SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Objective

To determine the diagnostic utility of different spinal inflammatory lesions assessed by whole-body magnetic resonance imaging (MRI) in patients with ankylosing spondylitis (AS) or with recent-onset inflammatory back pain (IBP) compared with healthy controls.

Methods

We scanned 35 consecutive patients with AS fulfilling the modified New York criteria, 25 patients with IBP of <24 months' duration (both groups were age ≤45 years and had a Bath Ankylosing Spondylitis Disease Activity Index score ≥4), and 35 healthy age- and sex-matched volunteers using whole-body MRI STIR sequences of the spine. MRIs were independently assessed in random order by 3 readers blinded to patient identity. Inflammatory spinal lesions were recorded consistent with definitions proposed by the Canada/Denmark International MRI Working Group: vertebral corner inflammatory lesions (CIL) and noncorner inflammatory lesions in central sagittal slices and lateral inflammatory lesions (LIL) in lateral slices. Concordantly scored lesions for the 3 possible reader pairs were used in the analysis of sensitivity, specificity, likelihood ratios (LRs), and areas under the curve for the entire spine and by spinal segment.

Results

Diagnostic utility was optimal when ≥2 CIL were recorded (for patients with AS, values for sensitivity, specificity, and positive LR were 69%, 94%, and 12, respectively, and for patients with IBP were 32%, 96%, and 8, respectively). LIL had high specificity (97%) but low sensitivity (31%). Nine controls had ≥1 CIL, but only 2 controls had >2 CIL.

Conclusion

Diagnostic utility of STIR MRI for AS is optimal when ≥2 CIL are present. A single CIL can be found in up to 26% of healthy individuals.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Inflammation of the spine in patients with ankylosing spondylitis (AS) causes substantial morbidity and may eventually lead to disability by progressive ossification of the axial skeleton. In AS, the thoracic spine is the second most frequent region affected by inflammation after the sacroiliac (SI) joints (1, 2). Plain radiography is an inadequate imaging modality in early spondylarthritis (SpA), prior to the appearance of structural damage, because it does not directly detect inflammatory lesions of the spine, and also due to technical limitations such as the superposition of lung tissue on the thoracic spine.

A growing number of imaging studies in AS over nearly 2 decades have consistently shown that magnetic resonance imaging (MRI) is the most sensitive imaging modality for the detection of inflammatory lesions in the spine (3–5). The recently introduced technique of whole-body MRI, which is based on multichannel technology that uses several coils concurrently, permits scanning of the entire spine within <30 minutes (6). This imaging tool may both contribute to early diagnosis of AS and represent a candidate objective measure of the degree of inflammation in the entire spine and SI joints.

The widespread use of MRI for diagnostic or classification purposes requires standardized definitions for the morphologic appearances of inflammatory lesions typically observed in the spine of a patient with AS. The Canada/Denmark International MRI Working Group has recently developed standardized definitions of spinal inflammatory lesions (7, 8). Inflammatory lesions are first defined based on their presence in either central (including the spinal canal) or lateral (not including the spinal canal) sagittal slices and then, second, on their anatomic location in relation to the vertebral end plate (vertebral corner versus noncorner). Inflammatory lesions are also recorded in the posterior elements of the spine according to anatomic location (facet joint, spinous process).

The goal of this study was to determine the sensitivity, specificity, and diagnostic utility of different inflammatory lesions in the entire spine when assessed by whole-body MRI in patients with AS or with recent-onset inflammatory back pain (IBP) compared with a control group of healthy individuals.

SUBJECTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Subjects.

AS and IBP group.

We studied 35 consecutive patients with AS who fulfilled the modified New York criteria (9) and 25 patients with IBP of ≥3 months and ≤24 months' duration, all of whom were recruited from a single rheumatology outpatient clinic. The plain pelvic radiographs of all 60 patients were independently graded by 2 readers (ROK, UW) for sacroiliitis according to the radiographic modified New York criteria (9). Patients with IBP had to fulfill ≥2 of 4 IBP criteria based on patient history (morning stiffness >30 minutes' duration, improvement in back pain with exercise but not with rest, awakening because of back pain during the second half of the night, alternating buttock pain) (10), and additionally had to show ≥1 of the following clinical and laboratory features: good response to nonsteroidal antiinflammatory drugs, peripheral arthritis, enthesitis, dactylitis, uveitis, HLA–B27 positivity, elevated erythrocyte sedimentation rate or C-reactive protein level, and positive family history for SpA. The principal criterion used in this cross-sectional study to allocate patients to the IBP group was a self-reported IBP duration of ≤24 months, irrespective of the radiographic SI joint grading. The upper age limit for both patient groups was 45 years, and all patients had active disease as defined by a Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) global score ≥4 and/or a BASDAI item 2 score (assessing spinal pain) of ≥4 on a numeric rating scale ranging from 0–10 (11). All patients were enrolled in a national prospective observational AS cohort, the Swiss Clinical Quality Management in Ankylosing Spondylitis.

Control group.

Thirty-five healthy age- and sex-matched volunteers from the staff (mainly nurses, physiotherapists, and physicians) of the same university hospital that recruited patients served as controls for the AS group. The youngest 25 controls were selected as the control group for the patients with IBP. The status of a healthy control was defined by the Nordic questionnaire (12). Transient back pain any time during the participant's lifetime and back pain of 1–7 days' duration in the 12 months prior to study enrollment was permitted under the following conditions concerning consequences of back pain: no previous inpatient or outpatient treatment by a physician, physiotherapist, chiropractor, or similar medical professional, no sick leave, and no adaptation or change of vocational or everyday activities in the past as a consequence of back pain.

The study protocol was approved by the Zurich Cantonal Ethics Committee, Subcommittee for Orthopedics and Rheumatology. The patients and the healthy participants all gave written informed consent.

Exclusion criteria.

In the AS and IBP group, exclusion criteria were ongoing or previous (within the last 6 months) treatment with tumor necrosis factor α inhibitors or other biologic agents, malignancies or infections affecting the skeleton, previously performed surgery of the spine, pelvic girdle, or shoulder girdle, pregnancy, advanced spinal deformity due to AS or other disorders precluding an adequate MRI examination, and technical contraindications to MRI such as cardiac pacemakers or similar devices.

The same exclusion criteria applied to the control group, supplemented by a questionnaire to rule out symptoms of specific back pain (IBP symptoms as defined above, psoriasis, inflammatory bowel disease, symptoms of radiculopathy or spinal stenosis, previous vertebral fracture or major spinal injury, known osteoporosis, high grade spondylolisthesis, or major spinal deformities such as severe scoliosis).

MRI protocol.

Imaging was performed with a Siemens Avanto 1.5T magnet (Siemens Medical Solutions, Erlangen, Germany). The system can address up to 18 independent radiofrequency channels so that 6 coils can be plugged into the system simultaneously. For the purpose of this study, 3 coils built into the MRI table were employed (head, neck, and spine coil). Two body matrix coils with 6 elements each were placed on the patient's chest and abdomen, in addition to a flexible coil placed over the hips in tall patients. Sagittal turbo STIR images of the entire spine and sacrum were acquired with the following parameters: repetition time 6,270 msec, echo time 93 msec, and inversion time 130 msec. The turbo factor was 21, the parallel acquisition technique factor was 2, and the generalized autocalibrating partially parallel acquisition mode was used. The images were obtained in 2 steps with automated combination of the 2 sequences into a single image series. Twenty slices were acquired with a slice thickness of 3–4 mm and an interslice gap of 0.3–0.4 mm. The combined field of view was 780 × 450 mm, the pixel size was 1.0 × 1.0 mm, and the acquisition time was 2 × 2 minutes 49 seconds.

MRI analysis.

STIR sequences of the entire spine were read and scored independently by 3 readers blinded to patient identifiers and clinical characteristics. Readers did not have access to the coronal SI joint images. Two readers (RAK, WPM) were not involved in clinical assessments, whereas the third (UW) was responsible for the recruitment of patients and healthy volunteers. One reader (WPM) is a member of the Canada/Denmark International MRI Working Group, which developed standardized definitions for abnormalities of the spine on MRI in SpA. One reader (UW, a rheumatologist) was involved in the initiation of the whole-body MRI in SpA project and was trained by WPM and RGWL during video teleconference sessions that focused on the application of these standardized definitions. The third reader (RAK, a radiologist) participated in a calibration exercise in UW's institution based on reference cases and on whole-body MRI examples not included in this study.

The films were evaluated in random order on electronic work stations in the institution of each reader. The findings were recorded electronically on a separate screen using a Microsoft Excel (Microsoft, Redmond, WA) based worksheet containing all possible inflammatory lesions on the x-axis and the 46 vertebral end plates, from the C2 lower end plate to the S1 upper end plate, on the y-axis.

Increased STIR signal denoting active inflammation was recorded for the entire spine according to the definitions proposed by the Canada/Denmark International MRI Working Group, as demonstrated in reference images available online at www.arthritisdoctor.ca (7, 8). The following specific lesions were recorded in a dichotomous manner (present/absent) (13): vertebral corner inflammatory lesions (CIL; anterior and posterior), vertebral noncorner inflammatory lesions (NIL), lateral inflammatory lesions (LIL), and facet joint or other posterior element inflammatory lesions (FIL/PIL). CIL and NIL are detected in central sagittal slices (slices that include the spinal canal), whereas LIL are detected in lateral slices (slices that do not include the spinal canal) (Figure 1A). FIL/PIL are signal alterations adjacent to a facet joint or in another posterior structure (except the pedicle).

thumbnail image

Figure 1. A, 33-year-old male patient with ankylosing spondylitis (HLA–B27 positive, symptom duration 5 years). The sixth thoracic vertebra displays a lateral inflammatory lesion (LIL) on a STIR sequence in this patient with scoliosis. In this region of the spine, the sagittal slice does not include the spinal canal. In contrast, the spinal canal is displayed in the lumbar spine, which shows 2 anterior corner inflammatory lesions (aCIL) of the first and third lumbar vertebra. B, 31-year-old healthy male control subject. Two vertebral corner STIR signal alterations that meet the definition of a CIL (arrows) were visible in the eighth and tenth thoracic vertebrae on 4 consecutive sagittal slices.

Download figure to PowerPoint

Statistical analysis.

Prestatistical analysis.

Thirteen of 95 whole-body MRIs showed a lumbosacral transitional anomaly, and 2 of 95 MRIs displayed a partial vertebral fusion; the corresponding MRI scoring sheets were adapted accordingly. Only lesions that were scored concordantly between ≥2 observers were entered in summary tables of the 3 possible reader pairs. Concordance was defined as a lesion scored concordantly by 2 readers concerning the spinal level as well as the type of inflammatory lesion.

Records of any vertebral end plate with a difference in ≥1 inflammatory lesions between a reader pair were regarded as discordant. Discordance was expressed as mean percentage of vertebral end plates with discordant findings in relation to the total of 46 end plates per study subject.

Analysis.

Concordantly scored inflammatory lesions for the 3 reader pairs comparing 35 patients with AS and 25 patients with IBP versus 35 and 25 healthy sex- and age-matched controls were used to compute specificity, sensitivity, and likelihood ratios (LRs) according to the specific type and number of each inflammatory lesion for the entire spine and by spinal segment. For these values, 95% confidence intervals (95% CIs) were computed using the R package epiR (14, 15). Standard references were used for the 95% CI formulas for sensitivity and specificity (16) and for LRs (17). No adjustment for multiple testing was done in computation of the 95% CIs.

To assess the capacity of each lesion to discriminate between diagnostic categories and to compare the performance of the 3 reader pairs, receiver operating characteristic (ROC) curves and areas under the curve (AUC) were computed (18, 19). In general, ROC curves and AUC allow for a comparison of different diagnostic tests based on an ordinal or continuous variable. A nondiscriminative test delivers a diagonal line as an ROC curve, with a corresponding AUC of 0.5. The better the diagnostic quality of a test, the more pronounced the shift of the ROC curve into the upper left corner. The AUC of an ideal ROC curve, which passes through the upper left corner, is 1. This method of presentation does not need to define cutoff levels, because sensitivity and specificity are computed for all possible cutoffs jointly and plotted in a single curve.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Descriptive analysis.

The 35 patients with AS (28 men; 31 HLA–B27 positive) had a median age of 31.5 years (range 19.3–44.8 years) and a median symptom duration of 8.0 years (range 2.1–29 years) (Table 1). Thirty-four patients had primary AS and 1 had AS associated with Crohn's disease.

Table 1. Characteristics of the study participants and descriptive data of the inflammatory lesions*
 AS groupIBP groupControl group
  • *

    Values for patient characteristics are the median (range) unless otherwise indicated. AS = ankylosing spondylitis; IBP = inflammatory back pain; NA = not applicable; BASDAI = Bath Ankylosing Spondylitis Disease Activity Index (11); NRS = numeric rating scale; BASDAI 2 = second item on the BASDAI (reflects back pain); BASFI = Bath Ankylosing Spondylitis Functional Index (20); BASMI = Bath Ankylosing Spondylitis Metrology Index (21); ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; CIL = corner inflammatory lesion; tCIL = thoracic CIL; lCIL = lumbar CIL; NIL = noncorner inflammatory lesion; LIL = lateral inflammatory lesion.

  • HLA–B27 had not been determined in 1 patient (percentage based on 34 patients with AS).

  • Reference range 5 mg/liter.

  • §

    Values are medians for the 3 reader pairs (lowest and highest value of interquartile ranges for the 3 reader pairs).

Male:female28:718:728:7
Age, years31.5 (19.3–44.8)27 (17.3–44.8)30.8 (17.7–44.7)
Symptom duration8 years (2.1–29)10 months (4–24)NA
HLA–B27 positive, %9192NA
BASDAI, NRS4.7 (2.2–7)4.2 (1.2–7.2)NA
BASDAI 2, NRS7 (4–10)6 (2–10)NA
BASFI, NRS2.9 (0–7.7)2.2 (0–8.2)NA
BASMI, range 0–101 (0–7)0 (0–4)NA
Nocturnal pain, NRS5 (0–10)7 (0–10)NA
Lumbar morning stiffness, NRS   
 Intensity6 (0–10)6 (0–10)NA
 Duration4 (0–10)3 (0–10)NA
ESR (mm/1hour)12 (2–60)11 (1–72)NA
CRP level6.5 (1–46)5 (0–150)NA
CIL§5/6/5 (8–9.5)1/0/0 (2)0/0/0 (0–0.5)
tCIL§3/4/3 (7.5–8)1/0/0 (1–2)0/0/0 (0)
lCIL§1/1/1 (2–2.5)0/0/0 (1)0/0/0 (0)
NIL§0/0/0 (0)0/0/0 (0)0/0/0 (0)
LIL§0/0/0 (0.5–2)0/0/0 (0)0/0/0 (0)

The 25 patients with IBP (18 men; 23 HLA–B27 positive) had a median age of 27 years (range 17.3–44.8 years) and a median symptom duration of 10 months (range 4–24 months) (Table 1). The median age at symptom onset was 25.6 years (range 16.3–44.3 years). One AS patient had psoriatic arthritis and one IBP patient had ulcerative colitis. Despite the short self-reported symptom duration, 8 (32%) of the 25 patients with IBP fulfilled the radiographic modified New York classification criteria.

The median numbers and interquartile ranges of concordantly scored inflammatory lesions for the 3 reader pairs and for both patient groups are given in Table 1. CIL were the most frequent inflammatory lesions, particularly in the thoracic spine. LIL represented the second most common lesion. NIL and FIL/PIL were recorded only rarely. The mean percentages of discordant records between the 3 reader pairs were 15.2%, 19.7%, and 17% for the AS group, and were 7.7%, 10.3%, and 6.6% for the IBP group. No single reader pair performed systematically differently than the other pairs.

The median age of the 35 controls (7 women, 28 men) was 30.8 years (range 17.7–44.7 years). The median numbers and interquartile ranges of lesions on the STIR sequence for the controls are shown in Table 1. Among the 35 healthy individuals, CIL were reported in 26% (cervical spine CIL in 3%, thoracic CIL in 20%, and lumbar CIL in 9%; in any individual subject, MRI abnormalities could be present in >1 spinal region), NIL in 6%, and LIL in 3% for all 3 reader pairs combined. The 25 youngest volunteers, who served as controls for the IBP group, had CIL, NIL, and LIL in 9%, 6%, and 3% of the group, respectively. Ten controls with CIL, NIL, or LIL were identified by all 3 reader pairs, and 1 control was indicated by 1 reader pair only. The highest numbers of CIL among the 9 volunteers reported to have this lesion was 1 for 3 controls, 2 for 4 controls, and 3 for 2 controls. In contrast to the previously described distribution pattern of inflammatory lesions in the spine in patients with AS (1) or IBP (6), the control group showed scattered lesions along the entire spine, with a slight increase in frequency in the lower thoracic and the lumbar spine. The mean percentages of discordant records between the 3 reader pairs were 2.1%, 1.7%, and 2.4%.

Analysis of diagnostic utility.

Sensitivity, specificity, and positive and negative LRs based on combined data from the concordant observations of all 3 reader pairs are shown in Table 2. The highest positive LR for the discrimination of cases from controls was obtained when ≥2 CIL were present both for the AS group (positive LR 12, sensitivity 69%, specificity 94%) and the IBP group (positive LR 8, sensitivity 32%, specificity 96%). The highest LRs per spinal segment for both patient groups were found in the thoracic spine for CIL. The finding of an LIL, NIL, or FIL/PIL was highly specific and lacked sensitivity for both AS and IBP.

Table 2. Sensitivity, specificity, and positive and negative LRs based on concordant observations recorded by all 3 reader pairs*
 Sensitivity, %Specificity, %Positive LRNegative LR
  • *

    Values are the number (95% confidence interval). LR = likelihood ratio; FIL/PIL = facet joint or other posterior element inflammatory lesions; NC = not calculable. See Table 1 for additional definitions.

AS group    
 >0 CIL0.77 (0.61–0.88)0.77 (0.61–0.88)3.38 (1.79–6.37)0.3 (0.16–0.56)
 >1 CIL0.69 (0.52–0.81)0.94 (0.81–0.98)12 (3.07–46.96)0.33 (0.20–0.55)
 >2 CIL0.66 (0.49–0.79)0.94 (0.81–0.98)11.5 (2.93–45.11)0.36 (0.23–0.58)
 >0 tCIL0.69 (0.52–0.81)0.83 (0.67–0.92)4 (1.87–8.57)0.38 (0.23–0.63)
 >1 tCIL0.57 (0.41–0.72)0.94 (0.81–0.98)10 (2.53–39.59)0.45 (0.31–0.67)
 >2 tCIL0.46 (0.30–0.62)0.97 (0.85–0.99)16 (2.24–114.18)0.56 (0.41–0.76)
 >0 lCIL0.51 (0.36–0.67)0.94 (0.81–0.98)9 (2.26–35.91)0.52 (0.36–0.73)
 >0 NIL0.06 (0.02–0.19)0.97 (0.85–0.99)2 (0.19–21.06)0.97 (0.88–1.07)
 >0 LIL0.31 (0.19–0.48)0.97 (0.85–0.99)11 (1.50–80.69)0.71 (0.56–0.89)
 >0 FIL/PIL0.09 (0.03–0.22)1.00 (0.90–1.00)NC0.91 (0.83–1.01)
IBP group    
 >0 CIL0.4 (0.23–0.59)0.88 (0.70–0.96)3.33 (1.04–10.69)0.68 (0.48–0.97)
 >1 CIL0.32 (0.17–0.52)0.96 (0.80–0.99)8 (1.08–59.32)0.71 (0.54–0.94)
 >2 CIL0.12 (0.04–0.30)0.96 (0.80–0.99)3 (0.33–26.92)0.92 (0.78–1.08)
 >0 tCIL0.32 (0.17–0.52)0.88 (0.70–0.96)2.67 (0.80–8.90)0.77 (0.57–1.05)
 >1 tCIL0.24 (0.11–0.43)0.96 (0.80–0.99)6 (0.78–46.29)0.79 (0.63–1.00)
 >2 tCIL0.12 (0.04–0.30)0.96 (0.80–0.99)3 (0.33–26.92)0.92 (0.78–1.08)
 >0 lCIL0.24 (0.11–0.43)1.00 (0.87–1.00)NC0.76 (0.61–0.95)
 >0 NIL0.04 (0.01–0.20)0.96 (0.80–0.99)1 (0.07–15.12)1 (0.89–1.12)
 >0 LIL0.12 (0.04–0.30)0.96 (0.80–0.99)3 (0.33–26.92)0.92 (0.78–1.08)
 >0 FIL/PIL0.00 (0.00–0.13)1.00 (0.87–1.00)NC1.00 (1.00–1.00)

ROC curves and AUC values for the AS group are shown in Figure 2A. There was a high consistency between the 3 reader pairs. The highest diagnostic discrimination was observed for CIL (AUC 0.84 for all 3 reader pairs combined) and thoracic CIL (AUC 0.80). LIL (AUC 0.65) showed a lower discriminatory capacity compared with CIL. NIL (AUC 0.52) did not discriminate between patients with AS and controls.

thumbnail image

Figure 2. A, receiver operating characteristic (ROC) curves and areas under the curve (AUC) for corner inflammatory lesions (CIL), thoracic CIL (tCIL), lumbar CIL (lCIL), and lateral inflammatory lesions (LIL) in patients with ankylosing spondylitis (AS) based on concordant observations for each of the 3 reader pairs (RPs) and for the combined data. B, ROC curves and AUC for CIL, tCIL, lCIL, and LIL in patients with inflammatory back pain (IBP) based on concordant observations for each of the 3 RPs and for the combined data.

Download figure to PowerPoint

ROC curves and AUC values for the IBP group are displayed in Figure 2B. Reader pair 1, calibrated by several video teleconference sessions, performed differently from the other reader pairs. Again, diagnostic discrimination was best for CIL (AUC 0.65 for all 3 reader pairs combined, AUC 0.80 for reader pair 1) and for thoracic CIL (AUC 0.61 and 0.70, respectively); however, the values were not as high as for the AS group. LIL and NIL with AUC values around 0.5 could not be used to differentiate between IBP patients and controls.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

This cross-sectional study in patients with AS or IBP age ≤45 years and with active disease showed that among inflammatory lesions, the CIL had the best diagnostic utility. In particular, the presence of ≥2 CIL had the best combination of sensitivity and specificity. The finding of a LIL also carried high specificity for AS and IBP, but lacked sensitivity. The finding of an NIL or a PIL/FIL was not diagnostically useful. We also showed that a CIL may occasionally be present in 26% of healthy individuals.

The goal of this validation study was to determine which inflammatory MRI-detected lesion(s) had diagnostic utility for SpA, particularly in a group of recent-onset patients who might present diagnostic challenges in early disease, by using the standardized definitions for acute inflammatory lesions developed by the Canada/Denmark International MRI Working Group. The recently introduced whole-body MRI technique proved to be an ideal imaging method for this purpose through its ability to scan the entire spine in a single examination. Two readers were calibrated by video teleconference sessions and 2 readers by a face-to-face training meeting. After an independent readout by each observer, only concordantly scored lesions per reader pair were used for further analysis. This approach attempted to avoid taking into account equivocal MRI signal alterations such as low intensity or very small lesions. The concordance of the 3 reader pairs was high, ranging from 80–85% for the AS group and from 90–93% for the IBP group. A recent study on MRI changes in the SI joints of 68 patients with recent-onset IBP showed a concordance rate from 78–85% (22). The reader pair trained by several video teleconference sessions showed higher AUC values for CIL in the IBP group, which highlights the importance of a pre-readout calibration. However, despite this different calibration approach, the data look very comparable over all 3 study groups, thus reinforcing the validity of the conclusions.

Given a lifetime prevalence of back pain of up to 70% and of ∼50% during a given year (23), rare transient episodes of back pain were considered irrelevant for the control group when there was no need for treatment and no impact on vocational or everyday activities. Patients with IBP were selected according to a self-reported IBP duration of ≤24 months with ≥1 additional clinical feature characteristic for SpA, but interestingly, 8 (32%) of 25 patients with IBP already fulfilled the radiographic modified New York criteria. In a Dutch IBP cohort with a symptom duration of <2 years at study entry, only 14 (21%) of 68 patients showed radiographic evidence for definite sacroiliitis (24). A possible explanation for this discrepancy may be that patients in the Dutch study were primarily recruited on the basis of fulfilling the Calin criteria for IBP, whereas our cohort required an additional clinical feature of SpA and therefore might be considered to include those patients more likely to progress to radiographically definite AS. Our study analyzed the diagnostic utility of MRI lesions in a very early disease stage (IBP of ≤24 months' duration) irrespective of the radiographic SI joint findings.

Patients with IBP showed the same types of inflammatory lesions as were seen in patients with AS, but with a lower frequency. A positive LR of 8 for ≥2 CIL in the spine of patients with IBP is very close to the best positive LR (positive LR 9 for HLA–B27) in a literature review on the sensitivity, specificity, and LR for various clinical features, laboratory findings, and imaging techniques for diagnosing pre-radiographic SpA (25).

In clinical practice, the finding of ≥2 CIL may have a high diagnostic utility, particularly in patients with recent-onset back pain for whom there is a reasonable pretest probability for early SpA based on clinical grounds. Our study design of evaluating only concordantly scored lesions between 2 readers may be translated into clinical practice, with usually a single observer, by taking only clear-cut inflammatory lesions into consideration.

Prior systematic assessment of MRI lesions for diagnostic utility in AS has been limited. A recent study compared MRI findings in 52 patients with AS (recruited from an imaging database irrespective of disease duration) with a mixed control group of patients with specific back pain and with radiologically normal subjects (26). The sensitivity and specificity for the MRI corner sign, defined as a triangular and sharply marginated corner abnormality in a vertebral body not associated with osteophytes or Schmorl's nodes, were 44% and 96%, respectively. This retrospective study of patients over a period of 5 years had several limitations, such as no inclusion of recent-onset patients with IBP with inconclusive SI joint radiographs, an analysis of conventional lumbar spine MRI only, no availability of STIR sequences, a historical mixed control group of specific back pain patients and healthy subjects, and a readout by consensus. The main limitation of our own study was the lack of a comparator group with mechanical back pain. Possible MRI signal alterations reflecting spinal degenerative changes may affect the specificity of inflammatory MRI lesions as defined for AS. Another limitation of our study was that the findings may not be valid in a population of elderly patients with AS.

The high diagnostic utility of CIL needs to be confirmed in other cohorts of patients with AS or IBP with an additional comparator group of nonspecific back pain patients. Spinal inflammatory lesions as defined by the Canada/Denmark Working Group may also serve as indicators in long-term cohort studies to assess whether spinal inflammatory lesions are indeed predictive of future structural damage such as syndesmophyte formation. Future studies should also focus on definitions for inflammatory changes in the SI joint and their diagnostic utility compared with healthy controls and patients with mechanical back pain.

Despite their lower frequency compared with CIL, LIL are relevant in clinical practice because they often occur in the thoracic spine adjacent to the costovertebral joints and may be missed if the MRI assessment does not include sagittal slices that extend to the lateral edges, particularly of the thoracic vertebrae (27).

To our knowledge, very few studies have examined the histologic equivalent of spinal inflammatory lesions visible on STIR MRI sequences of patients with AS. A histopathologic analysis of zygapophyseal joints obtained from 8 patients with longstanding AS who underwent spinal extension surgery showed interstitial mononuclear cell infiltrates (28). Our cross-sectional study showed that inflammatory-like MRI lesions may also occur in healthy volunteers with a frequency of up to 26% (Figure 1B). The nature of inflammatory-like lesions in healthy volunteers remains speculative because biopsies are not feasible for obvious ethical reasons. HLA–B27 was not tested in the 10 controls who showed inflammatory-like lesions. We speculate that these lesions in controls may be mechanically induced signal alterations. Arguments for this hypothesis may be the increase in frequency of inflammatory-like lesions in the oldest 10 control subjects, suggesting early degenerative spinal changes and a different distribution pattern of spinal segments affected in controls than in patients. The lesions in controls were scattered along the entire spine, in contrast to a clear-cut clustering of inflammatory lesions in the lower thoracic spine in both patient groups. MRI signal alterations in STIR sequences are possibly not specific for truly inflammatory changes as opposed to mechanical lesions.

In conclusion, this cross-sectional study, which focused on recording concordantly scored inflammatory lesions on MRIs from patients with AS or IBP age ≤45 years with active disease, showed that discrimination from healthy controls was optimal when ≥2 CIL or ≥1 LIL were present. The finding of an NIL or a PIL/FIL was not diagnostically useful due to low sensitivity. Inflammatory-like lesions also occurred in healthy individuals with a frequency of up to 26%. Future studies should address diagnostic utility in relation to other causes of back pain in the demographic that develops AS.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Weber had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Weber, Hodler, Kubik, Rufibach, Lambert, Kissling, Pfirrmann, Maksymowych.

Acquisition of data. Weber, Hodler, Kubik, Kissling, Pfirrmann, Maksymowych.

Analysis and interpretation of data. Weber, Hodler, Kubik, Rufibach, Lambert, Kissling, Pfirrmann, Maksymowych.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

The authors thank the patients and the healthy volunteers for their participation, and the following Swiss rheumatologists, internists, and primary care physicians for referral of their patients: A. Achermann (Luzern), M. Altermatt (Basel), D. Amgwerd (Spreitenbach), G. Bickel (Rapperswil), D. Bloesch (Olten), C. Boetschi (Romanshorn), C. Brunner (Zurich), S. Buergin (Basel), P. De Vecchi (St. Moritz), B. Elmiger (Bern), P. Exer (Basel), D. Galovic (Pfaeffikon), T. Gerber (Zurich), M. Giger (Menzingen), D. Glenz (Visp), F. Haefelin (Schlieren), G. Hajnos (Zurich), U. Heusser (Winterthur), U. Hintermann (Brugg), M. Hofmann (Zurich), M. Hoppler (Zug), P. Imbach (Zurich), J. Imholz (Zurich), C. Jeanneret (Schwerzenbach), B. Kleinert (Zurich), M. Klopfstein (Biel), I. Kramers (Zurich), R. Maager (Aarau), N. Masina (Lugano), C. Merlin (Baden), S. Pfister (Buelach), R. Putzi (Regensdorf), A. Rapp (Zurich), J. Ryser (Zurich), M. Sager (Winterthur), A. Schmidt (Basel), H. Schwarz (Basel), S. Studer (Zurich), P. Sutter (Zurich), F. Tapernoux (Rueti), B. Weiss (Basel), R. Wuethrich (Brugg).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  • 1
    Baraliakos X, Landewe R, Hermann KG, Listing J, Golder W, Brandt J, et al. Inflammation in ankylosing spondylitis: a systematic description of the extent and frequency of acute spinal changes using magnetic resonance imaging. Ann Rheum Dis 2005; 64: 7304.
  • 2
    Weber U, Hodler J, Jurik AG, Pfirrmann CW, Rufibach K, Kissling RO, et al. Assessment of active spine inflammatory changes in patients with axial spondyloarthritis: validation of whole body MRI against conventional MRI. Ann Rheum Dis 2009. E-pub ahead of print.
  • 3
    Ahlstroem H, Feltelius N, Nyman R, Haellgren R. Magnetic resonance imaging of sacroiliac joint inflammation. Arthritis Rheum 1990; 33: 17639.
  • 4
    Lambert RG, Salonen D, Rahman P, Inman RD, Wong RL, Einstein SG, et al. Adalimumab significantly reduces both spinal and sacroiliac joint inflammation in patients with ankylosing spondylitis: multicenter, randomized, double-blind, placebo-controlled study. Arthritis Rheum 2007; 56: 400514.
  • 5
    Weber U, Kissling RO, Hodler J. Advances in musculoskeletal imaging and their clinical utility in the early diagnosis of spondyloarthritis. Curr Rheum Rep 2007; 9: 35360.
  • 6
    Weber U, Pfirrmann CW, Kissling RO, Hodler J, Zanetti M. Whole body MR imaging in ankylosing spondylitis: a descriptive pilot study in patients with suspected early and active confirmed ankylosing spondylitis. BMC Musculoskelet Disord 2007; 8: 20.
  • 7
    Lambert RG, Pedersen SJ, Maksymowych WP, Chiowchanwisawakit P, Ostergaard M. Active inflammatory lesions detected by magnetic resonance imaging in the spine of patients with spondyloarthritis: definitions, assessment system, and reference image set. J. Rheumatol. In press.
  • 8
    Pedersen SJ, Ostergaard M, Chiowchanwisawakit P, Lambert RG, Maksymowych WP. Validation of definitions for active inflammatory lesions detected by MRI in the spine of patients with spondyloarthritis. J Rheumatol. In press.
  • 9
    Van der Linden S, Valkenburg HA, Cats A. Evaluation of diagnostic criteria for ankylosing spondylitis: a proposal for modification of the New York criteria. Arthritis Rheum 1984; 27: 3618.
  • 10
    Rudwaleit M, Metter A, Listing J, Sieper J, Braun J. Inflammatory back pain in ankylosing spondylitis: a reassessment of the clinical history for application as classification and diagnostic criteria. Arthritis Rheum 2006; 54: 56978.
  • 11
    Garret S, Jenkinson T, Kennedy LG, Whitelock H, Gaisford P, Calin A. A new approach to defining disease status in ankylosing spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. J Rheumatol 1994; 21: 228691.
  • 12
    Kuorinka I, Jonsson B, Kilbom A, Vinterberg H, Biering-Sorensen F, Andersson G, et al. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Appl Ergon 1987; 18: 2337.
  • 13
    Maksymowych WP, Chiowchanwisawakit P, Clare T, Pedersen SJ, Ostergaard M, Lambert RG. Inflammatory lesions of the spine on magnetic resonance imaging predict the development of new syndesmophytes in ankylosing spondylitis. Arthritis Rheum 2009; 60: 93102.
  • 14
    Stevenson M, Nunes T, Sanchez J, Thornton R. EpiR: functions for analysing epidemiological data. R package version 0.9–10. 2008. URL: http://epicentre.massey.ac.nz.
  • 15
    R Development Core Team. R: a language and environment for statistical computing. Vienna (Austria): R Foundation for Statistical Computing; 2008. URL: www.R-project.org.
  • 16
    Altman DG, Machin D, Bryant TN, Gardner MJ. Statistics with confidence. 2nd ed. London: BMJ; 2000. p. 289.
  • 17
    Simel D, Samsa G, Matchar D. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol 1991; 44: 76370.
  • 18
    Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: 2936.
  • 19
    Pepe MS. The statistical evaluation of medical tests for classification and prediction: Oxford statistical science series. Vol. 28. New York: Oxford University; 2003.
  • 20
    Calin A, Garrett S, Whitelock H, Kennedy LG, O'Hea J, Mallorie P, et al. A new approach to defining functional ability in ankylosing spondylitis: the development of the Bath Ankylosing Spondylitis Functional Index. J Rheumatol 1994; 21: 22815.
  • 21
    Jenkinson TR, Mallorie PA, Whitelock HC, Kennedy LG, Garrett SL, Calin A. Defining spinal mobility in ankylosing spondylitis (AS): the Bath AS metrology index. J Rheumatol 1994; 21: 16948.
  • 22
    Heuft-Dorenbosch L, Weijers R, Landewe R, van der Linden S, van der Heijde D. Magnetic resonance imaging changes of sacroiliac joints in patients with recent-onset inflammatory back pain: inter-reader reliability and prevalence of abnormalities. Arthritis Res Ther 2006; 8: R11.
  • 23
    Lawrence RC, Felson DT, Helmick CG, Arnold LM, Choi H, Deyo RA, et al. Estimates of the prevalence of arthritis and other rheumatic conditions in the United States. Part II. Arthritis Rheum 2008; 58: 2635.
  • 24
    Heuft-Dorenbosch L, Landewe R, Weijers R, Wanders A, Houben H, van der Linden S, et al. Combining information obtained from magnetic resonance imaging and conventional radiographs to detect sacroiliitis in patients with recent onset inflammatory back pain. Ann Rheum Dis 2006; 65: 8048.
  • 25
    Rudwaleit M, van der Heijde D, Khan MA, Braun J, Sieper J. How to diagnose axial spondyloarthritis early. Ann Rheum Dis 2004; 63: 53543.
  • 26
    Kim NR, Choi JY, Hong SH, Jun WS, Lee JW, Choi JA, et al. “MR corner sign”: value for predicting presence of ankylosing spondylitis. AJR Am J Roentgenol 2008; 191: 1248.
  • 27
    Rennie WJ, Dhillon SS, Conner-Spady B, Maksymowych WP, Lambert RG. MRI assessment of spinal inflammation in ankylosing spondylitis: standard clinical protocols may omit inflammatory lesions in thoracic vertebrae. Arthritis Rheum. In press.
  • 28
    Appel H, Loddenkemper C, Grozdanovic Z, Ebhardt H, Dreimann M, Hempfing A, et al. Correlation of histopathological findings and magnetic resonance imaging in the spine of patients with ankylosing spondylitis. Arthritis Res Ther 2006; 8: R143.