Objective To determine the sensitivity and specificity of enthesis ultrasound for the diagnostic classification of early spondyloarthritis.
Methods A cross-sectional, blinded and controlled study. Standardised bilateral ultrasound of six entheses (Madrid sonography enthesitis index (MASEI)) was performed. Accepted diagnostic classification criteria were used as the gold standard. Validity was analysed by receiver operating characteristic (ROC) curves. Values of p<0.05 were considered significant.
Results 113 early spondyloarthritis patients were included (58 women/55 men), 57 non-inflammatory control individuals (29 women/28 men) and 24 inflammatory control individuals (11 women/13 men). The evolution time of spondyloarthritis was 10.9±7.1 months. At least some grade of sacroiliitis on x-ray was present in 59 patients, but only five fulfilled the radiographic sacroiliitis New York criteria. Human leucocyte antigen B27 (HLA-B27) was positive in 42% of patients. No statistical differences were found for the enthesis score among diagnostic spondyloarthritis subtypes form of presentation (axial, peripheral or mixed) or HLA-B27 positivity. The MASEI score achieved statistical significance for gender. The ultrasound score was 23.36±11.40 (mean±SD) in spondyloarthritis patients and 12.26±6.85 and 16.04±9.94 in the non-inflammatory and inflammatory control groups (p<0.001), respectively. The ROC area under the curve was 0.82, and a cut-off point of ≥20 points achieved a likelihood ratio of 5.30 and a specificity of 89.47%.
Conclusions Entheses are affected early in spondyloarthritis, and the incidence of involvement is higher in men and independent of the spondyloarthritis diagnostic subtype, HLA-B27 status or presentation pattern. The enthesis ultrasound score seems to have diagnostic accuracy and may be useful for improving the diagnostic accuracy of early spondyloarthritis.
Statistics from Altmetric.com
Spondyloarthritis is made up of a group of chronic inflammatory diseases that share clinical manifestations and an association with human leucocyte antigen B27 (HLA-B27). The established classification criteria for spondyloarthritis rely on the combination of clinical symptoms with unequivocal radiographic sacroiliitis.1 However, x-rays are often normal when symptoms first arise, and the diagnosis of ankylosing spondylitis (AS) is commonly delayed by an average of 6–9 years after the onset of symptoms.2–3 In addition, at 10 years from first presentation, 25–35% of patients still do not have radiographic sacroiliitis.4
Until recently, therapeutic options for patients with spondyloarthritis were very limited, but this has changed since the introduction of anti-tumour necrosis factor therapies, which have been demonstrated to reduce disease activity, including metrological parameters.5,–,10 Therefore, reducing the time between the onset of spondyloarthritis and its diagnosis could be crucial to improving outcomes.
MRI has recently evolved as an important diagnostic tool for patients without definitive radiographic sacroiliitis. In two studies, active inflammation of the sacroiliac joints on MRI was shown to precede the development of radiographic sacroiliitis by 311 and 8 years12 in a percentage of patients, thereby supporting the validity of MRI as an appropriate imaging instrument for early disease. MRI has thus recently been included in the new Assessment of SpondyloArthritis International Society (ASAS) criteria for the classification of axial spondyloarthritis.13
However, the inflammatory involvement of the entheses, a characteristic feature of spondyloarthritis,14 is undervalued in the effort to improve the early diagnosis of spondyloarthritis. In this sense, a non-invasive test would be desirable. Peripheral enthesitis produces pain but may also be asymptomatic, and the clinical examination lacks sensitivity and specificity, as has been demonstrated by several studies comparing clinical evaluations with new imaging techniques.15,–,17 To date, MRI and special ultrasound have remained the most accurate imaging techniques for assessing enthesis inflammatory processes. We have previously demonstrated that the Madrid sonography enthesitis index (MASEI) is useful for discriminating patients with long-standing spondyloarthritis.18 However, access to an MRI is often restricted because of the high cost and low availability, the opposite scenario for ultrasonography, a diagnostic tool that is readily available in many rheumatology settings. Furthermore, it has been suggested that ultrasound might be superior to MRI for detecting early signs of enthesopathy.19,–,21 The aim of this study was to explore the accuracy of entheses ultrasound examinations for the early diagnostic classification of spondyloarthritis patients.
We performed a case control standardised enthesis ultrasound study in patients attending an early referral programme for spondyloarthritis and controls in a cross-sectional baseline evaluation.
The study was conducted according to local regulations and the Declaration of Helsinki, and local approval was obtained from the ethical committee and institutional review board of our hospital. All patients and controls signed an informed consent.
The patient sample had a mixed probability of either having or not having the disease. All were selected consecutively from individuals attending the early spondyloarthritis unit (ESU) at our hospital.
The ESU sees patients from eight primary care centres, corresponding to an area of 242 784 inhabitants, as part of the ESPERANZA programme, a nationwide health management programme designed to provide excellence in care for early spondyloarthritis.22 The referral criteria included: (1) age below 45 years; (2) symptom duration between 3 and 24 months; and (3) at least one of the following: (a) inflammatory low back pain, defined as at least two of the following: insidious onset, morning stiffness for more than 30 min or clear improvement of the symptoms with physical activity but not relieved by rest, (b) asymmetric arthritis, preferably of the lower limbs or (c) low back pain or arthralgia and at least one of the following: psoriasis, inflammatory bowel disease (IBD), anterior uveitis, a family history of spondylitis, psoriasis, radiographic sacroiliitis or HLA-B27-positive status. During the course of the study, 156 patients were referred to our department from participating centres. Of these, 43 were excluded from the study for the following reasons: evolution time greater than 24 months;9 diseases other than spondyloarthritis,13 did not meet the clinical criteria for spondyloarthritis11 and unwilling to provide consent or did not complete follow-up studies.10
Non-inflammatory controls were apparently healthy persons, selected among hospital workers and friends of patients, all of whom volunteered to participate after receiving an explanation of the procedure. In addition, we selected nine patients with non-inflammatory lumbar pain from our general clinic and 15 patients with posterior uveitis unrelated to spondyloarthritis. Once patients were identified, controls were selected to match for both sex and age of patients.
Inflammatory controls were patients of the ESPERANZA study that met the inclusion criteria for early spondyloarthritis, but did not meet spondyloarthritis diagnostic criteria.
For patients who fulfilled the referral criteria, ESU rheumatologists performed a prospective detailed medical history and examination, both general and rheumatological, as well as joint counts and the Maastricht ankylosing spondylitis enthesitis score entheses index. In addition, laboratory and x-ray studies were performed that included a complete blood cell count, biochemistry, C-reactive protein, erythrocyte sedimentation rate, HLA-B27, rheumatoid factor, antinuclear antibodies, as well as radiographs of the pelvis, cervical, dorsal and lumbar spine. Healthy controls (friends of hospital workers or patients) without any known inflammatory or mechanical musculoskeletal disease were invited to participate. For ethical reasons, no analytical or radiological studies, only anamnesis, was performed on the healthy control group.
All subjects underwent an ultrasound examination to obtain a MASEI score, as previously described.18 The MASEI score systematically explored six bilateral enthesis locations in each patient, including the proximal plantar fascia, distal Achilles tendon, distal and proximal patellar ligament, distal quadriceps and brachial triceps tendons. Enthesis thickness, structure, calcification/bone proliferation, erosion, bursa and power Doppler signal in the cortical bone profile, tendon and bursa are scored (figure 1). The MASEI score is a weighted score previously calculated by logistic regression that overestimates the score of three elemental lesions: calcification (0–3), Doppler (0 or 3) and erosion (0 or 3), while scoring tendon structure, tendon thickness and bursa as either a 0 or 1.18 The score range is 0–136.
Ultrasonography was performed by one rheumatologist trained in enthesis ultrasound using a Logiq 9 machine (General Electric, Wauwatosa, Wisconsin USA) with a linear probe at 9–14 MHz. The ultrasound examiner was blinded to the status of the subject being tested. All participants were asked not to communicate with the ultrasound examiner.
All subjects were classified as either having or not having one of the spondyloarthritis subtypes, as follows: (1) AS if they fulfilled the modified New York criteria;1 (2) psoriatic arthritis if they fulfilled the classification criteria for psoriatic arthritis (CASPAR);23 (3) spondyloarthritis without definitive radiographic sacroiliitis (at least bilateral grade II or unilateral grade III) and undifferentiated spondyloarthritis if the European Spondylarthropathy Study Group (ESSG) preliminary criteria for the classification of spondyloarthritis were fulfilled without any other specific diagnostic criteria;24 (4) reactive arthritis if the patient fulfilled the ESSG criteria or had arthritis, confirmed by a rheumatologist, with recent evidence of related infection; (5) arthritis-associated IBD if IBD was present in a patient with the New York criteria or ESSG criteria; and (6) anterior uveitis if it had been diagnosed by an ophthalmologist. The diagnosis of IBD required typical histological findings of Crohn's disease or ulcerative colitis.
Quantitative data are expressed as the mean±SD, and qualitative data are expressed as percentages. One-way analysis of variance (ANOVA) analysis was performed to study the differences in the mean MASEI score between groups. A Student's t test was used to test comparisons between each group. Pearson's correlation coefficient was used for the analysis of correlation. Receiver operating characteristic (ROC) curves were used to calculate the different levels of sensitivity, specificity and likelihood ratio (LR) at every cut-off point. Values of p<0.05 were considered to be significant.
The study sample included 57 non-inflammatory control subjects and 137 successive patients: 113 early spondyloarthritis patients (cases) and 24 patients who had inflammatory low back pain or arthritis (inflammatory controls), all of whom were similar in age and gender, as shown in table 1. The 24 inflammatory controls fulfilled the early spondyloarthritis inclusion criteria but did not meet spondyloarthritis diagnostic classification criteria (11 women and 13 men with a mean age of 32±8 years).
The average evolution time of the spondyloarthritis group was 10.9±7.1 months. The mean BASDAI was 4.96±2.18, the Bath ankylosing spondylitis functional index was 4.56±2.05, erythrocyte sedimentation rate was 10.19±10.22 and C-reactive protein was 8.76±13.60.
Radiographic sacroiliitis was present in 59 patients (55.1%), and 24 had bilateral sacroiliitis (22.5%), but only five patients fulfilled the radiographic sacroiliitis spondyloarthritis New York modified criteria. Eighteen fulfilled the psoriasis CASPAR criteria, four had arthritis-associated IBD, three had reactive arthritis and 83 had undifferentiated spondyloarthritis. HLA-B27 positivity was found in 42% of the patients. In total, 113 patients fulfilled the criteria for any spondyloarthritis.
Differences between spondyloarthritis groups
The MASEI score was as follows (mean±SD): 22.20±7.22 in AS, 24.25±10.71 in undifferentiated spondyloarthritis, 19.00±6.00 in reactive arthritis, 26.75±27.93 in IBD and 19.56±11.70 in psoriatic arthritis. In the control group, the ultrasound score was 12.26±6.85, and the difference was statistically significant compared with general spondyloarthritis patients (p<0.001) and versus all spondyloarthritis subtypes, with the exception of reactive arthritis (p=0.101). By ANOVA analysis, there were no statistical differences among entheses scores between different diagnostic subtype classification criteria of patients.
In parallel with the subdivision of their disease into classic diagnostic subtypes, spondyloarthritis patients were also classified as having axial, peripheral or mixed presentation forms, according to the clinical distribution pattern. There were no statistical differences between the clinical presentation patterns and the global scores or elemental lesions (table 2).
When we analysed the entheses ultrasound scores between HLA-B27-positive and HLA-B27-negative patients, there was no statistical difference (24.67±13.49 in HLA-B27-negative and 24.59±9.76 in HLA-B27-positive groups).
One important finding related to gender in early spondyloarthritis patients. The global MASEI score and the elemental lesions, except for the bursa, demonstrated significant statistical differences between men and women.
The value of the ultrasound score was 23.36±11.40 (mean±SD) in spondyloarthritis patients, 12.26±6.85 in the non-inflammatory control group (p<0.001 vs spondyloarthritis patients) and 16.04±9.94 in the inflammatory control group (p<0.01 vs spondyloarthritis patients); ANOVA analysis between groups achieved statistical significance (p<0.001). Patient inflammatory findings that fulfilled early spondyloarthritis inclusion criteria and that either met or did not meet spondyloarthritis diagnostic classification criteria are shown in table 3. Table 4 shows the different sensitivity, specificity, accuracy and LR for the different cut-off points of the MASEI score for early spondyloarthritis. The ROC area under the curve was 0.82 (95% CI 0.75 to 0.89). The choice of cut-off point is not simple; it depends on whether the objective is sensitivity or specificity. In table 4, we show different cut-off points to facilitate the selection depending on its intended clinical purpose. When we compared the spondyloarthritis patient group with the group of patients who had inflammatory low back pain or arthritis but without diagnostic classification criteria, the area under the ROC curve was 0.73 (95% CI 0.61 to 0.85) and the cut-off point of 20 had a sensitivity of 53.1%, a specificity of 83.3% and a positive LR of 3.26.
There is a growing interest in early diagnoses for patients with spondyloarthritis, especially because more effective treatment options are currently available. There are compelling data demonstrating that many patients with AS often do not show radiographic evidence of sacroiliitis at the start of their illness and for many years thereafter because of the time it takes to evolve. Moreover, better classification criteria are needed to avoid misclassification in clinical studies or drug trials of patients with early disease, particularly in patients at the preradiographic stage. As a result, new ASAS classification criteria have recently been presented13 with good results. However, the current sets of criteria undervalue either entheses or exclude them entirely. This can be an important lost opportunity because all experts agree that enthesitis is a hallmark of spondyloarthritis. To the best of our knowledge, this is the first study to investigate the prevalence of ultrasound entheses lesions in early spondyloarthritis and the validity of its subsequent criterion for the diagnosis of spondyloarthritis.
Our hypothesis was that if enthesitis is a true hallmark of spondyloarthritis, and if enthesis ultrasound is a sensible,15,–,17 specific16 18 25 and reliable tool,16 18 26 27 then ultrasound can be used to improve the accuracy of the spondyloarthritis classification diagnosis. There are interesting previous data suggesting that B mode ultrasound combined with Doppler ultrasound allowed for the detection of peripheral enthesitis in a majority of spondyloarthritis patients, thereby differentiating them from control populations; this finding could be very useful for the diagnosis of spondyloarthritis.16 18 28 Our study goes one step further to describe its utility in an early spondyloarthritis cohort.
Our study represented a sample of early spondyloarthritis patients with a mean symptomatic evolution time of 10.9 months (range 3–23 months) and a low prevalence of radiological findings; only five patients (4.4%) had radiographic sacroiliitis of bilateral grade 2 or unilateral grades 3 or 4. This percentage is lower than others have previously reported, but our patients were specifically asked about the time of evolution at their first visit. Viewed another way, our cohort is interesting because it represents patients in a non-radiographic stage and demonstrates that enthesis ultrasound can reveal frequent and characteristic alterations in early spondyloarthritis. This early involvement of entheses is in agreement with the GESPIC cohort, which showed that enthesitis occurs frequently in early disease, even in non-radiographic spondyloarthritis.29
The data presented in table 2 are relevant because a previous publication reported that the presence of entheseal involvement is independent of spondyloarthritis subtype. The study was conducted in spondyloarthritis patients with a mean evolution time of 16 years.16 Confirmation of the same result in an early cohort is of interest to evaluate the use of ultrasound for early diagnosis because we validated its use independent of the axial or peripheral presentation pattern. In our study, there were no statistical differences between groups for the global MASEI score or for any elemental lesion. Analysis of each individual elemental lesion has not been reported in previous publications.
The gender-matched subanalysis of this study revealed that women had significantly fewer lesions in entheses than men, including for spondyloarthritis patients. In a previous study, we demonstrated that entheses alterations achieved statistical differences in control men compared with women, but this difference disappeared in spondyloarthritis patients.18 The patients in that study all had established spondyloarthritis with a 15-year mean evolution time and were seen in a hospital; a scenario that probably biased a selection for patients with the worst evolution, especially for women. In this early cohort, entheses ultrasound lesions were more severe in men, as occurs in radiological spondyloarthritis affectation.29 As a result of this observation, future studies, including a control population, should be gender matched.
There was no significant difference in the ultrasound entheses scores between HLA-B27-positive and HLA-B27-negative patients with regard to the involvement of elemental lesions. Similar data were recently presented regarding either the modified Stoke AS spine score or the proportion of patients with syndesmophytes.29
The main objective of this study was to explore the accuracy of enthesis ultrasound examination. Validity was evaluated by sensitivity and specificity. Sensitivity is the proportion of true positives that are correctly identified by a test, whereas specificity is the proportion of true negatives that are correctly identified by a test. The choice for the cut-off point is not simple; it depends on whether the test objective is for a very sensitive test to detect mostly diseased individuals or for a very specific test to rule out the disease in mostly healthy individuals. An ideal test would have 100% sensitivity and specificity. However, this is rarely possible, and as sensitivity increases, specificity often decreases. We therefore calculated the LR, using LR-positive (sensitivity/1−specificity) and LR-negative (1−sensitivity/specificity). An LR above 10 or below 0.1 is considered to be strong evidence, respectively, to rule in or rule out a diagnosis in most circumstances. An LR of 5–10 and 0.1–0.2 generates moderate shifts in pretest and post-test probabilities, an LR of 2–5 and 0.5–0.2 generates small (but sometimes important) changes in probability and an LR of 1–2 and 0.5–1 alters the probability to a small (and rarely important) degree. In this sense, a cut-off point of 20 points or more, which achieved an LR positivity greater than 5 with a specificity near 90% or higher (see table 4), might be the best cut-off point. In the original MASEI score, in long-standing spondyloarthritis, an LR positivity above 5 was achieved with a score of 20 or greater.
A limitation of this validity study that must be noted is the high percentage of undifferentiated spondyloarthritis and the use of the ESSG criteria as a gold standard for this subtype of spondyloarthritis. The new ASAS criteria should probably currently be used, but the new ASAS criteria did not exist when the study was designed 3 years ago, and the ESSG had been validated and used in early spondyloarthritis studies.24 Nevertheless, the mean time of evolution for the reference cohort of the new ASAS criteria was 6 years, which is not truly an early spondyloarthritis cohort. A second limitation of this study is that ultrasound alone is not sufficient to make the diagnosis of spondyloarthritis; for example, a pretest probability of 5% is increased by a positive MASEI score by only 20%. However, when used in a clinical setting or in a core set with more signs or symptoms, the ultrasound results might be decisive. A third limitation of this study is that we did not test other enthesopathy or inflammatory arthritis diseases as controls. For example, rheumatoid arthritis can affects entheses, resulting in a limited ability to discriminate between rheumatoid arthritis and spondyloarthritis with the enthesis score. However, in clinical practice, this is probably not relevant because the clinical problem is to distinguish between undifferentiated spondyloarthritis and other causes of low back pain or, in less severe cases, asymmetric arthritis. This was the reason we added the group of patients that fulfilled the inclusion criteria for inflammatory low back pain or arthritis, but not spondyloarthritis diagnostic classification criteria. The results in table 3 show that as ultrasound achieved sufficient sensitivity to discriminate between patients and controls, Doppler in particular achieved statistical differences between normal controls and inflammatory controls (table 3). The isolated presence of enthesopathy is probably not more relevant for a diagnosis than the presence of clinical arthritis in the differential diagnosis of arthritis; however, it can improve diagnostic accuracy. Another limitation is that the study population was not matched for body mass index, a factor that might influence the enthesis score. However, we did not detect clear differences between cases and controls for body mass index.
In summary, entheses involvement occurs early in spondyloarthritis, even in non-radiographic axial patients. Male gender, but not HLA-B27 positivity, was associated with a higher amount of entheses involvement. The presence of entheseal involvement was independent of spondyloarthritis subtype or presentation pattern. The entheses ultrasound score appears to be valid and may be useful for improving the diagnostic accuracy of early spondyloarthritis. Despite these conclusions, its ultimate utility in the core set of spondyloarthritis diagnostic criteria remains to be determined.
The authors would like to thank Drs Loreto Carmona and Mª Jesús García de Yébenes from the Spanish Rheumatology Foundation for statistical advice.
Funding This study was supported by an unrestricted grant from Spanish Rheumatology Foundation and Pfeizer (ESPERANZA programme).
Patient consent Obtained.
Ethics approval This study was conducted according to local regulations and the Declaration of Helsinki, and local approval was obtained from the ethical committee and institutional review board of the Hospital Universitario La Paz, Madrid, Spain.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.