Dual contrast in computed tomography allows earlier characterization of articular cartilage over single contrast

Cationic computed tomography contrast agents are more sensitive for detecting cartilage degeneration than anionic or non‐ionic agents. However, osteoarthritis‐related loss of proteoglycans and increase in water content contrarily affect the diffusion of cationic contrast agents, limiting their sensitivity. The quantitative dual‐energy computed tomography technique allows the simultaneous determination of the partitions of iodine‐based cationic (CA4+) and gadolinium‐based non‐ionic (gadoteridol) agents in cartilage at diffusion equilibrium. Normalizing the cationic agent partition at diffusion equilibrium with that of the non‐ionic agent improves diagnostic sensitivity. We hypothesize that this sensitivity improvement is also prominent during early diffusion time points and that the technique is applicable during contrast agent diffusion. To investigate the validity of this hypothesis, osteochondral plugs (d = 8 mm, N = 33), extracted from human cadaver (n = 4) knee joints, were immersed in a contrast agent bath (a mixture of CA4+ and gadoteridol) and imaged using the technique at multiple time points until diffusion equilibrium. Biomechanical testing and histological analysis were conducted for reference. Quantitative dual‐energy computed tomography technique enabled earlier determination of cartilage proteoglycan content over single contrast. The correlation coefficient between human articular cartilage proteoglycan content and CA4+ partition increased with the contrast agent diffusion time. Gadoteridol normalized CA4+ partition correlated significantly (P < .05) with Mankin score at all time points and with proteoglycan content after 4 hours. The technique is applicable during diffusion, and normalization with gadoteridol partition improves the sensitivity of the CA4+ contrast agent.


| INTRODUCTION
As a consequence of instantaneous impact (eg, related to sports accident), articular cartilage can become injured, leading to the development of post-traumatic osteoarthritis (PTOA). 1 Erosion of articular cartilage, bone remodeling, and joint inflammation are the major characteristic features of osteoarthritis (OA). 2 Often, only in advanced stages of the disease patients experience symptoms, such as pain and limited mobility. Therefore, PTOA is often diagnosed after irreversible damage to the cartilage has already occurred, limiting any possibility of early intervention. Early detection of cartilage damage could enable pharmaceutical or surgical interventions for preventing the progression of OA. 3,4 Early OA is characterized by loss of proteoglycans (PGs), leading to decreased cartilage fixed charge density, lower swelling pressure, and subsequently reduced matrix stiffness. 5 Fibrillation, due to collagen network disruption, also leads to decreased stiffness and increased tissue deformation under physiological loading predisposing the tissue to further degeneration. 6 Today's medical imaging modalities provide multiple methods on how to quantify OA. However, they all suffer from limitations. Ultrasonography provides real-time image acquisition cost-effectively. However, the challenge in achieving perpendicularity between the ultrasound beam angle and the naturally curving cartilage surface limits accurate diagnosis. 7 Magnetic resonance imaging (MRI) is great for soft tissues (eg, cartilage), but it suffers from relatively long scan times and high costs. 8 Computed tomography (CT) imaging is substantially more accessible and affordable, and the image acquisition is swift, and the resolution superior to MRI. Further, significant advancement has been achieved with dose optimization techniques and imaging strategies in CT to reduce the radiation doses involved. 9,10 However, poor soft-tissue contrast prevents separating native cartilage tissue from the surrounding synovial fluid, and, thus, requires the use of contrast agents. 11,12 Delayed contrast-enhanced CT (CECT) has been applied for imaging human articular cartilage in vivo to assess tissue morphology and composition. 13 The diagnosis is based on the evaluation of an anionic contrast agent distribution within cartilage after intra-articular administration. Recently, a cationic CT agent was introduced. 14-16 Cartilage fixed negative charge, created by PGs, provides strong electrostatic attraction to cationic agents. These distribute inside the cartilage in direct relation to the cartilage PG content. 17 For this reason, cationic agents offer a more sensitive technique for diagnosing the distribution of PGs in cartilage compared with conventional anionic agents. 15,[18][19][20] Higher uptake of the cationic agents in cartilage provides higher X-ray attenuation and improves contrast allowing better visualization of PG distribution and its variation within cartilage. Thus, the detection of subtle changes in PG content is possible at different stages of cartilage degeneration. 16,17 Contrast agents diffusion in early time points is fast, especially in degenerated cartilage, due to increased permeability. The uptake of a cationic agent depends both on the electrostatic attraction between the positively charged molecule and the negative fixed charge in cartilage, and the passive diffusion controlled by cartilage water content and permeability. 21 Thus, in degenerated cartilage, the uptake of the cationic agent is simultaneously reduced due to the decrease of negatively charged PGs, and enhanced due to the increase in permeability and water content. These opposite effects limit the diagnostic effectiveness of the cationic agents, especially in the first hours of diffusion, which is vital for the clinical feasibility of the agent. After intraarticular administration, contrast agents diffuse into cartilage, while simultaneously, the body clears out the agent from the joint cavity, lowering the concentration as time progresses. The concentration of anionic ioxaglate in joint cavity has been reported to be adequate for delayed-CECT until 2 hours after the administration, while the agent concentration in patellar and femoral cartilage reached the maximum 30 and 60 minutes after the administration, respectively. 13 The molar concentration of cationic and non-ionic agents in cartilage increases faster compared with an anionic agent. 19,22 Thus, considering the diffusion in cartilage and the clearing out of the contrast agents from the joint cavity, the 30 to 60 minutes imaging time window could be clinically feasible for the application of both cationic and non-ionic agents.
Contrast agent partition in cartilage is quantified as a ratio of contrast agent-induced X-ray attenuation in the cartilage relative to the attenuation in the bath. 23 Normalization (division) of an iodinebased cationic agent (CA4+) partition with an electrically neutral gadolinium-based agent (gadoteridol) partition improved the sensitivity of CA4+ to probe cartilage PG content after 72 hours of diffusion. 24,25 Because water content and permeability of cartilage control the diffusion of the non-ionic gadoteridol the normalization minimizes the effect of these factors on the diffusion of the cationic agent. 25,26 In early diffusion time points, contrast agent diffusion flux is high. 15,22 Further, the agent fluxes are even higher in a degenerated cartilage due to loss of collagen network integrity and reduced PG, resulting in increased permeability. 12,25 Considering this, we hypothesize that the improvement in the sensitivity of the cationic agent after normalization is even more substantial in a degenerated cartilage at early time points. Here we study the diffusion of the agents at clinically relevant time points (<1 hour after contrast agent administration) and at later diffusion time points close to diffusion equilibrium. Further, we examine the validity of the hypothesis by evaluating the sensitivity of normalized CA4+ partition to reflect variation in histopathological and biomechanical properties of human articular cartilage samples. Improvement in the sensitivity of the cationic agent would enable early detection of minor injuries and lesions, allowing timely selection of treatment, thus reducing the risk for PTOA. This quantitative technique is based on the simultaneous diffusion of two contrast agents (iodine-based CA4+ and gadoliniumbased gadoteridol) into cartilage. Accurate simultaneous quantification of concentrations of two contrast agents using single X-ray tube voltage is not possible, as X-ray attenuation of both agents contributes to the attenuation. Hence, as iodine and gadolinium have different x-ray attenuation properties as a function of energy imaging with two separate X-ray tube voltages allows quantitative determination of the concentration of the elements in the mix. Determining the concentration of CA4+ and gadoteridol in cartilage is possible by using the Beer-Lambert law and Bragg's additive rule for mixtures as described in literature [24][25][26] and also in the materials and methods chapter of this paper. As the contrast agents are constantly diffusing in cartilage during a scan, CT acquisition at two different X-ray tube voltages must be nearly instantaneous for accurate determination of contrast agent tissue partition. In this study, we also quantify the error in the partition of the contrast agents arising from the ongoing diffusion in the cartilage when the imaging is performed separately with two X-ray tube voltages.

| Biomechanical measurements
Samples were thawed at room temperature. A custom-made, high precision material testing device (resolution: 0.1 µm, 0.005 N, PM500-1 A; Newport, Irvine, CA) was employed for biomechanical testing of the osteochondral plugs. 30 32 Based on literature, the Poisson ratios were set to ν = (0.3(Tibia), 0.2(Femur)) for E equilibrium and ν = 0.5 for E instantaneous . 33,34 The plugs were then again frozen, cut to two halves, and stored in a freezer (−22°C). One half was thawed for contrast-enhanced microCT imaging experiment, and the other half was prepared for reference histological analyses. Instruments, MA). The contrast agent bath was gently stirred throughout the immersion of the samples. The stirring assembly was placed inside a refrigerator (4°C) to preserve the cartilage and prevent bacterial and fungal growth. For microCT imaging, the samples were removed from the bath, gently blotted on the edges with blotting paper, and placed inside a humidified plastic tube. Scanning was performed using two X-ray energies (tube voltages of 90 and 50 kVp). Gd and I have well-separated K-absorption edges of 50.2 and 33.1 keV, respectively. When using a 50 kVp tube voltage, the maximum fraction of the spectrum was selected to be between the K-edges of I and Gd to maximize the ratio of X-ray absorption caused by I and Gd (µ I /µ Gd ). Similarly, when using 90 kVp, the maximum fraction of the spectrum was selected to be above 50 kVp to maximize the µ Gd /µ I ratio ( Figure S10). The tube current (0.2 mA) was set to the maximum value allowed by the manufacturer to improve the signal to noise ratio. Immediately after imaging, the samples were

| MicroCT imaging
where α is X-ray attenuation in a medium at energy E (tube voltages of 90 and 50 kVp) as, Gd 90kV I 50kV 50kV I 90kV Gd 90kV I 50kV Gd 50kV I 90kV The mass attenuation coefficients for CA4+ (µ I,E ) and gadoteridol

| Histological analysis and Mankin scoring
The osteochondral samples were decalcified in EDTA.

| Error simulation
Error in contrast agent concentrations arising from the progressing diffusion during the time between image acquisitions with two X-ray tube voltages (90 and 50 kVp) was studied using a numerical simulation. To describe the contrast agents diffusion in cartilage, equation was fitted to the experimental data (all the samples in the present study), where C represents I and Gd concentrations in mgI/ml and mgGd/ml, respectively, t is the diffusion time (minutes) and τ is the time required to reach 63.2% of the maximum concentration (C max ). 15 The error simulation was implemented in steps, as follows: Step 1. Fitting was done for each sample individually, after which a mean of the parameters (C max vs τ) for both contrast agents was calculated ( Figure 1).
Step 2. Using equation 1, X-ray attenuation was simulated with both tube voltages, based on the contrast agent concentrations ob- Step 3. Using the simulated data gathered in step 2, concentrations of the contrast agents were calculated (Equations 2 and 3).
Step 4. The simulated concentration values were then compared with the true concentration values (from the fit) to get the relative error, as illustrated in Figure 5.

| Statistical analysis
The statistical analyses were conducted using SPSS (v. 23 (Figures 1 and S11).

| DISCUSSION
In this ex vivo study, simultaneous diffusion of two contrast agents (CA4+ and gadoteridol) into cartilage was evaluated using a microCT scanner at multiple time points to probe cartilage composition and structural integrity. Normalization of the CA4+ partition with that of the gadoteridol improved the correlation with the Mankin score at all time points (Figure 3). Assessing cartilage structural integrity using only cationic agents at early time points is challenging as the uptake is comparably high in both intact and degenerated cartilage, due to high PG content, and increased permeability, respectively. This limits the sensitivity of cationic agents to quantify reduced PG content especially during the early points of contrast agent diffusion. Normalizing the CA4+ partition with that of gadoteridol improves its sensitivity to detect PG content. In this study, a similar effect is seen between CA4+ and PG content where the normalization with gadoteridol partition reveals a significant correlation six hours earlier, beginning at the 4-hour diffusion time point ( Figure 4B).
During very early diffusion (<1 hour time points), the CA4+ significantly correlates with PG content (ie, OD, P < .05) in the superficial zone (10% of the cartilage thickness) ( Figure 4A). Upon inspecting the first 20% of cartilage thickness, the correlation starts to be significant only after 6 hours of diffusion. In this zone, the normalization does not improve correlation with PG content (Table S1). This is likely due to the partial volume effect. In the full thickness cartilage, the correlation with PG content is relatively weak at early time points. However, with the diffusion of CA4+ into deep cartilage, the correlation becomes stronger and is significant (P < .05) after 10 hours of diffusion. The partition of CA4+ increases towards the deep cartilage at later diffusion time points (Figure 2). This is due to the increased electrostatic attraction, resulting from high PG content in the deep cartilage ( Figure S13). 38,39 Concurrently, the water content in cartilage decreases towards the cartilage-bone interface. 5 Thus, expecting the gadoteridol partition to follow the trend of water content in cartilage, it is surprising to see the higher partitions in the deeper zones after the 21-hour diffusion time point.
We suspect that this is a result of X-ray beam hardening, as very high uptake of the cationic agent is observed post 21-hour imaging timepoint in the PG rich deep cartilage (Figure 2). Based on the present experiments and the results, we cannot determine whether the high CA4+ flux could have caused drag influencing the gadoteridol diffusion. 40 Additionally, the overall gadoteridol partition is not observed to rise after a 10-hour imaging time-point ( Figure 1A). The correlation between the equilibrium modulus and CA4+ partition was significant and strengthened by the normalization after 21 hours until diffusion equilibrium. This was expected as the cationic agent's uptake is mostly due to the attraction to PG's, which controls cartilage biomechanical equilibrium response. 41 Therefore, it is natural that the CA4+ which is attracted by the PGs correlates strongly with equilibrium modulus.
Previously, we applied the QDECT technique to evaluate the cartilage PG and water contents at diffusion equilibrium (ie, after 72 hours of diffusion). 24,25 In the current study, we demonstrate the simultaneous   42 However, as all the samples were frozen and thawed following a uniform protocol, any changes in the mechanical/biological state between the samples should be similar.
The samples were immersed in the contrast agent bath maintained at 4ºC. This temperature is lower than that during the intended clinical application of the agents in the human body (37ºC).
The time constant τ of CA4+ was 1032 minutes, being much higher than the value reported for diffusion in bovine cartilage at room temperature (τ =104.4 minutes, cartilage edges not sealed) ( Figure 2). 15,43 With an increase in temperature, the diffusion rate of the contrast agent will also increase. Hence, at a warmer temperature, contrast agent diffusion will be faster, and reliable assessment of cartilage integrity may be conducted at earlier diffusion time point.
The challenges associated with the clinical application of the QDECT are yet to be explored. In the future, the QDECT should be tested on full knee joints using a clinical CT device, to obtain quantitative information