Contrast Enhanced Computed Tomography can predict theglycosaminoglycan content and biomechanical propertiesof articular cartilageP. N. Bansalyz, N. S. Joshix, V. Entezariy, M. W. Grinstaffzx* and B. D. Snydery*yOrthopedic Biomechanics Laboratory, Beth Israel Deaconess Medical Center andHarvard Medical School, Boston, MA, United StateszDepartment of Biomedical Engineering, Boston University, MA, United StatesxDepartment of Chemistry, Boston University, MA, United States

Summary

Objective: An early hallmark of osteoarthritis (OA) is the progressive loss of glycosaminoglycans (GAGs), the extracellular matrix (ECM)component of articular cartilage that confers it with compressive stiffness. Our aim in this work is to establish the feasibility of using ContrastEnhanced Computed Tomography (CECT) with an anionic iodinated contrast agent e Cysto Conray II e as a minimally invasive tool tomeasure the changes in the GAG content as well as the compressive stiffness of articular cartilage.

Methods: The GAG content of mated osteochondral plugs excised from bovine patello-femoral joints was progressively degraded usingchondroitinase ABC. The mated plugs were then immersed in an anionic, tri-iodinated contrast agent, imaged using peripheral quantitativecomputed tomography (pQCT), subjected to an unconfined compressive stress relaxation test and the GAG content measured using 1,9-dimethylmethylene blue (DMMB) assay. Partial correlation analysis was performed to compare the variation in X-ray attenuation measuredby pQCT to the variation in GAG content and in equilibrium compressive modulus.

Results: The X-ray attenuation of cartilage exposed to an anionic, tri-iodinated, contrast agent measured by quantitative computed tomogra-phy (QCT) accounted for 83% of the variation in GAG content (r2! 0.83, P< 0.0001) and 93% of the variation in the equilibrium compressivemodulus (r2! 0.93, P< 0.0001).

Conclusion: Using a mated osteochondral plug model to evaluate the biochemical composition and biomechanical properties of cartilage, thisstudy demonstrates the interrelationships between X-ray attenuation, GAG content, and equilibrium compressive modulus, and that CECT canbe used to monitor and quantify changes in the GAG content and biomechanical properties of articular cartilage.ª 2009 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Key words: constrast agnet, computed tomography, cartilage, equilibirum compressive modulus, CECT, GAG.

Introduction

Hyaline cartilage is the load bearing tissue that supportsand evenly distributes mechanical loads across articularjoints and provides joints with a smooth, nearly frictionlessgliding surface. On a molecular level, the composition andarchitecture of cartilage determine its tissue structural prop-erties. Articular cartilage is comprised primarily of proteogly-cans (5e10% wet weight.), type II collagen (10e20% wetweight.) and water (68e85%). By binding interstitial water,the negatively charged proteoglycans, which form the

extracellular matrix (ECM), confer cartilage with its com-pressive stiffness. Swelling of the ECM by bound water isresisted by collagen fibrils arranged in a dense network ofhighly oriented fibrils, and this network opposes the tensileand shear forces applied to articular surfaces1,2.Osteoarthritis (OA) is a disease that degrades articular

cartilage. It has been established that the concentration ofglycosaminoglycan (GAG) in cartilage is an indicator of car-tilage health3. In early OA, GAGs are lost from the ECM andthis decrease in GAG concentration reduces the resilienceof articular cartilage to compressive loads4. Unfortunately,this deterioration of articular cartilage occurs before a lossof cartilage volume is evident radiographically5,6. Conse-quently, imaging methods that can non-invasively or withminimal invasiveness measure changes in the GAG contentand structural properties of articular cartilage before clinicalsymptoms and radiological changes are evident may pro-vide physicians with an important opportunity to diagnoseOA early. These methods will also be important to seriallyevaluate the efficacy of disease modifying drugs andcartilage repair strategies.

*Address correspondence and reprint requests to: Brian D.Snyder, Orthopaedic Biomechanics Laboratory, Beth IsraelDeaconess Medical Center, 1 Overland Street, RN 115, Boston,MA 02215, United States, Mark W. Grinstaff, Departments ofBiomedical Engineering and Chemistry, Boston University, 590Commonwealth Ave, Boston, MA 02215, United States. Tel:1-617-358-3429; E-mail: mgrin@bu.edu, bsnyder@bidmc.harvard.eduReceived 19 May 2009; revision accepted 9 September 2009.

Osteoarthritis and Cartilage (2010) 18, 184e191ª 2009 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.joca.2009.09.003

184

Magnetic resonance imaging (MRI) has been used rou-tinely to measure the thickness of cartilage and to observejoint surface topography7e9. Several MRI-based techniques(e.g., T2 mapping, T1rho, and 23Na" mapping) have beendeveloped that quantitatively reflect the GAG and the colla-gen content of articular cartilage10e15. The delayed gadoli-nium enhanced magnetic resonance imaging of cartilage(dGEMRIC) uses a contrast agent, gadolinium (Gd-DTPA2#), that acts as a mobile anionic probe that partitionsitself throughout the ECM in inverse proportion to the fixednegative charge density of the proteoglycans comprisingthe ECM (Fig. 1). Since the T1 relaxation time is inverselyproportional to the concentration of Gd-DTPA2#, the map-ped T1 relaxation time is directly proportional to the concen-tration and spatial distribution of GAG throughout thearticular cartilage. Studies have confirmed that dGEMRICcan differentiate between healthy and arthritic cartilageboth in vitro and in vivo16e22. However, there have beenconflicting reports as to the reliability of dGEMRIC for pre-dicting the GAG concentration and compressive stiffnessof cartilage23e28. Further, widespread use of MRI-basedtechniques to measure the earliest biochemical changesin articular cartilage induced by OA has been hindered bythe high cost, long acquisition time and relatively low spatialresolution of clinical MRI scanners, as well as the need forspecialized pulse sequences and software programs to an-alyze and map the GAG content of cartilage.Recently, following a procedure analogous to dGEMRIC,

quantitative computed tomography (QCT) has been used tomeasure the concentration and spatial distribution of an an-ionic, iodinated contrast agent diffused into isolated articularcartilage explants29,30. Since the magnitude of ComputedTomography (CT) attenuation is directly proportional tothe concentration of iodine partitioned throughout theECM, the Contrast Enhanced Computed Tomography(CECT) attenuation is inversely proportional to the GAGcontent of the cartilage. Apart from the radiation exposureconcerns with repetitive CT scans, CT imaging is less ex-pensive, more widely accessible and able to achieve higherspatial resolution over shorter acquisition times comparedto MRI. Moreover, advances in CT technology, such as he-lical CT scanners using multiple row detectors, allow rapid3D reconstruction and segmentation of cartilage and sub-chondral bone without the need for specialized sequences.Unconfined compressive stress relaxation tests have

been widely used as a standard method to evaluate theequilibrium biomechanical properties of articular cartilage31.

It is known that the GAGs are responsible for the static com-pressive properties of articular cartilage4,32,33. Hence, it isreasonable to hypothesize that the CT attenuation obtainedfrom CECT images can be used as a gauge of the staticmechanical properties of articular cartilage. The equilibriummechanical properties are commonly obtained by separat-ing the cartilage surface from the subchondral bone. How-ever, employing osteochondral plugs circumvents theproblems of curling and swelling that occur when cartilageexplants are separated from the subchondral bone. Further-more, when cutting cartilage away from the subchondralbone, it is possible that the deep zone cartilage absentfrom the cartilage explant will affect the mechanical proper-ties measured for that explant. Thus, using congruent andconforming osteochondral plugs excised from opposingjoint surfaces may provide a more realistic model especiallyfor clinical applications of CECT to indirectly ascertain themechanical health of the joint, whereby two cartilage sur-faces interact.Consequently, obtaining information about the biochemi-

cal content and biomechanical properties of articularcartilage is important for basic studies as well as clinicalapplications. Herein we report the use of CECT with aniodinated, anionic contrast agent to measure the GAGcontent and compressive stiffness of articular cartilage.

Materials and methods

SPECIMEN PREPARATION

A diamond tipped cylindrical cutter irrigated with 0.9% saline at room tem-perature was used to extract mated, cartilage on cartilage, osteochondralplugs (7 mm diameter) from opposing articular surfaces of the five patella-femoral joints from 1 to 2 years old freshly slaughtered cows [Fig. 2(a)].Due to the requirement of harvesting matched osteochondral pairs, we har-vested between three and five pairs from the five patella-femoral joints.These 15 pairs of mated osteochondral plugs were used in this study. Pairsof conforming osteochondral plugs were randomly assigned to one of threegroups (five pairs/group): normal control or one of two GAG depleted groupsdegraded by immersing the osteochondral plugs in chondroitinase ABC(SigmaeAldrich, C3667, St. Louis, MO, USA) [0.1 U/mL in 50 mM Tris,60 mM NaOAc, 0.02% BSA, pH 8.0] at room temperature for 8 or 30 h. Atthe end of the digestion period the samples were washed twice in 200 mLof 0.9% saline and then immersed in 7 mL of contrast agent, which signifi-cantly reduced the concentration of the enzyme to continue any furtherdigestion. Chondroitinase ABC is a GAG specific hydrolase that has beenused to deplete GAGs from ECM of articular cartilage34e36. This particularenzyme was chosen, instead of other proteases like trypsin, because prelim-inary studies indicated that it gave the most reproducible GAG degradationpatterns in the osteochondral plug samples and it has been shown to bevery specific for GAG and does not affect the collagen content37,38.

Fig. 1. An anionic contrast agent diffuses into articular cartilage matrix in inverse proportion to the fixed negative charge density conferred byGAGs. In healthy cartilage diffusion of the anionic contrast agent into the matrix is limited due to the high GAG content. As GAG becomes

depleted, more of the anionic contrast agent can partition into the matrix.

185Osteoarthritis and Cartilage Vol. 18, No. 2

CECT

Each mated pair of osteochondral plugs was immersed in 7 mL of an an-ionic, tri-iodinated contrast agent (Cysto Conray II, Mallinckrodt, MO, USA)for at least 12 h to allow sufficient diffusion time for the contrast agent to par-tition itself throughout the ECM (see Supplementary material for details aboutthe diffusion time). A line profile across the diameter of the plug confirmedthat the contrast agent had completely diffused into the cartilage (see Sup-plementary material, for a representative figure). Cysto Conray II contains81 mg of iodine per mL and has a net charge of negative 1 [#1]. The contrastagent concentration was not altered from the commercial formulation(81 mgI/mL and 0.213 M) and mixed with an antibiotic cocktail (0.01%NaN3 and 10 mM EDTA) to prevent bacterial contamination. Cysto ConrayII has an osmolality of 403 mOsm/kg39,40, which is very close to the in situosmolality of articular cartilage (350e450 mOsm/kg)41. After blotting the car-tilage surface to remove excess contrast agent, the plugs were positioned inthe gantry of the peripheral quantitative computed tomography (pQCT) scan-ner (XCT Research SA", Stratec Medizintechnik GmbH, Pforzheim,Germany) using a custom airtight holder which contained saline soakedgauze to prevent drying of the cartilage. Four sequential, 100 mm thick, trans-axial pQCT images were obtained at 70 mm in plane resolution and 1 mm in-ter-slice distance. The image processing and thickness measurements wereperformed using a commercial image processing software (Analyze!, BIR,Mayo Clinic, Rochester, Minnesota, USA). The CT data sets were importedinto Analyze! and the cartilage was segmented from the subchondral boneusing a semi-automatic contour based segmentation algorithm. During thesegmentation process a seed point was placed in the region of interestand a contour was ‘grown’ around the seed point until the region of interest(cartilage) was segmented, the same process was applied to all the four sli-ces. The mean thickness of the cartilage surface and mean X-ray attenuationfor cartilage was obtained by averaging over the four transaxial CT images.The average CT attenuation in this study is reported in Hounsfield Units (re-fer Supplementary material for details on the Hounsfield Value computation)and was obtained without applying any thresholds or image manipulation.The average CT attenuation was assumed to reflect the GAG content ofthe tissue volume used in this study.

MECHANICAL TESTING

Scans for representative samples in this study confirmed that 6e8 h weresufficient to desorb the contrast agent from the cartilage. Hence, each pair ofosteochondral plugs was immersed in excess (15 mL) physiologic balancedsaline at room temperature overnight (w15 h) to desorb the contrast agent.Using a custom built fixture, each pair of osteochondral plugs was thenmounted co-axially in a mechanical testing machine (Instron 8511, Instron,Norwood, MA, USA). The cartilage surfaces of both plugs were completelyimmersed in saline [Fig. 2(b)] and incremental, stepwise compressive stressrelaxation tests were performed. An initial tare load of 0.25 N was appliedand allowed to ‘‘creep’’ to zero load to ensure congruent contact between op-posing cartilage surfaces. A series of four, incremental, 5% strain steps wereapplied to 20% of the total undeformed thickness of the mated cartilage sur-faces42. The average thickness for the femoral and patellar cartilages was1.36$ 0.19 mm and 1.84$ 0.19 mm respectively [mean$ standard

deviation] (Table I). After each strain step, the samples were allowed to relaxuntil the change in force was less than 0.05 N/min (<1250 Pa/min) (Fig. 3).The equilibrium compressive modulus (Eeq) was obtained from the slope ofthe line fit to the equilibrium stress-strain data using a least-squares tech-nique (MATLAB v7.6, Mathworks, Natick, MA, USA).

DETERMINATION OF GAG CONTENT

The articular cartilage was separated from the subchondral bone usinga razor blade in order to obtain the wet mass of cartilage for each osteochon-dral plug. The cartilage was lyophilized for 24 h and digested in papain(0.5 mg/mL in 50 mM sodium phosphate, 1 mM EDTA, 2 mM DTT, pH 8.0at 65%C) for 24 h and then diluted 10e100 times for the 1,9-dimethylmethy-lene blue (DMMB) colorimetric assay43. A GAG reference calibration curvewas generated using chondroitin-4-sulfate (SigmaeAldrich 27042, St. Louis,MO, USA) dissolved in PBS at concentrations ranging from 10 to 100 mg/mL.10 mL of each chondroitin-4-sulfate dilution was combined with 100 mL ofDMMB solution in a 96-well plate. The absorbance at 520 nm was measuredat each well using a plate reader (Beckman Coulter AD340, Fullerton, CA,USA). From the linear regression generated from the GAG calibration curveexpressing absorbance at 520 nm as a function of chondroitin-4-sulfate con-centration, the total GAG weight per mg wet weight of cartilage for each os-teochondral plug was calculated.

STATISTICAL ANALYSIS

Normality of the samples was determined by using the ShapiroeWilk test.To account for multiple dependant samples across the groups, partialcorrelation analysis was used to express the X-ray attenuation of the anioniccontrast agent partitioned throughout the cartilage measured by QCT as a lin-ear function of the GAG content measured by the DMMB assay and of theequilibrium compressive modulus (Eeq) measured mechanically. In addition,the same associations were reported for CT attenuation vs GAG correlationin the control group alone. The coefficient of determination was used to testthe strength of each association. Only statistically significant results,P< 0.05, are reported. All statistical analyses were conducted using SPSS17.0 (SPSS, Chicago, IL, USA).

Table IMean thickness data for each group used in the study. The data is

presented as mean$ standard deviation

Group Femoral groovecartilage thickness inmm (mean$SD)

Patellar cartilagethickness in

mm (mean$ SD)

Control 1.35$ 0.08 1.79$ 0.19Degraded e 8 h 1.38$ 0.13 1.83$ 0.16Degraded e 30 h 1.35$ 0.18 1.89$ 0.12

Fig. 2. (a) Mated pairs of conforming osteochondral plugs extracted from bovine patello-femoral joint where the highlighted region defineswhere plugs were harvested (dashed lines). The mated patello-femoral plugs were randomly assigned to three groups (5 pairs/group).(b) Schematic diagram showing the mechanical testing set-up. A e fixture to hold the top plug, B e aluminum molds to concentricallyhold the plugs and pot with PMMA, C e chamber to hold bottom plug and saline, D e set screws to hold the aluminum molds in place,

E e physiologic saline.

186 P. N. Bansal et al.: Contrast Enhanced Computed Tomography

Results

COMPARISON OF X-RAY ATTENUATION WITH THE GAGCONTENT OF CARTILAGE

The anionic, tri-iodinated contrast agent diffused into theECM of articular cartilage in inverse proportion to the GAGcontent. The X-ray attenuation values increased with de-creasing GAG content [Figs. 4 and 5(A)]. In degraded os-teochondral plugs, the increase in X-ray attenuation wasmost evident in the superficial zone and less evident inthe deep zone of the articular cartilage (Fig. 4). In the Tolu-idine-blue stained histological sections, the blue stainingprogressively decreased with increasing GAG degradation[Fig. 4(B)]. The change in X-ray attenuation was linearlyand inversely related to the GAG content of the articular

cartilage [r2! 0.83, P< 0.0001, Fig. 5(A)]. An equally highcorrelation was observed between X-ray attenuation andGAG content when the samples in only the control groupwere compared separately (Fig. 6). Figure 6 shows thatCT attenuation predicts 83% of the variation in the GAGcontent in the control group alone.

COMPARISON OF EQUILIBRIUM COMPRESSIVE MODULUSWITH GAG CONTENT OF CARTILAGE

The equilibrium compressive modulus (Eeq) was linearlyand directly related to the GAG content of the articular car-tilage [r2! 0.89, P< 0.0001, Fig. 5(B)]. The GAG contentfor this result was derived from the averaged GAG contentof the mated patellar and femoral osteochondral plugs (see

Fig. 3. Stress relaxation response for articular cartilage from normal control group (left) and after exposure to chondroitinase ABC for 30 h(right). Four incremental 5% strain steps (20% total cartilage thickness for each mated pair) were applied. After each strain step was applied,the sample was allowed to relax until the change in the force was 0.05 N/min. The slope of the line (dotted line) connecting the equilibrium

force after every step was used to compute the equilibrium compressive stiffness.

Fig. 4. CECT in bovine osteochondral plugs. (A) Representative CECT images of control and degraded samples (exposure to chondroitinaseABC for 8 h and for 30 h) demonstrate an increase in CT attenuation of articular cartilage with increased exposure to chondroitinase ABC. (B)Representative Toludine-blue stained sections indicate a progressive loss of GAG (blue staining) from the ECM as indicated by CECT in (A).

187Osteoarthritis and Cartilage Vol. 18, No. 2

Supplementary material). 89% of the variation in the com-pressive modulus of the articular cartilage was accountedfor by the variation in the GAG content of the ECM. Theequilibrium compressive modulus ranged from0.69$ 0.19 MPa for the normal control osteochondral plugsto 0.07$ 0.02 MPa for the GAG depleted osteochondralplugs, which had been exposed to chondroitinase ABC for30 h.

COMPARISON OF X-RAY ATTENUATION WITH EQUILIBRIUMCOMPRESSIVE MODULUS

The change in X-ray attenuation was linearly and in-versely related to the change in the compressive modulusof the cartilage [r2! 0.93, P< 0.0001, Fig. 5(C)]. The X-ray attenuation for this result was obtained by averagingthe CT attenuation for the patellar and femoral cartilage ofeach pair.

Discussion

Minimally invasive diagnostic methods capable of mea-suring early changes in the biochemical composition andbiomechanical properties of articular cartilage associatedwith OA are clinically important for optimizing treatment pro-tocols and assessing the efficacy of chondroprotectivedrugs and cartilage repair strategies. The variation in X-ray attenuation produced by diffusion of an FDA approved,tri-iodinated, anionic contrast agent into articular cartilagematrix measured by QCT accounted for 83% of the variationin GAG content in inverse proportion to the fixed negativecharge density [Fig. 5(A)] and 93% of the variation in theequilibrium compressive modulus [Fig. 5(C)] for mated os-teochondral plugs excised from bovine patella-femoraljoints. The results for the GAG vs CT attenuation relation-ship are consistent with the work of other investiga-tors29,30,44,45 who demonstrated that the X-ray attenuationof diffused iodinated anionic contrast agent into isolated ar-ticular cartilage specimens measured by micro-CT andpQCT was inversely correlated with the GAG content ofthe articular cartilage. In a recent study44 it was demon-strated that the X-ray attenuation of osteochondral plugs ex-posed to an anionic, iodinated, contrast agent measured byCT was negatively correlated with the GAG content of thearticular cartilage and that the subchondral bone did not al-ter the diffusion of the contrast agent into articular cartilageor interfere with the X-ray attenuation profile of the contrastagent in cartilage. However, none of these previously pub-lished studies has established that the X-ray attenuation ofthe diffused anionic contrast agent into the cartilage couldalso be used to determine the equilibrium compressivemodulus of the cartilage in one consistent sample set46,47.This work represents the first to explicitly generatea three-way correlation between CT attenuation, GAG con-tent, and compressive modulus in one consistent sampleset.In our study, the GAG content of the articular cartilage

measured by the DMMB assay accounted for 89% of thevariation in the equilibrium compressive modulus of themated osteochondral plugs [Fig. 5(B)]. In degradedcartilage, it is possible that the diffusion of the contrastagent may reflect changes in the permeability of the tissueindependent of changes in the negative fixed chargedensity48,49. The dependence of the compressive

Fig. 5. Linear regression plots: (A) CT attenuation vs GAG content measured by DMMB assay (individual plugs). (B) Equilibrium compressivemodulus vs GAG content measured by DMMB assay (pair mean). (C) Equilibrium compressive modulus vs CT Attenuation (pair mean). Thelight shaded curve shows the 95% confidence intervals for individual points and the dark shaded curve shows the 95% confidence interval

curve for the mean.

Fig. 6. Spatial distribution of GAG content as predicted by the cor-relation plot of CT attenuation and GAG content for the controlgroup (n! 10). The light shaded curve shows the 95% confidenceintervals for individual points and the dark shaded curve shows the

95% confidence interval curve for the mean.

188 P. N. Bansal et al.: Contrast Enhanced Computed Tomography

properties of articular cartilage on its biochemical constitu-ents and, specifically, on the GAG content has been previ-ously established50e52. Nevertheless, there have also beenreports suggesting that the variation in mechanical proper-ties across a joint surface may not be consistent with thevariation in proteoglycan content53. To also determine ifthe water content was affecting the correlation, we deter-mined the dry weight for all the cartilage samples and gen-erated the corresponding correlation graphs. A subsequentcorrelation plot between the GAG content normalized bywet weight against the GAG content normalized by dryweight gave an r2! 0.96 (see Supplementary material).This result indicates that water content was not a strongcontributor to the correlation.We evaluated the biomechanical properties of the articu-

lar cartilage using mated osteochondral plugs excised fromcongruent and conforming surfaces of bovine patella-femo-ral joints in an unconfined compressive stress relaxation ex-periment. The cartilage surfaces during the mechanicaltesting were treated as one composite stack of cartilagebeing subjected to compressive stress relaxation42. Themated osteochondral plugs simulate in vivo conditionswhere, under compression, local deformation of conformingchondral surfaces produces an interposed fluid film capableof supporting the load and lubricating the cartilage-on-carti-lage interface. The reported equilibrium compressive modu-lus is a composite measure for the two opposing jointsurfaces. This is more indicative of the biomechanical stateof the opposing surfaces as observed in the clinic where thehealth of joint is under investigation. To compare the valuesof the equilibrium compressive modulus obtained with theexperimental procedure used in our study with the morecommon procedure of measuring the equilibrium compres-sive modulus using cartilage against an artificial surface(metal or glass plate)10,26,27, we performed a series of addi-tional mechanical experiments. The differences betweenmoduli obtained from the cartilage on cartilage and thoseobtained from cartilage-on-steel were statistically insignifi-cant (P! 0.73) (see Supplementary material for further de-tails). However, it is important to note that this method oftesting does not include the effect of thickness or deforma-tion of each surface while determining the equilibrium mod-ulus. Additional studies are underway to better characterizethis model.In the present study, the diffusion kinetics may be differ-

ent than that found in vivo. However, CECT works by thesame physical principles as dGEMRIC, which has beenshown to measure successfully the GAG content and thespatial distribution of GAG in whole joints ex vivo andin vivo16e19. In clinical applications using dGEMRIC toevaluate the articular cartilage of the hip and knee, repet-itive loading of the joint enhanced the mass transfer of thecontrast agent into the cartilage by convection. Havingpatients walk or actively move their limbs after injectingthe joint with an anionic, iodinated contrast dye couldaccelerate the mass transfer of the contrast agent intothe cartilage in clinical applications. It should be notedhowever, that the toxicity of the contrast agent should betaken into account before advising patients to walk orperform any activity that will result in significant loads onthe tissue. This would be especially true for techniquesutilizing intra-articular injections.The correlation between the gadolinium enhanced T1 in-

dex measured in cartilage by MRI and the mechanical prop-erties of articular cartilage has been inconsistent. Usingindentation tests to measure the stiffness of human tibialplateaus, the gadolinium enhanced T1 index correlated

well with cartilage stiffness when T1 was computed at theregions where the cartilage was indented25,28. However,the T1 index value, when computed for the entire depth ofthe cartilage did not correlate with the stiffness. Using stan-dard unconfined compression tests to evaluate the mechan-ical properties of cartilage, the gadolinium enhanced T1

index correlated well with compressive stiffness measuredin human26 and bovine articular cartilage specimens27.However, there have been reports where the gadoliniumenhanced T1 index correlated poorly with the equilibriumcompressive modulus for patellar cartilage specimens23.The lack of correlation between the gadolinium enhancedT1 index and compressive modulus in some studies hasbeen attributed to the effect of collagen on the compressivestiffness of cartilage and the effect of variations in collagencontent on the diffusion of gadolinium into the cartilage (es-pecially in young animals and humans)24. The sensitivity ofCECT to the collagen content and collagen fibril orientationof articular cartilage will need further investigation espe-cially with tissues from different species and at differentmaturation stages. Nonetheless, in our study the X-ray at-tenuation of the diffused anionic iodinated contrast agentmeasured by QCT accounted for 93% of the variation inthe equilibrium compressive modulus.With the increasing availability of multi-detector, helical

CT scanners, we believe that the higher resolution(w200 mm) and faster acquisition times (w5 min) of theseinstruments will enable CECT to be a viable alternative todGEMRIC and other MRI-based techniques for determiningcartilage health. Post-processing of sequential transaxialCT images through whole joints allows the rapid generationof segmented 3D reconstructions and tomographic 2D im-ages of cartilage and subchondral bone without the needfor special sequences45. Since changes to the subchondralbone architecture are also a hallmark of OA54e56, the abilityof CECT to simultaneously image the subchondral bonemorphology and articular cartilage surface layer adds tothe attractiveness of this technique. However, repetitivescans using CECT may pose potential risks of radiation ex-posure, especially for imaging of the hip.In order to evaluate the potential of this technique to pre-

dict the spatial distribution of GAG content in the patello-femoral surfaces used in this study, we correlated theGAG content and the CT attenuation in the control groupalone. Figure 6 shows that CT attenuation predicts 83%of the variation in the GAG content in the control group.The significant correlation shows that CECT is able to pre-dict the narrow physiologic variation in the GAG content, butthis needs to be further validated in a larger cohort of sam-ples from different sources and species.The work presented here defines the interrelationships

between X-ray attenuation, GAG content, and equilibriumcompressive modulus as well as demonstrates that CECTusing an iodinated anionic contrast agent can quantify theGAG content and compressive stiffness of articular carti-lage in an ex vivo bovine osteochondral plug model.Ex vivo studies using human and bovine cadaveric jointsand in vivo studies to monitor GAG loss are underway tofurther validate the ability of CECT to monitor changes inthe GAG content and compressive modulus of articularcartilage as a measure of cartilage health and disease.Continued pre-clinical work using CECT based imaging pro-cedures will facilitate the transition of this diagnostic methodto the clinic, where it can aid physicians in assessing thebiochemical and biomechanical health of articular cartilagein order to monitor disease progression and to optimizetreatment protocols.

189Osteoarthritis and Cartilage Vol. 18, No. 2

Conflict of interest

The authors have no conflicts of interest.

Supplementary material

Supplementary material for this article can be found, inthe online version, at doi:10.1016/j.joca.2009.09.003.

References

1. Mow VC, Huiskes R. Basic Orthopaedic Biomechanics & Mechano-biol-ogy. 3rd edn. Philadelphia: Lippincott Williams & Wilkins; 2005.

2. Athanasiou KA, Rosenwasser MP, Buckwalter JA, Malinin TI, Mow VC.Interspecies comparisons of in situ intrinsic mechanical properties ofdistal femoral cartilage. J Orthop Res 1991;9(3):330e40.

3. Grushko G, Schneiderman R, Maroudas A. Some biochemical and bio-physical parameters for the study of the pathogenesis of osteoarthritis:a comparison between the processes of ageing and degeneration in hu-man hip cartilage. Connect Tissue Res 1989;19(2e4):149e76.

4. Brandt KD. Cartilage Changes in Osteoarthritis. Indianapolis, Ind: Indi-ana University School of Medicine; 1990.

5. Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG,Jordan JM, et al. Osteoarthritis: new insights. Part 1: the diseaseand its risk factors. Ann Intern Med 2000;133(8):635e46.

6. Buckwalter JA, Mankin HJ. Articular cartilage: degeneration and osteo-arthritis, repair, regeneration, and transplantation. Instr Course Lect1998;47:487e504.

7. Eckstein F, Buck RJ, Wyman BT, Kotyk JJ, Le Graverand MP,Remmers AE, et al. Quantitative imaging of cartilage morphology at3.0 Tesla in the presence of gadopentate dimeglumine (Gd-DTPA).Magn Reson Med 2007;58(2):402e6.

8. Eckstein F, Cicuttini F, Raynauld JP, Waterton JC, Peterfy C. Magneticresonance imaging (MRI) of articular cartilage in knee osteoarthritis(OA): morphological assessment. Osteoarthritis Cartilage 2006;14(Suppl A):A46e75.

9. Eckstein F, Mosher T, Hunter D. Imaging of knee osteoarthritis: data be-yond the beauty. Curr Opin Rheumatol 2007;19(5):435e43.

10. Kneeland JB, Reddy R. Frontiers in musculoskeletal MRI: articular car-tilage. J Magn Reson Imaging 2007;25(2):339e44.

11. Goebel JC, Watrin-Pinzano A, Bettembourg-Brault I, Odille F,Felblinger J, Chary-Valckenaere I, et al. Age-related quantitativeMRI changes in healthy cartilage: preliminary results. Biorheology2006;43(3-4):547e51.

12. Reddy R, Insko EK, Noyszewski EA, Dandora R, Kneeland JB,Leigh JS. Sodium MRI of human articular cartilage in vivo. MagnReson Med 1998;39(5):697e701.

13. Wheaton AJ, Borthakur A, Kneeland JB, Regatte RR, Akella SV,Reddy R. In vivo quantification of T1rho using a multislice spin-lockpulse sequence. Magn Reson Med 2004;52(6):1453e8.

14. WheatonAJ, BorthakurA, ShapiroEM,RegatteRR,AkellaSV,Kneeland JB,et al. Proteoglycan loss in human knee cartilage: quantitation with sodiumMR imagingefeasibility study. Radiology 2004;231(3):900e5.

15. Wheaton AJ, Dodge GR, Elliott DM, Nicoll SB, Reddy R. Quantificationof cartilage biomechanical and biochemical properties via T1rho mag-netic resonance imaging. Magn Reson Med 2005;54(5):1087e93.

16. Bashir A, Gray ML, Hartke J, Burstein D. Nondestructive imaging of hu-man cartilage glycosaminoglycan concentration by MRI. Magn ResonMed 1999;41(5):857e65.

17. Sawamura C, Takahashi M, McCarthy KJ, Shen Z, Fukai N,Rodriguez EK, et al. Effect of in ovo immobilization on developmentof chick hind-limb articular cartilage: an evaluation using micro-MRImeasurement of delayed gadolinium uptake. Magn Reson Med2006;56(6):1235e41.

18. Tiderius CJ, Tjornstrand J, Akeson P, Sodersten K, Dahlberg L,Leander P. Delayed gadolinium-enhanced MRI of cartilage (dGEM-RIC): intra- and interobserver variability in standardized drawing of re-gions of interest. Acta Radiol 2004;45(6):628e34.

19. Young-Jo K, Jaramillo D, Millis MB, Gray ML, Burstein D. Assessment ofearly osteoarthritis in hip dysplasia with delayed gadolinium-enhancedmagnetic resonance imaging of cartilage. J Bone Joint Surg Am 2003;85(10):1987.

20. Gray ML, Burstein D, Kim YJ, Maroudas A. 2007 Elizabeth Winston La-nier Award Winner. Magnetic resonance imaging of cartilage glycos-aminoglycan: basic principles, imaging technique, and clinicalapplications. J Orthop Res 2008;26(3):281e91.

21. Bashir A, Gray ML, Burstein D. Gd-DTPA2# as a measure of cartilagedegradation. Magn Reson Med 1996;36(5):665e73.

22. Tiderius CJ, Jessel R, Kim YJ, Burstein D. Hip dGEMRIC in asymptom-atic volunteers and patients with early osteoarthritis: the influence oftiming after contrast injection. Magn Reson Med 2007;57(4):803e5.

23. Lammentausta E, Kiviranta P, Nissi MJ, Laasanen SI, Kiviranta M,Nieminen T, et al. T2 relaxation time and delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) of human patellar cartilageat 1.5 T and 9.4 T: Relationships with tissue mechanical properties.J Orthop Res 2006;24(3):366e74.

24. Nissi MJ, Rieppo J, Toyras J, Laasanen MS, Kiviranta I, Nieminen MT,et al. Estimation of mechanical properties of articular cartilage withMRI e dGEMRIC, T2 and T1 imaging in different species with variablestages of maturation. Osteoarthritis Cartilage 2007;15(10):1141e8.

25. Baldassarri M, Goodwin JS, Farley ML, Bierbaum BE, Goldring SR,Goldring MB, et al. Relationship between cartilage stiffness anddGEMRIC index: correlation and prediction. J Orthop Res 2007;25(7):904e12.

26. Kurkijarvi JE, Nissi MJ, Kiviranta I, Jurvelin JS, Nieminen MT. Delayedgadolinium-enhanced MRI of cartilage (dGEMRIC) and T2 character-istics of human knee articular cartilage: topographical variation and re-lationships to mechanical properties. Magn Reson Med 2004;52(1):41e6.

27. Nieminen MT, Toyras J, Laasanen MS, Silvennoinen J, Helminen HJ,Jurvelin JS. Prediction of biomechanical properties of articular carti-lage with quantitative magnetic resonance imaging. J Biomech2004;37(3):321e8.

28. Samosky JT, Burstein D, Eric Grimson W, Howe R, Martin S, Gray ML.Spatially-localized correlation of dGEMRIC-measured GAG distribu-tion and mechanical stiffness in the human tibial plateau. J OrthopRes 2005;23(1):93e101.

29. Kallioniemi AS, Jurvelin JS, Nieminen MT, Lammi MJ, Toyras J. Con-trast agent enhanced pQCT of articular cartilage. Phys Med Biol2007;52(4):1209e19.

30. Palmer AW, Guldberg RE, Levenston ME. Analysis of cartilage matrixfixed charge density and three-dimensional morphology via contrast-enhanced microcomputed tomography. Proc Natl Acad Sci U S A2006;103(51):19255e60.

31. Armstrong CG, Lai WM, Mow VC. An analysis of the unconfined com-pression of articular cartilage. J Biomech Eng 1984;106(2):165e73.

32. Bader DL, Kempson GE. The short-term compressive properties of adulthuman articular cartilage. Biomed Mater Eng 1994;4(3):245e56.

33. Kempson GE, Tuke MA, Dingle JT, Barrett AJ, Horsfield PH. The effectsof proteolytic enzymes on the mechanical properties of adult humanarticular cartilage. Biochim Biophys Acta 1976;428(3):741e60.

34. Otsuki S, Brinson DC, Creighton L, Kinoshita M, Sah RL, D’Lima D, et al.The effect of glycosaminoglycan loss on chondrocyte viability: a studyon porcine cartilage explants. Arthritis Rheum 2008;58(4):1076e85.

35. Suzuki S, Saito H, Yamagata T, Anno K, Seno N, Kawai Y, et al. Forma-tion of three types of disulfated disaccharides from chondroitin sul-fates by chondroitinase digestion. J Biol Chem 1968;243(7):1543e50.

36. Sawamura C, Masaya T, Degoricija Lovorka, Patel Nipun, Fukai Naomi,Snyder Brian. Glycosaminoglycan content of osteoarticular speci-mens can be measured by micro CT using negatively charged iodin-ated contrast agent. In: Transactions of the Orthopaedic ResearchSociety 2007.

37. Asanbaeva A, Masuda K, Thonar EJ, Klisch SM, Sah RL. Mechanismsof cartilage growth: modulation of balance between proteoglycan andcollagen in vitro using chondroitinase ABC. Arthritis Rheum 2007;56(1):188e98.

38. Chih-Tung C, Kenneth WF, Peter AT, Amy H, Richard GSS,Walter Jr EH. Matrix fixed-charge density as determined by magneticresonance microscopy of bioreactor-derived hyaline cartilage corre-lates with biochemical and biomechanical properties. Arthritis Rheum2003;48(4):1047e56.

39. Hernanz-Schulman M, Vanholder R, Waterloos MA, Hakim R,Schulman G. Effect of radiographic contrast agents on leukocyte met-abolic response. Pediatr Radiol 2000;30(6):361e8.

40. Koneru P, Kaufman RA, Talati AJ, Jenkins MB, Korones SB. Successfultreatment of sodium polystyrene sulfonate bezoars with serial water-soluble contrast enemas. J Perinatol 2003;23(5):431e3.

41. Urban JP, Hall AC, Gehl KA. Regulation of matrix synthesis rates by theionic and osmotic environment of articular chondrocytes. J Cell Phys-iol 1993;154(2):262e70.

42. Malcom LL. An experimental investigation of the frictional and deforma-tional responses of articular cartilage interfaces to static and dynamicloading. Thesis Ph.D. University of California, San Diego Dept of Ap-plied Mechanics Engineering Sciences; 1976.

43. Farndale RW, Buttle DJ, Barrett AJ. Improved quantitation and discrim-ination of sulphated glycosaminoglycans by use of dimethylmethyleneblue. Biochim Biophys Acta 1986;883(2):173e7.

44. Silvast TS, Jurvelin JS, Lammi MJ, Toyras J. pQCT study on diffusionand equilibrium distribution of iodinated anionic contrast agent in hu-man articular cartilageeassociations to matrix composition and integ-rity. Osteoarthritis Cartilage 2009;17(1):26e32.

190 P. N. Bansal et al.: Contrast Enhanced Computed Tomography

45. Xie L, Lin AS, Levenston ME, Guldberg RE. Quantitative assessment ofarticular cartilage morphology via EPIC-microCT. Osteoarthritis Carti-lage 2009;17(3):313e20.

46. Aula AS, Jurvelin JS, Toyras J. Simultaneous computed tomography ofarticular cartilage and subchondral bone. Osteoarthritis Cartilage 2009;17(12):1583e8.

47. Silvast TS, Jurvelin JS, Aula AS, Lammi MJ, Toyras J. Contrast agent-en-hancedcomputed tomographyof articular cartilage: associationwith tis-sue composition and properties. Acta Radiol 2009;50(1):78e85.

48. Maroudas A. Biophysical chemistry of cartilaginous tissueswith special ref-erence to solute and fluid transport. Biorheology 1975;12(3e4):233e48.

49. Evans RC, Quinn TM. Solute diffusivity correlates with mechanical prop-erties and matrix density of compressed articular cartilage. Arch Bio-chem Biophys 2005;442(1):1e10.

50. Harris ED, Parker HG, Radin EL, Krane SM. Effects of proteolytic en-zymes on structural and mechanical properties of cartilage. ArthritisRheum 1972;15(5):497e503.

51. Treppo S, Koepp H, Quan EC, Cole AA, Kuettner KE, Grodzinsky AJ.Comparison of biomechanical and biochemical properties of carti-lage from human knee and ankle pairs. J Orthop Res 2000;18(5):739e48.

52. Frank EH, Grodzinsky AJ, Koob TJ, Eyre DR. Streaming potentials:a sensitive index of enzymatic degradation in articular cartilage.J Orthop Res 1987;5(4):497e508.

53. Franz T, Hasler EM, Hagg R, Weiler C, Jakob RP, Mainil-Varlet P. In situcompressive stiffness, biochemical composition, and structural integ-rity of articular cartilage of the human knee joint. Osteoarthritis Carti-lage 2001;9(6):582e92.

54. RadinEL,BurrDB,CatersonB,FyhrieD,BrownTD,BoydRD.Mechanicaldeter-minants of osteoarthrosis. Semin Arthritis Rheum 1991;21(3 Suppl 2):12e21.

55. Radin EL. Who gets osteoarthritis and why? An update. J Rheumatol2005;32(6):1136e8.

56. Brandt KD, Radin EL, Dieppe PA, van de Putte L. Yet more evidence that os-teoarthritis is not a cartilagedisease. AnnRheumDis 2006;65(10):1261e4.

191Osteoarthritis and Cartilage Vol. 18, No. 2