SpA is characterized by a higher frequency of extra-axial symptoms in children than in adults , and peripheral enthesitis in JIA often precedes the symptoms of axial involvement, in association or not with peripheral arthritis . In JIA, the pain caused by peripheral inflammatory enthesitis can be severe, disabling, and persistent. Nevertheless, due to diagnostic uncertainty, enthesitis is not always recognized and hence may not be treated appropriately [14, 23–27].
The diagnosis of enthesopathy is traditionally based on palpable tenderness at insertion sites . Site-specific tenderness of various origins (e.g., post-traumatic, psychosomatic, inflammatory, and transient growing pains) is common in children [28, 29]. The prevalence of enthesitis in the healthy population is unknown . Deep-seated entheses (e.g., on the iliac crest) are difficult to palpate and when entheseal insertions are close to the joint line (e.g., the patellar ligament and the quadriceps tendon) it may be difficult to attribute palpable pain to enthesitis or joint synovitis . Furthermore, apophysitis (e.g., Osgood-Schlatter disease) [31, 32], osteonecrosis (e.g., Legg-Calvé-Perthes disease), epiphysiolysis, coxitis simplex, or infection may in some cases simulate enthesis involvement in children and adolescents.
In patients with ERA, it has been shown that the number of active entheses and joints at onset can predict sacroiliitis at follow-up . Inflammatory back pain is rarely present at onset of JIA, but sacroiliac and spinal involvement occurs in up to two-thirds of children within 10 years of disease onset [25, 34, 35]. Sacroiliitis often remains clinically unrecognized , and, once it has developed, anti-inflammatory therapies cannot always prevent progression of the condition [37–39]. These observations suggest that it is important to implement early diagnosis and therapy of enthesitis in order to alter the course of the disease.
The present study is descriptive and was not designed to compare results from clinical and US assessments. The inclusion criterion was JIA and palpable tenderness at the posterior iliac crest, a part of the routine investigation of JIA patients at the Pediatric Rheumatology Outpatient Clinic, Rigshospitalet, Copenhagen. Patients without focal clinical symptoms were not presented to the US examiner, who was blinded to other aspects of the clinical status, subtype etc of the children. The same examiner performed the US examinations on patients and healthy controls and was not blinded to whether he was investigating JIA patients or healthy controls, which could be a bias.
Using the ILAR criteria for defining disease, our study shows that enthesitis can occur in other JIA subgroups than ERA as well (Table 1). We also noted that slightly more girls than boys were diagnosed with ERA, which does not agree with reports indicating that ERA usually affects boys more often than girls . The reason for the predominance of girls in our study, as well as in a recent retrospective investigation comparing the ILAR and Amor criteria for SpA in children , is not clear.
Nearly two decades ago, Lehtinen et al.  were the first to demonstrate that US could detect abnormalities associated with lower limb enthesitis in patients with SpA. Later studies have also shown that US is more sensitive than clinical assessment for diagnosing enthesitis in adult SpA patients [40–42] and that Doppler-US is a sensitive method for detecting abnormal blood flow in and around peripheral entheses in adults with SpA [43–47]. In 2005, the Outcome Measures in Rheumatology (OMERACT) network of working groups introduced the following definition of the US signs of enthesopathy: an abnormally hypoechoic and/or thickened tendon or ligament at its bony attachment seen in 2 perpendicular planes that may exhibit Doppler signal and/or bony changes . Other recognized US signs include focal or diffuse loss of normal tendon or ligament fibrillar structure, intratendinous or intraligamentary calcifications, bone erosions, new bone formation (enthesophytes), and associated abnormalities of adjacent bursae [42, 49].
The first MRI studies of peripheral joint involvement in SpA emphasized the extrasynovial nature of the inflammatory process, commonly enthesitis, and showed that this included peri-entheseal soft tissues and bone marrow adjacent to entheseal insertions [12, 50]. The MRI patterns of SpA enthesitis have the following characteristics: edema in the adjacent bone and the surrounding soft tissues, with a high signal on STIR or fat-suppressed T2w sequences, and a low signal on T1w sequences [18, 50, 51]; enhancement of the signal in the soft tissue part of the enthesis on fat-supressed T1w sequences after intravenous injection of gadolinium contrast .
US has been reported to be more sensitive than MRI for early detection of soft tissue involvement in Achilles tendon enthesitis . Also compared to MRI, US has the general advantages of being cheaper, mobile, instantly accessible at bedside, non-invasive, and easy to combine with the clinical assessment (interactivity), and it does not require sedation of young children [53, 54]. In addition, US allows assessment of multiple locations during the same session. Modern high-frequency US transducers provide images with unsurpassed resolution for examination of superficial musculoskeletal structures in children. On the other hand, US does not give a complete picture of all joint structures, because the ultrasound beams cannot penetrate bone. Some areas are hidden by overlying bony structures, such as axial ligament insertions in the spine. For the same reason bone edema is not detectable by US. The main disadvantages of MRI are the high cost of the equipment and that it is frequently inaccessible.
In adults with SpA, US and MRI have been shown to be more sensitive than clinical examination for detecting peripheral enthesitis in the limbs (early signs of inflammation and signs of structural damage) [40, 42, 44, 47, 51, 55, 56]. It is not yet clear what role these imaging methods can play in assessment of enthesitis in JIA, because thus far the pediatric literature contains only one case report  and one investigation comparing US and clinical examination . Our study is the first to concern US and MRI examinations performed on JIA patients to detect enthesitis in a small and deep-seated insertion, a situation in which clinical diagnosis is difficult and imaging would be of great importance.
Enthesopathy at the iliac crest can be of mechanical origin (chronic overuse or acute trauma) or arise secondary to inflammatory disease. Mechanical enthesopathy has been found to occur in both adolescents  and adults [60, 61] who participate in sports, primarily in running, and the changes that were observed in the cited investigations were localized to the anterior part of the iliac crest, and were visualized by MRI, US, and scintigraphy [59–62]. In our study, US signs of inflammatory enthesitis were demonstrated at the insertion of the gluteus medius muscle on the posterior aspect of the iliac crest. It is plausible that the posterior location of enthesitis in our JIA patients can be explained by the important role that the gluteus medius plays in maintaining normal postural control and hip joint motion during walking [63–65]. Peripheral inflammatory enthesitis is more frequently found in the weight-bearing lower limbs in both adults and children [1, 12–14]. The reason for this is not clear, although recent studies have suggested an association between mechanical factors and inflammatory disease [66, 67].
Normal and pathological US features have been described for the distal insertion of the gluteus medius tendon on the greater trochanter (tendinopathy and tears)  but never for the proximal insertion of this muscle. It is important that US appearance of the longitudinal reinforcements of the normal gluteus medius fascia are well recognized in order to avoid any confusion with signs of enthesitis. Our patients, who had hypoechoic entheses on US, had the same regularly distributed fascial reinforcements (Figure 10A), which suggests that there is a predisposition to fascial inflammation in these areas of reinforcement. To avoid diagnostic errors in differentiating normal and inflamed entheses, it is also essential to use a meticulous scanning technique that allows clear interpretation of possible anisotropic artifacts that can easily occur at the level of the entheseal insertion where fibers are curved towards the iliac crest.
Doppler activity was detected in only 37% of the iliac crests in patients with JIA. This suggests either poor sensitivity of the CD technique, possibly due to the deep location of the iliac crest insertion, and/or that a great number of the detected US anomalies were associated with chronic inactive disease. In our study we also displayed the normal US aspect of the posterior iliac crest of children, with a peripheral cartilaginous apophysis on the bony margin of the ilium at the level of the epiphysis (Figure 5A). No secondary ossification center was observed in any of the patients or controls, which can be explained by the fact that ossification usually appears in the anterior part of the iliac crest at the age of approximately 13-15 years . As a child matures, gradual ossification of cartilage occurs in a posterior direction until the age of 25 years .
In a small MRI pilot study of the iliac crest in healthy young individuals we found that the physiological edema associated with normal epiphyseal growth was impossible to distinguish from pathological edema caused by enthesitis. This has been demonstrated in previous studies, at the iliac crest  and at the wrist , and is the reason why MRI of the control subjects was not performed in our study. Consequently, it may be difficult to use MRI to detect any pathological edema caused by enthesitis in children and adolescents. The 38 MRI examinations we conducted revealed signs of enthesitis in only 12 iliac crests in 8 patients. These MRI examinations were performed as routine scannings to detect sacroiliitis (matrix, orientation of scanning planes) and not optimized for imaging of small entheseal insertions on the iliac crest. Because of low spatial resolution the sensitivity of MRI is decreasing with the size of the enthesis being examined. In our study, the detailed anatomical structures of the small soft tissue part of the gluteus medius enthesis that were detected by US were not visualized by MRI; this might have occurred because MRI mainly reveals edema in cartilaginous and bony parts, and hence this method is less sensitive to the changes seen in the soft tissue portion of the enthesis in enthesitis [71, 72]. Furthermore, the discrepancies we observed between the US and MRI findings might have been related to the time span between the two examinations in our study and to the fact that a number of patients received pharmacological treatments with a potential effect on enthesitis. MRI examinations were performed on the same day as US in only four patients, and thus the MRI results for those subjects represented the only true control examinations; those MRI examinations showed no signs of enthesitis or any contrast enhancement.
A weakness of the present study is that only one experienced musculoskeletal radiologist did evaluate the accuracy of the enthesitis US examination.