Can gadolinium contrast agents be replaced with saline for direct MR arthrography of the hip? A pilot study with arthroscopic comparison

Direct magnetic resonance arthrography (MRA) of the hip is widely considered the diagnostic gold standard for the detection of intra-articular lesions in patients with hip deformities such as hip dysplasia or femoroacetabular impingement (FAI) [1, 2]. The desired effect of joint distension is usually achieved with intra-articular injection of highly diluted gadolinium-based contrast agents (GBCA). This procedure is reportedly safe, being associated with minimal postprocedural pain secondary to an inflammatory response to the contrast agent, and does rarely lead to allergic reactions [3]. There has been mounting evidence that GBCA administration at systemic dose levels leads to gadolinium deposition in the body [4]. Yet to date, GBCA deposition on brain magnetic resonance imaging (MRI) has not been reported in the two studies in which MRI subsequent to intra-articular injection of highly diluted GBCA was performed [5, 6]. There has been controversy surrounding potentially chondrotoxic effects related to the intra-articular injection of GBCA [7, 8]. Coupled with the costs of GBCA, patient concerns exist, which can lead to prolonged preprocedural informed consent consultations. Furthermore, reducing costs and patient concerns along with improving patient safety would yield medical and economic benefits alike. Previously, the suitability of alternative agents for direct MRA, mostly using physiologic saline solutions, has been evaluated for the shoulder [9‐11] and the elbow [12]. Comparable diagnostic accuracy in detecting rotator cuff and glenoid labrum lesions was reported for saline and GBCA-MRA of the shoulder [13]. For MRA of the hip, a comparable image quality using hyaluronic acid as an alternative contrast agent was described [14].

Based on encouraging experience using saline as a contrast agent for direct shoulder MRA in our institution, saline MRA was included in our routine protocol for the hip joint.

The aim of this study was to compare image quality and diagnostic performance of direct MRA of the hip performed with GBCA and saline solution in patients undergoing hip arthroscopy.

Although severe adverse events related to intra-articular injection of gadolinium-based contrast agents for direct MR arthrography are extremely rare [3], the use of a saline solution could bypass patient concerns and reduce costs. Therefore, we compared image quality and diagnostic performance in the detection of intra-articular lesions of the hip between Saline- and GBCA-MRA (Figs. 5 and 6).

In addition, interobserver agreement for labrum lesions for Saline-MRA was only fair (ĸ = 0.31), while substantial (ĸ = 0.79) for GBCA-MRA. The higher variability between readers for the detection of labrum tears based on Saline-MRA may be explained by the potentially less intense signal interposition within or at the labral base as the intra-articular fluid was less bright compared to GBCA-MRA. This may be further complicated by the fact that labrum lesions reportedly present as intermediate signal alterations in up to 28% of patients on GBCA-MRA and thus can be confused with mucoid degeneration [30]. This may explain the increased variability between two different readers in diagnosing labrum lesions on Saline-MRA as opposed to the assessment of femoral and acetabular cartilage lesions that showed at least substantial interrater reliability for both techniques. Furthermore, we cannot rule out that differences in reader experience may have contributed to this finding. Literature about non-gadolinium-based contrast agents for direct hip MRA is sparse. One study compared the image quality of hip MR arthrography including fat-saturated fluid-sensitive images with hyaluronic acid against GBCA as an intra-articular contrast agent and reported comparable results for image contrast for cartilage, labrum, and hip capsule. Yet, intra-articular lesions were not compared to surgical findings [14].

In a previous study, Tiegs-Heiden et al. [31] tried to simulate hip MRA with saline by evaluating T2-weighted sequences thereby eliminating the GBCA effect versus assessment of T1-weighted images in 75 patients undergoing GBCA-MRA followed by hip arthroscopy [31]. They reported no inferiority (difference in the area under the curve—0.004; p = 0.90 for the labrum and 0.011; p = 0.79 for acetabular cartilage lesions) for analysis of T2-weighted sequences only in the detection of chondrolabral lesions. They proposed a direct comparison between GBCA- versus Saline-MRA in future studies since this comparison had not been possible in their study [31]. In contrast to the aforementioned studies [14, 31], we performed a comparison between age- and sex-matched patients undergoing either GBCA- or Saline-MRA of the hip followed by hip arthroscopy. Diagnostic performance was high for Saline-MRA and GBCA-MRA for the detection of labrum lesions, femoroacetabular damage, and extensive cartilage damage (all p > 0.05). For Saline-MRA, sensitivity was 99% for labrum, 86% for acetabular cartilage, and 90% for femoral cartilage damage. This is comparable to a previous pilot study [19] on GBCA-MRA of the hip performed under leg traction, which reported a sensitivity of 93% for labrum, 88% for acetabular cartilage, and 86% for femoral cartilage damage [19]. Using traction during MRA of the hip to improve visualization of intra-articular lesions enabled detection rates higher than previously reported in the literature. Specificity in diagnosing labrum lesions was low for either technique (each 33%), which is most likely related to the over-representation of labrum lesions as patients with lesions are more likely to undergo surgery [31, 32]. A meta-analysis of 12 studies including 828 cases reported a pooled sensitivity of 91% and 86% for the detection of labrum lesions with conventional MR arthrography and non-contrast MRI, respectively. Furthermore, a pooled sensitivity of 75% for chondral lesions detected with MR arthrography and 76% for non-contrast MRI was reported [32].

Our results showed that extensive cartilage damage can be detected with high accuracy regardless of the injection of GBCA or saline (93% for GBCA-MRA vs. 97% for Saline-MRA), which is an important predictor for the success of joint-preserving surgery at short and long term [26, 27].

There are several limitations of this retrospective study. First, we cannot directly extrapolate our findings to conventional MR arthrography of the hip, since the application of traction is part of our institutional routine, and imaging with and without traction was not feasible in a routine clinical setting. This should be the subject of future studies. Second, an a priori sample size calculation was not feasible due to a lack of data reporting on the diagnostic performance of non-GBCA-MRA of the hip. Third, no optimization of image contrast for Saline-MRA was performed. Although T1-w images with fat saturation are commonly performed, PD-w TSE imaging without fat saturation is preferred in our institution as it enables the detection of osseous changes and yields good contrast of the cartilage and labrum alike [33]. Although diagnostic accuracy was comparable to a previous study in which T1-w-based traction MRA of the hip was performed [19], our results need to be confirmed on different pulse sequences including T1-w TSE imaging. The application of fat saturation and the use of longer echo times to achieve intermediate-weighted imaging can potentially improve the contrast of the injected saline [33]. In addition, the use of next-generation MRI scanners and hardware may further yield improved image quality. Fourth, even though the radiologists were blinded regarding the use of GBCA/saline, differences in overall image quality may have introduced a bias during the image analysis. Finally, we could only perform a post hoc cost analysis which may differ among countries and healthcare systems and could not compare patient anxiety/adverse events due to the retrospective study design.

To conclude, the image quality and diagnostic accuracy of Saline-MRA and GBCA-MRA in assessing chondrolabral lesions are high. This underlines the potential of omitting GBCA in the future and reducing costs.