Anterior Delayed Gadolinium-enhanced MRI of Cartilage Values Predict Joint Failure After Periacetabular Osteotomy

Abstract

Background

Several available compositional MRIs seem to detect early osteoarthritis before radiographic appearance. Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) has been most frequently used in clinical studies and reportedly predicts premature joint failure in patients undergoing Bernese periacetabular osteotomies (PAOs).

Questions/Purposes

We asked, given regional variations in biochemical composition in dysplastic hips, whether the dGEMRIC index of the anterior joint would better predict premature joint failure after PAOs than the coronal dGEMRIC index as previously reported.

Methods

We retrospectively reviewed 43 hips in 41 patients who underwent Bernese PAO for hip dysplasia. Thirty-seven hips had preserved joints after PAOs and six were deemed premature failures based on pain, joint space narrowing, or subsequent THA. We used dGEMRIC to determine regional variations in biochemical composition. Preoperative demographic and clinical outcome score, radiographic measures of osteoarthritis and severity of dysplasia, and dGEMRIC indexes from different hip regions were analyzed in a multivariable regression analysis to determine the best predictor of premature joint failure. Minimum followup was 24 months (mean, 32 months; range, 24–46 months).

Results

The two cohorts were similar in age and sex distribution. Severity of dysplasia was similar as measured by lateral center-edge, anterior center-edge, and Tönnis angles. Preoperative pain, joint space width, Tönnis grade, and coronal and sagittal dGEMRIC indexes differed between groups. The dGEMRIC index in the anterior weightbearing region of the hip was lower in the prematurely failed group and was the best predictor.

Conclusions

Success of PAO depends on the amount of preoperative osteoarthritis. These degenerative changes are seen most commonly in the anterior joint. The dGEMRIC index of the anterior joint may better predict premature joint failure than radiographic measures of hip osteoarthritis and coronal dGEMRIC index.

Level of Evidence

Level II, prognostic study. See Instructions for Authors for a complete description of levels of evidence.

Appendix 1

MRI Techniques for Cartilage

Biochemical or compositional MRI techniques for cartilage hold the promise of detecting cartilage damage before irreversible structural damage occurs. At present, there are five techniques designed to probe different macromolecular components of the cartilage matrix: dGEMRIC, sodium MRI, diffusion-weighted imaging, T2 mapping, and T1ρ. Each technique has its own advantages and challenges to practical clinical application. The following is a brief synopsis of the most relevant biochemical imaging techniques for cartilage.

dGEMRIC is a contrast-based technique designed to specifically detect the loss of sulfated glycosaminoglycan (GAG) of articular cartilage. The MRI is performed after an intravenous or intraarticular injection of an anionic contrast agent chelate Gd-DTPA²⁻. This contrast agent is allowed sufficient time to partition into the cartilage before imaging is started. Because the contrast agent is negatively charged, it will partition into the articular cartilage in an inversely proportional manner. By quantitating the amount of contrast agent in cartilage using MRI, the charge density of cartilage can be calculated or inferred. This technique is highly specific for GAG loss in cartilage, mainly due to the use of a negatively charged contrast agent. However, the need for contrast injection poses challenges, which include the need for delay between contrast injection and imaging, need for a specific exercise protocol to facilitate contrast penetration into cartilage, and the risk of contrast reaction. Recently, concern has been raised with the use of contrast agents in patients with poor renal function due to the risk of developing nephrogenic systemic sclerosis, which is a rare and sometimes fatal syndrome. Despite these challenges, if the issues regarding timing of the scan after contrast administration [67] and exercise protocol [46] are respected, reproducible quantitative results can be obtained with a root-mean-square average coefficient of variation of less than 10%. Additionally, validated fast T1 mapping sequences are available so imaging times around 5 minutes are now feasible [37]. In addition to the hip studies outlined previously, multiple clinical studies have been performed using dGEMRIC, demonstrating the effect of exercise on knee cartilage [69], reconstitution of a more normal cartilage after autologous chondrocyte transplantation [24], and perhaps preservation of cartilage after high tibial osteotomy for knee osteoarthritis [56].

The main advantage of dGEMRIC is its specificity for assessing cartilage charge density. Similar specificity is possible but without the need for contrast injection using sodium MRI. In sodium MRI, instead of using gadolinium as the probe, sodium itself is used as the probe to measure charge density. Unlike the typical proton MRI, the nuclear spin momentum of the positively charged sodium ions is used to generate the MRI signal. However, due to the fact that there are far fewer sodium atoms in the body than water, sodium MRI requires minimum 3-T or higher field strength magnets, specialized hardware, and radiofrequency coils. At present, its use has been limited to strictly research purposes, although use of sodium MRI at 7 T was recently published and correlated with dGEMRIC after matrix-associated autologous chondrocyte transplantation [72]. In the future, widespread availability of high-field scanners may make this technique practical.

Some of the other noncontrast MRI techniques include diffusion-weighted MRI, which measures the diffusion characteristics of water molecules in tissue. The diffusivity of water is affected by intracellular and extracellular barriers and is most commonly utilized to detect early central nervous system cell necrosis. It can also provide information regarding the macromolecular environment, which includes GAG and collagen, as well as tissue ultrastructure. It is an attractive technique in the clinical setting because images can be obtained without contrast injection and scan times are relatively quick. However, it is sensitive to motion artifacts, and it is demanding to get absolute quantitative measurements. It has been utilized for evaluation of patients after matrix-associated autologous chondrocyte transplantation to show distinction between healthy cartilage and cartilage repair tissue, but it has had limited use in other clinical settings to evaluate cartilage [39, 75].

T2 mapping, like diffusion-weighted imaging, does not require contrast injection and scan times are relatively short. It assesses interactions between water and collagen fibers to quantify water content and collagen anisotropy [44]. High anisotropy of highly organized collagen will lead to shortened T2 relaxation times. Hence, T2 relaxation times vary from low in the deep layers to higher values in the transitional layers of cartilage due to anisotropy of collagen fibers, which in the deep layers is high with dense collagen matrix and low in the transitional layers [45, 63]. Additionally, disruption of collagen fiber matrices and increase in water content as a result of cartilage damage will increase T2 relaxation times [19, 25]. Areas of high T2 relaxation times in osteoarthritic knees correlate with findings by arthroscopy [8]. However, there are concerns that there is no linear relationship of T2 mapping with severity of osteoarthritis [32]. Nishii et al. [54] compared T2 values and distribution in normal volunteers and patients with hip dysplasia; graded plain radiographs into normal, prearthritic, and mildly arthritic hips; and found no difference in T2 relaxation times among the different radiographic grades. T2 mapping is also sensitive to the loading state of the joint, which may alter T2 signal independent of cartilage damage if joint unloading before scanning is not controlled [2, 40, 49, 52]. Finally, T2 values are sensitive to the orientation of the collagen fibers relative to the B0 magnetic field due to the magic angle effect. In a spherical structure such as the hip, this will need to be taken into account when interpreting the T2 mapping data.

The T1ρ technique is a noncontrast technique that detects low-frequency interactions between water molecules and macromolecular protons. This may allow T1ρ maps in cartilage to be more specific than T2 in assessing GAG loss; however, there is debate and conflicting data regarding this characteristic of T1ρ measurement in cartilage. Some studies have shown T1ρ is sensitive to both proteoglycan and/or collagen content [1, 20, 43]. T1ρ relaxation times have been shown in vivo to correlate with severity of radiographic and MR grading of osteoarthritis [35]. Other studies have shown T1ρ is specific to GAG loss [36, 59, 76]. The most recent study by Keenan et al. [30] found T2 and T1ρ values correlated, and when T2 effects were isolated by looking at tissue only with T2 within normal range, T1ρ correlated with GAG content. They concluded T2 relaxation time should be incorporated into a predictive model when using T1ρ data to estimate GAG content. Comparing T1ρ with T2 mapping, Regatte et al. [60] suggested T1ρ had higher dynamic range, which allowed greater accuracy than T2 mapping. Disadvantages of T1ρ mapping include the length of time required to obtain the multiple data sets and the requirement for the spin lock pulse may increase the specific absorption rate. Although to a lesser extent than T2 mapping, the magic angle effect is also observed in T1ρ [36].

At present, many of these biochemical MRI techniques hold the promise of improving diagnostics and patient management for clinical problems. Although each technique has its own challenges and limitations, the true value of these techniques will not be known until well-performed clinical studies demonstrate their value in the clinical setting.