0.053

−0.163

0.228

p

0.968

0.766

0.046

0.917

0.260

0.780

0.388

0.226

*Correlation is significant at the 0.05 level (2-tailed)

When BASMI and BASFI were taken as dependant variables, it was found that of all the radiological scores, only NoVI was associated with BASMI. NoVI explained 30.5% of the variance in BASMI score (Table 4). None of the radiographic scoring methods associated with BASFI (Table 5).
Table 4

Association between BASMI and radiographic score(s)

Model

 

Coefficients

r2 change

r2 total

p

β

Std. error

BASMI

NoVI

0.095

0.042

0.305

 

0.031

Constant

1.758

1.671

 

0.305

0.302

Only significant predictors are represented in the table

Dependent variable: BASMI

Table 5

Association between BASFI and radiographic score(s)

Model

 

Coefficients

r2 change

r2 total

p

β

Std. error

BASFI

NoVI

−0.011

0.129

0.024

 

SASSS

−0.010

0.029

0.007

 

BASRI–S

0.088

0.312

0.012

 

Constant

−0.171

1.621

 

Discussion

There are many radiologic scoring methods evaluating the magnitude of radiologic findings in patients with AS [47]. In these methods, AP or lateral X-rays of the spine can be scored. No comparative data exist to judge what would be the optimal choice to include in a scoring method [17]. The AS radiographic scoring system defined by Braun et al. [8] includes the thoracic spine in the evaluation. In AS, spinal changes show a characteristic ascending pattern. In early stages, sacroiliac changes are noted. This is followed by thoracolumbar and lumbosacral junction involvement. As the disease progresses, midlumbar, upper thorax, and cervical vertebrae are involved [18]. However, no method has so far been described to score radiologic changes in the thoracic spine [3]. Zygapophyseal joints cannot be reliably evaluated using thoracic spine X-rays [4]. Therefore, we evaluated both the AP and lateral X-rays of the thoracic spine. Zygapophyseal joint evaluation is possible in the lumbar spine [4]. In the lumbar spine, zygapophyseal joint evaluation was performed using AP, lateral, and right and left oblique radiograms. Spoorenberg et al. [19] evaluated the reliability of radiologic scoring methods. According to their data, combined SASSS scores were most reliable with intra- and interobserver intraclass correlation coefficients (ICC) between 0.90 and 0.96. The ICC of the combined BASRI scores were 0.85 to 0.95. Although all films were read by an experienced radiologist (FG), his intraobserver variability was not assessed in our study. We agree that the assessment of intraobserver variability would have improved the study.

In our study, the extent of spinal involvement (NoVI) showed a positive correlation with both SASSS and BASRI–S (Table 6). This finding may show that the magnitude of involvement increases as the extent of involvement increases. It is already well known that upper segments of the vertebral column are involved as the disease progresses. Therefore, an increase in the magnitude of the present lesions may be expected as the number of the involved segments increases [18]; and SASSS and BASRI–S scores are expected to be high. NoVI showed a more significant correlation with BASRI–S, probably because both lumbar and cervical spines are evaluated in BASRI–S [6]; whereas SASSS evaluates only the lumbar spine [7]. Radiologic scores also correlated with the disease duration (Table 2). Furthermore, NoVI showed a more significant correlation with spinal measurements such as modified Schober and BASMI as compared with the other radiologic scores (SASSS and BASRI–S) (Table 3). NoVI was the only radiologic score that correlated with occiput-to-wall distance and lateral spinal flexion (Table 3). Therefore, in the assessment of patients with AS, the inclusion of lumbar spine only in the radiologic scores used for outcome measurement does not seem to be sufficient. The extent of thoracic spine involvement should also be evaluated. However, the extent of radiologic involvement cannot be considered to be apart from the magnitude of involvement. In our study, these two features correlated with each other. Nevertheless, the extent of radiologic involvement (NoVI) has shown a more significant correlation with the spinal measurements (Table 3). Creemers et al. [20] have modified SASSS and developed modified SASSS (mSASSS), which includes the evaluation of cervical and lumbar spine involvement. According to their data, this radiographic scoring system has better validity than SASSS. However, mSASSS takes only the structural damage in the anterior part of the lumbar and cervical spine into account. Therefore, Wanders et al. [21] have suggested that the mSASSS is not a perfect score to represent all possible radiographic abnormalities and exclusion of the thoracic spine may result in underestimation of true structural damage.
Table 6

Correlation of the radiologic scores with each other

 

NoVI

SASSS

BASRI–S

NoVI

r

1

0.786*

0.939*

p

.

0.000

0.000

SASSS

r

0.786*

1

0.709*

p

0.000

.

0.000

BASRI–S

r

0.939*

0.709*

1

p

0.000

0.000

.

*Correlation is significant at the 0.01 level (2-tailed)

As a matter of fact, as BASRI–S grades spinal involvement, it also measures the extent of involvement; but it does not evaluate thoracic spine [6]. In our study BASRI–S correlated only with modified Schober and BASMI; however, this correlation was weaker than that between NoVI and these spinal measurements. SASSS, evaluating lumbar spine, showed negative correlation with modified Schober but was unrelated to lateral spinal flexion and other spinal measurements (Table 3). SASSS notes the changes in the anterior and posterior corners of lumbar vertebrae [3, 5, 7]; this feature may explain why it is correlated with the limitation of motion along the sagittal plane and why it is unrelated to lateral spinal flexion. Dawes [5] has investigated the relationship between SASSS and various spinal measurements and has found that SASSS was related to modified Schober, occiput-wall distance, and finger-floor distance. However, he has not studied the relation between SASSS and lateral spinal flexion. Wanders et al. [21] have suggested that not every patient with radiographic structural damage has reduced spinal mobility and not every patient with reduced spinal mobility has radiographic damage. However, according to their data, lateral spinal flexion is almost always abnormal if radiological damage is severe.

When BASMI was taken as dependant variable, NoVI was found to be associated with BASMI. This is probably due to the addition of the thoracic spine to the evaluation. NoVI explained 30.5% variance in BASMI score (Table 4). In this study, NoVI determines the extent, but not the magnitude of spinal involvement.

Five patients were excluded from the study due to difficulties encountered in deciding whether a vertebra was abnormal. There are several articles addressing this difficulty. For example, it was reported that it may be difficult to decide if a vertebra shows squaring [6]. In our series, the exclusion of the five patients from evaluation was mainly due to the inadequate quality of the radiograms.

The radiologic scores we have studied were not associated with BASFI (Table 5). In the evaluation of this finding, the following points should be taken into account: First of all, this index is a subjective method of measurement, relying on the information given by the patient [10]. Second, impaired spinal mobility may not always be associated with impaired function. Wanders et al. [21] have stated that, patients can compensate for impaired mobility due to severe structural damage in parts of the spine to a certain degree. Third, all changes encountered in the radiograms of patients with AS result from inflammation. However, it is not clear whether these changes represent disease activity or recovery process [17].

In conclusion, NoVI correlates with spinal measurements such as modified Schober, lateral spinal flexion, occiput-to-wall distance and BASMI. The only radiologic score associated with BASMI is NoVI. Because of the finding that the NoVI is associated with BASMI score, we can conclude that the extent of spinal involvement affects spinal measurements. This is probably due to the use of an evaluation method defined by Braun et al. [3], which also includes thoracic vertebrae. It is possible to understand and define the disease to a better extent by the evaluation of thoracic spine involvement. However, evaluation methods measuring not only the extent but also the magnitude of thoracic spine involvement are required.

Copyright information

© Clinical Rheumatology 2006