|Home | About | Journals | Submit | Contact Us | Français|
The ability to detect meniscus defects by magnetic resonance arthrography (MRA) can be highly variable. To improve the delineation of fine tears, we synthesized a cationic gadolinium complex, (Gd-DOTA-AM4)2+, that can electrostatically interact with Glycosaminoglycans (GAGs). The complex has a longitudinal relaxivity (r1) of 4.2 mM−1s−1 and is highly stable in serum. Its efficacy in highlighting soft tissue tears was evaluated in comparison to a clinically employed contrast agent (Magnevist) using explants obtained from adult bovine menisci. In all cases, Gd-DOTA-AM4 appeared to improve the ability to detect the soft tissue defect by providing increased signal intensity along the length of the tear. Magnevist shows a strong signal near the liquid-meniscus interface, but much less contrast is observed within the defect at greater depths. This provides initial evidence that cationic contrast agents can be used to improve the diagnostic accuracy of MRA.
Diagnostic imaging of joints usually begins with a radiographic evaluation, which can help detect obvious sources of disease such as advanced arthritis, tumor, dislocation, or impingement. However, X-ray and computed tomography (CT) scans cannot detect tears in cartilage and other soft tissues. The early diagnosis of a tear in the fibrocartilaginous knee meniscus is considered important in order to minimize the potential for the development of significant erosion of the articulating surfaces, with cartilage defects and meniscus tears potentially preceding and possibly leading to additional chondral damage.(1)
Conventional magnetic resonance (MR) imaging is currently the modality of choice for evaluating soft tissue with the ability to detect some intra-articular defects such as large deep cartilage deficiencies or avascular necrosis. However, despite being favored over CT, conventional MR still has a lower than desirable sensitivity at demonstrating tears in fibrocartilaginous tissues and in identifying partial thickness cartilage defects in the articulating surfaces.(2, 3) The diagnostic accuracy of MR imaging is improved by performing magnetic resonance arthrography (MRA), which involves the intra-articular injection of gadolinium (Gd)-based contrast agents, most commonly Magnevist® (i.e. Gd-diethylenetiramine pentaacetic acid, Gd-DTPA). For example, a meta-analysis covering 881 hips across nineteen studies revealed that MRA led to a statistically significant improvement in the sensitivity of detecting acetabular labral tears compared with conventional MR, 87% vs. 66%.(3, 4) However, there is fairly broad variability between these studies, with sensitivity ranging from 60% to 100% and specificity ranging from 44% to 100%.(5) Detecting joint pathologies becomes even more challenging following arthroscopic repair.(6) Importantly, poor/variable sensitivity and specificity is not limited to the hip, but is seen with other major joints including the knee, shoulder, wrist and elbow.(7)
This variability of MRA-based diagnoses of cartilage tears may stem from the use of anionic contrast agents; such as Gd-DTPA (Figure 1A), which has a net charge of 2−. Given their anionic nature, these agents experience electrostatic repulsion from the major constituents of articular cartilage and other fibrocartilages in the joint that possess a high concentration of negatively charged glycosaminoglycans (GAGs). This interaction at the tissue-fluid interface may predispose the agents to poor penetration into narrow crevices, making it difficult to identify fine features, tears, and thinning cartilage. Therefore, we hypothesized that a cationic MR contrast agent, Gd-DOTA-Am4 (Figure 1B; net charge of 2+), would interact with GAGs electrostatically and more efficiently penetrate into fine tears, allowing the liquid-meniscus interface and meniscal defects to be more clearly visualized (Figure 1C and D). Consistent with this hypothesis, two recent studies have shown that the introduction of cationic charges onto CT contrast agents could improve the ability to distinguish cartilage from surrounding tissue via CT, due to the increase in electrostatic interactions with GAGs.(8, 9)
Notably, a number of quantitative radiologic imaging techniques have been developed in recent years to measure changes in cartilage composition. These techniques include delayed Gd-enhanced MRI imaging of cartilage,(10) T2 mapping,(11) T1rho mapping,(12) ultrashort echo time,(13) GAG-specific chemical exchange saturation transfer,(14) and sodium MRI.(15) While these techniques show promise in identifying early stage osteoarthritis, they have primarily been developed to image the biochemical composition of cartilage rather than morphology.(16) Here, we describe the synthesis of a cationic MR contrast agent and specifically report on its ability to detect physical meniscus defects that were introduced into bovine tissue explants.
Gd-DOTA-Am4 was synthesized as outlined in Figure 2. Gd-DOTA-Am4 possesses a longitudinal relaxivity of 4.2 mM−1s−1 (Figure S1A) at 1.41 T. The longitudinal relaxivity of Gd-DTPA was measured to be 3.9 mM−1s−1 (Figure S1B). The pKa values of Gd-DOTA-Am4 were determined to be 2.1, 6.2, and 8.4 according to pH-titration measurements (Figure S2A). Zeta potential measurements revealed that Gd-DOTA-Am4 possessed a 2+ charge at physiological pH (Figure S2B).
To assess whether Gd-DOTA-AM4 interacts electrostatically with GAG-rich cartilage, articular cartilage from juvenile bovine femurs were ground into microparticles. These microparticles were then mixed with Gd-DOTA, Gd-DTPA, or Gd-DOTA-AM4. The microparticles were pelleted and the T1 relaxation time of the supernatant was measured. Measurements were compared to Gd complexes in the absence of cartilage microparticles (Figure S3). Only the supernatant of the cationic Gd-DOTA-AM4 sample that was mixed with microparticles had a longer T1 relaxation time than the analogous sample without microparticles. This finding is consistent with our hypothesis that Gd-DOTA-AM4 interacts electrostatically with the negatively charged cartilage and as a result was partially depleted from the supernatant, resulting in a longer T1 relaxation time. The other Gd complexes, which are both anionic, did not interact with the cartilage microparticles and thus remained entirely in the supernatant, leading to no change in the relaxation time.
The kinetic inertness of the Gd-DOTA-Am4, Gd-DTPA and Gd-DOTA was assessed via a transmetallation assay using equimolar concentrations of the Gd complexes and ZnCl2, in fetal bovine serum. Consistent with previous reports (17–20), the relaxation ratio R1(t)/R1(0) rapidly increased for samples containing Gd-DTPA, indicative of transmetallation (Figure 3A).(21, 22) It should be noted that components of fetal bovine serum (e.g. phosphate, carbonate, etc.) partly contribute to this dissociation.(23) No transmetallation was detected in samples containing Gd-DOTA or Gd-DOTA-Am4 over the course of 72 hours. Therefore, it can be concluded that Gd-DOTA-Am4 is significantly more stable than Gd-DTPA. Moreover, the lack of any detectable transmetallation for at least 3 days, even in the presence of a large concentration of a competing element, suggests that transmetallation is unlikely to occur during a standard MRA procedure.
To further measure the resistance of Gd-DOTA-Am4 to dissociation, the extent of demetallation was monitored as a function of time in a highly acidic solution (1M HCl). Even under these harsh, non-physiologic conditions, Gd remained complexed with DOTA-Am4 for at least 24 hours (Figure 3B). In contrast, both Gd-DTPA and Gd-DOTA exhibited rapid demetallation as indicated by a change in the R1 relaxation rate constant towards that of free Gd. This study further confirms the extremely high stability of the DOTA-Am4 complex, and suggests that its stability may be even higher than that of Gd-DOTA.
To evaluate the efficacy of DOTA-Am4 in highlighting soft tissue tears in comparison to Gd-DTPA, menisci were removed from adult bovine stifle joints (n=9; 3 per group). Biopsy punches were used to produce 8 mm diameter full thickness samples in the axial plane. Well-defined defects were then introduced into these samples by using a 4 mm punch to create an internal core, which was left in place. The tissue explant blocks with defects were placed in a 48-well microplate and bathed in 2mM Gd-DTPA, Gd-DOTA-Am4, or Saline (pH 7.4) for ~30 min. T1-weighted images were then acquired in the axial plane (Figure 4A and Figure S4). Signal enhancement along the defects was most pronounced in explants that were incubated with Gd-DOTA-Am4. The enhancement was heterogeneous along the length of each defect and between samples, but all defects could be visualized. Signal enhancement could also be seen in explants that were incubated with Gd-DTPA, but the signal within the defects was noticeably lower than with Gd-DOTA-Am4 or absent. Little to no signal enhancement was observed in samples incubated with saline.
In addition to the signal enhancement observed within the defects, some signal heterogeneity within the unperturbed areas of the meniscus was also visible, in all samples. This is likely due to structural features within the meniscus (e.g. vascularity, radial tie fibers, etc.). It was easy to differentiate these signals from the actual defects, since they did not extend to the articular surface and typically ran perpendicular to the defects.
To quantitatively compare the contrast-enhancement within the defects, ROIs were drawn around the lateral defects and the signal intensity of each pixel was plotted following background subtraction (Figure 4B–D). The area under curve within the ROI was then quantified and normalized to the length of the tear (Figure 4E). Gd-DOTA-Am4 led to a statistically significant improvement in contrast along the defects compared with Gd-DTPA and saline (p < 0.05). Gd-DTPA led to a statistically significant improvement in contrast compared to saline (p < 0.05).
The safety profile of Gd-DOTA-Am4 was assessed via both cell- and tissue-based assays. When monolayers of isolated bovine meniscal fibrochondrocytes (MFCs) were exposed to basal media containing Saline, Gd-DTPA, or Gd-DOTA-Am4, cell morphology, viability, and proliferation over 24 hours were qualitatively similar between groups (Figure 5). Similarly, after intra-articular injection of Saline, Gd-DTPA, or Gd-DOTA-Am4 into the joint space of an intact bovine femorotibial explant, the area fractions of live and dead cells in the cartilage and meniscus were equivalent after 3 hours (p>0.05) (Figure 6).
The data presented here support the hypothesis that positively charged contrast agents can improve the identification of tears in the knee meniscus, and likely other GAG-rich cartilaginous tissues of the major joints. This is expected to translate to an improvement in the diagnostic accuracy of MRA. We hypothesize that the increase in contrast is due to electrostatic interactions between Gd-DOTA-Am4 and GAGs, which enables more efficient penetration into the fine tears or defects within the meniscus and other cartilaginous tissues. Conversely, it is likely that Gd-DTPA was subject to electrostatic repulsion from GAGs, which at least partially excluded this agent from entering the defect. Similar electrostatic interactions with cartilage were recently observed with cationic and anionic CT contrast agents. (8, 9)
One concern with the use of Gd for any imaging application is the possibility of causing nephrogenic systemic fibrosis (NSF), which has been associated with the intravascular injection of Gd-based contrast agents in renally impaired patients. However, it is important to note that MRAs only require an intra-articular injection of Gd and the total injected dose is orders of magnitude less than what is used for studies requiring intravascular injection. Moreover, our preliminary findings suggest that Gd-DOTA-Am4 is remarkably inert in serum as well as in harsh, non-physiologic acidic conditions. We did not observe any cytotoxicity following a 24 hr incubation with isolated cells or after 3 hrs of incubation in the synovial fluid of the femorotibial joint. Although, studies have thus far been limited to tissue explants, in future work we plan to develop an in vivo model of cartilage defects to further evaluate the efficacy of Gd-DOTA-Am4 as well as its stability, toxicity, and elimination from living subjects.
Compound 1 (Figure 2) and bromine were purchased from Fisher Scientific (Philadelphia, PA) and 1,4,7,10-tetraazacyclododecane (4) was purchased from Strem Chemical (Newburyport, MA). Diethylenetriaminepentaacetic acid (DTPA) gadolinium complex was purchased from Aldrich (St. Louis, MO) while 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) gadolinium complex was purchased from Macrocyclics (Dallas, Texas). HyClone™ Fetal Bovine Serum was purchased from GE Healthcare Life Sciences (Logan, Utah).
Elemental analysis was performed on C, H, N and Gd by Intertek. 1H NMR spectra were acquired with a Bruker Avance-360 spectrometer. The relaxometric studies were performed using a Bruker mq60 tabletop MR relaxometer operating at 1.41 T.
Compound 2 was prepared as reported in a literature.(25) 1H NMR (CDCl3) δ 3.76 (3H, -OCH3, s), 6.21 (1H, HC=CBr, d), 6.73 (1H, HC=CBr, d).
To a stirred solution of compound 2 (19.1g; 0.11 mol) in 200 mL of diethylether at 0°C was slowly added diethylamine (7.81 g; 11.0 mL; 0.011 mol). The solution was stirred for 3h at room temperature and dried under vacuum. The resulting brown liquid was used immediately for the next step, due to instability.
1, 4, 7, 10-tetraazacyclododecane (4; 1.0 g; 5.81 mmol; Strem Chemicals, Newburyport, MA), anhydrous potassium phosphate, tribasic, (5.0 g; 23.2 mmol) and compound 3 (5.85 g; 23.2 mmol) in 150 mL of dry acetonitrile were stirred at 60°C for 48 h. The solid base was filtered and the solvent was dried under vacuum. The title compound was purified by chromatography on silica column (chloroform:methanol 7:1) to give final product (45%). m/z (ESI); 801 (M+H). Elemental analysis: (C40H80N8O8.HBr) calculated C 54.6, H 9.19, N 12.71 ; found C 54.81, H 9.07, N 12.93. 1H NMR (360 MHz, CDCl3) 0.8–1.5 (12H, -CH3, m), 2.4–2.8 (8H, -CH2-, m, br), 2.9–3.0 (8H, -CH2N-, m, br), 3.1–3.3 (8H, -NCH2CH2N-, m), 3.3–3.5 (4H, -CH-, m), 3.6–3.8 (12H, -OCH3, br).
Compound 5 (1.0g; 1.3 mmol) was stirred in 100 mL of methanol:0.5M NaOH (40:60) at room temperature for 48 h. The reaction mixture was neutralized with 2M HCl and dried under vacuum. The solid residue was re-dissolved in methanol, filtered and the solvent was removed under vacuum to yield the title compound (60%). m/z (ESI); 261 (M+2H). Elemental analysis: (C20H40N8O8.NaBr) calculated C 38.5, H 6.42, N 17.98 ; found C 38.21, H 6.27, N 17.64. 1H NMR (360 MHz, (CD3)2SO) 1.0–1.4 (8H, -CH2-, t), 2.7–2.9 (4H, -CH-, q), 3.0–3.8 (8H, -NCH2CH2N-, m, br), 8.8–9.1 (8H, -NH2, s, br).
Gd-DOTA-Am4 complexes were synthesized by refluxing the metal (Gd(NO3)3.6H2O) and ligand 6 in absolute methanol as previously described.(24) m/z (ESI); 337 (M-2). Elemental analysis: (C20H36N8O8Gd) calculated C 35.6, H 5.34, N 16.6, Gd 23.4; found C 35.1, H 4.95, N 16.1, Gd 23.1. To purify Gd-DOTA-Am4, the solvent was reduced to 1mL and 10 mL dichloromethane was added. The top solvent layer was decanted. The bottom oil was redissolved in a minimum amount in methanol and 10 mL of dichloromethane was added. The top layer was decanted and the bottom oil was dried. Gadolinium concentration was determined by ICP-OES analysis using an Elan 6100 ICP-MS (Perkin-Elmer, Shelton, CT) at the New Bolton Center Toxicology Laboratory, University of Pennsylvania, School of Veterinary Medicine, Kennett Square, PA, USA. The presence of free metal ion was assessed by xylenol orange assay (26). The complex was found to exhibit a purity of >98.5%.
Gadolinium concentration was determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES) (Spectro Analytical Instruments GMBH; Kleve, Germany). Longitudinal relaxation times (T1) were measured in saline at pH 7.4 using a Bruker mq60 tabletop MR relaxometer operating at 1.41 T. The longitudinal relaxivity (r1) of the complex was calculated by plotting the reciprocal of the T1 relaxation time versus gadolinium concentration.
pH titration was performed using an Orion Research Digital Analyser 501 equipped with an Accumet® semi-micro pH electrode. A 10 mL solution of 1mM Gd-DOTA-Am4 and 0.1 M KCl was prepared and its pH was adjusted to 3.0 using HCl solution. A 1M NaOH solution, purchased from Aldrich (St. Louis, MO), was diluted to 50 mM and used throughout the titration.
All chemicals in this study were purchased from Aldrich (St. Louis, MO). The zeta potential were recorded as a function of charge at pH 7.4 using a Malvern Zetasizer Nano ZS as previously described.(27) The pKa values of the various compounds have previously been reported.(27) All samples were measured in triplicate. 100 mM stock solutions were prepared and the pH was adjusted to 7.4 using KOH and HCl. Samples were prepared by dilution of stock solution to 1.0 mM with 1.0 mM solution of KCl at pH 7.4 and allowed to equilibrate overnight.
Articular cartilage from juvenile bovine femurs were minced and lyophilized overnight to remove water content. The dried tissue was disrupted and homogenized using a compact bead mill, where samples were shook at 50 Hz with stainless steel beads (TissueLyser LT, Qiagen). A 1 mM stock solution of Gd-DOTA-Am4, Gd-DOTA and Gd-DTPA were prepared in saline at pH 7.4 and their T1 relaxation times were recorded. 600 L of the above-prepared Gd solutions were added to 50 mg of the semi-powder cartilage microparticle samples, respectively. The suspension was sonicated for 1 min and allowed to stand at room temperature for 5 min. The suspension was then centrifuged and 300 L of the supernatant was removed and the T1 relaxation times were recorded. All measurements were performed in triplicate.
To compare the inertness of the Gd-DOTA-Am4 complex to Gd-DTPA and Gd-DOTA, transmetallation and demetallation assays were performed. Transmetallation, i.e. replacement of Gd3+ with Zinc2+, was evaluated in fetal bovine serum. The solution contained 4 mL of 2.5 mM Gd-complex and 2.5 mM ZnCl2 at pH 7.4. The mixture was stirred at 37°C and 300 µL aliquots were removed at various times for measurement of the longitudinal relaxation time. Relaxation times were measured using a Bruker mq60 tabletop MR relaxometer operating at 1.41T. The ratio of the relaxation time at each time point (R1(t)) and relaxation time at time zero (R1(0)) (i.e. in the absence of ZnCl2) was plotted against time.
Demetallation, i.e. loss of Gd from the complexes, was induced by incubating 300 µL of 2.5 mM Gd-compound at 37°C in 1M HCl. Longitudinal relaxation times were acquired at various time points. Demetallation was monitored by changes in the T1 relaxation time, since T1 relaxation time could not be acquired at t=0 for this assay.
Adult bovine stifle joints were purchased from a commercial vendor (Animal Technologies, Tyler, TX), dissected in a sterile field, and the menisci removed. Biopsy punches (Miltex, York, PA) were used to produce 8 mm diameter full thickness samples in the axial plane. Well-defined defects were then introduced by using a 4 mm punch to create an internal core, which was left in place. The meniscus samples were placed in a 48-well plate, bathed in 2mM [Gd- DOTA-AM4]+3, [Gd- DTPA] −1, or saline (pH 7.4) for 30 minutes and imaged via MR using a T1-weighted sequence.
Magnetic Resonance images were acquired using a 4.7 T small animal horizontal bore Varian INOVA system. T1-weighted images were acquired in the axial plane using the following parameters: repetition time (TR) = 2000 ms, echo time (TE) = 20 ms, FOV = 40 × 40 mm, flip angle = 90°, slice thickness = 1.0 mm, number of requisition = 2, matrix = 256 × 256 pixels.
A region of interest (ROI) was drawn around the defect for each MR image slice (n=6 per contrast agent, i.e. 2 slices per sample). All ROIs had a width of 27 pixels, which encompassed the entire tear for all slices. The ROI height was determined for each slice individually, such that it could span the sample from top to bottom and exclude media and unrelated bright areas within the cartilage (e.g. cavities near the liquid-cartilage interface). In samples where there was clearly a pool of solvent within the tear, the ROI spanned the longer of the two segments between the pool and the top or bottom of the cartilage.
The signal intensities for each line within an ROI were plotted against their position and the bottom 80% of values were used to determine a zero-intensity baseline for that line. The trapezoid method was used to integrate the area under the curve. All of the integrations in an ROI were then averaged to quantify that ROI's average signal intensity.
To determine whether the DOTA contrast agent was cytotoxic, both cell- and tissue-based assays were employed. Using a cell-based approach, confluent monolayers of meniscal fibrochondrocytes (MFCs) isolated from adult bovine menisci were cultured in 6-well plates and exposed to basal media (BM; Dulbecco’s Modified Eagle’s Medium with 10% Fetal Bovine Serum and 1% Penicillin/Streptomycin/Fungizone). Three experimental conditions were tested: BM diluted 1:1 with either saline (Saline; pH 7.4), Gd-DTPA in saline (2 mM final concentration), or DOTA-Am4 in saline (2 mM final concentration). Cells incubated in BM were used as a positive control. MFCs were maintained at 37°C and images were captured at 1, 4, and 24 hours with a light microscope at 4× magnification (n=2/group).
To further examine biocompatibility, Gd-DOTA-Am4 was injected directly into juvenile bovine femorotibial joints. Three experimental conditions were tested: Saline, Gd-DTPA in saline (2mM), and DOTA-Am4 in saline (2 mM). Using a lateral para-patellar approach, ~15 mL of solution was injected into the femorotibial joint space using a 22 gauge needle (n=1/group). Joints were maintained at 25°C and flexed and extended repeatedly every hour post-injection. An intact joint was used as a non-injected control. After 3 hours, joints were dissected and the medial meniscus and cartilage from the medial femoral condyle were isolated with 4 mm biopsy punches, with care to keep the superficial tissue surface (areas in contact with the synovial fluid) undamaged. Cell viability was assessed via a LIVE/DEAD® Viability/Cytotoxicity Kit (Invitrogen). Images of live and dead cells at the superficial tissue surface were captured in the green (FITC) and red (TRITC) channels, respectively, at 10× magnification (n=4/group). These images were binarized in ImageJ and the area fraction of each signal in their respective channel was computed for a 300 × 300 µm2 region of interest.
All data is presented as the mean ± standard deviation. A student t-test was used to determine significance (p < 0.05) between the average MR signal intensity along the defects in the meniscus, for each contrast agent. t-tests were performed in Microscrot Excel.
All statistical analyses related to live/dead cell measurements were performed using SYSTAT (Chicago, IL). Significance was assessed by two-way ANOVA with Tukey’s HSD post hoc test to make comparisons between groups (p≤0.05). Quantitative measures include the area fractions of live and dead fluorescent signal.
We would like to thank David R. Vann, Department of Earth and Environmental Science, University of Pennsylvania, for assistance with ICP-OES.
Contract/grant Sponsors: National Institute of Health NIBIB R01EB012065 (AT), NCI R01CA157766 (AT), and Department of Veterans Affairs VA RR&D I01 RX000174 (RLM).