Magnetic resonance imaging in polymyalgia rheumatica—contrast enhancement is not always needed

Polymyalgia rheumatica (PMR) is a prevalent rheumatic disease in patients older than 50 years of age and is associated with pain and stiffness, most often affecting the pelvis, hips, legs, neck, shoulders and upper arms [1]. The proposed classification criteria [2, 3] have mainly relied on clinical findings but European Alliance of Associations for Rheumatology (EULAR) imaging [4] and treatment recommendations [5] were recently published.

Although visualisation of extracapsular inflammation by magnetic resonance imaging (MRI, [6‐12]) and ¹⁸F‑fluorodeoxyglucose positron emission tomography with computed tomography (FDG-PET/CT) have been shown to provide characteristic imaging findings suggestive of additional diagnostic value for PMR [13‐21], imaging is currently not recommended in patients with a suspected diagnosis of PMR [22].

Nevertheless, inflammatory changes described to be typical for PMR are, as recently shown [11, 12], displayed as contrast enhancement around tendinous and capsular structures that most likely reflect inflammation of the peritendineum and the outer neurovascular layer of fibrous capsules rather than the synovium making synovitis a mere secondary phenomenon [6, 11, 12]. However, these MRI findings have not been histopathologically confirmed to date.

The inflammatory pattern in the pelvic girdle described to be characteristic for PMR usually consists of numerous bilateral symmetric peritendinitis and pericapsulitis with rather constant involvement of the tendinous origins of the rectus femoris or adductor longus muscles. In a previous study, using contrast-enhanced (ce)MRI bilateral inflammation of four tendinous and capsular sites including proximal rectus femoris or adductor longus muscle origin were very characteristic for PMR and discriminated convincingly between PMR cases and controls [12]. Similar observations were recently published for the shoulder girdle [23].

In clinical routine, however, musculoskeletal MRI beyond specialised rheumatologic imaging is usually and often targeted to structural, i.e. degenerative changes, and is therefore often performed without contrast enhancement. In patients with suspected PMR, especially in the nonspecialised outpatient setting, contrast-free MRI scans of the shoulder or pelvic girdle are often performed to rule out other reasons explaining joint and muscle pain [24]. Using fat-saturated T2-weighted (T2w) MRI extracapsular inflammation is displayed as peritendinous and pericapsular oedema [6, 12].

Based on our experience with PMR and MRI we observed that clearly detectable peritendinous contrast enhancement frequently correlates with only limited or even no peritendinous oedema (Fig. 1). Therefore, noncontrast-enhanced MRI scans are likely to underestimate the extent of inflammatory lesions with the consequence of misinterpretation of this finding.

In this retrospective study we examined peritendinous and pericapsular inflammation in the pelvic girdle by comparing two MRI techniques, one with contrast enhancement and one concentrating on fat saturation without using a contrast agent. The aim was to assess whether the use of contrast agent is necessary to provide evidence for the presence of MRI typical changes at predefined sites in the pelvis of patients with PMR compared to controls.

In this study we show that the ability of fat-saturated contrast-enhanced T1w imaging is in general superior to oedema-sensitive T2w TIRM imaging in the detection of extracapsular inflammation as a potentially discriminating imaging marker for PMR. Furthermore, we analysed the congruity of inflammation detected by these different MRI techniques. To our knowledge this is the first study that compares oedema-sensitive and contrast-enhanced MRI in PMR.

In our previous study we described and defined a stereotypic pattern of extracapsular inflammation in contrast-enhanced pelvic girdle MRI of patients with PMR and concluded that bilateral inflammation of four extracapsular sites in the pelvic girdle including either the rectus femoris or adductor longus muscle origin could be used as a diagnostic test to identify PMR [12]. Here in this revaluation with all MRI rated a second time by radiologists blinded to clinical data we confirm the high diagnostic capability of contrast-enhanced MRI in PMR by showing that cases were discriminated from controls with a sensitivity of 100% and a specificity of 97.1% if inflammation was visible at only three bilateral sites in the pelvic girdle including one key site. All tests performed differed only slightly from our previous results clearly confirming that the characteristic imaging feature of PMR in pelvic girdle MRI is a bilateral inflammation at the key site, rectus femoris or adductor longus muscle origin—in combination with other bilaterally inflamed extracapsular sites. Bilateral inflammation of a key site alone would already discriminate with a sensitivity and specificity of 100% and 95.0%, respectively, while the addition of a varying number of more bilaterally involved sites would mainly contribute to specificity.

Oedema-sensitive imaging correlated moderately with contrast-enhance imaging. Thus, there were 20.7% fewer inflammatory changes detected in the PMR group. Moreover, the reproducibility of oedema-sensitive imaging was inferior to contrast-enhanced imaging as expressed by inferior intra- and interrater correlations.

However, if oedema-sensitive MRI revealed bilateral inflammation at three sites including a key site, patients with PMR were identified with a sensitivity of 80.8% and specificity of 93.3%. In other words, in roughly only 1 out of 5 patients contrast-enhanced MRI will be of additional value. Taking into account the downsides of intravenous gadolinium-based contrast agent use, such as the additional time and effort needed for the longer scanning protocol and the patients’ discomfort by the inevitable intravenous cannula, routine application of a contrast agent does not seem necessary in most pelvic MRI scans when searching for a PMR pattern of inflammation. However, if the presumptive diagnosis of PMR is not confirmed by oedema-sensitive imaging additional application of a contrast agent is likely to increase the diagnostic gain.

No study has so far directly compared the performance of MRI and 18F-FDG PET-CT to provide additional evidence for a diagnosis of PMR. The high sensitivity and specificity values reported for ceMRI suggest that it is as least as good as the technique using radiation with an effective dose after i.v. application of 350 MBq FDG leading, according to the Medical Internal Radiation Dose (MIRD) Committee, to at least 6.7 mSv, possibly reaching 14–18 mSV [24]. This exposure, however, can be minimized by using, for example, low-dose CT. There is a theoretical advantage of using 18F-FDG PET-CT for a potential discovery of malignoma which may mimic PMR symptoms [21]. However, malignomas are rather rarely discovered when searching for the cause of PMR symptoms. Nevertheless, even one case detected early may be reason enough to perform 18F-FDG PET-CT but it is unclear whether additional clinical symptoms cannot provide guidance as to which imaging technique to perform. The value of 18F-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin has been recently established [25].

Finally, we like to stress that there is evidence that the characteristic inflammatory MRI signals can also help to discriminate patients with PMR at other sites including rheumatoid arthritis [26] and that these signals will likely vanish after successful therapy [27, 28].

Contrast-enhanced pelvic magnetic resonance imaging (MRI) uncovers more extracapsular inflammatory lesions in polymyalgia rheumatica (PMR). It is more reliable than oedema-sensitive MRI, but oedema-sensitive MRI shows the typical pattern similarly in most cases. Therefore, application of contrast agents is only necessary if oedema-sensitive imaging is not sufficient to show the typical MRI pattern.